Water resource supply and demand balance intelligent management and control method based on multi-source data fusion

Abstract

The invention discloses a water resource supply and demand balance intelligent management and control method based on multi-source data fusion, and relates to the technical field of water resource management. The time-space range of data acquisition is calibrated in advance, and meteorological, hydrological, water consumption, social economy, infrastructure and ecological multi-dimensional data are acquired in a classified manner; after preprocessing, multi-source data fusion is realized by adopting a mode of combining Kalman filtering and a deep learning automatic encoder, and a standardized data set is generated; and constructing a hybrid model comprising a system dynamics module, a machine learning module and a digital twinning module, wherein the hybrid model is used for realizing simulation of water resource supply and demand balance. By means of a hybrid model constructed by system dynamics, machine learning and digital twinning, accurate simulation of a dynamic evolution rule of a water resource supply and demand system is reserved, and high-precision prediction of future demands, rapid identification of abnormal events and virtual simulation of extreme scenes are realized; and the adaptability and toughness of the system to complex working conditions and emergency situations are obviously enhanced.

Description

Technical Field

[0001] This invention relates to the field of water resource management technology, and more specifically, to an intelligent management and control method for water resource supply and demand balance based on multi-source data fusion. Background Technology

[0002] Water resources, as a fundamental strategic resource supporting socio-economic development and maintaining ecosystem balance, directly impact the overall sustainable development of a region through the dynamic balance of its supply and demand. With the intensification of global climate change and the frequent occurrence of extreme hydrological events such as prolonged droughts and concentrated torrential rains, the uneven spatial and temporal distribution of water resources has become increasingly prominent. Increased fluctuations in surface runoff and unstable groundwater recharge pose severe challenges to the stability of water supply. Simultaneously, rapid industrialization, urbanization, and large-scale agricultural development have led to a continuous increase in demand for industrial, domestic, and agricultural water, resulting in ongoing adjustments to water use structures. Traditional water resource management models are struggling to adapt to the dynamic needs of diverse water users, leading to a persistent and acute supply-demand imbalance in some regions.

[0003] There are still many pain points in the field of water resource management that urgently need to be addressed: traditional management methods rely heavily on experience-based decision-making and static planning, lacking a systematic consideration of multi-dimensional factors such as meteorology, hydrology, and socio-economic factors, making it difficult to accurately capture the dynamic evolution of the water resource supply and demand system; data collection is scattered and standards are inconsistent, with heterogeneity, noise interference, and information redundancy among data from different sources, failing to fully explore the value of data and resulting in insufficient accuracy in supply and demand forecasting; existing control models mostly focus on single-objective optimization or can only achieve simulation analysis of local links, making it difficult to take into account multiple objectives such as water supply reliability, cost control, and ecological protection, and lacking adaptability and resilience in the face of extreme scenarios or emergencies, thus failing to provide scientific and efficient decision support for the refined and intelligent management of water resources.

[0004] There are currently no effective solutions to the problems in the relevant technologies. Summary of the Invention

[0005] To address the problems in related technologies, this invention proposes an intelligent management and control method for water resource supply and demand balance based on multi-source data fusion, in order to overcome the aforementioned technical problems existing in existing related technologies.

[0006] The technical solution of this invention is implemented as follows:

[0007] A method for intelligent management and control of water resource supply and demand balance based on multi-source data fusion includes the following steps:

[0008] The spatiotemporal scope of data collection is predefined, and multi-dimensional data on meteorology, hydrology, water use, socio-economics, infrastructure and ecology are collected in categories. After preprocessing, Kalman filtering and deep learning autoencoder are combined to achieve multi-source data fusion and generate a standardized dataset.

[0009] A hybrid model comprising system dynamics, machine learning, and digital twin modules is constructed to simulate, predict, and simulate extreme scenarios of water resource supply and demand balance.

[0010] A multi-objective optimization configuration model is constructed, and the output results of the hybrid model are input. With the objectives of maximizing water supply reliability, minimizing total configuration cost, and maximizing ecological benefits, the Pareto optimal solution set is obtained and the comprehensive optimal configuration scheme is selected to complete the intelligent management and control of water resource supply and demand balance.

[0011] Furthermore, the preprocessing includes using linear interpolation to supplement missing data, identifying and correcting abnormal data using the 3σ criterion, and normalizing data based on the Min-Max standardization formula; Kalman filtering is used to eliminate noise interference in the real-time time-series monitoring data, and a deep learning autoencoder is used to achieve feature extraction and fusion of multi-dimensional heterogeneous data to generate a standardized dataset.

[0012] Furthermore, the system dynamics module is used to establish the total water supply equation, the total demand equation, and the feedback regulation equation, and output the dynamic trend of supply and demand changes.

[0013] The equation for total water supply is expressed as follows:

[0014] ;

[0015] In the formula, Let t be the total water supply at time t. Let be the surface water supply at time t. Let be the groundwater supply at time t. Let be the amount of reclaimed water supplied at time t. Let t be the amount of water supplied from outside the region.

[0016] The surface water supply equation is expressed as:

[0017] ;

[0018] Where A is the area of ​​the target region. Let be the surface runoff at time t. Let t be the water supply from the reservoir at time t;

[0019] The groundwater supply equation is expressed as:

[0020] ;

[0021] in, This is the groundwater extraction coefficient. This is the minimum permissible groundwater level. This represents the maximum permissible extraction volume of groundwater.

[0022] The equation for the supply of reclaimed water is expressed as:

[0023] ;

[0024] In the formula, Let t be the total wastewater volume generated in the target area at time t; Let be the treatment efficiency of the wastewater treatment plant at time t; Let t be the reclaimed water reuse rate;

[0025] The equation for the external water supply is expressed as:

[0026]

[0027] In the formula, This represents the maximum water transfer capacity of the inter-regional water transfer project. Let t represent the water supply and demand gap in the target area at time t. This refers to the external water supply coefficient. Let t be the planned water transfer volume at time t;

[0028] The equation for total water resource demand is expressed as follows:

[0029] ;

[0030] In the formula, Let be the total water resource demand at time t;

[0031] Industrial water consumption equation , represented as:

[0032] ;

[0033] In the formula, Let be the total industrial output at time t. Let t be the water consumption per 10,000 yuan of industrial output. , To improve the efficiency of industrial water use;

[0034] Agricultural water consumption equation , represented as:

[0035] ;

[0036] In the formula, Let be the agricultural planting area at time t. Let t be the gross agricultural irrigation quota. Let be the irrigation water utilization coefficient at time t;

[0037] Domestic water consumption equation , represented as:

[0038] ;

[0039] In the formula, Let be the average domestic water consumption at time t;

[0040] Ecological water demand equation , represented as:

[0041] ;

[0042] In the formula, Let t be the river's ecological water demand. Let t be the wetland's ecological water demand. Let t be the vegetation ecological water demand; Let t be the soil ecological water demand.

[0043] Among them, the feedback regulation submodule can adjust water use efficiency and industrial structure to carry out feedback regulation when water supply and demand are imbalanced.

[0044] The water efficiency adjustment coefficient is expressed as:

[0045]

[0046] In the formula, This is the water efficiency adjustment coefficient;

[0047] The industrial structure adjustment coefficient is expressed as:

[0048]

[0049] In the formula, This is an industrial structure adjustment coefficient used to reduce the proportion of water-intensive industries.

[0050] Furthermore, the machine learning module includes an LSTM water resource demand prediction model and a GAN anomaly detection model; the LSTM water resource demand prediction model predicts the total water resource demand and the water consumption of each component at different time scales in the future by inputting the feature vector of the fused multi-source data; the GAN anomaly detection model realizes the identification of abnormal events and the generation of extreme scenarios through adversarial training between the generator and the discriminator.

[0051] Wherein, the generator G is denoted as: ;

[0052] In the formula, It is a random noise vector. The generator weight matrix, This is the generator bias vector;

[0053] Discriminator D is represented as: ;

[0054] In the formula, x represents the input data. This is the discriminator weight matrix. For the discriminator bias vector, This is the output of the discriminator.

[0055] The objective function of the GAN anomaly detection model is expressed as:

[0056]

[0057] in, The probability distribution of the real data. For the probability distribution of random noise, It expresses expectation.

[0058] Furthermore, the digital twin module constructs a three-dimensional geometric model based on GIS geographic information data and water conservancy engineering design drawings, embeds a mathematical model of system dynamics and machine learning module, and realizes real-time data mapping and state synchronization between the physical system and the virtual system through Internet of Things communication technology, which is used to complete the simulation of supply and demand changes under different control schemes and extreme scenarios.

[0059] Furthermore, the constraints of the multi-objective optimization configuration model include water balance constraints, water supply capacity constraints, water demand constraints, water quality constraints, and ecological constraints.

[0060] Water balance constraints are expressed as:

[0061] ;

[0062] Water supply capacity constraints are expressed as:

[0063] ;

[0064] ;

[0065] In the formula, , , , These are the maximum supply capacities for surface water, groundwater, reclaimed water, and water transferred from other regions, respectively.

[0066] Water demand constraints are expressed as:

[0067] ;

[0068] ;

[0069] In the formula, , , , These are the minimum necessary water consumption for industry, agriculture, domestic use, and ecology, respectively.

[0070] Water quality constraints, meaning that the quality of the supplied water must meet relevant national standards, are expressed as:

[0071]

[0072]

[0073] Ecological constraints, namely that groundwater extraction must not exceed groundwater recharge, are expressed as:

[0074]

[0075] In the formula, Let be the groundwater recharge at time t; the ecological water recharge shall not be less than the minimum ecological water requirement, i.e. .

[0076] Furthermore, the process of obtaining the Pareto optimal solution set and selecting the comprehensive optimal allocation scheme includes: using a non-dominated sorting genetic algorithm with an elitist strategy to solve the multi-objective optimization allocation model, obtaining the Pareto optimal solution set, and calculating the closeness of each scheme through the approximation of the ideal solution sorting method, and selecting the scheme with the largest closeness as the comprehensive optimal water resource allocation scheme.

[0077] The beneficial effects of this invention are:

[0078] This invention significantly improves data quality and information utilization by collecting heterogeneous data from multiple dimensions, including meteorology, hydrology, socio-economics, and ecology, and combining Kalman filtering with a deep learning autoencoder fusion strategy, providing reliable data support for subsequent prediction and optimization. Through a hybrid model built using system dynamics, machine learning, and digital twins, it retains accurate simulation of the dynamic evolution of the water resource supply and demand system while achieving high-precision prediction of future demand, rapid identification of abnormal events, and virtual simulation of extreme scenarios, significantly enhancing the system's adaptability and resilience to complex operating conditions and emergencies. Through a multi-objective optimization model, it achieves a comprehensive balance between water supply reliability, cost control, and ecological benefits, ensuring that water resource allocation schemes meet supply and demand balance requirements, improving the scientific and comprehensive nature of decision-making, and realizing intelligent and refined management of water resource supply and demand balance. This invention possesses significant technical advantages and application value. Attached Figure Description

[0079] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0080] Figure 1 This is a flowchart illustrating an intelligent management and control method for water resource supply and demand balance based on multi-source data fusion according to an embodiment of the present invention. Detailed Implementation

[0081] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the scope of protection of the present invention.

[0082] According to an embodiment of the present invention, a method for intelligent management and control of water resource supply and demand balance based on multi-source data fusion is provided.

[0083] like Figure 1 As shown, the intelligent management and control method for water resource supply and demand balance based on multi-source data fusion according to an embodiment of the present invention includes the following steps:

[0084] Step S1: Pre-determine the spatiotemporal range of data collection, classify and collect meteorological data, hydrological data, water use data, socioeconomic data, infrastructure data and ecological data, and after cleaning and standardization preprocessing, use Kalman filtering and deep learning autoencoder to achieve multi-source data fusion and generate a standardized dataset D.

[0085] This technical solution pre-defines the geographical boundaries, administrative scope, and water resource system coverage of the target control area, and determines the time span and spatial resolution of data collection.

[0086] Specifically, during implementation, meteorological data is acquired through regional meteorological stations and satellite remote sensing imagery, including precipitation (P), evaporation (E), temperature (T), relative humidity (RH), and wind speed (V). w Sunshine duration S; Hydrological data are monitored in real time through hydrological monitoring stations, water level gauges, and flow meters, including river flow Q. r Groundwater level H g Surface water storage W s Groundwater resources W g Transit water resources W t Water usage data is collected through smart water meters and enterprise water metering systems, including industrial water consumption (W).ind Agricultural water consumption W agr Domestic water consumption (W) res Ecological water replenishment volume W eco Wastewater utilization W recl Water efficiency Socioeconomic data were obtained from statistical yearbooks and publicly available government databases, including population size N, population growth rate rN, total GDP G, and industrial structure share. urbanization rate Infrastructure data is collected through water conservancy project archives and IoT sensor monitoring, including the length of the water supply network (L). p Pipeline leakage rate β, number of pumping stations N p Pump station efficiency Reservoir capacity V r Water supply capacity Q of water conservancy projects eng Ecological data were obtained through environmental monitoring stations and satellite remote sensing inversion, including water quality indicators such as pH, chemical oxygen demand (COD), and ammonia nitrogen (NH3−N).

[0087] Among them, linear interpolation was used to supplement missing data, the 3σ criterion was used to identify and correct outlier data, and data normalization preprocessing was completed based on the Min-Max standardization formula.

[0088] Meanwhile, a fusion strategy combining Kalman filtering and deep learning feature extraction is adopted, with the following specific steps:

[0089] For real-time monitoring data with time-series characteristics, noise interference is eliminated through Kalman filtering. The state equation and observation equation are as follows:

[0090] The state equation is expressed as:

[0091]

[0092] The observation equation is expressed as:

[0093]

[0094] In the formula, Let A be the system state vector at time k, and let A be the state transition matrix. Let B be the control input vector at time k, and let B be the control input matrix. This is the process noise vector. Let H be the observation vector at time k, and H be the observation matrix. This is the observed noise vector.

[0095] The Kalman filter update process includes two stages: prediction and update.

[0096] Prediction phase:

[0097]

[0098]

[0099] In the formula, The predicted state at time k is based on the observations at time k−1. This is the state estimate at time k−1. For the prediction error covariance matrix, Let be the estimation error covariance matrix at time k−1.

[0100] Update phase:

[0101]

[0102]

[0103]

[0104] In the formula, Let I be the Kalman gain matrix, and I be the identity matrix. The state estimate at time k. Let be the estimation error covariance matrix at time k.

[0105] For multi-dimensional, heterogeneous, multi-source data, a deep learning autoencoder is used for feature extraction and fusion. The autoencoder consists of an encoder and a decoder. The encoder maps high-dimensional input data to a low-dimensional feature space, and the decoder maps the low-dimensional features back to the original data space, achieving feature learning by minimizing the reconstruction error.

[0106] The mathematical expression for the encoder is:

[0107]

[0108] The mathematical expression for the decoder is:

[0109]

[0110] Where x is the input multi-source data vector, and h is the extracted low-dimensional feature vector. To reconstruct the data vector, , These are the weight matrices for the encoder and decoder, respectively. , Here, σ represents the bias vectors, and σ is the activation function.

[0111] Specifically, during training, the loss function that minimizes the reconstruction error is calibrated as follows:

[0112]

[0113] In the formula, N is the sample size. Let i be the original data vector of the i-th sample. Let be the reconstructed data vector of the i-th sample.

[0114] Through the above fusion process, a standardized dataset D with a unified format, low noise, and high correlation is generated, denoted as: , where M is the number of data samples.

[0115] Step S2: Construct a hybrid model based on system dynamics (SD) and machine learning (ML). The hybrid model includes a system dynamics module, a machine learning module, and a digital twin module, which are used to simulate and predict the balance of water supply and demand.

[0116] Among them, the system dynamics module is used to input preprocessed historical data, establish the total water supply equation, total demand equation and feedback regulation equation, and output the dynamic trend of supply and demand.

[0117] Specifically, the equation for total water supply is expressed as:

[0118] ;

[0119] In the formula, Let t be the total water supply at time t. Let be the surface water supply at time t. Let be the groundwater supply at time t. Let be the amount of reclaimed water supplied at time t. Let t be the amount of water supplied from outside the region at time t.

[0120] The surface water supply equation is expressed as:

[0121] ;

[0122] Where A is the area of ​​the target region. Let be the surface runoff at time t. Let t be the water supply from the reservoir at time t;

[0123] The groundwater supply equation is expressed as:

[0124] ;

[0125] in, This is the groundwater extraction coefficient. This is the minimum permissible groundwater level. This represents the maximum permissible extraction volume of groundwater.

[0126] The equation for the supply of reclaimed water is expressed as:

[0127] ;

[0128] In the formula, The total wastewater volume generated in the target area at time t includes industrial wastewater, domestic sewage, and agricultural drainage, which is obtained by summing the drainage volumes of each water-using sector. Let t be the treatment efficiency of the wastewater treatment plant at time t, which is the proportion of the total treated water volume that meets the reclaimed water quality standard. Let t be the reuse rate of reclaimed water, which is the proportion of reclaimed water that has been treated to meet the standards and is actually used for water resource supply.

[0129] The equation for the external water supply is expressed as:

[0130]

[0131] In the formula, The maximum water transfer capacity of an inter-regional water transfer project is determined by the design standards and operating status of the project. The calculation time interval at time t is set according to the control cycle; Let t represent the water supply and demand gap in the target area at time t. This is the external water transfer replenishment coefficient, which reflects the proportion of external water transfer that replenishes the supply-demand gap; The planned water volume for external water transfer at time t is pre-determined based on annual / quarterly water transfer agreements and regional water resource allocation plans.

[0132] The equation for total water resource demand is expressed as follows:

[0133] ;

[0134] In the formula, Let be the total water resource demand at time t;

[0135] Industrial water consumption equation , represented as:

[0136] ;

[0137] In the formula, Let be the total industrial output at time t. Let t be the water consumption per 10,000 yuan of industrial output. , To improve the efficiency of industrial water use;

[0138] Agricultural water consumption equation , represented as:

[0139] ;

[0140] In the formula, Let be the agricultural planting area at time t. Let t be the gross agricultural irrigation quota. Let be the irrigation water utilization coefficient at time t;

[0141] Domestic water consumption equation , represented as:

[0142] ;

[0143] In the formula, Let be the average domestic water consumption at time t;

[0144] Ecological water demand equation , represented as:

[0145] ;

[0146] In the formula, Let be the ecological water demand of the river at time t, that is, the minimum amount of water required to maintain the basic river runoff and ensure the survival of aquatic organisms, expressed as: ,in, The minimum base flow of a river is determined by river ecological protection standards. The calculation time interval is t. Let t be the wetland ecological water demand, which is the amount of water needed to maintain the wetland hydrological environment and protect wetland vegetation and biodiversity. The calculation formula is: ,in This represents the total area of ​​wetlands. The suitable water depth for the wetland at time t is determined based on the wetland type. Let t be the vegetation ecological water demand, which is the amount of water consumed by natural vegetation evapotranspiration in the region. The calculation formula is: ,in This represents the total area covered by vegetation. The reference crop evapotranspiration at time t is calculated using meteorological data. This is the crop coefficient, determined based on vegetation type; Let be the soil ecological water requirement at time t, which is the amount of water needed to maintain suitable soil moisture and ensure the growth of plant roots. The calculation formula is: [Formula omitted for brevity].

[0147]

[0148] In the formula, For the area of ​​soil requiring water replenishment, To increase the depth of soil watering, For soil bulk density, To achieve the appropriate soil moisture content, Let t be the actual soil moisture content at time t.

[0149] Among them, the feedback regulation submodule can adjust water use efficiency and industrial structure to carry out feedback regulation when water supply and demand are imbalanced.

[0150] The water efficiency adjustment coefficient is expressed as:

[0151]

[0152] In the formula, This is the water efficiency adjustment coefficient;

[0153] The industrial structure adjustment coefficient is expressed as:

[0154]

[0155] In the formula, This is an industrial structure adjustment coefficient used to reduce the proportion of water-intensive industries.

[0156] Among them, the machine learning module is used to input the standardized dataset D and the output results of the system dynamics module. It achieves accurate supply and demand prediction through the LSTM water resource demand prediction model, completes anomaly detection and extreme scenario generation through the GAN anomaly detection model, and outputs prediction results, anomaly alarms and virtual scene data.

[0157] Specifically, a water resource demand forecasting model is constructed using a Long Short-Term Memory (LSTM) network to predict the total water resource demand and the water consumption of each component at different time scales in the future.

[0158] The LSTM water resource demand prediction model's network structure includes an input layer, hidden layers, a fully connected layer, and an output layer. The input vector of the input layer is the fused multi-source data feature vector d. i The data includes historical water consumption, meteorological data, and socioeconomic data; the hidden layers consist of 2-4 LSTM layers with 64-256 neurons per layer, and Dropout technology is used to prevent overfitting with a Dropout rate of 0.2-0.5; the fully connected layers use the ReLU activation function; and the output layer uses a linear activation function to output the predicted water consumption.

[0159] Specifically, the LSTM water resource demand forecasting model is as follows:

[0160] The Forgot Gate is represented as: ;

[0161] Input gate, represented as: ;

[0162] Cell state candidate values ​​are represented as follows: ;

[0163] Cell state update is represented as: ;

[0164] Output gate, represented as: ;

[0165] Hidden state output, represented as; ;

[0166] In the formula, Output for the forget gate. For input gate output, Candidate values ​​for cell state. This represents the updated cell state. For output gate output, Output in hidden state Let be the input vector at time t. Let be the hidden state at time t−1. The cell state at time t−1. , , , These are the weight matrices for each gate. , , , These are the bias vectors for each gate. It is the sigmoid activation function. It is the hyperbolic tangent activation function.

[0167] Additionally, it includes: using Generative Adversarial Networks (GANs) to build GAN anomaly detection models to identify anomalous events in water resource systems, such as pipe ruptures, water use violations, and extreme droughts, while generating extreme scenarios to test system resilience.

[0168] The GAN anomaly detection model consists of a generator G and a discriminator D. The generator is used to generate fake data that is consistent with the distribution of real data, and the discriminator is used to distinguish between real data and generated fake data. The two are continuously optimized through adversarial training.

[0169] The generator G is represented as: ;

[0170] In the formula, It is a random noise vector. The generator weight matrix, This is the generator bias vector.

[0171] Discriminator D is represented as: ;

[0172] In the formula, x represents the input data. This is the discriminator weight matrix. For the discriminator bias vector, This is the output of the discriminator.

[0173] The objective function of the GAN anomaly detection model is expressed as:

[0174]

[0175] in, The probability distribution of the real data. For the probability distribution of random noise, It expresses expectation.

[0176] In anomaly detection, the reconstruction error between real data and normal data generated by the generator is calculated. When the reconstruction error exceeds a set threshold, anomaly detection is performed. When this occurs, it is determined to be an abnormal event, and the reconstruction error is calculated and represented as follows:

[0177] ;

[0178] In the formula, For the actual input data, For data reconstructed using GAN, It is an L2 norm.

[0179] In the implementation of this technical solution, extreme drought scenarios, peak demand scenarios, and infrastructure failure scenarios are generated by adjusting the input noise vector z of the generator G during the extreme scenario generation process.

[0180] Among them, the digital twin module is used to input the output data of the system dynamics module and the machine learning module, as well as the real-time state data of the physical system. Through geometric modeling, physical modeling, data mapping and synchronization, and scene simulation, it outputs a virtual system and scene simulation results that are mapped 1:1 to the physical system.

[0181] Specifically, a digital twin model of the water resource system in the target area is constructed to achieve real-time mapping and bidirectional interaction between the physical and virtual systems. The steps for constructing the digital twin model are as follows:

[0182] Based on GIS geographic information data and water conservancy engineering design drawings, a three-dimensional geometric model of the topography, water conservancy projects, and water use units of the target area is constructed, and BIM technology is used to realize the refined modeling of water conservancy projects.

[0183] By embedding the mathematical models of the system dynamics module and the machine learning module into the digital twin model, the physical behavior simulation of the water resource supply and demand process can be realized, including the dynamic simulation of water flow, water allocation, and water use processes.

[0184] By using IoT communication technology, a real-time data mapping channel is established between the physical system and the digital twin model, synchronizing the real-time monitoring data of the physical system to the digital twin model, updating the state of the virtual system, and achieving state consistency between the physical system and the virtual system.

[0185] Based on the digital twin model, it is possible to simulate changes in water supply and demand under different control schemes and extreme scenarios, analyze the system's response characteristics and resilience level, and provide a virtual test platform for subsequent optimization and management.

[0186] Step S3: Construct a multi-objective optimization configuration model, input the prediction results and scenario data output by the hybrid model, and take maximizing water supply reliability, minimizing total configuration cost, and maximizing ecological benefits as the objective strategies. Solve to obtain the Pareto optimal solution set and select the comprehensive optimal configuration scheme.

[0187] Maximizing water supply reliability, i.e., the probability that the actual water supply meets the demand, is expressed as:

[0188]

[0189] In the formula, The total duration of the planning period, For indicator functions, when hour, ,otherwise .

[0190] Minimizing the total configuration cost, which includes the construction and operation costs of water conservancy projects, the cost of water resource extraction, and the cost of wastewater treatment, is expressed as:

[0191]

[0192] In the formula, The construction and operation costs of water conservancy projects are expressed as... , For the number of water conservancy projects, Let be the water supply of the i-th water conservancy project. , Let be the cost coefficient of the i-th water conservancy project; Cost of water resource extraction; Wastewater treatment costs;

[0193] Maximizing ecological benefits, which is measured by improving ecological carrying capacity and reducing ecological footprint, is expressed as:

[0194] ;

[0195] Define the constraints as follows:

[0196] Water balance constraints are expressed as:

[0197] ;

[0198] Water supply capacity constraints are expressed as:

[0199] ;

[0200] ;

[0201] In the formula, , , , These represent the maximum supply capacity of surface water, groundwater, reclaimed water, and water transferred from other regions, respectively.

[0202] Water demand constraints are expressed as:

[0203] ;

[0204] ;

[0205] In the formula, , , , These are the minimum necessary water consumption for industry, agriculture, domestic use, and ecology, respectively.

[0206] Water quality constraints, meaning that the quality of the supplied water must meet relevant national standards, are expressed as:

[0207]

[0208]

[0209] Ecological constraints, namely that groundwater extraction must not exceed groundwater recharge, are expressed as:

[0210]

[0211] In the formula, Let be the groundwater recharge at time t; the ecological water recharge shall not be less than the minimum ecological water requirement, i.e. .

[0212] Specifically, the non-dominated sorting genetic algorithm with elitist strategy (NSGA-II) is used to solve the multi-objective optimization allocation model, obtaining the Pareto optimal solution set. This solution set contains multiple non-dominated optimization allocation schemes, each corresponding to a different combination of objective weights. The specific steps include:

[0213] Initialize the population and generate N initial solutions, representing water resource allocation schemes. Each solution represents the water supply allocation ratio for different water-using sectors.

[0214] Non-dominated ranking: Perform non-dominated ranking on each individual in the population, and calculate the dominance level and crowding distance of each individual;

[0215] The selection process employs a combination of roulette wheel selection and crowding selection to choose superior individuals for the next generation of the population.

[0216] Crossover operations can be performed using single-point or two-point crossover, with crossover probabilities set.

[0217] The mutation operation uses random mutation and sets the mutation probability;

[0218] Iterative updates are performed, and the above steps are repeated until the maximum number of iterations is reached, ultimately yielding the Pareto optimal solution set.

[0219] Specifically, from the Pareto optimal solution set, the Top-Optimal Solution Ranking Method (TOPSIS) is used to select the overall optimal water resource allocation scheme, which includes the following steps:

[0220] Construct a standardized decision matrix, represented as follows:

[0221] ;

[0222] In the formula, Let m be the value of the j-th objective in the i-th scheme, m be the number of schemes, and n be the number of objectives.

[0223] Determine the ideal solution Z + and negative ideal solution Z − , respectively represented as:

[0224] This indicates that for maximizing the objective or This indicates that for the minimization objective;

[0225] This indicates that for maximizing the objective or This indicates the goal of minimization.

[0226] The distances from each scheme to the positive and negative ideal solutions are calculated and expressed as follows:

[0227] ;

[0228] ;

[0229] The similarity between the various schemes is calculated and expressed as follows:

[0230]

[0231] The solution with the highest degree of similarity is selected as the overall optimal solution.

[0232] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Those skilled in the art, upon considering the disclosure in the specification and embodiments, will readily conceive of other embodiments of this disclosure. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

[0233] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A method for intelligent management and control of water resource supply and demand balance based on multi-source data fusion, characterized in that, Includes the following steps: The spatiotemporal scope of data collection is predefined, and multi-dimensional data on meteorology, hydrology, water use, socio-economics, infrastructure and ecology are collected in categories. After preprocessing, Kalman filtering and deep learning autoencoder are combined to achieve multi-source data fusion and generate a standardized dataset. A hybrid model comprising system dynamics, machine learning, and digital twin modules is constructed to simulate, predict, and simulate extreme scenarios of water resource supply and demand balance. A multi-objective optimization configuration model is constructed, and the output results of the hybrid model are input. With the objectives of maximizing water supply reliability, minimizing total configuration cost, and maximizing ecological benefits, the Pareto optimal solution set is obtained and the comprehensive optimal configuration scheme is selected to complete the intelligent management and control of water resource supply and demand balance.

2. The intelligent management and control method for water resource supply and demand balance based on multi-source data fusion according to claim 1, characterized in that, The preprocessing includes using linear interpolation to supplement missing data, identifying and correcting abnormal data using the 3σ criterion, and normalizing data based on the Min-Max standardization formula. Kalman filtering is used to eliminate noise interference in real-time time-series monitoring data, and deep learning autoencoders are used to extract and fuse features from multi-dimensional heterogeneous data to generate a standardized dataset.

3. The intelligent management and control method for water resource supply and demand balance based on multi-source data fusion according to claim 1, characterized in that, The system dynamics module is used to establish the total water supply equation, total demand equation, and feedback regulation equation, and output the dynamic trend of supply and demand. The equation for total water supply is expressed as follows: ； In the formula, Let t be the total water supply at time t. Let be the surface water supply at time t. Let be the groundwater supply at time t. Let be the amount of reclaimed water supplied at time t. Let t be the amount of water supplied from outside the region. The surface water supply equation is expressed as: ； Where A is the area of ​​the target region. Let be the surface runoff at time t. Let t be the water supply from the reservoir at time t; The groundwater supply equation is expressed as: ； in, This is the groundwater extraction coefficient. This is the minimum permissible groundwater level. This represents the maximum permissible extraction volume of groundwater. The equation for reclaimed water supply is expressed as: ； In the formula, Let t be the total wastewater volume generated in the target area at time t; Let be the treatment efficiency of the wastewater treatment plant at time t; Let t be the reclaimed water reuse rate; The equation for the external water supply is expressed as: In the formula, This represents the maximum water transfer capacity of the inter-regional water transfer project. Let t represent the water supply and demand gap in the target area at time t. This refers to the external water supply coefficient. Let t be the planned water transfer volume at time t; The equation for total water resource demand is expressed as follows: ； In the formula, Let be the total water resource demand at time t; Industrial water consumption equation , is represented as: ； In the formula, Let be the total industrial output at time t. Let t be the water consumption per 10,000 yuan of industrial output. , To improve the efficiency of industrial water use; Agricultural water consumption equation , is represented as: ； In the formula, Let be the agricultural planting area at time t. Let t be the gross agricultural irrigation quota. Let be the irrigation water utilization coefficient at time t; Domestic water consumption equation , is represented as: ； In the formula, Let be the average domestic water consumption at time t; Ecological water demand equation , is represented as: ； In the formula, Let t be the river's ecological water demand. Let t be the wetland's ecological water demand. Let t be the vegetation ecological water demand. Let t be the soil ecological water demand. Among them, the feedback regulation submodule can adjust water use efficiency and industrial structure to carry out feedback regulation when water supply and demand are imbalanced. The water efficiency adjustment coefficient is expressed as: In the formula, This is the water efficiency adjustment coefficient; The industrial structure adjustment coefficient is expressed as: In the formula, This is an industrial structure adjustment coefficient used to reduce the proportion of water-intensive industries.

4. The intelligent management and control method for water resource supply and demand balance based on multi-source data fusion according to claim 3, characterized in that, The machine learning module includes an LSTM water resource demand prediction model and a GAN anomaly detection model. The LSTM water resource demand prediction model predicts the total water resource demand and the water consumption of each component at different time scales by inputting the feature vector of the fused multi-source data. The GAN anomaly detection model realizes the identification of abnormal events and the generation of extreme scenarios through adversarial training between the generator and the discriminator. Wherein, the generator G is denoted as: ; In the formula, It is a random noise vector. The generator weight matrix, This is the generator bias vector; Discriminator D is represented as: ; In the formula, x represents the input data. This is the discriminator weight matrix. For the discriminator bias vector, This is the output of the discriminator. The objective function of the GAN anomaly detection model is expressed as: in, The probability distribution of the real data. For the probability distribution of random noise, It expresses expectation.

5. The intelligent management and control method for water resource supply and demand balance based on multi-source data fusion according to claim 4, characterized in that, The digital twin module constructs a three-dimensional geometric model based on GIS geographic information data and water conservancy engineering design drawings, embeds a mathematical model of system dynamics and machine learning module, and realizes real-time data mapping and state synchronization between the physical system and the virtual system through Internet of Things communication technology, which is used to complete the simulation of supply and demand changes under different control schemes and extreme scenarios.

6. The intelligent management and control method for water resource supply and demand balance based on multi-source data fusion according to claim 1, characterized in that, The constraints of the multi-objective optimization configuration model include water balance constraints, water supply capacity constraints, water demand constraints, water quality constraints, and ecological constraints. Water balance constraints are expressed as: ； Water supply capacity constraints are expressed as: ； ； In the formula, , , , These are the maximum supply capacities for surface water, groundwater, reclaimed water, and water transferred from other regions, respectively. Water demand constraints are expressed as: ； ； In the formula, , , , These are the minimum necessary water consumption for industry, agriculture, domestic use, and ecology, respectively. Water quality constraints, meaning that the quality of the supplied water must meet relevant national standards, are expressed as: Ecological constraints, namely that groundwater extraction must not exceed groundwater recharge, are expressed as: In the formula, Let be the groundwater recharge at time t; the ecological recharge shall not be less than the minimum ecological water requirement, i.e. .

7. The intelligent management and control method for water resource supply and demand balance based on multi-source data fusion according to claim 1, characterized in that, The process of obtaining the Pareto optimal solution set and selecting the comprehensive optimal allocation scheme includes: using a non-dominated sorting genetic algorithm with an elitist strategy to solve the multi-objective optimization allocation model, obtaining the Pareto optimal solution set, calculating the closeness of each scheme through the approximation of the ideal solution sorting method, and selecting the scheme with the largest closeness as the comprehensive optimal water resource allocation scheme.

Images