滤料气水反冲洗性能的人工神经网络模拟研究

对3组不同滤料的气水反冲洗性能的正交试验结果,用神经网络方法进行建模研究。建模实践表明:人工神经网络方法适合于滤料反冲洗的灰箱和黑箱系统的建模。建立了滤料气水反冲洗耗水量的BP人工神经网络模型,以便对变量进行重要度和灵敏度分析。在研究的空气反冲洗强度、水反冲洗强度、气水同时反冲洗历时、单水漂洗强度和池型5个变量中,水反冲洗强度最重要,空气反冲洗强度最不重要,为研究在保证清洁度的前提下有效降低气水反冲洗耗水量提供了理论依据。

Simulation on property of air-water backwashing using artificial neural networks

According to the orthogonal test results of three different filtering media, the properties of air-water backwashing were modeled using artificial neural networks. The modeling practice shown that the artificial neural networks were very suitable for modeling the air-water backwashing process. The BP model of water consumption in air-water backwashing was established and thus could be used to analyze the influence and sensitivity of each variable on water consumption. To the five variables, including the intensity of air backwashing, intensity of water backwashing, time of air-water backwashing, intensity of surface water washing and figure of the filter, the intensity of air-water backwashing was the most important one and the intensity of air backwashing the least. The theoretical and practical measures of decreasing the water consumption with the insurances of the cleanness of filtering media were put forward.