Researsh Of Food Cold Chain Logistics Service Quality Evaluation Based On PCA-BP Neural Network

The rhythm of our daily life lead the demand of Cold chain food constantly growing at the same time, peoples' pursuit of food was change from amounts to quality, but, at the present, the development of cold chain logistics is still in a relatively backward level, therefore, in order to visually express the logistics company's activities can guarantee to achieve customer satisfaction, food cold chain logistics service quality evaluation is particularly important. Since the cold chain logistics is a system, so food cold chain logistics service quality will be influenced by many factors, therefore, we need to consider the characteristics of products to analyze the food cold chain logistics activities. Based on this research, the paper point to build a food cold chain logistics service quality evaluation system and use method to evaluated.First of all, the paper analyzes the characteristics of food cold chain logistics and the connotation of quality of service, and then, adjust the SERVQUAL model and LSQ model's dimensions to quality, timeliness, reliability and tangibles, and combine the food(yogurt) cold chain logistics process to determine the affect factors of the services' quality, thus, the evaluation system of service qualitye was established, it contains 14 factors and the explanation for the value of index. Use the data services of an enterprise of Guangdong Province for an example to build BP neural network evaluation, it was found that the result of the evaluation is not satisfactory, from the literature, it can be found that a high correlation of input layer indexes will have a bad effect when use the BP network. Therefor, using the statistical methods of principal component analysis to reduce the correlation between the indexs. In order to reduce the dimension, using the principal component analysis to the raw data, then, the raw data will change into seven main components, the main component score formula can be calculated, According to this formula, the score of each main component can be calculated. After that, using the main component of the score as BP neural network input scalar, and use the part of data as the training data, the other part of data as the forecast data. At last, building a new BP neural network model, and implement training network in MATLAB, using the trained network predict the forecast data, the prediction error can be calculated. Then, compara with previous results, it can be found that aftert through PCA dimensionality reduction process the evaluation result will be better than before.In this paper, the research of food cold chain logistics service quality evaluation can enrich the content of cold chain logistics and the evaluation methods of cold chain logistics service quality.