论文部分内容阅读
针对集中式滤波算法存在计算效率不高、容错性差,引入融合滤波的思想,提出采用非线性融合的联邦式扩展卡尔曼滤波器进行发动机气路健康性能预测。子滤波器根据量测参数完成发动机部分健康性能的局部估计,主滤波根据子滤波器估计参数完成融合滤波估计,并将状态估计值和协方差反馈至子滤波器用于下一步健康预测。通过某型涡扇发动机仿真表明:融合EKF滤波器能准确地预测发动机的健康状态,估计稳定收敛时间短、计算时间短、效率高。
In view of the low computational efficiency and poor fault tolerance of the centralized filtering algorithm, the idea of fusion filtering is introduced. A federated extended Kalman filter with nonlinear fusion is proposed to predict the performance of the engine’s air circuit. The sub-filter performs a partial estimation of the health performance of the engine based on the measurement parameters. The main filter completes the fusion filter estimation based on the sub-filter estimation parameters, and feeds back the state estimation and covariance to the sub-filter for the next health prediction. The simulation of a turbofan engine shows that the fusion EKF filter can accurately predict the state of the engine, estimate the stable convergence time, short calculation time and high efficiency.