基于IQPSO-EKF的多传感器融合姿态测量方法研究

Attitude measurement of multi-sensor fusion based on IQPSO-EKF

为解决自动化竖井掘进设备的定位调姿精度对竖井、孔桩挖掘效率与质量的影响,提出了一种基于改进量子粒子群(IQPSO)-扩展卡尔曼滤波(EKF)的姿态测量算法,以提高微机电系统(MEMS)传感器测量精度。首先,对MEMS传感器数据进行了预处理(除噪、滤波、校准等);然后,参考现有飞行器的坐标系,建立了姿态解算模型,通过姿态角数学模型及运动学分析,构建了EFK状态方程,针对EKF方法参数估计不准确的问题,以分段混沌映射优化初始种群,引入平均位置最优值来避免陷入局部最优的IQPSO-EFK算法,优化EKF的系统、测量噪声的协方差参数;最后,对改进算法和三组姿态误差估计进行了对比实验。研究结果表明:对比三种典型目标函数,IQPSO-EFK相较于普通粒子群算法(QPSO-EFK)具有更强的寻优能力与收敛精度;对比三组旋转速度姿态测量误差,基于IQPSO-EKF算法的姿态测量方法在测量误差时比真实测量误差减少了约86.3%,比扩展卡尔曼滤波减少了约68.7%,比普通粒子群算法减少了约28.2%,证明该算法有效地提高了MEMS传感器测量精度。

To solve the influence of the positioning and attitude adjustment accuracy of automatic shaft boring equipment on the excavation efficiency and quality of shaft and hole pile,an attitude measurement algorithm based on improved quantum particle swarm optimization(IQPSO)-extended Kalman filter(EKF)was proposed to improve the measurement accuracy of micro-electro-mechanical system(MEMS)sensors.First,the MEMS sensor data was preprocessed(noise removal,filtering,calibration,etc.).Then,the attitude solving model was established with reference to the coordinate system of the existing aircraft,the EFK state equation was constructed through the attitude angle mathematical model and kinematic analysis,and the initial population was optimized by piecewise chaotic mapping for the problem of inaccurate estimation of EKF method parameters.The average position optimal value was introduced to avoid falling into the local optimal IQPSO-EFK algorithm,and the EKF system was optimized to measure the covariance parameters of noise.Finally,the comparative experiments were conducted on the improved algorithm and the three sets of attitude error estimation.The research results show that comparing with the three typical objective functions,IQPSO-EFK has stronger optimization ability and convergence accuracy than ordinary particle swarm operation(QPSO-EFK).Comparing with the three sets of rotational velocity attitude measurement errors,the attitude measurement method based on IQPSO-ECKF algorithm reduces the measurement error by about 86.3%compared with the real measurement error,about 68.7%compared to the extended Kalman filter,and about 28.2%compared to the ordinary particle swarm algorithm,which proves that the algorithm effectively improves the measurement accuracy of MEMS sensors.