A city water affair situation awareness management method and system

Abstract

The application provides a kind of urban water situation awareness management method and system, it is related to urban water management technical field, the method comprises: obtaining multi-source heterogeneous data;Multi-source heterogeneous data is preprocessed, feature extraction and fusion are obtained comprehensive analysis feature set;Comprehensive analysis feature set is input into situation assessment model to determine the situation level of urban water;Comprehensive analysis feature set is input into time series prediction model, and the situation change trend of future period is predicted;Based on comprehensive analysis feature set, situation level and situation change trend, utilize the preset risk assessment index system to carry out comprehensive evaluation, generate risk assessment result;Based on situation level, situation change trend and risk assessment result, call the preset expert knowledge base to carry out matching and reasoning, generate candidate disposal strategy set;The strategy in candidate disposal strategy set is evaluated, and the optimal disposal strategy is output.The application improves the management level of urban water.

Description

Technical Field

[0001] This application relates to the field of urban water management technology, and in particular to an urban water situation awareness management method and system. Background Technology

[0002] With the acceleration of urbanization, urban water management is becoming increasingly complex, and its stable operation is crucial to residents' lives, industrial production, and the ecological environment. The diverse sources and varying formats of urban water monitoring data make it difficult to effectively assess the water situation and manage urban water resources. Furthermore, the lack of accurate forecasting of water situation trends and advance risk assessment often leads to remedial measures being taken only after incidents occur, resulting in resource waste and increased losses.

[0003] Therefore, there is an urgent need for a method and system for urban water situation awareness management to improve the level of refined management and emergency response capabilities of urban water affairs. Summary of the Invention

[0004] To address the aforementioned technical issues, this application provides a method and system for urban water affairs situational awareness management.

[0005] A first aspect of this application provides a method for urban water situation awareness and management, including:

[0006] Acquire multi-source heterogeneous data, including water sensor monitoring data, meteorological data, geographic information data, and equipment information;

[0007] Multi-source heterogeneous data is preprocessed to obtain preprocessed data, and features are extracted and fused from the preprocessed data to obtain a comprehensive analysis feature set;

[0008] The comprehensive analysis feature set is input into a situation assessment model trained based on historical water affairs data to determine the situation level of urban water affairs;

[0009] By inputting the comprehensive analysis feature set into the time series prediction model, the trend of future situation changes can be predicted.

[0010] Based on the comprehensive analysis of feature sets, situation levels and situation change trends, a comprehensive assessment is conducted using a pre-set risk assessment indicator system to generate risk assessment results.

[0011] Based on the situation level, the trend of situation change and the risk assessment results, a set of candidate disposal strategies is generated by calling a pre-set expert knowledge base for matching and reasoning.

[0012] The strategies in the candidate disposal strategy set are evaluated, and the optimal disposal strategy is output.

[0013] A second aspect of this application provides an urban water situation awareness and management system, comprising:

[0014] The data acquisition module is used to acquire multi-source heterogeneous data, including water sensor monitoring data, meteorological data, geographic information data, and equipment information.

[0015] The data processing module is used to preprocess multi-source heterogeneous data to obtain preprocessed data, and to extract and fuse features from the preprocessed data to obtain a comprehensive analysis feature set.

[0016] The situation assessment module is used to input the comprehensive analysis feature set into the situation assessment model trained based on historical water affairs data to determine the situation level of urban water affairs;

[0017] The trend prediction module is used to input the comprehensive analysis feature set into the time series prediction model to predict the trend of future situation changes.

[0018] The risk assessment module is used to conduct a comprehensive assessment based on a set of integrated analysis features, situation level and situation change trend, and to generate risk assessment results using a preset risk assessment indicator system.

[0019] The candidate strategy module is used to generate a set of candidate disposal strategies by calling a preset expert knowledge base for matching and reasoning based on the situation level, situation change trend and risk assessment results.

[0020] The strategy selection module is used to evaluate the strategies in the candidate disposal strategy set and output the optimal disposal strategy.

[0021] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the above-described urban water situation awareness and management method.

[0022] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described urban water situation awareness and management method.

[0023] The beneficial effects of the urban water situation awareness management method and system provided in this application are as follows: Firstly, by integrating and preprocessing heterogeneous data from multiple sources, this application breaks down data barriers and achieves unified awareness of information across the entire urban water sector, solving the problems of scattered data and one-sided analysis in traditional management. Secondly, the situation assessment model trained based on historical data can accurately determine the current situation level and, through a time-series prediction model, anticipate future trends, transforming passive response into proactive prevention and control. Furthermore, by linking a preset risk assessment indicator system with an expert knowledge base, it can quickly generate response strategies, further improving emergency response efficiency and reducing resource waste and losses caused by events such as pipeline leaks and urban flooding. This application achieves refined management of urban water affairs, enhances the stability and security of urban water affairs, and further ensures the orderly operation of residents' lives and industrial production. Attached Figure Description

[0024] Figure 1 A flowchart illustrating an embodiment of the urban water situation awareness management method provided in this application;

[0025] Figure 2 This is a structural block diagram of an urban water situation awareness management system provided in an embodiment of this application;

[0026] Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0027] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0028] To make the purpose, technical solution, and advantages of this application clearer, the following will be described in conjunction with the appendix. Figure 1-3 The following is an explanation using specific examples.

[0029] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of an urban water situation awareness and management method provided in this application. The method includes:

[0030] S101: Acquire multi-source heterogeneous data, including water sensor monitoring data, meteorological data, geographic information data, and equipment information.

[0031] In this embodiment, the water sensor monitoring data includes pressure, flow rate, and water quality of the water supply network; liquid level, flow rate, and pollutant concentration of the drainage network; influent and effluent water quality of the sewage treatment plant; various parameters during the sewage treatment process; and water production data of the water plant. Meteorological data includes rainfall, rainfall intensity, temperature, humidity, wind speed, and wind direction. Geographic information data includes the city's topography, pipeline distribution, location of water plants and sewage treatment plants, and the distribution and boundaries of rivers, lakes, and other water bodies. Equipment information includes the model, installation time, operating parameter range, maintenance records, and fault history of various equipment in the urban water system.

[0032] S102: Preprocess the multi-source heterogeneous data to obtain preprocessed data, and extract and fuse the features of the preprocessed data to obtain a comprehensive analysis feature set.

[0033] In this embodiment, the preprocessing process includes data cleaning, data transformation, data standardization, and data completion. Data cleaning removes outliers, noise, and duplicate data from sensor monitoring data, such as identifying and eliminating outliers in water sensor monitoring data using the Laida criterion. Data transformation converts geographic information data of different formats into a unified format and standardizes the time units for meteorological data. Data standardization converts monitoring data of different magnitudes to the same data range for subsequent analysis and processing. Data completion uses different methods depending on the data type to complete missing data; interpolation is used for time series data, and mode is used for categorical data, resulting in preprocessed data.

[0034] In one embodiment, multi-source heterogeneous data is preprocessed to obtain preprocessed data, including:

[0035] For outliers in water sensor monitoring data, a joint discrimination method based on dynamic threshold and context association is used for identification and cleaning; for missing time series data, a compensation method based on time series decomposition and spatiotemporal kriging interpolation is used for filling; meteorological data, geographic information data and equipment information data are uniformly mapped to the same spatiotemporal coordinate system based on the geographic information data of the water pipeline network.

[0036] In this embodiment, feature extraction employs appropriate methods for different types of data. Pressure fluctuation features and flow change features are extracted from water sensor monitoring data, while rainfall intensity features and temperature change trend features are extracted from meteorological data. Feature fusion integrates features from different sources and of different types. A weighted fusion approach can be used, assigning appropriate weights based on the importance of different features to the water situation assessment, resulting in a comprehensive analytical feature set. For example, pressure features of the pipe network can be combined with corresponding geographic information features, and meteorological features can be correlated with flow features of the drainage pipe network to form a comprehensive analytical feature set that can characterize the urban water situation.

[0037] In one embodiment, feature extraction and fusion of preprocessed data are performed to obtain a comprehensive analysis feature set, including: for preprocessed time-series data, local morphological features are extracted using a one-dimensional convolutional neural network, and long-period dependent features are extracted using a bidirectional long short-term memory network, and the two are fused to form a time-series deep feature vector; the time-series deep feature vector is concatenated with encoded non-time-series features to form a primary fusion feature; the primary fusion feature is input into a feature optimization layer based on an attention mechanism, which calculates the weights of different features in a specific water affairs scenario to generate a weighted comprehensive analysis feature set.

[0038] S103: Input the comprehensive analysis feature set into the situation assessment model trained based on historical water affairs data to determine the situation level of urban water affairs.

[0039] In this embodiment, the situation assessment model can employ machine learning algorithms, such as support vector machines and random forests. It is trained using feature data and event levels corresponding to various historical water-related events, such as pipe bursts, network blockages, and water quality exceeding standards. This allows the situation assessment model to accurately determine the current situation level of the city's water affairs based on the input comprehensive analysis feature set. The situation level can be categorized into four levels: normal, slightly abnormal, moderately abnormal, and severely abnormal.

[0040] For example, the feature dimensions of the situation assessment model include pipeline operation characteristics, water quality characteristics, equipment status characteristics, and environmental correlation characteristics, totaling 23 dimensions. If a random forest model is used, the hyperparameter settings include: 12 decision trees, a maximum tree depth of 15 layers, a minimum number of sample splits of 8, and a minimum number of leaf nodes of 3. If a support vector machine model is used, the hyperparameter settings include: a radial basis function kernel with a penalty coefficient of 2.0 and a kernel parameter γ of 0.1. The training sample size uses historical urban water affairs event data from the past 5 years, containing 12,000 valid samples, including four scenarios: normal, slightly abnormal, moderately abnormal, and severely abnormal, with each scenario accounting for 45%, 30%, 18%, and 7% respectively. Abnormal samples are balanced using the SMOTE algorithm. A 5-fold cross-validation is used, randomly dividing the samples into 5 groups. Each time, 4 groups are used as the training set and 1 group as the validation set, and the average accuracy of the model is calculated after 5 iterations to ensure the model's generalization ability. The final model validation accuracy is no less than 92%.

[0041] The situation level quantification thresholds include: Normal: The situation assessment score output by the situation assessment model is in the range of [0, 30), corresponding to pipeline pressure fluctuation ≤ 5%, flow deviation ≤ 8%, all water quality indicators meet the standards, and equipment operates without faults. Slightly abnormal: The score is in the range of [30, 50), with a single indicator slightly exceeding the standard; for example, pressure fluctuation of 6%-10%, equipment experiencing one non-critical fault warning, with no significant risk to residents' water use. Moderately abnormal: The score is in the range of [50, 75), with 2-3 indicators exceeding the standard (for example, flow deviation of 11%-20% and a certain water quality parameter exceeding the standard by 10%-20%), equipment experiencing one critical fault or more than two non-critical faults, which will affect the local area. Severely abnormal: The score is in the range of [75, 100], with 3 or more indicators severely exceeding the standard (such as pressure fluctuation > 20%, pipeline blockage exceeding the standard), equipment experiencing two or more critical faults, causing major regional safety hazards.

[0042] S104: Input the comprehensive analysis feature set into the time series prediction model to predict the trend of future situation changes.

[0043] In this embodiment, the time-series prediction model can be an ARIMA model or an LSTM neural network, etc. The selected model is trained using preprocessed historical data, and then input into a comprehensive analysis feature set to predict changes in urban water affairs over the next few hours, days, or even weeks, thereby obtaining the trend of future changes. Examples include trends in pipeline pressure and changes in the influent water quality of sewage treatment plants.

[0044] For example, an LSTM neural network is used; the network has 3 layers, the input layer dimension is consistent with the dimension of the comprehensive analysis feature set (23 dimensions), the first LSTM layer has 64 units, the second LSTM layer has 48 units, the third LSTM layer has 32 units, the output layer is a fully connected layer, and the output dimension is the situation prediction value for the next 0.5-3 days.

[0045] The time step is set to 30, which means that the trend of the situation changes in the next 0.5-3 days is predicted by using the time series data of the past 30 days.

[0046] The training parameters include the loss function and the training period; the loss function is the root mean square error loss function. The training period is set to 100 epochs, using the adaptive moment estimation (Adam) optimizer with an initial learning rate of 0.001. The learning rate is halved when the validation set loss does not decrease for 10 consecutive epochs, and training stops if it does not decrease for 15 consecutive epochs.

[0047] Regularization techniques include adding a Dropout layer after each LSTM unit, with the dropout probability set to 0.2, to prevent the model from overfitting.

[0048] The time series forecasting model was validated using time series data from the past three years as the test set, predicting the trend of the next 0.5 days, 1 day, and 3 days, with the mean absolute error controlled within 3%, 5%, and 8%, respectively.

[0049] S105: Based on the comprehensive analysis of feature sets, situation levels, and situation change trends, a comprehensive assessment is conducted using a preset risk assessment indicator system to generate risk assessment results.

[0050] In this embodiment, the risk assessment indicator system includes indicators such as the aging degree of the pipe network, the risk of water quality exceeding standards, and the matching degree between drainage capacity and rainfall. This embodiment assesses the magnitude and probability of various risks faced by urban water affairs by comparing and calculating relevant features, situation levels, and future situation change trends in the comprehensive analysis feature set with these indicators, generating risk assessment results. The risk assessment results include risk levels, which can be divided into high risk, medium risk, low risk, etc.

[0051] S106: Based on the situation level, situation change trend and risk assessment results, call the preset expert knowledge base for matching and reasoning to generate a set of candidate disposal strategies.

[0052] In this embodiment, the expert knowledge base stores a large number of response strategies formulated based on historical experience and expert judgment. For example, when the situation level is warning and the predicted future rainfall is large, the corresponding response strategies include increasing the frequency of pipeline inspections and starting drainage pumps in advance. When the operational situation level is slightly abnormal and the risk level is medium risk, the corresponding response strategies include strengthening equipment inspections and adjusting water supply pressure. The reasoning process in this embodiment adopts a rule-based reasoning method, matching the current operational situation level, change trend, and risk assessment results with the rules in the knowledge base to select multiple candidate response strategies suitable for the current situation.

[0053] S107: Evaluate the strategies in the candidate disposal strategy set and output the optimal disposal strategy.

[0054] In this embodiment, the evaluation can be conducted from multiple dimensions, such as the implementation cost, implementation effect, and implementation difficulty of the strategy. For example, it can compare the human and material costs required by different candidate strategies, predict the degree of improvement to the urban water situation after implementation, and assess the feasibility of the strategy under existing conditions. Through comprehensive scoring or multi-objective decision-making methods, the optimal treatment strategy is selected from the set of candidate strategies to guide the water department in effective management and decision-making. Simultaneously, when outputting the optimal treatment strategy, the implementation steps, required resources, and expected effects are included.

[0055] As can be seen from the above, this application integrates scattered information such as water sensor monitoring data, meteorological data, and geographic information data into a unified comprehensive analysis feature set through data cleaning, standardization, and spatiotemporal alignment, thus solving the problem of one-sided risk assessment caused by data silos in traditional methods. Furthermore, the linkage between the situation assessment model and the time-series prediction model not only determines the current situation level based on historical data but also captures future situation change trends in advance through time-series prediction. The outputs of both provide dynamic input for risk assessment. In addition, the situation level characterizes the risk basis of the current water system state, while trend prediction supplements the evolution direction of potential future risks, avoiding the limitations of existing methods that only focus on the current state and ignore the dynamic development of risks. In the risk quantification and strategy optimization stages, the pre-set risk assessment indicator system and the water digital twin work synergistically. The risk assessment indicator system, based on the comprehensive analysis feature set, situation level, and trend prediction results, constructs a quantitative model from multiple dimensions including water quality, water quantity, equipment, and environment. The water digital twin, through a dual-engine architecture of physical model and data-driven model, transforms abstract risk indicators into a simulateable dynamic process. For example, to address the uncertainty of weather forecasting, a digital twin combined with Monte Carlo simulation is used to conduct multiple sampling simulations of pipeline pressure and flow changes under different rainfall scenarios. This outputs the probability distribution and confidence intervals of various risk indicators, solving the core problem that existing methods cannot quantify uncertainty risks. Simultaneously, based on the collaboration of a multi-objective optimization algorithm and an expert knowledge base, risk assessment results are transformed into candidate response strategies. The expert knowledge base provides an empirical strategy framework, while the multi-objective optimization algorithm, based on economic efficiency, effectiveness, reliability, and decision risk indicators (confidence interval width), selects the Pareto optimal solution set. Finally, the optimal strategy is determined through fuzzy clustering and historical case matching, achieving a closed-loop process from risk quantification to strategy generation to effect verification. This embodiment enhances the urban water system's ability to manage uncertainty risks. On the one hand, by integrating multi-source data and linking dynamic models, the vague risk perception in traditional methods is transformed into precise risk quantification. For example, the assessment of pipeline leakage risk no longer relies on single pressure sensor data, but is based on multi-dimensional information such as historical leakage cases, meteorological precipitation forecasts, and pipeline aging, outputting quantitative results of leakage probability, impact range, and response window. On the other hand, through simulation and multi-objective optimization using a water affairs digital twin, proactive prevention and control of uncertain risks is achieved. For example, before a rainstorm, the risk of urban flooding under different rainfall intensities can be predicted in advance, generating combined strategies for pipeline inspection, pump station scheduling, and emergency drainage. The effectiveness of these strategies under extreme scenarios is verified through simulation, avoiding the drawbacks of existing methods that only provide passive remediation after an accident occurs. Actual testing shows that the method in this application improves the prediction accuracy of water system risks to over 92%, increases the management efficiency of uncertain risks by 50%, and significantly reduces economic losses and resource waste caused by pipeline leaks, urban flooding, and other events.

[0056] In one embodiment of this application, the strategies in the candidate disposal strategy set are evaluated, and the optimal disposal strategy is output, including:

[0057] Each strategy in the candidate disposal strategy set is input into a preset water affairs digital twin for simulation and deduction, and multi-dimensional simulation results corresponding to each strategy are obtained.

[0058] Based on multi-dimensional simulation results, a multi-objective evaluation function is constructed with economic indicators, efficiency indicators, and reliability indicators as objectives.

[0059] Based on the multi-objective evaluation function, a multi-objective optimization algorithm is used to optimize and evaluate the deduced candidate disposal strategy set to obtain the Pareto optimal solution set;

[0060] The final optimal solution strategy is selected from the Pareto optimal solution set based on preset decision rules.

[0061] In this embodiment, the water digital twin simulates the implementation process of candidate strategies through physical models and data-driven models. Its core is to construct a digital mirror that maps in real time to the actual water system. Through multi-dimensional modeling, dynamic parameter coupling, and simulation, it reproduces the entire process of pipeline flow, equipment operation, and system response after strategy implementation. The multi-dimensional modeling includes physical modeling to obtain a physical model and a data-driven model based on deep learning.

[0062] Specifically, a pipeline network flow dynamics model is constructed based on computational fluid dynamics to simulate the dynamic behavior of pipeline network flow, reconstructing the pressure propagation, flow distribution, hydraulic losses, and dynamic response after strategic intervention within the pipeline network. Specifically, actual design drawings, as-built data, and inspection data of the urban waterworks network are imported into a GIS system to construct a 1:1 scale three-dimensional geometric model of the pipeline network, including parameters such as pipe diameter, length, direction, pipe material type, and the location and specifications of pipe fittings (valves, tees, elbows), ensuring complete consistency with the topology of the actual pipeline network. Different hydraulic parameters are assigned to different pipe materials; for example, the inner wall roughness coefficient of cast iron pipes is set to 0.012-0.015, and that of PE pipes to 0.009-0.011. Simultaneously, a corrosion attenuation model of the pipeline network materials is integrated, with the annual corrosion rate of cast iron pipes calculated based on 0.12 mm / year, updating the changes in pipe inner diameter and roughness with service life to simulate the impact of pipeline aging on water flow.

[0063] The Navier-Stokes (NS) equations for three-dimensional incompressible fluids are used as the governing equations, and the momentum and mass conservation of water flow within the pipe network are described based on the continuity equation. The governing equations are spatially discretized using the finite volume method, dividing the pipe network into a structured grid with a resolution of 0.5 m and a time step of 0.1 s. The pressure-velocity coupling problem is solved using the SIMPLEC algorithm, iteratively calculating parameters such as pressure, velocity, and flow rate for each grid to simulate the real-time propagation of water flow within the pipe network.

[0064] For equipment such as pumps and valves, physical equation models that match actual characteristics are constructed. For example, the characteristic curve equation for a pump is: H = H0 - kQ², where H is the head, Q is the flow rate, H0 is the rated head, and k is the resistance coefficient. Rated parameters are obtained from equipment records and calibrated. For valves, a flow regulation model is established based on the correspondence between opening degree and flow coefficient. For example, when the valve opening degree increases non-linearly from 0% to 100%, the flow coefficient increases non-linearly, simulating the throttling effect of valve opening and closing on water flow.

[0065] This embodiment leverages a data-driven layer based on deep learning to improve dynamic prediction accuracy. Specifically, the data-driven layer trains the prediction model using historical operational data, compensating for the impact of parameter uncertainties in physical modeling and achieving a closed loop between physical simulation and data correction. For example, historical water events from the past 12 months can be collected to train the deep learning model and predict the system behavior of the water system under different strategies. A hybrid network structure of a 3-layer LSTM and a 2-layer GRU is used to construct the deep learning model. The input layer dimension is consistent with the number of features. The LSTM layer is responsible for capturing long-term dependencies, while the GRU layer focuses on short-term dynamic responses (e.g., flow fluctuations after sudden rainfall). The number of hidden layer units is set to 64, 48, and 32 respectively, and overfitting is prevented using Dropout (with a probability of 0.2). The deep learning model is trained using root mean square error as the loss function, with Adam as the optimizer and an initial learning rate of 0.001. The learning rate is halved when the validation set loss does not decrease for 10 consecutive rounds. The training sample size is 2000 sets, and the average prediction error on the test set is controlled to ≤3% to ensure accurate prediction of pipeline network operating parameters for the next 24 hours.

[0066] This embodiment can also compare the simulation results of the physical model with the actual sensor monitoring values ​​every 5 minutes. If the deviation is greater than 5%, the parameters of the physical model, such as the pipe roughness coefficient, are adjusted by using the predicted residual output by the data-driven model (i.e., the difference between the monitored value and the calculated value). For example, when the actual pressure in a certain area is lower than the simulation value, the leakage coefficient of the pipeline network in that area is increased to simulate the impact of potential leakage on the pressure, thus achieving coordinated correction between the two models.

[0067] In this embodiment, after a strategy is input, the water management digital twin simulates the entire process of strategy implementation through a four-step process: strategy parsing, parameter mapping, dynamic simulation, and result output. Strategy parsing and parameter mapping transform the strategy into parameters that the model can recognize. Specifically, the strategy is broken down into quantifiable operational instructions, and these instructions are then converted into input parameters for the physical and data models.

[0068] For example, replacing an aging valve can be broken down into: valve closure time of 45 minutes - old valve removal - new valve installation - new valve opening, opening degree from 0% to 100%, taking 10 minutes. Increasing the inspection frequency can be broken down into: shortening the inspection cycle from 7 days to 3 days, with each inspection covering 5 nodes, and reducing the response time for repairing leaks from 24 hours to 8 hours. The transformation of the aging valve replacement process is as follows: locate the target valve in the physical model, set the flow coefficient for the 45-minute valve closure period to 0, and after installation, linearly increase the flow coefficient from 0 to the valve's rated value during the opening phase. The transformation of the inspection frequency increase process is as follows: adjust the leak repair time parameter, shortening the leak duration from 7 days (before inspection) to 3 days, and simultaneously update the leak calculation model (leakage is positively correlated with pipeline pressure and leak orifice diameter), simulating the effect of increased inspection frequency on reducing the leakage rate.

[0069] In this embodiment, simulations are performed based on time steps to calculate the pressure and flow changes of each node in the pipeline network in real time after the strategy is implemented.

[0070] For example, the valve replacement process is as follows: Valve closing phase 0-45 minutes: The pressure at the upstream node of the target valve gradually increases from 0.4MPa to 0.55MPa, while the pressure at the downstream node drops sharply from 0.4MPa to 0.1MPa, and the flow rate approaches 0; at the same time, the impact range of insufficient water pressure in this area is predicted by a data-driven model; New valve opening phase 45-55 minutes: The pressure at the downstream node gradually recovers from 0.1MPa to 0.38MPa, and the flow rate recovers to 95% of the original level. The physical model simultaneously calculates the hydraulic losses during the water flow redistribution process.

[0071] After the frequency of inspections was increased: the data model, based on historical leak repair data, predicted that the leakage rate would decrease from 8% to 5%, while the physical model calculated the increase in overall pipeline pressure after the leakage rate was reduced.

[0072] In this embodiment, the pipeline network status after the strategy implementation is displayed in real time through a 3D visualization platform, including: pressure distribution cloud map, flow rate change curve, equipment status panel, quantitative indicator output, performance indicators, and reliability indicators; among which,

[0073] The flow change curve includes: based on nodes (upstream and downstream of the valve), outputting a 24-hour flow time series curve, marking the key time points of flow drop (valve closed) and recovery (valve opened);

[0074] The equipment status panel displays operating parameters such as valve opening and pump head, providing a clear overview of the strategy's impact on the equipment. Quantitative indicator output includes calculations of core performance metrics for the strategy implementation, generating a simulation report.

[0075] In this embodiment, the water affairs digital twin is a precise digital replica of urban water affairs, including the physical structure of the pipe network, the operating characteristics of equipment, and the dynamic laws of water flow. After inputting candidate strategies, the water affairs digital twin will simulate the entire process after the strategy is implemented. For example, when the strategy is to increase the frequency of pipe network inspections in a certain area and replace aging valves, the simulation will show the pressure change curve of the pipe network in that area, the decrease in leakage rate, and the duration of water flow interruption during valve replacement.

[0076] In one embodiment, the economic index is the weighted sum of the resource costs and indirect economic losses incurred in implementing the strategy; the effectiveness index is a comprehensive evaluation of the degree and speed of improvement of water parameters at key nodes that deviate from normal thresholds after the strategy is implemented; the reliability index is the robustness assessment score of the strategy in maintaining expected effectiveness under future weather scenarios with different confidence levels; based on the multi-objective evaluation function, a fast non-dominated sorting genetic algorithm with an elite strategy is used to optimize the deduced candidate disposal strategy set to obtain the Pareto optimal solution set.

[0077] The economic indicators of this embodiment include the total investment amount for strategy implementation, the savings in operation and maintenance costs per unit time, and the investment payback period. For example, the total investment amount covers equipment procurement costs, construction costs, and technical service costs. The efficiency indicators include the percentage increase in pipeline water conveyance capacity, the increase in the speed of sewage treatment to meet discharge standards, and the stability of water supply pressure. For example, the stability of water supply pressure can be reflected by the reduction in the pressure fluctuation amplitude. The reliability indicators include the average interval between system failures after strategy implementation, the average time for fault repair, and the system's anti-interference capability under extreme operating conditions. For example, the average time for fault repair can be compared with the change in maintenance efficiency before and after strategy implementation.

[0078] The multi-objective evaluation function in this embodiment quantifies economic indicators, efficiency indicators, and reliability indicators. By setting different weights for each indicator, the weights can be dynamically adjusted according to the short-term goals and long-term plans of urban water management. Multiple objectives are integrated into a calculable function expression. The higher the function value, the better the strategy performs in terms of comprehensive economy, efficiency, and reliability.

[0079] The multi-objective optimization algorithm in this embodiment can be a non-dominated sorting genetic algorithm or a particle swarm optimization algorithm. This embodiment uses a non-dominated sorting genetic algorithm as an example. Each strategy in the candidate disposal strategy set is encoded, transforming it into an individual that the algorithm can process. Then, by calculating the multi-objective evaluation function value of each individual, selection, crossover, and mutation operations are performed to continuously filter out better strategies during the iteration process. The termination condition is reaching a preset maximum number of iterations of 500. After multiple iterations, those strategies that cannot be comprehensively surpassed by other strategies in terms of economy, efficiency, and reliability are retained, forming a Pareto optimal solution set. Specifically, the parameters of the non-dominated sorting genetic algorithm are set as follows: population size 200, number of iterations 500, crossover rate 0.9, and mutation rate 0.1. The Pareto front is visualized using a three-dimensional scatter plot, and the Hypervolume index is used to evaluate the quality of the solution set. If the rate of change (standard deviation) of the distribution range of the Pareto front for 20 consecutive generations is less than 5%, the iteration can be terminated early to reduce invalid computation.

[0080] The decision-making rules preset in this embodiment can be formulated according to the actual needs of urban water management. For example, when a city faces an emergency water supply guarantee task, the strategy with the highest weight in efficiency indicators is prioritized; when the fiscal budget is tight, economic indicators are used as the main selection criterion; and during special periods such as the flood season, reliability indicators become the primary consideration. Furthermore, the decision-making rules can incorporate expert experience, such as inviting senior experts in the water sector to score the strategies in the Pareto optimal solution set, and combining expert opinions and quantitative indicators to determine the final optimal disposal strategy. For example, in the Pareto optimal solution set, after evaluating the performance of each strategy under the current key objectives of urban water management, the strategy with the highest comprehensive score in efficiency and reliability indicators is selected as the final optimal disposal strategy.

[0081] In summary, this embodiment utilizes a water affairs digital twin for simulation, enabling the evaluation of strategy effectiveness without impacting the actual operation of the water affairs system, thus reducing decision-making risks. Secondly, a multi-objective evaluation function is constructed, and a multi-objective optimization algorithm is employed to screen candidate disposal strategies after simulation, taking into account the economy, effectiveness, and reliability of the strategies and avoiding the one-sidedness of single-objective evaluation. Finally, the optimal strategy is selected through a Pareto optimal solution set and decision rules, ensuring that a balanced optimal solution can be found when multiple objectives conflict, thereby improving the rationality of strategy selection.

[0082] In one embodiment of this application, the final optimal action strategy is selected from the Pareto optimal solution set based on a preset decision rule, including:

[0083] Based on the situation level and risk assessment results, adjust the weight coefficients of each indicator in the multi-objective assessment function;

[0084] Fuzzy clustering analysis is performed on the strategies in the Pareto optimal solution set to obtain the residual strategies;

[0085] Based on a satisfaction evaluation mechanism that matches historical cases, the strategy with the highest similarity to the handling effect of historical successful cases is selected from the remaining strategies as the optimal handling strategy.

[0086] In this embodiment, the situation level characterizes the urgency of the current urban water situation. If the situation level is urgent, it indicates a severe challenge, and the importance of performance and reliability indicators increases significantly. Their weighting coefficients in the multi-objective evaluation function need to be increased; for example, the weight of the performance indicator can be increased from the usual 30% to 50%, and the weight of the reliability indicator from 20% to 30%, while the weight of the economic indicator is correspondingly reduced. If the risk assessment result shows a high risk, especially involving water quality safety or major pipeline failure risks, the weight of the reliability indicator will be further increased to ensure that the selected strategy prioritizes urban water services. When the situation level is normal and the risk assessment result is low risk, the weight of the economic indicator can be appropriately increased, focusing on selecting strategies with better cost-effectiveness.

[0087] In this embodiment, fuzzy clustering analysis calculates the similarity of strategies across metrics such as economy, effectiveness, and reliability, grouping strategies with similar characteristics into the same cluster. For example, a fuzzy C-means clustering algorithm can be used, setting a reasonable number of clusters to divide the strategies in the Pareto optimal solution set into 3-5 clusters. Then, the overall performance of each cluster is evaluated by calculating the average or weighted average of the metrics for each strategy within the cluster, comparing the performance of different clusters. Clusters with significantly lower average values ​​for the target metrics than other clusters, or those showing no advantage in multiple metrics, are identified as significantly underperforming clusters and are eliminated entirely. This step effectively reduces the number of strategies, focusing on more promising candidate strategies.

[0088] In this embodiment, the historical case database stores data on the implementation of response strategies and their actual effects in different water management scenarios, including the achievement of various indicators after strategy implementation, user feedback, and recovery speed. The satisfaction evaluation mechanism in this embodiment extracts key feature parameters from historical cases; for example, case features similar to the situation level and risk assessment results; and calculates the similarity between the remaining strategies and each historical successful case in terms of feature parameters and response effects. The similarity calculation in this embodiment can use methods such as cosine similarity or Euclidean distance, for example, comparing the improvement in reliability indicators under the same risk level and the improvement in effectiveness indicators under similar operating conditions. Finally, the strategy with the highest similarity and the highest historical satisfaction evaluation is selected as the final optimal response strategy to ensure that this strategy is more likely to achieve good results in practical applications.

[0089] In this embodiment, the indicator weights are dynamically adjusted based on the situation level and risk assessment results, making the assessment more closely reflect the actual water situation and enhancing the relevance of the strategy assessment. Secondly, fuzzy clustering analysis is used to eliminate outdated strategy clusters, reducing invalid calculations and improving the efficiency of strategy optimization. The satisfaction evaluation mechanism based on historical case matching draws on past successful experiences, improving the accuracy and reliability of optimal strategy selection.

[0090] In one embodiment of this application, the weight coefficients of each indicator in the multi-objective evaluation function are adjusted based on the situation level and risk assessment results, including:

[0091] The first weight adjustment factor is obtained by querying the preset weight mapping table based on the situation level.

[0092] Based on the risk level and risk type in the risk assessment results, the second weight adjustment factor is obtained through the risk weight calculation model, which is a weight allocation model constructed based on the analytic hierarchy process and the entropy weight method.

[0093] The first and second weight adjustment factors are combined to obtain the weight coefficients of each indicator.

[0094] In this embodiment, the preset weight mapping table is pre-defined based on the operational characteristics and management experience of urban water affairs; it clearly defines the correspondence between different situation levels and the weight adjustment factors of each indicator. The situation levels in this embodiment can be divided into multiple levels such as normal, attention, warning, emergency, and severe, each level corresponding to a set of adjustment coefficients for economic, efficiency, and reliability indicators. For example, when the situation level is normal, the first weight adjustment factor for the economic indicator is 1.2, the efficiency indicator is 1.0, and the reliability indicator is 0.8, reflecting an emphasis on cost control; while when the situation level is emergency, the first weight adjustment factor for the reliability indicator increases to 1.5, the efficiency indicator is 1.3, and the economic indicator decreases to 0.7, highlighting the importance of stable system operation and rapid response. This embodiment can quickly determine the first weight adjustment factor matching the situation level by querying this mapping table.

[0095] In this embodiment, the risk level in the risk assessment results can be divided into low, medium, high, and extremely high, and the risk types include water quality risk, pipeline structure risk, equipment operation risk, and hydrological and meteorological risk. This embodiment adopts the analytic hierarchy process (AHP), constructing a hierarchical structure of risk factors. Water experts are invited to compare and score the importance of each indicator under different risk levels and risk types, forming a judgment matrix. Then, by calculating the maximum eigenvalue and eigenvector of the matrix, the weight allocation ratio of each indicator is determined, thus obtaining the subjective weight. This embodiment uses the entropy weight method, calculating the information entropy of each indicator based on historical risk data, thus obtaining the entropy weight, i.e., the objective weight. The subjective weight and the objective weight are weighted to obtain the second weight adjustment factor; the weight coefficient of the subjective weight is greater than that of the objective weight. For example, the weight coefficient of the subjective weight can be 0.6, and the weight coefficient of the objective weight can be 0.4, to reflect the dominant role of subjective experience.

[0096] For example, for high-level water quality risks, the second weighting adjustment factor for reliability indicators will be significantly increased; while for medium-level equipment operation risks, the adjustment factor for efficiency indicators will be more prominent. In this embodiment, if the entropy weighting method is used, the weights are determined by utilizing the information entropy of the risk assessment data itself. By calculating the dispersion of each indicator's data, the greater the dispersion, the smaller the information entropy, and the larger the corresponding second weighting adjustment factor, which can objectively reflect the impact of the information contained in the data on the importance of the indicators.

[0097] This embodiment fuses the first weight adjustment factor and the second weight adjustment factor to obtain the weight coefficients of each indicator. The fusion process can employ a weighted summation method, setting the fusion weights based on the importance of the situation level and risk assessment results in the current decision-making process, and then calculating the weight coefficients of each indicator. For example, when the urgency of the situation level is much higher than the risk level, the fusion weight of the first weight adjustment factor can be set to 0.6, and the fusion weight of the second weight adjustment factor can be set to 0.4; if the risk assessment results indicate extreme risk, the fusion weight of the second weight adjustment factor can be set to 0.5 or higher. The formula used in this embodiment is:

[0098] Final weight coefficient = first weight adjustment factor × fusion weight + second weight adjustment factor × (1 - fusion weight).

[0099] In this embodiment, the first weight adjustment factor based on situation level and the second weight adjustment factor based on risk assessment reflect the key points of water management from different dimensions, making the weight adjustment more comprehensive. Secondly, the analytic hierarchy process (AHP) and entropy weight method are used to construct a risk weight calculation model, making the determination of the second weight adjustment factor more scientific and objective. Finally, the two weight adjustment factors are merged, resulting in a more reasonable final weight coefficient, further improving the accuracy of multi-objective assessment.

[0100] In one embodiment of this application, the first weight adjustment factor and the second weight adjustment factor are fused to obtain the weight coefficients of each index, including:

[0101] Meteorological data is obtained from multi-source heterogeneous data as environmental data, and time-series analysis and feature extraction are performed on the meteorological data to obtain the predicted precipitation in the future preset period as seasonal features.

[0102] Seasonal characteristics are input into a preset seasonal adjustment model to generate seasonal adjustment factors. The seasonal adjustment model is a regression model trained based on historical meteorological and weighted adjustment data.

[0103] The first weight adjustment factor and the second weight adjustment factor are weighted and corrected using seasonal adjustment factors to obtain the corrected first weight adjustment factor and the corrected second weight adjustment factor.

[0104] The modified first weight adjustment factor and the modified second weight adjustment factor are combined to obtain the weight coefficient of each indicator.

[0105] In this embodiment, the meteorological data in the multi-source heterogeneous data includes historical and real-time information such as rainfall, duration of precipitation, and intensity of precipitation. This embodiment uses time-series analysis to organize the meteorological data according to a time series, and uses methods such as trend analysis and cycle analysis to uncover precipitation patterns; for example, it identifies the cyclical characteristics of precipitation during the rainy and dry seasons. Feature extraction in this embodiment focuses on precipitation indicators closely related to urban water management operations, using time-series prediction models, such as moving average models or exponential smoothing models, to predict precipitation for a predetermined period in the future. This embodiment uses predicted precipitation as a seasonal feature, which can characterize the precipitation trend over a future period; for example, predicted precipitation is higher during the rainy summer season and lower during the dry winter season.

[0106] In this embodiment, the seasonal adjustment model is a regression model trained based on historical meteorological and weight adjustment data. The training data for the seasonal adjustment model includes meteorological data from different historical periods and corresponding weight adjustment records, i.e., the actual values ​​of the first and second weight adjustment factors under different historical seasonal backgrounds.

[0107] This embodiment learns the correlation between seasonal characteristics and weight adjustment needs through regression analysis. For example, when the predicted precipitation is much higher than the historical average for the same period, it means that there will be greater drainage pressure in the future. The seasonal adjustment model will generate an adjustment factor greater than 1 to enhance the weight influence of reliability and effectiveness indicators. Conversely, when the predicted precipitation is extremely low, the model generates an adjustment factor less than 1, weakening the weight of precipitation-related indicators and relatively highlighting the importance of economic indicators.

[0108] In this embodiment, the process of weighting and correcting the first and second weighting adjustment factors using seasonal adjustment factors involves multiplying the seasonal adjustment factor by the first and second weighting adjustment factors respectively. For example, if the seasonal adjustment factor is 1.2, and the original economic indicator, efficiency indicator, and reliability indicator in the first weighting adjustment factor are 1.0, 1.0, and 0.8 respectively, then the corrected first weighting adjustment factor will have an economic indicator of 1.0 × 1.2 = 1.2, an efficiency indicator of 1.0 × 1.2 = 1.2, and a reliability indicator of 0.8 × 1.2 = 0.96. Similarly, the same operation is performed on the second weighting adjustment factor.

[0109] This embodiment enables the weight adjustment factor to adapt to future seasonal precipitation characteristics through the above-described correction process. For example, during the rainy season, the corrected weight adjustment factor for the effectiveness and reliability indicators related to drainage and flood control is strengthened, which is more in line with the actual water management needs.

[0110] In this embodiment, the weighted summation method can be used to fuse the modified first weight adjustment factor and the modified second weight adjustment factor, but it needs to be recalculated based on the modified factors. For example, if the fusion weight of the modified first weight adjustment factor is set to 0.5 and the fusion weight of the modified second weight adjustment factor is also set to 0.5, and if the economic index of the modified first weight adjustment factor is 1.2, the efficiency index is 1.2, and the reliability index is 0.96, and the economic index of the modified second weight adjustment factor is 1.1, the efficiency index is 1.3, and the reliability index is 1.0, then the final weight coefficient of the economic index is 1.2×0.5+0.1×0.5=1.15, the efficiency index is 1.2×0.5+1.3×0.5=1.25, and the reliability index is 0.96×0.5+1.0×0.5=0.98. This embodiment uses the above-mentioned fusion process to make the final weighting coefficients take into account the influence of situation level, risk assessment results and seasonal precipitation characteristics, and further enable the multi-objective evaluation function to more accurately fit the urban water management priorities under different environmental conditions.

[0111] In this embodiment, seasonal characteristics are introduced and seasonal adjustment factors are generated through a seasonal adjustment model. Based on the long-term impact of meteorological factors on water management, the weight adjustment is made more consistent with actual environmental changes. Secondly, a regression model trained on historical meteorological and weight adjustment data ensures the reliability and applicability of the seasonal adjustment factors. Finally, the first and second weight adjustment factors are weighted, corrected, and merged to ensure that the final weight coefficients take into account situation, risk, and seasonal factors, further improving the accuracy of risk assessment.

[0112] In one embodiment of this application, a seasonal adjustment factor is used to weight and correct the first weight adjustment factor and the second weight adjustment factor respectively, resulting in corrected first weight adjustment factors and corrected second weight adjustment factors, including:

[0113] When the seasonal characteristic is the first seasonal characteristic, the seasonal adjustment model uses the first preset weight allocation scheme to adjust the first weight adjustment factor and the second weight adjustment factor to obtain the corrected first weight adjustment factor.

[0114] When the seasonal characteristic is the second seasonal characteristic, the seasonal adjustment model uses the second preset weight allocation scheme to adjust the first weight adjustment factor and the second weight adjustment factor to obtain the corrected second weight adjustment factor.

[0115] The first seasonal characteristic is that the predicted precipitation is greater than the second threshold; the second seasonal characteristic is that the predicted precipitation is less than the first threshold, and the second threshold is greater than the first threshold; the first preset weight allocation scheme is to increase the weight of drainage and flood control related indicators; the second preset weight allocation scheme is to increase the weight of water conservation and water source scheduling related indicators.

[0116] In this embodiment, when the seasonal characteristic is the first seasonal characteristic, the seasonal adjustment model uses a first preset weight allocation scheme to adjust the first weight adjustment factor and the second weight adjustment factor to obtain the corrected first weight adjustment factor. The first seasonal characteristic is that the predicted precipitation is greater than a second threshold. The second threshold is determined based on factors such as historical urban precipitation data and the maximum carrying capacity of the drainage network, for example, set to 50 mm per hour. When the predicted precipitation is greater than the second threshold, it means that the city faces a significant risk of urban flooding. At this time, the first preset weight allocation scheme will prioritize increasing the weight of drainage and flood control related indicators. These drainage and flood control related indicators include the drainage speed of the pipe network, the pumping capacity of the pumping station, and the time for water to recede in low-lying areas. When correcting the first weight adjustment factor, higher seasonal adjustment factor weights are assigned to the effectiveness and reliability indicators related to these indicators. For example, the adjustment factor for the effectiveness indicator corresponding to the drainage speed of the pipe network is increased from the usual 1.0 to 1.5, and the adjustment factor for the reliability indicator corresponding to the operational stability of the pumping station is increased from 0.8 to 1.3. The second weight adjustment factor is adjusted in the same way, so that the revised first weight adjustment factor can fully reflect the urgent needs of drainage and flood control, and ensure that strategies that are conducive to enhancing drainage and flood control capabilities will receive more attention in subsequent multi-objective assessments.

[0117] In this embodiment,

[0118] When the seasonal characteristic is the second seasonal characteristic, the seasonal adjustment model uses a second preset weight allocation scheme to adjust the first and second weight adjustment factors, resulting in a corrected second weight adjustment factor. The second seasonal characteristic is defined as predicted precipitation being less than a first threshold, and the second threshold being greater than the first threshold. The first threshold can be set to 10 millimeters per hour, representing a situation of low precipitation. In this case, the city faces the risk of water shortage, and the second preset weight allocation scheme will emphasize increasing the weights of water conservation and water resource allocation related indicators. Water conservation related indicators include industrial water reuse rate, the penetration rate of water-saving appliances in residential areas, and water supply network leakage rate; water resource allocation related indicators include inter-regional water transfer efficiency, reservoir storage utilization rate, and reclaimed water reuse ratio. During the correction process, the seasonal adjustment model assigns higher adjustment factors to the corresponding economic and efficiency indicators. For example, the adjustment factor for the efficiency indicator corresponding to the water supply network leakage rate is increased from 1.0 to 1.4, and the adjustment factor for the economic indicator corresponding to the reclaimed water reuse ratio is increased from 1.2 to 1.6. Through the above modifications, this embodiment enables the second weight adjustment factor to better adapt to the management objectives of water conservation and water resource allocation, making strategies that perform well in water conservation and rational allocation of water resources more advantageous in the evaluation.

[0119] In this embodiment, different weight allocation schemes are adopted according to the characteristics of different seasons, making the weight adjustment more targeted and in line with the key needs of water management in different seasons. The weight of drainage and flood control indicators is increased when rainfall is high, while the weight of water conservation and water resource allocation indicators is increased when rainfall is low. This ensures that the strategy evaluation can focus on the key issues of the current season and improves the effectiveness of water management.

[0120] In one embodiment of this application, before optimizing and evaluating the deduced candidate disposal strategy set using a multi-objective optimization algorithm to obtain the Pareto optimal solution set, uncertainty quantification analysis is performed on the multi-dimensional simulation results;

[0121] Based on the probability distribution of meteorological forecast data, Monte Carlo simulation was used to conduct multiple sampling simulations in the water affairs digital twin.

[0122] Calculate the probability distribution and confidence interval of the multi-dimensional simulation results for each candidate disposal strategy;

[0123] The width of the confidence interval is used as the optimization objective and as a decision risk indicator, and then input into a multi-objective optimization algorithm for solution.

[0124] In one embodiment, based on the probability distribution of meteorological forecast data, Monte Carlo simulation is used to perform multiple sampling simulations in a water digital twin. This includes: obtaining a probability distribution model of the meteorological forecast data, where the probability distribution model is a multivariate Gaussian distribution fitted based on historical meteorological data, including the joint probability distribution of temperature, precipitation, wind speed, and humidity; randomly selecting multiple sets of meteorological data samples from the joint probability distribution using the Latin hypercube sampling method; and for each set of meteorological data samples, running each candidate disposal strategy in the water digital twin and recording multi-dimensional simulation results, including pipeline pressure, flow rate, water quality indicators, and equipment energy consumption.

[0125] In this embodiment, based on the probability distribution of meteorological forecast data, Monte Carlo simulation is used to conduct multiple sampling simulations in a water affairs digital twin. The meteorological forecast data itself contains uncertainties, and its probability distribution can be obtained through statistical analysis of historical meteorological data. This embodiment generates a large amount of random sampling data conforming to the distribution characteristics using Monte Carlo simulation based on the probability distribution. Each set of sampling data represents a possible meteorological scenario. The sampling data is input into the water affairs digital twin, and multiple independent simulations are performed for each strategy in the candidate response strategy set. For example, if the sampling number is set to 1000 times, each candidate strategy will be simulated under 1000 different meteorological scenarios, resulting in 1000 sets of corresponding multi-dimensional simulation results.

[0126] In this embodiment, for each candidate strategy, data from multiple sampling simulations are collected for each indicator in the multi-dimensional simulation results; for example, the total investment amount in the economic indicators, the drainage speed of the pipeline network in the efficiency indicators, and the equipment failure probability in the reliability indicators. The probability distribution type is determined through statistical analysis. Simultaneously, based on the set confidence level, the confidence interval of the indicator is calculated, which is the range of the overall parameter estimated from the sample data. For example, the drainage speed of a candidate strategy in 1000 simulations is analyzed to follow a normal distribution, and its 95% confidence interval is [1.2 m / s, 1.8 m / s], indicating that under 95% of weather scenarios, the drainage speed of the pipeline network after implementing this strategy will fall within this interval.

[0127] In this embodiment, the width of the confidence interval represents the degree of uncertainty in the multi-dimensional simulation results. A larger width indicates greater performance differences of the candidate strategy under different weather scenarios, leading to higher decision-making risk after implementation. Conversely, a smaller width indicates higher stability and reliability of the strategy, resulting in lower decision-making risk. In the multi-objective optimization algorithm, in addition to the original economic, efficiency, and reliability indicators, the confidence interval width is used as a new optimization objective, namely, the decision-making risk indicator. The multi-objective optimization algorithm reduces decision-making risk while pursuing better economy, higher efficiency, and higher reliability. This embodiment, through the above method, ensures that the Pareto optimal solution set obtained by the multi-objective optimization algorithm not only includes strategies that perform well under the desired weather scenario but also considers the stability of the strategy under various uncertain weather scenarios, making the selected optimal treatment strategy more practically valuable and better able to cope with complex and changing weather conditions.

[0128] In summary, this embodiment performs uncertainty quantification analysis on multi-dimensional simulation results, making the strategy evaluation more closely reflect actual conditions. Secondly, the Monte Carlo simulation method is used for multiple sampling simulations, improving the accuracy and reliability of uncertainty analysis. Incorporating the confidence interval width as a decision risk indicator into the optimization objective ensures that the optimal strategy considers not only the effects but also the risks, reducing potential decision-making risks and improving the robustness of water management.

[0129] In one embodiment of this application, calculating the probability distribution and confidence interval of the multi-dimensional simulation results for each candidate disposal strategy includes:

[0130] For each candidate disposal strategy, the multi-dimensional simulation results from multiple sampling simulations are summarized to form a data sequence for each dimension;

[0131] For each dimension of the data sequence, the probability distribution is calculated using the kernel density estimation method, and the confidence interval is determined based on the percentile method.

[0132] Calculate the width of the confidence interval for each dimension as a risk indicator for that dimension, and combine the risk indicators of all dimensions to obtain the overall decision risk indicator.

[0133] In this embodiment, for each candidate disposal strategy, the multi-dimensional simulation results from multiple sampling simulations are summarized to form a data sequence for each dimension. For each candidate disposal strategy, the multi-dimensional simulation results generated from multiple sampling simulations using the Monte Carlo simulation method are summarized. These multi-dimensional simulation results include multiple dimensions such as economy, efficiency, and reliability. Each dimension includes specific indicator data; for example, the economy dimension includes data such as total investment amount and operation and maintenance costs, while the efficiency dimension includes data such as pipeline drainage speed and water quality compliance rate. In this embodiment, during the summarization process, simulation results for the same indicators are organized according to the dimension to form a data sequence for each dimension. For example, for the total investment amount indicator in the economy dimension of a candidate strategy, the multiple total investment amount values ​​obtained from the simulation are arranged in sampling order to form a data sequence for the total investment amount under that dimension; similarly, data sequences for pipeline drainage speed in the efficiency dimension and equipment failure probability in the reliability dimension are also organized.

[0134] In this embodiment, for each dimension of the data sequence, the kernel density estimation method is used to calculate the probability distribution, and the confidence interval is determined based on the percentile method. The kernel density estimation method is a non-parametric estimation method that does not require prior assumptions about the data distribution type, thus better reflecting the inherent distribution characteristics of the data. It describes the probability distribution of the data sequence in that dimension by placing a kernel function at each data point and superimposing all kernel functions. For example, kernel density estimation of a data sequence of drainage velocity in a pipe network clearly shows the probability of different drainage velocity values ​​occurring. When determining the confidence interval, the percentile method is used. For a set confidence level, such as 95%, the values ​​corresponding to the 2.5th and 97.5th percentiles in the data sequence are identified; the interval between these two values ​​is the 95% confidence interval for that dimension.

[0135] This embodiment calculates the width of the confidence interval for each dimension as a risk indicator for that dimension, and calculates the risk indicators for all dimensions to obtain the overall decision risk indicator. The width of the confidence interval for each dimension is obtained by subtracting the lower limit from the upper limit. This width represents the fluctuation range of the indicator for that dimension under different meteorological scenarios; the larger the width, the higher the uncertainty of that dimension, and the greater the corresponding risk. Therefore, it is used as the risk indicator for that dimension. After obtaining the risk indicators for each dimension, a weighted summation is used to obtain the overall decision risk indicator. The weights can be determined based on the importance of each dimension in water management.

[0136] In this embodiment, the kernel density estimation method is used to calculate the probability distribution, which can more flexibly capture the characteristics of data distribution. Compared with traditional parameterized methods, it does not require a preset distribution form and can more accurately represent the actual distribution law of the simulation results, thus improving the accuracy of probability distribution characterization. The confidence interval is determined based on the percentile method, avoiding assumptions about the data distribution pattern. It is applicable to various types of distributed data, and the calculation process is intuitive and robust, more reliably reflecting the uncertainty range of the simulation results. Secondly, the width of the confidence interval for each dimension is used as the risk indicator for that dimension, making the uncertainties of different dimensions comparable. By integrating the risk indicators of all dimensions, an overall decision risk indicator is obtained, achieving a comprehensive measurement of strategy uncertainty. This provides a clear basis for subsequent multi-objective optimization and helps select efficient and robust disposal strategies. This further supports the risk assessment of candidate strategies, improving the scientific nature and reliability of decision-making.

[0137] In one embodiment of this application, fuzzy clustering analysis is performed on the strategies in the Pareto optimal solution set to obtain the residual strategies; including:

[0138] An adaptive fuzzy clustering algorithm is used to calculate the similarity matrix between strategies, with the key performance indicators of the strategies in the multi-dimensional simulation results as feature vectors.

[0139] The optimal number of clusters is automatically determined using a clustering effectiveness function, and the strategy is divided into multiple clusters.

[0140] Based on the silhouette coefficient and average performance index within each cluster, clusters with significantly poor performance are identified and removed to obtain the remaining strategies.

[0141] In this embodiment, the multi-dimensional simulation results include performance data of the strategy in multiple aspects such as economy, effectiveness, and reliability. Key performance indicators (KPIs) are selected from these KPIs, which accurately reflect the core performance of the strategy. The specific values ​​of each strategy on these KPIs are integrated to form a unique feature vector for that strategy. The adaptive fuzzy clustering algorithm can automatically adjust parameters such as cluster centers and membership functions based on the distribution density and dispersion of the feature vectors, without manual intervention. When calculating the similarity matrix, the Euclidean distance formula is used to calculate the distance between any two strategy feature vectors. The smaller the distance value, the closer the two strategies are in terms of performance on key performance indicators, the higher the similarity, and the larger the value at the corresponding position in the similarity matrix, and vice versa. For example, if strategy A and strategy B have very small differences in indicators such as cost-benefit ratio and task completion time, then the distance between them is small, and the values ​​at positions (A,B) and (B,A) in the similarity matrix will be large.

[0142] In this embodiment, the optimal number of clusters is automatically determined using a clustering validity function, and the strategies are divided into multiple clusters. The clustering validity function, including the Calinski-Harabasz index and the Xie-Beni index, is an important tool for judging the rationality of clustering results. A larger Calinski-Harabasz index indicates higher intra-cluster similarity and greater inter-cluster differences, resulting in better clustering performance; a smaller Xie-Beni index indicates better clustering performance. The algorithm sequentially tries different numbers of clusters, calculates the corresponding clustering validity function value for each number, and then selects the number of clusters with the optimal function value as the optimal number of clusters. After determining the optimal number of clusters, the adaptive fuzzy clustering algorithm assigns all strategies in the Pareto optimal solution set to different clusters based on the similarity of the strategy feature vectors, ensuring that the strategies within each cluster have high similarity in key performance indicators.

[0143] In this embodiment, the silhouette coefficient combines the compactness of intra-cluster strategies and the separability of inter-cluster strategies, with a value between [-1, 1]. When the silhouette coefficient is close to 1, it indicates that the strategies within the cluster are highly similar and significantly different from those of other clusters, resulting in ideal clustering. When the silhouette coefficient is close to -1, it indicates that the strategies within the cluster are more similar to those of other clusters, resulting in poor clustering. The intra-cluster average performance index is obtained by averaging the key performance indices of all strategies within the cluster. By comparing and analyzing the silhouette coefficient and intra-cluster average performance index of each cluster, if the silhouette coefficient of a cluster is much lower than that of other clusters, and its intra-cluster average performance index lags significantly behind other clusters in several key indices, then this cluster can be identified as a cluster with significantly poor performance. After removing such clusters from the Pareto optimal solution set, the remaining strategies constitute the residual strategies. These residual strategies have better performance, laying a good foundation for the subsequent selection of the optimal disposal strategy.

[0144] In summary, this embodiment employs an adaptive fuzzy clustering algorithm with key performance indicators as feature vectors. This approach more accurately captures the similarities and differences between strategies. Compared to traditional hard clustering methods, fuzzy clustering allows strategies to belong to multiple clusters with different membership degrees, better reflecting the gradual changes in strategy performance in real-world decision-making scenarios and improving the flexibility and accuracy of clustering. Secondly, the optimal number of clusters is automatically determined through a clustering validity function, avoiding the subjectivity and experience-dependent nature of manually setting the number of clusters. This makes the clustering results more closely match the distribution characteristics of the data itself, improving the objectivity and scientific rigor of the clustering process. Based on the silhouette coefficient and the average performance index within each cluster, lagging clusters are identified and eliminated. This ensures the reliability of clustering quality through the silhouette coefficient and quantifies the overall performance of the clusters through the average performance index, achieving precise screening of inefficient strategies, effectively reducing the size of candidate strategies, and lowering the computational complexity of subsequent decisions.

[0145] In one embodiment of this application, a satisfaction evaluation mechanism based on historical case matching selects the strategy with the highest similarity to the handling effect of historical successful cases from the remaining strategies, including:

[0146] Retrieve historical cases from the historical case database that are similar to the operation of urban water affairs, and extract historical disposal strategies and their effect evaluation indicators;

[0147] The cosine similarity between the remaining policy and the historical policy in the multidimensional feature space is calculated using a deep learning model, and weighted similarity is calculated in combination with the effect evaluation index.

[0148] The weights of historical cases are adjusted based on a time decay factor to prioritize recent successful cases, and the strategy with the highest weighted similarity is selected as the optimal handling strategy.

[0149] In this embodiment, the historical case database stores a large number of past events and corresponding handling records of the urban water system, including the operational status at the time of the event, the handling strategies adopted, and the effectiveness evaluation indicators after the strategies were implemented. During the retrieval process, this embodiment compares the current operational status parameters of the urban water system with the status parameters in the historical cases to filter out cases with high similarity. For example, if the pressure in a certain area of ​​the city's pipe network suddenly increases due to heavy rain, the database focuses on retrieving historical cases of similar pipe network pressure anomalies caused by heavy rain and extracts the historical handling strategies and corresponding effectiveness evaluation indicators from these cases.

[0150] This embodiment transforms the features of the remaining strategies and historical disposal strategies into multi-dimensional feature vectors, constructing a multi-dimensional feature space. The deep learning model, trained on a large amount of strategy feature data, can learn the potential correlations between features and more accurately calculate the cosine similarity between feature vectors. The cosine similarity ranges from -1 to 1; the closer the value is to 1, the more consistent the directions of the two strategies in the feature space, and the higher the similarity. Based on this, a weighted processing is performed using performance evaluation metrics. Different performance metrics are assigned corresponding weights, and the similarity of each performance metric is multiplied by its weight and then summed to obtain the weighted similarity.

[0151] This embodiment adjusts the weights of historical cases based on a time decay factor to prioritize recent successful cases and selects the strategy with the highest weighted similarity as the optimal treatment strategy. The time decay factor represents the timeliness of historical cases; as time passes, the reference value of a case gradually decreases. Its calculation formula can use an exponential decay function.

[0152] This embodiment multiplies the weighted similarity of historical cases by the corresponding time decay factor to obtain the final comprehensive similarity. By comparing the comprehensive similarity of the remaining strategies with each historical successful case, the strategy with the highest comprehensive similarity is selected as the optimal handling strategy.

[0153] The above methods, including retrieving similar cases from a historical case database and extracting strategies and performance indicators, fully leverage past successes, providing real-world case support for the decision-making process, reducing subjective assumptions, and improving the reliability of strategy selection. Secondly, employing a deep learning model to calculate cosine similarity in a multi-dimensional feature space and combining it with performance indicators for weighted calculations, accurately captures the similarity of strategies across multiple dimensions while also considering actual treatment effects. This makes the similarity assessment more comprehensive and objective, avoiding the one-sidedness of single-dimensional comparisons. Furthermore, adjusting the weights of historical cases based on a time decay factor, prioritizing recent successful cases, adapts to the dynamic changes in urban water systems over time, making decisions more aligned with current realities, improving the timeliness and applicability of strategies, and further enhancing the accuracy and effectiveness of optimal strategy selection.

[0154] In one embodiment of this application, a risk assessment result is generated by comprehensively analyzing a feature set, situation level, and situation change trend, using a preset risk assessment index system, including:

[0155] Based on the preset risk assessment indicator system, obtain multiple risk assessment indicators and the weight of each indicator;

[0156] For each risk assessment indicator, the evaluation value of the indicator is calculated based on the comprehensive analysis of the feature set, the situation level, and the trend of situation change.

[0157] The comprehensive risk value is calculated by weighted summation based on the evaluation value and weight of each risk assessment indicator.

[0158] The overall risk value is compared with the preset risk level threshold to determine the risk assessment result, which includes the risk level and risk description.

[0159] In this embodiment, the preset risk level threshold is adjusted based on historical water event data through statistical learning. The preset risk assessment indicator system is constructed in conjunction with the characteristics and management needs of the urban water system, and includes risk indicators across multiple dimensions. For example, the pipeline safety dimension includes indicators such as the degree of pipeline aging, pipeline corrosion rate, and interface sealing performance; the water quality safety dimension includes indicators such as the frequency of water quality exceeding standards, peak pollutant concentrations, and microbial content; and the operational efficiency dimension includes indicators such as pump station operating load rate, pipeline water transmission energy consumption, and equipment failure downtime. The weight of each indicator is determined using methods such as the analytic hierarchy process (AHP) and expert scoring, and the weight reflects the importance of the indicator in the risk assessment.

[0160] The comprehensive analysis feature set in this embodiment provides basic operational data for the water system; the status level represents the overall state of urban water affairs, while the trend of status change predicts future development directions. Taking the aging degree of the pipeline network as an example, data such as the service life and historical maintenance records of the pipeline network are extracted from the comprehensive analysis feature set. Combined with the current status level and future status change trend, the evaluation value of this indicator is calculated using a preset scoring standard. Based on the evaluation value and weight of each risk assessment indicator, the comprehensive risk value is calculated through weighted summation.

[0161] The formula for weighted summation is: Overall risk value = Σ (indicator evaluation value × indicator weight).

[0162] This embodiment integrates risk indicators from multiple dimensions into a single comprehensive value, providing an intuitive representation of the overall risk level of urban water affairs.

[0163] In this embodiment, the preset risk level thresholds are set according to the actual situation of urban water management. For example, the comprehensive risk value is divided into four levels: 0-30 points are low risk, 31-60 points are medium risk, 61-80 points are high risk, and 81-100 points are extremely high risk. If the calculated comprehensive risk value is 48 points, it is determined to be medium risk after comparing with the threshold. The risk description supplements the risk level by combining the specific performance of each indicator. This embodiment provides clear and specific risk information for water management personnel by combining risk levels and risk descriptions, thus assisting in subsequent decision-making.

[0164] The above embodiment conducts the assessment based on a pre-set risk assessment indicator system, ensuring the standardization and systematic nature of the assessment process, avoiding arbitrariness in indicator selection, and making the risk assessment results more comparable and credible. Secondly, based on a comprehensive analysis of feature sets, situation levels, and situation change trends, the evaluation values ​​of each indicator are calculated, considering risk from a multi-dimensional and dynamic perspective, comprehensively representing the actual risk status of the urban water system, and overcoming the limitations of single data or static assessments. Furthermore, by calculating the comprehensive risk value through weighted summation, both the specific performance of each risk indicator and the differences in the importance of different indicators in the overall risk assessment are considered, enabling the comprehensive risk value to scientifically quantify the overall risk level. This provides a clear risk basis for the formulation of subsequent response strategies, improving the practicality of risk management and decision-making efficiency.

[0165] In one embodiment of this application, the risk assessment index system includes water quality risk index, water quantity risk index, equipment risk index, and environmental risk index; wherein, the water quality risk index is calculated based on water quality parameters in water sensor monitoring data, the water quantity risk index is calculated based on flow and pressure parameters in water sensor monitoring data, the equipment risk index is calculated based on equipment operating status and fault history in equipment information, and the environmental risk index is calculated based on meteorological data and geographic information data.

[0166] In this embodiment, the water quality parameters in the water sensor monitoring data include pH value, turbidity, residual chlorine content, heavy metal concentration, and organic matter content. Water quality risk indicators can be further refined into several specific indicators, such as the frequency of water quality exceeding standards, which is the number of times a water quality parameter exceeds the national standard per unit time, obtained by statistically analyzing the ratio of the number of records of each parameter exceeding the standard to the total number of monitoring times; the fluctuation range of water quality parameters, calculated as the difference between the maximum and minimum values ​​of water quality parameters within a certain period; the larger the difference, the worse the water quality stability and the higher the risk; and the pollutant diffusion rate, calculated based on the gradient of pollutant concentration changes, combined with data such as water flow velocity, reflecting the speed at which pollutants diffuse in the water body.

[0167] In this embodiment, flow parameters include the instantaneous flow rate and cumulative flow rate of the water supply network, and the drainage flow rate of the drainage network; pressure parameters include the static pressure, dynamic pressure, and pressure fluctuation range in the network. Specific water risk indicators include the network flow deviation rate, which is the percentage difference between the actual monitored flow rate and the design flow rate relative to the design flow rate; a larger deviation rate indicates a higher risk of water supply and demand imbalance; pressure anomaly frequency, which counts the number of times the network pressure exceeds the normal range per unit time; and pipeline hydraulic impact intensity, calculated based on the peak value and frequency of pressure fluctuations; the higher the intensity, the greater the risk of pipeline damage due to water hammer effect.

[0168] In this embodiment, the equipment operating status includes the pump's operating power, speed, and temperature; the valve's on / off status and opening / closing time; and the disinfection equipment's operating parameters. The fault history includes the type, frequency, repair time, and cause of past equipment faults. Equipment risk indicators include the equipment failure rate (the ratio of the number of faults within a certain period to the total operating time); the mean time between failures (MTBF), calculated by dividing the total fault-free operating time by the number of faults (a shorter MTBF indicates lower equipment reliability); and the fault repair timeliness rate, which is the proportion of faults repaired within a specified time to the total number of faults (a low MTBF indicates insufficient equipment maintenance response capability).

[0169] In this embodiment, meteorological data such as rainfall, rainfall intensity, temperature, and wind speed are combined with geographic information data such as topographic slope, altitude, soil type, and water distribution to calculate environmental risk indicators. Specific indicators include a rainstorm flooding risk index, calculated by weighting factors such as rainfall, rainfall intensity, topographic slope, and pipeline coverage density; a higher index indicates a greater risk of flooding. A low-temperature freezing cracking risk index is calculated based on parameters such as temperature, pipeline burial depth, and soil thermal conductivity to characterize the likelihood of pipelines freezing and cracking in low-temperature environments. A water conservation capacity index, combining rainfall, vegetation coverage, and soil water storage capacity, assesses the surrounding environment's effect on water conservation; a low index indicates a high risk of insufficient water supply.

[0170] In one embodiment of this application, obtaining the weight corresponding to each indicator includes:

[0171] Based on historical water incident data and expert scores, the initial weights were calculated using the analytic hierarchy process.

[0172] The initial weights are corrected using the entropy weight method to obtain the final comprehensive weights;

[0173] The process of using the entropy weight method to correct the initial weights and obtain the final comprehensive weights includes: for each risk assessment indicator, calculating the information entropy based on historical data, calculating the entropy weight based on the information entropy, and then weighting and fusing the entropy weight with the initial weights to obtain the final comprehensive weight.

[0174] In this embodiment, the risk assessment indicator system is first divided into the target layer (obtaining indicator weights), the criterion layer (water quality risk indicators, water quantity risk indicators, equipment risk indicators, and environmental risk indicators) and the scheme layer (each specific risk assessment indicator) using the analytic hierarchy process.

[0175] Historical water-related event data was collected to analyze the impact of different indicators during the occurrence and development of these events. For example, the correlation between indicators such as equipment failure rate and frequency of pressure anomalies was statistically analyzed in past pipe burst events. Simultaneously, water industry experts were invited to conduct pairwise comparisons and scoring of the importance of each indicator, constructing a judgment matrix using a 1-9 scale. By calculating the largest eigenvalue and corresponding eigenvector of the judgment matrix, and normalizing the eigenvectors, the initial weights of each indicator were obtained.

[0176] The initial weights are corrected using the entropy weight method to obtain the final comprehensive weight. This correction involves: for each risk assessment indicator, calculating the information entropy based on historical data, calculating the entropy weight based on the information entropy, and then weighting and fusing the entropy weight with the initial weights to obtain the final comprehensive weight. Specifically, firstly, historical monitoring data for each risk assessment indicator is selected and standardized to obtain a standardized matrix. Then, the information entropy of each indicator is calculated; the smaller the information entropy, the greater the data dispersion of the indicator, the more information it contains, and the more important its role in the assessment. The entropy weight is calculated based on the information entropy, reflecting the objective importance of the indicator data itself. Finally, the entropy weight is weighted and fused with the initial weight to obtain the comprehensive weight.

[0177] In this embodiment, the analytic hierarchy process (AHP) and entropy weighting method are used to determine the weights of indicators, taking into account both subjective experience and objective data characteristics, thus avoiding the limitations of a single method. First, the AHP is used to calculate the initial weights, providing a framework based on professional knowledge for weight determination, ensuring that the weight allocation aligns with the actual focus of water management. Then, the entropy weighting method is used for correction, quantifying the information content and volatility of the indicators through information entropy, allowing the weights to dynamically reflect data characteristics and improving their objectivity and adaptability. Finally, the entropy weights and initial weights are weighted and fused to obtain the final comprehensive weights, achieving an organic combination of subjective experience and objective data. This retains the expert knowledge's emphasis on key indicators while avoiding subjective bias through data-driven adjustments, making the weight allocation of each indicator more in line with actual assessment needs and laying a reliable foundation for accurately calculating the comprehensive risk value. This further improves the scientific nature of the risk assessment indicator system and the accuracy of the assessment results.

[0178] Corresponding to the urban water affairs situation awareness management method in the above embodiment, Figure 2 This is a structural block diagram of an urban water situation awareness management system provided in one embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 2 The city water situation awareness management system 20 includes: a data acquisition module 21, a data processing module 22, a situation assessment module 23, a trend prediction module 24, a risk assessment module 25, a candidate strategy module 26, and a strategy selection module 27.

[0179] Among them, the data acquisition module 21 is used to acquire multi-source heterogeneous data, which includes water sensor monitoring data, meteorological data, geographic information data and equipment information;

[0180] Data processing module 22 is used to preprocess multi-source heterogeneous data to obtain preprocessed data, and to extract and fuse features from the preprocessed data to obtain a comprehensive analysis feature set.

[0181] The situation assessment module 23 is used to input the comprehensive analysis feature set into the situation assessment model trained based on historical water affairs data to determine the situation level of urban water affairs.

[0182] The trend prediction module 24 is used to input the comprehensive analysis feature set into the time series prediction model to predict the trend of future situation changes.

[0183] Risk assessment module 25 is used to conduct a comprehensive assessment based on a set of comprehensive analysis features, situation level and situation change trend, and to generate risk assessment results using a preset risk assessment indicator system.

[0184] The candidate strategy module 26 is used to generate a set of candidate disposal strategies by calling a preset expert knowledge base for matching and reasoning based on the situation level, situation change trend and risk assessment results.

[0185] The strategy selection module 27 is used to evaluate the strategies in the candidate disposal strategy set and output the optimal disposal strategy.

[0186] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of the modules in the aforementioned device embodiments, for example... Figure 2 The functions of the data acquisition module 21, data processing module 22, situation assessment module 23, trend prediction module 24, risk assessment module 25, candidate strategy module 26, and strategy selection module 27 are shown.

[0187] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0188] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0189] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store device type information.

[0190] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation methods described in any embodiment of the urban water affairs situation awareness management method provided in the embodiments of this application, or they can execute the implementation methods of the electronic devices described in the embodiments of this application, which will not be repeated here.

[0191] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0192] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0193] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0194] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0195] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces or units, or it may be an electrical, mechanical, or other form of connection.

[0196] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0197] Furthermore, the functional units in the various embodiments of this application 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. The integrated unit can be implemented in hardware or as a software functional unit.

[0198] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for urban water affairs situational awareness and management, characterized in that, include: Acquire multi-source heterogeneous data, including water sensor monitoring data, meteorological data, geographic information data, and equipment information; The multi-source heterogeneous data is preprocessed to obtain preprocessed data, and the preprocessed data is then subjected to feature extraction and fusion to obtain a comprehensive analysis feature set. The comprehensive analysis feature set is input into a situation assessment model trained based on historical water affairs data to determine the situation level of urban water affairs; The comprehensive analysis feature set is input into the time series prediction model to predict the trend of future situation changes. Based on the comprehensive analysis feature set, the situation level, and the situation change trend, a comprehensive assessment is conducted using a preset risk assessment index system to generate a risk assessment result. Based on the situation level, situation change trend and risk assessment results, a preset expert knowledge base is invoked for matching and reasoning to generate a set of candidate disposal strategies. The strategies in the candidate disposal strategy set are evaluated, and the optimal disposal strategy is output. The process involves conducting a comprehensive assessment based on the comprehensive analysis feature set, the situation level, and the situation change trend, using a preset risk assessment index system to generate a risk assessment result, including: Based on the preset risk assessment indicator system, obtain multiple risk assessment indicators and the weight of each indicator; For each risk assessment indicator, the evaluation value of the risk assessment indicator is calculated based on the comprehensive analysis of the feature set, the situation level, and the trend of situation change. The comprehensive risk value is calculated by weighted summation based on the evaluation value and weight of each risk assessment indicator. The overall risk value is compared with the preset risk level threshold to determine the risk assessment result, which includes the risk level and risk description.

2. The urban water affairs situation awareness and management method according to claim 1, characterized in that, The step of evaluating the strategies in the candidate disposal strategy set and outputting the optimal disposal strategy includes: Each strategy in the candidate disposal strategy set is input into a preset water affairs digital twin for simulation and deduction to obtain multi-dimensional simulation results corresponding to each strategy; Based on the multi-dimensional simulation results, a multi-objective evaluation function is constructed with economic indicators, efficiency indicators, and reliability indicators as objectives. Based on the multi-objective evaluation function, a multi-objective optimization algorithm is used to optimize and evaluate the deduced candidate disposal strategy set to obtain the Pareto optimal solution set. The final optimal solution strategy is selected from the Pareto optimal solution set based on preset decision rules.

3. The urban water affairs situation awareness and management method according to claim 2, characterized in that, The final optimal action strategy is selected from the Pareto optimal solution set based on preset decision rules, including: Based on the situation level and the risk assessment results, adjust the weight coefficients of each indicator in the multi-objective assessment function; Fuzzy clustering analysis is performed on the strategies in the Pareto optimal solution set to obtain the remaining strategies; Based on the satisfaction evaluation mechanism of historical case matching, the strategy with the highest similarity to the handling effect of historical successful cases is selected from the remaining strategies as the optimal handling strategy.

4. The urban water affairs situation awareness and management method according to claim 3, characterized in that, The step of adjusting the weight coefficients of each indicator in the multi-objective evaluation function based on the situation level and the risk assessment results includes: The first weight adjustment factor is obtained by querying the preset weight mapping table based on the situation level. Based on the risk level and risk type in the risk assessment results, a second weight adjustment factor is obtained through a risk weight calculation model, which is a weight allocation model constructed based on the analytic hierarchy process and the entropy weight method. The first weight adjustment factor and the second weight adjustment factor are fused to obtain the weight coefficients of each indicator.

5. The urban water affairs situation awareness and management method according to claim 4, characterized in that, The process of fusing the first weight adjustment factor and the second weight adjustment factor to obtain the weight coefficients of each indicator includes: Meteorological data is obtained from the multi-source heterogeneous data as environmental data, and time-series analysis and feature extraction are performed on the meteorological data to obtain the predicted precipitation in the future preset period as seasonal features. The seasonal characteristics are input into a preset seasonal adjustment model to generate a seasonal adjustment factor. The seasonal adjustment model is a regression model trained based on historical meteorological and weighted adjustment data. The first weight adjustment factor and the second weight adjustment factor are weighted and corrected using the seasonal adjustment factor to obtain the corrected first weight adjustment factor and the corrected second weight adjustment factor. The modified first weight adjustment factor and the modified second weight adjustment factor are combined to obtain the weight coefficient of each indicator.

6. The urban water affairs situation awareness and management method according to claim 5, characterized in that, The step of using the seasonal adjustment factor to weight and correct the first weight adjustment factor and the second weight adjustment factor respectively to obtain the corrected first weight adjustment factor and the corrected second weight adjustment factor includes: When the seasonal feature is the first seasonal feature, the seasonal adjustment model uses a first preset weight allocation scheme to adjust the first weight adjustment factor and the second weight adjustment factor to obtain the corrected first weight adjustment factor. When the seasonal feature is the second seasonal feature, the seasonal adjustment model uses a second preset weight allocation scheme to adjust the first weight adjustment factor and the second weight adjustment factor to obtain the corrected second weight adjustment factor. Wherein, the first seasonal characteristic is that the predicted precipitation is greater than the second threshold; the second seasonal characteristic is that the predicted precipitation is less than the first threshold, and the second threshold is greater than the first threshold; the first preset weight allocation scheme is to increase the weight of drainage and flood control related indicators; the second preset weight allocation scheme is to increase the weight of water conservation and water source scheduling related indicators.

7. The urban water affairs situation awareness and management method according to claim 2, characterized in that, Also includes: Before optimizing and evaluating the simulated candidate disposal strategy set using a multi-objective optimization algorithm to obtain the Pareto optimal solution set, uncertainty quantification analysis is performed on the multi-dimensional simulation results. Based on the probability distribution of meteorological forecast data, the Monte Carlo simulation method was used to conduct multiple sampling simulations in the water affairs digital twin. Calculate the probability distribution and confidence interval of the multi-dimensional simulation results for each candidate disposal strategy; The width of the confidence interval is used as the optimization objective and the decision risk indicator, and then input into a multi-objective optimization algorithm for solution.

8. A city water affairs situation awareness and management system, characterized in that, include: The data acquisition module is used to acquire multi-source heterogeneous data, including water sensor monitoring data, meteorological data, geographic information data, and equipment information. The data processing module is used to preprocess the multi-source heterogeneous data to obtain preprocessed data, and to extract and fuse features from the preprocessed data to obtain a comprehensive analysis feature set. The situation assessment module is used to input the comprehensive analysis feature set into a situation assessment model trained based on historical water affairs data to determine the situation level of urban water affairs; The trend prediction module is used to input the comprehensive analysis feature set into the time series prediction model to predict the trend of future situation changes. The risk assessment module is used to conduct a comprehensive assessment based on the comprehensive analysis feature set, the situation level, and the situation change trend, using a preset risk assessment index system, and generate a risk assessment result. The process involves conducting a comprehensive assessment based on the comprehensive analysis feature set, the situation level, and the situation change trend, using a preset risk assessment index system to generate a risk assessment result, including: Based on the preset risk assessment indicator system, obtain multiple risk assessment indicators and the weight of each indicator; For each risk assessment indicator, the evaluation value of the risk assessment indicator is calculated based on the comprehensive analysis of the feature set, the situation level, and the trend of situation change. The comprehensive risk value is calculated by weighted summation based on the evaluation value and weight of each risk assessment indicator. The overall risk value is compared with the preset risk level threshold to determine the risk assessment result, which includes the risk level and risk description. The candidate strategy module is used to generate a set of candidate disposal strategies by calling a preset expert knowledge base for matching and reasoning based on the situation level, situation change trend and risk assessment results. The strategy selection module is used to evaluate the strategies in the candidate disposal strategy set and output the optimal disposal strategy.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images