Online estimation of system parameters for artificial intelligence applications to load frequency control

Making balance between generation and demand is the operating principle of the load frequency control. As the automatic generation control (AGC) has been commissioned to serve load-generation balance for several decades, the developmental roadmap of the AGC is leading to the adaptive or artificial intelligence (AI) applications to the controller design. Among the AI applications, the genetic algorithm (GA) and the fuzzy inference system (FIS) are often adopted to optimise AGC gains under the time-varying system conditions. Relevant studies indicate that the real-time acquisition of system states would be advantageous for AGC to drive system to the optimal condition. However, some of the system parameters are not easily accessible. The object of getting good estimate of system parameters in real-time basis would definitely enhance the performance of the intelligence applications to AGC. This study explores the use of recursive least square algorithm for acquiring the system parameters in real time. As such, a more accurate reference model would be gainfully used in the AI controller design. Issues related to the model validation in the simulation and field test are presented. Following that, the use of the reference model in the GA as well as the FIS applications for the optimal gain scheduling of the AGC is demonstrated.