Abnormal state detection of production system based on the support vector data description

Abnormal state detection is very helpful for managers to find production system´s uncommon conditions timely, it can greatly reduce potential loss and increase the enterprise´s economic profits. To improve the work quality, this thesis puts forward an abnormal state detection model; the model is based on the theory of support vector data description. Firstly the production system´s evaluation indexes are defined, and the thesis points out that for different types of production system, the indexes may be quite different. Secondly, the relative abnormal state detection model is built, and then, the thesis briefs the basic theory of support vector data description. Thirdly, the production system´s abnormal state detection model is built and verified by an experiment. The result of the experiment shows that the model proposed by the thesis is effective and helpful.