An intelligent management system and method applied to intelligent water affairs

Abstract

The application discloses an intelligent management system and method applied to intelligent water affairs, relates to the technical field of intelligent water affairs scheduling, and comprises the following steps: collecting sewage treatment multidimensional operation monitoring data, calculating the margin value of three types of criteria of topology availability, bottleneck capacity and disturbance tolerance, generating a scheduling state coefficient and judging whether an improvement process is triggered; based on the scheduling state coefficient, combining peak and trough period characteristics, adopting a random forest algorithm to construct a filter urgency evaluation model, outputting an urgency score and dividing a priority and a delayable backwashing filter set; generating a candidate reconstruction scheme set and screening an optimal reconstruction scheme, arranging the backwashing time sequence of each filter under multiple constraints based on a time expansion maximum flow model, and outputting a final scheduling scheme; after verification by a digital twin platform, the final scheduling scheme is executed and the filter state and model parameters are updated; and the application realizes fine scheduling and dynamic adaptive control of filter backwashing.

Description

Technical Field

[0001] This invention relates to the field of smart water management technology, and more specifically, to an intelligent management system and method for smart water management. Background Technology

[0002] Smart water systems are driving urban water treatment facilities toward digitalization and intelligence. The maturity of technologies such as digital twins, the Internet of Things for water, artificial intelligence, and cloud computing is continuously improving the technological environment for smart water management, and the integration of the water conservancy industry with smart technologies has become a trend. However, current digital twin water conservancy systems are still at the stage of static mapping using technologies such as GIS and BIM, and have not yet achieved true symbiosis between physical water conservancy projects and virtual models. Traditional information technology methods cannot meet the operational needs of reservoir scheduling, dam safety management, and other related tasks.

[0003] In wastewater influent facilities, filter backwashing is a crucial step in ensuring filtration performance. During operation, filters accumulate suspended solids and particulate matter, which need to be removed periodically through backwashing to restore filtration capacity. Existing backwashing processes mainly target the entire filter unit and lack the ability to perform refined analysis and control on individual filters or modules.

[0004] Existing smart water systems have the following problems in backwashing operations:

[0005] Coarse-grained scheduling leads to blind spots in backwash tank allocation. Traditional systems typically start and stop backwashing according to fixed schedules or overall load conditions, ignoring multi-dimensional factors such as the degree of clogging of individual filters, filter media lifespan, and water quality disturbance. This can easily result in some filters not being washed for a long time while other filters are frequently backwashed.

[0006] The lack of redundancy assessment and dynamic adaptation capabilities means that forcibly scheduling backwashing during peak influent demand periods may lead to influent pressure drop, water quality fluctuations, and even safety risks. Existing systems lack a redundancy evaluation mechanism that comprehensively considers topology availability, bottleneck capacity, and disturbance tolerance, making it impossible to adaptively adjust the number of backwashes based on system operating conditions.

[0007] The application of digital twins is insufficient. Existing digital twin models are mostly used for static monitoring and fault early warning, lacking deep coupling with scheduling decisions. Backflushing plans are often formulated in business systems, lacking dynamic simulation and rapid verification based on digital twin platforms, which affects the reliability of scheduling.

[0008] To address the above problems, this invention proposes a solution. Summary of the Invention

[0009] The purpose of this invention is to provide an intelligent management system and method for smart water management, which solves the technical problem that existing smart water management systems do not perform detailed analysis on individual filter tanks during backwashing operations, resulting in blind spots in backwash tank allocation.

[0010] The objective of this invention can be achieved through the following technical solutions:

[0011] A smart management system and method for smart water management collects multi-dimensional operation monitoring data of sewage treatment, analyzes three criteria: topology availability, bottleneck capacity, and disturbance tolerance, and calculates their corresponding redundancy values; generates scheduling status coefficients based on the redundancy values ​​of each criterion; if all redundancies are sufficient, the original scheduling scheme is maintained; otherwise, an improvement process is triggered.

[0012] After the improved process is triggered based on the scheduling status coefficient, the filter urgency assessment model is constructed using a random forest algorithm, taking into account peak and off-peak periods and the clogging risk characteristic value of each filter as input. The model outputs an urgency score and generates a filter flushing urgency queue. Based on the comparison between the urgency score and the preset segmentation threshold, the filters in the filter flushing urgency queue with an urgency score not lower than the preset segmentation threshold are divided into a priority backwashing filter set, and the rest are divided into a delayed backwashing filter set.

[0013] A set of candidate reconfiguration schemes is generated based on a set of finite operation steps. The incremental number of available backwash tanks and the valve position switching complexity of each candidate reconfiguration scheme are calculated. The optimal reconfiguration scheme that meets the preset coverage threshold and whose valve position switching complexity does not exceed the preset upper limit is selected. Based on the optimal reconfiguration scheme, the maximum flow model is expanded in time. Under the constraints of the upper limit of tank resources, the upper limit of disturbance budget and the upper limit of parallel backwashing, the backwashing start tank and backwashing end tank of each filter in the priority backwashing filter set and the delayed backwashing filter set are arranged in stages. The final scheduling scheme containing the backwashing scheduling scheme and the corresponding valve position switching command is output.

[0014] The final scheduling scheme output is quickly verified on the digital twin platform. After verification, it is sent to the equipment control system for execution, and the filter status and parameters of each model are updated after backwashing is completed.

[0015] As a further aspect of this invention: multi-dimensional operational monitoring data includes: various real-time operational data acquired by intelligent sensors deployed in each structural unit and key pipeline of the wastewater treatment plant process flow, namely: influent and effluent flow rates and pipe section flow rates collected by flow meters; pipe section pressure and influent and effluent pressure difference of each filter collected by differential pressure gauges; influent and effluent turbidity and effluent turbidity of each filter collected by online turbidity meters; equipment energy consumption collected by power meters; valve position opening and closing status, pump operating frequency and switching status of each device collected by valve and pump station status monitors; historical operation and maintenance information of filter backwashing synchronously collected by the water IoT edge gateway; and resource limit data statistically analyzed by time slot, including aeration air volume limit, backwash water volume limit and sludge discharge capacity limit.

[0016] As a further aspect of this invention: Three criteria—topology availability, bottleneck capacity, and disturbance tolerance—are analyzed, and their corresponding redundancy values ​​are calculated. Specifically, the wastewater treatment plant process flow is modeled as a weighted directed graph, denoted as the wastewater treatment topology. Each node represents a structural unit and equipment, and each edge represents various types of pipes, valves, and pump stations. When certain filters enter the backwashing state, relevant nodes and edges are temporarily removed from the wastewater treatment topology, and a currently available wastewater treatment topology is generated based on real-time valve positions and pump status. A depth-first search is used to check whether the contact disinfection tank outlet and the receiving water body still maintain connectivity. If the path is broken, the topology availability criterion is determined to be zero. If the path is connected, the maximum flow algorithm is used to calculate the maximum effluent capacity of the currently available topology from the contact disinfection tank outlet to the receiving water body, and this capacity is compared with the actual influent demand for backwashing. The difference between the two is the topology availability redundancy value.

[0017] The system identifies bottleneck components in the current wastewater treatment topology, calculates the ratio of actual flow to rated capacity for each bottleneck component, takes the maximum load utilization rate, and subtracts the maximum load utilization rate from 1 to obtain the bottleneck capacity margin. A water quality disturbance model, using a hybrid mechanistic model and long short-term memory network, predicts the peak turbidity and short-term decrease in disinfection indicators of the effluent shortly after backwashing, based on backwashing intensity, duration, initial water quality, and historical data on downstream water quality changes. The predicted results are compared with various water quality standard thresholds to calculate turbidity and disinfection tolerances, with the smaller of the two used as the disturbance tolerance margin. The water quality standard thresholds include both turbidity and disinfection standard thresholds. Finally, scheduling state coefficients are generated based on the margin values ​​of various criteria.

[0018] As a further aspect of the present invention: generating scheduling status coefficients based on the redundancy values ​​of various criteria, including: converting topology availability redundancy values... Bottleneck capacity margin value Disturbance tolerance margin value As input, a scheduling state coefficient model is constructed by combining preset upper and lower bound thresholds; the expression for the scheduling state coefficient model is: In the formula This is the scheduling status coefficient. , , Preset upper bound thresholds for various redundancy indicators; , , To represent the preset lower bound thresholds corresponding to various redundancy indices that are close to instability;

[0019] And based on the above scheduling status coefficients The value of is used to output the scheduling status according to the following logic:

[0020] When the scheduling status coefficient PSC=0, it means that the system has sufficient redundancy and the existing schedule should be maintained.

[0021] When the scheduling status coefficient PSC=2, it indicates that the system redundancy is seriously insufficient and it needs to enter the topology reconfiguration mode; when the scheduling status coefficient PSC=1, it indicates that it is in a transitional state and it needs to enter the fine-tuning mode.

[0022] Based on the output value of the scheduling status coefficient, different scheduling strategies are executed, such as maintaining the existing schedule, entering fine-tuning, or triggering topology reconfiguration.

[0023] As a further aspect of the present invention: the logic for dividing peak period characteristics and off-peak periods is as follows: when the scheduling status coefficient PSC=1, it is in a transitional state and enters the fine-tuning mode. The historical backwashing time interval data in the multi-dimensional operation monitoring data is used to associate each backwashing time interval with a preset time period to construct a time period-backwashing time interval mapping dataset.

[0024] The percentile statistical method is used to calculate the backwash time interval, lower percentile value, upper percentile value, and average backwash time interval of each preset time period in the time period-backwash time interval mapping dataset. The maximum value of the lower percentile and the minimum value of the upper percentile for all preset time periods are calculated. A peak time threshold is defined as the minimum value of the upper percentile, and a trough time threshold is defined as the maximum value of the lower percentile. The defined peak and trough time thresholds are then compared with the corresponding average backwash time interval.

[0025] When the average backwash interval of a preset period is less than the low period threshold, the corresponding preset period is determined to be a low period; when the average backwash interval of a preset period is greater than or equal to the peak period threshold, the corresponding preset period is determined to be a peak period.

[0026] As a further aspect of this invention: a filter bed urgency assessment model is constructed using a random forest algorithm to output an urgency score and generate a filter bed backwashing urgency queue. Specifically, when the scheduling state coefficient PSC=1 and the current time period is a low period, a filter bed urgency assessment model is constructed using real-time collected filter bed operation data. The filter bed clogging risk characteristics selected by the filter bed urgency assessment model are: the clogging rate obtained based on the ratio of the filter bed inlet and outlet pressure difference to the initial pressure difference; the water quality index obtained based on the relative growth rate of turbidity of the filter bed effluent and inlet; and the operating time of the filter bed since the last backwash. A random forest classification model is trained using filter bed operation samples from historical operation and maintenance records, and the probability that each filter bed needs to be backwashed immediately is output as an urgency score. A filter bed backwashing urgency queue is generated in descending order of urgency scores.

[0027] As a further aspect of the present invention: after classifying filters in the filter backwashing urgency queue with urgency scores not lower than the preset segmentation threshold into a priority backwashing filter set, and the rest into a delayed backwashing filter set, based on a comparison of urgency scores with preset segmentation thresholds, the invention further includes:

[0028] The maximum number of filter tanks that can be backwashed simultaneously, as set in advance, is called the parallel backwashing number.

[0029] The filter combination that actually performs backwashing in this fine-tuning operation is determined based on the number of parallel backwashes, including:

[0030] If the number of filters in the priority backwash filter set is less than or equal to the number of parallel backwashes, then all filters in the priority backwash filter set will be selected for backwashing first. The remaining available slots will be filled by selecting filters from the set of filters that can be delayed backwashing in descending order of urgency score. If the number of filters in the priority backwash filter set is greater than the number of parallel backwashes, then filters equal to the number of parallel backwashes will be selected from the priority backwash filter set in descending order of urgency score and included in this backwashing combination. The priority backwash filters that are not selected will be marked as key targets to be backwashed.

[0031] Simultaneously, based on the real-time monitored disturbance tolerance margin value, the backwashing operation parameters of individual filters are dynamically adjusted: when the disturbance tolerance margin value is in the preset high margin range, the backwashing time of the filter is extended; when the disturbance tolerance margin value is in the preset middle margin range, the standard backwashing time of the filter is adopted; when the disturbance tolerance margin value is in the preset low margin range, the backwashing time of the filter is shortened and the backwashing intensity is reduced; if the water quality index is detected to be close to the safety threshold during the backwashing process, the backwashing operation of the current filter is terminated and switched to the standby filter.

[0032] As a further aspect of this invention: a set of candidate reconfiguration schemes is generated based on a set of finite operation steps. The incremental number of available flushing tanks and the valve position switching complexity of each candidate reconfiguration scheme are calculated. Specifically, when the topology reconfiguration mode is triggered, several reconfiguration schemes are enumerated within the limited operation steps to form a set of candidate reconfiguration schemes. For each candidate reconfiguration scheme, its impact on the pipeline connectivity is analyzed using a digital twin model, and the number of available flushing tanks that can be increased without changing the upper limit of parallel backwashing is calculated. At the same time, the valve position switching complexity required to execute the candidate reconfiguration scheme is calculated, which can be obtained by weighting and accumulating the number of valve opening and closing operations, the number of cross-zone operations, and the number of key valve actions. Candidate reconfiguration schemes that meet the conditions are selected based on a preset coverage threshold and the upper limit of valve position switching complexity. The preset coverage threshold refers to the number of filters that can serve priority backwashing within the number of parallel backwashing operations. The candidate reconfiguration scheme with the largest incremental number of available flushing tanks and the smallest valve position switching complexity is selected from the candidate reconfiguration schemes that meet the conditions as the optimal reconfiguration scheme, and the corresponding valve switching command is generated.

[0033] As a further aspect of this invention: Based on the time-expanded maximum flow model, and constrained by the upper limit of slot resources, the upper limit of disturbance budget, and the upper limit of parallel backwashing, the backwashing start and end slots of each filter in the priority backwashing filter set and the delayed backwashing filter set are arranged in stages to output the final scheduling scheme. Specifically, under the optimal reconstruction scheme, the backwashing execution time window is discretized into several equal-length slots according to the monitoring cycle, a time-expanded maximum flow model is constructed, and the upper limit of slot resources, the upper limit of disturbance budget, and the number of parallel backwashings are used as hard constraints. A staged scheduling strategy is adopted, that is, according to... The urgency scores are ranked from highest to lowest to assign backwashing start slots to the filters in the priority backwashing filter set. After the priority filters are scheduled, the filters that can be delayed backwashed are scheduled within their allowable window based on the remaining slot resources. If there are insufficient remaining slot resources, the backwashing start slots can be postponed in order of urgency from highest to lowest to obtain the backwashing start slots corresponding to each filter in the set of filters that can be delayed backwashed. After completing the phased scheduling strategy, the backwashing start slots and backwashing end slots of each filter are obtained, and the corresponding valve position switching instructions, the number of parallel slots, and the air volume and water volume schedules are generated to form the final scheduling scheme.

[0034] A smart management system for smart water management includes a data acquisition module for deploying smart sensors and edge computing gateways to collect multi-dimensional operational monitoring data such as flow rate, pressure, water quality, energy consumption, valve position, pump station status, and historical operation and maintenance records. This module performs time synchronization, anomaly detection, and data completion, and uploads the data to the cloud. A processing module calculates topology availability redundancy, bottleneck capacity redundancy, and disturbance tolerance redundancy, generates scheduling state coefficients, and trains a random forest model when the scheduling state is transitional, outputting filter bed urgency scores and filter bed flushing urgency queues. A scheduling module is used for... The system divides the scheduling process into peak and off-peak periods. During peak periods, it generates candidate topology reconfiguration schemes, evaluates the incremental number of available backwash tanks and the complexity of valve position switching, and selects the optimal reconfiguration scheme. It constructs a time-expanded maximum flow model, schedules the backwashing start and end tanks for each filter in stages, and generates backwashing scheduling schemes and valve position switching instructions. The execution module is used to quickly simulate and verify the final scheduling scheme output by the scheduling module on the digital twin platform. After verification, it sends the scheme to the equipment control system to implement the backwashing schedule and updates the filter status and model parameters after backwashing is completed.

[0035] The beneficial effects of this invention are:

[0036] (1) This invention constructs an urgency model by using the clogging rate, water quality index and running time of each filter to accurately divide the priority backwash set and the delayable backwash set; combined with topology availability, bottleneck capacity, disturbance tolerance margin assessment and time segment division, it dynamically adjusts the number and duration of parallel backwashes to avoid the risk of peak water intake; topology reconstruction and time unfolding maximum flow model optimize resources and timing, digital twin verification and closed-loop update model parameters improve scheduling reliability and water intake stability, reduce the situation of excessive or insufficient backwashing of filters, reduce equipment wear and tear, and solve the problem of allocation blind spots caused by the lack of fine analysis of individual filter cells in smart water backwashing. Attached Figure Description

[0037] The invention will now be further described with reference to the accompanying drawings.

[0038] Figure 1 This is a schematic diagram illustrating the implementation process of an intelligent management method for smart water affairs according to the present invention;

[0039] Figure 2 This is a schematic diagram of a deep-bed filter structure for an intelligent management system applied to smart water affairs, according to the present invention.

[0040] Figure 3 This is a schematic diagram of the overall architecture of an intelligent management system applied to smart water affairs according to the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] Example: Please refer to Figure 1 As shown, this invention is an intelligent management method applied to smart water affairs, comprising the following steps:

[0043] Step 1: Collect multi-dimensional operation monitoring data of wastewater treatment, analyze three criteria: topology availability, bottleneck capacity, and disturbance tolerance, and obtain the redundancy value of each criterion; generate scheduling status coefficients based on the redundancy value. If all redundancies are sufficient, maintain the original scheduling scheme; otherwise, trigger the subsequent improvement process.

[0044] Please see Figure 2 The image shows the structure of a deep-bed filter in the digital twin interface of a wastewater treatment plant.

[0045] Multi-dimensional operational monitoring data of various structures in the wastewater treatment plant are collected in conjunction with the plant's process flow. These structures include the inlet, coarse screen, aerated grit chamber, flocculation tank, primary sedimentation tank, biological treatment tank, secondary sedimentation tank, high-efficiency sedimentation tank, ozone contact tank, and effluent outlet. Smart sensors are deployed in key units and pipelines, and full-process visual perception is achieved through SCADA and IoT platforms. The data collection content and methods are as follows:

[0046] Flow meters, water level gauges, and online water quality analyzers are installed in the aforementioned key units to collect parameters such as influent and effluent flow rate, liquid level, turbidity, COD, ammonia nitrogen, total phosphorus, pH, and dissolved oxygen in real time; the data is time-marked based on the time axis to achieve time-series tracking and obtain the operation monitoring data of each process unit;

[0047] Sludge flow meters, moisture content sensors, and torque sensors are installed in sludge pipelines to monitor sludge return flow, dewatering machine status, and sludge pump load. Pressure sensors, air flow meters, and dissolved oxygen probes are installed in aeration pipelines (fans and pipes) to collect data on air supply, aeration pressure, and dissolved oxygen in the aeration tank. Flow meters, valve position sensors, and concentration monitors are installed in chemical dosing pipelines (coagulants, flocculants, disinfectants) to record dosage, valve opening, and chemical residue. Water treatment pipelines (filter backwash sections) use differential pressure meters, online turbidity meters, and power meters to collect influent and effluent flow rates, pressures, temperatures, differential pressures, and the current, voltage, and start-up frequency of the backwash pump / influent pump. This yields specialized pipeline operation monitoring data, including pipeline pressure in each filter section, influent and effluent pressure differentials, influent and effluent turbidity in each filter, effluent turbidity, and equipment energy consumption.

[0048] Real-time reading of the valve opening and closing status, pump operating frequency, and switching status of each device in the entire plant to construct a pipeline topology view; synchronously collecting historical operation and maintenance records, including the last backwash time / duration / intensity of the filter and inspection records, as the basis for model training and status assessment; statistical data on the upper limit of tank resources by time and tank, including the upper limit of aeration air volume, the upper limit of backwash water volume, and the upper limit of sludge discharge capacity, and forming a tank resource upper limit table;

[0049] The sampling period for the multi-dimensional operation monitoring data obtained above is set according to the dynamic characteristics of the sensors, such as water quality parameters at the minute level, flow and pressure at the second level, and energy consumption data at the 10-second level. The raw data is synchronized in time through the edge computing gateway, and noise data is removed by anomaly detection algorithms (such as Z-score and isolated forest). Missing data is filled in by neighborhood averaging or time series interpolation. Finally, the data is uploaded to the cloud platform and the digital twin model is updated synchronously.

[0050] Furthermore, the wastewater treatment plant process flow is modeled as a weighted directed graph, denoted as the wastewater treatment topology graph G=(V,E), where nodes V represent various structural units and equipment, and edges E represent various pipes, valves, and pump stations. The topology availability criterion A is calculated through the following steps. top and its corresponding topological available redundancy value r top Among them, topology availability criterion A top This is used to measure whether the system can maintain connectivity and water intake capacity under a given backwashing task; the specific calculation is as follows:

[0051] Topology Availability Criterion A top Based on real-time valve positions and pump status, a currently available wastewater treatment topology graph Gt is generated; when some filters enter backwashing mode, related pipelines or pump stations will temporarily cease service; these nodes and edges are removed from the residual network; depth-first search (DFS) is used to detect the connectivity between important nodes, such as the contact disinfection tank outlet and the receiving water body; if a break exists, A... top =0; otherwise, continue calculating;

[0052] The maximum flow algorithm, such as Edmonds-Karp, is used to calculate the maximum outflow capacity Fmax from the influent source to the receiving water body in the residual network. This capacity is then compared with the current backwashing demand D, and the available topology redundancy value rtop is defined as follows: It should be noted that the current backwashing demand D represents the total water demand per unit time for all filters planned for backwashing in the current period. Its calculation basis is the backwashing design flow rate of a single filter preset by the engineer, which is dynamically determined in combination with the actual number of filters participating in backwashing in the current period. It refers to the total backwashing influent required per unit time for all filters planned to perform backwashing operations within a specific scheduling period, i.e., the time window in which it is necessary to determine whether to trigger the improvement process or execute backwashing scheduling, such as one time slot, including the filters that need to be started in the current period in the priority backwashing filter set and the filters that are planned to be included in the current period in the set of filters that can be delayed backwashing. The unit is consistent with the maximum effluent capacity of the currently available topology and is the core water demand indicator for evaluating whether the system can support backwashing operations.

[0053] When r top The topology is considered usable when the value is greater than or equal to 0, and the usable redundancy value is r. top Conversely, it indicates that the topology does not meet the requirements.

[0054] The bottleneck capacity criterion A is calculated using the following steps. cap and its corresponding bottleneck capacity margin value r cap Among them, bottleneck capacity criterion A cap Used to measure the remaining capacity of critical equipment or pipe sections to avoid local overload; based on the pipe network topology in the current available wastewater treatment topology map Gt and the impact analysis of previous backwashing, identify pump stations, pipe sections or valves that are likely to become bottlenecks during the water intake process; for example, in an area with water intake from two main pipelines, if closing one pipeline would cause the other pipeline to overload, then that pipeline is the bottleneck.

[0055] Based on the collected flow rate, pressure difference, and pump curve model, calculate the actual flow rate of the i-th bottleneck component. With rated capacity proportion In the formula, i represents the bottleneck component identifier; the bottleneck capacity margin is defined as... That is, the value obtained by subtracting the maximum load utilization rate of all bottleneck components from 1; if the bottleneck capacity margin value r cap If ≥0, the capacity margin is considered sufficient;

[0056] Historical analysis shows that when the bottleneck component load exceeds 90%, the system is prone to voltage drop and service interruption. Therefore, the bottleneck capacity margin threshold... It can be set to 0.1;

[0057] The disturbance tolerance criterion A is calculated using the following steps. dist and disturbance tolerance margin value r dist The system's tolerance to water quality fluctuations was assessed; among which, disturbance tolerance criterion A... distUsed to assess the system's tolerance to fluctuations in water quality and quantity, reflecting the system's ability to withstand additional disturbances when backwashing is scheduled:

[0058] A water quality disturbance model was constructed by using a hybrid model of mechanistic model (filter kinetic equation and mass balance equation) and long short-term memory network (LSTM) to predict the impact of backwashing on the turbidity and disinfection index of the treated water. The model training data consisted of monitoring records during historical backwashing periods (including backwashing intensity, duration, initial water quality, and downstream water quality changes). The model was evaluated using mean absolute error (MAE) and root mean square error (RMSE), and hyperparameters such as hidden layer size and learning rate were adjusted through cross-validation.

[0059] Water quality disturbance model predicts the peak turbidity T of the effluent within a certain time after backwashing. peak and the short-term decrease in disinfection indicators C drop In this embodiment, a certain time period can be within 5 minutes, which is consistent with the water quality standard threshold, including the turbidity standard threshold. and disinfection standard threshold In comparison, for example, if the turbidity index does not exceed the permissible limit, the limit of the controlled particulate matter index, such as 10 mg / L, is taken; if the disinfection index meets the permissible / design requirements, the lowest disinfection capacity is taken to represent the disinfection capacity. Among these, if UV is used for disinfection indicators, the disinfection indicators represent the UV dose; if chlorination is used, the disinfection indicators represent the CT value.

[0060] Turbidity tolerance is defined as Disinfection tolerance is In the formula, the minimum of the two values ​​is taken as the disturbance tolerance margin value. ;

[0061] When the disturbance tolerance A value less than 0 indicates that water quality fluctuations exceed the standard, and the disturbance tolerance margin threshold is exceeded. It can be set according to regulatory requirements, such as... ;

[0062] Furthermore, the obtained three types of redundancy indicators, namely topological availability redundancy values, are... Bottleneck capacity margin value Disturbance tolerance margin value As input to the scheduling status coefficient model, corresponding preset upper and lower bound thresholds will be introduced to divide the scheduling status into three situations: normal, transitional, and critical. The establishment and implementation process of the scheduling status coefficient model is as follows: Define the scheduling status coefficient. The mathematical expression is:

[0063] In the formula, , , Preset upper bound thresholds for various redundancy indicators to ensure sufficient redundancy in the system; , , To represent the preset lower bound thresholds corresponding to various redundancy indices that are close to instability;

[0064] It should be noted that the above-mentioned preset upper bound thresholds... , , For safe / compliant samples, the point with a false positive rate ≤5% and maximum F1 score is selected using ROC or PR curves; if there is no label, the 10th percentile of the compliance period distribution is used as a conservative upper bound; various preset lower bound thresholds are also used. , , For critical / exceeding-standard samples, select the maximum threshold with a recall rate ≥ 95%; when there are no labels, use the 90th percentile of the distribution during the exceeding period (converted after taking the absolute value of the negative remainder) or use Bayesian discrimination to take the intersection of the two distributions as the preset lower bound threshold.

[0065] Based on scheduling status coefficient The output results are used to implement different scheduling strategies:

[0066] When PSC=0, it means that there is sufficient redundancy and the scheduling status is normal, and the current scheduling plan should be maintained.

[0067] When PSC=1, it indicates that the scheduling status is in transition and has entered fine-tuning mode;

[0068] When PSC=2, it indicates that the redundancy is severely insufficient and the scheduling status is critical, requiring the topology reconfiguration mode to be entered.

[0069] Step 2: Based on the output scheduling status data and the characteristics of peak and trough periods, generate an urgent backwashing queue for the filter beds. Based on the clogging risk characteristic value of each filter bed, construct a filter bed urgency assessment model and output an urgency score. Based on the comparison between the urgency score and the segmentation threshold, divide the urgent backwashing queue for the filter beds into two sets: a priority backwashing filter set and a delayed backwashing set. Based on the corresponding set type, combined with the redundancy value and the fluctuation intensity, adaptively adjust the number of parallel backwashing filters.

[0070] This step is designed for the transition state when PSC=1. Based on the characteristics of peak and off-peak periods and the degree of clogging of each filter, an urgent backwashing queue is generated and the number of parallel backwashes is adaptively adjusted.

[0071] First, using the historical backwashing time interval data from the multi-dimensional operational monitoring data collected in Step 1, and taking the filter backwashing time interval as the core indicator, the operating periods are divided using statistical analysis methods to clarify the differentiated treatment strategies for transitional states. Specifically, the following steps are taken: Backwashing event records for the past 12 months (sample size ≥ 360 groups, meeting statistical significance requirements) are extracted from the multi-dimensional operational monitoring data collected in Step 1, and the time interval for each backwash is calculated. Each backwashing time interval is then associated with a preset time period to form a time period-backwashing time interval mapping dataset. The preset time period is defined as m preset time periods divided into 24 hours, for example, 6 preset time periods: 0:00-4:00, 4:00-8:00, 8:00-12:00, 12:00-16:00, 16:00-20:00, 20:00-24:00.

[0072] A period segmentation model is constructed using percentile statistics to ensure the objectivity and verifiability of the segmentation criteria: For each preset period in the time-backwashing time interval mapping dataset, the lower percentile value, higher percentile value, and average backwashing time interval of the filter are calculated. The percentile calculation adopts linear interpolation. The specific percentile ratios of the lower and higher percentile values ​​can be set according to the water demand fluctuation characteristics of the project location. The initial values ​​can be 25th percentile and 75th percentile, respectively.

[0073] Calculate the maximum value of the lower percentile and the minimum value of the upper percentile for all preset time periods. Define the peak time threshold as the minimum value of the upper percentile and the trough time threshold as the maximum value of the lower percentile. Then, compare the defined peak and trough time thresholds with the corresponding average backwash time intervals.

[0074] When the average backwash interval of a preset period is less than the threshold of the low period, the corresponding preset period is determined to be the low period. The backwash interval is longer in the preset period, indicating that the water demand is low and the system redundancy can support local fine-tuning.

[0075] When the average backwashing time interval of a certain preset period is greater than or equal to the peak period threshold, the corresponding preset period is determined to be a peak period. During this period, the backwashing time interval is shorter, the blockage is faster, the water demand is higher, and the continuity of water intake should be prioritized.

[0076] The historical inflow and outflow data collected in step one are monitored in real time by the flow meter to verify the division results. If a preset time period is determined to be a peak period, but the average inflow of the corresponding time period is less than or equal to 60% of the historical average inflow, then the time period is corrected to a low period to ensure that the division results are consistent with the actual water demand.

[0077] Based on the above quantitative division results, the following diversion logic is executed for the transition state of PSC=1 to ensure the accuracy of the handling strategy:

[0078] If PSC=1 and the current preset time period is a low-temperature period, start the fine-tuning process to balance the backwashing needs of the filter and the safety of the system influent through local optimization.

[0079] If PSC=1 and the current preset time period is a peak period, terminate the fine-tuning process and directly mark it as entering the topology reconstruction process;

[0080] For scenarios where PSC=1 and is in a low-period period, based on the data collected by the monitoring equipment deployed in step one, a filter urgency assessment model is constructed to accurately identify filters that require priority backwashing. The specific steps are as follows:

[0081] The goal of the filter urgency assessment model is to assess the clogging risk of individual filters, rather than the overall redundancy. Therefore, the selected clogging risk characteristic values ​​for each filter include:

[0082] The clogging rate, defined as the ratio of the pressure difference between the filter inlet and outlet to the initial pressure difference, reflects the degree of increase in filter media resistance; the water quality index, defined by the relative growth rate of turbidity in the filter effluent and influent, assesses the degree of decline in filtration efficiency; and the continuous operating time of the filter since the last backwash is defined as the operating time, reflecting the cumulative load on the filter media.

[0083] Using these features, a filter urgency assessment model was constructed. Filter operation data extracted from historical operation and maintenance records was used to filter samples that required immediate backwashing (marked as 1, with the criteria being excessive turbidity in the filter effluent or a pressure difference exceeding the design threshold, the specific design threshold being set according to the wastewater treatment plant effluent standards). Samples that did not require immediate backwashing were marked as 0, with the criteria being that both turbidity in the filter effluent and the pressure difference met the design requirements. These samples were used as training samples, with each training sample containing the above three features and their corresponding labels, and a sample size of ≥500 groups. The samples were divided into training and testing sets in a 7:3 ratio. The training set was used to construct a filter urgency assessment model based on random forest. Model hyperparameters such as the number of trees and maximum depth were automatically selected through cross-validation, and optimized through 5-fold cross-validation. The objective function was to maximize the F1 score of the testing set.

[0084] After training, the current filter clogging rate, water quality index and running time are input into the trained filter urgency assessment model, and the probability that the filter needs to be backwashed immediately is output as the urgency score, which ranges from [0,1]. The closer it is to 1, the higher the urgency of the filter needing to be backwashed.

[0085] Based on the urgency scores on the training set, an ROC curve was plotted, and the urgency score corresponding to the point where the true positive rate and false positive rate reached a preset balance requirement was selected as the segmentation threshold. All filter beds were arranged in descending order of urgency score to form an urgency queue. The urgency queue was then divided into two categories according to the segmentation threshold:

[0086] Filters with a stress score greater than or equal to the segmentation threshold are classified as priority backwashing filters and form a priority backwashing filter set. Backwashing should be prioritized in this fine-tuning to avoid further deterioration of filter performance.

[0087] Conversely, if the backwashing is not performed, it will be classified as a delayed backwashing filter and form a delayed backwashing filter set. Backwashing will not be scheduled for the time being. The urgency score will be reassessed during the next low-pressure period before it is determined whether to backwash.

[0088] The threshold that minimizes the difference between the true positive rate and the false positive rate was determined using ROC curves and was used as the urgency threshold. Based on the urgency threshold, the filter sets that must be backwashed first and those that can be delayed were divided, as shown below:

[0089] During periods of low PSC=1, to avoid significant fluctuations in system redundancy, the number of parallel backwashing filters is kept constant. This number is set as a preset proportion of the maximum number of filters that can be safely backwashed in parallel within the pipeline topology of Step 1, reserving some redundancy to handle sudden disturbances. The specific proportion is set according to system redundancy design requirements. Local optimization is achieved only by adjusting the backwashing sequence and the duration of a single backwash. The specific steps are as follows: Based on the constraints of the urgent queue and the number of parallel backwashing filters, the backwashing filter combination for this fine-tuning is determined according to the following rules:

[0090] If the number of filters in the priority backwash filter set is less than or equal to the number of parallel backwashes, then all filters in the priority backwash filter set are selected for backwashing first. The remaining parallel slots are selected from the set of filters that can be delayed backwashed in descending order of urgency score, ensuring that the number of parallel backwashes is exactly the same as the number of parallel backwashes.

[0091] If the number of filters in the priority backwash filter set is greater than the number of parallel backwashes, then the first n filters in the priority backwash filter set, ranked from largest to smallest according to their urgency score, are selected for backwashing. The priority backwash filters that are not selected are marked as key objects to be backwashed, and their urgency score and current status are recorded for fine-tuning priority scheduling during the next off-peak period.

[0092] Combined with the disturbance tolerance margin value calculated in real time in step one Dynamically adjust the backwashing time of individual filters to ensure that water quality indicators meet standards during the backwashing process, including:

[0093] Extract the perturbation tolerance margin value at the current time step one. and the corresponding preset upper bound threshold Preset lower threshold ;

[0094] When the disturbance tolerance margin value is within the preset high margin range, that is, when the disturbance tolerance margin value is within the preset high margin range... Extend the backwashing time to In the formula, k1 is a preset time extension coefficient, ranging from 1.05 to 1.15, which is set according to the filter media type to improve the filter media rinsing effect. This is the standard backwashing time for the filter; where Δγ is the high redundancy range offset preset in step one, and its value is... and 10% of the difference The standard backwashing time is extracted from the historical operation and maintenance records of the filter in step one, which is the average duration of the historical normal backwashing of the filter.

[0095] When the disturbance tolerance margin value is within the preset intermediate margin range, that is, when the disturbance tolerance margin value is within the preset intermediate margin range... Perform backwashing according to the standard backwashing duration without adjustment;

[0096] When the disturbance tolerance margin value is within the preset low margin range, i.e., the disturbance tolerance margin value... Shorten backwash time to In the formula The preset duration reduction coefficient ranges from 0.7 to 0.8. The specific value is set according to the water quality warning threshold to avoid water quality fluctuations during backwashing that exceed the water quality standard thresholds such as turbidity and disinfection indicators set in step one.

[0097] After adjusting the backwashing time, monitor the effluent water quality of the filter in real time. If the water quality approaches the warning threshold for exceeding the standard, immediately terminate the backwashing of the filter and switch to the standby filter to ensure the safety of the effluent water quality.

[0098] Step 3: Based on the diversion decision results of Step 2, reconstruct the connection relationships, resource allocation, and timing: Generate a set of candidate reconstruction schemes according to the system-level topology reconstruction strategy, calculate the incremental number of available backwash tanks and valve position switching complexity of each candidate scheme, and select the optimal reconstruction scheme that meets the preset coverage threshold and whose valve position switching complexity does not exceed the preset upper limit; and expand the maximum flow model based on time, with the upper limit of tank resources, the upper limit of disturbance budget, and the upper limit of parallel backwashing as constraints, arrange the backwashing start tank and backwashing end tank of each filter in the priority backwashing filter set and the delayed backwashing filter set in stages, and output the final scheduling scheme;

[0099] Based on the current wastewater treatment topology Based on the list of controllable components (valve opening and closing, pump / fan switching, backwash inlet / return path assignment) and the project's existing safety templates / historical feasible configurations, a set of candidate reconfiguration schemes {Ck} is generated by setting a finite operating step size K. Each candidate reconfiguration scheme Ck consists of only a finite combination of the above primitives. Each candidate reconfiguration scheme expands the backwash water supply path by switching specific valve positions, activating standby pumps, or using temporary water sources. For example, if some filters have already entered the backwashing state, activating the standby inlet pump or switching valves can provide additional backwashing tanks.

[0100] For each candidate reconfiguration scheme Ck, the following index is calculated: the increase in the number of backwash tanks that each filter can use under the candidate reconfiguration scheme is recorded as the increment of the number of available backwash tanks; the larger this value is, the better the scheme supports the set of filters that must be backwashed first.

[0101] The total number of valve switching actions required for candidate reconfiguration schemes is counted. The switching actions of valves switching across process units and key valves are included in the total number of actions. The valve position switching complexity is directly quantified by the cumulative number of actions, i.e., by weighted accumulation with preset weights.

[0102] The candidate reconstruction scheme set {Ck} is filtered and sorted based on the incremental number of available flushing tanks and the valve position switching complexity, and the final result is recorded as the optimal reconstruction scheme. The specific filtering and sorting process is as follows: for each candidate reconstruction scheme Ck, the incremental number of available flushing tanks and the valve position switching complexity are calculated.

[0103] If the increase in the number of available flushing tanks is ≤0 or the valve position switching complexity exceeds the preset valve position switching complexity upper limit, the corresponding candidate reconstruction scheme will be eliminated.

[0104] For the retained candidate reconfiguration schemes, their coverage of the filter set that must be backwashed first within the number of parallel backwashes is checked, and those below the preset coverage threshold are also eliminated; the sorting is done in descending order of the number of available backwashing tanks, in ascending order of valve position switching complexity, and in descending order of coverage; when in parallel, the schemes with fewer cross-zone actions, lower critical valve weights, and larger increments of the backwash set that can be delayed are given priority, and the unique optimal reconfiguration scheme is determined.

[0105] It should be noted that the preset upper limit of the above project refers to the upper limit of valve position switching complexity allowed for a single candidate reconfiguration scheme, which is obtained by accumulating the total number of actions + the number of cross-zone actions + the number of key valve actions. The number of parallel backwashings refers to the maximum number of backwashing filters allowed to be performed simultaneously per time slot; this value is determined by step two under the current preset time period, based on... The resource capacity timetable is calculated accordingly; a preset coverage threshold is used as the criterion for determining whether the coverage is qualified; the preset coverage threshold refers to the number of filters that can serve the priority backwashing within the number of parallel backwashing operations; the threshold is determined based on historical samples and digital twin verification to ensure... and The minimum feasible coverage shall not be lower than its corresponding preset lower limit, and the coverage shall be rolled over seasonally.

[0106] Based on the obtained optimal reconstruction scheme, the task timing is optimized using a time-expanded maximum flow model. This model arranges the sets of filter tanks that must be backwashed first and the sets that can be delayed in step two on the time axis, ensuring not only the constraints but also minimizing valve position switching and delay costs. Key implementation points are as follows:

[0107] The backwash execution window is discretized into equal-length tanks according to the monitoring cycle of step one. The time discretization unit; where T represents the total number of slots into which the backwash execution window is divided, and the corresponding pipeline sensor monitoring data, including the aeration air volume, backwash water volume, and sludge discharge capacity limit of each time slot, are obtained based on the pipeline sensors to generate a slot resource limit table. ;in, Indicates the first The upper limit of aeration air volume for each slot. Indicates the first The maximum backwash water volume for each tank. Indicates the first The maximum sludge removal capacity of each tank;

[0108] And slot resource limit table Disruption budget cap The upper limit of the parallel backwashing in step two is used as the boundary constraint parameter;

[0109] Each filter will be Inherit the allow starting window from step two Continuous slot count Based on the water quality disturbance model (mechanism-LSTM mixing model) from step one, its position in the backwash initiation tank was determined. Time for subsequent slots to Discrete disturbance usage of slot and unit resources Parameters for filter backwashing are generated;

[0110] Under the optimal reconstruction scheme, this section will map the set of filter tanks that must be backwashed first and the set of backwashable tanks to discrete time slots. The arrangement will be completed with the above slot resource limit, disturbance budget limit and parallel backwashing quantity as hard constraints, and the backwashing start slot and backwashing end slot of each filter tank will be output.

[0111] Based on the optimal reconstruction scheme, a time-unfolded network is constructed, and an existing maximum flow model is used to build a time-unfolded maximum flow model. This enables phased scheduling, ensuring that resources are not overloaded and disturbances do not exceed limits, as detailed below:

[0112] Phase A: Based on the urgency score from Step 2, from highest to lowest, find the earliest and continuously occupyable number of slots within the window for each filter in the priority backwash filter set. The starting position for backwashing is determined; once determined, the corresponding position is locked, and the aeration capacity, backwash water capacity, and number of parallel backwashes for that position are deducted; if a filter in the filter set is to be backwashed first... If there is no feasible starting point within its window, execute the following deterministic adjustment sequence item by item in order, each item must meet the upper limit of equipment capacity and disturbance budget given in step one: a. Capacity fine-tuning: upper limit of aeration air volume for conflicting tanks. Or the upper limit of backwash water volume a) Increase by no more than 5% of the equipment's rated capacity and not exceeding the rated upper limit; b) Window extension: within the allowed starting window. Under the premise of, the backwash starting slots of the filter tanks that conflict with the delayed backwash set will be uniformly moved one slot to the right and rechecked; c. The only filter tanks that do not have a delayed backwash set scheduled in this round are those that conflict with this filter tank, which will be removed from this window and recorded in the candidate list for the next low period; after executing a→b→c, the earliest feasible starting point for this filter tank will be searched again and recorded as the backwash starting slot until all filters in the priority backwash filter set have been scheduled;

[0113] Phase B: Under the premise that the results of Phase A remain unchanged, based on the urgency score from high to low, the earliest feasible slot is prioritized for each filter in the delayed backwash set; if there are multiple feasible starting points, the starting point that does not change the predetermined valve and pump state of the optimal reconfiguration scheme is selected first, followed by the starting point in the same slot as the arranged starting point, and finally the earliest starting point is selected as the backwash starting slot; each placement immediately deducts the aeration air volume capacity, backwash water volume capacity and the number of parallel backwashes for the corresponding slot until there are no feasible starting points.

[0114] The implementation process of Phase A and Phase B is as follows: first, place the priority backwash filter set, and then place the delay backwash set. The earliest feasible tank position is given priority. After each placement, the corresponding capacity is deducted. The final scheduling scheme is obtained through the above-mentioned Phase A and Phase B, which includes the backwash start / end time of each filter and the corresponding specific tank position and valve position switching instructions.

[0115] Step 4: Quickly verify the output scheduling scheme from Step 3 on the digital twin platform. After verification, send it to the equipment control system. During execution, the minimum dwell time constraint shall be observed. If the redundancy turns negative in real time, the parallel number shall be automatically downgraded and the queue shall be reordered. After backwashing is completed, update the filter clogging status, the set of available backwash tanks and the water quality disturbance model parameters to form an adaptive control mechanism for cross-wheel closed loop.

[0116] Read the scheduling scheme output in step three under the optimal reconstruction scheme, which includes the start slot to backwash end slot of each filter, the number of parallel operations for each slot, the air volume capacity time schedule, the backwash water volume capacity time schedule, the sludge discharge capacity time schedule, the disturbance budget curve, and the valve and pump station status sequence. Perform minute-level simulation on the digital twin platform: verify and record all time slots one by one, ensuring that the air volume usage in any time slot does not exceed the air volume capacity time schedule, the backwash water usage does not exceed the backwash water volume capacity time schedule, the sludge discharge capacity usage does not exceed the sludge discharge capacity time schedule, the disturbance occupancy does not exceed the disturbance budget curve, the number of parallel operations does not exceed the number of parallel backwashes, and the topology availability margin, bottleneck capacity margin, and disturbance tolerance margin are all not lower than their respective preset lower bound thresholds. After all passes, generate valve position opening / closing commands, pump station start / stop and frequency commands, and filter start and end commands, and send them to the PLC or DCS with timestamps for archiving.

[0117] Continuously collect water quality, water quantity, water pressure, dissolved oxygen, energy consumption, and valve / pump status; when the topology availability margin, bottleneck capacity margin, or disturbance tolerance margin of any time slot is less than one-tenth of its preset lower limit threshold, execute the following in a fixed order: reduce the number of parallel backwashes in the current slot and re-verify the above four types of upper limits until they are met; the affected delayed backwash set is shifted one slot to the right first, according to the earliest feasible slot; for the filter currently in operation, according to the rules in step two, multiply the number of continuous slots by the duration reduction coefficient and reduce the backwash intensity by one level until the disturbance occupancy does not exceed the disturbance budget curve; if a path failure or pump station failure is detected, switch to the backup path and backup pump station marked in the optimal reconfiguration scheme, and immediately re-verify the four types of upper limits and the three types of margin; if not met, stop the start of delayed backwash sets that have not yet started in that slot and shift one slot to the right; each of the above adjustments must be verified again through digital twin before it takes effect;

[0118] After each filter is completed, the following data are written back: differential pressure recovery rate, improvement in effluent turbidity, and runtime reset; the clogging rate and water quality index are updated as inputs for the next round of urgency assessment in step two; the impulse response function is updated online based on the monitored disturbance response, and the subsequent discrete disturbance usage is corrected; the safety margin distribution of each time slot in this cycle is summarized: when more than 80% of the time slots simultaneously meet the four types of upper limit redundancy greater than 1 to 20%, the number of parallel backwashes in the next cycle is increased by one, or the air volume capacity schedule or backwash water volume capacity schedule is slightly increased without exceeding the equipment rating; when the proportion of slots approaching the lower threshold exceeds 20% for two consecutive times, the number of parallel backwashes is tightened accordingly or the disturbance budget curve is lowered.

[0119] At the end of the cycle, an assessment report is generated (completion rate, average redundancy, water quality fluctuations, energy consumption), triggering: retraining of the urgency assessment model (using the original data caliber and feature definitions), online calibration of the water quality disturbance model, and rolling recalibration of preset upper and lower bound thresholds and capacity schedules; the updated parameters and models are only effective in the next cycle, and are enabled only upon passing digital twin playback verification. Thus, step four achieves the safety verification, online assurance, and cross-cycle closed-loop of the timing sequence in step three, bringing the scheduling status coefficient back to the safe range.

[0120] Please see Figure 3 A smart management system for smart water management includes a data acquisition module, a processing module, a scheduling module, and an execution module.

[0121] The data acquisition module deploys smart sensors and edge computing gateways to collect multi-dimensional operational monitoring data, including flow rate, pressure, water quality, energy consumption, valve position, pump station status, and historical maintenance records. It performs time synchronization, anomaly detection, and data completion, and uploads the data to the cloud. The processing module calculates topology availability redundancy, bottleneck capacity redundancy, and disturbance tolerance redundancy, generates scheduling state coefficients, and trains a random forest model when the scheduling state is transitional, outputting filter urgency scores and filter flushing urgency queues. The scheduling module divides the scheduling state into peak and off-peak periods. During peak periods, it generates candidate topology reconfiguration schemes, evaluates the incremental increase in the number of available flushing tanks and the complexity of valve position switching, selects the optimal reconfiguration scheme, constructs a time-expanded maximum flow model, schedules the backwashing start and end slots for each filter in stages, and generates backwashing scheduling schemes and valve position switching instructions. The execution module performs rapid simulation verification of the final scheduling scheme output by the scheduling module on the digital twin platform. After verification, it sends the scheme to the equipment control system for backwashing scheduling and updates the filter status and model parameters after backwashing is completed.

[0122] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0123] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0124] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive 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 implementation should not be considered beyond the scope of this application.

[0125] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0126] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included 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.

[0127] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent management method applied to smart water affairs, characterized in that, Includes the following steps: Collect multi-dimensional operation monitoring data of wastewater treatment, analyze three criteria: topology availability, bottleneck capacity, and disturbance tolerance, and calculate their corresponding redundancy values; generate scheduling status coefficients based on the redundancy values ​​of each criterion; if all redundancies are sufficient, maintain the original scheduling scheme; otherwise, trigger the improvement process. After the improved process is triggered based on the scheduling status coefficient, the filter urgency assessment model is constructed using a random forest algorithm, taking into account peak and off-peak periods and the clogging risk characteristic value of each filter as input. The model outputs an urgency score and generates a filter flushing urgency queue. Based on the comparison between the urgency score and the preset segmentation threshold, the filters in the filter flushing urgency queue with an urgency score not lower than the preset segmentation threshold are divided into a priority backwashing filter set, and the rest are divided into a delayed backwashing filter set. A set of candidate reconfiguration schemes is generated based on a set of finite operation steps. The incremental number of available backwash tanks and the valve position switching complexity of each candidate reconfiguration scheme are calculated. The optimal reconfiguration scheme that meets the preset coverage threshold and whose valve position switching complexity does not exceed the preset upper limit is selected. Based on the optimal reconfiguration scheme, the maximum flow model is expanded in time. Under the constraints of the upper limit of tank resources, the upper limit of disturbance budget and the upper limit of parallel backwashing, the backwashing start tank and backwashing end tank of each filter in the priority backwashing filter set and the delayed backwashing filter set are arranged in stages. The final scheduling scheme containing the backwashing scheduling scheme and the corresponding valve position switching command is output. The final output scheduling plan is quickly verified on the digital twin platform. After the verification is passed, it is sent to the equipment control system for execution, and the filter status and parameters of each model are updated after the backwashing is completed. This includes generating scheduling status coefficients based on the redundancy values ​​of various criteria, including: constructing a scheduling status coefficient model, the expression of which is: In the formula This is the scheduling status coefficient. , , Preset upper bound thresholds for various redundancy indicators; , , To represent the preset lower bound thresholds corresponding to various redundancy indices that are close to instability; For topology availability margin value, Bottleneck capacity margin value, This represents the disturbance tolerance margin value. And based on the above scheduling status coefficients The value of is used to output the scheduling status according to the following logic: When the scheduling status coefficient PSC=0, it means that the system has sufficient redundancy and the existing schedule should be maintained. When the scheduling status coefficient PSC=2, it indicates that the system redundancy is seriously insufficient and it needs to enter the topology reconfiguration mode; when the scheduling status coefficient PSC=1, it indicates that it is in a transitional state and it needs to enter the fine-tuning mode. The final scheduling scheme is output through the following process: The backwash execution time window is discretized into several equal-length slots according to the monitoring cycle. A time-expanded maximum flow model is constructed, with the upper limit of slot resources, the upper limit of disturbance budget, and the number of parallel backwashes as hard constraints. A phased scheduling strategy is adopted, that is, backwash starting slots are allocated to the filters in the priority backwash filter set in descending order of urgency score. After the priority filters are scheduled, the filters that can be delayed backwashed are scheduled within their allowable window according to the remaining slot resources. If the remaining slot resources are insufficient, the backwash starting slots are postponed according to urgency to obtain the backwash starting slots corresponding to each filter in the set of filters that can be delayed backwashed. After the phased scheduling strategy is completed, the backwash starting slots and backwash ending slots of each filter are obtained, and the corresponding valve position switching instructions, the number of parallel slots, and the air volume and water volume time schedule are generated to form the final scheduling scheme.

2. The intelligent management method for smart water affairs according to claim 1, characterized in that, Multi-dimensional operational monitoring data includes: real-time operational data acquired by smart sensors deployed in various structural units and key pipelines of the wastewater treatment plant process, namely, influent and effluent flow rates and pipe flow rates collected by flow meters; pipe pressure and influent and effluent pressure differences of each filter collected by differential pressure gauges; influent and effluent turbidity and effluent turbidity of each filter collected by online turbidity meters; equipment energy consumption collected by power meters; valve opening and closing status, pump operating frequency, and switching status of each device collected by valve and pump station status monitors; historical operation and maintenance information of filter backwashing collected synchronously by the water IoT edge gateway; and resource limit data statistically analyzed by time slot, including aeration air volume limit, backwash water volume limit, and sludge discharge capacity limit.

3. The intelligent management method for smart water affairs according to claim 2, characterized in that, The analysis considers three criteria: topology availability, bottleneck capacity, and disturbance tolerance, and calculates their corresponding redundancy values. Specifically, the wastewater treatment plant process flow is modeled as a weighted directed graph, denoted as the wastewater treatment topology. Each node represents a structural unit and equipment, and each edge represents various types of pipes, valves, and pump stations. When some filters enter the backwashing state, relevant nodes and edges are temporarily removed from the wastewater treatment topology, and a currently available wastewater treatment topology is generated based on real-time valve positions and pump status. The connection between the outlet of the contact disinfection tank and the receiving water body is checked by depth-first search. If the connection is broken, the topology availability criterion is zero. If the connection is established, the maximum flow algorithm is used to calculate the maximum water discharge capacity of the current available topology from the outlet of the contact disinfection tank to the receiving water body, and it is compared with the actual water inflow demand of backwashing. The difference between the two is the topology availability margin. Identify components that are likely to become bottlenecks in the current wastewater treatment topology, calculate the ratio of actual flow to rated capacity for each bottleneck component, take the maximum load utilization rate, and then subtract the maximum load utilization rate from 1 to obtain the bottleneck capacity margin value. Using a water quality perturbation model modeled by a hybrid of mechanistic model and long short-term memory network, and utilizing historical data on backwashing intensity, duration, initial water quality, and downstream water quality changes, predict the peak turbidity of the effluent and the short-term decrease in disinfection indicators shortly after backwashing. Compare the prediction results with the various water quality standard thresholds to calculate turbidity tolerance and disinfection tolerance, and use the smaller of the two as the perturbation tolerance margin value. The water quality standard thresholds include both turbidity standard thresholds and disinfection standard thresholds.

4. The intelligent management method for smart water affairs according to claim 1, characterized in that, The logic for dividing peak and off-peak periods is as follows: when the scheduling status coefficient PSC=1, it is in a transitional state and enters the fine-tuning mode. The historical backwashing time interval data in the multi-dimensional operation monitoring data is used to associate each backwashing time interval with the preset time period to construct a time period-backwashing time interval mapping dataset. The percentile statistical method was used to calculate the backwash time interval corresponding to each preset time period in the time period-backwash time interval mapping dataset, as well as the low percentile value, high percentile value, and average value of the filter backwash time interval. Calculate the maximum value of the lower percentile and the minimum value of the upper percentile for all preset time periods. Define the peak time threshold as the minimum value of the upper percentile and the trough time threshold as the maximum value of the lower percentile. Then, compare the defined peak and trough time thresholds with the corresponding average backwash time intervals. When the average backwashing time interval of a certain preset period is less than the threshold of the trough period, the corresponding preset period is determined to be the trough period. When the average backwashing time interval of a preset period is greater than or equal to the peak period threshold, the corresponding preset period is determined to be a peak period.

5. The intelligent management method for smart water affairs according to claim 1, characterized in that, A random forest algorithm is used to construct a filter bed urgency assessment model, output urgency scores, and generate a filter bed backwashing urgency queue. Specifically, when the scheduling state coefficient PSC=1 and the current time period is a low period, the filter bed urgency assessment model is constructed using real-time collected filter bed operation data. The filter bed clogging risk characteristics selected by the filter bed urgency assessment model are: the clogging rate obtained based on the ratio of the filter bed inlet and outlet pressure difference to the initial pressure difference; the water quality index obtained based on the relative growth rate of turbidity of the filter bed effluent and inlet; and the operating time of the filter bed since the last backwash. The random forest classification model is trained with filter bed operation samples from historical operation and maintenance records, and the probability that each filter bed needs to be backwashed immediately is output as the urgency score. The filter bed backwashing urgency queue is generated in descending order of urgency scores.

6. The intelligent management method for smart water affairs according to claim 1, characterized in that, After classifying filters in the filter backwashing urgency queue with urgency scores not lower than the preset threshold into a priority backwashing filter set, and the rest into a delayed backwashing filter set, based on a comparison of urgency scores with preset thresholds, the process also includes: The maximum number of filter tanks that can be backwashed simultaneously, as set in advance, is called the parallel backwashing number. The filter combination that actually performs backwashing in this fine-tuning operation is determined based on the number of parallel backwashes, including: If the number of filters in the priority backwash filter set is less than or equal to the number of parallel backwashes, then all filters in the priority backwash filter set will be selected for backwashing first. The remaining available slots will be filled by selecting filters from the set of filters that can be delayed backwashing in descending order of urgency score. If the number of filters in the priority backwash filter set is greater than the number of parallel backwashes, then filters equal to the number of parallel backwashes will be selected from the priority backwash filter set in descending order of urgency score and included in this backwashing combination. The priority backwash filters that are not selected will be marked as key targets to be backwashed. Simultaneously, based on the real-time monitored disturbance tolerance margin value, the backwashing operation parameters of individual filters are dynamically adjusted: when the disturbance tolerance margin value is in the preset high margin range, the backwashing time of the filter is extended; when the disturbance tolerance margin value is in the preset middle margin range, the standard backwashing time of the filter is adopted; when the disturbance tolerance margin value is in the preset low margin range, the backwashing time of the filter is shortened and the backwashing intensity is reduced; if the water quality index is detected to be close to the safety threshold during the backwashing process, the backwashing operation of the current filter is terminated and switched to the standby filter.

7. The intelligent management method for smart water affairs according to claim 1, characterized in that, A set of candidate reconfiguration schemes is generated based on a set of finite operation steps. The incremental number of available backwash tanks and the valve position switching complexity of each candidate reconfiguration scheme are calculated. Specifically, when the topology reconfiguration mode is triggered, several reconfiguration schemes are enumerated within the limited operation steps to form a set of candidate reconfiguration schemes. For each candidate reconfiguration scheme, its impact on pipeline connectivity is analyzed using a digital twin model, and the number of available backwash tanks that can be increased while keeping the upper limit of parallel backwashing unchanged is calculated. At the same time, the valve position switching complexity required to execute the candidate reconfiguration scheme is calculated, which can be obtained by weighting and accumulating the number of valve opening and closing operations, the number of cross-zone operations, and the number of key valve actions. Candidate reconfiguration schemes that meet the conditions are selected based on a preset coverage threshold and the upper limit of valve position switching complexity. The preset coverage threshold refers to the number of backwash tanks that can be served within the number of parallel backwashing operations. The candidate reconfiguration scheme with the largest incremental number of available backwash tanks and the smallest valve position switching complexity is selected as the optimal reconfiguration scheme, and the corresponding valve switching command is generated.

8. An intelligent management system for smart water management, used to implement the intelligent management method for smart water management as described in any one of claims 1-7, characterized in that: The data acquisition module deploys smart sensors and edge computing gateways to collect multi-dimensional operational monitoring data, including flow rate, pressure, water quality, energy consumption, valve position, pump station status, and historical maintenance records. It performs time synchronization, anomaly detection, and data completion, and uploads the data to the cloud. The processing module calculates topology availability redundancy, bottleneck capacity redundancy, and disturbance tolerance redundancy, generates scheduling state coefficients, and trains a random forest model when the scheduling state is transitional, outputting filter urgency scores and filter flushing urgency queues. The scheduling module divides the scheduling state into peak and off-peak periods. During peak periods, it generates candidate topology reconfiguration schemes, evaluates the incremental increase in the number of available flushing tanks and the complexity of valve position switching, selects the optimal reconfiguration scheme, constructs a time-expanded maximum flow model, schedules the backwashing start and end slots for each filter in stages, and generates backwashing scheduling schemes and valve position switching instructions. The execution module performs rapid simulation verification of the final scheduling scheme output by the scheduling module on the digital twin platform. After verification, it sends the scheme to the equipment control system for backwashing scheduling and updates the filter status and model parameters after backwashing is completed.

Images