Development Of Chain Resistance Butt Welding Monitor Based On STM32 And LabVIEW

As a basic mechanical component,the chain has been widely used in ships,metallurgical mines,agricultural machinery and other fields.Because the links of the chain are interlocked,as long as one of the links does not meet the welding standards,the entire chain will be scrapped.Therefore,it is very important to ensure the welding quality of the links.However,there are many factors that affect the welding quality of resistance butt welding,such as current,voltage,upset,etc.,and the destructive tensile test cannot determine the specific cause of welding failure.Therefore,the goal of this article is to establish a monitoring system,collect and store chain welding signals,and then analyze the signals to provide a basis for post-weld diagnosis.This paper designs the overall architecture of the system according to the monitoring needs of resistance butt welding,and divides the system into three parts: single-chip microcomputer acquisition part,Lab VIEW program and Python offline analysis.The signals collected by the monitoring system include electrical signals(current,voltage,differential)and mechanical signals(electrode displacement,spindle rotation angle).After analyzing the characteristics of the above signals,the corresponding sensing scheme is designed,and the corresponding sensor interface is configured on the single chip microcomputer.The signal acquisition part adopts STM32H7 as the working core,cooperates with the peripheral analog quantity processing and TCP communication module to complete the signal acquisition and transmission tasks.The Lab VIEW host computer program receives and displays the signals and stores them in the My SQL database for offline analysis after welding.The analog signal of the monitor system is isolated,amplified,offset adjusted and filtered before it is connected to the AD of the microprocessor;The digital signals are isolated by optocouplers to ensure the safety of the system.The circuit gain,reference voltage,filter frequency,etc.can be adjusted by the digital potentiometer.The cut-off frequency adjustment range of the filter circuit is(5.1KHz,850KHz),which can meet the requirements of signal filtering.The system has also designed an integral recovery circuit for measuring the welding current using Roche coils,which can suppress integral drift by means of compensation,discharge and integration.After that,the logic processing scheme of CPLD is designed,which delay-filters the output signal of the comparator,extracts the conduction angle and separates the multi-section welding.The software design of the monitoring system includes the ARM microcontroller software part,Lab VIEW host computer part and Python database management part.MCU interacts with labview to divide the system into oscilloscope mode,normal welding mode and parameter setting mode.The system transmits the instantaneous value of the welding signal in the oscilloscope mode,transmits the preprocessed welding signal in the welding mode,and adjusts the reference voltage and offset voltage in the parameter setting mode.In addition to querying historical signals,the Python database platform has also designed signal statistical graph drawing and stability evaluation functions based on dynamic resistance tolerance bands.The developed monitor is installed on the production line of the industrial site,and the tests show that the various functions of the system are basically realized.Onsite working condition test shows that the welding machine spindle rotation is relatively stable with a fluctuation range of 0.11%,while the grid voltage stability is very poor,with a fluctuation range of up to 8.2%.Further,the principal component analysis(PCA)was used to reduce the dimension of the dynamic resistance curve,and it was found that the data after the dimension reduction still retained the original resistance curve characteristics.The K-means and Gaussian Mixture Model(GMM)clustering analysis of the dynamic resistance curve after dimensionality reduction shows that the clustering method can separate the resistance curves with obvious differences.Finally,the isolated forest algorithm is used to clean the data and distinguish the electrode displacement curve abnormally,and the detection effect is good.