CN-121990660-A - Dynamic calibration and multi-agent cooperative control method for intelligent mine water dosing system

CN121990660ACN 121990660 ACN121990660 ACN 121990660ACN-121990660-A

Abstract

The invention relates to the field of sewage treatment, in particular to a dynamic calibration and multi-agent cooperative control method of an intelligent mine water dosing system, which comprises the steps of S1 collecting multi-source data, S2 extracting characteristic alum images, S3 injecting standard liquid into a turbidity meter measuring pool to obtain a slope for first calibration, combining feed flow and characteristic alum images to carry out second calibration through a back propagation neural network soft measuring model, S4 inputting flow, concentration, turbidity and images into a Gaussian mixture model to judge working conditions, feeding the flow, concentration, turbidity and images into a time sequence convolution network-back propagation neural network mixture model to obtain total basic dosing quantity, S5 calculating total compensation dosing quantity and adjusting the total basic dosing quantity through a fuzzy proportion-integral-derivative algorithm according to turbidity and target deviation and ideal floccule characteristics and actual deviation, S6 constructing a 9-grid reference proportion lookup table to obtain a reference dosing ratio, outputting a correction coefficient according to a hard correction rule (cost priority, extreme dose protection and oscillation inhibition), and finally calculating the total basic dosing quantity of a main flocculant and a coagulant aid.

Inventors

MA YONGZHUANG
QIAO GUOWEI
XU DONGQUAN

Assignees

国家能源集团宁夏煤业有限责任公司

Dates

Publication Date: 20260508
Application Date: 20251230

Claims (10)

1. The dynamic calibration and multi-medicament cooperative control method of the intelligent mine water dosing system is characterized by comprising the following steps of: step S1, multi-source data are collected; S2, extracting characteristics of the alum blossom images in the multi-source data to obtain characteristic alum blossom images; S3, performing first calibration on the online turbidity meter through a hardware calibration method, acquiring turbidity confidence according to the feeding flow in the multi-source data and the characteristic alum blossom image, and performing second calibration on the turbidity of the overflow water according to the turbidity confidence to obtain the turbidity of the overflow water after calibration; s4, judging real-time working conditions according to the multi-source data, the characteristic alum blossom images and the calibrated overflow water turbidity, and acquiring total basic dosage according to the multi-source data and the real-time working conditions; s5, calculating the total compensation dosage according to the calibrated turbidity of the overflow water, and adjusting the total basic dosage according to the total compensation dosage; and S6, constructing a reference proportioning table, acquiring a reference main flocculant adding ratio and a reference coagulant aid adding ratio according to the reference proportioning table and a real-time working condition, outputting a correction coefficient according to a hard correction rule, calculating a main flocculant adding amount and a coagulant aid adding amount according to the correction coefficient, the total basic adding amount, the reference main flocculant adding ratio and the reference coagulant aid adding ratio, and outputting the main flocculant adding amount and the coagulant aid adding amount as a medicament cooperative control strategy.
2. The method for dynamically calibrating and cooperatively controlling a plurality of reagents in an intelligent mine water dosing system according to claim 1, wherein the step S2 of extracting features from alum images in multi-source data comprises: S21, performing noise reduction treatment on the alum blossom images in the multi-source data to obtain alum blossom images after noise reduction; S22, performing enhancement treatment on the alum blossom image after noise reduction to obtain an enhanced alum blossom image; And S23, extracting quantitative characteristics of the enhanced alum blossom images by using a computer vision algorithm to obtain characteristic alum blossom images, wherein the quantitative characteristics comprise average floc particle size, uniformity of particle size distribution and floc profile definition.
3. The method for dynamically calibrating and cooperatively controlling a plurality of reagents in a mine water intelligent dosing system according to claim 1, wherein the step S3 of performing a first calibration on an online turbidity meter by a hardware calibration method comprises: Step S311, standard liquid with preset concentration is injected into a turbidity meter measuring tank, and the rising slope Zc of the measured turbidity in five minutes of an online turbidity meter is obtained; Step S312, comparing the rising slope Zc with a preset rising slope Z0, judging the response condition of the online turbidimeter according to the comparison result, and performing a first calibration on the online turbidimeter according to the judgment result, wherein: When Zc is more than or equal to Z0, judging that the response condition of the online turbidimeter is quick in response, and not performing first calibration on the online turbidimeter; When Zc < Z0, judging that the response condition of the online turbidimeter is slow in response, and performing first calibration on the online turbidimeter, wherein the first calibration comprises triggering an online turbidimeter fault alarm and repairing the online turbidimeter by staff.
4. The method for dynamically calibrating and cooperatively controlling a mine water intelligent dosing system according to claim 3, wherein in the step S3, the turbidity confidence is obtained according to the feeding flow rate in the multi-source data and the characteristic alum blossom image, and the second calibrating the turbidity of the overflow water according to the turbidity confidence comprises: inputting the feeding flow and the characteristic alum flow images into a preset turbidity soft measurement model, obtaining turbidity confidence coefficient T_virtual output by the preset turbidity soft measurement model, calculating the calibrated overflow water turbidity T_control according to the overflow water turbidity T_sensor, the turbidity confidence coefficient T_virtual, a first calibration weight coefficient w1 and a second calibration weight coefficient w2, and setting T_control=w1×T_sensor+w2×T_virtual.
5. The method for dynamically calibrating and cooperatively controlling a plurality of medicaments for an intelligent dosing system for mine water according to claim 4, wherein the preset turbidity soft measurement model is an on-line fused lightweight counter-propagating neural network; The preset turbidity soft measurement model comprises a turbidity input layer, a turbidity hiding layer and a turbidity output layer; The turbidity input layer comprises a feeding flow node, a feeding concentration node, a floc average particle diameter node, a mud layer height node and an artificial assay value node; The turbidity hiding layer is of a single-layer structure and is provided with 10 hiding nodes, and the activation function of the turbidity hiding layer is a Tanh activation function; the node number of the output layer is 1, and the turbidity confidence is directly mapped.
6. The method for dynamically calibrating and cooperatively controlling a mine water intelligent dosing system according to claim 5, wherein the step S4 of determining the real-time conditions according to the multisource data, the characteristic alum blossom images and the calibrated turbidity of the overflow water comprises: inputting the feeding flow, feeding concentration, calibrated overflow water turbidity, average floc particle size, uniformity of particle size distribution and definition of floc outline into a pre-trained working condition judgment model, and obtaining a real-time working condition output by the pre-trained working condition judgment model; the real-time working conditions comprise a high-load working condition, a normal-load working condition and a low-load working condition.
7. The method for dynamically calibrating and cooperatively controlling a plurality of reagents in a mine water intelligent dosing system according to claim 6, wherein in step S4, the total basic dosing amount is further obtained according to the multi-source data and the real-time working conditions, and the method comprises the following steps: and inputting the feeding flow, the feeding concentration and the real-time working condition into a pre-trained feedforward predictor, and obtaining the total basic dosage output by the feedforward predictor.
8. The method for dynamically calibrating and cooperatively controlling a plurality of agents in a mine water intelligent dosing system according to claim 7, wherein the step S5 of calculating a total compensation dosage according to the turbidity of the overflow water after calibration and adjusting a total basic dosage according to the total compensation dosage comprises: The method comprises the steps of obtaining target effluent turbidity, ideal floc average particle size and ideal particle size distribution uniformity, calculating turbidity deviation according to the target effluent turbidity and calibrated overflow turbidity, setting turbidity deviation=target effluent turbidity-calibrated overflow turbidity, calculating two-dimensional deviation according to the ideal floc average particle size, ideal particle size distribution uniformity, floc average particle size and particle size distribution uniformity, setting two-dimensional deviation= [ ideal particle size distribution uniformity-floc average particle size, ideal particle size distribution uniformity-particle size distribution uniformity ], inputting the turbidity deviation and the two-dimensional deviation into a preset multivariable feedback controller, obtaining total compensation dosage output by the preset multivariable feedback controller, adjusting the total basic dosage according to the total compensation dosage, obtaining adjusted total basic dosage, setting the value of the adjusted total basic dosage=total basic dosage+total compensation dosage, and replacing the value of the total basic dosage with the value of the adjusted total basic dosage.
9. The method for dynamically calibrating and cooperatively controlling a plurality of agents in a mine water intelligent dosing system according to claim 8, wherein the step S6 is characterized by constructing a reference proportioning look-up table, and acquiring a reference main flocculant dosing ratio and a reference coagulant aid dosing ratio according to the reference proportioning look-up table and a real-time working condition, and specifically comprises the following steps: And constructing a reference proportioning lookup table according to the dosing proportion corresponding to the high load working condition, the normal load working condition, the low load working condition and the historical working condition to obtain a total basic dosing amount and a reference proportioning lookup table of 9 grids of three-gear dosing amount of the real-time working condition and three working conditions, inputting the real-time working condition into the reference proportioning lookup table, and obtaining a reference main flocculant dosing ratio and a reference coagulant aid dosing ratio output by the reference proportioning lookup table.
10. The method for dynamically calibrating and controlling the intelligent dosing system for mine water in cooperation with multiple agents according to claim 9, wherein the step S6 outputs the correction coefficient according to the hard correction rule and the multi-source data, calculates the main flocculant dosing amount and the coagulant aid dosing amount according to the correction coefficient, the total basic dosing amount, the basic main flocculant dosing ratio and the basic coagulant aid dosing ratio, and outputs the main flocculant dosing amount and the coagulant aid dosing amount as a agent cooperative control strategy, and specifically comprises: the hard correction rule comprises a cost priority rule, an extreme drug quantity protection rule and an oscillation suppression rule; the cost priority rule includes: Calculating a price ratio A3 according to a main flocculant price A1 and a coagulant aid price A2 in multi-source data, setting A3=A1/A2, and outputting a correction coefficient J according to the price ratio A2, wherein: when A3>3, the correction coefficient j= -0.08 is output; When 1.5< A3 is less than or equal to 3, outputting a correction coefficient J=0; The extreme dose protection rules include: Comparing the total basic dosage B with a preset dosage B0, outputting a correction coefficient J according to the comparison result, and setting b0=70 ppm, wherein: When B is less than or equal to B0, outputting a correction coefficient J=0; When B > B0, outputting a correction coefficient j= -9.02; The oscillation suppression rule includes: Outputting a correction coefficient J according to a main flocculant addition amount change value C of the last period in the multi-source data, wherein: when |c| >0.15, the correction coefficient j= -0.5×sign (C) is output; outputting a correction coefficient J=0 when the absolute value C is more than or equal to 0.15; And calculating the main flocculant addition amount P3 and the coagulant aid addition amount P4 according to the correction coefficient J, the total basic addition amount Q, the reference main flocculant addition ratio L1 and the reference coagulant aid addition ratio L2, and setting P3=Q× (L1+J), P4=Q× [1- (L1+J) ].

Description

Dynamic calibration and multi-agent cooperative control method for intelligent mine water dosing system Technical Field The invention relates to the field of sewage treatment, in particular to a dynamic calibration and multi-agent cooperative control method of an intelligent dosing system for mine water. Background The existing intelligent dosing technology for the slime water adopts a single floc feedback model to predict the dosage, but has three major bottlenecks that firstly, a core sensor such as an online turbidity meter and the like is easy to be polluted and scaled in complex water quality, so that monitoring data drift or failure is caused, an online calibration means is lacked, a system is in decision error due to 'perception distortion', secondly, a single model structure is insufficient in self-adaption to water inlet load step and water quality nonlinear change, response lag is large, long-term stable operation is difficult, thirdly, multi-agent dosing lacks a dynamic cooperative mechanism aiming at total cost optimization, and main flocculant and coagulant aid are independently controlled to manufacture medicine consumption waste. The problems of high cost of ton water medicament, high standard exceeding rate of effluent water quality and poor system robustness are caused, and a closed-loop control method integrating double calibration, double model self-adaption and economic rule cooperation is needed. CN114545985B discloses a dosing system and method based on floc feature monitoring and process feedback, and relates to the field of sewage treatment chemical dephosphorization and flocculation. The device comprises a water quality monitoring module, a floc particle size characteristic monitoring module, a digital-analog analysis module, a calculation module, a dosing control module and a dosing control module, wherein the water quality monitoring module is used for water quality monitoring of inflow water and outflow water, the floc particle size characteristic monitoring module is used for continuously monitoring the floc particle size characteristics of dephosphorization/flocculation in the flocculation process, the digital-analog analysis module is connected with the water quality monitoring module and the floc particle size characteristic monitoring module, the water quality monitoring module is used for collecting inflow feedforward according to inflow water quality parameters, collecting outflow water feedback according to outflow water quality parameters, feeding back according to dynamic change of the floc particle size characteristics, the three feedback processes are used for carrying out digital-analog analysis on dosing requirements, the calculation module is connected with the digital-analog analysis module for dosing self-adjustment optimization, and the dosing control module is connected with the calculation module for realizing real-time regulation dosing of dosing. The invention dynamically monitors the effect of the chemical dephosphorization/flocculation process in real time, and can effectively feed back the effect of the treatment process and the prejudgement control. However, the scheme still has the problems of inaccurate dosing, higher drug consumption and poor system robustness caused by lack of dual calibration of a sensor, insufficient self-adaptive capacity of a single model and no economic and synergistic mechanism of multiple medicaments. Disclosure of Invention Therefore, the invention provides a dynamic calibration and multi-agent cooperative control method for an intelligent mine water dosing system, which is used for solving the problems of inaccurate dosing, higher drug consumption and poor system robustness caused by lack of dual calibration of a sensor, insufficient self-adaptive capacity of a single model and no economic cooperative mechanism of multiple agents in the prior art. In order to achieve the above purpose, the invention provides a dynamic calibration and multi-medicament cooperative control method of an intelligent dosing system for mine water, which comprises the following steps: step S1, multi-source data are collected; S2, extracting characteristics of the alum blossom images in the multi-source data to obtain characteristic alum blossom images; S3, performing first calibration on the online turbidity meter through a hardware calibration method, acquiring turbidity confidence according to the feeding flow in the multi-source data and the characteristic alum blossom image, and performing second calibration on the turbidity of the overflow water according to the turbidity confidence to obtain the turbidity of the overflow water after calibration; s4, judging real-time working conditions according to the multi-source data, the characteristic alum blossom images and the calibrated overflow water turbidity, and acquiring total basic dosage according to the multi-source data and the real-time working conditions; s5, calculating the total compensati