论文部分内容阅读
BP神经网络和遗传神经网络是混合气体识别中常用的方法,但在实际应用仍然存在一些缺陷与不足。针对存在的问题,提出了1种改进自适应遗传算法,该算法根据进化过程种群中未产生更优解的代数,自适应调整变异率和变异量。利用该改进自适应遗传算法优化BP神经网络的连接权和阈值,构成改进自适应遗传神经网络,并应用于混合气体的识别中。实验结果表明:改进自适应遗传神经网络收敛成功率由40%提高到80%,平均识别误差H_2S由4.66 mL/m~3降为3.69 mL/m~3,CH_4由17.14 mL/m~3降为15.77 mL/m~3,CO由4.38 mL/m~3降为4.19 mL/m~3。
BP neural network and genetic neural network are commonly used methods in gas mixture recognition, but there are still some shortcomings and deficiencies in practical application. Aiming at the existing problems, an improved adaptive genetic algorithm is proposed. The algorithm adaptively adjusts the mutation rate and the variation according to the algebra which does not produce better solutions in the evolutionary population. The improved adaptive genetic algorithm is used to optimize the connection weights and thresholds of BP neural network to form an improved adaptive genetic neural network and to apply to the identification of mixed gas. The experimental results show that the success rate of the improved adaptive genetic neural network is improved from 40% to 80%, the average recognition error H_2S decreased from 4.66 mL / m 3 to 3.69 mL / m 3, and that of CH 4 from 17.14 mL / m 3 to 3 15.77 mL / m ~ 3, CO decreased from 4.38 mL / m ~ 3 to 4.19 mL / m ~ 3.