Quality stability evaluation model based on intelligent control

Abstract

Abstract At present, statistical process control in process industry is mainly off-line, which is incompetence in handling problems such as process control deviations and quality instability. In this work, an intelligent process control methodology is proposed for the on-line quality evaluation, by using GA-BP hybrid algorithm. Given that fact that the practical applications in industries are usually muti-objective, the multi-output neural network model is studied in order to compensate the deficiency of the single-output GA-BP algorithm. The proposed methodology was applied to a real operation unit to illustrate its effectiveness and feasibility. In the case study, two quality indicators were chosen as online evaluation objectives of the chosen operation unit. The algorithm was constructed, with 12 nodes in the input layer and 2 nodes in the output layer, which is then trained by 800 groups of real-time data and validated by 200 groups of test data.