Optimization Method for Input Parameters of Plasma Jet Generator based on the Coupling Model of SVR and GA

The change of ambient temperature can exert effects on length and width of plasma jet flow while voltage peak value, voltage frequency and gas flow are the adjustable input parameters of plasma jet generator that can also exert effects on quality of jet flow. In order to seek the input parameter values for jet generator to generate the optimal jet flow quality under any ambient temperature, in this paper, the optimization method for input parameters of plasma jet generator based on the coupling model of Support Vector Regression (SVR) and Genetic Algorithm (GA) is proposed and practically applied to the research of a common plasma jet generator. The method consists of three steps, including confirmation of the optimization model, establishment of jet flow quality prediction model based on SVR, parameter optimization based on jet flow quality prediction model and GA. In practical application, with the product of jet flow length and width as the objective function and voltage peak value, voltage frequency and gas flow as the design variables, the constraints of voltage peak value, voltage frequency and gas flow and temperature are confirmed by the preliminary tests in this paper. Orthogonal table can be obtained by 25 tests to establish the prediction model training set while the jet flow quality prediction model based on SVR can be established through MATLAB. When the ambient temperature is known, GA is adopted to solve the model. The optimal jet flow parameters under a certain ambient temperature can be eventually confirmed by a limited number of parameter optimization. The results show that the plasma jet flows generated under the parameters confirmed by this method are better in length and width compared to other parameters, indicating that the input parameter optimization of the jet generator can be realized. The method proposed in this paper is applicable to parameter optimization of the plasma jet generator of various structures and is a highly feasible and universal algorithm.