Research Of Status Assessing Methods Of Equipment Lubricating And Wearing Based On Oil Monitoring

Mechanical device condition monitoring and preventative maintenance are the main content of modern management of machinery. With the rapid development of modern production, they become more and more important. As one of the main techniques of equipment condition monitoring, oil monitoring is playing an extremely important role in assessing the status of lubricating oil and machines, forecasting the trend of machines, finding out the potential problem of machines in time, reducing the cost of maintenance and the time of labor-hours lost and improving the utilization rate and safety of machines.The paper mainly researches how to use these kinds of data that hail from oil monitoring for assessing the status of lubricating oil and equipment. That was set in Guangzhou Mechanical Engineering Institute Equipment Lubricating & Wear Condition Monitoring Center (namely Mechanical Industry Oil Testing & Assessing Center). These data include spectroscopic analysis data, physical and chemical analysis data, direct-reading ferrograph data and analytical ferrograph image etc.The paper firstly introduced the survey of technology of oil monitoring and the processing methods of the data of oil monitoring. And then some methods, which is used to assess mechanical device lubricating and wear status, are put forward according to these data. As far as those assessing methods based on single monitoring parameter are concerned, the paper improves on the generally used method of threshold computing, and proposes two kinds of graph as the assistant analysis tools, which are used in the technology of exploratory data analysis. As far as those assessing methods based on multiple monitoring parameters are concerned, the paper extends fuzzy integration method from assessing the wear status of mechanical device to the lubricating status, and introduces the support vector classifier into oil monitoring for status assessing while the thresholds of oil monitoring parameters are not established. It specially integrates the ferrograph image with these methods for analysis example. Lastly, the paper analyzes five kinds of artificial neural network and support vector regressor. After comparing their performance of forecasting, these winners are chosen for application in oil monitoring. Every method mentioned above is proved by some illustrations and examples.Also, the paper gives the software system analysis, design, some functional examples and the introduction of running. In the end, summary of the paper and a vista of future are provided.