Font Size: a A A

Design Of Soft Measurement Instrument For Bod Parameters Based On Neural Network

Posted on:2015-03-28Degree:MasterType:Thesis
Country:ChinaCandidate:N GuoFull Text:PDF
GTID:2298330452453516Subject:Control Science and Engineering
Abstract/Summary:PDF Full Text Request
Biochemical Oxygen Demand (BOD) is one of the key parameters in evaluatingwater quality and sewage effect in the process of activated sludge wastewatertreatment system. Measuring BOD is very difficult because it involves many complexbiochemical reaction processes. Specific evidence is that international standard testmethod of measuring BOD has big hysteresis. In the majority of current sewagetreatment plant, microorganism aerobic respiration method is applied to get the valueof BOD. And this progress needs5days at constant temperature of20degrees aftercollecting water samples. There are other ways to measure BOD, for example, usingmicrobial membrane electrodes. But in this way, the instrument would be tooexpansive and the results might not be so precise.In order to solve the problems existing in the measurement of BOD, neuralnetwork soft measurement technology has been used to construct an online model ofBOD. And then, a new kind instrument has been designed based on the neuralnetwork for BOD soft measurement, which makes it possible to predict BOD on-linerapidly and accurately. In this paper, the main work is embodied in the following fouraspects:1. In this master’s thesis, detailed analysis of the auxiliary variables selection hascarried on for BOD soft measurement through the principles of microbial growth insewage treatment system. In order to make the choices of auxiliary variables morereasonable, existing testing instrument in sewage treatment plant was studied. pH、ORP、DO and temperature had been chosen at last.2. A three layer feed forward BP neural network has been designed for modelingand predicting of BOD. Special matlab experiments are designed to verify theperformance of the neural network. BP neural network is easy to fall into localminimum when it is training. Vector momentum item and adaptive network wasintroduced, which make the neural network more stable and gives higher predictionprecision.3. A kind of instrument which combines software and hardware has been devisedto soft measure water quality parameters of BOD rapidly. After that, we test this BODsoft measure instrument in one of the sewage treatment plants in Beijing. Very goodresults had been achieved.
Keywords/Search Tags:Neural network, Software measurement, Instrument
PDF Full Text Request
Related items