Resource scheduling method and system based on water affair data

Abstract

The invention discloses an intelligent resource scheduling method and system based on water affair data, and the method comprises the steps: achieving the real-time collection and preprocessing of multi-modal water affair data through an edge calculation node, guaranteeing the data quality and safety through an adaptive sampling strategy and differential privacy protection, and achieving the non-tampering evidence storage of the water quality data; a demand prediction model based on a space-time diagram neural network is adopted, accurate prediction of partitioned water consumption is achieved through space-time convolution and a gating attention mechanism, an improved multi-target particle swarm algorithm is adopted, water supply stability, energy consumption minimization and leakage suppression are taken as targets, and an optimal scheduling scheme is screened in combination with a dynamic inertia weight and a fuzzy membership function; virtual deduction is carried out through a digital twin system, pressure abnormity is visualized, an intervention strategy is recommended, pump set operation is optimized in combination with a time-of-use electricity price strategy, and agent adding is dynamically adjusted based on water quality prediction. According to the method, the water supply stability can be improved, the energy consumption and the leakage rate are reduced, and the reliability and economy of a water affair system are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a resource scheduling method and system based on water data. BACKGROUND

[0002] With the acceleration of urbanization, the water system is facing severe challenges such as imbalance between supply and demand, aging of pipe network, high energy consumption, and lag in response to sudden accidents. The traditional scheduling method has the following technical bottlenecks:

[0003] Existing systems cannot effectively integrate cross-domain data such as pressure sensors (fixed sampling rate), smart water meters (privacy sensitive), and water quality monitoring (tamper-proofing requirements). For example, a provincial capital city misjudged water demand during a rainstorm in 2024 because it did not correlate with weather data, resulting in a 12-hour water outage in three urban areas;

[0004] The conventional SCADA system uses centralized data processing, and in a pipe explosion event at a water plant, the response delay from data collection to the central server was up to 8 minutes, missing the best disposal opportunity;

[0005] Water quality monitoring data stored in traditional databases have been tampered with to avoid punishment, such as altering pH records.

[0006] The technical solution of the present application addresses the above-mentioned pain points by ensuring data quality through edge computing and blockchain, implementing spatiotemporal joint prediction with ST-GNN, balancing multi-objective optimization with improved MOPSO algorithm, and realizing second-level emergency response with digital twin system, forming a complete technical closed loop. Application in a pilot city shows that this system can improve water supply stability by 40%, reduce energy consumption by 18%, and decrease leakage rate by 35%, with significant technological progress. SUMMARY

[0007] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0008] According to the first aspect of the present application, a resource scheduling method based on water data is claimed, comprising:

[0009] S1, collecting multi-modal water data, and performing data preprocessing on the multi-modal water data through an edge computing node;

[0010] S2, inputting the preprocessed multi-modal water data into a water demand prediction model, and outputting a partition water demand prediction result;

[0011] S3, setting an optimization target, and correcting the partition water demand prediction result to obtain a water resource scheduling scheme using a multi-objective particle swarm optimization algorithm;

[0012] S4, performing virtual deduction on the water resource scheduling scheme, and dynamically adjusting the water resource scheduling scheme according to a deduction result;

[0013] S5, monitoring an execution effect of the adjusted water resource scheduling scheme in real time, and feeding back to the water demand prediction model for online learning.

[0014] Further, the S1 further comprises:

[0015] real-time acquisition of pressure data of a water supply network, user-end intelligent water meter water consumption data, water source water quality monitoring data, meteorological environment data and industrial user water consumption plan data;

[0016] data preprocessing is performed by an edge computing node, including noise filtering, space-time alignment and missing value interpolation;

[0017] The S2 further comprises: inputting the preprocessed multi-modal water data into a water demand prediction model based on a space-time graph neural network (ST-GNN) model;

[0018] dynamically adjusting feature weights of different regions and time periods in combination with an attention mechanism, and outputting a future 24-hour partition water demand prediction result;

[0019] The S3 further comprises:

[0020] Taking water supply stability, energy consumption minimization and pipe network leakage suppression as optimization objectives, an improved multi-objective particle swarm optimization (MOPSO) algorithm is used to correct the partition water demand prediction result to generate a water resource scheduling scheme, and the multi-objective particle swarm optimization (MOPSO) algorithm introduces water equipment physical constraints and a dynamic reward and punishment mechanism;

[0021] The S4 further comprises:

[0022] The water resource scheduling scheme is synchronized to a water digital twin system for virtual deduction, and the water resource scheduling scheme is dynamically adjusted according to a deduction result and then issued to an execution terminal.

[0023] Further, the S1 further comprises:

[0024] The edge computing node adopts an adaptive sampling strategy, and a sampling frequency of the pressure data is positively correlated with a pipe network pressure fluctuation rate;

[0025] Differential privacy protection is implemented on user water consumption data, and noise conforming to a Laplace distribution is added;

[0026] The water quality data of the water source is ensured to be non-tamperable through blockchain storage, and the storage information includes collection time, position and sensor ID.

[0027] Further, the S2 further comprises:

[0028] The space-time graph neural network ST-GNN model comprises:

[0029] a space-time convolution module for extracting pipe network topology features;

[0030] a gated attention module for dynamically weighting the differences in water usage patterns between industrial users and residential users;

[0031] an online learning module for updating model parameters in real time according to scheduling feedback data;

[0032] When the pipe network leakage rate is monitored to exceed a threshold value, a hierarchical emergency scheduling is triggered to prioritize the water supply of key nodes;

[0033] An automatic valve closing scheme is generated to isolate the leakage area, and an optimal water replenishment path is calculated in combination with a hydraulic model.

[0034] Further, the S3 further comprises:

[0035] The improved multi-objective particle swarm optimization algorithm MOPSO comprises:

[0036] a dynamic inertia weight: the weight value is adaptively adjusted according to the variance of water supply pressure;

[0037] a constraint processing mechanism: the pump station start-stop frequency limit is converted into a penalty function;

[0038] Pareto solution screening: the optimal scheduling scheme is selected based on a fuzzy membership function.

[0039] Further, the S4 further comprises:

[0040] Robustness of the water resource scheduling scheme under extreme weather is simulated in the water digital twin system;

[0041] An abnormal pipe network pressure area is visualized through an AR interface, and an artificial intervention strategy is recommended;

[0042] A time-of-use electricity price strategy is introduced in the water resource scheduling scheme, and high-energy-consumption pump groups are preferentially started during off-peak electricity periods;

[0043] Based on the water quality prediction result, the water treatment chemical dosage is dynamically adjusted.

[0044] According to the second aspect of the present application, the present application claims to protect a water data-based resource scheduling system, comprising:

[0045] one or more processors;

[0046] A memory having one or more programs stored thereon, when the one or more programs are executed by the one or more processors, cause the one or more processors to implement the resource scheduling method based on water data.

[0047] The application discloses an intelligent resource scheduling method and system based on water data, which realizes real-time collection and preprocessing of multi-modal water data through an edge computing node, adopts an adaptive sampling strategy and differential privacy protection to ensure data quality and safety, and realizes tamper-proof evidence of water quality data; adopts a demand prediction model based on a space-time graph neural network to realize accurate prediction of partition water consumption through space-time convolution and a gated attention mechanism, adopts an improved multi-objective particle swarm algorithm to take water supply stability, energy consumption minimization and leakage suppression as targets, combines a dynamic inertia weight and a fuzzy membership function to filter an optimal scheduling scheme; performs virtual deduction through a digital twin system, visualizes pressure abnormalities and recommends intervention strategies, optimizes pump set operation in combination with a time-of-use electricity price strategy, and dynamically adjusts reagent addition based on water quality prediction. The method can improve water supply stability, reduce energy consumption and leakage rate, and significantly improve reliability and economy of a water system. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 A workflow diagram of a resource scheduling method based on water data claimed by the embodiments of the application;

[0049] Figure 2 A structural diagram of a resource scheduling method based on water data claimed by the embodiments of the application. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0051] The terms "first", "second", "third", etc. in the present application are only for descriptive purpose and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second", "third" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "a plurality of" is at least two, such as two, three, etc., unless otherwise explicitly and specifically limited. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present application are only used to explain the relative position relationship, movement condition, etc. between components in a certain posture (as shown in the drawings), and if the certain posture changes, the directional indications also change accordingly. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units is not limited to the listed steps or units, but can optionally include steps or units not listed, or can optionally include other steps or units inherent to the process, method, product or device.

[0052] Reference herein to "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment can be included in at least one embodiment of the application. The appearance of the phrase in various places in the specification does not necessarily all refer to the same embodiment, nor is it necessarily mutually exclusive of other embodiments. It is explicitly and implicitly understood that the embodiments described herein can be combined with other embodiments.

[0053] According to a first embodiment of the present application, with reference to Figure 1 , the present application claims a water affair data-based resource scheduling method, comprising:

[0054] S1, collecting multi-modal water affair data, and performing data preprocessing on the multi-modal water affair data through an edge computing node;

[0055] S2, inputting the preprocessed multi-modal water affair data into a water affair demand prediction model, and outputting a partition water demand prediction result;

[0056] S3, setting an optimization target, and correcting the partition water demand prediction result to obtain a water resource scheduling scheme by using a multi-objective particle swarm algorithm;

[0057] S4, virtually deducing the water resource scheduling scheme, and dynamically adjusting the water resource scheduling scheme according to a deduction result;

[0058] S5, monitoring an execution effect of the adjusted water resource scheduling scheme in real time, and feeding back to the water affair demand prediction model for online learning.

[0059] Further, the S1 further comprises:

[0060] Real-time acquisition of pressure data of water supply network, user-side intelligent water meter water consumption data, water source water quality monitoring data, meteorological environment data and industrial user water consumption plan data;

[0061] Data preprocessing is performed by the edge computing node, including noise filtering, space-time alignment and missing value interpolation;

[0062] The S2 further comprises: inputting the preprocessed multi-modal water affair data into a water affair demand prediction model based on a space-time graph neural network (ST-GNN) model;

[0063] The feature weights of different regions and time periods are dynamically adjusted in combination with an attention mechanism, and a 24-hour partitioned water demand prediction result is output;

[0064] The S3 further comprises:

[0065] Taking water supply stability, energy consumption minimization and pipe network leakage suppression as optimization objectives, an improved multi-objective particle swarm optimization (MOPSO) algorithm is used to correct the partitioned water demand prediction result to generate a water consumption resource scheduling scheme, and the multi-objective particle swarm optimization (MOPSO) algorithm introduces water affair equipment physical constraints and a dynamic reward and punishment mechanism;

[0066] The S4 further comprises:

[0067] The water consumption resource scheduling scheme is synchronized to a water affair digital twin system for virtual deduction, and the water consumption resource scheduling scheme is dynamically adjusted according to the deduction result and then issued to an execution terminal.

[0068] Further, the S1 further comprises:

[0069] The edge computing node adopts an adaptive sampling strategy, and the sampling frequency of the pressure data is positively correlated with the pipe network pressure fluctuation rate;

[0070] Differential privacy protection is implemented on the user water consumption data, and noise conforming to Laplace distribution is added;

[0071] The water quality data of the water source is ensured to be non-tamperable through blockchain storage, and the storage information includes collection time, location and sensor ID.

[0072] In this embodiment, the hardware deployment for water supply network pressure monitoring is to deploy 20 wireless pressure sensors (range 0-1.6MPa, accuracy ±0.3%FS) at key nodes of the main pipe network, and to use an NB-IoT transmission protocol.

[0073] The sensor is built-in with a three-axis vibration detection module, which can distinguish between real pressure fluctuation and mechanical vibration interference.

[0074] Normal pressure range 0.25-0.45 MPa, sampling reference frequency 5 minutes / once.

[0075] When the fluctuation rate of adjacent two data is >8%, automatically switch to high-frequency sampling of 1 second / once, and continue until the fluctuation rate <5% to restore the reference frequency.

[0076] The user end uses smart water meter data, and the data sources include:

[0077] Residential users: 500,000 ultrasonic water meters (±1% accuracy), upload data every 15 minutes.

[0078] Industrial users: 200 electromagnetic flowmeters (±0.5% accuracy), upload data every 5 minutes.

[0079] Privacy protection measures include:

[0080] Add Laplace noise (ε=0.1) to residential user data, and only aggregate and encrypt industrial user data for accurate measurement.

[0081] After adding noise, it is ensured that individual user data cannot be traced back, but the total error of the region is <3%.

[0082] The monitoring parameters of water source water quality monitoring include:

[0083] Basic five parameters (pH, turbidity, residual chlorine, COD, ammonia nitrogen), collected every 10 minutes.

[0084] Heavy metal indicators (lead, arsenic, etc.) are detected every 2 hours.

[0085] Each piece of data stored on the blockchain contains:

[0086] Timestamp (UTC standard time);

[0087] GPS coordinates (accuracy ±3m);

[0088] Sensor device ID (such as WQ-SENSOR-007);

[0089] Digital signature (ECDSA algorithm);

[0090] Storage delay <500ms, block confirmation time <2 seconds.

[0091] Meteorological environmental data are fused, and the API of the Meteorological Bureau is accessed to obtain 72-hour forecasts (temperature, precipitation, humidity) and real-time monitoring of local micro-meteorological stations (wind speed, evaporation) (used to correct the water consumption prediction model).

[0092] The data interface of the industrial user water plan includes:

[0093] Through the interface with the enterprise ERP system, obtain the production plan and water reservation (e.g., XX factory will increase production tomorrow from 9:00 to 12:00, and needs to increase 2000 m³ of water supply).

[0094] The data format adopts the ISO 8601 time standard, and the water quantity unit is accurate to 0.1 m³.

[0095] The edge computing node preprocessing includes noise filtering. For pressure data, a sliding window median filter (window width = 5 sampling points) is used to eliminate transient pulse interference.

[0096] A lightning strike caused the sensor P-205 data to suddenly change to 0 MPa, which was automatically identified and removed by the system.

[0097] For water quality data, based on the historical data distribution (e.g., pH is usually 6.5-8.5), data exceeding 3σ range triggers a review mechanism.

[0098] For spatio-temporal alignment, when time alignment, all data is aligned to the whole minute timestamp (e.g., 08:00:00), and lagging data is compensated by linear interpolation; when spatial alignment, scattered monitoring points are mapped to the pipe network GIS topology map, and regional interpolation is realized through Delaunay triangulation.

[0099] For missing value processing, the paint fan processing rules table is used, as shown in Table 1;

[0100] Table 1 Missing value processing rules table Type of deletion Treatment method Single point deletion (<5 minutes) Linear interpolation Long-term deletion (> 30 minutes) Hydraulic model-based simulation completion Water quality data deletion Take the average of other sensor data from the same water source

[0101] The adaptive sampling strategy is a dynamic adjustment logic:

[0102] The reference sampling interval is 5 minutes, and when the pressure change rate ΔP / P0>8%, the sampling frequency is increased to 1 Hz; continuous monitoring until ΔP / P0<5% for 10 minutes;

[0103] In a certain pipeline leakage event, the system switched to high-frequency sampling within 2 seconds after a pressure drop of 12%, accurately capturing the leakage characteristics.

[0104] For differential privacy protection, the noise addition rule is that when the true value of the resident user water quantity Q is 100 m³, the output value range after adding noise is: 90-110 m³ (ε=0.1);

[0105] Industrial user data is only added with noise after regional aggregation (to ensure that the total amount error is less than 1%).

[0106] For blockchain storage

[0107] Storage data structure: { "timestamp": "2023-11-20T08:00:00Z", "location": {"lat": 34.0522, "lng": -118.2437}, "sensor_id": "WQ-7890", "parameters": {"pH": 7.2, "turbidity": 0.8}, "signature": "0x3a7b...e2c4" }

[0108] When tamper-proof verification is performed, if the pH data is maliciously modified from 7.2 to 8.5, the blockchain hash verification will fail (hash collision probability <10^-18).

[0109] The following is illustrated with specific test cases:

[0110] Data processing under heavy rain weather:

[0111] Input data:

[0112] Pressure data: 3 monitoring points are damaged due to flood, data missing for 2 hours;

[0113] Water quality data: turbidity increases from 1.2 NTU to 12.8 NTU;

[0114] System response:

[0115] The edge node starts the hydraulic model simulation to complete the missing pressure data (error <5%);

[0116] Turbidity data triggers blockchain emergency evidence (evidence frequency increases to 1 minute / once);

[0117] Adjust the coagulant dosage of downstream water plant dynamically (from 20 mg / L to 35 mg / L);

[0118] Industrial user data privacy protection:

[0119] Test method:

[0120] Attackers obtain the annual water consumption data sequence of a resident user (with added noise);

[0121] Try to restore the true value through reverse engineering;

[0122] Result:

[0123] The average deviation of the restored data from the actual value is 22%, meeting the privacy protection requirement of ε=0.1.

[0124] The embodiment realizes reliable collection and safe preprocessing of multi-source water data by the above technical combination, and provides a high-quality data basis for subsequent intelligent scheduling.

[0125] Further, the S2 further comprises:

[0126] The space-time graph neural network ST-GNN model comprises:

[0127] A space-time convolution module is configured to extract a pipe network topology feature;

[0128] A gate attention module is configured to dynamically weight the water use mode difference between industrial users and residential users;

[0129] An online learning module is configured to update model parameters in real time according to scheduling feedback data;

[0130] When it is monitored that the pipe network leakage rate exceeds a threshold value, hierarchical emergency scheduling is triggered to preferentially guarantee water supply of key nodes;

[0131] A valve closing scheme is automatically generated to isolate a leakage area, and an optimal water supplement path is calculated in combination with a hydraulic model.

[0132] In the embodiment, space-time graph construction is performed during input data preparation, the nodes are 185 pipe network nodes (including water source plants, booster stations and user access points), the edges are 230 pipelines, and the weight = 1 / pipeline diameter (the weight of a DN200 pipeline = 0.005, and the weight of a DN400 = 0.0025);

[0133] According to static characteristics (pipeline diameter, material and elevation) and dynamic characteristics (pressure and flow in the past 24 hours);

[0134] Prediction process (taking City A as an example):

[0135] Spatial convolution is performed to identify the water supply correlation between a commercial area (nodes 23-45) and an industrial area (nodes 78-92);

[0136] For example, the spatial influence coefficient of industrial area node 81 on commercial area node 35 = 0.17;

[0137] Time convolution is used to detect that the water consumption of a residential area in the evening peak (18:00-20:00) exists a 15-minute delay propagation;

[0138] For an emergency scheduling response mechanism, for example, a DN300 main pipe burst (leakage rate exceeds a threshold value of 15%), a hierarchical response process is used;

[0139] Pressure sensor P-208 is used to detect a 0.2 MPa sudden drop (lasting for 5 minutes) for leakage detection;

[0140] ST-GNN identifies the abnormal node as N-67 (confidence 99%);

[0141] Prioritize key nodes, including hospitals (node 12, pressure maintained > 0.25 MPa), fire stations (node 55, 10% increase in flow);

[0142] Automatically close valves V15 / V22 / V37 to isolate the leakage area; the number of affected users decreases from 12,000 to 1,500.

[0143] Use an online learning mechanism triggered by feedback data, such as when the actual water consumption exceeds the predicted value by 10% for three consecutive times:

[0144] For example, the late peak of residential area water consumption is delayed by 30 minutes due to the occupancy of a new residential area, and the system automatically marks the data in this period as high-priority training samples;

[0145] Dynamic adjustment items include: time decay coefficient of LSTM unit (original value 0.9 → adjusted to 0.85), initial value of industrial user attention weight (0.7 → 0.65)

[0146] After model update, the prediction error of the new community decreases from 12% to 4%

[0147] Specific test cases include sudden changes in holiday water consumption patterns:

[0148] Input conditions:

[0149] Before the Spring Festival, industrial water consumption decreases by 60%, and residential water consumption increases by 45%

[0150] Model response:

[0151] The gate attention automatically reduces the industrial area weight to 0.1;

[0152] The spatio-temporal convolution detects an increase in spatial correlation of residential area water consumption (inter-node influence coefficient +20%);

[0153] Output the corrected prediction: the commercial area peak is moved to 16:00 (error <3%);

[0154] Multiple pipe leaks:

[0155] Simulate the artificial triggering of nodes N-12 / N-45 to zero pressure;

[0156] The system generates two levels of emergency plans within 45 seconds:

[0157] First isolation: close V8 / V19;

[0158] Secondary water supply: start backup water source S3;

[0159] The key node pressure fluctuation is controlled within ±0.03 MPa.

[0160] Through the multi-level feature extraction and dynamic adjustment capability of the ST-GNN, high-precision water demand prediction and rapid emergency response are realized. The system has been stably running in B City Water Group for 12 months, and has cumulatively reduced the leakage water volume by 187,000 tons / year.

[0161] Further, the S3 further comprises:

[0162] The improved multi-objective particle swarm optimization algorithm MOPSO comprises:

[0163] Dynamic inertia weight: the weight value is adaptively adjusted according to the supply water pressure variance;

[0164] Constraint processing mechanism: the pump station start-stop number limit is converted into a penalty function;

[0165] Pareto solution screening: the optimal scheduling scheme is selected based on a fuzzy membership function.

[0166] In this embodiment, the multi-objective function defines the target as including: water supply stability, energy consumption minimization, and leakage suppression;

[0167] When the target is water supply stability, the deviation of each node pressure from the demand pressure is minimized; when the target is energy consumption minimization, the total power consumption of the pump station needs to be reduced (α is the electricity price coefficient); when the target is leakage suppression, the leakage of the most serious leakage pipeline needs to be controlled;

[0168] The decision variables include: pump station frequency (30-50 Hz), valve opening (0%-100%), and water tank scheduling water level (3-8 m);

[0169] The physical constraints include: pump station daily start-stop ≤5 times, pipe network node pressure ≥0.2 MPa, and water flow velocity ≤2.5 m / s;

[0170] The improved MOPSO algorithm is implemented with dynamic inertia weight adjustment. The adjustment rule is that the reference weight w=0.7, when the pressure variance σ²>0.05 MPa²: w=0.9 (enhance global search), when σ²<0.01 MPa²: w=0.4 (enhance local optimization);

[0171] During a certain peak period, the pressure variance reaches 0.08 MPa², and the algorithm automatically increases the weight to 0.88 to quickly find a new solution;

[0172] For the constraint processing mechanism adopted, the pump station start-stop penalty function is:

[0173] Penalty = 10^6 x max(0, N_{start-stop} - 5)^2;

[0174] 5 start-stop 7 times → penalty = 4 x 10 6 ( direct elimination of the particle ) ;

[0175] For the screening of Pareto solutions, different membership function types are used according to different objectives. When the objective is water supply stability, the type of membership function used is trapezoidal function; when the objective is energy minimization, the type of membership function used is exponential function; when the objective is leakage suppression, the type of membership function used is triangular function.

[0176] The scheduling scheme generation process includes an initialization phase. For particle coding, each particle contains: 3 pump frequencies (Hz), 8 valve openings (%) and 2 tank target water levels (m) ;

[0177] The population is set to 100 particles, with a maximum of 50 iterations;

[0178] The optimization process includes prediction result correction. The predicted value of the residential area received from the ST-GNN output (+25% demand) is adjusted by MOPSO to supply water to the industrial area (reduced by 8% to balance the total amount)

[0179] Dynamic rewards and penalties include a penalty value += 5 x 10 5 , when a scheme causes the pressure of a key hospital to be <0.22 MPa, while eliminating the particle;

[0180] When non-dominated sorting is performed, 32 Pareto solutions are obtained in the 20th generation, and 15 solutions that violate the flow rate constraint are eliminated;

[0181] The final scheme parameters are shown in Table 2.

[0182] Table 2 Final scheme parameter table Equipment Set value Constraint verification Pump station P3 42Hz (original 38Hz) Daily start-stop = 4 times Valve V7 65% opening Flow rate = 2.3m / s Water tank W2 6.8m water level Pressure = 0.28MPa

[0183] In a typical test case, summer water peak + pipeline aging leakage

[0184] Input conditions: residential area demand peak + 30%, DN400 pipeline leakage detected 3.5 m³ / h

[0185] Dynamic weight in MOPSO optimization process, pressure variance 0.12 → w = 0.92

[0186] When processing constraints, eliminate all schemes with start-stop > 5 times (penalty value > 10 6 ), and keep solutions with pressure ≥0.23 MPa;

[0187] Final selected solution: energy consumption increased by 8%, but leakage decreased by 12%;

[0188] After the implementation of this embodiment in the water system of City C, the average annual energy saving reached 150,000 kWh, and the leakage rate decreased from 18.7% to 14.2%, verifying the engineering practicability of the improved MOPSO algorithm.

[0189] Further, the S4 further comprises:

[0190] Robustness of the water resource scheduling scheme under extreme weather in the water digital twin system;

[0191] Visualize the abnormal area of pipe network pressure through the AR interface, and recommend manual intervention strategies;

[0192] Introduce time-of-use electricity price strategy in the water resource scheduling scheme, and preferentially start high-energy consumption pump sets during valley electricity period;

[0193] Based on the water quality prediction results, dynamically adjust the water treatment agent dosage.

[0194] In this embodiment, the virtual deduction process takes typhoon weather as an example and loads extreme scenarios;

[0195] Input parameters:

[0196] Wind speed 25 m / s (10 level wind);

[0197] Rainfall 60 mm / h for 6 hours;

[0198] 3 tree collapse risk areas (coordinates X:125, Y:348, etc.);

[0199] Robustness test includes pressure anomaly detection:

[0200] Simulate tree smashing leading to pipe P-09 rupture, and the digital twin system detects it within 12 seconds;

[0201] Node N-45 pressure drops from 0.32 MPa to 0.08 MPa;

[0202] Downstream flow anomaly increases by 220%;

[0203] Dynamic adjustment strategies include:

[0204] Automatic response to close valves V12 / V15 / V21;

[0205] Start standby water source S2 (replenishment rate 500 m³ / h);

[0206] Recommend sending a maintenance team to GPS coordinates (31.2304°N, 121.4737°E);

[0207] The inference result output is shown in Table 3;

[0208] Table 3 Inference result output table Index Original scheme Adjusted scheme Improvement effect Number of affected users 8,200 households 1,500 households 81.7%↓ Recovery time 4.5 hours 2.2 hours 51.1%↓ Risk of water quality exceeding standards 2 communities 0 Completely avoided

[0209] An AR assisted decision application is adopted to visualize pressure anomalies;

[0210] The display elements include a pipe network three-dimensional model and real scene superposition, pressure gradient coloring (blue 0.3MPa→red <0.15MPa) and abnormal point pulse warning (frequency 2Hz);

[0211] According to the artificial intervention guide, the inspector gazes at the abnormal area for 3 seconds to trigger the menu, and the display selectable operations include immediately closing the nearest valve (distance prompt: V15 needs to walk 23m), calling for reinforcement (automatically sending location + on-site video), and selecting to generate an electronic work order (including operation two-dimensional code);

[0212] For time-of-use electricity price strategy, including pump group scheduling optimization, as shown in Table 4;

[0213] Table 4 Pump group scheduling optimization table Time period Electricity price Original pump set operation Optimized operation 08:00-11:00 Peak electricity 1.2 yuan 3 x 55kW pumps fully open 2 running, 1 standby 13:00-15:00 Flat electricity 0.7 yuan 2 pumps intermittent operation Maintain 2, reduce frequency by 5% 22:00-06:00 Valley electricity 0.3 yuan 1 pump operation 3 pumps full load water replenishment

[0214] For actual effect, the chlorine residual rate of pipe network tail is improved from 89% to 97%, and the annual sodium hypochlorite consumption is reduced by 12.5 tons.

[0215] After the embodiment is deployed in the water bureau of D city, the influence range of pipe explosion is successfully reduced by 67% during the typhoon, verifying the reliability of the digital twin system under extreme conditions.

[0216] According to the second embodiment of the present application, referring to Figure 2 The present application claims a resource scheduling system based on water data, comprising:

[0217] One or more processors;

[0218] A memory having one or more programs stored thereon, when the one or more programs are executed by the one or more processors, so that the one or more processors implement the resource scheduling method based on water data.

[0219] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the division of the units is only a logical function division, and there can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, or the among different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0220] In addition, each function unit in the various embodiments of the present application can be integrated into a processing unit, or each unit can exist alone physically, or two or more units can be integrated into one unit. The integrated unit can be implemented in the form of hardware, or in the form of a software functional unit. The above is only an embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation using the content of the present application specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

[0221] The specific embodiments of the application are described in detail above, but they are only examples. The present application is not limited to the specific embodiments described above. Any equivalent modification or substitution made by those skilled in the art to the present application is also within the scope of the present application. Therefore, any equivalent transformation and modification, improvement, etc. made without departing from the spirit and principle range of the present application should be included in the scope of the present application.

Claims

1. A resource scheduling method based on water data, characterized in that, The method comprises the following steps: S1, collecting multi-modal water data, and performing data preprocessing on the multi-modal water data through an edge computing node; S2, inputting the preprocessed multi-modal water data into a water demand prediction model, and outputting a partitioned water demand prediction result; S3, setting an optimization target, and correcting the partitioned water demand prediction result to obtain a water resource scheduling scheme by using a multi-objective particle swarm optimization algorithm; S4, virtually simulating the water resource scheduling scheme, and dynamically adjusting the water resource scheduling scheme according to a simulation result; S5, monitoring an execution effect of the adjusted water resource scheduling scheme in real time, and feeding back to the water demand prediction model for online learning.

2. The resource scheduling method based on water data according to claim 1, characterized in that, The S1 further comprises the following steps: real-time acquisition of pressure data of a water supply network, user-end intelligent water meter water consumption data, water source water quality monitoring data, meteorological environment data and industrial user water plan data; data preprocessing through the edge computing node, including noise filtering, space-time alignment and missing value interpolation; The S2 further comprises the following steps: inputting the preprocessed multi-modal water data into a water demand prediction model based on a space-time graph neural network (ST-GNN) model; dynamically adjusting feature weights of different regions and time periods in combination with an attention mechanism, and outputting a partitioned water demand prediction result for 24 hours in the future; The S3 further comprises the following steps: taking water supply stability, energy consumption minimization and pipe network leakage suppression as optimization targets, correcting the partitioned water demand prediction result to generate a water resource scheduling scheme by using an improved multi-objective particle swarm optimization (MOPSO) algorithm, and introducing water equipment physical constraints and a dynamic reward and punishment mechanism into the multi-objective particle swarm optimization algorithm; The S4 further comprises the following steps: synchronizing the water resource scheduling scheme to a water digital twin system for virtual simulation, and issuing the water resource scheduling scheme after dynamic adjustment according to a simulation result to an execution terminal.

3. The resource scheduling method based on water data according to claim 2, characterized in that, The S1 further comprises the following steps: The edge computing node adopts an adaptive sampling strategy, and a sampling frequency of the pressure data is positively correlated with a pipe network pressure fluctuation rate; differential privacy protection is implemented on the user water consumption data, and noise conforming to a Laplace distribution is added; blockchain storage is used to ensure the non-tamperability of the water source water quality data, and storage information includes acquisition time, position and sensor ID.

4. The resource scheduling method based on water data according to claim 2, characterized in that, The S2 further comprises the following steps: The space-time graph neural network (ST-GNN) model comprises the following steps: a space-time convolution module for extracting pipe network topological structure features; a gated attention module for dynamically weighting differences in water consumption modes of industrial users and residential users; an online learning module for updating model parameters in real time according to scheduling feedback data; when a pipe network leakage rate exceeds a threshold value, a hierarchical emergency scheduling is triggered to preferentially guarantee water supply of key nodes; a valve closing scheme is automatically generated to isolate a leakage area, and an optimal water supplement path is calculated in combination with a hydraulic model.

5. The resource scheduling method based on water data according to claim 2, characterized in that, The S3 further comprises the following steps: The improved multi-objective particle swarm optimization (MOPSO) algorithm comprises the following steps: a dynamic inertia weight value is adaptively adjusted according to a water supply pressure variance; a constraint processing mechanism is used to convert pump station start-stop times into a penalty function; Pareto solution screening is based on a fuzzy membership function to select an optimal scheduling scheme.

6. The resource scheduling method based on water data according to claim 2, characterized in that, The S4 further comprises the following steps: Robustness of the water resource scheduling scheme under extreme weather is simulated in the water digital twin system; AR interface can visualize abnormal pressure areas in the pipe network and recommend manual intervention strategies; Time-of-use electricity pricing strategy is introduced in the water resource scheduling scheme to preferentially start high-energy-consumption pump sets during off-peak electricity periods; Water treatment chemical dosage is dynamically adjusted based on water quality prediction results.

7. A resource scheduling system based on water data, characterized in that, The method comprises the following steps: one or more processors; a memory having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement a water data-based resource scheduling method according to any one of claims 1 to 6.

Images