基于改进神经网络预测的智能污水处理监控系统设计

发布时间：2018-01-01 08:20

本文关键词：基于改进神经网络预测的智能污水处理监控系统设计　出处：《青岛科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：本设计以青岛市高新区污水处理厂为现场背景,在对其工艺流程、设备设施作了详细地介绍与分析情况下,根据信息物理融合的思想以及工业4.0的要求进行了监控系统的全面设计。总体设计上采用四层信息物理架构,分为感知交流、融合处理、推送、执行四大部分。融合处理部分采用神经网络智能算法实现溶解氧预测,根据预测值调整送氧量实现精确曝气,进而优化出水水质;推送、执行部分采用多参数监测实现设备平稳运行,保证系统安全。污水处理过程采用的是A2/O工艺,其净化机理主要是好氧池(即曝气池)污泥中附着的微生物在适当的氧气条件下,通过新陈代谢分解污染物,实现污水的净化,因此溶解氧的控制最为关键,所以本设计提出一种新的基于神经网络的溶解氧优化控制策略。通过试验以及历史数据,获取在出水较优的情况下的曝气池入水指标以及此时的溶解氧值作为样本,根据样本训练采用粒子群算法优化的BP神经网络,最后实现在不同入水条件下的溶解氧的精准预测。污水处理设备的平稳运行由下位机和上位机共同来完成。下位机设计中,首先对污水处理的各工艺段按照顺序配置了监测设施,全面采集各设备参数,监测关键设备的开启以及状态;然后设计了上位机与下位机和下位机与传感器之间的通讯网络;下位机采用PLC作为核心,通过STEP7对PLC进行编程,采用PID算法进行溶解氧控制。上位机用C#语言开发,实现用户登录、实时数据显示、超限以及故障报警、报表查询、用户管理,并且通过混合编程,将Matlab编写的溶解氧预测神经网络集成在上位机平台里,由上位机把预测出的精确值传给下位机实现溶解氧参数的设置。最后通过系统的现场实施应用,利用检测仪对水质数据进行检测,与之前出水水质数据进行对比,证明了本设计的优良特性。
[Abstract]:This design takes the sewage treatment plant of Qingdao High-tech Zone as the scene background, under the detailed introduction and analysis of its technological process, equipment and facilities. According to the idea of information physical fusion and the requirements of industry 4.0, the overall design of the monitoring system is carried out. In the overall design, four layers of information physical architecture are adopted, which are divided into perceptual communication, fusion processing and push. Four parts are implemented. In the fusion part, the neural network intelligent algorithm is used to predict the dissolved oxygen, and the oxygen delivery is adjusted according to the predicted value to realize the accurate aeration, and then the effluent quality is optimized. Push, the executive part uses the multi-parameter monitor to realize the equipment to run smoothly, guarantees the system safe. The sewage treatment process uses the A2 / O process. The main purification mechanism is that the microorganisms attached to sludge in aerobic tank (aeration tank) can decompose pollutants by metabolism under appropriate oxygen conditions, so the control of dissolved oxygen is the most important. Therefore, this design proposes a new neural network-based dissolved oxygen optimal control strategy, through experiments and historical data. The parameters of aeration tank and the dissolved oxygen value of the aeration tank were obtained as samples, and the BP neural network was optimized by particle swarm optimization according to the sample training. Finally, the accurate prediction of dissolved oxygen under different water entry conditions is realized. The smooth operation of sewage treatment equipment is completed by the lower computer and the upper computer. First, the monitoring facilities are arranged for each process section of sewage treatment according to the sequence, and the parameters of each equipment are collected, and the opening and status of the key equipment are monitored. Then the communication network between the upper computer and the lower computer and between the lower computer and the sensor is designed. The lower computer uses PLC as the core, PLC is programmed by STEP7, and dissolved oxygen is controlled by PID algorithm. The upper computer is developed with C # language to realize user login and real-time data display. Beyond the limit and fault alarm, report query, user management, and through mixed programming, the dissolved oxygen prediction neural network written by Matlab is integrated into the upper computer platform. The precise value of the prediction is transmitted to the lower computer by the upper computer to realize the setting of the dissolved oxygen parameter. Finally, through the field application of the system, the water quality data are detected by using the detector. Compared with the previous effluent quality data, the excellent characteristics of the design are proved.
【学位授予单位】：青岛科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：X703;TP277

【参考文献】