Research On Product Quality Prediction System Based On Improved Support Vector Machine

Nowadays the competition between enterprises is fiercer,it is more and more important for enterprises to improve their competitiveness.To achieve this goal,on the one hand,they have to design and produce products that are more suitable to the market,on the other hand,is the quality of the products should be qualified.It can be said that quality is the establishment that an enterprise can develop.After lots of years development,the technology of product quality management has been very adequate,it includes the quality management and traceability of the whole chain which contains from raw materials to finished products.What's more,if the production process can be predicted in advance before the product is finally completed and then provide early warning information for the next process of production,we can adjust the production process timely,then we can reduce the output of non-conforming products.By doing this,we can reduce enterprise's costs,enhance their competitiveness.The core of product quality prediction is to find a suitable prediction model.Several commonly used prediction models are summarized.By doing simulation experiments,the support vector machine model was selected for further researching and learning.The idea of using genetic algorithm to optimize two parameters the support vector machine needed:radial base core function parameter ? and error penalty parameters C is proposed.By selecting the parameter combination that is most suitable for the current input to make the model predict better result.After the simulation experiment,the results show that the support vector machine model based on genetic algorithm optimization is better than the support vector machine model without parameter optimization.This paper designs and preliminarily realizes the product quality prediction system based on the support vector machine model with genetic algorithm parameter optimization.This system can train and establish the prediction model according to the historical production data of the current product.The system can collect production data of the products which on the current production line and put the data into the prediction model to finally obtain the prediction value of product quality.Then system can display it in real time on monitor.The technicians can check the prediction result and adjust production parameters to ensure the product is qualified.