Development of neural network controllers for power industry applications: Volume 2, Final report
- Stanford Univ., CA (United States). Information Systems Lab.
This report details, in two volumes, an investigation in the development of the theory of the design of neural network controllers, and their application to power system problems. In Volume I current algorithms for the supervised training of neural networks are reviewed. The strengths of neural networks are brought out in this work, demonstrating that neural networks can solve highly nonlinear problems for which no analytical solutions exist. Demonstrating the ability of a neural network to estimate complex processes, neural networks are trained to predict void fractions Demonstrating the ability of a neural network to control dynamic systems, neural networks are trained to solve excitation control problems, regulating load-frequency and terminal voltage. Volume I contains the results of the applications of neural networks to the above described problems. This report, Volume II, contains two Ph.D. theses which present algorithms and discuss the science of neural networks. in the areas of control, nonlinear filtering, system identification and prediction. In the first thesis, algorithms are presented for training neural networks for terminal control problems which minimize the elapsed trajectory time, while achieving obstacle avoidance. In the second thesis, a dynamic neural network is proposed, and is applied to various examples of time series prediction. (It is interesting to note that this network won the time-series prediction contest sponsored by the Santa Fe Research Institute in 1991). The application of these developments to power systems control problems with constraints (such as power line constraints) and estimation problems. such as load forecasting, should lead to promising results and is a subject for future work.
- Research Organization:
- Electric Power Research Inst., Palo Alto, CA (United States); Stanford Univ., CA (United States). Information Systems Lab.
- Sponsoring Organization:
- Electric Power Research Inst., Palo Alto, CA (United States)
- OSTI ID:
- 200674
- Report Number(s):
- EPRI-TR--105533-V2
- Country of Publication:
- United States
- Language:
- English
Similar Records
Constraint methods for neural networks and computer graphics
An introduction to neural networks: A tutorial