Armored Forces Fuel Consumption Prediction Technology Research Based On The Grey Neural Network

With the development of new military revolution which uses information technology asits core, the accuracy and efficiency of fuel supply put forward newer and higher requirementsto our army’s logistics mode. Accurate prediction of fuel consumption directly affects thelogistics capability of armored forces, but the traditional prediction model of fuelconsumption is not accurate enough, and there are some limitations in the range ofapplications, so it is difficult to meet the support requirements of information war precisely.Based on this,‘precise support’ has already become major issues need to be resolved of thefuture war, and also become an important content of modern military logistics.First of all, the paper discusses the basis of our armored forces’ fuel consumptionprediction, analyzes the nine main factors that affect the armored forces’ fuel consumption,and takes them as the inputs of the prediction model, introduces the basic stages of modeling,steps of prediction and related technology.Secondly, the paper expounds the basic principle and problem-solving steps of greycorrelation analysis, discusses the advantages of grey correlation analysis, then outlines basicthoughts and prediction steps of GM(1,1) model, and uses the principle of ‘the new replacesthe old information’ to improve the model, makes it more suitable to the fuel consumptionrules. The paper introduces the structure and training methods of BP neural network, aimingat the defects of traditional BP algorithm, uses four kinds of methods to improve theprediction precision.Thirdly, an improved series grey neural network combination prediction model was putforward, uses the grey model’s prediction result and the main factors as the inputs of neuralnetwork, and uses the grey correlation degree of fuel consumption factors as weightcoefficient, and uses BP neural network model to achieve the best fit of predicted value andactual value.Finally, the paper inspects the prediction models and improved methods throughexperiment respectively. The results showed that gray neural network combination model can predict armored forces’ fuel consumption with high precision, so the model can better guidethe army for fuel supply management in the future.