The Research Of Refrigeration Compressor Sales Prediction Based On A New General Regression Neural Network

With the rapid development of food frozen, demand for refrigeration compressors also will rapidly increase. How to get an impregnable position in the process of rapid development of the industry is becoming the primary problem of each enterprise. And how to occupy the market is a key problem, but the market occupation is not a simple subjective speculation, but to rely on scientific and effective prediction algorithm. And the effective market prediction is based on a research of various market factors.In this paper, we have introduced the research background and research significance of refrigeration compressor sales prediction, and analyzed a several factors which mainly influencing the refrigeration compressor sales. Then, this paper introduces several kinds of Traditional algorithm of market sales prediction, such as linear regression prediction algorithm and neural network prediction algorithm, etc. And analyzes their advantages and disadvantages respectively.According to the disadvantage of various Traditional prediction algorithm, this paper has proposed a new prediction algorithm of Generalized regression neural network(GRNN) which is based on the improved genetic algorithm optimization prediction algorithm. While the performance of GRNN neural network prediction is based on the parameter of smooth factor setting, So the value of smooth factor is calculated by the improved genetic optimization in order to get the best performance of GRNN neural network.At last, through analysis of the MATLAB simulation results, we can get a result that the method in our paper has a high precision and fast convergence, train the sales data of refrigeration compressor for the last years, and get the future market forecast for a period of time the overall error is within 5%.