Research On Spot Welding Quality Monitoring System Based On BP Neural Network

Resistance spot welding technology,with its advantages of simple operation,good economy and high production efficiency,is widely used in automobile body-in-white manufacturing and aerospace manufacturing,and the quality of solder joint will directly affect whether industrial products meet national standards,so it is very important to ensure the quality of solder joint qualification rate.However,in the process of welding,the fusion core is always in a closed state and the quality of welding spot cannot be directly monitored.In addition,spot welding process is difficult to be described by an accurate mathematical model because of its complexity,multi-parameter coupling and many random factors.Therefore,it is of great significance to comprehensively analyze the parameters of spot welding process and establish the accurate mapping relationship between them and welding spot quality.The main contents of this paper are as follows:First of all,in many process parameters affecting the quality of solder joint,selection of welding current,electrode pressure and welding time as the input parameters of spot welding quality monitoring system,and the sensor of the above process parameters for real-time acquisition,data information bamboozled by the signal conditioning circuit and modulus conversion circuit processing operations to quality monitoring model of judgement,So as to achieve the real-time prediction of welding spot core diameter and the goal of judging welding spot quality.The hardware system of spot welding quality monitoring mainly includes data acquisition module and data analysis module.The data acquisition module is composed of Rogowski coil current sensor,strain type pressure sensor,signal conditioning circuit and ADC signal acquisition circuit.The data analysis module realizes the prediction of welding spot core diameter according to the spot welding quality monitoring model.In addition,the power supply circuit of the spot welding quality monitoring system and the peripheral circuit of the spot welding quality unqualified alarm are designed based on the Helper A133 main control chip,The latter is used to sound and light alarm for welding process with unqualified predicted results..Secondly,a 3×4×1 spot welding quality monitoring model was constructed based on BP neural network.The welding current,interelectrode pressure and welding time were taken as input,and the diameter of welding spot core was taken as output.In order to make up for the inherent defects of BP neural network standard algorithm,genetic algorithm was used to select the optimal initial weight and threshold value to optimize and improve the spot welding quality monitoring optimization model.Finally,the spot welding quality monitoring software system is designed based on Linux system,and the interactive interface of spot welding quality monitoring system is designed based on Qt development software,including user login interface,spot welding quality monitoring model training interface and prediction interface,which is used to realize the transmission,storage and display of welding process parameter information.The spot welding quality monitoring model program optimized by genetic algorithm is used to predict the diameter of welding spot core.In this paper,the spot welding quality is indirectly judged by monitoring the process parameters of spot welding,and a spot welding quality monitoring model based on BP neural network is established.At the same time,the inherent defects of BP neural network standard algorithm are remedied by genetic algorithm.In the spot welding quality monitoring system,the mapping relation between welding current,pressure between electrodes,welding time and the diameter of welding spot core is established,so as to achieve real-time prediction and determine whether the quality of welding spot is qualified.The average relative error between the predicted value and the actual value is 5.31%.The results show that the spot welding quality monitoring system designed based on BP neural network can accurately predict the diameter of welding spot and can be used to judge the quality of welding spot,which lays a foundation for the subsequent improvement of welding spot pass rate.