Ozone Technology In Neural Network Soft Sensor Modeling

Ozone oxygenized water processing system is the most important application field of ozone technology, by years of development; it has improved both widely and deeply in this field. As a industry process, there is still no maturity auto controlling scheme in our country. The delay of ozone concentration measurement and the lag of water quality standard examination are mainly factors that result in lower quality of ozone-oxygenized water processing system. In this paper, based on the analysis of theory and practice of research, we build the model for ozone concentration soft measurement initiately in the country. We also make discussion and research for the ozone oxygenized water processing experiment system modeling and effluent COD forecasting. Followings are main study content and result:1. According to the experiment result and mechanism analysis of the ozone producing process, select peak voltage value Vpp, frequency f and raw gas flux Q (converted to atmosphere pressure) as assistant variable in ozone concentration soft measurement. In order to get appropriate experiment data for modeling, adopt the equality experiment scheme.2. As times of simulations and experiments, use LM algorithm, which is one of improved BP algorithm, to build up soft measurement model. And give out the implementation of visual neural network modeling software. As^ the presentation of simulation and verification, the soft measurement modeling method proposed in this paper has some precision and spreading ability, it is prove to be feasible and efficient.3. In the project of industry pollution water processing research and development that is joined with Dalian chemical group ammonia synthesis factory, practical data was gained in the series system experiments. With these data, we have set up the neural network relation model of ozone amounts, influent COD and effluent COD for ozone-oxygenized water processing system. According to this model, we forecast the effluent COD value by influent COD, and ozone amounts, and get the best ozone amounts and its working curve. Finally give out a economy and technology feasibility study result for ozone oxygenized water processing system.The soft measurement model and ozone oxygenized water processing system neural network model introduced in this paper has proposed theory study andexperiment demonstration to realize a whole auto-controlled ozone oxygenized water processing system based on Soft Measurement. The latter also provides forecasting reference for industry pollution water ozone processing technology design. The result of the study shows that it is feasible and efficient using neural network soft measurement method to resolve the delay of ozone concentration measurement and the lag of water quality standard examination in ozone oxygenized water-processing system. And it worth of furthering study and research.