Advanced wide-area monitoring system design for electrical power system

A monitoring system is an essential part of power system operation and planning. This system model is used to observe the system stability behaviour, to detect the unforeseen power system instabilities and to prevent the system from being collapsed. Since the power system scheme is a complex and non-linear system, employing the conventional method based on a mathematical model for designing a monitoring system is quite complicated. This paper proposes an advanced wide-area monitor design using a machine learning method called a multi-output least-square support vector machine to overcome the problem above. The application of the kernel approach in a multi-output least-square support vector machine provides higher accuracy in predicting result and flexibility. The kernel technique will transform the input data into a higher dimensional feature subspace as an input data of a multi-output least-square support vector machine. Then this non-linear pattern will be transformed into a linear pattern and intensive computing time can also be reduced. The major problem of multi-output least-square support vector machine parameters is on finding the best parameter. If the trial and error method is used to find the parameter of multi-output least-square support vector machine parameters, the machine is ineffective in obtaining good performance for a wide range of operating conditions and various load change scenarios in a multi-area power system. A quantum-inspired evolutionary algorithm is employed to achieve the machine’s appropriate parameters automatically and accurate output by minimizing the objective function. In order to measure the rigorousness of multi-output least-square support vector machine output, mean square error is used as an objective function in this study. The simulation results show that the amalgamation of the multi-output least-square support vector machine and quantum-inspired evolutionary algorithm provides a satisfactory result when compared to standard multi-output least-square support vector machine output. In addition, this proposed method can be considered as a promising technique in the future as a wide-area monitor model.