A connectionist approach to real-time control and simulation of switching operations in power systems substations

This paper applies an artificial neural network (ANN) model to the online control and simulation of switching operations in power system substations. Both problems are considered in relation to a complex substation configuration that uses a switching arrangement with double busbar selection and one transfer busbar. The first section presents the design of an operator training simulator based on the ANN approach. For simplicity, the principles of the ANN model are described using a simple test-case. Then, the more complex substation configuration is considered in order to design two ANNs to simulate operations with different apparatus in the substation using simple multilayer perceptron architectures. For the online control of switching operations, the authors propose a generalized circuit that can be easily particularized to obtain any real circuit in the substation. Eleven ANNs were trained separately for the operation with each apparatus in the generalized circuit. These ANNs were joined to form a unique ANN able to control all operations in the substation. Both ANN models were implemented as software applications by Turbo-Pascal into PC-AT compatible systems. The advantage of the proposed connectionist approach is that it is reliable and faster than other traditional methods.