Research on dual-threshold detection based adaptive fault-tolerant kalman filtering algorithm for dynamic 6-DOF measurement

To improve the performance of dynamic six degrees of freedom (6-DOF) measurement and overcome the limitations in terms of dynamic inaccuracy, signal unavailability and error accumulation for large-scale metrology (LSM). A hybrid LT/PMTS/SINS 6-DOF measurement system is constructed and a dual -threshold adaptive fault -tolerant kalman filtering algorithm (DTAFTKF) composed of four modules (DFDM, D-RBFNN ECM, CFDM, AFM) working in a tightly coupled mode is proposed as the fault -tolerant multi -sensor fusion scheme. In the proposed scheme, a dual -threshold detection structure is formed by constructing an LSSVR-based fault detection algorithm combined with residual chi-square (RCS) test, and a dynamic-RBFNN (D-RBFNN) network is proposed to predict and compensate for the system pose error during fault durations when the fault detection value exceeds the high threshold. Moreover, a filtering system covariance adaptive adjustment algorithm based on covariance adjustment factor (CAF) is studied to improve the fault tolerance capacity and estimation accuracy when the fault detection value is at the middle threshold. Finally, experiments under different dynamic motions are carried out. The experimental results verified that based on the proposed method, the LT/PMTS/SINS system is capable of providing accurate and stable estimates of 6-DOF pose under various dynamic conditions with the maximum angular error below 0.047 and maximum positional error below 0.35 mm, which showed significant improvement compared to other comparative methods.