Research And Implementation Of Insulation Monitoring Technology For Aluminum Plants's Crane Based On Neural Network

The crane is the core electrical equipment in the aluminum electrolysis process,with complex and huge electrical structure.Most aluminum electrolysis plants are in the harsh environment of strong magnetic fields,strong conductive dust and damp heat.Therefore,the insulating mat of the multifunctional crane may have an insulation fault in which the insulating surface is broken.The manual inspection method may cause problems that are not timely.In this paper,the insulation of the crane is taken as the research object,and the neural network algorithm is incorporated.It is used to classify and identify the fault of the insulation state of the crane,and combine it with the intelligent industrial Internet of Things platform with both intelligence and intelligence functions.The fault is monitored online.The IoT platform can realize fault diagnosis and visual monitoring based on real-time monitoring data for the insulation performance of the crane during the aluminum electrolysis process.This paper analyzes the characteristics of leakage current when the insulation material of aluminum electrolysis crane is affected by external factors.The leakage current collected by the sensor was analyzed by Fourier transform.The data analysis provided by the aluminum plant indicates the characteristic performance of the leakage current at different faults.From the results,it can be found that there is a significant difference in the magnitude of the leakage current occurring in the case of different faults in the range of 0.1 Hz to 0.5 Hz.Aiming at the fault classification of the aluminum electrolysis plant,the characteristic type of leakage current is taken as the characteristic input of the neural network,and the fault type is used as the characteristic output,and the modified BP network model is established.And in the multiple training,the network structure model with input neurons of 6,hidden layer of 8,and output neurons of 3 is determined.The data provided by an aluminum electrolysis plant is used to verify the effect of the improved neural network model,and the classification of faults has a high accuracy rate,which is in line with the requirements of actual industrial operations.This paper encapsulates the Matlab neural network toolbox of M language into components,links into the cloud platform based on java,and deviates from the working environment of Matlab,and designs the fault monitoring mode with two main modules of operation and maintenance and monitoring.