Research On Noise Reducing Of Dynamic Testing Signals And Dynamic Compensation For Testing System

Sensor is the first link in the measuring control system, and its accuracy play a decisiverole in the measuring control system, thereby correcting and compensating sensors’ staticcharacteristics and dynamic characteristics has already become an important subject both athome and abroad. In order to improve sensor’s dynamic characteristics，in this paper，therelevant problem such as dynamic testing signal denoising, sensor nonlinear fitting,identification of dynamic compensation filter etc. Each content is achieved in differentmethods, and its characteristics and applicability are analyzed.In this paper, the used of the traditional digital filter in signal processing of dynamicmeasurement is briefly introduces, and the shortcomings of the method is analysis. on thisbasis, the wavelet threshold denoising method is proposed. The choice of wavelet basis, thefix of decomposition level of wavelet, and the determination of the threshold value in theWavelet noise suppressor are discussed. Further more，a new denoising method base on sparsedecomposition is proposed. Standard matching pursuit algorithm is time-consuming, in thispaper an adaptive genetic algorithm is introduced to matching pursuit method, which makecomputation speed improved effectivey. Experimental results indicate that this method caneffectively remove the high frequency noise.This dissertation proposes a new type of nonlinear correction method base on BP neuralnetwork after briefly introducing the traditional least square principle methods. A detaileddescription and some application problems such as the choice of hidden nodes and parametersare analyzed in the process of modeling. This method under the conditions of that less in thesample data and uneven distribution is still have good fitting effect.Finally, discusses the method of dynamic compensation, especially expound theidentification for the order and parameter of the dynamic compensation filter. Comparison andanalysis the least squares method and the particle swarm optimization algorithm, dynamic compensation used those methods are carried on a PCB pressure sensor and a thermocouple,which show that those methods can improve dynamic performance effectively.