The Prediction of Chemical Oxygen Demand (Cod) In Waste Water by Uv-Visible Absorption Spectrum – Neural Network

Yasunobu Kato²*, Tetsu Kumagai¹, Hiroshi Nishioka¹ and Yosohiro Sugie²

¹The Faculty of Engineering, Himeji Institute of Technoloy 2167 Syosya Himeji-shi, Hyogo 670-2201 (Japan) ²The School of Humanity for Environment Policy and Technology Himeji Institute of Technology 1-1-12 Shinzaike-honcho Himeji-shi, Hyogo 670-0092 (Japan)

A rapid measurement of Chemical Oxygen Demand (COD) was studied based on the UV absorption spectra and the Neural Network. The COD calibration models for the mixture of some organic compounds were made from the ultra-violet absorption spectra and the visible absorption spectrum. Sample solutions for calibration were prepared in order to obtain required concentration (1~10ppm) for the experiment. Firstly, sixty-one samples for database, ten samples for test samples, the samples added the inorganic compound with the absorption in equal ultraviolet region were prepared. Next as an interference, sodium nitrate, sodium nitrite, sodium chloride, and sodium sulfate were mixed at various proportions so that COD may vary. Sixty samples for database, ten samples for test samples. To simulate the contamination of the haze, 1~100ppm of turbidity by kaolin mixture, sixty samples for database, ten samples for test samples, were prepared. These samples were measured between 200nm and 300nm by spectrophotometer (HITACHI U-3500) and they were saved as an ASCII mode. It was assumed wastewater include pigment 1-30ppm of concentration by pigment, sixty samples for database, ten samples for test samples, were prepared. Absorption of these samples was measured between 200nm and 700nm. The COD values of ten test samples were predicted by the calibration database calculated in the Neural Network. The COD values of samples were then predicted with accuracy.

Chemical oxygen demand; wastewater; neural network