Increase of PLC computability with neural network for recovery of faults in electrical distribution substation

Abstract

Due to the increasing in technological development and modernization of industrial process, control techniques to address high performance are being developed. These not only solve new problems in more complicated plants, but also improve the performance of existing controllers. To improve the control performance of electrical distribution substations, embedded optimization techniques have been developed utilizing the programmable logic controllers (PLC). In this work an industrial artificial neural controller, that is the union between industrial controller and neural network, in associated with an intelligent electronic device (IED) for data acquisition is developed. The PLC execute operational tasks. The neural network controller performs data processing in a MATLAB environment that communicates with the PLC, to receive and to send data, via OPC protocol. The proposed methodology is evaluated in virtual electrical power substations, where the automation devices are: 1)Smart Relays, Remote Transmission Units (RTU) and PLC are real devices and 2) transformers, circuit breakers and capacitor banks simulated with software or hardware.