【摘要】 【目的】为探索接触辉光放电过程中的伏安特性和等离子体中·OH的含量变化。【方法】构建一种CrossBP(Back propagation network)神经网络模型和遗传算法（Genetic Algorithm）优化的GA-BP神经网络模型，预测在不同电压、放电时间、电解质初始浓度等条件下的伏安特性和·OH浓度、pH和电导率。【结果】建立的Cross-BP模型（结构2-11-1）和GA-BP（结构3-8-3）模型经精度（调整后的R~2≥0.992 1）和试验验证能在一定误差内预测放电过程中伏安特性和·OH的变化，且GA-BP模型预测结果的MSE、MAE值更低。【结论】所构建的GA-BP模型对等离子体中·OH含量变化的预测具有高精确度和高效率。更多还原

【Abstract】 【Objective】 To study the magnitude of the volt-ampere characteristic and the content of hydroxyl radicals in the plasma during the contact discharge process.【Method】 Constructed a Cross-BP(Back Propagation Network)neural network model and genetic algorithm(Genetic Algorithm) optimized GA-BP neural network model, predicted the volt-ampere characteristics and hydroxyl radical concentration, pH,and conductivity of the electrolyte at different voltages, discharge times, electrolyte initial concentrations using two models respectively.【Result】 Both the established Cross-BP model(structure 2-11-1) and GABP model(structure 3-8-3) can predict the changes of volt-ampere characteristic and hydroxyl radical during the discharge process in a certain error through accuracy(adjusted≥0. 992 1) and experimental verification, the MSE and MAE value of the GA-BP model is lower.【Conclusion】 The obtained GA-BP model has high accuracy and efficiency in predicting the change of hydroxyl radical content in the plasma.更多还原