带约束卡尔曼滤波对涡扇发动机状态估计

提出了一种加入线性不等式约束的卡尔曼滤波方法,并用于涡扇发动机的健康状况估计。涡扇发动机数字模型包含10个状态变量、12个量测量、6个控制输入量以及8个健康状况参数。不等式约束不仅保证了状态变量估计在用户自定义的范围内随时间变化平稳缓慢,而且还提高了滤波计算效率,改善了滤波估计精度。同时系统还允许滤波器沿确定的方向修正状态变量估计,以保持状态变量真值恒定。对比传统的无约束卡尔曼滤波,线性化滤波结果显示,该方法对涡扇发动机的健康状况估计尤其行之有效。

Kalman Filter with Inequality Constraints for Turbofan Engine Health Estimation

A method for incorporating linear state inequality constraints in a Kalman filter is proposed and applied to turbofan engine health estimation. The digital turbofan engine model contains 10 state variables, 12 measurements, 6 control inputs and 8 component health parameters.Inequality constraints not only maintain the state variable estimates vary slowly with time in a user-defined limit, but also increase the computational effort of the filter and improve its estimation accuracy. And the system allows the filter to correct state variable estimates in a direction that the true state variables might never change. The linearized filtering results demonstrate the filter with inequality constraints offered a great improvement over the traditional unconstrained filter, particularly for turbofan engine health estimation.