变循环发动机自适应无迹卡尔曼滤波器设计

Adaptive Unscented Kalman Filter Design for Variable Cycle Engine

关键词:  变循环发动机  参数估计  卡尔曼滤波器  自适应无迹卡尔曼滤波器  概率密度函数

Key words:  Variable cycle engine  Parameter estimation  Kalman filter  Adaptive unscented Kalman Filter  Probability density function

摘要:

针对变循环发动机健康参数估计问题，设计了一种自适应无迹卡尔曼滤波器。该算法通过最大化后验密度函数来建立过程噪声协方差和测量噪声协方差的自适应更新方程，以改善传统无迹卡尔曼滤波器设计中先验参数需要根据经验来设置，进而导致滤波器性能受人为因素影响较大的问题。以带核心机驱动风扇级的变循环发动机为对象，进行了不可测参数估计仿真试验，对所设计的自适应无迹卡尔曼滤波器算法进行了仿真对比验证。结果表明：在单参数退化条件下，健康参数平均估计误差不大于2%；多参数退化条件下，健康参数平均估计误差不大于1.8%；该算法性能优于增广卡尔曼滤波器、传统无迹卡尔曼滤波器，相较于传统无迹卡尔曼滤波器性能提升9.5%。

Abstract:

An adaptive unscented Kalman filter is designed for variable cycle engine health parameter estimation. The algorithm establishes adaptive update equations for process noise covariance and measurement noise covariance by maximizing the posteriori density function. Unlike the traditional unscented Kalman filter design， where prior parameters need to be set according to experience， the designed adaptive unscented Kalman filter can reduce the impact of human factors on the filter performance. A simulation test of health parameter estimation was conducted for a variable cycle engine with CDFS， and the designed adaptive unscented Kalman filter algorithm was verified by simulation comparison. The results show that the average estimation error of health parameter was no more than 2% under single-parameter degradation condition， and no more than 1.8% under multi-parameter degradation condition. The performance of this algorithm is better than that of the augmented Kalman filter and the traditional odorless Kalman filter， and the performance is improved by 9.5% compared to the traditional unscented Kalman filter.