Research On The Method Of Detecting Dissolved Oxygen Concentration In Water Based On Machine Vision

At present,there are equipment that can monitor the river water quality in real time at home and abroad,but the price of such water quality monitoring equipment is expensive and the maintenance cost is high.Therefore,the testing of river water quality is still based on laboratory testing methods.However,the operation process of laboratory testing methods is generally complex,and the water quality cannot be obtained in real time.A new idea of detecting dissolved oxygen concentration in water by using machine vision technology based on the image of water sample solution was proposed,which based on the analysis of the existing water quality monitoring technology and machine vision detection technology at home and abroad,and combined iodimetry.A soft measurement model including chemical reaction operation module,image acquisition module,image processing module and data processing module was designed to explore the mapping relationship between the color characteristic value of water sample solution and the concentration of dissolved oxygen.(1)The method of water sample solution configuration was designed based on iodine quantity method,and the experimental parameters were optimized by experiment comparison.Experimental parameters include laboratory temperature,chemical reagent concentration,dilution ratio of water sample and sampling speed.The selection of each experimental parameter value is based on the parameter variation interval provided by three dominant factors such as iodine quantity method,chemical experimental device and physical experimental device,and obtained by adding part of the preliminary experiment,finally a linear model reflecting the linear relationship between chroma value and dissolved oxygen concentration is obtained.(2)The image acquisition device is composed of stabilized voltage light source,camera,darkroom and computer,and the most appropriate spacing is determined by adjusting the distance between camera,colorimetric dish and light source and comparing the effect of the acquired image,so as to complete the adjustment of darkroom structure.Image preprocessing is carried out on PC,and the image is processed to reduce the error of color eigenvalue.(3)A data processing model was established based on the relationship between the color characteristic value of water sample image and the dissolved oxygen concentration of water sample solution by using artificial neural network.In the neural network model,the color characteristic value of water sample image was used as the input of neural network,and the dissolved oxygen concentration of water sample solution was used as the output of neural network.The BP neural network established the nonlinear mapping relation between the color characteristic value of water sample solution image and the dissolved oxygen concentration of water sample solution.By comparing the predicted value of the neural network with the target value,the accuracy of the neural network is above 96%.Finally,the BP neural network model is compared with the linear model.