Design Of The Photoelectric Detection System For Organic Wastewater COD And Realization

Organic wastewater COD detection technology can be divided into two kinds of traditional chemical method and optical method.Potassium dichromate method is national standard method in many chemical methods.It’s high precision,but there are flaws of long testing period,secondary pollution,etc.So it is not applied to actual production application.In optical detection methods,single wavelength method can realize rapid on-line detection,but it doesn’t have a certain anti-interference ability.The prediction model according to least square method to establish is unable to effectively eliminate the interference of nonlinear factors in the waste water and the model is very unstable.At present,combine the spectral quantitative analysis technology with chemometrics methods,and establish a prediction model,to solve nonlinear factors,suspended solids and other interference on the result of prediction,improve the mathematical model between spectral data and the value of COD,improve the accuracy and efficiency of model prediction is the vital problem that we need to solve urgently in the field of organic wastewater detection.In this paper,the least squares method,BP neural network,and interval partial least squares method,as three different chemometrics methods,were used for modeling respectively in the process of detecting the waste water of a chemical plant situated in Zhejiang province.Through experimental comparison,the advantages and disadvantages of the three methods were summarized:the prediction accuracy of the prediction model based on interval partial least squares was up to 2%,proving it was obviously superior to another two methods.The major contents and innovative points of this paper are shown below:First,the current domestic and foreign research situations of COD detection technology and the significance of applying optical methods to the detection of organic wastewater COD were discussed.A combination of chemometrics methods with spectral quantitative analysis technology was emphasized;the design idea for flexibly choosing chemometrics methods based on concrete detection objects and actual targets was raised.Second,the basic principle of optical detection technology was studied.Centered on Lambert Beer’s law,the least squares modeling principle and algorithm of single wavelength method,UV wavelength scanning correction method,and the principle and calculation steps of dual wavelength method to exclude suspended solids using visible light were introduced.Then,the principle of multi-wavelength method to establish a model using BP neural network and partial least squares was emphatically introduced.Third,the hardware of the optical detector was designed:the hardware modules included light path system,circuit amplifier system,single chip microcomputer system,and PC system;the component selection criteria was analyzed.A/D module was simulated using PROTUES;GSM communication system was added into the overall structural design for fulfilling the remote water quality monitoring and making water quality management easier.The water sampling device integrating water sample pretreatmeng,cleaning and testing was designed,and the PLC program was designed according to the requirements of the system.Fourth,the characteristic wavelengths were chosen;a waste water model was established using the least squares method,interval neural network,and interval partial least squares method respectively.The prediction effect was detected using the test water sample to provide an error analysis;the main commands of MATLAB algorithm was introduced.Fifth,the concept of ultraviolet spectrum interval method was proposed,which could improve the prediction result deviation caused by the instability of light source or the fluctuation of the wave peaks of spectral scanning curves.