Quality Prediction Of Multi-varieties And Small Batch Products Based On Support Vector Machine

With the rapid social and economic development, people’s living standards have been continuously improving.The competition countries and enterprises are becoming more and more fierce.At the same time,with the consumer tastes continue to improve, the product features and characteristics can not meet consumers’ needs.Therefore, many enterprises in order to adapt to this demand, have to give up large quantities of production, instead of using a variety of small batch production.Product quality is the lifeblood of manufacturing enterprises, Only good quality products can be in the increasingly fierce competition in the international market may stand out.When the enterprise changes the modeof its production,it is impossible to collect a large quantity of quality data, which is difficult to predict the quality of the product. According to the characteristics of multi variety and small batch production mode, this paper presents a prediction model of support vector machine based on the theory of learning machine.First of all, the quality of the product and forecasting problems are introduced and analyzed,shortages and limitations of product quality prediction methods are pointed out.The new theory of support vector machine is proposed to predict the quality of multi varieties and small batch products.According to the characteristics of multi varieties and small batch production mode, analyzes the method of support vector machine based on the theory of machine learning to predict the quality of the feasibility,which is the theoretical basis for this study.Then based on the Mercer theorem, select several commonly used SVM kernel function, and combined with the actual business data, respectively with the kernel function on the quality of fitting regression experiment,the kernel function of the support vector machine model is finally determined. The next, according to the influence of model parameters on the prediction performance, and thus roughly determine ranges of the model parameters.By using genetic algorithm and particle swarm optimization algorithm, the model parameters are selected, and the complete quality prediction model isdetermined.Finally, this paper uses multiple linear regression method, artificial neural network and support vector machine to compare the results of the experimental verification,in order to verify the support vector machine prediction method of good results.