Experimental Research And Practical Discussion Of COD Monitoring With UV Absorption Spectrum

Chemical oxygen demand monitoring technology of UV method utilizes the physical measure,based on Lambert-Beer law according to the relation between UVA and COD.Compare with the conventional measure,COD monitoring technology of UV is a quick monitoring measure, especially in the field of consecutive monitoring.However,rigor request of the stability of the water quality is a obstacle for popularize the UV method.Base on the predominance and problem ,in this paper,the technic detail and application of the UV monitoring is rearched:(1)Based on contrastexperimen between UV method and chemical method,the feasibility of COD monitoring technology of UV method is tested.The experiment deal with 23 types of sample including 8 living sewage and 15 industrial liquid waste.For the living sewage with the steady quality,the UVA can reflect COD accurately.But for the industrial liquid waste of complicated and changefully quality,the relativity is bad.The concentration of COD higher,the correlativity worse.(2)By the experiment, the influence law of turbidity, temperature, PH, interferential ion and the background of apparatus isdiscussed.(3)This paper starts with organic substance's ultraviolet absorption spectrum and chose some organic substances which have certain representaiveness.The correlation between ultraviolet absorption law and organic matter content are disscussed and BP-ANN(back propagation Artificial Neural Network) are used to establish UV spectrum's chemical computation method mesuring COD in waste water.COD in simulater waste water and real waste water is forecasted in this paper.LmnerPf(V2.3) is used in this paper to finish the modeling and correlater mathematical operation.Collection composed of 36 samples is set up a BP-ANN model, which is used to forecast simulated waster water and get good comeout and the relavive error is less than 5%.The model can't meet the analytical demand when used to forecast the real water and it mainly because that the composition of training collection is quite different from the complex real water. Suggestions are made aimed at and analysis of waste water,the composition of training samples,the lack of BP-ANN model and the improvement of model,etc in this paper.This paper provides references of account for further research in determination of COD by BP artificial neural network and UV absorption spectrum.