Research On Spindle Health Monitoring Method For Multi-Line Slicer

As a key equipment in the photovoltaic semiconductor industry,the high maintenance cost of multiline slicers seriously restricts the development of the entire industry.As one of the core components of a multi-line slicer,the spindle’s performance directly affects the production quality of silicon wafers and its maintenance costs account for over 60%of the overall maintenance cost.As the running time increases,the cutting rollers,bearings,and other components in the spindle caused the abrasion in different degrees,resulting in abnormal vibration,increased noise,and decreased surface quality of silicon wafers.Therefore,the spindle is the main inspection object for the multi-line slicer.At present,manual regular inspection is the main inspection method for multi-line slicing machines,but it has been criticized for its low efficiency and difficulty in avoiding missed and erroneous inspections.Sensor-based monitoring technology has become mainstream with the maturity of monitoring technology.However,due to the nonstationary and complex vibration signals of the spindle of the multi-line slicer,conventional monitoring methods are no longer applicable,and a single factor is difficult to comprehensively and objectively evaluate the spindle health status,which poses a challenge to the health monitoring of the spindle of the multi-line slicer.This article focuses on studying the health monitoring of multi-line slicer spindles based on vibration sensors.To address the above issues,a microelement kurtosis method is proposed.This method utilizes the integration principle to decompose the spindle vibration signal into multiple segments,with each segment approximately treated as a stationary signal.The kurtosis of each segment of data is used to characterize the magnitude of the spindle vibration.When the kurtosis is greater than 3 and periodic impacts occur,the spindle health status is abnormal;A multi-source data fusion health monitoring model is constructed using the average time between failures,spindle vibration kurtosis,and silicon wafer processing accuracy of the multi-line slicer spindle to monitor the health of the slicer spindle.This model first utilizes Analytic Hierarchy Process(AHP)to configure the weights of each element,then constructs the triangular membership function of each indicator through fuzzy mathematics principles,and fuses each indicator to comprehensively and objectively evaluate the health of the multi-line slicer spindle.Finally,in order to effectively extract fault features from the signal,a VMD-SWT adaptive threshold wavelet denoising algorithm is proposed to preprocess the spindle signal.In order to verify the feasibility and effectiveness of the algorithm,the algorithm was validated on a multi-line slicer.The experimental results showed that after VMD-SWT adaptive threshold denoising,fault features can be effectively extracted from the data;Even under strong background noise,using the microelement kurtosis method can still accurately and effectively monitor the fault of the slicer spindle in real-time,with a fault recognition rate of 94%;At the same time,the health monitoring model based on multi-source data fusion can scientifically and accurately evaluate the spindle performance,and has good stability.Finally,the proposed microelement kurtosis method was encapsulated using C#winform to develop system software for monitoring the health of the multi-line slicer spindle.After multiple tests,it was shown that the software can quickly and stably monitor the health status of the multi-line slicer spindle.