Optimization Of Technics Parameters Of Welding Robot CO₂ Gas Protecting And Welding Based On The Neural Network

In order to further forecast the geometrical figure of welding line, the paper puts forward the necessity and practicability of the parameter optimization system of CO₂ gas protecting and welding based on the development demand of CO₂ gas protecting and welding technology. Utilizing the non-linear feature of Artificial Neural Networks and its technology advantage in self-organise study and fuzzy reasoning, the paper introduces the neural networks technology into this system, builds a parameter optimization system based on the neural networks for the sake of further resolving non-linear mapping problem existing between CO₂ gas protecting and welding technics parameter and geometrical figure of welding line and overcoming difficulty in building mathematics model. The paper pioneers a new way for forecasting the geometrical figure of welding line of CO₂ gas protecting and welding.For collecting the network training stylebooks, the paper adopts the orthogonal method to arrange experimentation and the range analysis to deal with the data, testifies it by an example and assures the representation of those training stylebooks.For confirming the parameters of the network structure, the paper simulates and trains the built BP neural network by using the neural network tool box of Matlab software and attains ideal parameters with the incessant adjust.In the paper the development of parameter optimization system adopts modularization design, uses VC++ as development tool under Windows environment, adopts design method facing object, builds man-machine mutual interface, network training module and data simulation module. In the network training module, complete the instauration of BP network, the setting of systemic parameters, the input of stylebook data and the training of the network, saves the trained network. The CMatrix class, which is user-defined, completes concrete calculation during the course of training. The data simulation module can input the trained network and simulated data to complete data simulation. Run the program and accomplish network training module and data simulation module, then can complete the forecast of geometrical figure of welding line, realize the optimization of welding parameters. The building process of the whole system provides available thought and method for the development of parameter optimization of other domains.