Filtering method, application, device and storage medium based on signal sparsity classification

Abstract

The application relates to a filter method, application, device and storage medium based on signal sparsity classification, which comprises the following steps: collecting vehicle original signals; measuring the sparsity of the vehicle original signal data and classifying the signals; converting the filter problem into an optimization problem by adopting a full-difference TVD method for the sparse type signal, solving the optimization problem to obtain the filtered signal; comprehensively considering the sparse component, low-pass component and noise in the mixed type signal to construct an optimization problem, solving the optimization problem to obtain the sparse component in the signal; obtaining the low-pass component in the signal based on a low-pass filter; and outputting the filtered signal. Compared with the prior art, the application realizes better signal differentiation classification and differentiated filtering of the classified signals by evaluating and measuring the signal sparsity; the filtered signal retains signal peak information and is smoother, the performance of a vehicle state estimator is improved, and more accurate and reliable vehicle state estimation is realized.

Images