Road Traffic Accidents Prediction Based On Optimal Weighted Combined Methed

Road traffic accident is a person, vehicle, road based uncertainty in the process, there are many factors which influence traffic safety, action mechanism, therefore the road traffic accidents have largely sudden and accidental.Scientific and reasonable to predict the future trend of road traffic accident is the first condition of related departments to take timely measures in advance.Traffic accidents according to the traditional regression forecast method, time series method, the BP neural network to forecast, because each model has certain limitation, so the prediction results tend to be larger deviation.Grey forecasting method don’t need a lot of sample size, easy operation, suitable for "poor information", stochastic volatility, the smaller the accident data.When using a single grey prediction model for volatile or medium accident forecast, there will be a fitting degree and low reliability prediction problem.Markov chain prediction method is through the judgment variable between the state and the state transition probability, thus draws the model predicted value, suitable for stochastic volatility is larger, medium and long term variable of accident.Therefore, according to the grey prediction model to master the law of development of accident sequences, by markov chain to grasp the fluctuation characteristics of accident sequences, grey markov chain prediction model is set up, you can get more accurate forecasts.In order to further improve the precision of the model, the optimal weighted combination forecast method, using the grey prediction model from the valuable information, has carried on the forecast to the future trend of accidents in urumqi.This research use urumqi city traffic accident casualties from 2007 to 2013, the first to establish the grey GM(1, 1) model and grey verhulst model, and calculate the average relative error of 4.32% and 4.57% respectively;Secondly by using markov chain model of grey GM(1, 1) model and the model of verhulst preliminarily error correction, found that the error reduced to 1.67% and 1.71%, respectively;The use of optimal weighted combination forecast method, the gray GM(1, 1) model and markov chain combination after verhulst grey markov chain model, the average relative error is further reduced to 1.33%, and it is concluded that the urumqi from 2014 to 2019 the number of traffic accident casualty forecast.This study through the grasp of urumqi accident development trend of the future, to make the traffic laws, and the affecting factors to provide theoretical basis for effective control of traffic accident.Level of road traffic safety, and minimize the future traffic accident economic loss is immeasurable realistic significance.