monte carlo parameter estimation

The key objective of modern power generation, being dynamic and multifarious in nature, is to maintain power exchanges and system frequency to their contractual values to meet growing energy needs. This can be achieved by Load-Frequency Regulation using adaptive controllers. The present study illustrates an innovative approach for adaptive tuning of Support Vector Machine (SVM), a supervised machine learning algorithm, which can be used as a controller for Load Frequency Control (LFC) problem of electric grid to regulate the frequency and tie-line power flows in an interconnected power system. Primarily, an optimized Proportional Integral Derivative (PID) controller is designed for a two interconnected non-reheat thermal areas in which Monte Carlo parameter estimation method is used for sampling the values of uncertain parameters randomly. The input-output data set of optimized-PID controller is used to design a PID based Support Vector Machine (SVMPID) controller. The simulation studies are conducted to find the deviations in frequency and tie-line power exchanges resulting from a step load perturbation in each area. The comparative results are presented with conventional controller, optimized-PID controller and SVMPID controller. The efficacy of the trained SVMPID controller is tested on a three area interconnected thermal-thermal-hydropower system by considering generation rate constraint (GRC), dead band (DB) and boiler dynamics (BD) to represent real-time situation. The results show the performance of the proposed SVMPID controller and its capability to ensure zero steady state error by sustaining minimum over/undershoot and settling time in the system dynamic responses under multi-operating conditions.