With the rapid development of modern manufacturing, welding robot plays a more and more important role in the industrial production. When use the robots for welding production, it is very important for choosing the welding process parameters, since the weld joint quality determines the product is qualified or not. Aiming at the welding process has highly non-linear and difficult to describe by mathematical model, in this paper, the artificial neural network was used to build the mapping model between the process parameters and the weld geometry size, then established the welding parameters optimization system by using the Genetic Neural Network, thus, provides a new method for the robotic process parameters optimization and weld quality prediction.Used the existing OTC welding robot system as a test platform,then the orthogonal experimental method was used to arrange experiment to obtain the sample data. By this way, it can not only shorten the test cycle, reduce test cost and difficulty, but also ensure that the sample data is representative and meet to the requirements of network training.Firstly, take advantage of the genetic algorithm global search function to find the optimal BP neural network initial weights and thresholds, then trained on the sample data, established the relationship model between the welding process parameters and the weld geometry size. The model can be used to predict the weld geometry size, at the same time, in order to realize the optimization of process parameters, the model can transformed into the fitness function of genetic algorithm to find the welding process parameters which fit to the given weld size requirements.In order to facilitate welding process parameter optimization, use the GUIDE of MATLAB software to design the welding parameters optimization interface system. Through this system, we can establish the prediction model, predict weld geometry size, optimize process parameters quickly. |