Research on Kalman Filter for One-dimensional Discrete Data

Abstract Kalman filter processes the input and observation signals with noise on the basis of linear state space representation to obtain the system state or real signal. In the one-dimensional model, due to the lack of multi-dimensional description of the target, the prior estimate of the state is often the measured value of the previous moment. When the target state changes, the divergence phenomenon will occur. Aiming at the problem that the one-dimensional traditional Kalman filter lacks the target observation dimension, which leads to the divergence or imprecision of the filter, this paper focuses on improving the estimation method of the target state, and proposes a real-time prediction model based on the cascade structure. This model can improve the response of Kalman filter to the change of target state and dynamically adjust the Kalman iterative domain to improve the measurement accuracy. The digital signal filtering simulation is carried out and the performance of the filter is verified based on LabVIEW. Experimental results show that the algorithm can maintain the accuracy and real-time performance of filtering when only one dimension observation results are obtained.