A Multivariate Robust Soft Sensing Method for Water Quality Indicators of Wastewater Treatment

Abstract

The invention provides a multivariate robust soft-sensing method for sewage treatment effluent water quality indicators, and relates to the technical field of sewage treatment automation control. This method is based on the parameters obtained by real-time measurement of conventional detection equipment on the industrial site as the input data of the model; it establishes a random weight neural network model that can simultaneously perform multivariate dynamic prediction of the main parameters that measure the quality of sewage treatment effluent, and simultaneously realizes BOD content, The robust soft measurement of COD content and TSS content sewage quality parameters comprehensively describes the sewage water quality parameters and avoids the lag of offline testing and the uncertainty caused by manual operation. The present invention utilizes sparse partial least squares and Schweppe-type generalized M estimation at the same time, eliminates the influence of multicollinearity on modeling, reduces the adverse influence of outliers and leverage points in data on modeling, and also achieves the accuracy of variable selection The purpose is to more accurately give the estimated value of the multivariate sewage treatment effluent water quality parameters in the specified dynamic time interval.

Description

technical field [0001] The invention relates to the technical field of automatic control of sewage treatment, in particular to a multivariate robust soft-sensing method for water quality indicators of sewage treatment effluent. Background technique [0002] In recent years, the demand for fresh water resources in the industrial society and daily life has been increasing day by day, and the damage to the water environment has become more and more serious. How to efficiently treat discharged sewage and realize the sustainable utilization and virtuous cycle of fresh water resources is an eternal topic in the development of modernization in the world today. Sewage treatment is a very complex non-linear dynamic process in which the sewage generated in daily production and life is subjected to a series of treatments to achieve discharge targets, and it involves the sustainable development of the environment and resources. [0003] The activated sludge process is the most widely u...

AI Technical Summary

Problems solved by technology

[0006] The method reported in the above-mentioned patents and similar methods related to other relevant literatures are only for forecasting and soft measurement of a single sewage effluent water quality parameter (such as BOD content, COD content, etc.), but fail to predict the main parameters that characterize sewage water quality, namely BOD content, COD content, etc. Content and TSS content are multivariately forecasted at the same time, so it cannot fully reflect the level of sewage water quality, and the practicability is poor

Moreover, the sewage data is high-dimensional data and there is multi-collinearity in the data, and the soft measurement for the abnormal data needs to be processed; in addition, in the actual sewage treatment process, the environment is harsh, the failure of the detection instrument and other devices and other abnormalities Effect of noise, outliers included in the measured data

These methods mainly consider the soft measurement of sewage water quality parameters under ideal conditions, and their robustness is poor. When the modeling data contains outliers, these methods cannot suppress the interference of outliers and predict the sewage quality parameters more accurately.

To sum up, there is currently no multivariate robust soft-sensing method for sewage quality parameters (BOD content, COD content, and TSS content) in the sewage treatment process at home and abroad.

Images