Pipe network drainage scheduling strategy recommendation method based on collaborative filtering algorithm

Abstract

The invention discloses a pipe network drainage scheduling strategy recommendation method based on a collaborative filtering algorithm, and the method comprises the following steps: 1, determining a research region, and collecting and sorting pipe network data, rainwater and sewage catchment partition data, lake canal system data and historical scheduling data in the research region; step 2, constructing a drainage scheduling model of the research area; step 3, constructing a scheduling scheme pool and a scheduling scene pool; 4, recommending a drainage scheduling scheme; step 5, based on the recommended drainage scheduling scheme, performing calculation and deduction in combination with the drainage scheduling model; a calculation deduction result and an evaluation index are obtained, after a drainage scheduling scheme is adjusted by scheduling personnel according to actual requirements, the scheme and a scene are stored in a data pool, calculation deduction is carried out again to obtain an expected scheduling result, and arrangement and implementation of a scheduling strategy are guided; according to the method, the defect of experience failure or insufficiency of drainage scheduling personnel caused by unknown scenes can be overcome.

Description

Technical Field

[0001] The invention relates to the technical field of pipe network drainage scheduling, and in particular to a pipe network drainage scheduling strategy recommendation method based on a collaborative filtering algorithm. Background Art

[0002] In recent years, due to the frequent occurrence of extreme weather, the number of rainstorms or extreme rainfall in cities has gradually increased, and the disasters and impacts they bring are often more serious and far-reaching. Therefore, in rainfall scenarios, the decision-making of plant network drainage scheduling for urban flood control and drainage and water environment improvement involving urban water safety and water environment management is particularly important.

[0003] At present, when conducting pipeline drainage scheduling, the scheduling strategy is usually arranged and adjusted mainly based on manual experience after referring to meteorological information, pipeline monitoring information and video information. The plan is often not rehearsed through the drainage scheduling model, and there is a lack of methods and systems that can intelligently make scheduling strategy recommendations and provide them to dispatchers for reference.

[0004] Specifically, there are two main types of existing technical solutions: One type does not use model calculations. Instead, through relevant empirical formulas, combined with the dispatcher's manual experience and on-site monitoring data information, the dispatcher directs and responds at any time to arrange and adjust the dispatch plan. The quality of the dispatch plan depends largely on the dispatcher's business ability and adaptability.

[0005] The other type only provides a drainage scheduling model for scheme calculation and rehearsal. The scheduling personnel are still required to provide scheduling strategies based on their own experience. After the model calculation, various indicators of the scheduling scheme are obtained. This cannot solve the current situation of over-reliance on the experience of existing experts.

[0006] Therefore, there are two problems with the existing technology: one is that the scheduling effect is unclear: it is unknown what impact the scheduling operation will have on the entire drainage system; the other is that the plan is not targeted enough: it can only execute a fixed plan based on the rainfall level, and there is no precise scheduling for different types of rainfall types, pipeline network status, storage tank capacity, etc. Summary of the invention

[0007] The purpose of the present invention is to overcome the above-mentioned shortcomings and provide a method for recommending a pipe network drainage scheduling strategy based on a collaborative filtering algorithm to solve the problems raised in the background technology.

[0008] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for recommending a pipe network drainage scheduling strategy based on a collaborative filtering algorithm, which comprises the following steps: Step 1: Determine the study area, collect and organize the pipe network data, rainwater and sewage catchment division data, lake and canal water system data, and historical dispatch data in the study area; Step 2: Based on the pipe and canal hydrodynamics model and the gate and pump hydrodynamics model, and according to the pipe network data, rainwater and sewage catchment division data, and lake and canal water system data collected and sorted out in step 1, a drainage scheduling model for the study area is constructed; Step 3: Based on the collected historical dispatch data and the dispatch plans customized by dispatchers according to the dispatch model, a dispatch plan pool and a dispatch scenario pool are constructed; Step 4: Based on the data of the scheduling scheme pool and the scenario pool, a collaborative filtering algorithm is used to calculate the correlation of each scheme and the correlation of each scenario; based on the current or recent predicted rainfall data and pipe network liquid level data, a drainage scheduling scheme is specifically recommended; Step 5: Based on the recommended drainage scheduling plan, the drainage scheduling model is combined for calculation and deduction; the calculation and deduction results and evaluation indicators are obtained. After the scheduling personnel adjust the drainage scheduling plan according to actual needs, the plan and scenario are saved in the data pool, and the calculation and deduction are performed again to obtain the expected scheduling results to guide the layout and implementation of the scheduling strategy.

[0009] Furthermore, in step 1, the pipe network data includes data such as the structure, material, burial depth, inspection well node topology information of the rainwater and sewage pipe network, as well as gate pump attributes and gate pump design information; the rainwater and sewage catchment zoning data and lake and canal water system data include the rainwater and sewage zoning GIS data, water system vector data, and lake and canal DEM data in the study area.

[0010] Furthermore, in step 1, the historical dispatching data includes historical rainfall monitoring data, pipe network liquid level data at the time of rainfall, and the executed drainage dispatching strategy.

[0011] Furthermore, in step 2, the specific method for constructing the drainage scheduling model of the study area is as follows: Step 2.1, based on the pipe network data, rainwater and sewage catchment zoning data, and lake and canal water system data collected and sorted in step 1, write the pipe network topology and pipe network attributes in the study area into a structured data file, write the relevant geographic elevation and burial depth information into the file according to the pipe network topology and node coordinates, and express the location of the gate pump and the connection relationship with the pipe network in the data file, forming a parameter input file of the pipe network grid and attributes in the study area; Step 2.2: Based on the canal hydrodynamic model and the sluice pump hydrodynamic model, fill in the relevant parameters in the model formula according to the parameter input file for calculation.

[0012] Furthermore, the canal hydrodynamic model is developed and constructed based on the one-dimensional Saint-Venant equations, which can be used to describe the unsteady flow evolution process of one-dimensional rivers and canals. The equations are composed of the continuity equation and the momentum equation, which are expressed as: ; ; Where: A is the area of ​​the water-passing section of the river / pipeline; x is the distance; y is the water depth of the river / pipeline. The liquid level and water level can be calculated based on the buried depth of the pipeline or the geographical elevation of the open channel; g is the acceleration of gravity; S f is the friction slope coefficient, and Q is the flow rate through the section.

[0013] Furthermore, the pump-sluice hydrodynamic model is composed of a gate model and a pump station model; Gate model: By dividing the gate flow process into orifice flow and weir flow process, the hydrodynamic calculation of the gate can be performed; Among them, when the quotient of the gate opening e and the water depth H0 in front of the gate (e / H0) is less than 0.65, it is regarded as orifice flow, and the formula of orifice flow process is expressed as: ; In the formula, Q0 represents the orifice flow rate, σ represents the flooding coefficient; m represents the comprehensive orifice flow coefficient; b represents the width of the water-passing section; e represents the gate opening; and g represents the acceleration of gravity. When the quotient of the gate opening e and the water depth H0 in front of the gate (e / H0) is greater than 0.65, it is regarded as weir flow. The formula for the weir flow process is expressed as: ; Where Q1 represents the weir flow rate, C w represents the weir flow coefficient; Pump station model: Considering that the flow rate of the pump station changes continuously with the head difference between the inlet and outlet nodes, the flow value corresponding to the head difference is obtained through linear interpolation in actual calculation.

[0014] Furthermore, the specific process of constructing the scheduling solution pool and the scheduling scenario pool in step 3 is as follows: Step 3.1: Based on the historical dispatch data collected and collated in step 1, as well as some rainfall scenarios and dispatch plans customized by dispatch personnel to supplement the data content, they are classified into one category according to the matching of rainfall data and pipe network liquid level, and the drainage dispatch plans corresponding to the rainfall and pipe network liquid level are separately classified into one category; the historical dispatch data includes historical rainfall monitoring data, pipe network liquid level data at the time of rainfall, and the implemented drainage dispatch strategy, and the drainage dispatch plan includes the dispatch plan of pump station and gate; Step 3.2, format the rainfall data-pipeline liquid level data, group them according to the maximum rainfall intensity, and divide them into two groups: small and medium rain and extreme rainstorm, and save them in the scheduling scene pool after numbering; Step 3.3, format the drainage scheduling scheme data, and calculate the scheduling results of each scenario and scheduling scheme through the drainage scheduling model constructed in step 2; Step 3.4: Based on the calculation results of each plan, key indicators are counted, including overflow water volume in each period, total overflow water volume, overflow duration, pump station energy consumption, pump station start-stop times and pipe network empty capacity percentage, pump station start-stop times and sewage pipe network empty capacity percentage data, and the key indicators are associated with the corresponding plan. The plan is numbered and saved in the scheduling plan pool.

[0015] Furthermore, the specific process of step 4 is as follows: Step 4.1, extract each scene data in the scheduling scene pool, and use the Pearson correlation coefficient to calculate the correlation between the scenes in the two groups; Step 4.2: Extract the data of each plan in the scheduling plan pool, and use the Pearson correlation coefficient to calculate the correlation between the scheduling plans based on the key indicators of each plan, including the overflow water volume in each period, the total overflow water volume, the overflow duration, the energy consumption of the pump station, the number of pump station starts and stops, and the percentage of the vacant capacity of the sewage pipe network; the calculation process of the correlation between the plans is the same as step 4.1, and the physical quantities therein are replaced by the key indicator quantities of the plan; Step 4.3: Based on the current or recent predicted rainfall data and pipe network liquid level data, determine which group it belongs to based on the maximum rainfall intensity, calculate the correlation between the data and other data in the group, find the three dispatching scenario data with the highest correlation with the data, extract and recommend the dispatching plan data corresponding to these three scenarios; Step 4.4: If the current or recent predicted rainfall data and pipe network liquid level data are extreme unknown scenarios and no satisfactory correlation can be found with any scheduling scenario data, then based on the key indicators of the scheduling plan set or required by the dispatcher, find the three scheduling plans closest to the key indicators, extract and recommend the data of these three scheduling plans.

[0016] Furthermore, the calculation process of the correlation in step 4.1 is as follows: For the small and medium rainfall groups, consider the rainfall duration T, rainfall intensity q at each moment, and average rainfall intensity , peak rainfall intensity , and the average value m of the percentage of the liquid level in the pipeline network. The Pearson correlation coefficient is calculated for each of the above physical quantities. The formula of the coefficient is expressed as follows: ; In the formula, r represents the correlation coefficient; n represents the number of data items of the physical quantity; and Respectively represent the kth data in scene x and scene y; and Represent the data averages of scene x and scene y respectively; Finally, the correlation coefficient between the scene data in this group is obtained through weighted average. It is expressed as: ; Among them, a and b represent any two different scene data, and J represents the total number of statistical physical quantities of the group; Represents the Pearson correlation coefficient of the jth physical quantity between data a and b; For the extreme rainstorm group, consider the rainfall duration T, rainfall intensity q at each moment, and average rainfall intensity , peak rainfall intensity , the average rainfall depth i of the rainstorm and the average value m of the percentage of the pipe network liquid level.

[0017] Furthermore, the specific process of step 5 is as follows: Step 5.1: Based on the drainage scheduling scheme recommended in step 4, combined with the current or recent predicted rainfall data and pipe network liquid level data, the drainage scheduling numerical model constructed in step 2 is called to calculate, the results corresponding to the scheduling scheme are obtained, and its key indicators are statistically calculated; Step 5.2: The dispatcher selects a dispatch plan and adjusts it according to the actual needs and conditions to meet the actual working conditions; Step 5.3: Based on the adjusted dispatching plan, combined with the current or recent predicted rainfall data and pipe network liquid level data, the drainage dispatching numerical model constructed in step 2 is called to calculate, and the results corresponding to the dispatching plan are obtained. The key indicators are statistically calculated, and the dispatching personnel determine whether to repeat step 5.2 based on actual needs; Step 5.4: After determining the dispatching plan, associate the rainfall data, pipe network liquid level data and dispatching plan strategy of the plan, and add them to the dispatching scenario pool and dispatching plan pool to expand and improve the scenario and plan pool, and guide subsequent dispatching personnel to arrange and implement dispatching strategies in other dispatching scenarios.

[0018] Beneficial effects of the present invention: The present invention can be based on the constructed drainage scheduling model, by pre-storing a large number of scheduling plans, combining the scheduling strategies and recommendation algorithms in similar scenarios, and giving the drainage scheduling recommendation strategy in the current scenario, so as to solve the current situation of over-reliance on the manual experience of drainage scheduling personnel; and in unknown extreme emergency scenarios, according to the changes in key indicators calculated by the scheduling model and the combination of historical experience, give the corresponding drainage scheduling recommendation strategy, which can make up for the shortcomings of invalid or insufficient experience of drainage scheduling personnel due to unknown scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1It is a flowchart of a method for recommending a drainage scheduling strategy for a pipe network based on a collaborative filtering algorithm; Figure 2 The remaining capacity diagram of the rainwater and sewage pipe network is a schematic diagram of the calculation results of the drainage scheduling model; Figure 3 The scheduling scenario and corresponding scheduling scheme management diagram of the drainage scheduling strategy recommendation system; Figure 4 This is a comparison chart of key indicators of the recommended scheduling scheme shown in the drainage strategy recommendation system. DETAILED DESCRIPTION

[0020] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0021] Example 1: In a riverside city in central my country, there are many lakes and rivers in the urban area. It is easy to have water environment and water safety problems in the urban mountainous area due to the influence of short-term heavy rain or continuous rainfall. Therefore, it is necessary to develop and build a set of drainage scheduling models and scheduling strategy recommendation methods suitable for the urban area of ​​the city to assist local relevant supervision and enforcement department personnel in discussing drainage scheduling strategies and simulating and analyzing scheduling results, and support the completion of urban drainage scheduling plans under various rainfall scenarios.

[0022] like Figure 1 As shown, a method for recommending a pipe network drainage scheduling strategy based on a collaborative filtering algorithm includes the following steps: Step 1. According to the planning information of the local urban area, the research area where scheduling is required is determined, and the design and survey data of the pipeline network and gate pumps in the study area are further collected and organized, including data such as the structure, material, burial depth, and topological information of the inspection well node of the rainwater and sewage pipeline network, as well as gate pump scheduling rules, gate pump location information, gate pump structural properties, and water flow capacity curve data; the rainwater and sewage catchment zoning data and lake and canal water system data in the study area are collected and organized, including the rainwater and sewage zoning GIS data, water system vector data, and lake and canal DEM data in the study area; the historical scheduling data are collected and organized, including historical rainfall monitoring data, the pipe network liquid level data at the time of rainfall, and the drainage scheduling strategy implemented.

[0023] Step 2: Based on the collected data, the drainage scheduling model is constructed; the network topology and network attributes of the study area are written into a structured data file, and the relevant geographic elevation and burial depth information are written into the file according to the network topology and node coordinates, and the location of the gate pump and the connection relationship with the network are expressed in the data file to form a parameter input file of the network grid and attributes in the study area; based on the canal hydrodynamics and gate pump hydrodynamics model, the drainage scheduling model is constructed; specifically, the canal hydrodynamic model used is developed and constructed based on the one-dimensional Saint-Venant equations, which can be used to describe the evolution process of unsteady flow in one-dimensional rivers and canals. The equations are composed of continuity equations and momentum equations, which can be expressed as:

[0024]

[0025] Where: A is the area of ​​the water-passing section of the river / pipeline (㎡); x is the distance (m); y is the water depth of the river / pipeline (m). The liquid level and water level can be calculated based on the buried depth of the pipeline or the geographical elevation of the open channel; g is the gravitational acceleration (㎡ / s); Sf is the friction slope coefficient.

[0026] The pump-sluice hydrodynamic model used consists of a gate model and a pump station model; Gate model: By dividing the gate flow process into orifice flow and weir flow process, the hydrodynamic calculation of the gate can be performed; Among them, when the quotient of the gate opening e and the water depth H0 in front of the gate (e / H0) is less than 0.65, it is regarded as orifice flow, and the formula of orifice flow process can be expressed as:

[0027] In the formula, Q0 represents the orifice flow rate, σ represents the flooding coefficient; m represents the comprehensive orifice flow coefficient; b represents the width of the water-passing section (m); e represents the gate opening (m); and g represents the acceleration of gravity. When the quotient of the gate opening e and the water depth H0 in front of the gate (e / H0) is greater than 0.65, it is regarded as weir flow. The formula of weir flow process can be expressed as:

[0028] Where Q1 represents the weir flow rate, and Cw represents the weir flow coefficient.

[0029] Pump station model: Considering that the flow rate of the pump station changes continuously with the head difference between the water inlet and outlet nodes, by defining the pump characteristic curve (i.e. the relationship curve between the head difference and the pump station flow rate), in the embodiment of this project, the flow value corresponding to the head difference is obtained by linear interpolation during actual calculation.

[0030] In this embodiment, the constructed model includes more than 400 generalized pipelines, 1 open channel, several pumping stations and gates.

[0031] The model constructed in this embodiment can calculate the scheduling plan results based on the input rainfall data, the initial pipe network liquid level status and the scheduling process of the pump and gate, combined with the parameters defined during construction, including the time series data of the liquid level or water level at various locations in the pipe network or open channel, as well as the time series data of the pipe network overflow water volume and the time series data of the gate pump energy consumption, etc., which can be used for statistical calculation of key indicators of the drainage scheduling plan.

[0032] Step 3: Based on the collected historical rainfall, pipe network level and dispatch strategy data, as well as some rainfall scenarios and dispatch plans customized by dispatch personnel to supplement the data content, they are matched according to rainfall data-pipe network level and classified into one category, and the drainage dispatch plans corresponding to the rainfall and pipe network level (including the dispatch plans of pump stations and gates) are classified into one category separately; the rainfall data-pipe network level data are formatted and grouped according to the maximum rainfall intensity (i.e., peak rainfall intensity), divided into two groups: small and medium rain and extreme rainstorm, and saved in the dispatch scenario pool after numbering; the drainage dispatch plan data is formatted, and the dispatch results of each scenario and dispatch plan are calculated through the constructed drainage dispatch model; based on the calculation results of each plan, key indicators are counted, including overflow water volume (mm), total overflow water volume (mm), overflow duration (h), pump station energy consumption (Kw.h), pump station start and stop times and percentage of sewage pipe network vacant capacity in each period, and the key indicators are associated with the corresponding plans, and the plans are numbered and saved in the dispatch plan pool.

[0033] In this embodiment, considering that the peak rainfall intensity is the local rainstorm intensity that occurs once every 10 years, after data collection and collation, there are about 40 scheduling scenario data in total, including 38 small and medium-sized rains; 3 corresponding scheduling plan data are missing, and the dispatching personnel supplement the scheduling plan according to the rules and perform the scheduling model calculation, and then organize them into the scheduling plan pool; to supplement the data, in the rainfall and pipeline liquid level scenarios randomly generated by the system, as well as various randomly generated scheduling plans and model calculation results, the dispatching personnel supplemented and calculated more than 40 data into the scheduling scenario and plan pool based on the possible rainfall and liquid level scenarios, and the appropriate scheduling plans under the corresponding rules.

[0034] Step 4: Extract each scene data in the scheduling scene pool. For each scene in the two groups, use the Pearson correlation coefficient to calculate the correlation between the scenes in the group. For the small and medium rain groups, consider the rainfall duration T (h), rainfall intensity q (L / s.ha) at each moment, and average rainfall intensity (L / s.ha), peak rainfall intensity (L / s.ha), and the average value m of the percentage of the liquid level in the pipe network. The Pearson correlation coefficient is calculated for each of the above physical quantities. The formula of the coefficient can be expressed as:

[0035] In the formula, r represents the correlation coefficient; n represents the number of data items of the physical quantity (some physical quantities have only one data item); and Respectively represent the kth data in scene x and scene y; and Represent the data averages of scene x and scene y respectively.

[0036] Finally, the correlation coefficient between the scene data in this group is obtained through weighted average. (a, b represent any two different scene data) can be expressed as

[0037] Among them, J represents the total number of statistical physical quantities in the group (e.g., 5 for the small and medium rain group); Represents the Pearson correlation coefficient of the jth physical quantity between data a and b.

[0038] For the extreme rainstorm group, consider the rainfall duration T (h), rainfall intensity q (L / s.ha) at each moment, and average rainfall intensity (L / s.ha), peak rainfall intensity (L / s.ha), average rainfall depth of rainstorm i (mm / min) and average value m of percentage of pipe network liquid level.

[0039] Extract the data of each plan in the scheduling plan pool, based on the key indicators of each plan, including the overflow water volume in each period (m 3 )、Total overflow water volume(m 3 ), overflow duration (h), pump station energy consumption (Kw.h), pump station start and stop times and percentage of network vacant capacity, etc., and the Pearson correlation coefficient is used to calculate the correlation between the various scheduling schemes; the calculation process of the correlation between the various schemes refers to the above process, and the physical quantities therein are replaced by the key indicator quantities of the scheme; Based on the current or recent predicted rainfall data and pipe network liquid level data, determine which group it belongs to based on the maximum rainfall intensity, calculate the correlation between the data and other data in the group, find the three dispatch scenario data with the highest correlation with the data, extract and recommend the dispatch plan data corresponding to these three scenarios; If the current or near-term predicted rainfall data and pipe network liquid level data are extreme unknown scenarios, and no satisfactory correlation can be found with any scheduling scenario data, then according to the scheduling plan key indicators set or required by the dispatcher, find the three scheduling plans closest to the key indicators, extract and recommend the data of these three scheduling plans. In this embodiment, the absolute value of the Pearson correlation coefficient of the scheduling scenario is less than 0.2, which is regarded as no scenario correlation. In actual application, they are basically correlated in the small and medium rain groups, and there is a certain probability of irrelevant scenarios in the extreme rainstorm group, and it is necessary to find correlation from the scheduling plan pool.

[0040] Step 5: Based on the three recommended schemes, combined with the current or recent predicted rainfall data and pipe network liquid level data, the constructed drainage scheduling numerical model is called to calculate, the results corresponding to the scheduling scheme are obtained, and its key indicators are statistically calculated; the scheduling personnel select a scheduling scheme and adjust it according to actual needs and conditions to meet the actual working conditions; According to the adjusted dispatching plan, combined with the current or recent predicted rainfall data and pipe network liquid level data, the constructed drainage dispatching numerical model is called for calculation to obtain the results corresponding to the dispatching plan, and its key indicators are statistically calculated. The dispatching personnel will determine whether it is necessary to repeat the steps of plan adjustment and model calculation based on actual needs; After determining the scheduling plan, the system associates the rainfall data, pipeline liquid level data and scheduling plan strategy of the plan, and adds them to the scheduling scenario pool and scheduling plan pool to expand and improve the scenario and plan pool, and guide subsequent scheduling personnel to arrange and implement scheduling strategies in other scheduling scenarios.

[0041] In this embodiment, the dispatching personnel can use the dispatching scheme recommended by the technical solution of the present invention based on the future medium- and short-term rainfall forecast data, combined with the current pipe network liquid level monitoring data, to quickly respond and complete the dispatching consultation decision, thereby supporting the deployment and implementation of the dispatching scheme strategy in the rainfall scenario.

[0042] In the above embodiment, the drainage scheduling simulation model includes a combination of the pipeline hydrodynamic model and the pump and sluice hydrodynamic model; this embodiment constructs a scheduling scenario pool and a scheduling scheme pool, and each scenario and scheme has its quantitative indicators; finally, the scheduling scheme recommendation algorithm of this embodiment can give corresponding recommended scheduling schemes according to the quantitative indicators of the scheduling scenarios and scheme pools, as well as different scenario ranges.

[0043] In addition, in this embodiment, Figure 2The remaining capacity diagram of the rainwater and sewage pipe network is a schematic diagram of the calculation results of the drainage scheduling model. After the drainage scheduling model is calculated in the study area, the rainwater and sewage pipe networks in each area can show different remaining capacities, and the remaining capacity values ​​range from 1.0 (completely empty) to 0.0 (completely full); Figure 3 The diagram shows the scheduling scenario and the corresponding scheduling scheme management diagram of the drainage scheduling strategy recommendation system. The table on the right side of the diagram shows the timing operation rules of each pump and gate station recommended after calculation by the drainage scheduling strategy recommendation system. Figure 4 This is a comparison chart of key indicators of the recommended scheduling schemes shown in the drainage strategy recommendation system. By comprehensively comparing the indicators such as pipe network overflow flow, internal waterlogging volume and pipeline vacant storage capacity of the three recommended schemes, the total score and key indicators of each scheme are obtained, which can provide guidance for users' scientific scheduling.

[0044] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A method for recommending a drainage scheduling strategy for a pipe network based on a collaborative filtering algorithm, characterized in that: It includes the following steps: Step 1: Determine the study area, collect and organize the pipe network data, rainwater and sewage catchment division data, lake and canal water system data, and historical dispatch data in the study area; Step 2: Based on the pipe and canal hydrodynamics model and the gate and pump hydrodynamics model, and according to the pipe network data, rainwater and sewage catchment division data, and lake and canal water system data collected and sorted out in step 1, a drainage scheduling model for the study area is constructed; Step 3: Based on the collected historical dispatch data and the dispatch plans customized by dispatchers according to the dispatch model, a dispatch plan pool and a dispatch scenario pool are constructed; Step 4: Based on the data of the scheduling scheme pool and the scenario pool, a collaborative filtering algorithm is used to calculate the correlation of each scheme and the correlation of each scenario; based on the current or recent predicted rainfall data and pipe network liquid level data, a drainage scheduling scheme is specifically recommended; Step 5: Based on the recommended drainage scheduling plan, the drainage scheduling model is combined for calculation and deduction; the calculation and deduction results and evaluation indicators are obtained. After the scheduling personnel adjust the drainage scheduling plan according to actual needs, the plan and scenario are saved in the data pool, and the calculation and deduction are performed again to obtain the expected scheduling results to guide the layout and implementation of the scheduling strategy.

2. The method for recommending a pipeline drainage scheduling strategy based on a collaborative filtering algorithm according to claim 1, characterized in that: In step 1, the pipe network data includes data such as the structure, material, burial depth, inspection well node topology information, and gate pump attributes and gate pump design information of the rainwater and sewage catchment area; the rainwater and sewage catchment area data and lake and canal water system data include the rainwater and sewage area GIS data, water system vector data, and lake and canal DEM data in the study area.

3. The method for recommending a pipeline network drainage scheduling strategy based on a collaborative filtering algorithm according to claim 1, characterized in that: In step 1, the historical dispatching data includes historical rainfall monitoring data, pipe network liquid level data at the time of rainfall, and the executed drainage dispatching strategy.

4. The method for recommending a pipeline network drainage scheduling strategy based on a collaborative filtering algorithm according to claim 1, characterized in that: In step 2, the specific method of constructing the drainage scheduling model of the study area is as follows: Step 2.1, based on the pipe network data, rainwater and sewage catchment zoning data, and lake and canal water system data collected and sorted in step 1, write the pipe network topology and pipe network attributes in the study area into a structured data file, write the relevant geographic elevation and burial depth information into the file according to the pipe network topology and node coordinates, and express the location of the gate pump and the connection relationship with the pipe network in the data file, forming a parameter input file of the pipe network grid and attributes in the study area; Step 2.2: Based on the canal hydrodynamic model and the sluice pump hydrodynamic model, fill in the relevant parameters in the model formula according to the parameter input file for calculation.

5. The method for recommending a pipeline network drainage scheduling strategy based on a collaborative filtering algorithm according to claim 4 is characterized in that: The canal hydrodynamic model is developed and constructed based on the one-dimensional Saint-Venant equations, which can be used to describe the unsteady flow evolution process of one-dimensional rivers and canals. The equations are composed of the continuity equation and the momentum equation, which are expressed as: ； ； Where: A is the area of ​​the water-passing section of the river / pipeline; x is the distance; y is the water depth of the river / pipeline. The liquid level and water level can be calculated based on the buried depth of the pipeline or the geographical elevation of the open channel; g is the acceleration of gravity; S f is the friction slope coefficient, and Q is the flow rate through the section.

6. The method for recommending a pipeline network drainage scheduling strategy based on a collaborative filtering algorithm according to claim 4 is characterized in that: The pump-sluice hydrodynamic model consists of a gate model and a pump station model; Gate model: By dividing the gate flow process into orifice flow and weir flow process, the hydrodynamic calculation of the gate can be performed; Among them, when the quotient of the gate opening e and the water depth H0 in front of the gate (e / H0) is less than 0.65, it is regarded as orifice flow, and the formula of orifice flow process is expressed as: ； In the formula, Q0 represents the orifice flow rate, σ represents the flooding coefficient; m represents the comprehensive orifice flow coefficient; b represents the width of the water-passing section; e represents the gate opening; and g represents the acceleration of gravity. When the quotient of the gate opening e and the water depth H0 in front of the gate (e / H0) is greater than 0.65, it is regarded as weir flow. The formula for the weir flow process is expressed as: ； Where Q1 represents the weir flow rate, C w represents the weir flow coefficient; Pump station model: Considering that the flow rate of the pump station changes continuously with the head difference between the inlet and outlet nodes, the flow value corresponding to the head difference is obtained through linear interpolation in actual calculation.

7. The method for recommending a pipeline network drainage scheduling strategy based on a collaborative filtering algorithm according to claim 1, characterized in that: The specific process of constructing the scheduling solution pool and the scheduling scenario pool in step 3 is as follows: Step 3.1: Based on the historical dispatch data collected and collated in step 1, as well as some rainfall scenarios and dispatch plans customized by dispatch personnel to supplement the data content, they are classified into one category according to the matching of rainfall data and pipe network liquid level, and the drainage dispatch plans corresponding to the rainfall and pipe network liquid level are separately classified into one category; the historical dispatch data includes historical rainfall monitoring data, pipe network liquid level data at the time of rainfall, and the implemented drainage dispatch strategy, and the drainage dispatch plan includes the dispatch plan of pump station and gate; Step 3.2, format the rainfall data-pipeline liquid level data, group them according to the maximum rainfall intensity, and divide them into two groups: small and medium rain and extreme rainstorm, and save them in the scheduling scene pool after numbering; Step 3.3, format the drainage scheduling scheme data, and calculate the scheduling results of each scenario and scheduling scheme through the drainage scheduling model constructed in step 2; Step 3.4: Based on the calculation results of each plan, key indicators are counted, including overflow water volume in each period, total overflow water volume, overflow duration, pump station energy consumption, pump station start-stop times and pipe network empty capacity percentage, pump station start-stop times and sewage pipe network empty capacity percentage data, and the key indicators are associated with the corresponding plan. The plan is numbered and saved in the scheduling plan pool.

8. The method for recommending a pipeline network drainage scheduling strategy based on a collaborative filtering algorithm according to claim 1, characterized in that: The specific process of step 4 is as follows: Step 4.1, extract each scene data in the scheduling scene pool, and use the Pearson correlation coefficient to calculate the correlation between the scenes in the two groups; Step 4.2: Extract the data of each plan in the scheduling plan pool, and use the Pearson correlation coefficient to calculate the correlation between the scheduling plans based on the key indicators of each plan, including the overflow water volume in each period, the total overflow water volume, the overflow duration, the energy consumption of the pump station, the number of pump station starts and stops, and the percentage of the vacant capacity of the sewage pipe network; the calculation process of the correlation between the plans is the same as step 4.1, and the physical quantities therein are replaced by the key indicator quantities of the plan; Step 4.3: Based on the current or recent predicted rainfall data and pipe network liquid level data, determine which group it belongs to based on the maximum rainfall intensity, calculate the correlation between the data and other data in the group, find the three dispatching scenario data with the highest correlation with the data, extract and recommend the dispatching plan data corresponding to these three scenarios; Step 4.4: If the current or recent predicted rainfall data and pipe network liquid level data are extreme unknown scenarios and no satisfactory correlation can be found with any scheduling scenario data, then based on the key indicators of the scheduling plan set or required by the dispatcher, find the three scheduling plans closest to the key indicators, extract and recommend the data of these three scheduling plans.

9. The method for recommending a pipeline network drainage scheduling strategy based on a collaborative filtering algorithm according to claim 8, characterized in that: The calculation process of the correlation in step 4.1 is as follows: For the small and medium rainfall groups, consider the rainfall duration T, rainfall intensity q at each moment, and average rainfall intensity , peak rainfall intensity , and the average value m of the percentage of the liquid level in the pipeline network. The Pearson correlation coefficient is calculated for each of the above physical quantities. The formula of the coefficient is expressed as follows: ； In the formula, r represents the correlation coefficient; n represents the number of data items of the physical quantity; and Respectively represent the kth data in scene x and scene y; and Represent the data averages of scene x and scene y respectively; Finally, the correlation coefficient between the scene data in this group is obtained through weighted average. It is expressed as: ； Among them, a and b represent any two different scene data, and J represents the total number of statistical physical quantities of the group; Represents the Pearson correlation coefficient of the jth physical quantity between data a and b; For the extreme rainstorm group, consider the rainfall duration T, rainfall intensity q at each moment, and average rainfall intensity , peak rainfall intensity , the average rainfall depth i of the rainstorm and the average value m of the percentage of the pipe network liquid level.

10. The method for recommending a drainage scheduling strategy for a pipe network based on a collaborative filtering algorithm according to claim 1, characterized in that: The specific process of step 5 is as follows: Step 5.1: Based on the drainage scheduling scheme recommended in step 4, combined with the current or recent predicted rainfall data and pipe network liquid level data, the drainage scheduling numerical model constructed in step 2 is called to calculate, the results corresponding to the scheduling scheme are obtained, and its key indicators are statistically calculated; Step 5.2: The dispatcher selects a dispatch plan and adjusts it according to the actual needs and conditions to meet the actual working conditions; Step 5.3: Based on the adjusted dispatching plan, combined with the current or recent predicted rainfall data and pipe network liquid level data, the drainage dispatching numerical model constructed in step 2 is called to calculate, and the results corresponding to the dispatching plan are obtained. The key indicators are statistically calculated, and the dispatching personnel determine whether to repeat step 5.2 based on actual needs; Step 5.4: After determining the dispatching plan, associate the rainfall data, pipe network liquid level data and dispatching plan strategy of the plan, and add them to the dispatching scenario pool and dispatching plan pool to expand and improve the scenario and plan pool, and guide subsequent dispatching personnel to arrange and implement dispatching strategies in other dispatching scenarios.

Images