Research On Health State Estimation Of Heavy-duty CNC Based On The Hidden Markov Model

The heavy-duty CNC machine tools were widely used in core national industries andkey projects for processing large and oversize parts.Once the machines broke up, it wouldcause great loss. In order to avoid the accident, the enterprises were urgently eager toobtain the health status of the machines and gave immediately repair. And it was essentialto improve the efficiency of CNC machine tools and to keep it running normally. So thehealth assessment technology for heavy-duty CNC machines was important in practical.In this paper，research on health estimate and Hidden Markov Model were summedup. The health estimate technology and the theory of the Hidden Markov Model werediscussed. The problems which HMM can solve and the algorithms of HMM had beenelaborated.Combined with the characteristics of the heavy-duty CNC machine tools, the healthstate of the heavy-duty CNC machine tools was defined, and its health status was graded.The HMM has been initialized by K-means clustering algorithm, whose parameterstend to reach global optimal solution. HMM’s algorithm was improved to build up healthassessment model facing the multiple performance parameters and multi-observationsequence. And it solved the problem that Hidden Markov model was limited for solvingthe single parameter observation sequence models.In this work, Heavy-duty CNC machine tool performance data was collected andhealth status assessment model of the multi-performance parameters observation sequencewas applied for ball screw. What’s more, heavy-duty CNC health status assessmentprototype system was developed with JAVA and the function of managing parameters dataof heavy-duty CNC was realized. Its health status was stored in the system. At last, thesummarization and the outlook were presented.