Vehicle-to-Grid Frequency Regulation Signal Optimization Based on Inhomogeneous Hidden Markov Model

Abstract

As one of the potential frequency regulation (FR) service providers for the independent system operator (ISO) or regional transmission organization (RTO), electric vehicles (EVs) can response the FR control signals generated by ISO or RTO by changing their real-time charging or discharging power. The target is to keep the area control error (ACE) at a low level while minimizing the cost. Recently, many of the ISOs and RTOs such as CAISO and PJM have offered performance-based pricing schemes for FR service providers. Yet, how to estimate the total FR capacity of EV owners managed by the EV aggregators (AGGs) and optimize the FR signals accordingly still need extensive research. In this paper, we proposed an optimal strategy for ISO (or RTO) to allocate differentiated FR signals to EV AGGs according to their actual FR capacity. Thereby, an inhomogeneous (or time-variant) hidden Markov model (HMM) is developed to estimate the FR capacity through observing the historical FR responses. In this paper, traditional BaumWelch algorithm is extended to be applied in the inhomogeneous scenario through decoupling the transition matrix. In this way, the computational complexity can be significantly reduced. The performance of the proposed algorithm is evaluated through extensive simulations based on the real FR signals from PJM.