Microscopic Parameters Response To Activated Sludge Settling Ability Based On Image Analysis Technology

Activated sludge settling ability is considered of critical importance to the stability of activated sludge system. Up to date, the traditional method of measur ing Sludge Volume Index(SVI) is widely used, which is time-consuming, difficult to get rid of the manual operation and hard to realize online automatic detection. Therefore, this paper aims at realizing activated sludge settling ability characterization, discrimination and soft-measurement with the microscopic image analysis technology, and providing a technology for online detecting of activated sludge settling ability.With the conception of microscopic concentration, two parameters named Microscopic Volume Index(MVI) and Sludge Micro-floc Index(SMI) are proposed to characterize activated sludge settling ability. The internal relation between new parameters and SVI are discussed and experimentals have been tested. As results, MVI has a significant correlation with SVI in terms of domestic wastewater and coking wastewater. Thereafter, micro flocs have an obvious influence on sludge settling ability on the condition of large floc proportion is low in quantity, and SMI has significant correlation with SVI.Three comprehensive indexes named size factor, morphology factor and concentration factor can be achieved from 20 microscopic parameters through the Principal Component Analysis(PCA) method. Based on the indexes, a discriminant model is established to make Discriminant Analysis(DA) on the sludge setting ability. The discriminant model can provide a simple result of "normal" or "bulking". To learn the reliability of the model, monitoring data from two wastewater treatment plants during three months are used to test and the accuracy rates is 80.6%.With the method of Partial Least Squares(PLS) regression, three components are extracted from 20 microscopic parameters and regarded as the input layer nodes of radial basis function network. The output layer of the activated sludge settling ability discriminant model is "normal" and "bulking". The accuracy rate of this model is 84.6%.Five comprehensive indexes can be extracted from 27 wastewater treatment plant operation parameters, which named size factor, morphology factor, concentration factor, water quality factor and flow factor. These indexes as the input layer nodes of radial basis function network and a soft-measurement model of activated sludge settling ability is established. The model response fast, and has good generalization performance.