Gas power plant circulating water and sewage treatment system and method

Abstract

The invention relates to the technical field of power plant water treatment intelligent diagnosis, and discloses a gas power plant circulating water and blow-down water treatment system and method. The method comprises the following steps: collecting and staged processing original monitoring data of a circulating water processing flow to form a processing stage data block with a label; and inputting the data blocks into a pollutant migration simulation network, dynamically simulating form transformation and concentration attenuation of total nitrogen, chemical oxygen demand, total phosphorus and suspended solids among continuous processing units, and generating a predicted pollutant migration trajectory map. The trajectory deviation degree is calculated by comparing the trajectory map with actual monitoring data, and abnormality diagnosis is triggered when the trajectory deviation degree exceeds a threshold value. The diagnosis program can identify specific treatment units and pollutant types which cause deviation and generate a diagnosis report. And searching a historical case library according to the diagnosis report, and matching to generate a targeted regulation and control scheme set. According to the invention, dynamic simulation and abnormal accurate traceability of the sewage treatment process are realized, and the predictability and the regulation and control accuracy of operation management and control are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent diagnostic technology for power plant water treatment, specifically to a circulating water wastewater treatment system and method for gas-fired power plants. Background Technology

[0002] In the treatment of circulating water wastewater from gas-fired power plants, conventional techniques primarily rely on setting up monitoring points at the inlet and outlet of each independent treatment unit to measure and record the concentrations of key pollutants such as total nitrogen, chemical oxygen demand, total phosphorus, and suspended solids at fixed points and discretely. This monitoring model treats the entire continuous treatment process as a series of isolated links, and its management logic is based on comparing the static thresholds of each monitoring point. When the data at a certain monitoring point exceeds the standard, it can usually only locate the anomaly at the outlet of the unit where that point is located, but cannot reveal the source and evolution process of the anomaly in the process.

[0003] The shortcomings of existing technologies lie in the lack of overall dynamic perception and simulation of the migration and transformation process of pollutants between continuous treatment units. When system fluctuations or malfunctions occur, operators find it difficult to distinguish whether the anomaly originates from the performance degradation of the current unit or from the transmission of impact loads from upstream units. This leads to delayed fault diagnosis, unclear root cause identification, and control measures that often target symptoms rather than the underlying cause, relying heavily on trial and error or comprehensive adjustments, resulting in low efficiency and the potential for secondary system fluctuations. A technological solution is needed that can comprehensively simulate the migration trajectory of pollutants and accurately trace and locate the source when anomalies occur, thereby improving the predictability and precision of the treatment process. Summary of the Invention

[0004] The purpose of this invention is to provide a wastewater treatment system and method for circulating water in gas-fired power plants to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method for treating circulating water wastewater from a gas-fired power plant, the method comprising:

[0006] Collect and aggregate raw monitoring sequences from the circulating water treatment process of a gas-fired power plant;

[0007] The original monitoring sequence is segmented and labeled in stages. The continuous data stream is divided into multiple independent processing stage data blocks according to the timestamp and processing unit identifier, and each processing stage data block is labeled with the processing unit type from which it originates.

[0008] The data blocks from the processing stage are input into a pollutant migration simulation network to simulate the speciation and concentration decay of pollutants between consecutive processing units, generating a pollutant migration trajectory map. The pollutant migration trajectory map records the predicted concentration change paths of total nitrogen, chemical oxygen demand, total phosphorus, and suspended solids between different processing units.

[0009] The pollutant migration trajectory map is compared with the actual pollutant concentration readings for the corresponding time window to calculate the trajectory deviation value. When the trajectory deviation value exceeds the preset deviation threshold, an anomaly diagnosis procedure for that time window is triggered.

[0010] The anomaly diagnosis procedure identifies the specific processing unit and type of contaminant that caused the trajectory deviation, and generates a diagnostic report that includes the anomaly unit identifier, the type of anomalous contaminant, and the degree of deviation.

[0011] Based on the diagnostic report, the historical control case database is queried to retrieve historical control records that match the abnormal unit identifier and abnormal pollutant type, and a set of control schemes is generated.

[0012] Preferably, the step of performing staged segmentation and labeling processing on the original monitoring sequence includes:

[0013] The original monitoring sequence includes periodically recorded readings of operating parameters and pollutant concentrations in denitrification tanks, secondary sedimentation tanks, flocculation sedimentation tanks, ozone advanced oxidation units, aerated biological filters, and fiber disc filters.

[0014] Read the timestamp information and data source channel identifier of the original monitoring sequence;

[0015] Based on the data source channel identifier, the data stream is separated into data sub-streams corresponding to the denitrification tank, the secondary sedimentation tank, the flocculation sedimentation tank, the ozone advanced oxidation unit, the aerated biological filter, and the fiber disc filter.

[0016] For each data substream, the time window length is set according to the standard hydraulic residence time of its corresponding processing unit. The data substream is then divided into multiple preliminary data segments by a fixed-length sliding window using this time window length.

[0017] Check the time span of each preliminary data segment. If the time span is incomplete, search for adjacent data points forward or backward to fill in the gaps and form a complete processing stage data block.

[0018] Based on the data source channel identifier, each complete processing stage data block is assigned a processing unit type label of its source. The processing unit type label includes denitrification tank stage, secondary sedimentation tank stage, flocculation sedimentation tank stage, ozone advanced oxidation stage, aerated biological filter stage, and fiber disc filter stage.

[0019] Preferably, the step of inputting the processing stage data block into the pollutant migration simulation network to simulate the speciation and concentration decay process of pollutants between consecutive processing units and generate a pollutant migration trajectory map includes:

[0020] Construct a directed network containing sequentially connected nodes, where each node represents a specific processing unit, and the directed edges between nodes represent the sewage flow direction and processing order;

[0021] Extract the pollutant concentration vector at the effluent end from the data block corresponding to the upstream treatment unit, and use it as the simulation input for the downstream treatment unit;

[0022] The processing unit response sub-model in the pollutant migration simulation network is invoked. The processing unit response sub-model selects the corresponding simulation logic according to the processing unit type label. The simulation logic simulates the input pollutant concentration vector based on the physicochemical and biological action principle of the unit and predicts the output pollutant concentration vector.

[0023] The simulated output of the upstream node is used as the simulated input of the downstream node and is passed sequentially along the edges of the directed network until the last node.

[0024] Record the pollutant concentration vector changes before and after each node's simulated treatment, and connect the concentration changes of all nodes in chronological order to draw a complete pollutant migration trajectory map. The pollutant migration trajectory map graphically displays the change curve of each pollutant concentration along the treatment process.

[0025] Preferably, the step of comparing the pollutant migration trajectory map with the actual pollutant concentration readings for the corresponding time window and calculating the trajectory deviation value includes:

[0026] From the pollutant migration trajectory map, extract the predicted pollutant concentration value of each processing unit node at the end of the simulation;

[0027] From the data block of the processing stage, read the actual online monitoring pollutant concentration value at the outlet of the same processing unit within the corresponding time window;

[0028] For each pollutant in total nitrogen, chemical oxygen demand, total phosphorus, and suspended solids, calculate the absolute difference between the predicted concentration value and the actual concentration value;

[0029] The absolute difference for the same pollutant is normalized to the allowable deviation range of the standard treatment efficiency for that pollutant to obtain the standardized deviation score for that pollutant.

[0030] The standardized deviation scores of total nitrogen, chemical oxygen demand, total phosphorus, and suspended solids are weighted and summed. The weighting coefficients are set according to the stringency of the emission standards for each pollutant. The summation result is the trajectory deviation value corresponding to the data block of this processing stage.

[0031] Preferably, the step of identifying the specific processing unit and pollutant type causing the trajectory deviation through the anomaly diagnosis procedure, and generating a diagnostic report including the anomaly unit identifier, the type of anomalous pollutant, and the degree of deviation, includes:

[0032] The anomaly diagnosis program is initiated, which receives the trajectory deviation value exceeding the deviation threshold and its corresponding processing stage data block;

[0033] The anomaly diagnosis program parses the standardized deviation scores of each pollutant that constitute the trajectory deviation value from the data block of the processing stage.

[0034] Identify the pollutant type with the largest standardized deviation score and mark it as the main anomalous pollutant type;

[0035] The simulated and actual changes in the concentration of the main anomalous pollutant type between each treatment unit node in the pollutant migration trajectory map were calculated retrospectively.

[0036] The adjacent processing unit node pair with the largest difference between the simulated change and the actual change is located. The upstream node is determined to be the unit where the anomaly occurs and the downstream node is determined to be the unit where the anomaly effect is manifested.

[0037] The percentage difference between the actual removal efficiency and the expected removal efficiency of the unit where the anomaly occurred is calculated as the degree of deviation of that unit.

[0038] The system summarizes the identifier of the initiating unit of the anomaly, the main types of abnormal pollutants, and the degree of deviation to generate a structured diagnostic report.

[0039] Preferably, the step of querying the historical control case database based on the diagnostic report, retrieving historical control records that match the abnormal unit identifier and abnormal pollutant type, and generating a control scheme set includes:

[0040] Analyze the diagnostic report to extract the abnormal unit identifiers and the main abnormal pollutant types;

[0041] Using the abnormal unit identifier and the main abnormal pollutant type as joint query keywords, the historical control case database is retrieved;

[0042] The historical control case database stores records of past abnormal events, records of control measures taken, and evaluation records of the effects of post-control treatment.

[0043] The search yielded all historical case records containing the same or similar anomalous unit identifiers and the main anomalous pollutant types.

[0044] From each matching historical case record, extract the details of the control measures taken, including the name of the adjusted parameter, the direction of adjustment, the magnitude of adjustment, and the duration of execution;

[0045] All extracted details of control measures are deduplicated and merged to form a set of control schemes.

[0046] Preferably, the method further includes:

[0047] The set of control schemes is subjected to multi-constraint feasibility verification, which includes verifying the compatibility of the schemes with the hydraulic impact on upstream and downstream treatment units, chemical dosing, and changes in the overall energy consumption of the treatment system, and screening out candidate control schemes that pass the verification.

[0048] The diagnostic report is integrated with the validated candidate control schemes to form an optimized instruction set that includes specific execution parameters and expected correction targets.

[0049] Preferably, the multi-constraint feasibility verification of the control scheme set includes:

[0050] Multiple sets of verification constraints were set, including hydraulic impact constraints, reagent compatibility constraints, and system energy consumption constraints.

[0051] For each control scheme in the set of control schemes, simulate its impact on the entire processing system after execution;

[0052] Assess whether the aforementioned control scheme will cause the hydraulic load of its downstream treatment unit to exceed the design range; if it does, it violates the hydraulic shock constraint.

[0053] Assess whether changes in the chemical dosage involved in the control scheme will produce adverse chemical reactions or precipitation with other agents being added in the system. If such reactions occur, it violates the agent compatibility constraints.

[0054] After the control scheme is implemented, evaluate whether the total energy consumption change of the processing system exceeds the preset energy consumption fluctuation threshold. If it does, the system energy consumption constraint is violated.

[0055] Check in sequence whether each control scheme simultaneously meets the hydraulic impact constraint, reagent compatibility constraint and system energy consumption constraint;

[0056] The control schemes that simultaneously meet all constraints are selected and marked as candidate control schemes that have passed the verification.

[0057] Preferably, the integration of the diagnostic report with the validated candidate control scheme to form an optimized instruction set including specific execution parameters and expected correction targets includes:

[0058] Read the degree of deviation from the diagnostic report regarding the starting unit of the abnormality;

[0059] Based on the degree of deviation, the parameter adjustment range in the verified candidate control scheme is scaled proportionally; the greater the degree of deviation, the larger the scaling factor.

[0060] For each candidate control scheme after scaling adjustment, a target correction is set, wherein the target correction is to increase the actual removal efficiency of the anomaly initiation unit for the main anomalous pollutant type to a specified percentage range of the expected removal efficiency.

[0061] The parameter adjustment suggestions and execution time after scaling are bound to the corresponding expected correction targets to form a single optimization instruction;

[0062] All optimization instructions for the current abnormal event are summarized and sorted based on the estimated time required to achieve the expected correction target, forming the final set of optimization instructions.

[0063] Preferably, when the processor executes the computer program, it implements the steps of the gas-fired power plant circulating water wastewater treatment method as described in any of the above-mentioned methods.

[0064] Compared with the prior art, the beneficial effects of the present invention are:

[0065] By inputting discrete, phased monitoring data from each treatment unit into a pollutant migration simulation network, the concentration change paths of key pollutants such as total nitrogen throughout the entire treatment chain can be dynamically extrapolated. This enables continuous dynamic simulation of the circulating water treatment process, replacing traditional isolated point data comparisons. Operators can thus gain insight into the transfer and transformation trends of pollutants between units, and detect potential deviations in the system state before indicators actually exceed standards, thereby shifting operation management from reactive response to proactive prediction.

[0066] By calculating the overall deviation between the predicted trajectory and the actual monitoring data throughout the entire process to trigger diagnosis, it is possible to reverse-engineer the specific responsible unit and pollutant type when the system malfunctions. This elevates fault identification from monitoring point alarms to precise source tracing of the anomaly. Based on this accurate source localization, the control schemes generated from retrieving historical cases directly target the malfunctioning unit and specific pollutant, avoiding the response delays and operational redundancy caused by traditional experience-based global adjustments, thus improving the effectiveness of control measures and the stability of system operation. Attached Figure Description

[0067] Figure 1This is a schematic diagram illustrating the working principle of the wastewater treatment method for circulating water in a gas-fired power plant according to the present invention.

[0068] Figure 2 A flowchart for staged segmentation and labeling processing;

[0069] Figure 3 A flowchart for calculating trajectory deviation;

[0070] Figure 4 A bar chart showing the distribution of pollutants in abnormal units of the wastewater treatment system for circulating water in a gas-fired power plant.

[0071] Figure 5 A bar chart showing the pollutant removal efficiency of the wastewater treatment unit in a gas-fired power plant's circulating water system. Detailed Implementation

[0072] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0073] Please see Figure 1This invention provides a method for treating circulating water wastewater from a gas-fired power plant. The method includes: continuously collecting and aggregating raw monitoring sequences from various treatment units in the circulating water treatment process of a gas-fired power plant, the sequences containing various operating parameters and pollutant concentration readings. The system performs staged segmentation and labeling of the raw monitoring sequences, dividing the continuous data stream into multiple independent treatment stage data blocks based on timestamps and treatment unit identifiers, and attaching a label indicating the treatment unit type from which each treatment stage data block originates. Subsequently, these treatment stage data blocks are input into a pre-constructed pollutant migration simulation network, which simulates the speciation and concentration decay processes of pollutants between consecutive treatment units, outputting a pollutant migration trajectory map. This map records in detail the predicted concentration change paths of total nitrogen, chemical oxygen demand, total phosphorus, and suspended solids between different treatment units. The system compares the pollutant migration trajectory map with the actual pollutant concentration readings within the same time window, calculating a quantified trajectory deviation value. When this value exceeds a preset deviation threshold, an anomaly diagnosis procedure for that time window is automatically triggered. The anomaly diagnosis program analyzes and compares the results to identify the specific treatment units and pollutant types that caused the trajectory deviation, and then generates a diagnostic report containing the anomaly unit identifier, the type of anomalous pollutant, and the degree of deviation. Finally, based on the generated diagnostic report, the system automatically queries the historical control case database, retrieves historical control records that match the anomaly unit identifier and anomalous pollutant type in the report, and integrates these records to generate a set of operable control schemes.

[0074] Example 1: See Figure 2The collected raw monitoring sequences include periodically recorded readings of operating parameters and pollutant concentrations in the denitrification tank, secondary sedimentation tank, flocculation sedimentation tank, ozone advanced oxidation unit, aerated biological filter, and fiber disc filter. The system reads the timestamp information and data source channel identifier of the raw monitoring sequences. Based on the data source channel identifier, it separates the continuous data stream into sub-streams corresponding to the denitrification tank, secondary sedimentation tank, flocculation sedimentation tank, ozone advanced oxidation unit, aerated biological filter, and fiber disc filter. For each independent data sub-stream, the system sets a fixed time window length based on the standard hydraulic retention time of its corresponding treatment unit and uses this time window length to perform a fixed-length sliding window segmentation of the data sub-stream, obtaining multiple preliminary data segments. The system checks the time span of each preliminary data segment; if its time span is incomplete, it automatically searches for adjacent data points forward or backward to complete the segment, thus forming a complete treatment stage data block. Based on the data source channel identifier, the system assigns a treatment unit type label to each complete treatment stage data block. The treatment unit type labels include denitrification tank stage, secondary sedimentation tank stage, flocculation sedimentation tank stage, ozone advanced oxidation stage, aerated biological filter stage, and fiber disc filter stage.

[0075] When constructing the pollutant migration simulation network, a directed network containing sequentially connected nodes is first built. Each node in the directed network represents a specific treatment unit, and the directed edges between nodes represent the wastewater flow direction and treatment sequence. The simulation process begins with the upstream treatment unit. The system extracts the pollutant concentration vector at the effluent end from the treatment stage data block corresponding to the upstream treatment unit and uses this vector as the simulation input for the downstream treatment unit. The system calls the pre-set treatment unit response sub-model in the pollutant migration simulation network. The treatment unit response sub-model selects the corresponding simulation logic based on the treatment unit type label attached to the input treatment stage data block. This simulation logic simulates the input pollutant concentration vector based on the physicochemical and biological principles of the treatment unit and predicts the pollutant concentration vector at its output end. The simulation output of the upstream node is used as the simulation input of the downstream node, and the simulation is passed sequentially along the edges of the directed network until the last node. The system records the change of pollutant concentration vector at each node before and after the simulation treatment and connects the concentration changes of all nodes in chronological order to draw a complete pollutant migration trajectory map. This pollutant migration trajectory map graphically displays the change curve of each pollutant concentration along the entire treatment process.

[0076] In practice, the original monitoring sequence includes periodic records from the effluent outlets of the denitrification tank, secondary sedimentation tank, flocculation sedimentation tank, ozone advanced oxidation unit, aerated biological filter, and fiber disc filter. These records include readings of operating parameters such as dissolved oxygen, pH, and oxidation-reduction potential, as well as pollutant concentrations such as total nitrogen, chemical oxygen demand, total phosphorus, and suspended solids. The system reads the timestamp and data source channel identifier attached to each line of data in the original monitoring sequence. The data source channel identifier clearly indicates whether the data originates from the monitoring point in the denitrification tank, secondary sedimentation tank, flocculation sedimentation tank, ozone advanced oxidation unit, aerated biological filter, or fiber disc filter. Based on the data source channel identifier, the system separates the continuous raw monitoring sequence data stream into six independent data sub-streams. The six independent data sub-streams correspond to the data sub-streams of the denitrification tank, the secondary sedimentation tank, the flocculation sedimentation tank, the ozone advanced oxidation unit, the aerated biological filter, and the fiber disc filter, respectively.

[0077] In practical implementation, for the data substream corresponding to the denitrification tank, the system sets the time window length based on the standard hydraulic retention time of two hours for the denitrification tank, and uses this two-hour time window to divide the data substream corresponding to the denitrification tank into fixed-length sliding windows. For the data substream corresponding to the secondary sedimentation tank, the system sets the time window length based on the standard hydraulic retention time of 1.5 hours for the secondary sedimentation tank. For the data substream corresponding to the flocculation sedimentation tank, the system sets the time window length based on the standard hydraulic retention time of one hour for the flocculation sedimentation tank. For the data substream corresponding to the ozone advanced oxidation unit, the system sets the time window length based on the standard hydraulic retention time of the ozone advanced oxidation unit (0.5 hours). For the data substream corresponding to the aerated biological filter, the system sets the time window length based on the standard hydraulic retention time of 1.2 hours for the aerated biological filter (1.2 hours). For the data substream corresponding to the fiber disc filter, the system sets the time window length based on the standard hydraulic retention time of 0.3 hours for the fiber disc filter (0.3 hours). The system uses the set time window length to divide the corresponding data substream into sliding windows, obtaining multiple preliminary data segments. The system checks whether the data points within each preliminary data segment cover the complete time window span. When an incomplete time span is found in a preliminary data segment, the system searches for adjacent timestamps to fill in the gaps, forming a complete processing stage data block. Based on the data source channel identifier, the system assigns a processing unit type label to each complete processing stage data block. These labels include labels for denitrification tanks, secondary sedimentation tanks, flocculation sedimentation tanks, ozone advanced oxidation tanks, aerated biological filters, and fiber disc filters.

[0078] In practical implementation, when constructing the pollutant migration simulation network, the system builds a directed network containing six sequentially connected nodes. These six nodes represent the denitrification tank treatment unit node, the secondary sedimentation tank treatment unit node, the flocculation sedimentation tank treatment unit node, the ozone advanced oxidation unit treatment unit node, the aerated biological filter treatment unit node, and the fiber disc filter treatment unit node, respectively. The directed edges between the nodes represent the treatment sequence of wastewater flowing from the denitrification tank to the secondary sedimentation tank, from the secondary sedimentation tank to the flocculation sedimentation tank, from the flocculation sedimentation tank to the ozone advanced oxidation unit, from the ozone advanced oxidation unit to the aerated biological filter, and from the aerated biological filter to the fiber disc filter. When the simulation starts, the system extracts the pollutant concentration vector from the effluent end of the treatment stage data block labeled with the denitrification tank stage. The pollutant concentration vector includes the total nitrogen concentration, chemical oxygen demand (COD) concentration, total phosphorus concentration, and suspended solids concentration. The extracted pollutant concentration vector serves as the simulation input for the secondary sedimentation tank treatment unit node.

[0079] In some embodiments, the system invokes a pre-defined treatment unit response sub-model in the pollutant migration simulation network. The treatment unit response sub-model selects the corresponding simulation logic based on the treatment unit type label attached to the input treatment stage data block. In a specific implementation, the treatment unit response sub-model, as a pre-defined component in the pollutant migration simulation network, internally stores a mapping relationship between simulation logic and different treatment unit type labels. When the system invokes the treatment unit response sub-model, the sub-model first reads the treatment unit type label attached to the input treatment stage data block, such as a denitrification tank stage label, a secondary sedimentation tank stage label, or an ozone advanced oxidation stage label. Then, based on the label content, it retrieves and activates the corresponding simulation logic from the mapping relationship. For data blocks labeled with a secondary sedimentation tank stage label, the sub-model selects a sedimentation separation simulation logic based on the principle of gravity settling. This logic simulates the settling process of suspended solids in the secondary sedimentation tank to predict the effluent suspended solids concentration. For data blocks labeled with an ozone advanced oxidation stage label, the sub-model selects an advanced oxidation simulation logic based on the principle of hydroxyl radical oxidation. This logic simulates the degradation process of organic pollutants under the action of ozone to predict the effluent chemical oxygen demand (COD) concentration. For data blocks tagged with aerated biofilter stages, the sub-model selects a biodegradation simulation logic based on the principle of microbial aerobic metabolism. This logic simulates the decomposition process of organic and nitrogenous pollutants by microorganisms to predict the chemical oxygen demand (COD) and total nitrogen (TNO) concentrations in the effluent. Each simulation logic dynamically calculates the input pollutant concentration vector based on the specific physicochemical or biological principles of its corresponding treatment unit, thereby generating the predicted pollutant concentration vector at the output of that treatment unit. For the simulation of COD at the aerated biofilter treatment unit node, the simulation logic is expressed using the following formalized mathematical relationship:

[0080]

[0081] Where: symbol This represents the chemical oxygen demand (COD) concentration value of the input node of the aerated biological filter treatment unit, with the symbol... This represents the chemical oxygen demand (COD) concentration value output by the simulated node of the aerated biological filter treatment unit, with the symbol [symbol missing]. This represents the biodegradation rate constant of the aerated biological filter unit relative to chemical oxygen demand (COD), with the symbol [symbol missing]. This indicates the standard hydraulic retention time of the aerated biological filter treatment unit.

[0082] In practical implementation, the system uses the simulated pollutant concentration vector output from the denitrification tank treatment unit node as the simulated input from the secondary sedimentation tank treatment unit node, and the simulated output from the secondary sedimentation tank treatment unit node as the simulated input from the flocculation sedimentation tank treatment unit node. The simulation process is passed sequentially along the edges of the directed network until the fiber disc filter treatment unit node completes the simulated treatment. The system records the changes in pollutant concentration vectors before and after the simulated treatment for the denitrification tank treatment unit node, the secondary sedimentation tank treatment unit node, the flocculation sedimentation tank treatment unit node, the ozone advanced oxidation unit treatment unit node, the aerated biological filter treatment unit node, and the fiber disc filter treatment unit node.

[0083] Example 2: See Figure 3 When calculating the trajectory deviation, the system extracts the predicted pollutant concentration value for each treatment unit node at the end of the simulation from the generated pollutant migration trajectory map. Simultaneously, the system reads the actual online monitored pollutant concentration values ​​at the outlet of the same treatment unit within the corresponding time window from the data block of the treatment stage that triggered the diagnosis. For each pollutant among total nitrogen, chemical oxygen demand (COD), total phosphorus, and suspended solids, the system calculates the absolute difference between its predicted and actual concentration values. The system normalizes the absolute difference for the same pollutant to the allowable deviation range of its standard treatment efficiency, obtaining the standardized deviation score for that pollutant. The system then performs a weighted summation of the standardized deviation scores for the four pollutants—total nitrogen, COD, total phosphorus, and suspended solids—with the weighting coefficients set according to the stringency of the emission standards for each pollutant. The summation result is the trajectory deviation value corresponding to the data block of that treatment stage.

[0084] When the trajectory deviation value exceeds a preset threshold, triggering an anomaly diagnosis procedure, the procedure receives this trajectory deviation value and its corresponding processing stage data block. The anomaly diagnosis procedure parses the standardized deviation scores of each pollutant constituting the trajectory deviation value from the processing stage data block, identifies the pollutant type with the largest standardized deviation score, and marks it as the primary anomalous pollutant type. The procedure backtracks and calculates the simulated and actual changes in the concentration of this primary anomalous pollutant type between each processing unit node in the pollutant migration trajectory map. The procedure locates the adjacent processing unit node pair with the largest difference between the simulated and actual changes, identifying the upstream node as the anomaly initiation unit and the downstream node as the anomaly impact manifestation unit. The system calculates the percentage difference between the actual and expected removal efficiency of the primary anomalous pollutant for the anomaly initiation unit, using this percentage difference as the deviation degree of that unit. The system summarizes the anomaly initiation unit's identifier, the primary anomalous pollutant type, and the calculated deviation degree to generate a structured diagnostic report.

[0085] In practical implementation, the system extracts the predicted pollutant concentration values ​​for each treatment unit node at the end of the simulation from the generated pollutant migration trajectory map. For example, it extracts the predicted total nitrogen concentration value of the secondary sedimentation tank effluent from the map position representing the secondary sedimentation tank treatment unit node, and the predicted chemical oxygen demand (COD) concentration value of the flocculation sedimentation tank effluent from the map position representing the flocculation sedimentation tank treatment unit node. The system reads the actual online monitoring pollutant concentration values ​​at the outlet of the same treatment unit within the corresponding time window from the treatment stage data block that triggers the anomaly diagnosis procedure. For example, it reads the actual total nitrogen concentration value of the secondary sedimentation tank effluent recorded in the treatment stage data block with the secondary sedimentation tank stage label, and the actual COD concentration value of the flocculation sedimentation tank effluent recorded in the treatment stage data block with the flocculation sedimentation tank stage label. For each pollutant among total nitrogen, COD, total phosphorus, and suspended solids, the system calculates the absolute difference between its predicted and actual concentration values. For example, it calculates the absolute difference between the predicted total nitrogen concentration value and the actual total nitrogen concentration value of the secondary sedimentation tank node. The system normalizes the absolute difference of the same pollutant against the allowable deviation range of the standard treatment efficiency for that pollutant to obtain the standardized deviation score. The allowable deviation range of the standard treatment efficiency is a pre-defined range of allowable concentration fluctuations based on design specifications or historical operating data. The system then performs a weighted summation of the standardized deviation scores for total nitrogen, chemical oxygen demand, total phosphorus, and suspended solids. The weighting coefficients are set according to the stringency of the emission standards for each pollutant; for example, total nitrogen is assigned a higher weighting coefficient due to its stricter emission limits. The summation result is the trajectory deviation value corresponding to the data block in that treatment stage. The process of calculating the trajectory deviation value can be expressed using the following formalized mathematical relationship:

[0086]

[0087] Where: symbol This represents the calculated trajectory deviation value, with the sign... The weighting coefficient for total nitrogen pollutants is represented by the symbol. The standardized deviation fraction of total nitrogen pollutants is represented by the symbol. The weighting coefficient for chemical oxygen demand (COD) pollutants is represented by the symbol. The standardized deviation fraction of chemical oxygen demand (COD) pollutants is represented by the symbol. The weighting coefficient for total phosphorus pollutants is represented by the symbol. The standardized deviation fraction of total phosphorus pollutants is represented by the symbol. The weighting coefficient for suspended particulate matter pollutants is represented by the symbol. This represents the standardized deviation score of suspended particulate pollutants.

[0088] In some embodiments, when the calculated trajectory deviation value exceeds a preset deviation threshold, the system automatically triggers an anomaly diagnosis program for that time window. The anomaly diagnosis program receives the trajectory deviation value exceeding the deviation threshold and its corresponding processing stage data block. The anomaly diagnosis program parses the standardized deviation scores of each pollutant constituting the trajectory deviation value from the processing stage data block, namely the standardized deviation scores of total nitrogen, chemical oxygen demand, total phosphorus, and suspended solids. The anomaly diagnosis program identifies the pollutant type with the largest standardized deviation score; for example, if the standardized deviation score of total nitrogen is identified as the largest, total nitrogen is marked as the primary anomalous pollutant type. The anomaly diagnosis program backtracks and calculates the simulated and actual changes in the concentration of the primary anomalous pollutant type between processing unit nodes in the pollutant migration trajectory map. The simulated change refers to the difference between the predicted concentration values ​​of adjacent nodes in the pollutant migration trajectory map, and the actual change refers to the difference between the actual concentration values ​​of adjacent nodes recorded in the processing stage data block. The anomaly diagnosis program locates the adjacent treatment unit node pair with the largest difference between the simulated change and the actual change. The upstream node is identified as the starting unit of the anomaly, and the downstream node is identified as the unit where the anomaly effect manifests. For example, if the difference between the aerated biological filter treatment unit node and the fiber disc filter treatment unit node is the largest, the aerated biological filter treatment unit node is identified as the starting unit of the anomaly.

[0089] The anomaly diagnosis program calculates the percentage difference between the actual removal efficiency and the expected removal efficiency of the unit where the anomaly occurred as the degree of deviation. The actual removal efficiency is calculated based on the actual inlet and outlet concentrations of the unit where the anomaly occurred in the data block of the treatment stage, while the expected removal efficiency is calculated based on the simulated inlet and outlet concentrations of the unit where the anomaly occurred in the pollutant migration trajectory map. The system summarizes the identifier of the unit where the anomaly occurred, the main abnormal pollutant type, and the calculated degree of deviation, and generates a structured diagnostic report. The diagnostic report records in the form of a data file that the unit where the anomaly occurred is an aerated biological filter, the main abnormal pollutant type is total nitrogen, and the degree of deviation is -15%.

[0090] Optionally, the allowable deviation range for standard treatment efficiency can be set differently for different treatment units. For example, the allowable deviation range for the standard treatment efficiency of suspended solids in a secondary sedimentation tank is set to 5 mg / L, and the allowable deviation range for the standard treatment efficiency of chemical oxygen demand in an ozone advanced oxidation unit is set to 10 mg / L. In some embodiments, the weighting coefficient is directly related to national or local pollutant emission standards, assigning higher weighting coefficients to pollutants with stricter emission limits to reflect their importance in the trajectory deviation value. It can be understood that the process of retrospectively calculating the simulated change and the actual change involves traversing every directed edge in the pollutant migration trajectory map and calculating the concentration change difference of the main abnormal pollutant types on each edge. Optionally, the calculation of the deviation degree can be expressed as dividing by the expected removal efficiency and then multiplying by 100%. A positive result indicates that the actual removal efficiency is higher than expected, and a negative result indicates that the actual removal efficiency is lower than expected.

[0091] Example 3: When generating a set of control schemes based on a diagnostic report, the system first parses the diagnostic report, extracting the abnormal unit identifier and the main abnormal pollutant type. The system uses the abnormal unit identifier and the main abnormal pollutant type as joint query keywords to search the historical control case database. This database stores records of past abnormal events, records of control measures taken in response to these events, and evaluation records of post-control treatment effects. The system retrieves all historical case records containing the same or similar abnormal unit identifiers and main abnormal pollutant types. From each matching historical case record, the system extracts details of the control measures taken. These details typically include the name of the adjusted parameter, the direction of adjustment, the adjustment magnitude, and the execution duration. The system deduplicates and merges all extracted control measure details to form a preliminary set of control schemes.

[0092] In practical implementation, when the system queries the historical control case library based on the generated diagnostic report, it first parses the diagnostic report, which is a structured data file. The system extracts the content of the abnormal unit identifier field and the main abnormal pollutant type field from the diagnostic report; for example, it extracts the abnormal unit identifier "aerated biological filter" and the main abnormal pollutant type "total nitrogen" from the diagnostic report. The system uses the extracted abnormal unit identifier "aerated biological filter" and the main abnormal pollutant type "total nitrogen" as joint query keywords to search the historical control case library. The historical control case library is a relational database that stores records of past abnormal events, records of control measures taken in response to past abnormal events, and evaluation records of the post-control treatment effects. The system searches the historical control case library for all historical case records containing the same or similar abnormal unit identifier and main abnormal pollutant type; for example, the search criteria are that the abnormal unit identifier field contains "aerated biological filter" and the main abnormal pollutant type field contains "total nitrogen".

[0093] In practice, the system extracts details of the control measures taken from each matching historical case record. These details include the name of the adjusted parameter, the direction of adjustment, the magnitude of adjustment, and the execution duration. For example, from one historical case record, the extracted control measures details are: the adjusted parameter name is "backwashing frequency of aerated biological filter," the direction of adjustment is "increase," the magnitude of adjustment is "20%," and the execution duration is "four hours." From another historical case record, the extracted control measures details are: the adjusted parameter name is "carbon source dosage of aerated biological filter," the direction of adjustment is "increase," the magnitude of adjustment is "15 mg / L," and the execution duration is "six hours." From a third historical case record, the extracted control measures details are: the adjusted parameter name is "aeration rate of aerated biological filter," the direction of adjustment is "increase," the magnitude of adjustment is "10%," and the execution duration is "five hours." The system deduplicates and merges all extracted control measures details. For records with the same adjusted parameter name but different magnitudes or execution durations, the merging process retains all different value options, forming a preliminary control scheme set containing multiple control measures.

[0094] In some embodiments, records in the historical control case database are linked through a correlation function, which can be expressed as:

[0095]

[0096] Where: symbol This represents a historical case record. Represents an "associative function", symbol The feature vector representing historical anomalous events, with symbols The vector represents the control measures taken, with the symbol... This represents the evaluation vector indicating the effectiveness of post-control measures. After parsing the diagnostic report, the extracted diagnostic report information forms a current event feature vector. The system calculates the similarity between the current event feature vector and the feature vectors of historical abnormal events in historical case records, and uses this similarity as an auxiliary sorting criterion, prioritizing the display of historical case records with high similarity in the search results. Optionally, deduplication means that only one record is retained for details of completely identical control measures.

[0097] Optionally, the construction of joint query keywords can include more dimensions. For example, combining the numerical range of deviation in the diagnostic report, the system can use a deviation greater than 10% as an additional query condition. In some embodiments, for retrieved historical case records, the system simultaneously extracts evaluation records of the post-treatment effects, which include the percentage decrease in the concentration of major abnormal pollutants or the recovery rate of unit treatment efficiency after treatment. The system binds all extracted control measure details with the corresponding post-treatment effect evaluation records to form control measure details labeled with expected effects. When merging the extracted control measure details, the system removes records with the same adjustment parameter name but opposite adjustment directions. For example, if one record suggests "increasing carbon source dosage" and another suggests "reducing carbon source dosage," the system considers these two records to be fundamentally conflicting and will not include them in the final control scheme set.

[0098] See Figure 4 This is a bar chart showing the distribution of pollutants in abnormal units of the wastewater treatment system of a gas-fired power plant. The aerated biological filter (ABF) had the most abnormal chemical oxygen demand (COD) cases, while the ozone advanced oxidation unit (OOU) had 5 abnormal suspended solids (SSD) cases. The secondary sedimentation tank and denitrification tank showed a relatively balanced number of abnormal cases for various pollutants. This chart reflects that the ABF and OOU are high-risk and key pollutant units in the wastewater treatment process of gas-fired power plants, with the ABF and OOU being the most frequently occurring units for pollutant anomalies. COD and SSDs are the key pollutant types that need to be monitored in the current treatment process, and the control schemes for these units need to be optimized accordingly.

[0099] Example 4: After generating the control scheme set, the system performs multi-constraint feasibility verification on the control scheme set. The multi-constraint feasibility verification process sets multiple sets of verification constraints, including hydraulic impact constraints, chemical compatibility constraints, and system energy consumption constraints. For each control scheme in the control scheme set, the system simulates its impact on the entire wastewater treatment system after execution. The system evaluates whether the control scheme will cause the hydraulic load of its downstream treatment unit to exceed the design range. If it does, it is determined to violate the hydraulic impact constraint. The system evaluates whether the changes in the chemical dosage involved in the control scheme will produce adverse chemical reactions or precipitation with other chemicals being added in the system. If so, it is determined to violate the chemical compatibility constraint. The system evaluates whether the change in the total energy consumption of the wastewater treatment system after the execution of the control scheme exceeds the preset energy consumption fluctuation threshold. If it does, it is determined to violate the system energy consumption constraint. The system sequentially checks whether each control scheme simultaneously satisfies the hydraulic impact constraint, chemical compatibility constraint, and system energy consumption constraint. The system selects the control scheme that simultaneously satisfies all constraints and marks it as a candidate control scheme that has passed verification.

[0100] In practical implementation, when verifying the feasibility of the control scheme set under multiple constraints, the system sets multiple sets of verification constraints, including hydraulic impact constraints, chemical compatibility constraints, and system energy consumption constraints. Hydraulic impact constraints define the maximum and minimum allowable hydraulic load range for each treatment unit; chemical compatibility constraints define the rules prohibiting the simultaneous addition of different chemical agents or establishing dosage limitations; and system energy consumption constraints define the allowable energy consumption fluctuation threshold for the entire wastewater treatment system. For each control scheme in the control scheme set, the system simulates its impact on the entire wastewater treatment system after execution, based on the wastewater treatment plant's process model and material balance calculations.

[0101] In some embodiments, to assess whether a control scheme violates hydraulic shock constraints, the system checks whether the control scheme would cause the hydraulic load of its downstream treatment unit to exceed the design range. For example, a control scheme that suggests increasing the "recirculation ratio in the denitrification tank" would increase the hydraulic load of the subsequent secondary sedimentation tank. The system calculates the expected surface hydraulic load of the secondary sedimentation tank after the implementation of this scheme. If the expected surface hydraulic load exceeds 95% of the maximum design value of the secondary sedimentation tank, the control scheme is determined to violate the hydraulic shock constraint. To assess whether a control scheme violates reagent compatibility constraints, the system checks whether the changes in the chemical reagent dosage involved in the control scheme would produce adverse chemical reactions or precipitation with other reagents being added in the system. For example, a control scheme that suggests "increasing the dosage of aluminum salt coagulant in the flocculation sedimentation tank" would be evaluated by querying the reagent compatibility constraint rule base. If the rule base contains a rule that "aluminum salts and the specific phosphate scale inhibitor currently added to the system easily form precipitates at high pH," and the current system pH value is higher than the rule trigger threshold, the control scheme is determined to violate the reagent compatibility constraint. To assess whether a control scheme violates system energy consumption constraints, the system calculates the total energy consumption change of the treatment system after the control scheme is implemented. Total energy consumption includes the energy consumption of all moving equipment such as pumps, fans, and agitators. The system compares the calculated total energy consumption change with a preset energy consumption fluctuation threshold. A control scheme aimed at "increasing the ozone dosage in the advanced ozone oxidation unit" will significantly increase the energy consumption of the ozone generator. If the simulated increase in total energy consumption exceeds 5% of the allowable energy consumption fluctuation threshold, the control scheme is deemed to violate system energy consumption constraints. The system sequentially checks whether each control scheme simultaneously satisfies hydraulic impact constraints, reagent compatibility constraints, and system energy consumption constraints. The system selects control schemes that simultaneously satisfy all constraints and marks them as candidate control schemes that have passed verification. The simulation evaluation process for multi-constraint feasibility verification can be expressed using the following formalized mathematical relationship:

[0102]

[0103] Where: symbol Indicates a violation of the comprehensive evaluation value, symbol This indicates a violation of the hydraulic impact constraint indicator value; the value is 1 when a violation occurs and 0 otherwise. This indicates a violation of the drug compatibility constraint indicator value. The value is 1 if a violation occurs and 0 otherwise. This indicates a violation of the system's energy consumption constraint; the value is 1 when a violation occurs and 0 otherwise. Only control schemes with a value of 0 are marked as validated. In practice, the system integrates diagnostic reports with validated candidate control schemes to form an optimized instruction set. See Table 1 for a summary of validation results for a typical control scheme.

[0104] Table 1: Validation Results of Candidate Control Schemes under Multiple Constraints

[0105] Description of the regulation plan Hydraulic shock constraint verification Drug compatibility constraint verification System energy consumption constraint verification Comprehensive verification results Increase the backwashing frequency of the aerated biological filter by 20% and continue for 4 hours. pass pass pass pass Increase the carbon source dosage in the aerated biological filter by 15 mg / L and continue for 6 hours. pass pass pass pass Increase the aeration intensity of the aerated biological filter by 30%. pass pass Failed (energy consumption exceeds limit) Not approved Add coagulant aid to flocculation sedimentation tank pass Failed (Drug incompatibility) pass Not approved Significantly increase the internal recirculation ratio to dilute the influent load. Failed (downstream secondary sedimentation tank overload). pass Failure (surge in booster pump energy consumption) Not approved

[0106] Optionally, the hydraulic load can be calculated based on the pipe flow rate and the cross-sectional area of ​​the tank, and the reagent compatibility rule base is stored in "IF-THEN" format. It is understood that the simulation calculation of system energy consumption needs to consider the product of equipment power and operating time, and the energy consumption fluctuation threshold can be dynamically set according to the wastewater treatment plant's monthly electricity consumption plan. In some embodiments, the selected and validated candidate control schemes will be accompanied by a brief description of their validation process, such as "Scheme A: The simulated energy consumption increase is 2.3%, lower than the threshold of 5%, passing the energy consumption constraint validation."

[0107] Example 5: When integrating diagnostic reports and candidate control schemes to form an optimized instruction set, the system reads the deviation value of the starting unit of the anomaly from the diagnostic report. Based on the magnitude of the deviation, the system scales the parameter adjustment range suggested in the validated candidate control schemes proportionally; the greater the deviation, the larger the scaling factor applied. The system sets a clear expected correction target for each candidate control scheme after scaling adjustment. The expected correction target is stated as increasing the actual removal efficiency of the starting unit of the anomaly for the main abnormal pollutant type to a specified percentage range of its expected removal efficiency. The system binds the specific parameter adjustment suggestions after scaling adjustment, the suggested execution time, and the corresponding expected correction target to form a complete optimization instruction. The system summarizes and sorts all optimization instructions for the current anomaly event based on the estimated time required for each optimization instruction to achieve its expected correction target, ultimately forming a structured optimized instruction set.

[0108] In practical implementation, when integrating diagnostic reports and validated candidate control schemes to form an optimized instruction set, the system reads the deviation degree field of the anomaly initiation unit from the diagnostic report. For example, if the diagnostic report indicates that the anomaly initiation unit is an aerated biological filter, and its deviation degree for the main anomalous pollutant type, total nitrogen, is -15%, the system scales the parameter adjustment range in the validated candidate control schemes according to the magnitude of the deviation degree. The larger the absolute value of the deviation degree, the larger the scaling factor. The system internally presets a scaling factor mapping relationship; for example, a scaling factor of 1.0 corresponds to an absolute deviation degree in the range of 5% to 10%, and a scaling factor of 1.2 corresponds to an absolute deviation degree in the range of 10% to 20%. The system sets an expected correction target for each candidate control scheme after scaling adjustment. The expected correction target is expressed as increasing the actual removal efficiency of the anomaly initiation unit for the main anomalous pollutant type to a specified percentage range of the expected removal efficiency. For example, the target is set to increase the actual removal efficiency of the aerated biological filter for total nitrogen to between 95% and 105% of its expected removal efficiency.

[0109] In practical implementation, the system binds the scaled parameter adjustment suggestions, execution time, and corresponding expected correction targets to form a single optimization instruction. The process of adjusting the parameter range of the candidate control scheme through scaling can be expressed by the following formalized mathematical relationship:

[0110]

[0111] Where: symbol Indicates the adjustment range of the parameter after scaling, symbol Indicates the original parameter adjustment range in the candidate control scheme, with the sign... Indicates the system's preset proportional coefficient, symbol This indicates the degree of deviation read from the diagnostic report. For example, a validated candidate control scheme originally suggested "increasing the carbon source dosage in the aerated biological filter by 15 mg / L". The deviation read in the diagnostic report is -15%, the system's preset proportional coefficient is 0.015, and the calculated adjustment range after scaling is 17.25 mg / L. The system rounds this value to 17 mg / L. The single optimization instruction formed by binding includes the adjustment parameter "carbon source dosage in the aerated biological filter", the adjustment direction "increase", the adjustment range "17 mg / L", the execution time "six hours", and the expected correction target "to make the actual removal efficiency of total nitrogen in the aerated biological filter reach between 95% and 105% of the expected removal efficiency".

[0112] In some embodiments, the system aggregates all optimization instructions for the current anomalous event, resulting in a set of multiple independent instructions. The system sorts the aggregated optimization instruction set based on the estimated time required to achieve the expected correction target. For example, if one optimization instruction has an estimated execution time of four hours and another has an estimated execution time of six hours, the instruction with an estimated execution time of four hours will precede the one with an estimated execution time of six hours. The system then forms the final optimization instruction set, presented in list form. Each instruction in the list clearly specifies the adjustment target, adjustment action, adjustment amount, duration, and expected target. It can be understood that the scaling function allows the intensity of the control suggestion to automatically match the severity of the anomalous deviation; the greater the deviation, the stronger the adjustment suggestion. Optionally, when setting the expected correction target, the specified percentage range can be fine-tuned based on the main anomalous pollutant type and the stability of the treatment unit. For treatment units with high volatility, the range can be appropriately widened. In some embodiments, for optimization instructions with long execution times, the system can break them down into multiple sub-instructions executed in stages and set stage-specific expected correction targets for each stage. Understandably, the final optimized instruction set is directly geared towards the operators, providing a complete closed-loop output from anomaly diagnosis to specific operational steps.

[0113] See Figure 5 This is a bar chart showing the pollutant removal efficiency of the wastewater treatment unit in a gas-fired power plant's circulating water system. With "treatment unit" on the x-axis and "removal efficiency (%)" on the y-axis, it displays the removal effect of four types of pollutants in different units. The chart illustrates the functional positioning and advantages of each treatment unit. The secondary sedimentation tank, aerated biological filter, and ozone advanced oxidation unit are the core treatment links for SS, TN, and COD, respectively, and their stable operation must be ensured. If a pollutant exceeds the standard, the operating parameters of its corresponding high-removal-rate unit should be checked first.

[0114] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for treating circulating water wastewater in a gas-fired power plant, characterized in that, The method includes: Collect and aggregate raw monitoring sequences from the circulating water treatment process of a gas-fired power plant; The original monitoring sequence is segmented and labeled in stages. The continuous data stream is divided into multiple independent processing stage data blocks according to the timestamp and processing unit identifier, and each processing stage data block is labeled with the processing unit type from which it originates. The data blocks from the processing stage are input into a pollutant migration simulation network to simulate the speciation and concentration decay of pollutants between consecutive processing units, generating a pollutant migration trajectory map. The pollutant migration trajectory map records the predicted concentration change paths of total nitrogen, chemical oxygen demand, total phosphorus, and suspended solids between different processing units. The pollutant migration trajectory map is compared with the actual pollutant concentration readings for the corresponding time window to calculate the trajectory deviation value. When the trajectory deviation value exceeds the preset deviation threshold, an anomaly diagnosis procedure for that time window is triggered. The anomaly diagnosis procedure identifies the specific processing unit and type of contaminant that caused the trajectory deviation, and generates a diagnostic report that includes the anomaly unit identifier, the type of anomalous contaminant, and the degree of deviation. Based on the diagnostic report, the historical control case database is queried to retrieve historical control records that match the abnormal unit identifier and abnormal pollutant type, and a set of control schemes is generated.

2. The method for treating circulating water wastewater in a gas-fired power plant according to claim 1, characterized in that, The staged segmentation and labeling process for the original monitoring sequence includes: The original monitoring sequence includes periodically recorded readings of operating parameters and pollutant concentrations in denitrification tanks, secondary sedimentation tanks, flocculation sedimentation tanks, ozone advanced oxidation units, aerated biological filters, and fiber disc filters. Read the timestamp information and data source channel identifier of the original monitoring sequence; Based on the data source channel identifier, the data stream is separated into data sub-streams corresponding to the denitrification tank, the secondary sedimentation tank, the flocculation sedimentation tank, the ozone advanced oxidation unit, the aerated biological filter, and the fiber disc filter. For each data substream, the time window length is set according to the standard hydraulic residence time of its corresponding processing unit. The data substream is then divided into multiple preliminary data segments by a fixed-length sliding window using this time window length. Check the time span of each preliminary data segment. If the time span is incomplete, search for adjacent data points forward or backward to fill in the gaps and form a complete processing stage data block. Based on the data source channel identifier, each complete processing stage data block is assigned a processing unit type label of its source. The processing unit type label includes denitrification tank stage, secondary sedimentation tank stage, flocculation sedimentation tank stage, ozone advanced oxidation stage, aerated biological filter stage, and fiber disc filter stage.

3. The method for treating circulating water wastewater in a gas-fired power plant according to claim 1, characterized in that, The step of inputting the data blocks from the processing stage into the pollutant migration simulation network to simulate the speciation and concentration decay of pollutants between consecutive processing units and generate a pollutant migration trajectory map includes: Construct a directed network containing sequentially connected nodes, where each node represents a specific processing unit, and the directed edges between nodes represent the sewage flow direction and processing order; Extract the pollutant concentration vector at the effluent end from the data block corresponding to the upstream treatment unit, and use it as the simulation input for the downstream treatment unit; The processing unit response sub-model in the pollutant migration simulation network is invoked. The processing unit response sub-model selects the corresponding simulation logic according to the processing unit type label. The simulation logic simulates the input pollutant concentration vector based on the physicochemical and biological action principle of the unit and predicts the output pollutant concentration vector. The simulated output of the upstream node is used as the simulated input of the downstream node and is passed sequentially along the edges of the directed network until the last node. Record the pollutant concentration vector changes before and after each node's simulated treatment, and connect the concentration changes of all nodes in chronological order to draw a complete pollutant migration trajectory map. The pollutant migration trajectory map graphically displays the change curve of each pollutant concentration along the treatment process.

4. The method for treating circulating water wastewater in a gas-fired power plant according to claim 1, characterized in that, The step of comparing the pollutant migration trajectory map with the actual pollutant concentration readings for the corresponding time window and calculating the trajectory deviation value includes: From the pollutant migration trajectory map, extract the predicted pollutant concentration value of each processing unit node at the end of the simulation; From the data block of the processing stage, read the actual online monitoring pollutant concentration value at the outlet of the same processing unit within the corresponding time window; For each pollutant in total nitrogen, chemical oxygen demand, total phosphorus, and suspended solids, calculate the absolute difference between the predicted concentration value and the actual concentration value; The absolute difference for the same pollutant is normalized to the allowable deviation range of the standard treatment efficiency for that pollutant to obtain the standardized deviation score for that pollutant. The standardized deviation scores of total nitrogen, chemical oxygen demand, total phosphorus, and suspended solids are weighted and summed. The weighting coefficients are set according to the stringency of the emission standards for each pollutant. The summation result is the trajectory deviation value corresponding to the data block of this processing stage.

5. The method for treating circulating water wastewater in a gas-fired power plant according to claim 1, characterized in that, The anomaly diagnosis procedure identifies the specific processing unit and contaminant type causing the trajectory deviation, and generates a diagnostic report including the anomaly unit identifier, the type of anomalous contaminant, and the degree of deviation, including: The anomaly diagnosis program is initiated, which receives the trajectory deviation value exceeding the deviation threshold and its corresponding processing stage data block; The anomaly diagnosis program parses the standardized deviation scores of each pollutant that constitute the trajectory deviation value from the data block of the processing stage. Identify the pollutant type with the largest standardized deviation score and mark it as the main anomalous pollutant type; The simulated and actual changes in the concentration of the main anomalous pollutant type between each treatment unit node in the pollutant migration trajectory map were calculated retrospectively. The adjacent processing unit node pair with the largest difference between the simulated change and the actual change is located. The upstream node is determined to be the unit where the anomaly occurs and the downstream node is determined to be the unit where the anomaly effect is manifested. The percentage difference between the actual removal efficiency and the expected removal efficiency of the unit where the anomaly occurred is calculated as the degree of deviation of that unit. The system summarizes the identifier of the initiating unit of the anomaly, the main types of abnormal pollutants, and the degree of deviation to generate a structured diagnostic report.

6. The method for treating circulating water wastewater in a gas-fired power plant according to claim 5, characterized in that, Based on the diagnostic report, the system queries the historical control case database, retrieves historical control records that match the abnormal unit identifier and abnormal pollutant type, and generates a control scheme set, including: Analyze the diagnostic report to extract the abnormal unit identifiers and the main abnormal pollutant types; Using the abnormal unit identifier and the main abnormal pollutant type as joint query keywords, the historical control case database is retrieved; The historical control case database stores records of past abnormal events, records of control measures taken, and evaluation records of the effects of post-control treatment. The search yielded all historical case records containing the same or similar anomalous unit identifiers and the main anomalous pollutant types. From each matching historical case record, extract the details of the control measures taken, including the name of the adjusted parameter, the direction of adjustment, the magnitude of adjustment, and the duration of execution; All extracted details of control measures are deduplicated and merged to form a set of control schemes.

7. The method for treating circulating water wastewater in a gas-fired power plant according to claim 1, characterized in that, The method further includes: The set of control schemes is subjected to multi-constraint feasibility verification, which includes verifying the compatibility of the schemes with the hydraulic impact on upstream and downstream treatment units, chemical dosing, and changes in the overall energy consumption of the treatment system, and screening out candidate control schemes that pass the verification. The diagnostic report is integrated with the validated candidate control schemes to form an optimized instruction set that includes specific execution parameters and expected correction targets.

8. The method for treating circulating water wastewater in a gas-fired power plant according to claim 7, characterized in that, The multi-constraint feasibility verification of the control scheme set includes: Multiple sets of verification constraints were set, including hydraulic impact constraints, reagent compatibility constraints, and system energy consumption constraints. For each control scheme in the set of control schemes, simulate its impact on the entire processing system after execution; Assess whether the aforementioned control scheme will cause the hydraulic load of its downstream treatment unit to exceed the design range; if it does, it violates the hydraulic shock constraint. Assess whether changes in the chemical dosage involved in the control scheme will produce adverse chemical reactions or precipitation with other agents being added in the system. If such reactions occur, it violates the agent compatibility constraints. After the control scheme is implemented, evaluate whether the total energy consumption change of the processing system exceeds the preset energy consumption fluctuation threshold. If it does, the system energy consumption constraint is violated. Check in sequence whether each control scheme simultaneously meets the hydraulic impact constraint, reagent compatibility constraint and system energy consumption constraint; The control schemes that simultaneously meet all constraints are selected and marked as candidate control schemes that have passed the verification.

9. The method for treating circulating water wastewater in a gas-fired power plant according to claim 8, characterized in that, The integration of the diagnostic report with the validated candidate control schemes forms an optimized instruction set containing specific execution parameters and expected correction targets, including: Read the degree of deviation from the diagnostic report regarding the starting unit of the abnormality; Based on the degree of deviation, the parameter adjustment range in the verified candidate control scheme is scaled proportionally; the greater the degree of deviation, the larger the scaling factor. For each candidate control scheme after scaling adjustment, a target correction is set, wherein the target correction is to increase the actual removal efficiency of the anomaly initiation unit for the main anomalous pollutant type to a specified percentage range of the expected removal efficiency. The parameter adjustment suggestions and execution time after scaling are bound to the corresponding expected correction targets to form a single optimization instruction; All optimization instructions for the current abnormal event are summarized and sorted based on the estimated time required to achieve the expected correction target, forming the final set of optimization instructions.

10. A wastewater treatment system for circulating water in a gas-fired power plant, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the wastewater treatment method for circulating water in a gas-fired power plant as described in any one of claims 1 to 9.

Images