Propulsion performance diagnostics and optimization based on the estimation of measurement biases and hybrid optimization

Abstract

This article deals with performance diagnostics and optimization of aircraft propulsion systems. To eliminate the adverse effects of unknown measurement biases on performance diagnostics, a new design approach based on measurement biases estimation is proposed. In this approach, measurement biases and the component deviation parameters (CDPs) are treated as a part of the augmented state vector, and multiple Kalman filters are designed based on different biases hypotheses to estimate the augmented states. Through the estimated measurement biases, the effects of actual biases are eliminated and accurate performance diagnostics is achieved. Moreover, considering the synthesis influence of linearization errors and initial values on the conventional linear programming (LP) method, a hybrid optimization algorithm based on LP and sequential quadratic programming (SQP) is investigated to improve the optimization accuracy and convergence speed. This algorithm uses a parallel subregion LP method to obtain the global feasible initial value, and then utilizes the SQP algorithm to search for the optimal solution. Simulations are carried out at cruise-operating conditions. The results demonstrate that the CDPs are estimated accurately, and the propulsion performance and online optimization ability are also improved.