Intelligent carbon system based on water treatment

Abstract

The invention discloses an intelligent carbon system based on water treatment, which belongs to the field of urban management and comprises an acquisition processing module, a treatment map module, a training prediction module, an evaluation prediction module, a pollution traceability module, a resource optimization module, a scheme evaluation module, a scheduling control module and a public information platform. According to the method, the risk of privacy disclosure can be avoided, the possibility of data cooperation is improved, the generalization ability and accuracy of a prediction model are remarkably improved, data barriers among cities are broken, cross-regional cooperative governance is promoted, and the consistency and collaboration of governance schemes are improved; complex relations behind pollution events can be comprehensively analyzed, a scientific basis is provided for traceability, a plurality of potential pollution sources and diffusion paths thereof can be efficiently evaluated, real-time response capability and information transmission efficiency are improved, and an intelligent carbon system is helped to better coordinate multi-aspect treatment requirements.

Description

Technical Field

[0001] The present invention relates to the field of urban management, and in particular to an intelligent carbon system based on water management. Background Art

[0002] Water is the source of life in cities, but it is also a key factor in resource shortages and limited environmental carrying capacity in urban development. Urban water governance is an important part of modern urban management, covering water resource utilization, water pollution prevention and control, and water ecological protection. Current urban water governance often faces problems such as information dispersion, isolated governance strategies, and inefficient resource utilization. At the same time, due to the dependence of the water governance process on energy, a large amount of greenhouse gas emissions have been emitted, further exacerbating the pressure of carbon emission reduction. Urban water governance must integrate advanced technologies and green development concepts on the basis of traditional management models and move towards intelligence and low carbonization. Against the background of global climate change and accelerated urbanization, urban water governance faces multiple challenges such as resource shortages, environmental pollution, and carbon emissions. On the one hand, efficient utilization and optimal allocation of water resources have become the key to coping with the imbalance between supply and demand; on the other hand, water governance activities are closely related to carbon emissions, such as the large amount of energy consumption generated in sewage treatment and water resource transportation. Therefore, how to integrate carbon emission reduction goals into water governance and achieve the dual goals of water resource protection and low-carbon development has become an urgent problem to be solved.

[0003] After searching, Chinese patent number CN117605119A discloses a smart carbon system based on water governance, which provides a variety of energy-saving implementation paths with strong feasibility, optimizes urban water system planning, improves the ecological benefits of water system operation, and effectively promotes the construction of low-carbon cities. However, the risk of privacy leakage is high, which reduces the possibility of data cooperation, and the generalization ability and accuracy of the prediction model are poor, which is not conducive to promoting cross-regional collaborative governance; in addition, the existing smart carbon system based on water governance cannot fully analyze the complex relationship behind pollution events, reducing real-time response capabilities and information transmission efficiency. For this reason, we propose a smart carbon system based on water governance. Summary of the invention

[0004] The purpose of the present invention is to solve the defects in the prior art and to propose a smart carbon system based on water treatment.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A smart carbon system based on water management, including a collection and processing module, a management map module, a training and prediction module, an evaluation and prediction module, a pollution source tracing module, a resource optimization module, a program evaluation module, a scheduling and control module, and a public information platform;

[0007] The acquisition and processing module is used to collect water quality, water quantity, rainfall, climate and carbon emission data through multi-source sensors, and process each set of collected data;

[0008] The governance graph module is used to integrate domain knowledge and collected data to establish a causal relationship network between water resource flow and carbon emissions;

[0009] The training prediction module is used to train a high-precision prediction model and perform water quality anomaly detection and trend prediction in real time based on the prediction model;

[0010] The evaluation and prediction module is used to predict the impact of different governance schemes on carbon emissions;

[0011] The pollution source tracing module is used to locate pollution sources and simulate pollutant diffusion paths;

[0012] The resource optimization module is used to simulate various water resource allocation schemes and optimize water supply and water use scheduling strategies;

[0013] The scheme evaluation module is used to generate multiple governance schemes, evaluate their effects in terms of pollution control, water conservation and carbon emission reduction, and select the scheme with the highest comprehensive benefit;

[0014] The dispatching control module is used to combine with IoT devices to perform intelligent control of water pumps, sluice gates, and treatment equipment, and accurately operate sewage treatment and rainwater dispatching;

[0015] The public information platform is used to publish water governance progress, governance plans and carbon emission data, and collect public feedback to optimize the system.

[0016] As a further solution of the present invention, the specific steps of the governance map module to establish a causal relationship network associated with water resource flow and carbon emissions are as follows:

[0017] S1.1: Collect structured, semi-structured and unstructured data from multiple sources, perform semantic analysis on text information through natural language processing technology, annotate text data using CRF model and identify entities, attributes and relationships in urban water governance;

[0018] S1.2: Generate the basic triple form of the knowledge graph based on the relationship between entities extracted from the data, determine the causal relationship between entities through Granger causality test, and construct a weighted directed graph to represent the causal network based on water resource flow and carbon emission data;

[0019] S1.3: The constructed triples are embedded and represented, the causal network is mapped to a low-dimensional vector space, a graph database is used to store the triple embedding results, and the implicit relationship between each group of water resource flow and carbon emission data is obtained through causal network reasoning.

[0020] As a further solution of the present invention, the specific steps of the training prediction module for water quality anomaly detection and trend prediction are as follows:

[0021] S2.1: Each city or region stores the collected water quality, water quantity, rainfall, climate and carbon emission data locally. The central server initializes the global model, regards the smart carbon system of each city or region as a participating node, and sends the global model to each participating node;

[0022] S2.2: Each participating node initializes its own local model data based on the received global model information, trains the local model through the stored local data, and updates the local model architecture and parameters through the back propagation algorithm and gradient descent algorithm with the goal of minimizing the local loss function;

[0023] S2.3: The central server collects the updated model parameters of all participating nodes, aggregates them into a new global model in a weighted average manner, inputs the historical water management data into the new global model, and calculates the loss value between the global model prediction value and the actual value to evaluate the performance of the global model. If the model performance reaches the preset target, the global model is sent to each participating node and the local model information of each node is updated. If it does not meet the target, the global model is re-sent and the latest local data is used for training and aggregation;

[0024] S2.4: The final global model is sent to each participating node, and the original local model information is updated to the latest global model information. After that, each local model accepts the latest collected data sets, and forward propagates each set of data through the local model to obtain the abnormal values ​​of the corresponding water quality parameters and future water quality trends.

[0025] As a further solution of the present invention, the specific steps of the pollution source tracing module to locate the pollution source and simulate the pollutant diffusion path are as follows:

[0026] S3.1: Obtain pollution information from sensor data, and match each set of pollution information obtained with the causal network to determine each pollution path and the corresponding entity. According to the matching results, a search tree of the pollution path is constructed, with the detected pollution point as the root node. The branches of the tree represent the diffusion path or pollution source. Each node in the search tree represents the pollution state of a certain location, and the edge represents the diffusion path from one node to another.

[0027] S3.2: Initialize the search tree, set the value of each node to zero, start from the root node, use the confidence upper bound formula to calculate the UCB value of each edge, and select the edge with the highest UCB value to explore downward layer by layer, expand an unexplored node on the selected path to generate multiple groups of new nodes, and add each group of new nodes to the search tree;

[0028] S3.3: Simulate the pollutant diffusion path from the expanded node, and use the diffusion rules in the random walk or causal network to generate sample paths. If the simulated path successfully locates the pollution source, increase the benefit of the corresponding path and update the number of visits to all passed nodes. Repeat the selection, expansion, simulation and backtracking until the preset number of iterations is reached or the path benefit change value converges to the preset range.

[0029] S3.4: After the iteration, the path with the highest UCB value is selected as the optimal solution for pollution source location and diffusion path. The simulated pollution source location and diffusion path are visualized, and a detailed causal relationship description is provided in combination with the causal relationship network.

[0030] As a further solution of the present invention, the specific steps of optimizing the water supply and water use scheduling strategy by the resource optimization module are as follows:

[0031] S4.1: Collect water supply and demand data in the region, including reservoir storage, river water volume, water demand data, and constraints, and use control curves to generate initial decision paths based on the supply and demand relationship of water resource allocation;

[0032] S4.2: Initialize the control points of the control curve based on the historical data of water resource allocation and demand forecast, set the starting point of the control curve to the current supply and demand status, set the end point of the curve to the system equilibrium point, and set the initial distribution of intermediate control points according to experience or rules;

[0033] S4.3: Search and generate multiple allocation paths, simulate water supply and water use scheduling strategies in different time periods, calculate the utility of each path and optimize it, then simulate the utility values ​​of different paths, select the path with higher utility value in the initial path, adjust the control points in the control curve by gradient descent method, and improve the local path;

[0034] S4.4: Select the path with the highest utility value among all candidate paths as the final water resource allocation plan, and verify its adaptability under different demand scenarios, map the final path into a specific water supply and water use scheduling strategy, and generate a scheduling schedule and regional allocation plan.

[0035] As a further solution of the present invention, the specific steps of the solution evaluation module selecting the solution with the highest comprehensive benefit are as follows:

[0036] S5.1: Set the objective function based on pollution control effect, water resource conservation and carbon emission reduction benefits, add economic cost and implementation constraints, and randomly generate initial treatment plans based on the optimized water supply and water use scheduling strategy, including a combination of pollution control technology, water resource management measures and carbon emission reduction strategies;

[0037] S5.2: Set the initial temperature and cooling rate. Based on the current plan, modify the technical parameters and replace the governance measures, locally adjust and generate the neighborhood plan, calculate the objective function values ​​of the current plan and the new plan, and evaluate the benefit changes. If the benefit of the new plan is higher than that of the current plan, update the current plan to the new plan. If the benefit of the new plan is lower than that of the current plan, calculate the probability of selecting the new plan, and select and update the current plan based on the probability value.

[0038] S5.3: After each round of solution update is completed, the current temperature is updated according to the preset cooling rate, and the solution search and update are repeated until the current temperature drops to the set threshold or the solution benefit changes converge to the preset range after multiple times. Then, the update is stopped and the current best solution is output. The generated best solution is evaluated in detail and the governance plan for implementation is output.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] 1. In the present invention, each city or region stores the collected water quality, water quantity, rainfall, climate and carbon emission data locally. The central server initializes the global model, regards the smart carbon system of each city or region as a participating node, and sends the global model to each participating node. Each participating node initializes its own local model data according to the received global model information, trains the local model through the stored local data, and updates the local model architecture and parameters through the back propagation algorithm and the gradient descent algorithm with the goal of minimizing the local loss function. The central server collects the updated model parameters of all participating nodes, aggregates them into a new global model in a weighted average manner, inputs the historical water management data into the new global model, and calculates the loss value between the predicted value and the actual value of the global model. , to evaluate the performance of the global model. If the model performance reaches the preset target, the global model will be sent to each participating node, and the local model information of each node will be updated. If it does not meet the target, the global model will be re-sent, and the latest local data will be used for training and aggregation. The final global model will be sent to each participating node, and the original local model information will be updated to the latest global model information. After that, each local model accepts the latest collected data sets, and forward propagates each set of data through the local model to obtain the abnormal values ​​of the corresponding water quality parameters and future water quality trends. This can avoid the risk of privacy leakage, increase the possibility of data cooperation, significantly improve the generalization ability and accuracy of the prediction model, break down the data barriers between cities, promote cross-regional collaborative governance, and improve the consistency and synergy of governance solutions.

[0041] 2. The present invention obtains pollution information from sensor data, and matches each group of pollution information obtained with the causal network, determines each pollution path and the corresponding entity, constructs a search tree of the pollution path according to the matching results, takes the detected pollution point as the root node, and the branches of the tree represent the diffusion path or the pollution source. The search tree is initialized, and the value of each node is set to zero. Starting from the root node, the UCB value of each edge is calculated using the confidence upper bound formula, and the edge with the highest UCB value is selected to explore downward layer by layer, and an unexplored node is expanded on the selected path to generate multiple groups of new nodes, and each group of new nodes is added to the search tree, and the pollutant diffusion path is simulated starting from the expanded node, and the random walk or the diffusion rule in the causal network is used to generate samples. Path. If the simulated path successfully locates the pollution source, the benefit of the corresponding path is increased, and the number of visits to all passed nodes is updated. Then the selection, expansion, simulation and backtracking are repeated until the preset number of iterations is reached, or the change value of the path benefit converges to the preset range. After the iteration, the path with the highest UCB value is selected as the optimal solution for locating the pollution source and the diffusion path. The simulated pollution source location and diffusion path are visualized, and a detailed causal relationship description is provided in combination with the causal relationship network. It can comprehensively analyze the complex relationship behind the pollution incident, provide a scientific basis for tracing the source, and efficiently evaluate multiple potential pollution sources and their diffusion paths, improve real-time response capabilities and information transmission efficiency, and help the smart carbon system better coordinate various governance needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0043] Figure 1 This is a system block diagram of a smart carbon system based on water management proposed in the present invention. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0045] Example 1

[0046] Reference Figure 1 , a smart carbon system based on water governance, including collection and processing module, governance map module, training prediction module, evaluation prediction module, pollution source tracing module, resource optimization module, program evaluation module, scheduling control module and public information platform.

[0047] The collection and processing module is used to collect data on water quality, water quantity, rainfall, climate and carbon emissions through multi-source sensors, and process each set of collected data; the governance map module is used to integrate domain knowledge and collected data to establish a causal network linking water resource flow and carbon emissions.

[0048] Specifically, structured, semi-structured and unstructured data are collected from multiple sources, and semantic analysis of text information is performed through natural language processing technology. The CRF model is used to annotate text data and identify entities and their attributes and relationships in urban water governance. According to the relationship between entities extracted from the data, the basic triple form of the knowledge graph is generated, and the causal relationship between entities is judged through the Granger causality test. Based on water resource flow and carbon emission data, a weighted directed graph is constructed to represent the causal network, and the constructed triples are embedded and represented. The causal network is mapped to a low-dimensional vector space. A graph database is used to store the triple embedding results, and the implicit relationship between each group of water resource flow and carbon emission data is obtained through causal network reasoning.

[0049] The training and prediction module is used to train high-precision prediction models and perform real-time water quality anomaly detection and trend prediction based on the prediction models.

[0050] Specifically, each city or region will store the collected water quality, water quantity, rainfall, climate and carbon emission data locally. The central server will initialize the global model, regard the smart carbon system of each city or region as a participating node, and send the global model to each participating node. Each participating node will initialize its own local model data based on the received global model information, train the local model with the stored local data, and update the local model architecture and parameters through the back propagation algorithm and gradient descent algorithm with the goal of minimizing the local loss function. The central server will collect the updated model parameters of all participating nodes and aggregate them into a new global model in a weighted average manner. , input the historical water governance data into the new global model, and calculate the loss value between the global model prediction value and the actual value to evaluate the performance of the global model. If the model performance reaches the preset target, the global model will be sent to each participating node, and the local model information of each node will be updated. If it does not meet the target, the global model will be re-sent, and the latest local data will be used for training and aggregation. The final global model will be sent to each participating node, and the original local model information will be updated to the latest global model information. After that, each local model accepts the latest collected data sets, and forward propagates each data set through the local model to obtain the abnormal values ​​of the corresponding water quality parameters and future water quality trends.

[0051] The evaluation and prediction module is used to predict the impact of different governance options on carbon emissions.

[0052] Example 2

[0053] Reference Figure 1 , a smart carbon system based on water governance, including collection and processing module, governance map module, training prediction module, evaluation prediction module, pollution source tracing module, resource optimization module, program evaluation module, scheduling control module and public information platform.

[0054] The pollution source tracing module is used to locate pollution sources and simulate the diffusion path of pollutants.

[0055] Specifically, pollution information is obtained from sensor data, and each group of pollution information obtained is matched with the causal network to determine each pollution path and the corresponding entity. A search tree of the pollution path is constructed based on the matching results. The detected pollution point is used as the root node, and the branch of the tree represents the diffusion path or pollution source. Each node in the search tree represents the pollution state of a certain location, and the edge represents the diffusion path from one node to another. The search tree is initialized, and the value of each node is set to zero. Starting from the root node, the UCB value of each edge is calculated using the confidence upper bound formula, and the edge with the highest UCB value is selected to explore downward layer by layer, and an unexplored node is expanded on the selected path to generate multiple groups. New nodes are created and each group of new nodes is added to the search tree. The pollutant diffusion path is simulated starting from the expanded nodes. The sample path is generated using the diffusion rules in the random walk or causal network. If the simulated path successfully locates the pollution source, the benefit of the corresponding path is increased and the number of visits to all passed nodes is updated. Then, the selection, expansion, simulation and backtracking are repeated until the preset number of iterations is reached or the change in the path benefit converges to a preset range. After the iteration, the path with the highest UCB value is selected as the optimal solution for the pollution source location and diffusion path. The simulated pollution source location and diffusion path are visualized, and a detailed causal description is provided in combination with the causal network.

[0056] The resource optimization module is used to simulate various water resource allocation schemes and optimize water supply and water use scheduling strategies.

[0057] Specifically, water supply and demand data in the region are collected, including reservoir reserves, river water volume, water demand data, and constraints. According to the supply and demand relationship of water resource allocation, the control curve is used to generate an initial decision path. The control points of the control curve are initialized based on the historical data of water resource allocation and demand forecasts. The starting point of the control curve is set to the current supply and demand state, the end point of the curve is set to the system equilibrium point, and the initial distribution of intermediate control points is set according to experience or rules. Multiple allocation paths are searched and generated, and water supply and water use scheduling strategies in different time periods are simulated. The utility of each path is calculated and optimized. Then, by simulating the utility values ​​of different paths, the path with a higher utility value in the initial path is selected, and the control points in the control curve are adjusted by the gradient descent method. The local path is improved, and the path with the highest utility value among all candidate paths is selected as the final water resource allocation plan, and its adaptability under different demand scenarios is verified. The final path is mapped to a specific water supply and water use scheduling strategy, and a scheduling schedule and regional allocation plan are generated.

[0058] The solution evaluation module is used to generate a variety of governance solutions, evaluate their effects in pollution control, water conservation and carbon emission reduction, and select the solution with the highest overall benefits.

[0059] Specifically, an objective function is set based on pollution control effects, water resource savings and carbon emission reduction benefits, and economic costs and implementation constraints are added. According to the optimized water supply and water use scheduling strategy, an initial governance plan is randomly generated, including a combination of pollution control technology, water resource management measures and carbon emission reduction strategies, and the initial temperature and cooling rate are set. According to the current plan, the neighborhood plan is locally adjusted by modifying technical parameters and replacing governance measures. The objective function values ​​of the current plan and the new plan are calculated, and the benefit changes are evaluated. If the benefit of the new plan is higher than that of the current plan, the current plan is updated to the new plan. If the benefit of the new plan is lower than that of the current plan, the probability of selecting the new plan is calculated, and the current plan is updated according to the probability value. After each round of plan update is completed, the current temperature is updated according to the preset cooling rate, and the plan search and update are repeated until the current temperature drops to the set threshold or after multiple plan benefit changes converge to the preset range, the update is stopped and the current best plan is output. The generated best plan is evaluated in detail, and the governance plan for implementation is output.

[0060] The dispatching and control module is used to combine IoT devices to perform intelligent control of water pumps, sluice gates, and treatment equipment, and to accurately operate sewage treatment and rainwater dispatching; the public information platform is used to publish water governance progress, governance plans, and carbon emission data, and collect public feedback to optimize the system.

Claims

1. A smart carbon system based on water management, characterized in that: It includes collection and processing module, governance map module, training prediction module, evaluation prediction module, pollution source tracing module, resource optimization module, program evaluation module, dispatching control module and public information platform; The acquisition and processing module is used to collect water quality, water quantity, rainfall, climate and carbon emission data through multi-source sensors, and process each set of collected data; The governance graph module is used to integrate domain knowledge and collected data to establish a causal relationship network between water resource flow and carbon emissions; The training prediction module is used to train a high-precision prediction model and perform water quality anomaly detection and trend prediction in real time based on the prediction model; The evaluation and prediction module is used to predict the impact of different governance schemes on carbon emissions; The pollution source tracing module is used to locate pollution sources and simulate pollutant diffusion paths; The resource optimization module is used to simulate various water resource allocation schemes and optimize water supply and water use scheduling strategies; The scheme evaluation module is used to generate multiple governance schemes, evaluate their effects in terms of pollution control, water conservation and carbon emission reduction, and select the scheme with the highest comprehensive benefit; The dispatching control module is used to combine with IoT devices to perform intelligent control of water pumps, sluice gates, and treatment equipment, and accurately operate sewage treatment and rainwater dispatching; The public information platform is used to publish water governance progress, governance plans and carbon emission data, and collect public feedback to optimize the system.

2. According to claim 1, a smart carbon system based on water treatment is characterized in that: The specific steps of the governance map module to establish a causal relationship network between water resource flow and carbon emissions are as follows: S1.1: Collect structured, semi-structured and unstructured data from multiple sources, perform semantic analysis on text information through natural language processing technology, annotate text data using CRF model and identify entities, attributes and relationships in urban water governance; S1.2: Generate the basic triple form of the knowledge graph based on the relationship between entities extracted from the data, determine the causal relationship between entities through Granger causality test, and construct a weighted directed graph to represent the causal network based on water resource flow and carbon emission data; S1.3: The constructed triples are embedded and represented, the causal network is mapped to a low-dimensional vector space, a graph database is used to store the triple embedding results, and the implicit relationship between each group of water resource flow and carbon emission data is obtained through causal network reasoning.

3. The intelligent carbon system based on water treatment according to claim 2 is characterized in that: The specific steps of training the prediction module for water quality anomaly detection and trend prediction are as follows: S2.1: Each city or region stores the collected water quality, water quantity, rainfall, climate and carbon emission data locally. The central server initializes the global model, regards the smart carbon system of each city or region as a participating node, and sends the global model to each participating node; S2.2: Each participating node initializes its own local model data based on the received global model information, trains the local model through the stored local data, and updates the local model architecture and parameters through the back propagation algorithm and gradient descent algorithm with the goal of minimizing the local loss function; S2.3: The central server collects the updated model parameters of all participating nodes, aggregates them into a new global model in a weighted average manner, inputs the historical water management data into the new global model, and calculates the loss value between the global model prediction value and the actual value to evaluate the performance of the global model. If the model performance reaches the preset target, the global model is sent to each participating node and the local model information of each node is updated. If it does not meet the target, the global model is re-sent and the latest local data is used for training and aggregation; S2.4: The final global model is sent to each participating node, and the original local model information is updated to the latest global model information. After that, each local model accepts the latest collected data sets, and forward propagates each set of data through the local model to obtain the abnormal values ​​of the corresponding water quality parameters and future water quality trends.

4. The intelligent carbon system based on water treatment according to claim 2 is characterized in that: The specific steps of the pollution source tracing module to locate the pollution source and simulate the pollutant diffusion path are as follows: S3.1: Obtain pollution information from sensor data, and match each set of pollution information obtained with the causal network to determine each pollution path and the corresponding entity. According to the matching results, a search tree of the pollution path is constructed, with the detected pollution point as the root node. The branches of the tree represent the diffusion path or pollution source. Each node in the search tree represents the pollution state of a certain location, and the edge represents the diffusion path from one node to another. S3.2: Initialize the search tree, set the value of each node to zero, start from the root node, use the confidence upper bound formula to calculate the UCB value of each edge, and select the edge with the highest UCB value to explore downward layer by layer, expand an unexplored node on the selected path to generate multiple groups of new nodes, and add each group of new nodes to the search tree; S3.3: Simulate the pollutant diffusion path from the expanded node, and use the diffusion rules in the random walk or causal network to generate sample paths. If the simulated path successfully locates the pollution source, increase the benefit of the corresponding path and update the number of visits to all passed nodes. Repeat the selection, expansion, simulation and backtracking until the preset number of iterations is reached or the path benefit change value converges to the preset range. S3.4: After the iteration, the path with the highest UCB value is selected as the optimal solution for pollution source location and diffusion path. The simulated pollution source location and diffusion path are visualized, and a detailed causal relationship description is provided in combination with the causal relationship network.

5. The intelligent carbon system based on water treatment according to claim 4 is characterized in that: The specific steps of optimizing the water supply and water use scheduling strategy by the resource optimization module are as follows: S4.1: Collect water supply and demand data in the region, including reservoir storage, river water volume, water demand data, and constraints, and use control curves to generate initial decision paths based on the supply and demand relationship of water resource allocation; S4.2: Initialize the control points of the control curve based on the historical data of water resource allocation and demand forecast, set the starting point of the control curve to the current supply and demand status, set the end point of the curve to the system equilibrium point, and set the initial distribution of intermediate control points according to experience or rules; S4.3: Search and generate multiple allocation paths, simulate water supply and water use scheduling strategies in different time periods, calculate the utility of each path and optimize it, then simulate the utility values ​​of different paths, select the path with higher utility value in the initial path, adjust the control points in the control curve by gradient descent method, and improve the local path; S4.4: Select the path with the highest utility value among all candidate paths as the final water resource allocation plan, and verify its adaptability under different demand scenarios, map the final path into a specific water supply and water use scheduling strategy, and generate a scheduling schedule and regional allocation plan.

6. The intelligent carbon system based on water treatment according to claim 5 is characterized in that: The specific steps of the scheme evaluation module for selecting the scheme with the highest comprehensive benefit are as follows: S5.1: Set the objective function based on pollution control effect, water resource conservation and carbon emission reduction benefits, add economic cost and implementation constraints, and randomly generate initial treatment plans based on the optimized water supply and water use scheduling strategy, including a combination of pollution control technology, water resource management measures and carbon emission reduction strategies; S5.2: Set the initial temperature and cooling rate. Based on the current plan, modify the technical parameters and replace the governance measures, locally adjust and generate the neighborhood plan, calculate the objective function values ​​of the current plan and the new plan, and evaluate the benefit changes. If the benefit of the new plan is higher than that of the current plan, update the current plan to the new plan. If the benefit of the new plan is lower than that of the current plan, calculate the probability of selecting the new plan, and select and update the current plan based on the probability value. S5.3: After each round of solution update is completed, the current temperature is updated according to the preset cooling rate, and the solution search and update are repeated until the current temperature drops to the set threshold or the solution benefit changes converge to the preset range after multiple times. Then, the update is stopped and the current best solution is output. The generated best solution is evaluated in detail and the governance plan for implementation is output.

Images