Soft measurement modeling method of A2O municipal sewage treatment process based on constraint theory

Abstract

The invention discloses a soft measurement modeling method of an A2O municipal sewage treatment process based on a constraint theory. The method comprises the following steps of establishing an ASM2d reaction model and two settling tank models; executing the sewage steady simulation for the municipal sewage of the A2O process by utilizing steady data, carrying out the constraint planning according to each reaction process and parameter constraint relation between every two components, and acquiring a steady simulation result; carrying out the time integration for the discharged water quality parameter concentration equation of each reaction tank, establishing the dynamic discharged water concentration equation set of each tank, carrying out the nonlinear planning solution for the dynamic discharged water concentration equation based on the constraint plan to acquire a dynamic simulation result, comparing the dynamic simulation result with the real discharged water result, adjusting the parameter and correcting the model of each tank until the dynamic simulation result is consistent with the data of the real discharged water result. Through the constraint plan of ASM2d mechanism model parameter in the mechanism-constraint A2O process, the precision of solution speed of the soft measurement model of each reaction tank can be improved.

Description

technical field [0001] The invention relates to the field of urban sewage index detection, in particular to an A 2 O Soft-sensing modeling method for urban sewage treatment process. Background technique [0002] At present, some important effluent water quality indicators in the urban sewage treatment process lack reliable online instruments, and can only be measured by manual experiments. The soft sensor technology is used to predict the effluent water quality indicators in real time by building a soft sensor model, that is, from known process parameters. Estimate values ​​of target parameters that are not measurable online. Most of the existing technical solutions use methods based on multiple linear regression (MLR) and artificial neural networks, such as BP neural network, Elman neural network, process neural network, etc., sample data interpolation and multi-step memory. [0003] for A 2 O Sewage treatment process, focusing on the BOD of effluent quality which is dif...

AI Technical Summary

Problems solved by technology

[0013] 1) The constraint relationship between variables is not fully considered;

[0014] 2) The mean square error of regression and the mean square error of prediction evaluation are both large;

[0015] 3) Most of the auxiliary variables are water inflow parameters, and the measurement has hysteresis, and the use time is long and uneven. There are few auxiliary variables, and the process is nonlinear

Images