Drift error compensation for laser displacement measurement system based on Bayesian lag regression model and new information integration

Drift exhibits time-varying, cumulative, and stochastic characteristics, which degrade the accuracy of high-precision measuring instruments, especially in long-term measurements. This paper presents a drift error compensation method for laser displacement measurement systems by integrating a Bayesian lag regression model with new information integration. The primary factors influencing the drift error were analyzed, and characteristic analyses were conducted on the drift error data and key influencing factor data collected during non-measurement intervals. A Bayesian lag regression model was subsequently established. Model parameters were estimated using a recursive Bayesian tracking and prediction algorithm initialized with the non-informative prior distributions. The drift error values predicted by the model have been used to compensate for drift errors during operational periods, while outputs from idle periods have been utilized to update the model parameters. Experimental validation has been performed on a homemade laser Doppler displacement measurement system, demonstrating an average reduction of 75% in variation and 92% in root mean square error (RMSE), across six repeated experiments. Furthermore, the proposed method achieves average reductions in RMSE of 72% and 71% compared to conventional multiple regression and support vector machine (SVM) methods, respectively. These results confirm the effectiveness and robustness of the proposed compensation methodology.