Research On Fault Diagnosis Of AGC Cylinder Internal Leakage In Large Rolling Mill

The AGC cylinder is an important part of the rolling mill system.The main function is to provide rolling force for the rolling mill.The rolling mill system is a system with high coupling between hydraulic and electromechanical.The performance of the AGC cylinder directly affects the quality of the rolling.Once the AGC cylinder fails,it not only directly affects the quality of the rolled strip,but also affects other parts of the rolling mill system,resulting in rolling mill equipment vibration,strip slip,deviation,broken belt,pile steel,etc.,which may lead to major accidents..Therefore,it is very meaningful to carry out AGC cylinder fault diagnosis to ensure the normal operation of the mill.In view of the current fault diagnosis process of the servo hydraulic cylinder of the rolling mill,the fault feature extraction is difficult,the signal nonlinearity changes,and the data volume is large.The research work and results carried out in this paper are as follows:In this paper,based on the hydraulic principle of AGC system,the AGC system simulation model is established in AMESIM software to analyze the influence of internal leakage coefficient on the rodless cavity pressure and AGC cylinder displacement of AGC cylinder during rolling process.The research shows that with the increase of the internal leakage coefficient,the response time of the rolling mill system becomes longer and the displacement peak begins to decrease.On the basis of the set input,the fault data of the AGC cylinder under different internal leakage coefficients are obtained,which provides samples for AGC in-cylinder leakage intelligent diagnosis.A method for fault diagnosis of rolling mill servo hydraulic cylinder based on deep confidence network is proposed.According to the working principle of the rolling mill system,the rolling mill system simulation model is established to simulate the leakage fault condition in the rolling mill.Using the superiority of the deep confidence network in intelligent fault diagnosis,the signal is normalized and put into the deep confidence network for training,and then through the back propagation learning,optimize the network parameters and improve the diagnostic accuracy.The deep confidence network model consists of a multi-layer Boltzmann machine and a top-level BP classifier.The results show that the deep confidence network model can be applied to the diagnosis of leakage faults in servo hydraulic cylinders under the condition that the training sample data is sufficient.