Study On Online Monitoring Of Gearbox Lubricant Based On Multi-source Information Fusion

With the rapid and intelligent development of mechanical equipment,the technical level of condition monitoring and fault diagnosis has been improved constantly.Owing to the capacity of evaluating the operating condition of equipment accurately,oil analysis technology has become an effective means to provide early warning in fault diagnosis.There are problems of low efficiency,single monitoring parameters,and low intelligence in the current oil monitoring technology.Based on the theory of multisource information fusion,the multi-source parameter online monitoring system of lubricant is designed,which can monitor the characteristic information of the gearbox lubricant and fuse the monitored information in real-time to obtain the operating condition of the gearbox.Based on the comprehensive analysis of the common fault types and generation mechanisms of the gearbox,the main monitoring content of the lubricant online monitoring system is determined as the information on viscosity,moisture,pollution,and abrasive particle of the gearbox lubricant.From the perspective of information theory,the feasibility and effectiveness of multi-source information fusion in the field of state monitoring are demonstrated.On this basis,the level,processing structure,and process of multi-source information fusion are studied.The structure of the distributed fusion system is adopted to fuse the sensor data at the feature level.Based on the determined characteristics information parameters of lubricant monitoring,a lubricant online monitoring system composed of modules of characteristic information acquisition and signal conversion and transmission was designed.In view of the actual operating conditions of the gearbox,the high-pressure diaphragm pump is used to sample the gearbox lubricant.Combined with the principle of lubricant online monitoring sensor,the applicable online viscosity sensor,moisture sensor,particle counter,and abrasive particle sensor are selected to complete the collection of lubricant feature information.According to the theory of multi-sensor detection fusion,the four sensors in the system are arranged in a parallel distributed structure to achieve optimal distributed detection.According to the output data type of the selected sensor,CAN-USB converter and RS485-USB converter are used to complete the signal conversion and transmission of the sensor.According to the functional requirements of the system,the software of the lubricant online monitoring system is developed based on the Lab VIEW program.The functions of serial port debugging and communication control,data acquisition and data processing,data display and data storage are realized by multithread technology.The automatic acquisition,processing,analysis,and storage of sensor signals are completed,so as to meet the functional requirements of the human-computer interaction of the system.The repeatability and accuracy of the lubricant online monitoring system are evaluated by the fault simulation experiment of the gearbox.On this basis,the model of operating condition evaluation based on BP neural network and D-S evidence theory is built,which can make full use of the data of information source to evaluate the operating condition of the gearbox.The software of the lubricant online monitoring system has been developed,by constructing the model of operating condition evaluation based on BP neural network and D-S evidence theory.The results show that the lubricant online monitoring system can realize the monitoring of the lubricant characteristic information of the gearbox,the repeatability and accuracy of which can meet the monitoring requirements of the system.In addition,the model of operating condition evaluation can make full use of the data of the information source to realize the accurate evaluation of the operating condition of the gearbox.This paper contains 53 figures,17 tables,and 93 references.