Design And Application Of Intelligent Monitoring System For Gearbox In Heavy-Duty Machinery

The gearbox is one of the important components of rotating machinery.This paper mainly develops the remote monitoring and fault diagnosis system for heavy machinery gearboxes.It is one of the important contents of promoting intelligent manufacturing and industrial internet under the background of Industry 4.0.Heavy machinery such as mine cars and large passenger bus,due to their large load and complex operating environment,the gearbox failures occur frequently,resulting in a significant decline in the performance and service life of heavy machinery.Therefore,it is urgent to study the remote monitoring and failure of gearboxes for heavy machinery.A diagnostic method and an application system based on the method.The main contents of the thesis are:First,the overall architecture of the gearbox intelligent monitoring system was designed.Based on the requirements of the acttial industrial application of enginecringmasters,based on key technologies such as industrial Internet of Things,intelligent monitoring and deep learning,a hierarchical architecture model was designed.Secondly,the intelligent monitoring hardware based on NB-IOT was developed.The overall design of the hardware includes two parts:the data collector and the networking module.According to the needs of the project,the selection of temperature sensor and vibration sensor was completed.The current mainstream industrial Internet technology was analyzed and analyzed.The narrowband Internet of Things NB-IOT technology was determined,and the data acquisition and networking communication scheme based on the technology was designed.A specific circuit diagram is shown.Thirdly,the fault diagnosis algorithm based on deep learning is studied.Deep neural network is a self-learning model that can automatically extract fault features.This paper mainly proposes an unsupervised fault diagnosis method K-SAE.The vibration signal of the gearbox is mapped by Laplace features,and the marker samples are generated by K-means clustering,and the fault diagnosis model is trained by the stack self-encoder SAE.The method directly inputs the time domain vibration signal,overcomes the shortcomings of the traditional supervised learning that requires manual marking of the training sample,and improves the intelligence of the gearbox fault diagnosis.Experiments show that the accuracy of the method can be compared with the supervised method,which proves the feasibility and effectiveness of unsupervised learning in gearbox fault diagnosis,and lays a foundation for industrial application of real-time diagnosis.Finally,a software platform for intelligent monitoring of gearboxes was developed,and the above hardware,algorithms and platforms were effectively integrated for practical industrial applications.The data acquisition and networking module is installed on the gear box,the acceleration sensor collects the bearing vibration signal,the thermocouple collects the bearing temperature signal,and the above-mentioned deep learning gearbox fault diagnosis algorithm is deployed in the cloud software platform,which can be used for the gear box and the mechanical equipment.Real-time monitoring of real-time operational postures and health signs,thereby improving the overall management level,decision-making level and emergency response efficiency of heavy machinery and equipment.Finally,it summarizes the work done and prospects the next research directions.