Data-driven water resource allocation consensus decision support method and device

Abstract

The invention relates to a data-driven water resource allocation consensus decision support method and device, and belongs to the technical field of water resource management.The data-driven water resource allocation consensus decision support method comprises the steps that water resource monitoring data and decision maker information of a target water area are acquired; constructing a decision maker relation network based on decision maker information of the target water area, and performing group division on the decision maker relation network based on a Louvain community detection algorithm; and allocating a depth Q network to each group divided in the decision maker relation network, and determining a water resource allocation consensus decision of each group divided in the decision maker relation network based on the water resource monitoring data of the target water area and the depth Q network. According to the invention, the efficiency and accuracy of water resource allocation consensus decision can be effectively improved.

Description

Technical Field

[0001] This invention relates to the field of water resource management technology, and in particular to a data-driven consensus decision support method and apparatus for water resource allocation. Background Technology

[0002] Water resources, as a fundamental element of human social development, are facing increasingly complex management and decision-making processes. With the acceleration of global climate change, population growth, and industrialization, water scarcity and allocation conflicts have become international challenges. Existing water resource decision support systems primarily rely on traditional hydrological models, expert experience, and static optimization algorithms, such as linear programming-based water supply scheduling models or simulated annealing algorithms for resource allocation. While these methods demonstrate some effectiveness in small-scale decision-making environments, they reveal significant limitations in large-scale, multi-stakeholder consensus-based decision-making.

[0003] The limitations of existing technologies mainly include: the decision-making process lacks integration of real-time data, making it unable to cope with uncertainties such as sudden floods or droughts; consensus mechanisms rely on human intervention, making it difficult to handle heterogeneous opinions from a large number of stakeholders; and optimization algorithms have fixed parameters, making them unable to learn adaptively, which leads to increased decision-making bias.

[0004] Therefore, improving the efficiency and accuracy of consensus-based decision-making in water resource allocation has become an urgent technical problem to be solved. Summary of the Invention

[0005] In view of this, it is necessary to provide a data-driven consensus decision support method and device for water resource allocation to solve the problem of low efficiency and accuracy of existing consensus decision-making for water resource allocation.

[0006] To address the aforementioned problems, in a first aspect, the present invention provides a data-driven consensus decision support method for water resource allocation, comprising: Acquire water resource monitoring data and decision-maker information for the target water area. The water resource monitoring data includes water quality data, hydrological data, and environmental data, while the decision-maker information includes the decision-maker's identity information and historical decision-making records. A decision-maker relationship network is constructed based on decision-maker information in the target water area, and the network is divided into groups based on the Louvain community detection algorithm. Assign a deep Q-network to each group in the decision-maker relationship network, and determine the consensus decision on water resource allocation for each group in the decision-maker relationship network based on the water resource monitoring data of the target water area and the deep Q-network.

[0007] In one possible implementation, the construction of the decision-maker relationship network based on the decision-maker information of the target water area includes: Based on decision-maker information for the target water area, determine the historical cooperation records among decision-makers; Construct a network of relationships among decision-makers based on their historical records of collaboration.

[0008] In one possible implementation, the grouping of the decision-maker relationship network based on the Louvain community detection algorithm includes: Based on the Louvain community detection algorithm, the modularity index of each group is calculated to divide the decision-maker relationship network into groups.

[0009] In one possible implementation, determining the influence weight of each decision-maker based on the PageRank algorithm includes: The PageRank algorithm is used to determine the out-degree and in-degree of each decision-maker in the decision-maker relationship network. The influence weight of each decision-maker is determined based on the out-degree and in-degree of each decision-maker in the decision-maker relationship network.

[0010] In one possible implementation, the consensus decision-making process for water resource allocation, based on water resource monitoring data of the target water area and a deep Q-network to determine each group within the decision-maker relationship network, includes: Based on water resource monitoring data of the target water area and the deep Q network, the allocation actions of each group in the decision-maker relationship network are selected, and the allocation actions of each group are optimized based on the reward function until the decision-maker relationship network reaches the consensus threshold. Based on the allocation actions of each group when the decision-maker relationship network reaches the consensus threshold, the consensus decision on water resource allocation for each group is determined.

[0011] In one possible implementation, determining the consensus decision on water resource allocation for each group based on the allocation actions of each group when the decision-maker relationship network reaches a consensus threshold includes: When the decision-makers' relationship network reaches a consensus threshold, the allocation actions of each group and the influence weight of each decision-maker are determined as the consensus decision on water resource allocation for each group. The influence weight of each decision-maker is determined based on the PageRank algorithm.

[0012] In one possible implementation, the reward function is constructed based on harmony and the total number of iterations.

[0013] On the other hand, the present invention also provides a data-driven consensus decision support device for water resource allocation, comprising: The acquisition module is used to acquire water resource monitoring data and decision-maker information for the target water area. The water resource monitoring data includes water quality data, hydrological data, and environmental data, while the decision-maker information includes the decision-maker's identity information and historical decision-making records. The module is used to construct a decision-maker relationship network based on decision-maker information in the target water area, and to perform group segmentation of the decision-maker relationship network based on the Louvain community detection algorithm; The determination module is used to assign a deep Q-network to each group in the decision-maker relationship network, and to determine the consensus decision on water resource allocation for each group in the decision-maker relationship network based on the water resource monitoring data of the target water area and the deep Q-network.

[0014] Secondly, the present invention also provides a dispensing device, including a memory and a processor, wherein, The memory is used to store programs; The processor, coupled to the memory, is used to execute the program stored in the memory to implement the steps in the data-driven water resource allocation consensus decision support method described in any of the above implementations.

[0015] Thirdly, the present invention also provides a computer-readable storage medium for storing a computer-readable program or instruction, which, when executed by a processor, can implement the steps in the data-driven water resource allocation consensus decision support method described in any of the above implementations.

[0016] The beneficial effects of this invention are as follows: The data-driven water resource allocation consensus decision support method and apparatus provided by this invention first acquires water resource monitoring data and decision-maker information of the target water area to provide data support for subsequent water resource allocation consensus decision support, thereby ensuring the accuracy of water resource allocation consensus decision. Then, it realizes water resource allocation consensus decision support by constructing a decision-maker relationship network, thereby improving the efficiency of water resource allocation consensus decision. Finally, it optimizes the decision by using water resource monitoring data of the target water area and a deep Q-network, thereby determining the water resource allocation consensus decision for each group divided in the decision-maker relationship network. This invention can effectively improve the efficiency and accuracy of water resource allocation consensus decision. Attached Figure Description

[0017] Figure 1 A schematic flowchart of an embodiment of the data-driven water resource allocation consensus decision support method provided by the present invention; Figure 2 A schematic flowchart of an embodiment of the consensus decision support process for water resource allocation provided by the present invention; Figure 3 A schematic diagram of an embodiment of the data-driven water resource allocation consensus decision support device provided by the present invention; Figure 4 This is a schematic diagram of an embodiment of the dispensing equipment provided by the present invention. Detailed Implementation

[0018] 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 a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0019] In the description of the embodiments of the present invention, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0020] The terms "first," "second," etc., used in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a technical feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature.

[0021] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0022] Water resources, as a fundamental element of human social development, are facing increasingly complex management and decision-making processes. With the acceleration of global climate change, population growth, and industrialization, water scarcity and allocation conflicts have become international challenges. Existing water resource decision support systems primarily rely on traditional hydrological models, expert experience, and static optimization algorithms, such as linear programming-based water supply scheduling models or simulated annealing algorithms for resource allocation. While these methods demonstrate some effectiveness in small-scale decision-making environments, they reveal significant limitations in large-scale, multi-stakeholder consensus-based decision-making.

[0023] The current state of existing technologies is mainly reflected in the following aspects: On the one hand, traditional decision support systems monitor and predict water resources by collecting hydrological data, meteorological information, and socio-economic indicators. However, these systems are often limited to static data processing and cannot respond to dynamic environmental changes in real time. On the other hand, related research has introduced social network analysis and group decision-making models to handle the aggregation of opinions from multiple parties. However, when these methods are applied in the field of water resources, they ignore the dynamic adaptability of data-driven approaches, resulting in low efficiency in the consensus process.

[0024] The limitations of existing technologies mainly include: the decision-making process lacks integration of real-time data, making it unable to cope with uncertainties such as sudden floods or droughts; consensus mechanisms rely on human intervention, making it difficult to handle heterogeneous opinions from a large number of stakeholders; and optimization algorithms have fixed parameters, making them unable to learn adaptively, which leads to increased decision-making bias.

[0025] The shortcomings of existing technologies are as follows: 1. Existing water resource decision support systems are inefficient when handling large-scale group decisions. In the water resource allocation decision-making process in the Yangtze River Basin, 80 decision-makers are involved. Traditional group decision-making methods require 12-16 iterations to reach a consensus, with a decision-making cycle of 3-5 days. When the number of decision-makers exceeds 50, the computational complexity of traditional methods increases from O(n²) to O(n³), leading to a sharp decline in decision-making efficiency.

[0026] 2. Existing systems use a fixed adjustment rate for preference adjustments, which cannot adapt to sudden water resource events and changes in decision-makers' preferences. Traditional systems cannot adjust strategies in real time, leading to a 15-20% increase in decision-making bias.

[0027] 3. In water resource allocation decisions, agricultural users and industrial enterprises often resist preference adjustments, resulting in a consensus level that is 0.1-0.2 lower than the overall consensus level, which prolongs the overall consensus-reaching time by 40-60%. The existing system lacks an effective mechanism for identifying and handling non-cooperative behavior.

[0028] 4. Decision-makers have social network characteristics such as trust relationships and influence spread, but existing systems use simple weighted averaging, ignoring social network analysis, resulting in decision results lacking a social basis.

[0029] 5. The existing system mainly relies on static hydrological models and cannot integrate real-time monitoring data, resulting in a 10-15% increase in decision-making bias.

[0030] To address the above deficiencies, this invention provides a data-driven consensus decision support method and apparatus for water resource allocation, which will be described below.

[0031] Figure 1 A schematic flowchart of an embodiment of the data-driven water resource allocation consensus decision support method provided by the present invention is shown below. Figure 1 As shown, the data-driven consensus decision support method for water resource allocation includes: S101. Obtain water resource monitoring data and decision-maker information for the target water area. The water resource monitoring data includes water quality data, hydrological data, and environmental data. The decision-maker information includes the decision-maker's identity information and historical decision-making records.

[0032] It should be noted that the data-driven consensus decision support method for water resource allocation provided by this invention can be applied to water resource allocation scenarios, such as the allocation of domestic water and industrial water.

[0033] When providing consensus-based decision support for water resource allocation, the first step is to obtain water resource monitoring data and decision-maker information for the target water area. For example, real-time monitoring networks can be used to collect water quality data, including key indicators such as pH, dissolved oxygen, total phosphorus, ammonia nitrogen, and chemical oxygen demand. Simultaneously, hydrological information such as flow rate, water level, water consumption, and water storage can be collected, along with environmental data on pollution sources, including discharge volume, pollutant concentration, and discharge time. Meteorological data such as rainfall, temperature, humidity, and wind speed can also be obtained. Regarding the collection of decision-maker information, basic information on all parties involved in the decision-making process should be gathered, including names, organizations, professional backgrounds, and contact information. Historical decision-making records should be compiled, including past participation in water resource decision-making projects, decision preferences, and cooperation history. Decision-maker profiles should be established, including their professional fields, impact assessments, and stakeholder analysis. This data support provides crucial data for subsequent consensus-based decision support for water resource allocation, ensuring the accuracy of the decisions.

[0034] S102. Construct a decision-maker relationship network based on the decision-maker information of the target water area, and perform group division of the decision-maker relationship network based on the Louvain community detection algorithm.

[0035] It should be noted that after obtaining the decision-maker information for the target water area, social network analysis and community structure identification can be performed based on this information. First, a relationship network is established through historical cooperation records among decision-makers. Then, the Louvain community detection algorithm is used to automatically segment decision-maker groups, calculate modularity indices, and optimize the community segmentation results. By constructing a decision-maker relationship network to support consensus-based decision-making in water resource allocation, the efficiency of consensus-based decision-making in water resource allocation can be improved.

[0036] S103. Assign a deep Q-network to each group in the decision-maker relationship network, and determine the consensus decision on water resource allocation for each group in the decision-maker relationship network based on the water resource monitoring data of the target water area and the deep Q-network.

[0037] It should be noted that, in order to further improve the efficiency of consensus decision-making in water resource allocation, this invention also assigns a depth Q network to each group in the decision-maker relationship network, and optimizes the decision by using water resource monitoring data of the target water area and the depth Q network, thereby determining the consensus decision on water resource allocation for each group in the decision-maker relationship network.

[0038] In summary, the data-driven water resource allocation consensus decision support method provided in this embodiment of the invention first acquires water resource monitoring data and decision-maker information for the target water area to provide data support for subsequent water resource allocation consensus decision support, thereby ensuring the accuracy of water resource allocation consensus decision. Next, it constructs a decision-maker relationship network to achieve water resource allocation consensus decision support, improving the efficiency of water resource allocation consensus decision. Finally, it optimizes decisions using water resource monitoring data of the target water area and a deep Q-network, thereby determining the water resource allocation consensus decision for each group segmented in the decision-maker relationship network. This invention can effectively improve the efficiency and accuracy of water resource allocation consensus decision.

[0039] In some embodiments of the present invention, the construction of a decision-maker relationship network based on decision-maker information of the target water area includes: Based on decision-maker information for the target water area, determine the historical cooperation records among decision-makers; Construct a network of relationships among decision-makers based on their historical records of collaboration.

[0040] It should be noted that when constructing a decision-maker relationship network based on the decision-maker information of the target water area, the historical cooperation records between decision-makers can be determined based on the decision-maker information of the target water area, and then the decision-maker relationship network can be constructed through the historical cooperation records between decision-makers. In the decision-maker relationship network, the edge weights can represent the relationship strength.

[0041] In some embodiments of the present invention, the grouping of the decision-maker relationship network based on the Louvain community detection algorithm includes: Based on the Louvain community detection algorithm, the modularity index of each group is calculated to divide the decision-maker relationship network into groups.

[0042] It should be noted that when dividing the decision-maker relationship network into groups based on the Louvain community detection algorithm, the Louvain community detection algorithm can be used to automatically divide the decision-maker groups, calculate the modularity index, optimize the community division results, and determine the optimal number of communities.

[0043] In some embodiments of the present invention, determining the influence weight of each decision-maker based on the PageRank algorithm includes: The PageRank algorithm is used to determine the out-degree and in-degree of each decision-maker in the decision-maker relationship network. The influence weight of each decision-maker is determined based on the out-degree and in-degree of each decision-maker in the decision-maker relationship network.

[0044] It should be noted that when determining the influence weight of each decision-maker according to the PageRank algorithm, the PageRank algorithm can be used to calculate the influence of each decision-maker in the network, taking into account in-degree and out-degree, assessing the decision-maker's authority and dissemination ability, and then performing normalization processing to obtain a standardized influence weight.

[0045] In some embodiments of the present invention, the step of determining the consensus decision on water resource allocation for each group segmented in the decision-maker relationship network based on water resource monitoring data of the target water area and a deep Q-network includes: Based on water resource monitoring data of the target water area and the deep Q network, the allocation actions of each group in the decision-maker relationship network are selected, and the allocation actions of each group are optimized based on the reward function until the decision-maker relationship network reaches the consensus threshold. Based on the allocation actions of each group when the decision-maker relationship network reaches the consensus threshold, the consensus decision on water resource allocation for each group is determined.

[0046] It should be noted that when determining the consensus decision on water resource allocation for each group within the decision-maker relationship network based on water resource monitoring data of the target water area and the deep Q-network, the allocation actions for each group can be selected using the water resource monitoring data and the deep Q-network. Then, the allocation actions for each group are optimized using a reward function until the decision-maker relationship network reaches a consensus threshold. After the decision-maker relationship network reaches the consensus threshold, the consensus decision on water resource allocation for each group is determined based on its allocation actions.

[0047] In some embodiments of the present invention, determining the consensus decision on water resource allocation for each group based on the allocation actions of each group when the decision-maker relationship network reaches a consensus threshold includes: When the decision-makers' relationship network reaches a consensus threshold, the allocation actions of each group and the influence weight of each decision-maker are determined as the consensus decision on water resource allocation for each group. The influence weight of each decision-maker is determined based on the PageRank algorithm.

[0048] It should be noted that when determining the consensus decision on water resource allocation for each group based on the allocation actions of each group when the decision-maker relationship network reaches the consensus threshold, the allocation actions of each group when the decision-maker relationship network reaches the consensus threshold and the influence weight of each decision-maker can be used to determine the consensus decision on water resource allocation for each group. The influence weight of each decision-maker can be determined using the PageRank algorithm.

[0049] In some embodiments of the present invention, the reward function is constructed based on harmony and the total number of iterations.

[0050] It should be noted that the reward function used to optimize the allocation actions for each group can be constructed using harmony and the total number of iterations.

[0051] To address the shortcomings of existing consensus-based decision-making technologies for water resource allocation, this invention proposes a water resource consensus-based decision support system framework based on multi-agent reinforcement learning. This method automatically clusters large-scale decision-makers into communities using the Louvain algorithm, assigning a deep Q-network agent to each community to achieve parallel decision optimization and significantly improve decision-making efficiency. The invention also proposes a dynamic adaptive mechanism based on harmony. The reward function is designed to consider short-term harmony and the total number of iterations. Agents dynamically adjust their strategies through experience replay and soft update mechanisms, reducing the response time from 3-5 days in traditional methods to 1-2 days. Furthermore, this invention proposes a weighted penalty mechanism. A non-cooperative behavior identification algorithm is established, and weighted penalties are applied to non-cooperative communities. This mechanism increases the willingness of non-cooperative participants to cooperate by more than 60%. The invention also proposes an integrated social network analysis method. The PageRank algorithm is integrated to calculate decision-maker influence, and the Louvain community detection algorithm is combined to automatically determine the optimal number of communities, improving the social basis of the decision results. Finally, this invention proposes a data-driven decision optimization method. Establish a real-time data integration framework, which integrates multi-source data such as water quality monitoring data, water allocation data, and pollution source information, and realizes data-driven dynamic decision-making through a neural network architecture.

[0052] This invention proposes a data-driven water resources consensus decision support system design, which realizes the creation of a decision support system by acquiring domain knowledge from water resources monitoring data, and optimizes the consensus reaching process to improve decision efficiency and the accuracy of results. It includes a data acquisition and preprocessing module, a social network analysis module, a multi-agent reinforcement learning module, a consensus reaching module, and a decision output module.

[0053] The data acquisition and preprocessing module collects data on water quality, water quantity, and pollution sources through a real-time monitoring system. Then, through data cleaning and preprocessing, it achieves more comprehensive data integration. This integrated use of multi-source data can improve the efficiency and accuracy of data processing.

[0054] In the community detection process, the social network analysis module automatically segments decision-makers using the Louvain algorithm, avoiding the limitations of subjectively setting the number of communities. In the influence calculation process, the PageRank algorithm is used to consider the trust relationship and influence propagation among decision-makers, which can strengthen the social basis of the decision results.

[0055] The multi-agent reinforcement learning module implements the construction of agents based on deep Q-networks to complete state observation and action selection tasks. Each agent is responsible for managing the decision optimization of its corresponding community. A harmony-based reward function is designed to help the agent converge quickly and learn stably during training, thereby improving the accuracy and efficiency of decision-making.

[0056] The consensus-building module aggregates the preference matrices of each community based on influence weights, thus calculating the global consensus level. A weighted penalty mechanism is established to identify decision-makers' behaviors based on user input, and returns adjusted weight allocations according to certain rules, forming a dynamic consensus-building system.

[0057] The decision output module sorts and weights the final water resource management plans, generating specific decision recommendations and implementation paths. The data-driven water resources consensus decision support system proposed in this invention includes the following steps: 1. Collection of water resources data.

[0058] Through real-time monitoring systems, historical databases, and other channels, highly reliable water resource-related data are collected. Ultimately, water quality monitoring data, water allocation data, pollution source information, and meteorological data are selected as data sources, including key indicators such as pH value, dissolved oxygen, total phosphorus, ammonia nitrogen, and chemical oxygen demand; hydrological information such as flow rate, water level, water consumption, and water storage; environmental data such as discharge volume, pollutant concentration, and discharge time; and meteorological forecast information such as rainfall, temperature, humidity, and wind speed.

[0059] 2. Collection of information on decision-makers.

[0060] Collect basic information on all parties involved in the decision-making process, including name, organization, professional background, and contact information; compile historical decision-making records, including past water resource decision-making projects, decision preferences, and cooperation history; establish decision-maker profiles, including professional fields, impact assessments, and stakeholder analysis. Each decision-maker conducts pairwise comparative evaluations of water resource management plans, constructing a reciprocal preference matrix to express their judgment of the relative importance of the plans, verifying the consistency of the preference matrix, and ensuring the rationality of the decision-making logic.

[0061] 3. Social network analysis.

[0062] A relationship network is established based on historical collaboration records among decision-makers. This network analyzes professional relevance, trust relationships, and influence propagation paths, constructing a weighted directed graph where edge weights represent relationship strength. The Louvain community detection algorithm is used to automatically segment decision-maker groups, calculate modularity indices, optimize community segmentation results, and determine the optimal number of communities, avoiding subjective settings. The PageRank algorithm is used to calculate the influence of each decision-maker in the network, considering in-degree and out-degree, assessing the decision-maker's authority and dissemination ability, and normalizing the results to obtain standardized influence weights.

[0063] 4. Construction of a multi-agent reinforcement learning system.

[0064] A deep Q-network agent is assigned to each detected community. Each agent is responsible for managing the decision optimization of its corresponding community. A communication mechanism is established between agents to support collaborative decision-making. The number of neurons in the input layer is set to the square of the number of schemes plus 2. The hidden layer structure is configured with 64 neurons in the first hidden layer and 32 neurons in the second hidden layer. The output layer is defined as 11 discrete action levels, corresponding to different adjustment rate selections. The network weights are initialized using a random initialization strategy. The learning rate is set to 0.001 to control the network weight update speed; the discount factor is 0.9 to balance short-term and long-term rewards; the exploration rate decays linearly from 1.0 to 0.01 to balance exploration and exploitation; the experience replay buffer size is set to 10,000 to 100,000; the mini-batch training size is set to 256; and the target network update rate is set to 0.01.

[0065] 5. Consensus-reaching process.

[0066] The agent observes the current decision-making environment state, obtains a flattened representation of the preference matrix, calculates the global and local consensus levels, and analyzes network topology features and community structure. Based on the current state, the agent selects adjustment actions, employing a strategy that balances exploration and exploitation, choosing from 11 discrete adjustment rates to execute preference adjustments, update the decision-maker's preference expression, and calculate the adjusted preference matrix. A reward signal is calculated based on harmony, with the reward function considering short-term harmony and the total number of iterations. Experience is stored in a replay buffer; when the buffer is large enough, mini-batch training is performed to update the target network parameters, implementing a soft update mechanism. A new global consensus level is calculated, non-cooperative communities are identified, a weighted penalty mechanism is applied, and it is checked whether a consensus threshold (0.90 to 0.95) has been reached. If the threshold is not reached, the process returns to the first step to continue iterating; if the threshold is reached, the process proceeds to the result output stage.

[0067] 6. Output of decision results.

[0068] The weights of each scheme are calculated based on the final consensus preference matrix, and the water resource management schemes are ranked to generate a priority list and weight distribution. The technical feasibility and economic rationality of the optimal scheme are analyzed, a specific implementation path and timeline are formulated, potential risks and countermeasures are identified, and a detailed decision recommendation report is generated. A time-series diagram of the decision-making process is provided, showing the convergence curve of the consensus-reaching process, visualizing the weight changes of each community, generating system performance comparison charts, and providing an interactive results display interface.

[0069] The specific algorithm of this invention in the consensus decision support process for water resource allocation is as follows:

[0070] Combination Figure 2 The water resource allocation consensus decision support process provided by this invention specifically includes the following steps: 1. Data input stage.

[0071] The data input phase primarily involves the collection and preprocessing of water resource monitoring data and decision-maker information. The system collects water quality monitoring data through a real-time monitoring network, including key indicators such as pH, dissolved oxygen, total phosphorus, ammonia nitrogen, and chemical oxygen demand. It also collects hydrological information such as flow rate, water level, water consumption, and water storage, as well as environmental data on pollution sources, including discharge volume, pollutant concentration, and discharge time. Furthermore, it acquires meteorological data such as rainfall, temperature, humidity, and wind speed, along with weather forecasts. Regarding decision-maker information collection, the system gathers basic information from all parties involved in the decision-making process, including name, organization, professional background, and contact information. It also compiles historical decision-making records, including past participation in water resource decision-making projects, decision preferences, and cooperation history, and establishes decision-maker profiles including professional fields, impact assessments, and stakeholder analysis. Each decision-maker conducts pairwise comparative evaluations of water resource management plans, constructing a reciprocal preference matrix to express their judgment of the relative importance of each plan, verifying the consistency of the preference matrix, and ensuring the rationality of the decision-making logic.

[0072] 2. Network analysis phase.

[0073] The network analysis phase utilizes collected decision-maker information to perform social network analysis and community structure identification. The system first establishes a relationship network based on historical cooperation records among decision-makers, analyzing professional relevance, trust relationships, and influence propagation paths to construct a weighted directed graph, with edge weights representing relationship strength. Then, the Louvain community detection algorithm is used to automatically segment decision-maker groups, calculate modularity indices, optimize community segmentation results, and determine the optimal number of communities, avoiding subjective settings. Finally, the PageRank algorithm is used to calculate the influence of each decision-maker in the network, considering in-degree and out-degree to assess the decision-maker's authority and propagation ability, and normalize the data to obtain standardized influence weights. The PageRank algorithm's network parameters have a damping factor β=0.85 and a convergence threshold ε=10^-6.

[0074] 3. Agent initialization phase.

[0075] During the agent initialization phase, a deep Q-network agent is assigned to each detected community. Each agent is responsible for managing the decision optimization of its corresponding community, establishing a communication mechanism between agents to support collaborative decision-making. Regarding neural network configuration, the system sets the number of input layer neurons to the square of the number of schemes plus 2, configures the hidden layer structure as follows: 64 neurons in the first hidden layer and 32 neurons in the second hidden layer. The output layer is defined as having 11 discrete action levels, corresponding to different adjustment rate selections. Network weights are initialized using a random initialization strategy. Reinforcement learning parameter settings include a learning rate of 0.001 to control the network weight update speed; a discount factor of 0.9 to balance short-term and long-term rewards; an exploration rate that linearly decays from 1.0 to 0.01 to balance exploration and exploitation; an experience replay buffer size of 10,000 to 100,000; a mini-batch training size of 256; and a target network update rate of 0.01.

[0076] 4. Consensus-reaching stage.

[0077] The consensus-building phase is the core processing stage of the system, achieving dynamic consensus through multi-agent reinforcement learning. The agents first observe the current decision-making environment, obtain a flattened representation of the preference matrix, calculate the global and local consensus levels, and analyze network topology and community structure. Then, based on the current state, they select adjustment actions, employing a strategy that balances exploration and exploitation, choosing from 11 discrete adjustment rates to execute preference adjustments, update the decision-makers' preference expressions, and calculate the adjusted preference matrix. In the reward calculation and learning phase, the system calculates reward signals based on harmony, with the reward function considering short-term harmony and the total number of iterations. Experience is stored in a replay buffer. When the buffer is large enough, mini-batch training is performed to update the target network parameters, implementing a soft update mechanism. Finally, consensus evaluation and iteration are performed, calculating the new global consensus level, identifying non-cooperative communities, applying a weighted penalty mechanism, and checking if the consensus threshold (0.90 to 0.95, penalty parameter φ=5, adjustment rate range [0,1]) has been reached. If the threshold is not reached, the system returns to the first step to continue iteration; if the threshold is reached, the system enters the result output stage.

[0078] 5. Results output stage.

[0079] The results output phase organizes and displays the system's processing results. First, the system calculates the weights of each scheme based on the final consensus preference matrix, ranks the water resource management schemes, and generates a priority list and weight distribution. Then, it analyzes the technical feasibility and economic rationality of the optimal scheme, formulates specific implementation paths and timelines, identifies potential risks and countermeasures, and generates a detailed decision-making recommendation report. Finally, it provides a time-series diagram of the decision-making process, displaying the convergence curve of the consensus-reaching process.

[0080] This invention proposes a framework for a water resources consensus decision support system based on multi-agent reinforcement learning. Using high-quality water resources monitoring data as the data source helps ensure the comprehensiveness, authority, and reliability of the data; a social network analysis module is created to systematically present the social relationships among decision-makers in a structured manner, improving the visualization of the decision-making process; and multi-agent reinforcement learning is combined with the consensus-building process to provide water resources managers with more professional and targeted decision support.

[0081] This invention proposes a dynamic adaptive mechanism based on harmony. Different adjustment strategies are selected according to changes in the decision-making environment, and an experience replay mechanism is used to further learn and optimize the data. The combined use of these two mechanisms can improve the efficiency of the decision-making process and the accuracy and stability of the results, contributing to improved decision quality and more efficient subsequent applications.

[0082] This invention proposes a weighted penalty mechanism. In the process of identifying non-cooperative behavior, it comprehensively employs consensus level comparison and weight adjustment methods. The penalty parameters are adjusted based on decision-maker behavior, task requirements, and algorithm functionality to achieve optimal processing results, realizing high accuracy in identifying non-cooperative behavior and reducing the complexity of subsequent tasks. In the weight adjustment stage, non-cooperative communities are processed using a combination of weighted penalties and normalization to obtain adjusted weight allocations. Weighted penalties ensure the accuracy of the processing results, while normalization allows for verification of the applicability of these weights through practical cases, thereby improving the credibility of the weight allocation. The combined use of both methods can more comprehensively cover various situations in the decision-making process. This weighted penalty mechanism improves the efficiency of consensus reaching and the fairness of the results.

[0083] This invention proposes an integrated method for social network analysis. During community detection, the Louvain algorithm is used to automatically segment decision-maker groups, avoiding the limitations of subjectively setting the number of communities. In influence calculation, the PageRank algorithm is used to consider trust relationships and influence propagation among decision-makers, strengthening the social basis of the decision outcomes. This integrated social network analysis method improves the social acceptability and credibility of decision results.

[0084] This invention proposes a data-driven decision optimization method. A real-time data integration framework is established, incorporating multi-source data such as water quality monitoring data, water allocation data, and pollution source information. A neural network architecture is used to achieve data-driven dynamic decision-making. This data-driven approach improves the adaptability and accuracy of decision-making, enabling it to better cope with complex and ever-changing water resource management environments.

[0085] This invention has achieved significant technical results in a water resources allocation case study of the Yangtze River Basin: the decision-making time has been shortened from 3.5 days to 1.8 days using traditional methods, the number of iterations has been reduced from 14.2 to 8.7, the final consensus level has increased from 0.87 to 0.93, the accuracy of non-cooperative identification has increased from 65.2% to 92.7%, and the system scalability has increased from supporting 30 participants to supporting 150 participants. This fully demonstrates the effectiveness and advancement of this invention in water resources consensus decision support.

[0086] To better implement the data-driven water resource allocation consensus decision support method in the embodiments of the present invention, based on the data-driven water resource allocation consensus decision support method, the corresponding method is as follows: Figure 3 As shown, this embodiment of the invention also provides a data-driven water resource allocation consensus decision support device. The data-driven water resource allocation consensus decision support device 300 includes: The acquisition module 301 is used to acquire water resource monitoring data and decision-maker information for the target water area. The water resource monitoring data includes water quality data, hydrological data and environmental data, and the decision-maker information includes the decision-maker's identity information and historical decision records. Module 302 is used to construct a decision-maker relationship network based on decision-maker information in the target water area, and to perform group segmentation of the decision-maker relationship network based on the Louvain community detection algorithm; The determination module 303 is used to assign a deep Q-network to each group in the decision-maker relationship network, and to determine the consensus decision on water resource allocation for each group in the decision-maker relationship network based on the water resource monitoring data of the target water area and the deep Q-network.

[0087] The data-driven water resource allocation consensus decision support device 300 provided in the above embodiments can realize the technical solutions described in the above data-driven water resource allocation consensus decision support method embodiments. The specific implementation principles of each module or unit can be found in the corresponding content in the above data-driven water resource allocation consensus decision support method embodiments, which will not be repeated here.

[0088] like Figure 4 As shown, the present invention also provides a dispensing device 400. The dispensing device 400 includes a processor 401, a memory 402, and a display 403. Figure 4 Only some components of the dispatching equipment 400 are shown, but it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead.

[0089] In some embodiments, processor 401 may be a central processing unit (CPU), microprocessor, or other data processing chip, used to run program code stored in memory 402 or process data, such as the data-driven water resource allocation consensus decision support method of the present invention.

[0090] In some embodiments, processor 401 may be a single server or a group of servers. The server group may be centralized or distributed. In some embodiments, processor 401 may be local or remote. In some embodiments, processor 401 may be implemented on a cloud platform. In one embodiment, the cloud platform may include a private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, internal cloud, multi-cloud, etc., or any combination thereof.

[0091] In some embodiments, memory 402 may be an internal storage unit of the allocation device 400, such as a hard disk or memory of the allocation device 400. In other embodiments, memory 402 may also be an external storage device of the allocation device 400, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., provided on the allocation device 400.

[0092] Furthermore, the memory 402 may include both internal storage units of the dispatching device 400 and external storage devices. The memory 402 is used to store application software and various types of data installed on the dispatching device 400.

[0093] In some embodiments, display 403 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an organic light-emitting diode (OLED) touchscreen. Display 403 is used to display information from the dispensing device 400 and to display a visual user interface. Components 401-403 of the dispensing device 400 communicate with each other via a system bus.

[0094] In one embodiment, when processor 401 executes the data-driven water resource allocation consensus decision support program in memory 402, the following steps can be implemented: Acquire water resource monitoring data and decision-maker information for the target water area. The water resource monitoring data includes water quality data, hydrological data, and environmental data, while the decision-maker information includes the decision-maker's identity information and historical decision-making records. A decision-maker relationship network is constructed based on decision-maker information in the target water area, and the network is divided into groups based on the Louvain community detection algorithm. Assign a deep Q-network to each group in the decision-maker relationship network, and determine the consensus decision on water resource allocation for each group in the decision-maker relationship network based on the water resource monitoring data of the target water area and the deep Q-network.

[0095] It should be understood that when the processor 401 executes the data-driven water resource allocation consensus decision support program in the memory 402, in addition to the functions mentioned above, it can also perform other functions, as can be found in the description of the corresponding method embodiments above.

[0096] Furthermore, this embodiment of the invention does not specifically limit the type of the dispatching device 400 mentioned. The dispatching device 400 can be a portable electronic device such as a mobile phone, tablet computer, personal digital assistant (PDA), wearable device, or laptop computer. Exemplary embodiments of portable electronic devices include, but are not limited to, portable electronic devices running iOS, Android, Microsoft, or other operating systems. The aforementioned portable electronic devices can also be other portable electronic devices, such as laptop computers with touch-sensitive surfaces (e.g., touch panels). It should also be understood that in some other embodiments of the invention, the dispatching device 400 may not be a portable electronic device, but rather a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0097] Accordingly, this application also provides a computer-readable storage medium for storing computer-readable programs or instructions. When the programs or instructions are executed by a processor, they can implement the steps or functions of the data-driven water resource allocation consensus decision support method provided in the above-described method embodiments.

[0098] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.), and the computer program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0099] The data-driven water resource allocation consensus decision support method and apparatus provided by the present invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A data-driven consensus decision support method for water resource allocation, characterized in that, include: Acquire water resource monitoring data and decision-maker information for the target water area. The water resource monitoring data includes water quality data, hydrological data, and environmental data, while the decision-maker information includes the decision-maker's identity information and historical decision-making records. A decision-maker relationship network is constructed based on decision-maker information in the target water area, and the network is divided into groups based on the Louvain community detection algorithm. Assign a deep Q-network to each group in the decision-maker relationship network, and determine the consensus decision on water resource allocation for each group in the decision-maker relationship network based on the water resource monitoring data of the target water area and the deep Q-network.

2. The data-driven water resource allocation consensus decision support method according to claim 1, characterized in that, The construction of a decision-maker relationship network based on decision-maker information from the target water area includes: Based on decision-maker information for the target water area, determine the historical cooperation records among decision-makers; Construct a network of relationships among decision-makers based on their historical records of collaboration.

3. The data-driven water resource allocation consensus decision support method according to claim 1, characterized in that, The process of grouping decision-maker relationship networks based on the Louvain community detection algorithm includes: Based on the Louvain community detection algorithm, the modularity index of each group is calculated to divide the decision-maker relationship network into groups.

4. The data-driven water resource allocation consensus decision support method according to claim 1, characterized in that, The determination of each decision-maker's influence weight based on the PageRank algorithm includes: The PageRank algorithm is used to determine the out-degree and in-degree of each decision-maker in the decision-maker relationship network. The influence weight of each decision-maker is determined based on the out-degree and in-degree of each decision-maker in the decision-maker relationship network.

5. The data-driven consensus decision support method for water resource allocation according to claim 1, characterized in that, The consensus decision-making process for water resource allocation, based on water resource monitoring data of the target water area and a deep Q-network, for each group segmented within the decision-maker relationship network, includes: Based on water resource monitoring data of the target water area and the deep Q network, the allocation actions of each group in the decision-maker relationship network are selected, and the allocation actions of each group are optimized based on the reward function until the decision-maker relationship network reaches the consensus threshold. Based on the allocation actions of each group when the decision-maker relationship network reaches the consensus threshold, the consensus decision on water resource allocation for each group is determined.

6. The data-driven water resource allocation consensus decision support method according to claim 5, characterized in that, The determination of water resource allocation consensus decisions for each group based on the allocation actions of each group when a consensus threshold is reached through the decision-maker relationship network includes: When the decision-makers' relationship network reaches a consensus threshold, the allocation actions of each group and the influence weight of each decision-maker are determined as the consensus decision on water resource allocation for each group. The influence weight of each decision-maker is determined based on the PageRank algorithm.

7. The data-driven water resource allocation consensus decision support method according to claim 5, characterized in that, The reward function is constructed based on harmony and the total number of iterations.

8. A data-driven consensus decision support device for water resource allocation, characterized in that, include: The acquisition module is used to acquire water resource monitoring data and decision-maker information for the target water area. The water resource monitoring data includes water quality data, hydrological data, and environmental data, while the decision-maker information includes the decision-maker's identity information and historical decision-making records. The module is used to construct a decision-maker relationship network based on decision-maker information in the target water area, and to perform group segmentation of the decision-maker relationship network based on the Louvain community detection algorithm; The determination module is used to assign a deep Q-network to each group in the decision-maker relationship network, and to determine the consensus decision on water resource allocation for each group in the decision-maker relationship network based on the water resource monitoring data of the target water area and the deep Q-network.

9. A dispensing device, characterized in that, Including memory and processor, among which, The memory is used to store programs; The processor, coupled to the memory, is used to execute the program stored in the memory to implement the steps in the data-driven water resource allocation consensus decision support method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store computer-readable programs or instructions, which, when executed by a processor, can implement the steps in the data-driven water resource allocation consensus decision support method according to any one of claims 1 to 7.

Images