On-line process fault diagnosis using neural network techniquesZhang, J.; Roberts, P.D.
doi: 10.1177/014233129201400402pmid: N/A
A neural-network based on-line fault-diagnosis system for industrial processes is presented in this paper. A multi-layer feed-forward neural network is developed and trained with symptom-fault pairs from experience of the operation of a process or from simulation analysis of that process. The trained network can be used to diagnose faults in that it can associate the abnormalities in on-line measurements with corresponding faults. Compared with diagnosis systems based on expert-systems techniques, which have several limitations such as the time consuming nature of developing the knowledge base and the inability to cope with situations not presented in the knowledge base, the neural-network based fault-diagnosis system is easy to develop and performs robustly. The feasibility of applying such a diagnosis system to industrial processes is demonstrated by applying it to a pilot-scale mixing process and in a simulation study of a continuously-stirred tank reactor (CSTR) system. A series of experiments is carried out to investigate the performance of the neutral-network based on-line diagnosis system and it is shown that it can perform satisfactorily with partially incorrect and partially unavailable information. Therefore, to some extent, the system can tolerate measurement noise and model-plant mismatch.
Improved transient fault simulation on a laboratory model of a turbine-generatorNouri, H.; Colfer, G.W.
doi: 10.1177/014233129201400403pmid: N/A
In order to enable a simulated fault current in the region of 4000 A to produce a pulsating magnetisation in the stator coil, an arrangement of a parallel tuned-circuit was constructed. The performance characteristics of the tuned-circuit were assessed from an analytical solution (Transient response method) and modelling on an Apollo computer using the transient response package.The inrush current which is associated with the starting of the tuned-circuit was controlled with a soft-start, which increased the input voltage across the coil from zero to full preset value in 50 cycles. The soft-start circuit consisted of two anti-parallel thyristors, current-limiting resistor, phase-angle trigger unit and its control circuitry.Spurious triggering of the thyristor was greatly reduced by placing the current-limiting resistor in between the supply and the thyristor. The possible causes are discussed.A detailed comparison of the measured flux for the fault current of 4000A at an air-gap of 75 mm, for conditions with and without the endbell, at the end of the rotor suggests that the conduction current is dependent on the contact resistance at the endbell and rotor interface.
A high-speed image acquisition system for roboticsSmith, J.S.; Tan, C.M.; Lucas, J.
doi: 10.1177/014233129201400404pmid: N/A
There are many applications in industry for imaging systems. The majority of these use a video camera as the input device. The output from the camera is sampled and stored in memory and a computer used to analyse the image. Until recently, the time for the analysis of the image was relatively long in comparison with the acquisition time of the image. However, with the development of new high-speed microprocessors decreasing the processing time, the acquisition time of the image will become the limiting factor of the analysis rate. This paper describes the development of a prototype system to make possible the capture of a full frame image (256 lines), in under 15 ms, for analysis by a computer. With the use of programmable logic devices (PLDS) and hardware windowing techniques, the acquisition time for a part frame (40 lines) is 2.8 ms. Shorter acquisition times can be obtained, but with reduned resolution. An application of the imaging system is also presented.
Robust and adaptive control of a multi-variable systemWu, Wen-Teng; Tseng, Ching-Guey
doi: 10.1177/014233129201400405pmid: N/A
An on-line robust control and robust adaptive control of a multi-variable system has been developed to solve the problem of plant/model mismatch. An on-line stability index is proposed for stability indication of the control system. Although very small mismatch can be handled by conventional feedback control, moderate and large mismatch must be corrected by on-line robust and robust adaptive control, respectively. Regarding the former, the control algorithm is based on tuning the parameters of designed filters, and the tuning method depends on the robust stability index which is on-line determined. Regarding robust adaptive control, system identification is carried out, and the identified model replaces the original one. Control of a continuous stirred tank reactor (CSTR) is presented to illustrate the proposed method.
A generalised algorithm for polarity correlation estimation using binary patternsHarba, M.I.A.
doi: 10.1177/014233129201400406pmid: N/A
A recently developed binary-pattern correlator uses the binary patterns of a group of successive samples of the polarity-sampled signal for correlation estimation. At high sampling rates (compared to the bandwidth of the signal), only a limited number of polarity changes may take place within a finite signal duration. Thus, a finite interval of a polarity-sampled signal may only have a limited set of binary patterns. The correlator estimates the frequency of occurrence of these patterns on-line and stores them in tables. The actual correlation coefficients are then deduced from these tables. Because the number of patterns is limited, efficient operation results.A general binary-pattern correlation algorithm is presented, with examples implemented on a Z-80 microprocessor. These examples represent different implementations of the algorithm. They differ in the number of signal samples effectively input each time, and in the size of the frequency of pattern-occurrence tables. As a result they have different sampling rates, and different off-line processing intervals. However, there is a trade-off between the achieved sampling rate and the amount of off-line processing.