The Algorithm Research On Traffic Flow Forecasting Based On Artificial Neural Network

The Intelligent Transportation System(ITS) is an economic,environmentttally-f- riendly,low energy consumption and high efficiency method in reducing city traffic pressure and traffic pollution at present. Traffic flow forecasting , an important part of ITS ,is used to predict a future time's traffic information base real-time traffic information.And ITS is mainly make use of the predication to control traffic signal light real-time and guide road traffic effiectively.So achieve better accuracy traffic information becoming a hot spots to expert scholars.The city cross-roads traffic flow is highly in nolinear,relativity and mutability.and traditional predict methods based on mathematical models are hard to predict the traffic flow of crossroads.Artificial neural networks is an effective and precise traffic flow foreacasting method at present.We make a city's crossroads as the subjects In this paper.And we gather a set of traffic flow data each 5 minutes.Here we make use of the chaos system with overall situation ergodic characteristic,and combine neural networks,wavelet functions and particle swarm optimization algorithm,advance a crossroads lowtime traffic flow indiceasting algorithm base on chaos and wavelet neural network to advance the precision of neural network algorithm.our research work is mainly containing four parts:1.short-term traffic flow forecasting based on BP neural networkHere the BP neural network is applied to predict the crossroads traffic flow.after. After analysing the limitations of BP network algorithm,we make use of genetic algorithm and particle swarm optimiza- tion to optimizing the dispatcherservlet of BP network.And than analyse the merits and faults of each improved algorithm.2. short-term traffic flow forecasting based on wavelet neural networkCombine the characteristic of the traffic flow with nolinear,mutability,small finite sets and better processing capacity of wavelet neural network to high-frequency and low-frequency signal.advance an traffic flow algorithm based on wavelet neural network.and make use of MATLAB 2009 to simulating the algorithm.and then analyse the performance of the new algorithm.3. short-term traffic flow forecasting based on PSO wavelet neural networkAnalyse the merits and faults of wavelet network algorithm. make use of particle swarm optimization algorithm to optimizing the dispatcherservlet of wavelet As to conquer the limitations of wavelet neural network.and then analyse the merits and faults of the improved algorithm.At last we make use of the algorithm to forecast the crossroads traffic flow.4. short-term traffic flow forecasting based on chaos algorithm and wavelet networkIn this paper we embed the chaos system into PSO algorithm and wavelet network's training by make use of the overall situation ergodic characteristic of chaos system .and then make use of the two improved algorithms into the traffic flow forecasting.In this paper we use MATLAB 2009 to simulate each predict algorithm.and the experimental shows that the predict algorithm based on chaos and PSO wavelet network can pop out the nolinear,mutability characteristics of traffic flow,and the algorithm has done a good job on globe convergence and predict precision,it is an effective predict algorithm.