Park water-energy collaborative management method and system based on digital twinning

Abstract

The invention provides a park water-energy collaborative management method and system based on digital twinning, and relates to the technical field of energy management, and the method comprises the steps: constructing a digital twinning body fusing plant water demand, pipe network hydraulic power and rainwater storage models, and synchronously processing the digital twinning body into multi-source real-time data in a space-time unified format; and a high-fidelity park water circulation virtual simulation environment is established. Multiple candidate schemes covering irrigation and energy distribution strategies are generated by using an intelligent algorithm, and virtual deduction and multi-target comprehensive scoring are performed on each scheme through a digital twinborn body, so that an optimal execution scheme is determined. According to the method, a data barrier between water and energy systems is effectively broken, and water-energy collaborative decision is realized. According to the method, the comprehensive effect of the scheme can be evaluated before the scheme is actually executed by utilizing the simulation prediction capability of the digital twinborn, so that the scientificity and the foresight of decision making are remarkably improved, and the overall efficiency and the intelligent level of park resource management are improved.

Description

Technical Field

[0001] This invention relates to the field of energy management technology, and in particular to a method and system for coordinated water and energy management in industrial parks based on digital twins. Background Technology

[0002] In existing technologies, water resource management (such as irrigation and rainwater harvesting) and energy management systems in industrial parks typically operate independently, resulting in functional fragmentation and "information silos." Irrigation systems cannot dynamically adjust based on the real-time water storage of rainwater harvesting systems, leading to water waste. Furthermore, the operation of energy-consuming equipment such as pumps is not optimized in conjunction with energy factors such as grid load and carbon emission costs. Simultaneously, existing management methods largely rely on pre-set static rules or single-dimensional data for decision-making, lacking a unified platform capable of integrating multi-source, heterogeneous data (such as meteorological, soil, and pipeline status) and performing pre-simulation, evaluation, and optimization of different scheduling strategies. This results in rigid overall management strategies that cannot dynamically adapt to environmental changes, making it difficult to achieve comprehensive optimization among multiple objectives such as water resource utilization efficiency, energy consumption costs, and environmental impact. Summary of the Invention

[0003] The purpose of this invention is to provide a digital twin-based method and system for collaborative water and energy management in industrial parks, so as to achieve multi-objective optimization of water conservation, energy consumption reduction and carbon emission reduction, thereby improving the overall efficiency and intelligence level of park resource management.

[0004] In a first aspect, the present invention provides a method for coordinated water-energy management in a park based on digital twins, comprising: constructing a digital twin of the park's water cycle system, wherein the digital twin integrates a plant water requirement model, a pipe network hydraulic model, and a rainwater storage model; receiving multi-source real-time monitoring data collected by the Internet of Things (IoT) sensing layer within the park, and processing the multi-source real-time monitoring data into a spatiotemporally unified format before synchronizing it to the digital twin; wherein the multi-source real-time monitoring data includes: environmental data reflecting the park's water cycle status and equipment operation data; based on the digital twin and its synchronized real-time data, generating multiple candidate water-energy scheduling schemes using a preset intelligent algorithm; wherein each candidate water-energy scheduling scheme includes: an irrigation control strategy for regulating the allocation of water resources in the park and an energy allocation strategy for optimizing the park's energy consumption; conducting virtual simulations of each candidate water-energy scheduling scheme through the digital twin, obtaining key data of each scheme in terms of water resource utilization, energy consumption, and environmental impact, and calculating the comprehensive score of each scheme according to a preset multi-objective comprehensive scoring rule, so as to select the scheme with the best comprehensive score as the water-energy scheduling execution scheme for the park.

[0005] In an optional implementation, a digital twin of the park's water cycle system is constructed, including: receiving park geospatial data, water cycle system equipment parameters, and historical operating data; wherein, the geospatial data includes: park topographic data and pipeline distribution coordinate data; the equipment parameters include: pump rated power and sensor monitoring range; the historical operating data includes: historical irrigation data and rainwater storage data; based on the geospatial data, a three-dimensional basic model of the park's water cycle system is constructed; the plant water requirement model, pipeline hydraulic model, and rainwater storage model are respectively embedded into the three-dimensional basic model to obtain a target three-dimensional model; wherein, the plant water requirement model is configured to receive meteorological data and plant growth stage data, and output dynamic irrigation demand; the pipeline hydraulic model is configured to receive pipeline flow data and pressure data, and simulate the dynamics of water flow within the pipeline network; the rainwater storage model is configured to receive historical rainfall data and rainwater collection device parameters, and predict changes in rainwater storage capacity; the target three-dimensional model is calibrated based on historical actual operating data, so that the deviation between the output data of the target three-dimensional model and the historical actual operating data is within a preset allowable range, thus obtaining a digital twin.

[0006] In an optional implementation, the multi-source real-time monitoring data is processed into a spatiotemporally unified format, including: cleaning the multi-source real-time monitoring data to remove outliers and supplementing missing values ​​using interpolation; extracting the timestamps and spatial location information of each monitoring data to establish a spatiotemporal index; and using a spatiotemporal interpolation algorithm based on the spatiotemporal index to match time-series data collected at different times with spatial data collected at different spatial locations, unifying the temporal granularity and spatial coordinate system of the data, and generating data in a spatiotemporally unified format.

[0007] In an optional implementation, a preset intelligent algorithm is used to generate multiple candidate water-energy scheduling schemes, including: constructing a state space based on real-time system state data output by a digital twin; the state space includes: soil moisture, rainwater storage, and pipeline pressure threshold; constructing an initial strategy set based on the state space, the initial strategy set includes: an initial irrigation strategy generated based on a preset irrigation rule base, and an initial energy allocation strategy generated based on a preset energy rule base; constructing a multi-objective reward function based on water-saving benefits, energy consumption costs, and carbon emission reduction gains, and using a reinforcement learning algorithm to dynamically optimize and explore the strategies in the initial strategy set to generate multiple candidate water-energy scheduling schemes.

[0008] In an optional implementation, a virtual simulation is conducted for each candidate water-energy scheduling scheme using a digital twin. This includes: taking the irrigation control strategy and energy allocation strategy in each candidate water-energy scheduling scheme as input, driving the plant water demand model, pipeline hydraulic model, and rainwater storage model in the digital twin to perform collaborative simulation; outputting the predicted data within the simulation period; the predicted data includes: the estimated rainwater recovery rate and the estimated effective utilization coefficient of irrigation water in the water resource utilization dimension, the estimated energy consumption and the estimated energy cost in the energy consumption dimension, and the estimated carbon emissions and the estimated carbon trading cost in the environmental impact dimension.

[0009] In an optional implementation, the comprehensive score is the result of a weighted sum of the water resource utilization efficiency score, energy consumption cost score, and environmental impact score.

[0010] In an optional implementation, after selecting the scheme with the best comprehensive score as the water-energy scheduling implementation scheme for the park, the method further includes: collecting actual execution data after the water-energy scheduling implementation scheme is implemented; the actual execution data includes: actual rainwater recovery rate, actual irrigation water effective utilization coefficient, actual energy consumption, actual energy consumption cost, actual carbon emissions, and actual carbon trading cost; comparing the actual execution data with the predicted data obtained from virtual simulation to generate deviation feedback; and using the deviation feedback to correct the parameters of the preset intelligent algorithm or the model parameters in the digital twin.

[0011] Secondly, this invention provides a digital twin-based water-energy collaborative management system for industrial parks, comprising: a construction module for constructing a digital twin of the park's water cycle system, wherein the digital twin integrates a plant water requirement model, a pipe network hydraulic model, and a rainwater storage model; and a receiving and synchronization module for receiving multi-source real-time monitoring data collected by the Internet of Things (IoT) sensing layer within the park, processing the multi-source real-time monitoring data into a spatiotemporally unified format, and synchronizing it to the digital twin; wherein the multi-source real-time monitoring data includes: environmental data reflecting the park's water cycle status and equipment operation data; The generation module is used to generate multiple candidate water-energy scheduling schemes based on the digital twin and its synchronized real-time data, using a preset intelligent algorithm. Each candidate water-energy scheduling scheme includes an irrigation control strategy for regulating the allocation of water resources in the park and an energy allocation strategy for optimizing the park's energy consumption. The simulation and determination module is used to conduct virtual simulations of each candidate water-energy scheduling scheme through the digital twin, obtain key data of each scheme in terms of water resource utilization, energy consumption, and environmental impact, and calculate the comprehensive score of each scheme according to the preset multi-objective comprehensive scoring rules, so as to select the scheme with the best comprehensive score as the water-energy scheduling execution scheme for the park.

[0012] Thirdly, the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the digital twin-based park water-energy collaborative management method as described in any of the foregoing embodiments.

[0013] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions, which, when executed by a processor, implement the digital twin-based park water-energy collaborative management method as described in any of the foregoing embodiments.

[0014] This invention establishes a high-fidelity virtual simulation environment for the park's water cycle by constructing a digital twin that integrates models of plant water requirements, pipeline hydraulics, and rainwater storage, and simultaneously processing multi-source real-time data into a unified spatiotemporal format. Based on this, intelligent algorithms generate multiple candidate schemes covering irrigation and energy allocation strategies. The digital twin then performs virtual simulations and multi-objective comprehensive scoring on each scheme to determine the optimal implementation scheme. This method effectively breaks down data barriers between water and energy systems, enabling collaborative water-energy decision-making. Utilizing the simulation and prediction capabilities of the digital twin, the overall effect of the scheme can be evaluated before actual implementation, significantly improving the scientific rigor and foresight of the decision-making process. Furthermore, it can achieve multi-objective optimization of water conservation, energy consumption reduction, and carbon emission reduction in a dynamically changing environment, improving the overall efficiency and intelligence level of park resource management. Attached Figure Description

[0015] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating a digital twin-based collaborative management method for water and energy in a park, provided as an embodiment of the present invention; Figure 2 A flowchart for processing multi-source real-time monitoring data into a spatiotemporally unified format is provided as an embodiment of the present invention; Figure 3 A functional module diagram of a park water-energy collaborative management system based on digital twins provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0018] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0019] The following detailed description of some embodiments of the present invention is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0020] Example 1 Figure 1 A flowchart of a digital twin-based water-energy collaborative management method for industrial parks is provided as an embodiment of the present invention, as follows: Figure 1 As shown, the method specifically includes the following steps: Step S102: Construct a digital twin of the park's water cycle system, which integrates a plant water requirement model, a pipe network hydraulic model, and a rainwater storage model.

[0021] The core of this step is to create a virtual model that accurately corresponds to the physical park's water cycle system, and to integrate specialized models to support subsequent decision-making. First, basic data related to the park's water cycle needs to be collected, including data reflecting the physical spatial morphology and equipment characteristics, as well as data demonstrating the system's historical operational patterns, providing a foundation for the construction of the digital twin. Next, 3D modeling tools are used to reconstruct the spatial distribution of physical entities such as the park's terrain, pipe network, and irrigation facilities, forming the basic framework of the digital twin. Subsequently, a plant water requirement model, a pipe network hydraulic model, and a rainwater storage model are embedded into this framework. The plant water requirement model dynamically calculates water demand based on plant characteristics and environmental factors; the pipe network hydraulic model simulates changes in water flow and pressure within the pipe network; and the rainwater storage model predicts fluctuations in rainwater storage. Finally, by comparing the simulated data with historical actual data, model parameters are adjusted to calibrate the digital twin, ensuring it accurately reflects the physical system's state.

[0022] Step S104: Receive multi-source real-time monitoring data collected by the IoT sensing layer within the park, process the multi-source real-time monitoring data into a spatiotemporally unified format, and then synchronize it to the digital twin.

[0023] The multi-source real-time monitoring data includes environmental data and equipment operation data that reflect the water cycle status of the park.

[0024] To transform the scattered data collected by the IoT sensing layer into standardized data that can be matched with the digital twin, the system first receives environmental data and equipment operation data reflecting the water cycle status of the park. This data comes from various sensors deployed within the park and directly reflects the real-time operating status of the system. Then, the received data undergoes preprocessing to remove outliers and supplement missing data, ensuring data quality. The key is to unify the temporal and spatial attributes of the data through technical means, eliminating differences in the temporal acquisition frequency and spatial acquisition location of different data, forming data in a unified spatiotemporal format. Finally, the processed standardized data is synchronized to the digital twin, enabling the digital twin to be updated in real-time to reflect the current state of the physical system, providing real-time data support for subsequent solution development.

[0025] Step S106: Based on the digital twin and its synchronized real-time data, a variety of candidate water-energy scheduling schemes are generated using a preset intelligent algorithm.

[0026] Each candidate water-energy scheduling scheme includes: an irrigation control strategy for regulating the allocation of water resources in the park and an energy allocation strategy for optimizing the park's energy consumption.

[0027] This invention generates multiple water-energy scheduling schemes by combining intelligent algorithms with the real-time status of a digital twin. First, real-time system operation status data is acquired using the digital twin. This data covers key information such as water supply and demand, pipeline operation, and energy-related aspects, forming the basis for scheme generation. Next, a pre-set intelligent algorithm is used to generate initial irrigation control and energy allocation strategies based on experience and rules. These initial strategies are then dynamically optimized using system feedback data. During optimization, the algorithm comprehensively considers multiple objectives such as water conservation, energy consumption, and carbon emission reduction. By iteratively adjusting strategy parameters, multiple candidate water-energy scheduling schemes are generated, providing options for subsequent selection of the optimal scheme.

[0028] Step S108: A virtual simulation is conducted on each candidate water-energy scheduling scheme using a digital twin to obtain key data on water resource utilization, energy consumption, and environmental impact of each scheme. The comprehensive score of each scheme is calculated according to the preset multi-objective comprehensive scoring rules, and the scheme with the best comprehensive score is selected as the water-energy scheduling execution scheme for the park.

[0029] After obtaining multiple candidate water-energy scheduling schemes, this embodiment of the invention further simulates the effects of the candidate schemes using a digital twin to select the optimal implementation scheme. First, each group of candidate water-energy scheduling schemes is input into the digital twin, which simulates the scheme's actual operation and outputs key data on water resource utilization, energy consumption, and environmental impact, thus predicting the scheme's implementation effect in advance. Then, based on a preset multi-objective comprehensive scoring rule, the key data of each scheme are scored. The scoring process balances the importance of different objectives, calculating the comprehensive score for each scheme. Finally, by comparing the comprehensive scores of all candidate schemes, the scheme with the highest score is determined as the park's water-energy scheduling implementation scheme, which achieves an optimal balance among multiple objectives.

[0030] This invention establishes a high-fidelity virtual simulation environment for the park's water cycle by constructing a digital twin that integrates models of plant water requirements, pipeline hydraulics, and rainwater storage, and simultaneously processing multi-source real-time data into a unified spatiotemporal format. Based on this, intelligent algorithms generate multiple candidate schemes covering irrigation and energy allocation strategies. The digital twin then performs virtual simulations and multi-objective comprehensive scoring on each scheme to determine the optimal implementation scheme. This method effectively breaks down data barriers between water and energy systems, enabling collaborative water-energy decision-making. Utilizing the simulation and prediction capabilities of the digital twin, the overall effect of the scheme can be evaluated before actual implementation, significantly improving the scientific rigor and foresight of the decision-making process. Furthermore, it can achieve multi-objective optimization of water conservation, energy consumption reduction, and carbon emission reduction in a dynamically changing environment, improving the overall efficiency and intelligence level of park resource management.

[0031] In an optional implementation, step S102, which involves constructing a digital twin of the park's water recycling system, specifically includes the following steps: Step S1021: Receive park geospatial data, water circulation system equipment parameters, and historical operating data; wherein, the geospatial data includes: park topographic data and pipeline distribution coordinate data; the equipment parameters include: pump rated power and sensor monitoring range; the historical operating data includes: historical irrigation data and rainwater storage data.

[0032] Specifically, this step is the data foundation stage for building the digital twin. Its core is to provide accurate and comprehensive basic information support for subsequent model building through multi-dimensional data collection, ensuring that the digital twin can accurately map the characteristics and operational patterns of the physical park's water cycle system. First, it receives the park's geospatial data. This type of data directly determines the spatial accuracy of the digital twin. The park's topographic data includes terrain features such as altitude, slope, and aspect in different areas of the park (e.g., the elevation difference between irrigation and rainwater harvesting areas affects rainwater runoff paths and irrigation water distribution). The pipe network distribution coordinate data includes details such as the laying location, diameter, pipe material, and node connection relationships of various water supply pipes, irrigation pipes, and rainwater harvesting pipes (e.g., the coordinates of key nodes in the pipe network serve as the spatial reference for subsequent hydraulic model simulations of pressure and flow distribution).

[0033] Secondly, the parameters of the water circulation system equipment are received. These parameters are key to the digital twin's reconstruction of the equipment's operating characteristics. The rated power of the water pump determines the pump's maximum water supply capacity and energy consumption calculation benchmark in the model. The sensor monitoring range (such as the measurement range of the soil moisture sensor and the detection accuracy of the pressure sensor) affects the accuracy of the model's perception of the physical system's state. It is necessary to ensure that the parameters are completely matched with the specifications of the actual deployed equipment to avoid distortion of the model simulation results due to parameter deviations.

[0034] Finally, historical operational data is received. This type of data is the core basis for subsequent model calibration and verification. Historical irrigation data records the changes in irrigation water volume in different seasons and different plant areas (such as high-frequency, high-volume irrigation in the tree area in summer and low-frequency, low-volume irrigation in the lawn area in winter), reflecting the pattern of water consumption in the park. Rainwater storage data includes the changes in water storage of rainwater collection devices under different rainfall conditions (such as rapid increase in storage after heavy rain and slow decrease in storage during drought), reflecting the utilization potential of rainwater resources. By integrating these three types of data, a complete data pool for the construction of a digital twin is formed.

[0035] Step S1022: Based on geospatial data, construct a three-dimensional basic model of the park's water cycle system.

[0036] This step aims to transform geospatial data into a visualized, structured digital spatial model, building the physical framework of the digital twin and providing a spatial carrier for subsequent domain model embedding. Optionally, based on the park's terrain data, a digital model of the park's terrain is first constructed using 3D modeling tools (such as BIM, Unity3D, etc.), reproducing the terrain undulations and regional divisions (such as the boundaries and locations of irrigation areas, green spaces, building areas, and rainwater harvesting areas), ensuring that the terrain model's elevation, slope, and other parameters are consistent with the actual park, laying the foundation for subsequent simulations of spatially related processes such as rainwater runoff and irrigation water infiltration.

[0037] Next, based on the coordinate data of the pipeline network distribution, the digital form of the pipeline network system is accurately restored in the terrain model, including the spatial direction of various pipelines, pipe diameter changes, and node locations (such as pump connection nodes, valve nodes, and sensor installation nodes). Key attributes (such as the hydraulic resistance coefficient corresponding to the pipe material and the equipment number corresponding to the node) are labeled for each pipeline network component, so that the pipeline network model not only has visualization features, but also can meet the functional requirements of subsequent hydraulic calculations.

[0038] Simultaneously, the core equipment in the water circulation system (such as water pumps, rainwater collection tanks, irrigation nozzles, and sensors) is digitally modeled and deployed in a 3D model according to its actual location and specifications. This ensures the accurate spatial connection between the equipment and the pipe network and terrain (such as the docking position between the water pump outlet and the water supply pipe, and the access point between the rainwater collection tank and the rainwater pipe network). Ultimately, a 3D basic model that can completely restore the spatial structure and equipment layout of the park's water circulation system is formed. This model is the core carrier for subsequent embedding of professional domain models and realizing "physical-virtual" mapping.

[0039] Step S1023: The plant water requirement model, the pipeline hydraulic model, and the rainwater storage model are embedded into the three-dimensional basic model to obtain the target three-dimensional model. The plant water requirement model is configured to receive meteorological data and plant growth stage data and output dynamic irrigation demand. The pipeline hydraulic model is configured to receive pipeline flow data and pressure data and simulate the dynamic flow of water in the pipeline. The rainwater storage model is configured to receive historical rainfall data and rainwater collection device parameters and predict changes in rainwater storage capacity.

[0040] By embedding three core domain models—plant water requirement model, pipeline hydraulic model, and rainwater storage model—into a three-dimensional basic model, the digital model is upgraded from a visualization tool to a functional decision-making tool, forming a target three-dimensional model with system simulation and prediction capabilities.

[0041] First, an embedded plant water requirement model is needed. This model requires pre-defined computational logic related to plant growth characteristics and environmental factors. It should be able to receive external meteorological data (temperature, humidity, wind speed, sunshine duration) and plant growth stage data (such as seedling stage, growth stage, dormancy stage). By calculating plant transpiration water consumption and root water absorption requirements, it should dynamically output precise irrigation needs for different plant regions and time periods. (For example, in a tree area during the growth stage with high temperature and humidity, the daily irrigation amount is 15m³.) 3 This provides a scientific basis for the formulation of irrigation strategies.

[0042] Secondly, an embedded pipeline hydraulic model is required. This model needs to integrate fluid dynamics calculation logic and be able to receive pipeline flow data (such as water pump outlet flow and irrigation nozzle outlet flow) and pressure data (such as pipeline node pressure and pipeline end pressure). It can simulate the dynamic processes of water flow velocity, pressure distribution, and head loss in the pipeline under different operating conditions (such as single pump operation, multi-pump linkage, and valve opening and closing). It can predict whether there are risks such as excessive pressure in the pipeline (which may lead to pipeline rupture) or excessive pressure (which may lead to uneven irrigation), and provide support for energy distribution (such as water pump power adjustment) and pipeline operation and maintenance.

[0043] Finally, a rainwater storage model is embedded. This model needs to establish the correlation logic between rainwater collection, consumption, and storage. It should be able to receive historical rainfall data (such as rainfall frequency, rainfall amount, and rainfall duration) and rainwater collection device parameters (such as volume, inflow rate, and outflow rate). By analyzing historical rainfall patterns, it should predict the amount of rainwater collected in the future. Combined with rainwater consumption (such as irrigation water and landscape water), it should calculate the storage volume change curve and output the prediction results of "when rainwater irrigation can be relied upon and when external water supply is needed", providing a basis for water resource collaborative utilization decisions.

[0044] After embedding the above three types of models into the three-dimensional basic model according to the principles of spatial matching and data interoperability, the three types of models can interact with the equipment parameters based on the spatial relationship of the three-dimensional model (such as the irrigation demand data of the plant water requirement model being transmitted to the pipeline hydraulic model to calculate the flow and pressure that the pipeline needs to provide), and finally form a target three-dimensional model with the ability to simulate the system, predict demand, and predict risks.

[0045] Step S1024: The target 3D model is calibrated based on historical actual operation data so that the deviation between the output data of the target 3D model and the historical actual operation data is within a preset allowable range, thereby obtaining a digital twin.

[0046] The core of this step is to verify the accuracy of the target 3D model through historical operational data, adjust model parameters to eliminate deviations between virtual simulation and physical reality, and ensure that the digital twin can reliably reflect the operating status and changing patterns of the physical system. First, the core indicators for model calibration are determined. Based on key parameters in historical operational data (such as historical irrigation volume, historical rainwater storage, and historical pipeline pressure and flow), verification benchmarks for the model output are set. For example, the "deviation rate between simulated and actual monthly irrigation volume," the "deviation rate between simulated and actual rainwater storage 24 hours after rainfall," and the "deviation rate between simulated and actual pipeline peak pressure" are used as core calibration indicators.

[0047] Next, the input conditions from historical operational data (such as meteorological data, plant growth stage, and rainfall data for a certain period) are input into the target 3D model. The model is run to obtain corresponding simulated output data (such as simulated irrigation volume, simulated rainwater storage, and simulated pipe network pressure for that period). Then, the simulated output data is compared with the actual output data from historical operational data to calculate the deviation values ​​of each core indicator. If the deviation value exceeds the preset allowable range (such as deviation rate > 5%), the cause of the deviation is analyzed and the model parameters are adjusted: for example, if the simulated irrigation volume is much lower than the actual value, it may be that the transpiration coefficient in the plant water requirement model is set too low, and the transpiration coefficient needs to be back-calculated and corrected based on the historical actual irrigation volume; if the simulated pipe network pressure is higher than the actual value, it may be that the pipe resistance coefficient in the pipe network hydraulic model is set too high, and the resistance coefficient needs to be optimized by combining historical pipe network pressure data; if the simulated rainwater storage value deviates greatly from the actual value, it may be that the inflow rate parameter in the rainwater storage model does not match the actual efficiency of the rainwater collection device, and the inflow rate calculation logic needs to be adjusted.

[0048] Repeat the above process of inputting historical conditions → obtaining simulated output → comparing and calculating deviations → adjusting model parameters until the deviations of all core indicators are within the preset allowable range. At this point, the target 3D model has the ability to accurately map the operating laws of the physical system and officially becomes a digital twin of the park's water cycle system, which can be used for subsequent data synchronization, scheme simulation, decision optimization, and other processes.

[0049] In one alternative implementation, such as Figure 2 As shown, in step S104 above, the multi-source real-time monitoring data is processed into a spatiotemporally unified format, specifically including the following steps: Step S1041: Perform data cleaning on the multi-source real-time monitoring data to remove outliers and use interpolation to supplement missing values.

[0050] To ensure the quality of multi-source real-time monitoring data and eliminate interference generated during data acquisition and transmission, providing reliable raw data for subsequent unified spatiotemporal processing, the following steps are taken: First, outliers in the data are removed. This can be done using preset anomaly detection rules, such as threshold methods based on normal data fluctuation ranges or the 3σ principle based on statistical distribution, to identify and filter data that does not conform to the actual system state, preventing outliers from misleading subsequent decisions. Next, this embodiment of the invention uses interpolation to supplement missing values. Specifically, based on the valid data sequence before and after the missing value, supplementary data consistent with the actual data change patterns is generated through linear interpolation, nearest neighbor interpolation, or trend interpolation based on historical data from similar periods, ensuring the continuity of the data sequence.

[0051] Step S1042: Extract the timestamps and spatial location information of each monitoring data and establish a spatiotemporal index for the data.

[0052] Specifically, the timestamp information of each monitoring data is first extracted. The timestamp needs to be accurate to the specific time of data collection (such as second or minute level), and the same time format (such as UTC time or local standard time) should be used to ensure that the data collected by different sensors are comparable in the time dimension.

[0053] Then, the spatial location information of each monitoring data is extracted. The spatial location information must be consistent with the spatial coordinate system of the park's digital twin. For example, the soil moisture sensor needs to be labeled with the specific irrigation area number and geographical coordinates where it is installed, the pipeline pressure sensor needs to be associated with the pipeline node number and location where it is located, and the rainwater level data needs to correspond to the specific rainwater collection device's identification and location, to ensure that each piece of data can be accurately mapped to a specific spatial location of the physical system.

[0054] Finally, based on the extracted timestamps and spatial location information, a spatiotemporal index is established. The spatiotemporal index adopts a time-space two-layer structure (such as grouping by timestamp first and then sorting by spatial location; or dividing by spatial region first and then arranging by time order), so that any data can be quickly located and retrieved by time and space keywords, providing efficient index support for the matching and unification of data from different dimensions in the future.

[0055] Step S1043: Based on the spatiotemporal index of the data, a spatiotemporal interpolation algorithm is used to match the time series data collected at different times with the spatial data collected at different spatial locations, unify the time granularity and spatial coordinate system of the data, and generate data in a spatiotemporally unified format.

[0056] First, to address the issue of inconsistent time granularity, a unified approach is needed. The collection frequencies of multi-source monitoring data may differ (e.g., meteorological data is collected every 10 minutes, soil moisture data every 5 minutes, and pipeline pressure data every 1 minute). Based on the time information in the spatiotemporal index, a time interpolation algorithm should be used (e.g., averaging high-frequency data with the time granularity of low-frequency data, or interpolating low-frequency data with the time granularity of high-frequency data to supplement it). All data should be unified to a preset time granularity (e.g., one data point every 5 minutes), so that different types of data can achieve "synchronization" in the time dimension. For example, it should be ensured that meteorological, soil moisture, pipeline pressure, and rainwater level data can be acquired simultaneously every 5 minutes.

[0057] Secondly, to address the issue of inconsistent spatial coordinate systems, a unified system needs to be established. If the local coordinates used by some sensors (such as "relative coordinates with the irrigation area entrance as the origin") differ from the global coordinate system (such as the geodetic coordinate system) of the digital twin of the park, a coordinate transformation algorithm is required to convert the local coordinates to global coordinates. At the same time, for sparsely distributed spatial data (such as an irrigation area with only one soil moisture sensor, which cannot reflect the spatial differences in soil moisture within the area), spatial interpolation algorithms (such as Kriging interpolation, inverse distance weighted interpolation, etc.) should be used based on the spatial location information in the spatiotemporal index to infer the system status data of areas where no sensors are deployed based on existing sensor data. This ensures that the data achieves "full coverage and consistent coordinates" in the spatial dimension, guaranteeing that every spatial area of ​​the digital twin has corresponding monitoring data support.

[0058] By unifying the temporal granularity and spatial coordinate system, data in a unified spatiotemporal format is generated. This type of data not only has consistency in the temporal and spatial dimensions, but also fully matches the temporal-spatial framework of the park's digital twin. It can be directly synchronized to the digital twin, providing standardized data input for the digital twin to accurately simulate and reflect the real-time operating status of the physical system.

[0059] In an optional implementation, step S106 above, which uses a preset intelligent algorithm to generate multiple candidate water-energy scheduling schemes, specifically includes the following steps: Step S1061: Based on the real-time system status data output by the digital twin, a state space is constructed; the state space includes: soil moisture, rainwater storage capacity, and pipeline pressure threshold.

[0060] Real-time system status data is key information output by the digital twin after synchronizing data from the IoT sensing layer and combining it with built-in models. It directly reflects the current resource supply and demand and equipment operation status of the system.

[0061] Among them, soil moisture data needs to cover real-time moisture values ​​of different irrigation areas and different soil depths (such as surface and root layer moisture) to determine the current water supply and demand gap of plants, which is the core basis for the formulation of irrigation strategies; rainwater storage data needs to include the real-time liquid level and total storage of each rainwater collection device to reflect the available potential of rainwater resources and determine whether to prioritize rainwater irrigation to reduce external water supply energy consumption; pipeline pressure threshold data needs to cover the real-time pressure values ​​and safe pressure range of key nodes in the pipeline network to determine the current operating load of the pipeline network and avoid pipeline pressure exceeding the standard (causing leakage risk) or insufficient pressure (affecting irrigation effect) due to strategy implementation.

[0062] By integrating the above three types of core data, a state space that can fully describe the system's operating status can be constructed. This space not only contains the real-time values ​​of each parameter, but also implies the correlation between parameters (such as when the soil moisture is low and the rainwater storage is sufficient, rainwater irrigation should be started first, and the pipeline pressure should be adjusted accordingly). This provides clear state input for the generation of subsequent initial strategies, ensuring that the strategies can accurately adapt to the current system operating conditions.

[0063] Step S1062: Based on the state space, construct an initial strategy set, which includes: an initial irrigation strategy generated based on a preset irrigation rule base and an initial energy allocation strategy generated based on a preset energy rule base.

[0064] Specifically, in this embodiment of the invention, an initial irrigation strategy is generated based on a preset irrigation rule base. The preset irrigation rule base is a set of rules constructed based on the park's water resource management experience, plant physiological characteristics, and historical operational data. It contains the corresponding logic of "state-action". For example, when the soil moisture is lower than the suitable threshold of the root layer and the rainwater storage is ≥50% and there is no rainfall forecast, rainwater irrigation is started and the irrigation duration is set to 30 minutes; when the soil moisture is higher than the suitable threshold or there is a rainstorm forecast, irrigation is suspended and the irrigation valve is closed, etc. The algorithm will match the corresponding rules in the rule base according to the soil moisture, rainwater storage, and other data in the current state space to generate an initial irrigation strategy for different irrigation areas.

[0065] Meanwhile, an initial energy allocation strategy is generated based on a preset energy rule base. The preset energy rule base focuses on the economy and low carbon emissions of energy utilization, and integrates the peak and valley characteristics of the power grid load, the carbon price fluctuation pattern and equipment energy consumption parameters. It includes the corresponding logic of "state-energy allocation". For example, when the power grid price is low and the carbon price is lower than the benchmark value, the power grid is used to drive high-power water pumps. When the carbon price is high and the rainwater storage is sufficient, the system is switched to rainwater recycling system to reduce the high carbon energy consumption of external water supply. When the pipeline pressure is lower than the safety threshold, the water pump power is adjusted to increase the pipeline pressure, and low-energy-consumption water pumps are selected for operation.

[0066] The pre-defined intelligent algorithm combines data such as pipeline pressure thresholds, carbon prices, and electricity prices in the current state space with rules in the energy rule base to generate an energy allocation strategy that coordinates with the initial irrigation strategy (e.g., if the irrigation strategy requires starting a water pump, the energy strategy specifies the energy supply type and operating power of the water pump). Finally, the initial irrigation strategy and the initial energy allocation strategy are combined to form an initial strategy set. Each strategy in this set is feasible and can be directly used as an initial sample for subsequent reinforcement learning optimization.

[0067] Step S1063: Construct a multi-objective reward function based on water-saving benefits, energy consumption costs, and carbon emission reduction gains, and use a reinforcement learning algorithm to dynamically optimize and explore the strategies in the initial strategy set to generate a variety of candidate water-energy scheduling schemes.

[0068] This step is the core of achieving dynamic optimization and diversified exploration of the strategy. By constructing a multi-objective reward function to guide the reinforcement learning algorithm, iterative optimization is performed based on the initial strategy to generate multiple sets of candidate solutions that take into account water conservation, energy conservation, and low carbon emissions, thus avoiding the imbalance of comprehensive benefits caused by single-objective optimization.

[0069] Specifically, a multi-objective reward function is pre-constructed, which simultaneously considers three core objectives: water-saving benefits, energy costs, and carbon emission reduction gains. Users can quantify the importance of each objective by setting weight coefficients (e.g., water-saving benefits have the highest weight to ensure efficient water resource utilization; energy costs are the second highest to control operating costs; and carbon emission reduction gains are the last to respond to low-carbon management requirements). The reward function value will dynamically change according to the actual effect after the strategy is implemented. For example, if the water-saving rate increases, energy costs decrease, and carbon emission reductions increase after the strategy is implemented, the reward value will increase; conversely, the reward value will decrease, thereby guiding the algorithm to optimize towards the optimal multi-objective direction.

[0070] Next, a reinforcement learning algorithm is used to dynamically optimize and explore the initial policy set. The algorithm inputs the initial policy into a digital twin and simulates the policy execution process through the digital twin to obtain feedback data after policy execution (such as actual water saving rate, energy cost, carbon emissions, and changes in pipeline pressure). The feedback data is then substituted into a multi-objective reward function to calculate the reward value. The algorithm judges the merits of the current policy based on the reward value, retains and fine-tunes the parameters of policies with good performance (high reward value), and corrects or replaces policies with poor performance (low reward value).

[0071] Simultaneously, the algorithm will explore diverse approaches, introducing small parameter perturbations (such as adjusting irrigation duration or switching energy supply types) based on the current strategy to generate new strategies. The effectiveness is then verified using a digital twin, preventing the algorithm from getting trapped in local optima. The iterative process of "strategy execution simulation - feedback data acquisition - reward value calculation - strategy optimization and exploration" is repeated until a predetermined number (e.g., 10 to 15 groups) of candidate water-energy scheduling schemes are generated. Each group of schemes presents a different balance between water conservation, energy consumption, and carbon emission reduction goals (e.g., some schemes emphasize water conservation, while others emphasize low carbon emissions), providing ample selection space for subsequent screening of the optimal comprehensive solution.

[0072] In an optional implementation, step S108 above, which involves conducting virtual simulations of each candidate water-energy scheduling scheme using a digital twin, specifically includes the following steps: Step S1081: The irrigation control strategy and energy allocation strategy in each candidate water-energy scheduling scheme are used as inputs to drive the plant water demand model, pipeline hydraulic model and rainwater storage model in the digital twin to perform collaborative simulation.

[0073] The essence of this step is to transform the abstract candidate water-energy scheduling schemes into simulation instructions executable by the digital twin. Through multi-model collaborative computation, the actual operation of the schemes in the physical system is simulated, ensuring that the simulation results accurately reflect the implementation effects of the schemes. First, the irrigation control strategy and energy allocation strategy in the candidate schemes need to be parsed. Irrigation-related rules such as "irrigation area division, irrigation duration and frequency, and rainwater utilization priority," as well as energy-related rules such as "pump start / stop times, energy supply type (grid power / renewable energy), and equipment operating power," are transformed into parameterized instructions recognizable by each model in the digital twin. For example, "prioritize rainwater irrigation" is parsed as "the rainwater storage model outputs flow and prioritizes allocation to the irrigation network," and "start pumps during off-peak electricity periods" is parsed as "the grid energy supply module outputs corresponding power during specified periods."

[0074] Next, the parsed instructions are input into the plant water requirement model, the pipeline hydraulic model, and the rainwater storage model in the digital twin, triggering multi-model collaborative simulation. The plant water requirement model calculates the actual water requirement and water absorption efficiency of the plants during the simulation period based on the irrigation volume and irrigation time period in the irrigation control strategy, combined with real-time meteorological data (such as temperature and humidity) and plant growth stage parameters, to determine whether the irrigation strategy meets the plant's needs. The pipeline hydraulic model simulates the pressure distribution and water flow velocity changes within the pipeline network based on the irrigation flow demand and the pump power parameters in the energy distribution strategy, verifying whether the pipeline network can withstand the operating load set by the scheme and whether there are problems such as excessive pressure leading to leakage or insufficient pressure leading to uneven irrigation. The rainwater storage model combines the rainwater utilization amount in the irrigation strategy with historical rainfall patterns to simulate the consumption and replenishment process of rainwater storage during the simulation period, determining whether rainwater resources can support irrigation needs and whether external water supply is required.

[0075] During the simulation, the three models do not operate independently, but rather collaborate through data interaction: the water requirement data output by the plant water requirement model is transmitted in real time to the pipeline hydraulic model as input for pipeline flow calculation; the pipeline pressure data output by the pipeline hydraulic model is fed back to the energy allocation strategy module to adjust the pump operating power; the available rainwater data output by the rainwater storage model affects the irrigation water allocation logic of the plant water requirement model, forming a closed-loop simulation of demand-supply-transmission, fully restoring the linkage relationship and state changes of each system during the execution of the scheme.

[0076] Step S1082: Output the prediction data for the simulation period. The prediction data includes: the estimated rainwater recycling rate and the estimated irrigation water effective utilization coefficient in the water resource utilization dimension; the estimated energy consumption and the estimated energy consumption cost in the energy consumption dimension; and the estimated carbon emissions and the estimated carbon trading cost in the environmental impact dimension.

[0077] The purpose of this step is to extract and output key predictive data that supports the comprehensive evaluation of the scheme based on the calculation results of multi-model collaborative simulation. This data covers three core dimensions: water resource utilization, energy consumption, and environmental impact, ensuring that the data directly reflects the scheme's performance on different objectives. First, in the dimension of water resource utilization, the estimated rainwater recovery rate and the estimated irrigation water effective utilization coefficient are output: the estimated rainwater recovery rate is calculated using a rainwater storage model, which is the proportion of rainwater actually used for irrigation during the simulation period to the total rainwater storage, reflecting the scheme's efficiency in utilizing rainwater resources; the estimated irrigation water effective utilization coefficient is calculated using a plant water requirement model, which is the proportion of irrigation water actually absorbed by plants to the total irrigation water, excluding water waste caused by excessive irrigation and pipeline leakage (simulated scenario), reflecting the scheme's water-saving effect.

[0078] Secondly, in terms of energy consumption, the estimated energy consumption and estimated energy cost are output: the estimated energy consumption is calculated collaboratively by the pipeline hydraulic model and the energy supply module, that is, the total energy consumed by all water system equipment (such as pumps and valves) during the simulation period (including grid electricity, renewable energy, etc.), distinguishing the consumption ratio of different energy types; the estimated energy cost combines the electricity price time data in the energy allocation strategy (such as the difference between off-peak and peak electricity prices) and the unit cost of different energy sources, converting the estimated energy consumption into economic cost, intuitively reflecting the energy economy of the scheme.

[0079] Finally, in terms of environmental impact, the projected carbon emissions and carbon trading costs are output: the projected carbon emissions are calculated using energy consumption data and a carbon footprint model, specifically the total carbon emissions corresponding to different energy types (such as thermal power and photovoltaics), combined with the carbon emission reduction from rainwater utilization to reduce external water supply, resulting in the net carbon emissions of the proposed solution; the projected carbon trading costs are calculated based on real-time carbon quota trading prices, according to the difference between the projected carbon emissions and the park's carbon quota (carbon revenue is generated if emissions are lower than the quota, and carbon costs are generated if emissions are higher than the quota), reflecting the impact of the proposed solution on the park's carbon asset management. These multi-dimensional predictive data not only comprehensively present the implementation effects of candidate solutions but also provide a quantitative basis for subsequent selection of the optimal solution based on multi-objective comprehensive scoring rules, ensuring that the selection results are scientific and objective.

[0080] In one alternative implementation, the comprehensive score is the weighted sum of the water resource utilization efficiency score, energy consumption cost score, and environmental impact score.

[0081] Specifically, the core of the comprehensive scoring is to integrate the scores of three dimensions—water resource utilization efficiency, energy consumption cost, and environmental impact—into a single quantitative indicator through weighted summation. This enables an objective assessment of the multi-objective balancing capability of candidate water-energy scheduling schemes. The calculation process revolves around four logical steps: defining scoring dimensions, generating single-dimensional scores, allocating weights, and weighted summation. This ensures that the scoring results reflect both the independent value of each dimension and the comprehensive benefits of the scheme.

[0082] First, the evaluation logic of the three core scoring dimensions is clarified to provide a basis for generating single-dimensional scores. Among them, the water resource utilization efficiency score focuses on the scheme's ability to conserve and efficiently utilize water resources. It is mainly calculated based on the estimated rainwater recovery rate and the estimated irrigation water effective utilization coefficient output from virtual simulations. A higher estimated rainwater recovery rate (meaning more efficient use of rainwater resources and reduced reliance on external water supply) and a higher estimated irrigation water effective utilization coefficient (meaning a higher proportion of irrigation water is absorbed by plants, avoiding water waste) results in a higher water resource utilization efficiency score. This dimension score is directly related to the core objective of this invention's embodiments in addressing the extensive water resource allocation in traditional systems and is a key indicator for measuring the water-saving effect of the scheme.

[0083] The energy cost score focuses on evaluating the energy economy of the solution. It is based on the estimated energy consumption and estimated energy cost output from the virtual simulation. A higher estimated energy consumption (meaning less energy is consumed by the water system equipment) and a lower estimated energy cost (meaning more economical energy expenditure after combining peak-valley electricity pricing strategies) results in a higher energy cost score. This score addresses the shortcomings of traditional technologies where pump start-up and shutdown are not linked to grid load peaks and valleys, demonstrating the energy cost control capability of this invention's implementation, and is an important consideration for ensuring the economic efficiency of park operations.

[0084] The environmental impact assessment focuses on the low-carbon attributes of the proposed solution, primarily based on the estimated carbon emissions and estimated carbon trading costs generated from virtual simulations. The lower the estimated carbon emissions (meaning less carbon emissions will be generated during the implementation of the solution) and the lower the estimated carbon trading costs (even generating carbon revenue, i.e., revenue when emissions are lower than the park's carbon allowance), the higher the environmental impact score. This dimension of the score addresses the shortcomings of existing technologies in quantifying carbon footprints and supporting trading, and is a core manifestation of the solution's responsiveness to low-carbon management and its ability to support carbon trading in the park.

[0085] Secondly, after generating scores for each dimension, reasonable weights need to be assigned to the three dimensions based on the park's management priorities and specific technical objectives. Weight allocation should adhere to the principle of prioritizing core objectives and balancing multiple objectives. Optionally, water resource utilization efficiency, as a core breakthrough point for addressing the low collaborative efficiency and water waste of traditional systems, should have the highest weight (e.g., set to 0.6); energy consumption costs directly affect the park's operational economics and are a key support for the feasibility of the solution, so its weight is second highest (e.g., set to 0.3); environmental impact, while an important objective, needs to achieve low carbon emissions based on water and energy conservation, so its weight is relatively low (e.g., set to 0.1). The weight values ​​need to be calibrated based on the actual needs of the park, but the core logic must prioritize water conservation while considering energy conservation and low carbon emissions, ensuring consistency with the overall concept of "water-energy-carbon synergistic optimization."

[0086] Finally, the comprehensive score is calculated using a weighted summation formula, which can be expressed as: Comprehensive Score = Water Resource Utilization Efficiency Score × Water Resource Weight + Energy Cost Score × Energy Weight + Environmental Impact Score × Environmental Weight. During the calculation, each individual dimension score must first be standardized (e.g., converted to a uniform score range of 0-100) to avoid weight invalidation due to differences in scoring scales across different dimensions. Simultaneously, the weighted summation result must retain a certain precision (e.g., retaining two decimal places) to effectively distinguish the differences in comprehensive benefits among different candidate schemes. This provides a clear and comparable quantitative basis for subsequently selecting the implementation scheme with the optimal comprehensive score, ultimately achieving a scheme selection logic that balances energy costs and carbon emission reduction targets while conserving water.

[0087] In one optional implementation, after selecting the scheme with the best comprehensive score as the water-energy scheduling implementation scheme for the park, the present invention further includes the following steps: Step S201: Collect actual execution data after the water-energy scheduling execution plan is implemented; the actual execution data includes: actual rainwater recovery rate, actual irrigation water effective utilization coefficient, actual energy consumption, actual energy consumption cost, actual carbon emissions, and actual carbon trading cost.

[0088] The core of this step is to establish a data comparison foundation between virtual simulation and actual execution. Through multi-dimensional, high-precision data collection, the actual implementation effect of the water-energy scheduling execution plan in the physical system is fully captured, providing an objective basis for subsequent deviation analysis and parameter correction. First, it is necessary to clarify the coverage of data collection, focusing on the core dimensions corresponding to the virtual simulation prediction data to ensure the comparability and correlation of the data. Among them, the water resource utilization dimension needs to collect the actual rainwater recovery rate and the actual irrigation water effective utilization coefficient: the actual rainwater recovery rate is calculated by comparing the liquid level change data of the rainwater collection device (the difference in liquid level before and after execution) with the actual amount of rainwater used for irrigation, reflecting the actual utilization efficiency of rainwater resources; the actual irrigation water effective utilization coefficient is estimated by comparing the plant growth status monitoring data (such as leaf humidity and growth rate) with the actual total irrigation volume, eliminating water waste caused by uneven irrigation and soil seepage, and accurately measuring the actual water-saving effect of the irrigation strategy.

[0089] In terms of energy consumption, it is necessary to collect actual energy consumption and actual energy cost: actual energy consumption is obtained through the power metering devices of water system equipment (pumps, valves, sensors, etc.), and the actual consumption of different energy types (such as grid electricity and renewable energy) is distinguished; actual energy cost is combined with the peak and valley electricity price periods of the grid and the unit cost of different energy sources to convert actual energy consumption into economic cost, and compare it with the estimated energy cost in the virtual simulation to determine whether the economic efficiency of the energy allocation strategy meets the standards.

[0090] In terms of environmental impact, actual carbon emissions and actual carbon trading costs need to be collected. Actual carbon emissions are calculated based on carbon emission coefficients for different energy types (such as carbon emission coefficients for thermal power and zero carbon emissions for photovoltaic power) and actual energy consumption, while also considering the carbon emissions reduced by rainwater utilization replacing external water supply, resulting in net carbon emissions. Actual carbon trading costs are calculated based on the park's carbon quota, real-time carbon trading prices, and the difference between actual carbon emissions and quotas (emissions below the quota generate carbon revenue, while emissions above the quota generate carbon costs), fully reflecting the actual impact of the plan on the park's carbon asset management. During the data collection process, the real-time nature and accuracy of the data must be ensured. Automatic data collection and synchronization should be achieved through sensors, metering devices, and interfaces with the carbon trading platform in the IoT sensing layer, avoiding errors caused by manual recording and laying a reliable data foundation for subsequent deviation analysis.

[0091] Step S202: Compare the actual execution data with the predicted data obtained from the virtual simulation to generate deviation feedback.

[0092] Specifically, the actual execution data collected in step S201 and the predicted data output in step S1082 are mapped one-to-one according to their dimensions. First, the deviation value and deviation rate of each indicator are calculated (deviation rate = (actual value - predicted value) / predicted value × 100%). By comparing the data, the differences between the virtual simulation and the actual execution are identified, and targeted deviation feedback is generated to clarify the direction for parameter correction and solve the problem of the disconnect between virtual simulation and physical reality. For example, in the dimension of water resource utilization, the actual rainwater recovery rate is compared with the estimated rainwater recovery rate, and the deviation rate between the two is calculated to determine whether the prediction of the rainwater storage model is accurate; in the dimension of energy consumption, the actual energy consumption is compared with the estimated energy consumption to analyze whether the deviation stems from the difference between the actual operating power of the water pump and the preset parameters of the model; in the dimension of environmental impact, the actual carbon emissions are compared with the estimated carbon emissions to investigate whether the deviation is caused by the discrepancy between the actual proportion of energy types and the simulation assumptions.

[0093] Next, an attribution analysis needs to be performed on the deviations to distinguish the types of sources of the deviations: If the deviation stems from inaccurate parameters of the digital twin model (e.g., the preset value of the transpiration coefficient in the plant water requirement model does not match the actual physiological characteristics of the plant, resulting in a deviation between the estimated irrigation amount and the actual demand), it is marked as "model parameter deviation"; if the deviation stems from the decision logic of the preset intelligent algorithm (e.g., the unreasonable setting of the reward function weight in reinforcement learning, resulting in the generated irrigation strategy having higher energy consumption in the actual scenario), it is marked as "algorithm parameter deviation"; if the deviation stems from sudden changes in the external environment (e.g., sudden rainfall causing the actual rainwater recovery rate to be higher than the estimate, which is an unforeseen meteorological change during the simulation), it is marked as "environmental interference deviation". This type of deviation does not require parameter correction and only needs to be used as a reference for environmental variables in the subsequent scheme generation.

[0094] Finally, based on the quantified deviation values ​​and attribution results, a structured deviation feedback report is generated, clearly defining the specific values, deviation rates, deviation types, and possible influencing factors for each dimension of deviation. For example, "The actual effective utilization coefficient of irrigation water is 0.75, while the estimated value is 0.85, resulting in a deviation rate of -11.8%. This is attributed to model parameter deviation, possibly due to an excessively high setting of the root capillary effect parameter in the plant water requirement model, leading to an overestimation of water absorption efficiency." This provides clear and actionable guidance for parameter correction in subsequent steps.

[0095] Step S203: Use deviation feedback to correct the parameters of the preset intelligent algorithm or the model parameters in the digital twin.

[0096] The core of this step is to adjust the key parameters of the preset intelligent algorithm and digital twin model based on the deviation feedback results, so as to narrow the gap between virtual simulation and actual execution and ensure the accuracy and adaptability of subsequent scheduling schemes.

[0097] To address the deviations in model parameters identified in the bias feedback, the core model parameters within the digital twin are optimized. If the plant water requirement model's transpiration coefficient and root capillary effect parameters are inaccurate, leading to a significant discrepancy between the actual and predicted irrigation water utilization coefficients, the corresponding parameters need to be corrected based on actual plant growth data. If the friction loss coefficient and local resistance coefficient of the pipeline hydraulic model are set unreasonably, resulting in actual energy consumption exceeding the prediction, these parameters are adjusted to match the actual pressure loss of the pipeline network. If the rainfall runoff coefficient and water storage capacity calculation parameters of the rainwater storage model are flawed, causing the actual rainwater recovery rate to differ from the prediction, the parameters need to be optimized based on actual rainfall and rainwater collection data to ensure that the model output better reflects the actual physical system.

[0098] To address the deviations in algorithm parameters identified by the bias feedback markers, key parameters of the preset intelligent algorithm are adjusted. If the carbon emission reduction gain weight in the multi-objective reward function of reinforcement learning is too low, resulting in actual carbon emissions exceeding the estimate, the weight ratio can be appropriately increased to guide the algorithm to generate scheduling strategies that prioritize low carbon emissions. If the pump start-stop time rules in the fuzzy rule base do not adequately adapt to the peak and off-peak electricity price characteristics, causing actual energy consumption costs to exceed the budget, the rule parameters are optimized to strengthen the decision-making logic of prioritizing pump start-up during off-peak electricity prices. If the algorithm is not sufficiently adaptable to abnormal operating conditions such as equipment aging, the reinforcement learning exploration rate parameter can be adjusted to increase the algorithm's strategy exploration for special operating conditions and improve decision robustness.

[0099] Parameter calibration must follow the principle of small-step adjustments and verification optimization. After each adjustment, a small-scale virtual simulation should be conducted using a digital twin to verify the calibration effect until the deviation is reduced to the preset allowable range. At the same time, the content, basis, and effect of each parameter adjustment should be recorded to form an adjustment log, providing data support for subsequent long-term optimization. This ensures that the digital twin and intelligent algorithms can continuously adapt to the dynamic changes of the park's water-energy system and maintain efficient scheduling capabilities.

[0100] Example 2 This invention also provides a digital twin-based park water-energy collaborative management system. This system is mainly used to execute the digital twin-based park water-energy collaborative management method provided in Embodiment 1 above. The following is a detailed description of the digital twin-based park water-energy collaborative management system provided in this invention.

[0101] Figure 3 A functional module diagram of a digital twin-based water-energy collaborative management system for a park is provided as an embodiment of the present invention, such as... Figure 3 As shown, the device mainly includes: a construction module 10, a receiving and synchronization module 20, a generation module 30, and a deduction and determination module 40, wherein: Module 10 is used to build a digital twin of the park's water cycle system. The digital twin integrates a plant water requirement model, a pipe network hydraulic model, and a rainwater storage model.

[0102] The receiving and synchronization module 20 is used to receive multi-source real-time monitoring data collected by the IoT sensing layer in the park, process the multi-source real-time monitoring data into a spatiotemporally unified format, and synchronize it to the digital twin; among which, the multi-source real-time monitoring data includes: environmental data reflecting the water cycle status of the park and equipment operation data.

[0103] The generation module 30 is used to generate multiple candidate water-energy scheduling schemes based on the digital twin and its synchronized real-time data using a preset intelligent algorithm; wherein each candidate water-energy scheduling scheme includes: an irrigation control strategy for regulating the allocation of water resources in the park and an energy allocation strategy for optimizing the energy consumption of the park.

[0104] The simulation and determination module 40 is used to conduct virtual simulations of each candidate water-energy scheduling scheme through a digital twin, obtain key data of each scheme in terms of water resource utilization, energy consumption and environmental impact, and calculate the comprehensive score of each scheme according to the preset multi-objective comprehensive scoring rules, so as to select the scheme with the best comprehensive score as the water-energy scheduling execution scheme of the park.

[0105] This invention establishes a high-fidelity virtual simulation environment for the park's water cycle by constructing a digital twin that integrates models of plant water requirements, pipeline hydraulics, and rainwater storage, and simultaneously processing multi-source real-time data into a unified spatiotemporal format. Based on this, intelligent algorithms generate multiple candidate schemes covering irrigation and energy allocation strategies. The digital twin then performs virtual simulations and multi-objective comprehensive scoring on each scheme to determine the optimal implementation scheme. This method effectively breaks down data barriers between water and energy systems, enabling collaborative water-energy decision-making. Utilizing the simulation and prediction capabilities of the digital twin, the overall effect of the scheme can be evaluated before actual implementation, significantly improving the scientific rigor and foresight of the decision-making process. Furthermore, it can achieve multi-objective optimization of water conservation, energy consumption reduction, and carbon emission reduction in a dynamically changing environment, improving the overall efficiency and intelligence level of park resource management.

[0106] Optionally, module 10 is specifically used for: It receives geospatial data of the park, equipment parameters of the water circulation system, and historical operating data. The geospatial data includes: park topography data and pipeline distribution coordinate data; the equipment parameters include: pump rated power and sensor monitoring range; and the historical operating data includes: historical irrigation data and rainwater storage data.

[0107] Based on geospatial data, a three-dimensional basic model of the park's water cycle system was constructed.

[0108] The plant water requirement model, the pipeline hydraulic model, and the rainwater storage model are embedded into the three-dimensional basic model to obtain the target three-dimensional model. The plant water requirement model is configured to receive meteorological data and plant growth stage data and output dynamic irrigation demand. The pipeline hydraulic model is configured to receive pipeline flow data and pressure data and simulate the dynamic water flow within the pipeline. The rainwater storage model is configured to receive historical rainfall data and rainwater collection device parameters and predict changes in rainwater storage capacity.

[0109] The target 3D model is calibrated based on historical actual operation data so that the deviation between the output data of the target 3D model and the historical actual operation data is within a preset allowable range, thus obtaining a digital twin.

[0110] Optionally, the receiving and synchronization module 20 is specifically used for: Data cleaning is performed on multi-source real-time monitoring data to remove outliers, and interpolation is used to supplement missing values.

[0111] Extract the timestamps and spatial location information of each monitoring data point to establish a spatiotemporal index for the data.

[0112] Based on spatiotemporal indexing, a spatiotemporal interpolation algorithm is used to match time-series data collected at different times with spatial data collected at different spatial locations, unifying the temporal granularity and spatial coordinate system of the data, and generating data in a spatiotemporally unified format.

[0113] Optionally, the generation module 30 is specifically used for: The real-time system status data output by the digital twin constitutes the state space; the state space includes: soil moisture, rainwater storage, and pipeline pressure threshold.

[0114] Based on the state space, an initial strategy set is constructed, which includes: an initial irrigation strategy generated based on a preset irrigation rule base and an initial energy allocation strategy generated based on a preset energy rule base.

[0115] A multi-objective reward function is constructed based on water-saving benefits, energy consumption costs, and carbon emission reduction gains. A reinforcement learning algorithm is then used to dynamically optimize and explore the strategies in the initial strategy set, generating a variety of candidate water-energy scheduling schemes.

[0116] Optionally, the deduction and determination module 40 is specifically used for: The irrigation control strategy and energy allocation strategy in each candidate water-energy scheduling scheme are used as inputs to drive the plant water demand model, pipeline hydraulic model and rainwater storage model in the digital twin to perform collaborative simulation.

[0117] Output the predicted data for the simulation period; the predicted data includes: the estimated rainwater recycling rate and the estimated irrigation water effective utilization coefficient in the water resource utilization dimension; the estimated energy consumption and the estimated energy consumption cost in the energy consumption dimension; and the estimated carbon emissions and the estimated carbon trading cost in the environmental impact dimension.

[0118] Optionally, the comprehensive score is the weighted sum of the water resource utilization efficiency score, energy consumption cost score, and environmental impact score.

[0119] Optionally, after selecting the scheme with the best comprehensive score as the water-energy scheduling execution scheme for the park, the system is also used for: Collect actual execution data after the water-energy scheduling implementation plan is implemented; the actual execution data includes: actual rainwater recovery rate, actual irrigation water effective utilization coefficient, actual energy consumption, actual energy consumption cost, actual carbon emissions, and actual carbon trading cost.

[0120] The actual execution data is compared with the predicted data obtained from the virtual simulation to generate deviation feedback.

[0121] Deviation feedback is used to correct the parameters of the preset intelligent algorithm or the model parameters in the digital twin.

[0122] Example 3 See Figure 4 This invention provides an electronic device, which includes a processor 60, a memory 61, a bus 62, and a communication interface 63. The processor 60, the communication interface 63, and the memory 61 are connected via the bus 62. The processor 60 is used to execute executable modules, such as computer programs, stored in the memory 61.

[0123] The memory 61 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 63 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc.

[0124] Bus 62 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 4 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0125] The memory 61 is used to store programs. After receiving an execution instruction, the processor 60 executes the program. The method executed by the apparatus defined by the process disclosed in any of the foregoing embodiments of the present invention can be applied to the processor 60 or implemented by the processor 60.

[0126] Processor 60 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 60 or by instructions in software form. Processor 60 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 61. Processor 60 reads the information in memory 61 and, in conjunction with its hardware, completes the steps of the above method.

[0127] The computer program product of the park water-energy collaborative management method and system based on digital twin provided in this invention includes a computer-readable storage medium storing non-volatile program code executable by a processor. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation, please refer to the method embodiments, which will not be repeated here.

[0128] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0129] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0130] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0131] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," "third," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0132] Furthermore, terms such as "horizontal," "vertical," and "sag" do not imply that components must be absolutely horizontal or suspended, but rather that they can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal relative to "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0133] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0134] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for coordinated water and energy management in a park based on digital twins, characterized in that, include: A digital twin of the park's water cycle system is constructed, which integrates a plant water requirement model, a pipe network hydraulic model, and a rainwater storage model. The system receives multi-source real-time monitoring data collected by the IoT sensing layer within the park, processes the multi-source real-time monitoring data into a spatiotemporally unified format, and then synchronizes it to the digital twin; wherein, the multi-source real-time monitoring data includes: environmental data and equipment operation data reflecting the water cycle status of the park; Based on the digital twin and its synchronized real-time data, a variety of candidate water-energy scheduling schemes are generated using a preset intelligent algorithm; wherein, each of the candidate water-energy scheduling schemes includes: an irrigation control strategy for regulating the allocation of water resources in the park and an energy allocation strategy for optimizing the energy consumption of the park. The digital twin is used to conduct virtual simulations of each of the candidate water-energy scheduling schemes, obtain key data on water resource utilization, energy consumption and environmental impact of each scheme, and calculate the comprehensive score of each scheme according to the preset multi-objective comprehensive scoring rules, so as to select the scheme with the best comprehensive score as the water-energy scheduling execution scheme of the park.

2. The method for coordinated water and energy management in a park based on digital twins according to claim 1, characterized in that, Constructing a digital twin of the park's water cycle system includes: The system receives geospatial data of the park, equipment parameters of the water circulation system, and historical operating data. The geospatial data includes: park topography data and pipeline distribution coordinates; the equipment parameters include: pump rated power and sensor monitoring range; and the historical operating data includes: historical irrigation data and rainwater storage data. Based on the geospatial data, a three-dimensional basic model of the park's water cycle system is constructed. The plant water requirement model, the pipeline hydraulic model, and the rainwater storage model are embedded into the three-dimensional basic model to obtain the target three-dimensional model. The plant water requirement model is configured to receive meteorological data and plant growth stage data, and output dynamic irrigation demand. The pipeline hydraulic model is configured to receive pipeline flow data and pressure data, and simulate the dynamic flow of water within the pipeline network. The rainwater storage model is configured to receive historical rainfall data and rainwater collection device parameters, and predict changes in rainwater storage capacity. The target 3D model is calibrated based on historical actual operating data so that the deviation between the output data of the target 3D model and the historical actual operating data is within a preset allowable range, thereby obtaining the digital twin.

3. The method for coordinated water and energy management in a park based on digital twins according to claim 1, characterized in that, Processing the multi-source real-time monitoring data into a spatiotemporally unified format includes: The multi-source real-time monitoring data is cleaned to remove outliers, and interpolation is used to fill in missing values. Extract the timestamps and spatial location information of each monitoring data point to establish a spatiotemporal index for the data. Based on the aforementioned spatiotemporal index, a spatiotemporal interpolation algorithm is used to match time-series data collected at different times with spatial data collected at different spatial locations, thereby unifying the temporal granularity and spatial coordinate system of the data and generating data in a spatiotemporally unified format.

4. The method for coordinated water and energy management in a park based on digital twins according to claim 1, characterized in that, Multiple candidate water-energy scheduling schemes are generated using a pre-set intelligent algorithm, including: The real-time system status data output by the digital twin constitutes a state space; the state space includes: soil moisture, rainwater storage capacity, and pipeline pressure threshold. Based on the state space, an initial strategy set is constructed, which includes: an initial irrigation strategy generated based on a preset irrigation rule base and an initial energy allocation strategy generated based on a preset energy rule base. A multi-objective reward function is constructed based on water-saving benefits, energy consumption costs, and carbon emission reduction gains. A reinforcement learning algorithm is then used to dynamically optimize and explore the strategies in the initial strategy set, generating a variety of candidate water-energy scheduling schemes.

5. The method for coordinated water and energy management in a park based on digital twins according to claim 1, characterized in that, Virtual simulations of each candidate water-energy scheduling scheme are conducted using the digital twin, including: The irrigation control strategy and energy allocation strategy in each candidate water-energy scheduling scheme are used as inputs to drive the plant water demand model, pipeline hydraulic model and rainwater storage model in the digital twin to perform collaborative simulation. Output the predicted data within the simulation period; the predicted data includes: the estimated rainwater recycling rate and the estimated irrigation water effective utilization coefficient in the dimension of water resource utilization, the estimated energy consumption and the estimated energy cost in the dimension of energy consumption, and the estimated carbon emissions and the estimated carbon trading cost in the dimension of environmental impact.

6. The method for coordinated water and energy management in a park based on digital twins according to claim 1, characterized in that, The comprehensive score is the result of a weighted sum of the water resource utilization efficiency score, energy consumption cost score, and environmental impact score.

7. The method for coordinated water and energy management in a park based on digital twins according to claim 1, characterized in that, After selecting the optimal solution based on the overall score as the water-energy dispatching implementation plan for the park, the following steps are also included: Collect actual execution data after the water-energy scheduling execution plan is implemented; the actual execution data includes: actual rainwater recovery rate, actual irrigation water effective utilization coefficient, actual energy consumption, actual energy consumption cost, actual carbon emissions, and actual carbon trading cost; The actual execution data is compared with the predicted data obtained from the virtual simulation to generate deviation feedback; The deviation feedback is used to correct the parameters of the preset intelligent algorithm or the model parameters in the digital twin.

8. A digital twin-based water-energy collaborative management system for industrial parks, characterized in that, include: A construction module is used to build a digital twin of the park's water cycle system, which integrates a plant water requirement model, a pipe network hydraulic model, and a rainwater storage model. The receiving and synchronization module is used to receive multi-source real-time monitoring data collected by the IoT sensing layer within the park, process the multi-source real-time monitoring data into a spatiotemporally unified format, and synchronize it to the digital twin; wherein, the multi-source real-time monitoring data includes: environmental data and equipment operation data reflecting the water cycle status of the park; The generation module is used to generate multiple candidate water-energy scheduling schemes based on the digital twin and its synchronized real-time data using a preset intelligent algorithm; wherein each candidate water-energy scheduling scheme includes: an irrigation control strategy for regulating the allocation of water resources in the park and an energy allocation strategy for optimizing the energy consumption of the park. The simulation and determination module is used to conduct virtual simulations of each of the candidate water-energy scheduling schemes through the digital twin, obtain key data of each scheme in terms of water resource utilization, energy consumption and environmental impact, and calculate the comprehensive score of each scheme according to the preset multi-objective comprehensive scoring rules, so as to select the scheme with the best comprehensive score as the water-energy scheduling execution scheme of the park.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, characterized in that, When the processor executes the computer program, it implements the digital twin-based water-energy collaborative management method for industrial parks as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions, which, when executed by a processor, implement the digital twin-based water-energy collaborative management method for industrial parks as described in any one of claims 1 to 7.

Images