Data driven fault diagnose method of polymerize kettle equipment

Aiming at the real-time fault diagnose and optimized monitoring requirements of the large-scale key polymerization equipment of PVC production process, a real-time fault diagnose strategy is proposed based on SOM neural networks. Through training of the SOM neural network, a layout is established to make each weight vector located in the centers of input vector clusters. When the training process is completed, the SOM neural network can be used to realize the fault diagnose for the training samples or other process data. Simulations experiments are carried out combining with the industry history datum to show the effectiveness of the proposed the fault diagnose method based on the SOM neural networks.