Efficient Load Scheduling Algorithm Using Artificial Neural Network in an Isolated Power System

Abstract

In this paper, an efficient load scheduling technique is presented to meet the unpredictable power supply requirements. The power consumption in upcoming days’ must be scheduled in a power system. The accuracy of the system significantly affects the economic operation and consistency of the system. The power generation system fails due to instability at the peak load time. Usually load shedding procedure is used to compensate demanded load. Unnecessary and extra loads are disconnected in load shedding. The proposed system overcomes this difficult by forecast the load based on the load affected constraints. To predict and schedule the load with the previous data is a challenging process when an unexpected change occurs - like days with extreme weather or special days. With the current advance of artificially intelligent tools, it is potentially possible to improve the existing demand of load. For optimal load scheduling, Artificial neural networks are used. The Levenberg-Marquardt backpropagation algorithm is used for the training purpose to minimize the error function. The results are compared by correlation analysis.