TY - JOUR
AU - Babar, Muhammad
AB - 1. Introduction Renewable energy plays a vital role in the energy sector. The renewable energy sector has witnessed phenomenal growth in recent times. Despite the recent pandemic and looming uncertainties, the development of renewables remains strong, and the industry saw an increase of 15% from January 2020 to October 2020 as opposed to for the same period. Predominantly led by China and the United States, the renewable energy sector is set to grow by 4% globally to reach up to 200 GW. Among them, the solar power PV sector is estimated to jump by 30% in many countries. IEA estimates that the sector will likely witness a surge in Europe and India by 2021 and is expected to peak by 10–12% [1]. Fig 1 shows that the total solar energy generated accounted for 2000 gigawatts, suggesting that renewable energy investments have increased significantly worldwide [2], with an annual growth rate above 20% in most emerging nations. This amazing rise also resulted in a 26 percent reduction in the weighted cost of electricity, with solar alone accounting for a 13 percent reduction [3]. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. Solar PV generation from 2008–2018. https://doi.org/10.1371/journal.pone.0286573.g001 Solar PV collectors are widely used for harnessing the energy emanated from the sun and are generally classified as active or passive. A photovoltaic cell or a combination of cells, also known as solar arrays or systems (manufactured through semiconductors), converts the captured light photons into power by employing a photoelectric effect. The application of photovoltaic cells has dominated the renewable energy sector and has found wide use in grid interconnected inverters, electric vehicles, consumer electronics, etc. This growth has prompted researchers to investigate and improve the system’s efficiency [4]. The sun’s photons’ collection is generally achieved using sun-tracking PV’s. A typical solar module has limited energy (DC supply), and the internal impedances of the cell vary throughout the day and largely depend on the solar irradiance on the module. The sun-tracking PVs maximize the collection by continually adjusting themselves to the global solar insolation shifts. The captured sunlight is then amplified to generate a constant steady voltage. The efficiency of the collection, therefore, becomes paramount. To achieve max efficiency, most solar systems employ an MPPT controller. A Maximum Power Point Tracking Controller (MPPT) is used as a tracking mechanism and is a crucial element of the solar energy system. 2. MPPT system characteristics 2.1. Maximum power point tracking and partial shading problem The general goal of any MPPT tracker is to ensure accuracy, precision, and robustness [6]. A typical problem that regularly hampers the collection is the partial shading effect. Two methods are widely adopted for point tracking: the mechanical rotational tracking method that can be implemented by embedded automation and the electric power traction method. Researchers have extensively researched these areas and compared various techniques [7, 8]. Various power point tracking methods have been employed to fix the partial shading effect. A variable INR-MPPT method for determining the scaling factor, which is crucial to the steady performance of an MPPT under partial shading conditions is proposed [9]. Handle the problem of partial shading by using single-state grid-connected voltage source inverters in their study [10]. As the partial shading conditions have more than one power point, the accurate tracking of the same using conventional algorithms is tricky. An attempt was made to address this problem by adopting a global search space differential algorithm by [11]. Their experiments indicate that the global maximum power point tracking using a differential algorithm has greater efficiency and accuracy of 99%, exhibits faster operation PSIM simulated environment, and responds to changing load variation at 0.1 sec [11]. Propose an improved phase shift PWM-based MPPT to overcome partial shading conditions in distributed PV systems [12]. Numerous researchers study various techniques and improve DC-DC boost converter technologies. The most common and adopted techniques employ DC-DC boost converters using MPPT and LPT (limited power tracking). As PV voltages are not entirely dependent on solar irradiance, the MPPTs can be programmed to reach high dynamic and static efficiencies when operating at maximum speeds. Similarly, high speed can be obtained in the LPT by controlling the inductor current. The main challenge in this operation is that both voltages and current depend on the array’s cell temperature, operation point, and solar irradiance. This could potentially cause interaction loops and instability. This loop and instability are generally handled by compensating input capacitance and employing a low-speed voltage controller. The method provides more dynamic response and better tracking ability due to the PV resistance’s ability and function without an extra current sensor [13]. Improved DC/DC model predictive control-based MPPT (MPC-MPPT). Unlike the PV DC/DC converter, where traditional MPC-based converters employ two voltage and one current sensor, the proposed model employs two sensors to reduce costs. The improved model is also said to operate on both fixed and variable switching frequencies. The test result of the suggested model is said to improve the gain by ten times the input voltage with an efficiency of 93% [14]. Apart from the above, several other MPPT techniques have been tested and used, including hill climbing, perturb and observation (P&O), incremental conductance, and resistance [15]. Researchers have also adopted and studied Artificial Intelligence and machine learning techniques to improve the efficiency of the MPPT algorithms. Among them, the fuzzy logic-based algorithms are most widely due to it has proven track record as an effective solution in non-linear PV systems without the need for accurate system data [16–23]. An incremental conductance based on a fuzzy duty cycle to overcome the problems of IC controller-based fuzzy logic system [24]. They implement and test the system on an STP085B panel to verify the accuracy using MATLAB [24]. Implementing AI-based systems in addressing partial shading effects offers better prospects due to its various advantages and capabilities, such as nonlinear mapping, response time, robustness, and minimum computations. AI-based systems are researched more to improve the system. Surveys various AI-based systems and propose a new ANN MPPT-based system to address the limitations of PV MPPTs. The results indicate the proposed system’s superiority over other surveyed systems [19]. Various MPPT controllers suggest a solution using AI techniques to improve the efficiency of the MPPT and address the partial shading effect [20]. Polynomial-based algorithms on the MPPTs to increase the performance and efficiency and establish their model on the Takagi-Sugeno fuzzy model. The model uses the minimum time to track, thereby increasing efficiency. Unlike the conventional methods, their model does not require the maximum power operational point [25]. An adaptive fuzzy logic-based system, demonstrate their model in a simulated C-Block PSIM software-based environment, and verify the feasibility of a floating-point digital signal processing controller (TMS320F28335). They conclude that the proposed model exhibits greater consistency and accuracy in the output power of the PV system [26]. Artificial Neuro-Fuzzy Inference System (ANFIS) based on a particle swarm optimization algorithm is proposed to achieve maximum power output with minimum oscillation tracking and demonstrate their model’s superiority due to its enhanced drive control to enhance PV extraction [5–27]. Though most controllers exhibit enhanced tracking abilities and efficiency, these control techniques are complex. The response time to load variations is also quite normal. An enhanced Digital Signal Processing (DSP) technique based independent MPPT control scheme was demonstrated [28]. Among the various techniques, the Perturb and observe technique is a popular hill-climbing method and is widely used due to its easy operation and the simplicity of the design [26]. The ease of implementation and enhanced performance makes the P&O algorithm popular [29]. Unlike the incremental conductance method, this method requires oscillating power output to determine the maximum power point under steady-state irradiance. To mitigate the oscillations’ limitations, Ahmed and Salam propose an enhanced adaptive perturb and observe (EA-P&O) method to mitigate the limitations of the diverged tracking. Using four algorithms, they simulate the experiment on a buck-boost converter with a dSpace DS1104 board in Matlab. The four algorithms artificial bee colony, modified incremental conduction, cuckoo search and hybrid ant colony. The test results indicate that the proposed method tracks global peak successfully under partial shading conditions while maintaining an efficiency of over 99% [30]. A self-predictive perturb and observe (SPP&O) algorithm, and demonstrate its function using Matlab. They obtain successful results in each iteration with a 90% reduction in step size to provide an oscillation free and enhanced steady performance [31]. Based on the self-predictive algorithm Kumar et al. further study and propose an incremental conductance algorithm (SA-I&C) with higher efficiency for tracking the maximum power point while minimizing the development cost [32]. Though P&O technique is effective and widely used, the optimization of the MPPT requires two design parameters, the perturbation frequency and the perturbation step. Both these require specifics, and the perturbation has a drawback as the combined energy conversion limits it while step-size must be high to differentiate the responses in the system. While this technique has explicit guidelines for a single loop, no algorithm governs the multi-loop MPPT structures. This was elucidated and researched by Kivimaki et al., who provided a solution by modifying the perturbation frequency. They conclude that the influence of PVG dynamic resistance is negligible and validates the proposed guidelines [29]. 2.2. Battery system characteristics Battery power management is the most crucial technique in the inverter. Numerous researchers have studied various technologies to improve battery efficiency, life, and storage [33–37]. Compared to Pb Acid batteries, Li-IoN battery exhibit better-charging efficiency, great cycling time and self-discharging capability, thus reliable in various applications like consumer electronics, EVs, renewable energy and grid energy storage. The extensive application of the batteries also means fast and efficient charging. While fast charging applications may increase the convenience of the battery application, it is also essential to protect them from damage. This technically means ensuring safety, less charging duration and a healthy battery. Zou et al. propose a health-aware fast charging based on advanced control theory and formulate the problem based on a linear time-varying model. The batteries, in this model, are ensured protection by controlling the physical variables. This method also results in fast charging of the batteries in real-time applications [38]. Constant Current and Voltage (CC & CV) based power converters etc., to name a few. The most used charging mechanisms employ an open-loop approach among the battery charging techniques. Closed-loop instantaneous cell charging techniques are essential to charge faster on the batteries. Therefore, a closed-loop Constant Temperature Constant-Voltage Charging Technique (CT-CV) charging control was proposed by Patnaik et al., where the battery cell temperature is maintained by implying a closed control temperature control mechanism. The results indicated faster charging time by over 20% than a conventional CC-CV charger [39]. One other typical problem is the charging and discharging of the batteries. The charging and discharging of the battery are determined by the State of Charge (SoC) calculation to avoid overcharging the batteries. The charge state in a battery is indirectly measured and estimated based on several measures like voltage, current, and temperature. Researchers have proposed many non-AI methods to measure SoC. These methods have serious and result in an inaccurate prediction. The ability of AI algorithms for greater efficiency and accuracy enhances the accurate determination of the SoC. Song et al. study Convolutional Neural Network and Long Short-Term Memory (CNN-LSTM) networks and train the network using data collected from discharge profiles. The trained results indicate that the network records the data, the relationship of the SoC and variables and exhibits smooth and accurate results. The experimental result of CNN and LSTM network record the real value of SoC even under nonlinear conditions and the Root Mean Square Error (RMSE) as low as 2% [40]. The open-loop methods pose uncertainties in the distribution, and the generated SoC value is likely to have errors that accumulate over time and have adverse estimation accuracy [41–44]. To overcome problems in SoC, like in cubature Kalman algorithms, [44] present a modified algorithm using Adaptive Weighting Cubature Particle Filter. The results presented indicate high estimation accuracy, strong robustness, fast convergence speed and an estimated error of less than 1% [44]. IoT based applications used in different secotors for the energy efficiency [45, 46]. Peak power management strategy plays an important role in allowing batteries to reconsider its VI characteristics during peak power consumption and energy-efficient power flow and reduce the development cost. The real-time state of health in the batteries is quite difficult to obtain due to the various parameters required for capacity measurements. Tan et al. study SoH estimation strategy dependent on the Equivalent Internal Resistance (EIR). They adapted the support vector regression method for computing the real-time SoH. The results indicate that the relationship between EIR and battery deterioration can be predicted with greater accuracy and ensure efficiency and monitoring of the batteries [47]. The review of the various technological application on MPPT and PV System components has indicated that closed-loop MPP tracking topologies, energy-efficient BMS and boost converter technologies are reliable. Every algorithm applied meets respective objectives and shows good performance. However, no system explains all the issues and proposes a single solution. This paper aims to propose a new design to address all the issues as a single solution. 2.1. Maximum power point tracking and partial shading problem The general goal of any MPPT tracker is to ensure accuracy, precision, and robustness [6]. A typical problem that regularly hampers the collection is the partial shading effect. Two methods are widely adopted for point tracking: the mechanical rotational tracking method that can be implemented by embedded automation and the electric power traction method. Researchers have extensively researched these areas and compared various techniques [7, 8]. Various power point tracking methods have been employed to fix the partial shading effect. A variable INR-MPPT method for determining the scaling factor, which is crucial to the steady performance of an MPPT under partial shading conditions is proposed [9]. Handle the problem of partial shading by using single-state grid-connected voltage source inverters in their study [10]. As the partial shading conditions have more than one power point, the accurate tracking of the same using conventional algorithms is tricky. An attempt was made to address this problem by adopting a global search space differential algorithm by [11]. Their experiments indicate that the global maximum power point tracking using a differential algorithm has greater efficiency and accuracy of 99%, exhibits faster operation PSIM simulated environment, and responds to changing load variation at 0.1 sec [11]. Propose an improved phase shift PWM-based MPPT to overcome partial shading conditions in distributed PV systems [12]. Numerous researchers study various techniques and improve DC-DC boost converter technologies. The most common and adopted techniques employ DC-DC boost converters using MPPT and LPT (limited power tracking). As PV voltages are not entirely dependent on solar irradiance, the MPPTs can be programmed to reach high dynamic and static efficiencies when operating at maximum speeds. Similarly, high speed can be obtained in the LPT by controlling the inductor current. The main challenge in this operation is that both voltages and current depend on the array’s cell temperature, operation point, and solar irradiance. This could potentially cause interaction loops and instability. This loop and instability are generally handled by compensating input capacitance and employing a low-speed voltage controller. The method provides more dynamic response and better tracking ability due to the PV resistance’s ability and function without an extra current sensor [13]. Improved DC/DC model predictive control-based MPPT (MPC-MPPT). Unlike the PV DC/DC converter, where traditional MPC-based converters employ two voltage and one current sensor, the proposed model employs two sensors to reduce costs. The improved model is also said to operate on both fixed and variable switching frequencies. The test result of the suggested model is said to improve the gain by ten times the input voltage with an efficiency of 93% [14]. Apart from the above, several other MPPT techniques have been tested and used, including hill climbing, perturb and observation (P&O), incremental conductance, and resistance [15]. Researchers have also adopted and studied Artificial Intelligence and machine learning techniques to improve the efficiency of the MPPT algorithms. Among them, the fuzzy logic-based algorithms are most widely due to it has proven track record as an effective solution in non-linear PV systems without the need for accurate system data [16–23]. An incremental conductance based on a fuzzy duty cycle to overcome the problems of IC controller-based fuzzy logic system [24]. They implement and test the system on an STP085B panel to verify the accuracy using MATLAB [24]. Implementing AI-based systems in addressing partial shading effects offers better prospects due to its various advantages and capabilities, such as nonlinear mapping, response time, robustness, and minimum computations. AI-based systems are researched more to improve the system. Surveys various AI-based systems and propose a new ANN MPPT-based system to address the limitations of PV MPPTs. The results indicate the proposed system’s superiority over other surveyed systems [19]. Various MPPT controllers suggest a solution using AI techniques to improve the efficiency of the MPPT and address the partial shading effect [20]. Polynomial-based algorithms on the MPPTs to increase the performance and efficiency and establish their model on the Takagi-Sugeno fuzzy model. The model uses the minimum time to track, thereby increasing efficiency. Unlike the conventional methods, their model does not require the maximum power operational point [25]. An adaptive fuzzy logic-based system, demonstrate their model in a simulated C-Block PSIM software-based environment, and verify the feasibility of a floating-point digital signal processing controller (TMS320F28335). They conclude that the proposed model exhibits greater consistency and accuracy in the output power of the PV system [26]. Artificial Neuro-Fuzzy Inference System (ANFIS) based on a particle swarm optimization algorithm is proposed to achieve maximum power output with minimum oscillation tracking and demonstrate their model’s superiority due to its enhanced drive control to enhance PV extraction [5–27]. Though most controllers exhibit enhanced tracking abilities and efficiency, these control techniques are complex. The response time to load variations is also quite normal. An enhanced Digital Signal Processing (DSP) technique based independent MPPT control scheme was demonstrated [28]. Among the various techniques, the Perturb and observe technique is a popular hill-climbing method and is widely used due to its easy operation and the simplicity of the design [26]. The ease of implementation and enhanced performance makes the P&O algorithm popular [29]. Unlike the incremental conductance method, this method requires oscillating power output to determine the maximum power point under steady-state irradiance. To mitigate the oscillations’ limitations, Ahmed and Salam propose an enhanced adaptive perturb and observe (EA-P&O) method to mitigate the limitations of the diverged tracking. Using four algorithms, they simulate the experiment on a buck-boost converter with a dSpace DS1104 board in Matlab. The four algorithms artificial bee colony, modified incremental conduction, cuckoo search and hybrid ant colony. The test results indicate that the proposed method tracks global peak successfully under partial shading conditions while maintaining an efficiency of over 99% [30]. A self-predictive perturb and observe (SPP&O) algorithm, and demonstrate its function using Matlab. They obtain successful results in each iteration with a 90% reduction in step size to provide an oscillation free and enhanced steady performance [31]. Based on the self-predictive algorithm Kumar et al. further study and propose an incremental conductance algorithm (SA-I&C) with higher efficiency for tracking the maximum power point while minimizing the development cost [32]. Though P&O technique is effective and widely used, the optimization of the MPPT requires two design parameters, the perturbation frequency and the perturbation step. Both these require specifics, and the perturbation has a drawback as the combined energy conversion limits it while step-size must be high to differentiate the responses in the system. While this technique has explicit guidelines for a single loop, no algorithm governs the multi-loop MPPT structures. This was elucidated and researched by Kivimaki et al., who provided a solution by modifying the perturbation frequency. They conclude that the influence of PVG dynamic resistance is negligible and validates the proposed guidelines [29]. 2.2. Battery system characteristics Battery power management is the most crucial technique in the inverter. Numerous researchers have studied various technologies to improve battery efficiency, life, and storage [33–37]. Compared to Pb Acid batteries, Li-IoN battery exhibit better-charging efficiency, great cycling time and self-discharging capability, thus reliable in various applications like consumer electronics, EVs, renewable energy and grid energy storage. The extensive application of the batteries also means fast and efficient charging. While fast charging applications may increase the convenience of the battery application, it is also essential to protect them from damage. This technically means ensuring safety, less charging duration and a healthy battery. Zou et al. propose a health-aware fast charging based on advanced control theory and formulate the problem based on a linear time-varying model. The batteries, in this model, are ensured protection by controlling the physical variables. This method also results in fast charging of the batteries in real-time applications [38]. Constant Current and Voltage (CC & CV) based power converters etc., to name a few. The most used charging mechanisms employ an open-loop approach among the battery charging techniques. Closed-loop instantaneous cell charging techniques are essential to charge faster on the batteries. Therefore, a closed-loop Constant Temperature Constant-Voltage Charging Technique (CT-CV) charging control was proposed by Patnaik et al., where the battery cell temperature is maintained by implying a closed control temperature control mechanism. The results indicated faster charging time by over 20% than a conventional CC-CV charger [39]. One other typical problem is the charging and discharging of the batteries. The charging and discharging of the battery are determined by the State of Charge (SoC) calculation to avoid overcharging the batteries. The charge state in a battery is indirectly measured and estimated based on several measures like voltage, current, and temperature. Researchers have proposed many non-AI methods to measure SoC. These methods have serious and result in an inaccurate prediction. The ability of AI algorithms for greater efficiency and accuracy enhances the accurate determination of the SoC. Song et al. study Convolutional Neural Network and Long Short-Term Memory (CNN-LSTM) networks and train the network using data collected from discharge profiles. The trained results indicate that the network records the data, the relationship of the SoC and variables and exhibits smooth and accurate results. The experimental result of CNN and LSTM network record the real value of SoC even under nonlinear conditions and the Root Mean Square Error (RMSE) as low as 2% [40]. The open-loop methods pose uncertainties in the distribution, and the generated SoC value is likely to have errors that accumulate over time and have adverse estimation accuracy [41–44]. To overcome problems in SoC, like in cubature Kalman algorithms, [44] present a modified algorithm using Adaptive Weighting Cubature Particle Filter. The results presented indicate high estimation accuracy, strong robustness, fast convergence speed and an estimated error of less than 1% [44]. IoT based applications used in different secotors for the energy efficiency [45, 46]. Peak power management strategy plays an important role in allowing batteries to reconsider its VI characteristics during peak power consumption and energy-efficient power flow and reduce the development cost. The real-time state of health in the batteries is quite difficult to obtain due to the various parameters required for capacity measurements. Tan et al. study SoH estimation strategy dependent on the Equivalent Internal Resistance (EIR). They adapted the support vector regression method for computing the real-time SoH. The results indicate that the relationship between EIR and battery deterioration can be predicted with greater accuracy and ensure efficiency and monitoring of the batteries [47]. The review of the various technological application on MPPT and PV System components has indicated that closed-loop MPP tracking topologies, energy-efficient BMS and boost converter technologies are reliable. Every algorithm applied meets respective objectives and shows good performance. However, no system explains all the issues and proposes a single solution. This paper aims to propose a new design to address all the issues as a single solution. 3. Proposed converter design To mitigate the problems in the section, the proposed model aims to provide the solution in four operation stages. Fig 2 describes the block diagram of the proposed work. The PV solar array is reconstructed to maximize the MPP level. The proposed system develops a low-cost and reliable multi-output boost converter to energize multiple rated batteries. The SoC and SoH estimation because of high performance with embedding sensors and the Micro Controller. The GSM and IoT technology used for the communication establishment. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. Energy management system. https://doi.org/10.1371/journal.pone.0286573.g002 3.1. DC-DC converter design To mitigate the problems in the section, the proposed model aims to provide the solution in four operation stages. Fig 2 describes the block diagram of the proposed work. (1) VIp = Input minimum voltage. VOp = output voltage at the desired level. Ƞ = converter side efficiency. The calculation adds the efficiency of the duty cycle due to the dissipated energy by the converter. This equation relies on more duty cyclic range compared with the efficiency factor. (2) The next process calculates the maximum switching current determined by the inductor ripple current. The converters catalogue specifies the range of inductors with their ICs. For the above reason, the inductance value is calculated concerning the ripple current. (3) II(Im) = minimum inductance current. ΔIL = ripple inductance value. Dcy = calculated duty cycle. The maximum output current of the selected chip, IMOP, is the maximum output current, and alternate IC is to limit the high switch current. The value for the IMOP is lower than the required one. Using the selected chip with the inductance within the recommended range is possible. A huge inductance avoids the ripple current and increases the selected chip’s output current. The calculated value is above the maximum current of the application, the switching current at the maximum system is determined, (4) IO(m) = output current at maximum level ΔIL = inductance ripple current The proposed system recommended assigning the inductance with a lower value and smaller size. The inductance must always have a higher current rating than the current maximum given by the equation with higher and lower inductance. For this part, we don’t know the required inductance value; the following equation will give the right inductance value: (5) VIP = Input voltage ΔIL = current at ripples by the inductor. fsm = switching frequency at the minimum level. (6) Io(m) = maximum ripple current required for the specific application. The above equations cannot determine the current at ripples by the inductor due to the presence of an unidentified inductance value. High accuracy for the ripple current is 20% to 40% of the current at the output level. Schottky diodes have huge peak current values than the average rating, therefore, higher current in the system should not pose a problem. The next parameter is to decide the power dissipation of the diode. (7) PFD = Power at Forward Diode. IFD = Current at Forward Diode. VFD = Voltage at Forward Diode. 3.2. Battery management system The battery open circuit voltage (Voc) is shown as an element of battery SoC, as is the interior obstruction R. SoC determines the battery limit. The SoC is used to determine the current status of charging and discharging conditions in the battery (4). The Fuzzy Logic Controller (FLC) determines the charging and discharging currents. The closed-loop control technique determines the load demand response over the battery. Fig 2 shows the EMS. Solar panels, load and lithium-ion batteries are the three blocks in the system. Wind and photovoltaics are non-linear systems. The FLC helps extend the battery life by determining the charging and discharging times of the battery. Different advantages like adaptability, simple design, and resemblance to human intellect make to choose of the FLC controller. Mathematical modelling is not needed for the FLC controller. 3.3. Fuzzy logic controller model basis Fuzzification, inference system and defuzzification blocks are the three basic segments for Fuzzy logic systems. The numerical input values should be converted to linguistic values as the fuzzy uses linguistic variables as inputs. The fuzzifier converts the crisp values into fuzzy values. The membership function is the degree of truth in fuzzy logic [7]. 3.1. DC-DC converter design To mitigate the problems in the section, the proposed model aims to provide the solution in four operation stages. Fig 2 describes the block diagram of the proposed work. (1) VIp = Input minimum voltage. VOp = output voltage at the desired level. Ƞ = converter side efficiency. The calculation adds the efficiency of the duty cycle due to the dissipated energy by the converter. This equation relies on more duty cyclic range compared with the efficiency factor. (2) The next process calculates the maximum switching current determined by the inductor ripple current. The converters catalogue specifies the range of inductors with their ICs. For the above reason, the inductance value is calculated concerning the ripple current. (3) II(Im) = minimum inductance current. ΔIL = ripple inductance value. Dcy = calculated duty cycle. The maximum output current of the selected chip, IMOP, is the maximum output current, and alternate IC is to limit the high switch current. The value for the IMOP is lower than the required one. Using the selected chip with the inductance within the recommended range is possible. A huge inductance avoids the ripple current and increases the selected chip’s output current. The calculated value is above the maximum current of the application, the switching current at the maximum system is determined, (4) IO(m) = output current at maximum level ΔIL = inductance ripple current The proposed system recommended assigning the inductance with a lower value and smaller size. The inductance must always have a higher current rating than the current maximum given by the equation with higher and lower inductance. For this part, we don’t know the required inductance value; the following equation will give the right inductance value: (5) VIP = Input voltage ΔIL = current at ripples by the inductor. fsm = switching frequency at the minimum level. (6) Io(m) = maximum ripple current required for the specific application. The above equations cannot determine the current at ripples by the inductor due to the presence of an unidentified inductance value. High accuracy for the ripple current is 20% to 40% of the current at the output level. Schottky diodes have huge peak current values than the average rating, therefore, higher current in the system should not pose a problem. The next parameter is to decide the power dissipation of the diode. (7) PFD = Power at Forward Diode. IFD = Current at Forward Diode. VFD = Voltage at Forward Diode. 3.2. Battery management system The battery open circuit voltage (Voc) is shown as an element of battery SoC, as is the interior obstruction R. SoC determines the battery limit. The SoC is used to determine the current status of charging and discharging conditions in the battery (4). The Fuzzy Logic Controller (FLC) determines the charging and discharging currents. The closed-loop control technique determines the load demand response over the battery. Fig 2 shows the EMS. Solar panels, load and lithium-ion batteries are the three blocks in the system. Wind and photovoltaics are non-linear systems. The FLC helps extend the battery life by determining the charging and discharging times of the battery. Different advantages like adaptability, simple design, and resemblance to human intellect make to choose of the FLC controller. Mathematical modelling is not needed for the FLC controller. 3.3. Fuzzy logic controller model basis Fuzzification, inference system and defuzzification blocks are the three basic segments for Fuzzy logic systems. The numerical input values should be converted to linguistic values as the fuzzy uses linguistic variables as inputs. The fuzzifier converts the crisp values into fuzzy values. The membership function is the degree of truth in fuzzy logic [7]. 4. Proposed working of the models 4.1. FLC working model The fuzzy controller for charging and discharging can achieve the desired SoC. The lithium-ion battery is used for energy storage. Eqs 7 and 8 indicate the inputs to the FLC. The input to the battery is a variable output current. (8)(9) The energy storage device is a lithium-ion battery, and the fuzzy controller is designed to be in either charging or discharging mode to reach the necessary SOC. As shown in equation ΔSOC and ΔP are the inputs to the fuzzy logic controller, and the variable output current is provided as an input to the battery model. The load power and solar panels generated power difference gives the total power. The input and output membership function are shown in Figs 3 and 4. The membership functions are divided into five grades. The five grades are NB (Negative Big), NS (Negative Small), ZO (Zero), PS (Positive Small), and PB (Positive Big). After establishing the membership function, the current for charge and discharge may be calculated by replacing it with the scaling factor. The fuzzy controller supplies the battery power to the load when the renewable energy system cannot supply the demand. The battery must be discharged when the SoC of the battery is higher than the command SoC. Table 1 demonstrates the proposed system’s fuzzy control rules (Jadhav & Nair, 2019). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. Membership functions for input. https://doi.org/10.1371/journal.pone.0286573.g003 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Energy management system. https://doi.org/10.1371/journal.pone.0286573.g004 Download: PPT PowerPoint slide PNG larger image TIFF original image Table 1. Fuzzy rules of the proposed system. https://doi.org/10.1371/journal.pone.0286573.t001 4.2. Functions and working of the proposed hardware The hardware block diagram is shown in Fig 5. The experimental setup states the rating of a 48V lithium-ion battery charging scheme with an ampere rating of 2.5 Ah. The initial stage of the model is the renewable charging scheme. The input charging section consists of solar panels with a reconfigurable structure adapted to the power generated by the panels. The panel feedback is fed to the microcontroller with the sensing unit Analog to Digital Converter (ADC) calibrated and based on the reference threshold value. If the panel power is insufficient for the converter’s input, it is adjusted with series and parallel connections to balance the power. The reconfigurable solar panel pattern is shown in Fig 6. The second part of the model is the charge controller, the converter side. The output of the solar panel is fed into the converter as input. The charging converter section has the toroidal core inductor, which generates the required battery charging voltage for the coil windings. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Hardware block diagram. https://doi.org/10.1371/journal.pone.0286573.g005 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 6. Real time implementation of the proposed system. https://doi.org/10.1371/journal.pone.0286573.g006 The driver circuit for the converter is designed with the SG3525 PWM generation IC. It has the technique of closed loop PWM generation. That is, the controller adapts to the battery current voltage. This adaptive charging technique eliminates the normal charging scheme. The duty cycle of the converter adjusted depends on the battery current voltage levels. The third part of the model is the SOH and SOC. In this part, battery maintenance and protection form the core. The battery parameters like voltage, current and temperature measurement are controlled using sensors. In this paper, the SOC is calibrated by the ESI reference. The voltage divider rule measures the battery voltage. The battery voltage is limited between the 5V for the microcontroller ATMEGA 2560 chip protection. As the input in battery varies, the ATMEGA chip will recalibrate. The LM 35 temperature sensor will continuously monitor temperature status, feeding the information to the AMTEGA controller. The AMTEGA controller controls any abnormal temperature change recorded by the system, regulating the charging, and discharging with the help of the relay driver circuits. This model uses the AMTEGA’s native source code as developed by the Arduino IDE software platform. The final part of the model is the communication and user interface medium. In this step, the IoT and GSM platforms continuously monitor the entire status like the charging, discharging, temperature status, battery voltage, and current levels. The IoT part of the Wi-Fi-enabled controller uses NODEMCU microcontroller. Its operation is to get data from the main ATMEGA 2560 microcontroller and publish it as “Think speak” to the web server. Think speak is a user-friendly cloud computing webpage. The webpage allows users to create user credentials and data publishing and monitoring accounts. Each user has specific API keys and is unique. The other communication medium used here is the GSM network. The user’s mobile number is programmed into the controller. Any malfunctions recorded by ATMEGA 2560 controller will trigger an SMS to the user. 4.1. FLC working model The fuzzy controller for charging and discharging can achieve the desired SoC. The lithium-ion battery is used for energy storage. Eqs 7 and 8 indicate the inputs to the FLC. The input to the battery is a variable output current. (8)(9) The energy storage device is a lithium-ion battery, and the fuzzy controller is designed to be in either charging or discharging mode to reach the necessary SOC. As shown in equation ΔSOC and ΔP are the inputs to the fuzzy logic controller, and the variable output current is provided as an input to the battery model. The load power and solar panels generated power difference gives the total power. The input and output membership function are shown in Figs 3 and 4. The membership functions are divided into five grades. The five grades are NB (Negative Big), NS (Negative Small), ZO (Zero), PS (Positive Small), and PB (Positive Big). After establishing the membership function, the current for charge and discharge may be calculated by replacing it with the scaling factor. The fuzzy controller supplies the battery power to the load when the renewable energy system cannot supply the demand. The battery must be discharged when the SoC of the battery is higher than the command SoC. Table 1 demonstrates the proposed system’s fuzzy control rules (Jadhav & Nair, 2019). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. Membership functions for input. https://doi.org/10.1371/journal.pone.0286573.g003 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Energy management system. https://doi.org/10.1371/journal.pone.0286573.g004 Download: PPT PowerPoint slide PNG larger image TIFF original image Table 1. Fuzzy rules of the proposed system. https://doi.org/10.1371/journal.pone.0286573.t001 4.2. Functions and working of the proposed hardware The hardware block diagram is shown in Fig 5. The experimental setup states the rating of a 48V lithium-ion battery charging scheme with an ampere rating of 2.5 Ah. The initial stage of the model is the renewable charging scheme. The input charging section consists of solar panels with a reconfigurable structure adapted to the power generated by the panels. The panel feedback is fed to the microcontroller with the sensing unit Analog to Digital Converter (ADC) calibrated and based on the reference threshold value. If the panel power is insufficient for the converter’s input, it is adjusted with series and parallel connections to balance the power. The reconfigurable solar panel pattern is shown in Fig 6. The second part of the model is the charge controller, the converter side. The output of the solar panel is fed into the converter as input. The charging converter section has the toroidal core inductor, which generates the required battery charging voltage for the coil windings. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Hardware block diagram. https://doi.org/10.1371/journal.pone.0286573.g005 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 6. Real time implementation of the proposed system. https://doi.org/10.1371/journal.pone.0286573.g006 The driver circuit for the converter is designed with the SG3525 PWM generation IC. It has the technique of closed loop PWM generation. That is, the controller adapts to the battery current voltage. This adaptive charging technique eliminates the normal charging scheme. The duty cycle of the converter adjusted depends on the battery current voltage levels. The third part of the model is the SOH and SOC. In this part, battery maintenance and protection form the core. The battery parameters like voltage, current and temperature measurement are controlled using sensors. In this paper, the SOC is calibrated by the ESI reference. The voltage divider rule measures the battery voltage. The battery voltage is limited between the 5V for the microcontroller ATMEGA 2560 chip protection. As the input in battery varies, the ATMEGA chip will recalibrate. The LM 35 temperature sensor will continuously monitor temperature status, feeding the information to the AMTEGA controller. The AMTEGA controller controls any abnormal temperature change recorded by the system, regulating the charging, and discharging with the help of the relay driver circuits. This model uses the AMTEGA’s native source code as developed by the Arduino IDE software platform. The final part of the model is the communication and user interface medium. In this step, the IoT and GSM platforms continuously monitor the entire status like the charging, discharging, temperature status, battery voltage, and current levels. The IoT part of the Wi-Fi-enabled controller uses NODEMCU microcontroller. Its operation is to get data from the main ATMEGA 2560 microcontroller and publish it as “Think speak” to the web server. Think speak is a user-friendly cloud computing webpage. The webpage allows users to create user credentials and data publishing and monitoring accounts. Each user has specific API keys and is unique. The other communication medium used here is the GSM network. The user’s mobile number is programmed into the controller. Any malfunctions recorded by ATMEGA 2560 controller will trigger an SMS to the user. 5. Results and discussion The reconstructed PV array with Series Parallel topology must satisfy the three factors. The proposed method should be suitable for uniform and non-uniform irradiance patterns. Module-level reconfiguration should be carried out. The proposed method can be applied to any PV array and strings. The experiment prototype has a battery bank of 48v /2.5Ah lithium-ion batteries, which are connected in series (i.e. each battery rating 3V/2.5Ah). Four segments of 12V rated battery backup meet the required 48V voltage for the experiment. After necessary calibration, the entire battery feedback is fed into the Arduino Mega 2560 microcontroller. Fig 7 shows the battery in an off state with Li-ion battery temperature, B1 and B2 battery voltage. Fig 8 shows the battery in an off state with Li-ion battery temperature, B3 and B4 battery total voltage connected in series. Fig 9 shows the Li-ion battery temperature, B1 and B2 voltage in the battery on condition. Fig 10 shows the Li-ion battery temperature, B3 and B4 total voltage connected in series when the battery at on condition. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 7. Battery in the OFF state and Li-ion battery temperature and B1 and B2 voltage. https://doi.org/10.1371/journal.pone.0286573.g007 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 8. Battery in OFF state with Li-ion battery temperature, B3 and B4 battery total voltage connected in series. https://doi.org/10.1371/journal.pone.0286573.g008 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 9. Battery in ON state with Li-ion battery temperature and B1 and B2 voltage. https://doi.org/10.1371/journal.pone.0286573.g009 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 10. Battery in ON state with Li-ion battery temperature and B3 and B4 and total battery voltage connected in series. https://doi.org/10.1371/journal.pone.0286573.g010 Automatic charging and a discharging system are required to increase battery life. Figs 11 and 12 shows the respective charging and discharging state operation. The state of charge has been calculated by standard cyclic battery power usage. The charging system is activated when the total battery voltage drops below 46V. The load resistance can explain both the charging and discharging system. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 11. Battery charging indication when it drops below 46V. https://doi.org/10.1371/journal.pone.0286573.g011 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 12. Battery discharging indication when it increases over 48V. https://doi.org/10.1371/journal.pone.0286573.g012 The system’s failure to charge and discharge in real-time failures may lead to a battery explosion. Over usage of the battery also likely decreases the battery life. To avoid such problems, testing was carried out for charging and discharging relay status. Fig 13 represents the charging relay problem. The modified closed-loop boost converter design is shown in Fig 14. The converter in this paper is designed to accommodate three different power ratings (12V, 24V and 48V). This increases the converter flexibility based on real-time requirements. The Fig 15 shows the working of the battery energy management system and control-relays operations and load status. Four pairs of Li-ion batteries were used to demonstrate the segmented energy storage system. Each pair consists of 12v (cell voltage 3.4v, 4*3.4 = 13.2 maximum voltage approximately), and the total battery voltage connected in series is around 48v. It clearly shows that the battery temperature will increase gradually when the load is added to the battery. By checking with XML sheet, first, initially single battery source relates to the converter. According to this project, the converter produces 14-16v. Depending on how many batteries are connected in series, converter output will change accordingly, in the range of 14v, 26v, 38V, and 50V, as seen in the XML Sheet. Initially, the batteries connected were 12; subsequently, all four were connected for testing. The last part of the experiment was tested with only two batteries. By checking with the above graph without load, the battery voltage stood at 25.63V. The other parameters, load status, charging relay status and discharging relay status, are identified by the digital state as 0 indicates the OFF state and 1 indicates ON state. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 13. Charging relay problem. https://doi.org/10.1371/journal.pone.0286573.g013 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 14. Proposed converter design. https://doi.org/10.1371/journal.pone.0286573.g014 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 15. IoT platform–Think speak monitoring plot of different parameters. https://doi.org/10.1371/journal.pone.0286573.g015 When the load is turned ON, the battery voltage gradually decreases in its state. The discharging relay state is 1, the charging relay state is 0 in this condition, and the battery state is in the discharging mode till the battery voltage reaches 22.08V. As the battery goes below 22.08V, the discharge state is turned OFF and charging changes the state to 1. This system uses IoT technology as a platform for monitoring and controlling automation. The result of IoT is shown in Fig 15. Battery voltage, load current, battery temperature level, charging relay status, discharging relay status, and Load ON/OFF status can be monitored wirelessly. “Think speak” web server provides the extra benefit of database management, trigger alert management and a graphical user interface option. Fig 16 shows the fault SMS alert. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 16. Fault SMS alert. https://doi.org/10.1371/journal.pone.0286573.g016 6. Conclusion The hardware prototype verified the overall system performance. The automatic reconfigurable pattern of the solar panel resulted in excellent MPP tracking ability under a partial shading effect. The mechanical panels largely depend on solar radiation tracking and do not support efficient tracking. Therefore, this experimental setup is designed to suit reconfiguration, thus achieving efficiency due to the efficient radiation capture. The converter can perform based on the rated battery connected as load. The converter is designed for 12v, 24v and 48v, respectively. The variable switching technique automatically achieves these voltage levels. Battery SoC and SoH are calculated for individual and entire series connected systems. In addition, it provides accurate results in automatic charging and discharging and indicates other problems like charging relay failure and discharging relay failure. IoT-based battery energy systems have restrictions that should be considered. It requires a constant internet connection and depends on a dependable power source. They may become inoperable due to internet or power outages, making them difficult to scale. Custom development or integration is needed to adapt to changes in fleet size or configuration. Acknowledgments The authors would like to acknowledge the support of Prince Sultan University for paying the Article Processing Charges (APC) of this publication.
TI - IoT based battery energy monitoring and management for electric vehicles with improved converter efficiency
JF - PLoS ONE
DO - 10.1371/journal.pone.0286573
DA - 2023-10-05
UR - https://www.deepdyve.com/lp/public-library-of-science-plos-journal/iot-based-battery-energy-monitoring-and-management-for-electric-1Auw33ZBDB
SP - e0286573
VL - 18
IS - 10
DP - DeepDyve
ER -