Design of support vector machine controller for hybrid power system automatic generation control

ABSTRACT

This paper describes an online adaptive tuning methodology for Automatic Generation Control (AGC) of grid-connected wind-diesel hybrid power systems using the Support Vector Machine (SVM) controller. Using the Simulink design optimization tool/MATLAB, an optimized PID controller is first designed. To acquire system uncertainty, the SVM controller is trained using the input-output data set of an optimized PID controller obtained by varying the system nominal parameters by 40%. Simulations are carried out for conventional power generating units of various control areas integrated with hybrid wind-diesel power system. Outcomes are presented in the form of frequency aberration response, deviation in tie-line power exchange, and variation in mechanical power, and the efficacy of the proposed controller is presented in view of settling time, undershoot and overshoot. The impact of the hybrid power system on the thermal generating system is noted, and the usefulness of the developed SVM controller in keeping frequency and tie-line power exchange deviations within the stipulated limits is explored. The proposed SVM controller diminishes the settling time of frequency deviation response from 19.24 seconds to 14.36 seconds, and the undershoot and overshoots are reduced from 4.42% to 2.83% and 0.85% to 0.28%, respectively, for a three area three machine power structure. Further, the stability of the power systems under study is studied using bode-plot analysis with the designed controllers. As a result, the results indicated the robustness of the proposed SVM controller in minimizing system frequency and tie-line power variations caused by generation and load mismatch, as well as enhancing system stability.

KEYWORDS:

• Support vector machine
• Automatic generation control
• Hybrid power system
• Bode-plot analysis

Disclosure statement

No potential conflict of interest was reported by the author(s).