Diesel Generator Model Development and Validation using Moving Horizon Estimation

Diesel hybrid power systems including inverter-based generation have faster and more stochastic dynamics than traditional systems. It is necessary to develop accurate models of the system components to ensure the stability of these systems and proper controller design. The parameters of the diesel generators in hybrid power systems, such as the inertia constant, are time-varying, requiring online parameter estimation techniques. This paper presents a simplified linear model developed to represent the frequency dynamics of the detailed diesel generator system and validated the model using a moving horizon estimation (MHE) approach. The proposed optimization-based MHE algorithm is employed to accurately provide an estimation of multiple parameters of a simplified diesel generator model. The proposed method extracts the parameters minimizing a cost function with a given set of constraints on the parameters. A non-intrusive square wave excitation signal generated by step changes in load is used to perturb the system with minimal impacts on power system operation. MHE estimates the parameters based on the power and frequency from the diesel generator system measured using the phase-locked loop (PLL) and provides reasonable estimates of unknown parameters. The estimated parameters are further verified by using them back in the simplified model and comparing them with the PLL measurements to represent the frequency dynamics of the diesel genset system.