CN-121981609-A - Industrial wastewater quality discriminating system based on multi-mode sensing

CN121981609ACN 121981609 ACN121981609 ACN 121981609ACN-121981609-A

Abstract

The invention discloses an industrial wastewater quality judging system based on multi-mode sensing, which comprises the steps of collecting multi-mode sensing data to generate an online data stream, executing sliding time window slicing, time alignment and generating a quality labeling time window data set, generating an initial mode evidence budget set based on dynamic mode dispatching, calculating cross-mode conflict strength and forming a conflict labeling evidence set, generating a dispatching correction factor based on conflict discount fusion, introducing the dispatching correction factor to update mode dispatching parameters and generating a fusion evidence representation, and outputting an online water quality judging record based on the fusion evidence representation. The invention realizes multi-mode evidence scheduling and conflict discount closed-loop updating, and improves the fusion stability and result consistency of industrial wastewater on-line water quality discrimination.

Inventors

SONG RUIMING
BIAN JUNPENG
Jiao Jianqi
MA BINQIANG
LIU LING

Assignees

山西华普检测技术有限公司

Dates

Publication Date: 20260505
Application Date: 20260126

Claims (8)

1. The industrial wastewater quality judging system based on multi-mode perception is characterized by comprising the following modules: The multi-mode acquisition module acquires multi-mode data at the industrial wastewater discharge port, adds a timestamp mark to the data and writes the timestamp mark into the online cache queue to obtain multi-mode online data flow; The time window labeling module is used for executing sliding time window slicing processing on the multi-modal online data stream and completing time alignment, calculating the noise level, the drift degree, the deletion proportion and the saturation proportion of the multi-modal data in each sliding time window and generating quality labeling information to obtain a quality labeling time window data set; the scheduling budget module inputs Dynamic Modality Scheduling the quality annotation time window data set into a processing flow, generates an initial modal scheduling parameter set according to the quality annotation information, and generates an initial modal evidence budget set based on the initial modal scheduling parameter set; the conflict calculation module is used for calculating modal evidence distribution differences based on the initial modal evidence budget set, generating a cross-modal conflict intensity set, and associating the cross-modal conflict intensity set with the initial modal evidence budget set to obtain a conflict labeling evidence set; the discount correction module inputs Discounting Belief Fusion the conflict labeling evidence set into the processing flow, introduces a conflict evidence discount mechanism and generates a scheduling correction factor set according to the cross-modal conflict strength set; The scheduling updating module is used for introducing the scheduling correction factor set into a scheduling parameter generating mechanism of Dynamic Modality Scheduling processing flow, executing structural updating on the initial modal scheduling parameter set, and regenerating an updated modal evidence budget set based on the updated modal scheduling parameter set after the structural updating is completed to obtain a fusion evidence representation; And the judging and storing module is used for inputting the fused evidence representation into a water quality judging process after the fused evidence representation is generated, outputting an online water quality judging result of the industrial wastewater discharge port, and storing the online water quality judging result and the updated modal evidence budget set in a correlated manner to obtain an online judging record.
2. The industrial wastewater quality discrimination system based on multi-modal sensing according to claim 1, wherein the multi-modal collection module specifically comprises: Electrochemical sensing data, optical sensing data and physical parameter sensing data are collected at an industrial wastewater discharge port to form a multi-mode original data sequence; generating a corresponding time stamp mark for each piece of data in the multi-mode original data sequence, and completing time stamp writing to obtain a multi-mode data sequence with the time stamp mark; performing time stamp integrity check on the multi-mode data sequence with the time stamp mark, deleting the time stamp missing data, and obtaining a time stamp complete data sequence; And writing the complete data sequence of the time stamp into the online cache queue according to the time sequence, generating the data sequence of the online cache queue, and outputting data from the data sequence of the online cache queue according to the time stamp sequence to obtain the multi-mode online data stream.
3. The industrial wastewater quality discrimination system based on multi-modal sensing according to claim 1, wherein the time window labeling module specifically comprises: executing time window slicing processing according to sliding time window parameters from the multi-modal online data stream to obtain a multi-modal time window data sequence; Establishing a cross-modal time index relation in the multi-modal time window data sequence according to the timestamp mark to obtain a time pair Ji Suoyin table; performing time alignment processing on different mode data in the same sliding time window based on the time alignment index table to generate a multi-mode alignment time window data set; Calculating the noise level and the drift degree of multi-modal data aiming at the multi-modal alignment time window data set, and generating a noise drift annotation information set; And calculating the deletion proportion and saturation proportion of the multi-modal data aiming at the multi-modal alignment time window data set, and combining the multi-modal data with the noise drift annotation information set to generate the quality annotation time window data set.
4. The industrial wastewater quality discrimination system based on multi-modal awareness according to claim 1, wherein the scheduling budget module specifically comprises: Based on the quality labeling time window data set, respectively extracting the noise level, the drift degree, the missing proportion and the saturation proportion corresponding to the electrochemical sensing data in each sliding time window, extracting the noise level, the drift degree, the missing proportion and the saturation proportion corresponding to the optical sensing data, and extracting the noise level, the drift degree, the missing proportion and the saturation proportion corresponding to the physical parameter sensing data to obtain a sliding time window quality labeling information set; Inputting Dynamic Modality Scheduling a sliding time window quality labeling information group into a processing flow, performing uniform processing on each mode quality labeling information, and outputting a mode quality labeling information sequence with uniform dimension; In Dynamic Modality Scheduling processing flow, based on the modal quality labeling information sequence, performing comparison operation on different modalities in the same sliding time window, generating modal ordering information and writing the modal ordering information into an initial modal scheduling parameter set to obtain a sliding time window initial modal scheduling parameter set; In Dynamic Modality Scheduling processing flow, executing time sequence consistency constraint update to the initial modal scheduling parameter set of the sliding time window and the initial modal scheduling parameter set of the previous sliding time window to obtain an initial modal scheduling parameter set after time sequence update; In Dynamic Modality Scheduling processing flow, performing evidence budget allocation operation on electrochemical sensing data, optical sensing data and physical parameter sensing data in the same sliding time window based on the initial modal scheduling parameter set after time sequence updating to generate an initial modal evidence budget set; And executing sliding time window index binding processing on the initial modal evidence budget set, and establishing association between a binding result and the quality labeling time window data set to obtain the initial modal evidence budget set with the sliding time window index.
5. The industrial wastewater quality discrimination system based on multi-modal awareness according to claim 1, wherein the conflict calculation module specifically comprises: based on the initial modal evidence budget set, respectively extracting initial modal evidence budgets corresponding to electrochemical sensing data, initial modal evidence budgets corresponding to optical sensing data and initial modal evidence budgets corresponding to physical parameter sensing data in each sliding time window to obtain a sliding time window modal evidence budget group; Performing evidence distribution normalization processing on initial modal evidence budgets of different modalities in the sliding time window modal evidence budget group to generate a normalized modal evidence budget group; performing difference degree calculation on evidence distribution differences between electrochemical sensing data and optical sensing data in a normalized modal evidence budget group, performing difference degree calculation on evidence distribution differences between electrochemical sensing data and physical parameter sensing data, and performing difference degree calculation on evidence distribution differences between the optical sensing data and the physical parameter sensing data to obtain a modal pair difference degree set; Generating a cross-modal conflict intensity set based on the modal pair difference degree set, and binding the cross-modal conflict intensity set according to the sliding time window index to obtain the cross-modal conflict intensity set with the sliding time window index; and establishing association between the cross-modal conflict intensity set with the sliding time window index and the sliding time window modal evidence budget set to obtain a conflict labeling evidence set.
6. The industrial wastewater quality discrimination system according to claim 1, wherein the discount modification module specifically comprises: Based on conflict labeling evidence sets, respectively extracting initial modal evidence budget corresponding to electrochemical sensing data, initial modal evidence budget corresponding to optical sensing data and initial modal evidence budget corresponding to physical parameter sensing data in each sliding time window, extracting cross-modal conflict intensity between the electrochemical sensing data and the optical sensing data, extracting cross-modal conflict intensity between the electrochemical sensing data and the physical parameter sensing data, extracting cross-modal conflict intensity between the optical sensing data and the physical parameter sensing data, and obtaining a sliding time window conflict evidence input group; inputting Discounting Belief Fusion the sliding time window conflict evidence input group into a processing flow, and executing conflict aggregation operation on cross-modal conflict intensity in the sliding time window conflict evidence input group to generate a modal conflict aggregation value set; In Discounting Belief Fusion processing flow, executing discount coefficient generation operation on the initial modal evidence budget based on the modal conflict aggregate value set to obtain a discount coefficient set; in Discounting Belief Fusion process flow, performing evidence discount update operation on the initial modal evidence budget based on the discount coefficient set to generate a discount modal evidence budget set; in the Discounting Belief Fusion process flow, a scheduling correction factor set is generated based on the discount modality evidence budget set, and a sliding time window index binding process is performed on the scheduling correction factor set, so that the scheduling correction factor set with the sliding time window index is obtained.
7. The industrial wastewater quality discrimination system based on multi-modal awareness according to claim 1, wherein the dispatch updating module specifically comprises: extracting a scheduling correction factor group corresponding to the current sliding time window according to the sliding time window index based on the scheduling correction factor set with the sliding time window index to obtain the sliding time window scheduling correction factor group; Reading a sliding time window initial mode scheduling parameter set in Dynamic Modality Scheduling processing flow, establishing a parameter corresponding relation between a sliding time window scheduling correction factor group and the sliding time window initial mode scheduling parameter set, and obtaining a scheduling parameter correction input group; In a scheduling parameter generating mechanism of Dynamic Modality Scheduling processing flow, executing structural updating operation on the scheduling parameter correction input group to generate a mode scheduling parameter set after structural updating; Executing modal scheduling parameter recalculation operation based on the structured updated modal scheduling parameter set in Dynamic Modality Scheduling processing flow to generate an updated modal scheduling parameter set; Performing evidence budget allocation operation on electrochemical sensing data, optical sensing data and physical parameter sensing data in the same sliding time window based on the updated modal scheduling parameter set in Dynamic Modality Scheduling processing flow to generate an updated modal evidence budget set; a sliding time window index binding process is performed on the updated modal evidence budget set and a fused evidence representation is generated based on the updated modal evidence budget set.
8. The industrial wastewater quality discriminating system based on multi-modal sensing according to claim 1, wherein the discriminating storage module specifically comprises: based on the fusion evidence representation, extracting the fusion evidence representation corresponding to the current sliding time window according to the sliding time window index to obtain a sliding time window fusion evidence input group; Inputting the sliding time window fusion evidence input group into a water quality judging process, and executing category judging operation on the sliding time window fusion evidence input group to generate a sliding time window water quality judging result; establishing an association relation between the water quality judging result of the sliding time window and the updated modal evidence budget set corresponding to the sliding time window, and generating sliding time window judging association data; Performing time sequence index binding processing on the sliding time window discrimination associated data to generate a discrimination associated sequence with a sliding time window index; and writing the discrimination association sequence into an online storage queue to obtain an online discrimination record.

Description

Industrial wastewater quality discriminating system based on multi-mode sensing Technical Field The invention relates to the technical field of industrial wastewater discharge port on-line water quality monitoring and multi-mode sensing information fusion and discrimination, in particular to an industrial wastewater quality discrimination system based on multi-mode sensing. Background The on-line water quality monitoring of the industrial wastewater discharge port is an important technical link of environmental supervision and pollution prevention and control, and the application scene covers the supervision of the industrial park discharge port, the effluent inspection of a sewage treatment plant, the self-inspection of enterprise discharge and the like. The existing engineering system adopts an electrochemical sensor to obtain indexes such as pH value, oxidation-reduction potential, dissolved oxygen, conductivity and the like, adopts an optical sensor to obtain indexes such as turbidity, chromaticity, ultraviolet absorption and the like, and combines physical parameter sensing data such as temperature, flow, pressure and the like to form an on-line monitoring data stream. In order to meet the continuous monitoring requirement, the system generally adopts timestamp recording, online caching and sliding time window processing to realize data synchronization and real-time calculation of cross-equipment. In the aspect of multi-source data fusion, the existing scheme performs time alignment on multi-mode data according to a sliding time window, and then outputs a water quality category or index discrimination result through weighted average, rule threshold, statistical regression or a fusion strategy based on probability. The partial scheme introduces evidence representation and confidence allocation, converts each mode observation into evidence budget and fuses the evidence budget to cope with deviation caused by single sensor abnormality. The other scheme identifies cross-mode conflict through consistency check or similarity calculation, and gives punishment weight or discount coefficient to the modes participating in fusion according to the conflict degree so as to reduce interference of conflict information to the judging process. The working condition of the industrial wastewater discharge port is frequently changed, the quality problems such as noise fluctuation, baseline drift, missing sampling, range saturation and the like are easy to occur when the sensor is operated on line for a long time, and the quality degradation of different modes has asynchronism and stage property. The existing fusion flow is often carried out under the constraint of fixed weight or fixed rule, dynamic scheduling of the mode contribution is difficult to carry out in the sliding time window scale, and when the cross-mode conflict is enhanced, a tightly-coupled linkage updating mechanism is lacking between evidence budget and discount strategies, so that instability of fusion characteristics occurs along with time window changes, and further the continuous consistency and traceability of follow-up discrimination records are affected. Therefore, how to provide an industrial wastewater quality judging system based on multi-modal sensing is a problem to be solved by those skilled in the art. Disclosure of Invention Aiming at the problem that fusion characteristics are unstable in the aspect of industrial wastewater multi-mode sensing information fusion modeling in the prior art, the invention provides a multi-mode online data stream with a timestamp, which is obtained by a multi-mode acquisition module, a time window marking module is used for finishing sliding time window slicing, time alignment and generating quality marking time window data sets of noise level, drift degree, deletion proportion and saturation proportion, a scheduling budget module is used for generating an initial mode scheduling parameter set and distributing an initial mode evidence budget set in Dynamic Modality Scheduling processing flow, a conflict calculation module is used for calculating mode evidence distribution difference and generating a cross-mode conflict intensity set to form a conflict marking evidence set, a discount correction module is used for executing discount on conflict evidence in Discounting Belief Fusion processing flow to generate a scheduling correction factor set, a scheduling update module is used for introducing the scheduling correction factor set into a scheduling parameter generation mechanism to generate fusion evidence representation, and finally a judging storage module is used for outputting online water quality judging result and storing in a correlation mode judging record. According to the embodiment of the invention, the industrial wastewater quality judging system based on multi-mode sensing comprises the following modules: The multi-mode acquisition module acquires multi-mode data at the industrial wastewater discha