针对传统组合导航仿真算法在MEMS惯性器件和工业级处理器搭建的硬件平台上实现时,面临处理器算力不够和有效位数少进而导致的解算实时性及精度较差的问题,通过量化MEMS惯性器件的噪声与误差项的量级,简化卡尔曼滤波误差方程,并采取UD分解算法,优化后的算法运算速度提升了17%,充分证明优化算法的有效性,并最终实现由MEMS器件组成的低成本姿态测量系统,经过跑车试验验证,航向角精度达到0.5°,水平姿态角精度达到0.2°,解决了导航数据输出实时性和解算精度较差的问题。

Aiming at the two problems of poor real-time performance and calculation accuracy caused by insufficient processor computing power and limited effective bits when implementing traditional integrated navigation simulation algorithms on hardware platforms built with MEMS inertial devices and industrial-grade processors, this study quantifies the noise and error magnitudes of MEMS inertial devices, simplifies Kalman filter error equations, and adopts UD decomposition algorithm. The optimized algorithm achieves 17% improvement in computational speed, effectively demonstrating its validity. Ultimately it is implemented on a low-cost attitude measurement system composed of MEMS devices. Vehicle tests verify that the heading angle accuracy reaches 0.5° and horizontal attitude angle accuracy reaches 0.2°, successfully resolving the issues of real-time navigation data output and computational precision.