The Research Of Vehicle Dynamic Weighing System Based On Intelligent Algorithm

Overloaded vehicles cause serious damage to traffic and break the order of transport market. Technology of vehicle dynamic weighing is the most efficient way to solve this problem. The intelligent algorithms have advantages of better fault tolerant and strong robust in dynamic weighing signal processing. They can solve the problems of heavy noise and non-linear efficiently. So, the study of vehicle dynamic weighing system based on intelligent algorithm is meaningful in application.Vehicle dynamic weighing system consists of two parts: hardware and software. This article focuses on the software of signal processing segment. The main contents and contributions can be summed up as follows:1) By analyzing the components of dynamic weighing signal, the pretreatment of signals is divided into two steps. The first step is using wavelet transform to filter the. high frequency noise, then reconstruction of wavelet is performed to the de-noised signal. The second step is using genetic algorithm to fit the dynamic load which mixed up with weighing signal in low frequency bandwidth, and filter the dynamic load. According to the features of the dynamic weighing signal, the multi-sections fit method using genetic algorithm is presented to improve the precision of the dynamic load curve fitting.2) Research on modeling of dynamic weighing system. Firstly, The ARX model of dynamic weighing system is introduced and simulations are performed. Then, The ARX model is aimed at the particular weighing platform, So Prony algorithm, which is a tool of system identification widely used in the electronic system analysis, is introduced in this article, the algorithm is developed to solve the problem of dynamic weighing. The precision of simulation results satisfies the demand of weighing system. Finally, the results of two algorithms are compared and analyzed.3) The method of multi-sections model is presented because the measure result of small weight vehicle in high speed is higher than the actual weight of vehicle. Different models are built according to high-speed section, middle-speed section and low-speed section. Because of the affection of stochastic resonance which can not be eliminated by common de-noise approaches, the neural network is introduced to modeling the high-speed section. The middle speed section and low speed section are using ARX model with some simple modifications. Applying fuzzy theory, the weighing results in two crossover segments are modified.