Prediction of settlement velocity of sludge area based on image analysis

Abstract

Abstract The sedimentation performance of sludge affects the operation of the entire treatment process. It is important to measure the zone settling velocity accuracy and high efficiency which can reflect the sedimentation performance of sludge. Traditional methods often rely entirely on manual implementation. Long hours of manual measurement work brings physical strain to the operator, low efficiency of the entire measurement work and a large error in the measurement results. Therefore, the method combining image processing and neural network to realize automatic recognition of the scale value of sludge water interface in sludge sedimentation video is proposed. Based on the recognized scale value, overall sedimentation change curve is drew, and the sludge sedimentation rate is calculated. Then, the activated sludge was examined by phase contrast microscope to obtain sludge image. The flocs, filamentous bacteria and other targets in the image are segmented with Labelme, the characteristics of flocs and sedimentation rate of the previous day are taken as input of deep stochastic configuration network, and the calculated sludge sedimentation rate is output for predicting layered sedimentation rate. Through experimental analysis, the prediction error is within a reasonable range. The intelligent prediction of sludge regional sedimentation velocity is realized.