Data-driven Methods Of Equipment Health Assessment And Maintenance Decision-making

With the rapid progress of industrial technology, machinery and equipment constantly become more and more large-scale, complicated, precise and intelligent. How to correctly identify the equipment running status and make maintenance decision according to the equipment reliability analysis is the key to ensure machinery equipment can run efficiently and safely.For numerical control equipment plays an important role in the normal production of a manufacturing enterprise, this paper puts forward a scheme design of numerical control equipment data acquisition system in the manufacturing plant, which constructs a foundation platform to meet the digital workshop’s requirement of production process data acquisition. Design scheme is designed to monitor and control numerical controlled running status of equipment in workshop. At the same time, it is used for acquiring the internal and external characteristic information of numerical control equipment. So it can provide basic data for mechanical equipment health assessment, equipment reliability analysis and reasonable maintenance decision-making.In the health state evaluation of mechanical equipment research, this paper combines the equipment health level and the equipment degradation state recognition method based on hidden markov models. So it can effectively judge the equipment’s health level according to the condition information of equipment. To overcome the shortcoming in training method of hidden markov models, which depends on initial value and is easy to fall into local optimal. This paper puts forward a new training method based on improved particle swarm algorithm. The test of bearing health state evaluation shows the effectiveness of the new method.For reliability analysis of mechanical equipment and maintenance decision-making, this paper uses the weibull distribution proportional hazards models to establish the equipment reliability analysis model firstly. Then, the calculation methods of reliability index estimate and confidence interval were given by the proportional hazards model based on machine tool cumulative failure frequency. So we can determine the optimum preventive maintenance cycle of the machine tool according to the principle of minimum maintenance cost. At last, the proportional hazards model based on equipment condition vibration feature considers simultaneously the equipment failure risk and maximum availability. So we can make condition-based maintenance decisions according to the vibration characteristics of the equipment. Case analysis shows the process of specific maintenance strategy formulation.