Solution method of three-phase voltage errors in the transformer based on the genetic algorithmZhong, Yao; Liu, Qingchan; Li, Tengbin; Chang, Junchao; Yang, Guangrun
doi: 10.1117/12.2680444pmid: N/A
In terms of the problem of solving the three-phase voltage errors in the transformer, a method for solving the three-phase voltage errors in the transformer is proposed in this paper based on the genetic algorithm. By encoding the numerical groups of the three-phase voltage errors in the transformer and setting constraints such as the three-phase imbalance rate and voltage difference value, errors are solved by multiple rounds of iteration. The results show that the method proposed in this paper can effectively solve the three-phase voltage errors in the transformer, providing support for the metering equipment management to transform from “on-site operation and maintenance + periodic rotation” to “online operation and maintenance + precise rotation.”
Design of key data monitoring software for underwater targetXue, Ting; Yang, Litao; Deng, Leilei
doi: 10.1117/12.2680511pmid: N/A
Real-time monitoring of key data of underwater target in navigation status is an important basis for mastering and intervening the navigation status of underwater target in the whole process and ensuring the navigation safety of underwater target. With the help of CAN bus communication technology, this paper designs a key data monitoring software for underwater target based on CAN bus, which can monitor and display the key data of underwater target such as their attitude, depth, motor voltage, motor speed, driving plate temperature, etc. in real time. On the premise of ensuring real-time performance, the software makes full use of CAN bus communication features, and has the advantages of good human-computer interaction, simple operation and high visibility.
Multi-channel dictionary learning speech enhancement based on power spectrumNi, Tongzheng; Wei, Junfeng; Wu, Jiarong; Zhang, Lanfang; Tang, Weidong
doi: 10.1117/12.2680516pmid: N/A
Algorithms that model and estimate noise based on statistical properties, such as spectral subtraction, can estimate the distribution of stationary noise, but their performance degrades when suppressing non-stationary noise. Dictionary learning and sparse representation algorithms have made great achievements in solving non-stationary noise suppression. However, the multi-channel speech enhancement algorithm based on dictionary learning needs to manually estimate the parameters of spectrum reduction threshold in practice. In order to obtain optimized noise reduction results, the adaptive estimation of spectrum reduction threshold is of great significance. According to the power spectrum of the signal, the algorithm of spectral subtraction threshold is defined and the spectral subtraction threshold is used to optimize and enhance the quality of speech. The experimental comparison shows that the spectral reduction threshold calculated based on the power spectrum is closer to the optimal result compared with the fixed threshold. In the -10dB noise environment, the multichannel dictionary learning algorithm based on improved power spectrum improves the segmental signal-to-noise ratio by 1-2dB compared with spectral subtraction and non-negative matrix decomposition, and improves the perceived speech quality assessment and short-term intelligibility by an average of 2.3 and 0.11 points respectively. The experimental results show that the multi-channel dictionary learning algorithm based on the improved power spectrum can effectively remove additive noise under both unsteady and steady state noise conditions.
Design and implementation of UAS interoperability based on different protocolsYe, Zhen'gan; Pei, Hailong; Cheng, Zihuan
doi: 10.1117/12.2680502pmid: N/A
Different unmanned aerial vehicles teams often work independently, leaving a “stove-piped” isolation problem, and the cost of collaboration among them is relatively high. Interoperability has been proposed to study the ability of unmanned aerial systems in working together, of which smooth communication is the fundamental requirement. To construct communication ability of different systems, the protocols they support should be essentially concerned. A novel, lightweight and flexible protocol, named as LAB protocol, is developed in this paper, where we study it and STANAG 4586 and MAVLink for comparision. We propose a bridging method to break isolation and verify interoperability among UASs using different protocols through experiments.
Research on click fraud prediction based on multi-algorithm fusionDuan, Ganglong; Liu, Jianjun; Kong, Weiwei; Cui, Bowen; Li, Jiahao
doi: 10.1117/12.2680157pmid: N/A
The detection of click fraud in online advertisements on the Internet for the purpose of extracting advertising fees is one of the important aspects of machine learning applications. In this paper, using the data information of 400000 ad click cheating cases, we use recursive feature elimination method to determine the predictors and use five algorithms of gradient boosted decision tree (GBDT), random forest (RF), Adaboost, KNN and LGbmclassifier to train a single classifier, compare the prediction performance of each type of classifier, and the first three with better prediction performance The top three with better prediction performance were fused with multiple algorithms for prediction. The experimental results show that the random forest, Lgbmclassifier and Adaboost algorithms have the highest prediction accuracy, 87%, 83% and 79%, respectively, with AUC values of 0.90, 0.87 and 0.81. The prediction accuracy of the multi-algorithm fusion model taken in this paper can improve by 3% compared to the single algorithm with the best prediction performance, reaching 90%.
Deep reinforcement learning-based path planning for net sealing robotYang, Mazan; Huang, Zhiqing; Zhang, Yi; Ye, Kai; Li, Yunlong
doi: 10.1117/12.2680389pmid: N/A
For the path planning problem of overhead transmission net sealing robot in the process of sealing network, a visual perception and decision method based on deep reinforcement learning is proposed. By combining the perceptual capability of convolutional neural networks with the decision-making capability of reinforcement learning, the method achieves direct output control from the visual perception input of the environment to the action through end-to-end learning, forming a closed loop between the system environment perception and decision control directly, and obtaining the optimal decision strategy by maximizing the cumulative reward return of the robot's interaction with the dynamical environment. Simulation experimental results prove that the method can meet the requirements of multi-task intelligent perception and decision making, and better solve the problems of traditional algorithms such as easily falling into local optimum, oscillating in narrow passages and unreachable targets near obstacles, which greatly improve the real-time and adaptability of trajectory tracking and dynamic obstacle avoidance of the net sealing robot and ensure the safe operation of the net sealing robot in transmission line sealing operations.
Electroencephalography artifact removal based on an autoencoder deep networkLuo, You; Wang, Siyuan; Shen, Hui
doi: 10.1117/12.2680455pmid: N/A
The electroencephalography (EEG) signal acquisition process is inevitably affected by a variety of physiological noise signals, including electrooculogram (EOG), electromyography (EMG). The traditional methods of removing EOG and EMG rely heavily on the subjective experience and prior knowledge of the user. However, the ambiguity of artificial judgments can lead to erroneous and misleading interpretations that are insufficient for qualitative analysis. This inaccurate denoising may affect the true information of the signals in the time domain and spectral domain, leading to a decline in the accuracy of the BCI system. In recent years, a variety of EEG denoising methods based on deep learning have been proposed, but their denoising performance needs to be further improved. In this paper, we design a novel autoencoder (AE) neural network to remove artifacts in EEG. The network includes an encoder and a decoder module. The encoder contains five convolutional layers with increasing feature dimension as depth increases, which are responsible for detecting and suppressing artifacts. The decoder contains five deconvolution layers, whose feature dimension decreases gradually, and is used for EEG reconstruction after denoising. The experimental results on semi-synthetic EEG datasets demonstrate that the proposed algorithm outperforms the four benchmark models.
Research on multidimensional risk assessment system of renewable energy distribution network based on value at riskLiu, Yang; Li, Lisheng; Sun, Yong; Zhang, Shidong; Wang, Di; Ai, Qian
doi: 10.1117/12.2680395pmid: N/A
In order to reflect the impact of source-load uncertainty on the overall operation risk of the power system in the future time interval, this paper establishes a multi-dimensional risk assessment system for renewable energy distribution networks based on value-at-risk. Firstly, the output probability model and load probability model of wind power photovoltaic are established, and then the sampling combination and power flow calculation are carried out based on the Monte Carlo method, and the corresponding node voltage and branch power flow results are obtained. Calculate the two risk indicators of branch power limit and node voltage limit at each prediction time of the distribution network, and use them as the first dimension risk assessment index; secondly, based on the value-at-risk theory to quantitatively evaluate the system risk, calculate the second dimension risk assessment Index; Finally, calculate the third dimension risk assessment index, calculate the comprehensive risk index through the entropy weight method, and comprehensively and reasonably analyze the operation risk of the distribution network under the access of different amounts of renewable energy. Finally, the rationality of the proposed model and method is verified by IEEE33 example system.
Optimization of docker container live migration solutionWang, Kang; Xie, Zhenping
doi: 10.1117/12.2680231pmid: N/A
Nowadays, all cloud platforms support the live migration technology of Docker containers, which plays an important role in improving the stability and quality of services. In this article, we will use the JNLP algorithm to determine which migration method to use in pre-copy and post-copy based on the actual network bandwidth and real-time monitoring of memory changes. Moreover, this article analyzes the storage mechanism of docker containers, using the hierarchical structure of docker to reduce the data that needs to be transferred during migration, thereby reducing the total migration time. Experimental results show that the live migration technique proposed in this paper is more general and reduces container data migration time by 40% and total migration time by 30% compared with state-of-the-art techniques.
Quantitative analysis of chemical pollution such as ammonia nitrogen in the Chaohu Lake basinLiu, Jiepeng; Liu, Zhenhua; Lin, Chuanwen; Li, Guoyun
doi: 10.1117/12.2680019pmid: N/A
According to the national policy of proposing water pollution prevention and control regulations for the Chaohu Lake basin, to solve the current problem of chemical pollution in the Chaohu Lake basin, the relevant environmental data of the Chaohu Lake basin provided by Chengxin Environmental Testing Company were used, and a random forest algorithm based on high correlation filtering was proposed to construct a prediction model for the concentration of chemical pollution such as ammonia nitrogen in the Chaohu Lake basin. The method uses high correlation filtering to eliminate the strong correlation between water pollution data and adjusts the parameters of the random forest model to achieve the best prediction effect for the training data. The model was tested by real data sets, and the Bayesian ridge regression model, ordinary linear regression model, elastic network regression model, support vector machine model, and correlation vector machine model were used for comparison experiments, and the average relative error was used as the evaluation index. Finally, the random forest algorithm model with high correlation filtering achieved the best prediction results in the study of quantitative analysis of eutrophic chemical pollution such as ammonia nitrogen in the Chaohu Lake basin.