Hidden Markov model-based smith predictor for the mitigation of the impact of communication delays in wide-area power systems

Abstract

The use of an open communication network in a wide-area power system (WAPS) introduces random delays into the transmission of frequency measurements and control signals, which can deteriorate the load frequency control (LFC) performance. Current studies are focusing on developing a suitable delay margin controller to maintain the stability of the WAPS. This paper introduces a new Smith predictor (SP) with tools to accurately predict the input time delays and consequently mitigate the LFC performance loss caused by unreliable communication. The time delays are predicted via the discrete hidden Markov model (DHMM) and the exponentially weighted moving average (EWMA) model. The predicted delays are then input to the SP. The DHMM adopts two different scalar quantization methods, i.e., the uniform technique and the K-means clustering technique. It then maps the time delays on to a discrete observational space. To validate the findings for practical applications, we conduct a case study on a test platform, namely a single-area WAPS with Ethernet, which is implemented via the Truetime simulator. The results indicate that the SP is more effective in eliminating load disturbances and enhancing the robustness against time delays than existing delay-margin-based controllers. The K-means DHMM-based SP achieved better LFC performance than the one with the EWMA. The stability of the LFC system with time-varying delay is discussed using a Lyapunov–Krasovskii-based delay-dependent criterion and the small gain theorem.