基于主元分析和神经网络的污水处理能耗分析

Analysis of Sewage Treatment Aeration Energy Consumption Based on PCA and BP Networks

目前水处理是一个高能耗行业,在污水处理厂的各种能耗中,污水处理曝气阶段的能耗是一个重要部分,几乎占到总能耗的60%。提供了一种简单有效的污水处理曝气能耗监测模型,该模型可以根据污水厂的入水水质参数及其他运行参数预测污水处理曝气能耗,有助于控制系统根据实时能耗调节控制策略。文中首先采用主元分析法(PCA)简化该模型的输入参数,其次采用神经网络模型对该模型的输入参数及输出参数进行映射。通过某污水厂的运行数据对该模型进行训练和测试,结果表明该模型性能学习精度高,有效预测污水处理曝气能耗。也表明污水处理曝气能耗不仅和入水流量、pH、BOD有关,也和曝气池温度有关。

Sewage treatment plant is a high energy consumption industry by now. Sewage aeration energy consumption is the main of total sewage treatment plant energy consumption, nearly accounting for 60% of the total. Present a simple, powerful aeration energy consumption monitor model, which supplies the real time energy consumption through the influent quality parameters and other parameters. The model is helpful for advanced control system to adjust its control opinion. To make the model simple, PCA is used to choose the influent parameters affecting energy consumption strongly as few as possible. The model established by back-propagation network is trained and tested by sets of a sewage treatment plant operational data. The test result shows that the model works well with high efficiency and accuracy. The test result also shows aeration energy consumption is not only affected by influent flow, pH ,biological oxygen demand （BOD） ,but also affected by temperature in aeration tank in the sewage treatment plant.