A method for optimizing the layout of monitoring points in urban water supply networks

By constructing a minimum response flow meter and a multi-objective optimization algorithm, the problem of the inhomogeneity of spatial partitioning of the monitoring point arrangement in the urban water supply pipeline network is solved, the efficiency and accuracy of explosive pipe positioning are improved, and the calculation amount is reduced.

CN114580786BActive Publication Date: 2024-11-12GUANGZHOU WATER SUPPLY CO +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210332084.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-03-30
Publication Date
2024-11-12
Estimated Expiration
2042-03-30

AI Technical Summary

Technical Problem

In urban water supply pipeline networks, traditional monitoring point arrangement methods are difficult to effectively consider the unevenness of the spatial partition of the pipe section, resulting in an increase in the workload of explosive pipe positioning, and the traditional hydraulic model has a large amount of calculation, which affects the optimization efficiency.

Method used

By constructing a minimum response flow table to replace traditional hydraulic model calculations, the minimum monitoring flow of the pipe segment is obtained, and the pipe segment monitoring coverage, spatial partition inhomogeneity and average similarity are set as the objective functions. The multi-objective optimization algorithm Borg MOEA is used for monitoring. Point layout optimization.

Benefits of technology

The feedback speed and accuracy of monitoring point arrangement are improved, the simulation calculation amount is reduced, and faster and more accurate explosive tube positioning is achieved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114580786B_ABST
    Figure CN114580786B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for optimizing the layout of monitoring points in urban water supply pipe networks, including constructing a minimum response flow table for retrieving and obtaining the minimum monitoring flow of all pipe sections in any monitoring point layout plan in the area to be optimized, and setting three pipe network monitoring network efficiency evaluation functions as optimization targets based on pipe section monitoring coverage, pipe section spatial partitioning non-uniformity, and average similarity between pipe sections, iteratively calculating and multi-objective optimizing the monitoring point layout plan, and obtaining a Pareto optimal solution set for determining the number and layout positions of optimized monitoring points. The method obtains the minimum monitoring flow of all pipe sections by retrieving the minimum response flow table to replace the traditional hydraulic model calculation to obtain the minimum monitoring flow of all pipe sections, thereby solving the problem of excessive simulation calculation in the process of pipe section spatial partitioning optimization, and improving the feedback speed and accuracy of the burst pipe positioning work of the final layout plan.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of urban water supply network monitoring points, and in particular to a step optimization method for urban water supply network monitoring points. Background Art

[0002] As one of the most important public infrastructures in the city, the reliable operation of the water supply network is related to the safety of drinking water for residents and the normal operation of industrial production. However, the water supply network system has complex pipe connection structures, corrosion, uneven ground drop, construction work and other pipe bursts. At the same time, there are a large number of water demand nodes in the network. On the one hand, it is not economical to install monitoring points on each node. On the other hand, there is a lot of redundant information in the data and more equipment storage and computing equipment are needed to save and process the corresponding data. Therefore, the problem of optimizing the layout of pressure monitoring points in the water supply network is to reasonably select some nodes to arrange monitoring points to achieve better monitoring and positioning effects.

[0003] The optimization layout of pressure monitoring points in water supply network is mainly divided into two categories: empirical method and theoretical method. The empirical method is generally used to arrange monitoring points at the most unfavorable points, important users and other nodes, but it is difficult to meet the development needs of water companies and is generally only used as an auxiliary method. The theoretical method uses mathematical knowledge to determine the installation nodes of monitoring points through calculation and analysis, which is divided into clustering method, sensitivity matrix analysis and multi-objective optimization method. The clustering method may have the problem that the nodes cannot be connected after clustering. At present, it is mostly used as an auxiliary step for monitoring points to improve efficiency or reduce sensor information redundancy; sensitivity matrix analysis may have the problem that the sensitivity of nodes near high-sensitivity nodes is also relatively high, resulting in the phenomenon of monitoring point aggregation; multi-objective optimization can consider multiple goals at the same time and is used in various fields. It is suitable for solving the problem of optimizing the layout of monitoring points and can meet the needs of modern water supply companies in monitoring and positioning of water supply network events.

[0004] When studying the problem of locating pipe bursts, it was found that the unevenness of the spatial partitioning of pipe segments will affect the efficiency of searching for pipe segments when a pipe burst occurs. If the unevenness of the spatial partitioning of pipe segments is not considered when arranging monitoring points, it may lead to an increase in the workload of subsequent pipe burst locating. Therefore, this objective function can be considered when arranging monitoring points, thereby reducing the workload of subsequent pipe burst locating. However, in traditional multi-objective function optimization, since there are a large number of nodes in the water supply network within a certain area where monitoring points can be selected, and there are tens of millions of potential layout plans, and the evaluation of each layout plan requires hydraulic simulation of each pipe segment of the water supply network from 0 to the maximum burst flow, it will greatly increase the difficulty of optimizing the spatial partitioning of pipe segments.

[0005] Patent document CN112989535A discloses a method for optimizing the layout of pressure monitoring points in a water supply network based on the benefits of pipe burst detection, including using the pipe network hydraulic model to calculate the pressure drop of each pipeline under different levels of pipe burst in EPANET, and constructing a pipe burst event coverage matrix based on the calculated pressure drop results; then, for a given number of water pressure monitoring points, the layout of the optimized water pressure monitoring points is constructed with the maximum pipe burst event coverage rate as the objective function and the distribution estimation algorithm is used to solve it, thereby determining the optimal layout of the monitoring points under a given number. This method uses the pipe burst event coverage rate and the minimum detectable flow rate of pipe burst as indicators to solve the problem of how many water pressure monitoring points should be arranged and where to arrange them in engineering practice. However, this layout method only considers the economic efficiency of the layout, but does not consider the subsequent pipe burst location problem.

[0006] Patent document CN112503400A discloses a multi-objective optimization layout method for pressure measuring points in a water supply network, including: (1) hydraulic calculation of the basic network model to obtain the basic water pressure of each point under normal working conditions; (2) establishing a pressure-driven model, setting the minimum service head, minimum outlet head and ejector index, simulating pipe bursts for all pipe sections, and obtaining the water pressure of each node under different pipe burst events and the flow rate of the corresponding pipe burst events; (3) setting three monitoring network effective performance evaluation functions based on the coverage, sensitivity and positioning accuracy of the monitoring network respectively. For a given monitoring point layout plan, the three evaluation function values ​​can be calculated based on the above calculation results; (4) setting the number of new monitoring points, using the BORG algorithm for multiple iterations, and performing multi-objective optimization on the layout plan to obtain the Pareto optimal solution set. This method can provide data support for subsequent pipe burst positioning, but it has high requirements on computing power and cannot complete positioning quickly. Summary of the invention

[0007] In order to solve the above problems, the present invention provides a layout optimization method for monitoring points in urban water supply pipe networks. The method obtains the minimum monitoring flow of all pipe sections by retrieving the minimum response flow table instead of the traditional hydraulic model calculation, thereby solving the problem of excessive simulation calculation in the process of pipe section spatial partition optimization and improving the feedback speed and accuracy of the burst pipe locating work of the final layout plan.

[0008] A method for optimizing the layout of monitoring points in an urban water supply network, comprising:

[0009] Step 1: Obtain the pipe network data in the area to be optimized, wherein the pipe network data includes pipe segment data and node numbers of the monitoring points that can be arranged;

[0010] Step 2: construct a minimum response flow table based on the set of monitoring points with pressure less than the threshold pressure in the simulated pipe burst test and the corresponding node numbers and pipe section data. The minimum response flow table is used to retrieve and obtain the minimum monitoring flow of all pipe sections in any monitoring point arrangement scheme in the area to be optimized;

[0011] Step 3: Set the effectiveness evaluation functions of the three pipe network monitoring networks based on the pipe segment monitoring coverage, the non-uniformity of the pipe segment spatial partitioning, and the average similarity between pipe segments. For a given pipe network monitoring point arrangement plan, calculate the corresponding three effectiveness evaluation function values ​​based on the pipe network data obtained in step 1 and the minimum response flow table constructed in step 2;

[0012] Step 4: Set the number range of new monitoring points, iteratively calculate the monitoring point layout plan, perform multi-objective optimization on the iterative calculation results according to the three energy efficiency evaluation functions in step 3, and obtain the Pareto optimal solution set for determining the number and layout locations of the optimized monitoring points.

[0013] Preferably, the threshold pressure in step 2 is set after statistical analysis of historical monitoring data. The threshold pressure is the difference between the pressure value under normal flow of the pipe section and twice the standard deviation of the monitoring data in any time period. The alarm threshold is selected in accordance with the normal distribution, so that the final result of pipe burst positioning is more accurate.

[0014] Specifically, the steps of constructing the minimum response flow table in step 2 are as follows:

[0015] Step 2.1, perform a pipe burst simulation test on the pipe section to obtain a collection of monitoring points that issue alarms in the area where the pipe section is located;

[0016] Step 2.2, according to the monitoring point set obtained in step 2.1, select all monitoring points whose pressure is less than the threshold pressure, and calculate the minimum monitoring flow set corresponding to all the monitoring points;

[0017] Step 2.3, in the pipe burst simulation test, a non-alarm flow value is uniformly set for all monitoring points that have not issued an alarm, and the non-alarm flow value is the sum of the maximum pipe burst flow value of the pipe section and a constant;

[0018] Step 2.4, according to the node numbers corresponding to all monitoring points, fill in the minimum flow value or non-alarm flow value corresponding to each monitoring point into the table in turn to obtain the minimum response flow sequence of the pipe section;

[0019] Step 2.5, repeat steps 2.1-2.4 to obtain the minimum response flow sequence of all pipe sections, and sort the minimum response flow sequences of all pipe sections according to the size of the pipe section index number to obtain a minimum response flow table, wherein the row and column titles of the minimum response flow table are the node number and the pipe section index number, respectively.

[0020] Specifically, the mathematical expressions of step 2.2 and step 2.3 are:

[0021] When i∈S alarm_sensor (j,t):

[0022]

[0023] when

[0024]

[0025] Among them, S alarm_sensor (j,t)={i:I[p i (t,j),p a (t,i)]=1}} represents the response sensor sequence, p a (t,i) represents the threshold pressure of node i at time t, p i (t,j) represents the water pressure at node i of pipe segment j at time t, q j represents the minimum burst flow rate of pipe section j, Q jmax represents the maximum burst flow rate of pipe section j, C represents a constant, C>0;

[0026] Preferably, the performance evaluation function in step 3 is specifically expressed as:

[0027] Taking the maximization of pipe section monitoring coverage as the objective function, the monitoring point layout plan X is solved. Assuming that there are K sensors in the layout plan, the expression of the objective function is:

[0028]

[0029]

[0030]

[0031] Among them, F1 X represents the monitoring coverage of the pipe network, j represents the pipe section number of the pipe burst, L represents the total number of pipe sections, Q j represents the flow rate of the pipe section, l j Indicates the length of the pipe section, S j (t,X) represents the function of whether the pipe section j can be monitored by any monitoring point k in the given monitoring point arrangement X at time t. If it is monitored, it is 1, otherwise it is 0. j (t,k) represents the water pressure at monitoring point k when the maximum flow burst occurs in pipe section j at time t, p a (t,k) is the threshold pressure set for sensor k. j (t,k) <pa When (t, k), the monitoring point alarms, otherwise it does not alarm, α and β are weight coefficients;

[0032] The objective function is to minimize the non-uniformity of the spatial partition of the pipe segment. The expression of the objective function is:

[0033]

[0034]

[0035] Among them, F2 X Indicates the unevenness of the spatial partitioning of the pipe segment. represents the total length of the pipeline divided into the spatial partition where the monitoring point k is located, represents the average pipe length of the spatial partition where the K monitoring points are located, S pipe (t,k) represents the pipeline space partition;

[0036] The objective function is to minimize the average similarity between pipe sections. The expression of the objective function is:

[0037]

[0038] Among them, F3 X represents the average similarity between pipe segments, γ j,j′ Represents the similarity between pipe segment j and pipe segment j′.

[0039] Specifically, the expression of the pipeline space partition is:

[0040] S pipe (t,k)={j:f sensor (j,t)=k,j∈{1,…,L.}}

[0041]

[0042] Among them, q j is the simulated minimum burst flow of pipe segment j. The minimum response flow table is retrieved according to the index number of pipe segment j to obtain the minimum monitoring flow corresponding to the simulated minimum burst flow.

[0043] Specifically, the similarity between the pipe segment j and the pipe segment j′ is obtained as follows:

[0044] Step 3.1, there are L pipe sections in the pipe network, N is the total number of nodes, and the pressure value of node i under normal working conditions is P i The pressure value P when the pipe burst occurs i The difference between ′ i , we can get the residual r at each node:

[0045] r=(r1,r2,…,r i,…,r N ), i=1,2,…,N

[0046] Step 3.2: Calculate the sensitivity matrix S′ based on the residual r and the burst flow rate of pipe section j:

[0047]

[0048] in, q j is the simulated minimum burst flow rate of pipe segment j, and the minimum response flow table is retrieved according to the index number of pipe segment j to obtain the minimum monitoring flow rate corresponding to the simulated minimum burst flow rate;

[0049] Step 3.3: Normalize the sensitivity matrix S′:

[0050]

[0051] in, Represents the maximum value of the jth row in the S matrix, s′ jmin Represents the minimum value of the jth row in the S matrix;

[0052] Step 3.4: Calculate the similarity between pipe segment j and pipe segment j′:

[0053]

[0054] Among them, s j It represents the impact of pipe burst on the water pressure change of each monitoring point, which is composed of each row vector of the normalized sensitivity matrix S′, that is, s j =(s j1 ,s j2 ,…,s jN ).

[0055] Preferably, the step 4 adopts the multi-objective optimization algorithm Borg MOEA for optimization calculation, and the number of iterations is set between 5000-15000. By setting the ε-box dominance archive, the search diversity is guaranteed and it is conducive to getting rid of the local optimum; at the same time, the ε-progress mechanism is introduced into the algorithm, so that the algorithm has the function of search stagnation and restart; it has a variety of recombination operation operators, and a feedback mechanism can be established according to the proportion of offspring generated by different operators. The application proportion of different operators can be adjusted online according to specific problems to adapt to the solution of problems in different fields.

[0056] Preferably, the Pareto optimal solution set of step 4 generates a corresponding set of monitoring point layout schemes based on the continuously increasing number of monitoring points, and multi-objective optimization is performed on each scheme in the set of monitoring point layout schemes to obtain the corresponding Pareto value. When the Pareto value no longer increases, the increase in the number of monitoring points is stopped, and the monitoring point layout scheme corresponding to the highest Pareto value is output as the final optimized layout scheme.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] (1) The three optimization objectives of pipe segment monitoring coverage, spatial partitioning non-uniformity of pipe segments, and average similarity between pipe segments are proposed to optimize the layout of monitoring points in the area, so that the feedback speed of pipe burst location work of the final layout optimization plan is faster and more accurate.

[0059] (2) Based on the optimization objective of the uneven spatial partitioning of pipe sections, it is proposed to obtain the minimum monitoring flow of all pipe sections by constructing a minimum response flow table to replace the traditional hydraulic model calculation, thereby solving the problem of excessive simulation calculation in the process of optimizing the spatial partitioning of pipe sections. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 The pipe network topology diagram of the area to be optimized in this embodiment;

[0061] Figure 2 A schematic diagram of a flow chart of a method for optimizing the arrangement of monitoring points in a city water supply network provided by the present invention;

[0062] Figure 3 It is a distribution diagram of the multi-objective optimization Pareto solution set in this embodiment;

[0063] Figure 4 The figure is a schematic diagram of the monitoring point arrangement after being optimized by the method provided by the present invention. DETAILED DESCRIPTION

[0064] like Figure 1 As shown, this is a pipe network topology diagram of an area to be optimized provided in this embodiment. The basic data of the pipe network includes: 550 water storage nodes, 735 pipe sections, 7 existing pressure monitoring points, and the node indexes (numbers) are: 436, 2303, 1394, 6383, 87, 6370 and 1026.

[0065] like Figure 2 As shown, the layout of monitoring points in the area is optimized by a layout optimization method for monitoring points in an urban water supply network, including:

[0066] Step 1: Obtain the pipe network data in the area to be optimized, including pipe segment data and node numbers where monitoring points can be arranged.

[0067] Perform pipe burst simulation to obtain the maximum burst flow rate Q of each pipe section jmax :Let each pipe section perform pipe burst simulation under complete fracture in turn. Set a sufficiently large initial water demand value before simulation. The actual water demand after iteration of hydraulic simulation software is the maximum pipe burst flow of the pipe section. The maximum pipe burst flow of each pipe section is shown in Table 1 (735 lines in total, part of the excerpt):

[0068] Table 1 Maximum burst flow rate of each pipe section

[0069]

[0070] The water heads of each node in the normal state are shown in Table 2 (a total of 550 columns, part of which is extracted):

[0071] Table 2 Water head at each node in normal state

[0072]

[0073] Each pipe section is subjected to pipe burst simulation at the maximum burst flow rate in turn to obtain the pressure of each node, as shown in Table 3 (the table size is 735 rows and 550 columns, which is a truncated part):

[0074] Table 3 Pressure of each node after the pipe burst at the maximum burst flow rate in each pipe section

[0075]

[0076] In this case, 1m 3 / h step size to increase the burst pipe flow rate, simulating 0-Q jmax Pipe burst events: 9,712,750 pipe burst events were simulated for 735 pipe sections; taking pipe section 1 as an example, as shown in Table 4 (a total of 86,846 rows, part of which is extracted):

[0077] Table 4 Pipe sections corresponding to 0-Q jmax Pressure at each node of pipe burst event

[0078]

[0079] Step 2: According to the set of monitoring points with pressure less than the threshold pressure in the simulated pipe burst test and the corresponding node numbers and pipe section data, a minimum response flow table is constructed, and the minimum response flow table is used to retrieve and obtain the minimum monitoring flow of all pipe sections in any monitoring point arrangement scheme in the area to be optimized;

[0080] The monitoring point layout scheme can be composed of any node combination, so the number of layout schemes is Where N is the total number of nodes, L is the number of pipe sections, and K is the number of sensors that need to be deployed. When the water supply network is large, there are tens of millions of potential layout options, which also makes optimization work in large areas very difficult.

[0081] The present invention proposes a minimum response flow table, which records the corresponding pipe section burst flow when the pipe section j burst node i is lower than the set threshold pressure Use Tab min_alarm It is expressed as follows:

[0082]

[0083] In order to determine the threshold p i (t,q j )>p a The burst flow rate of pipe section j at time (t,i) The enumeration method needs to be applied. The specific process is: the burst flow rate q of pipe section j j With a certain step size Δq j Until q j Equal to Q jmax .

[0084] When p i (t,q j ) is lower than the alarm pressure p a (t,i), burst flow The nodes that never give an alarm record a value (Q jmax +C), where C is a constant, indicating that no matter how big the pipe burst event is, it will not cause an alarm at this node.

[0085] The mathematical expression of the above is:

[0086] When i∈S alarm_sensor (j,t):

[0087]

[0088] when

[0089]

[0090] Among them, S alarm_sensor (j,t)={i:I[p i (t,j),p a (t,i)]=1}} represents the response sensor sequence, p a (t,i) represents the threshold pressure of node i at time t, p i (t,j) represents the water pressure at node i of pipe segment j at time t, q jrepresents the minimum burst flow rate of pipe section j, Q jmax represents the maximum burst flow rate of pipe section j, C represents a constant, C>0;

[0091] The method for constructing the minimum response flow table includes the following specific steps:

[0092] Step 2.1, perform a pipe burst simulation test on the pipe section to obtain a collection of monitoring points that issue alarms in the area where the pipe section is located;

[0093] Step 2.2, according to the monitoring point set obtained in step 2.1, select all monitoring points whose pressure is less than the threshold pressure, and calculate the minimum monitoring flow set corresponding to all the monitoring points;

[0094] Step 2.3, in the pipe burst simulation test, a non-alarm flow value is uniformly set for all monitoring points that have not issued an alarm, and the non-alarm flow value is the sum of the maximum pipe burst flow value of the pipe section and a constant;

[0095] Step 2.4, according to the node numbers corresponding to all monitoring points, fill in the minimum flow value or non-alarm flow value corresponding to each monitoring point into the table in turn to obtain the minimum response flow sequence of the pipe section;

[0096] Step 2.5, repeat steps 2.1-2.4 to obtain the minimum response flow sequence of all pipe sections, and sort the minimum response flow sequences of all pipe sections according to the size of the pipe section index number to obtain a minimum response flow table, wherein the row and column titles of the minimum response flow table are the node number and the pipe section index number, respectively.

[0097] After the minimum response flow table is established, for different monitoring point layout schemes, there is no need to perform hydraulic simulation calculations, but directly select Tab min_alarm The corresponding rows and columns are used to calculate the objective function value of the spatial partitioning unevenness of the pipe section, so that the optimization goal can be achieved.

[0098] The minimum response flow table is shown in Table 5 (the table size is 735 rows and 550 columns, which is a truncated part):

[0099] Table 5 Minimum response flow table

[0100]

[0101] Step 3: Set the effectiveness evaluation functions of the three pipe network monitoring networks based on the pipe segment monitoring coverage, the non-uniformity of the pipe segment spatial partitioning, and the average similarity between pipe segments. For a given pipe network monitoring point arrangement plan, calculate the corresponding three effectiveness evaluation function values ​​based on the pipe network data obtained in step 1 and the minimum response flow table constructed in step 2;

[0102] In a pipe network with K sensors, at time t, when the flow rate of pipe segment j gradually increases from a small flow rate, the pressure drop at different sensor locations is different due to the pipe burst at a specific flow rate in pipe segment j. The sensor that alarms first is the most sensitive to the pipe burst in this pipe segment. If sensor k alarms first, pipe segment j is divided into the area where sensor k is located. By calling f on L pipes in sequence, sensor (j, t) function, simulates the burst of each pipe section, divides the pipe sections into K regions one by one, and realizes the spatial partition of the pipe sections. Its mathematical expression is:

[0103] S pipe (t,k)={j:f sensor (j,t)=k,j∈{1,…,L.}}

[0104]

[0105] Among them, q j is the simulated minimum burst flow of pipe segment j. The minimum response flow table is retrieved according to the index number of pipe segment j to obtain the minimum monitoring flow corresponding to the simulated minimum burst flow.

[0106] The specific expressions of the three performance evaluation functions are:

[0107] Taking the maximization of pipe section monitoring coverage as the objective function, the monitoring point layout plan X is solved. Assuming that there are K sensors in the layout plan, the expression of the objective function is:

[0108]

[0109]

[0110]

[0111] Among them, F1 X represents the monitoring coverage of the pipe network, j represents the pipe section number of the pipe burst, L represents the total number of pipe sections, Q j represents the flow rate of the pipe section, l j Indicates the length of the pipe section, S j (t,X) represents the function of whether the pipe section j can be monitored by any monitoring point k in the given monitoring point arrangement X at time t. If it is monitored, it is 1, otherwise it is 0. j (t,k) represents the water pressure at monitoring point k when the maximum flow burst occurs in pipe section j at time t, p a (t,k) is the threshold pressure set for sensor k. j (t,k) <p a When (t, k), the monitoring point alarms, otherwise it does not alarm, α and β are weight coefficients;

[0112] The objective function is to minimize the non-uniformity of the spatial partition of the pipe segment. The expression of the objective function is:

[0113]

[0114]

[0115] Among them, F2 X Indicates the unevenness of the spatial partitioning of the pipe segment. represents the total length of the pipeline divided into the spatial partition where the monitoring point k is located, represents the average pipe length of the spatial partition where the K monitoring points are located, S pipe (t,k) represents the pipeline space partition;

[0116] The objective function is to minimize the average similarity between pipe sections. The expression of the objective function is:

[0117]

[0118] Among them, F3 X represents the average similarity between pipe segments, γ j,j′ Represents the similarity between pipe segment j and pipe segment j′.

[0119] The steps for obtaining the similarity between pipe segment j and pipe segment j′ are as follows:

[0120] There are L pipe sections in the pipe network, N is the total number of nodes, and the pressure value of node i under normal working conditions is P i The pressure value P when the pipe burst occurs i The difference between ′ i , we can get the residual r at each node:

[0121] r=(r1,r2,…,r i ,…,r N ), i=1,2,…,N.

[0122] Define the sensitivity matrix S′:

[0123]

[0124] in

[0125] Normalize the sensitivity matrix:

[0126]

[0127] in,

[0128] s′ jmax Represents the maximum value of the jth row in the S matrix, s′ jminRepresents the minimum value of the j-th row in the S matrix.

[0129] Among them, i represents the node number where the pressure measuring point is located, j represents the pipe section number where the pipe burst occurs, and q j is the burst flow rate, and each row constitutes a sensitivity vector s j =(s j1 ,s j2 ,…,s jN ), represents the impact of pipe burst on the water pressure change of each monitoring point. j,j′ To measure their similarity, the expression is:

[0130]

[0131] γ j,j′ The larger the value of is, the more similar the pressure changes caused by pipe segment j and pipe segment j′ at each pressure measuring point are, which also means that it is more difficult to distinguish whether the pipe burst occurs in pipe segment j or pipe segment j′.

[0132] The difference between the water head of each node in the normal state and the pressure of each node after the pipe section bursts at the maximum burst flow is calculated. The residual matrix is ​​shown in Table 6 (the matrix size is 735 rows and 550 columns, which is a truncated part):

[0133] Table 6 Residual matrix

[0134]

[0135] The pressure sensitive matrix is ​​shown in the following table (the matrix size is 735 rows and 550 columns, which is a truncated part):

[0136] Table 7 Pressure Sensitivity Matrix

[0137]

[0138] Step 4: Set the number range of new monitoring points, iteratively calculate the monitoring point layout plan, perform multi-objective optimization on the iterative calculation results according to the three energy efficiency evaluation functions in step 3, and obtain the Pareto optimal solution set for determining the number and layout locations of the optimized monitoring points:

[0139] The multi-objective optimization algorithm Borg MOEA is used for optimization calculation, and the number of monitoring points n is increased gradually to obtain the solution set after each optimization. The multi-objective optimization Pareto frontier solution set is shown in Table 8 (a total of 88 solutions, part of which is extracted):

[0140] Table 8 Multi-objective optimization Pareto frontier solution set

[0141]

[0142] Draw the corresponding solution set diagram according to the Pareto frontier solution set, such as Figure 3 shown.

[0143] When the number of monitoring points n is increased between 1 and 29, the optimization amplitude increases fastest. As the number of monitoring points increases, the optimization amplitude gradually decreases. When the monitoring point n=41, the optimization amplitude is already small, and it can be considered that a good optimization effect has been achieved.

[0144] From the change trend of the entire Pareto solution set, when n = 41, the difference in the function values ​​of each solution on different objectives is not large. Each solution can achieve good optimization results in each objective dimension and has good convergence. The monitoring point layout diagram with n = 41 is drawn (the solution set number in Table 8 is 83), as shown in Figure 4 As shown, the black dots are monitoring points.

[0145] Through the optimized monitoring point arrangement scheme of the present invention, the pipe segment space partition can be used to quickly locate the pipe burst, and the uniformity of the pipe segment space partition will affect the pipe segment search range when locating the pipe burst. Assuming that there are 24 pipe segments in an area, which are divided into 3 partitions respectively, and the probability of pipe burst in each pipe segment is set to be equal, and there are three monitoring points, the expression of the expected number of searched pipe segments is as follows:

[0146]

[0147] In the above formula, K is the number of pipe space partitions, i.e. the number of sensors; A is the number of alarm sensors; Indicates the number of times each pipe segment space partition is selected into the partition combination; M is the total number of alarm partition combinations; N m is the total number of pipe sections in the mth partition combination; is the pipe burst probability of the mth partition combination.

[0148] Assume that the number of partitioned pipe sections in layout plan 1 is 2, 7, and 15 (the pipe section space is unevenly partitioned), and the number of partitioned pipe sections in layout plan 2 is 8, 8, and 8 (the pipe section space is evenly partitioned):

[0149] 1. When only one monitoring point alarms, the number of search pipe sections for layout plan 1 is 11.58, and the number of search pipe sections for layout plan 2 is 8.

[0150] 2. When two monitoring points alarm, the number of search pipe sections for layout plan 1 is 17.79, and the number of search pipe sections for layout plan 2 is 16.

[0151] 3. When all three monitoring points are in alarm, the number of search pipe sections for layout plan 1 is 24, and the number of search pipe sections for layout plan 2 is 24.

[0152] Based on the above difference in the number of searched pipe sections, it can be seen that when the spatial partitioning of pipe sections in the area is more uniform, the number of pipe sections that need to be searched will be reduced accordingly when a pipe burst occurs, thereby achieving rapid feedback of pipe burst location.

Claims

1. A method for optimizing the layout of monitoring points in an urban water supply network, characterized in that: include: Step 1: Obtain the pipe network data in the area to be optimized, wherein the pipe network data includes pipe segment data and node numbers of the monitoring points that can be arranged; Step 2: According to the set of monitoring points with pressure less than the threshold pressure in the simulated pipe burst test and the corresponding node numbers and pipe section data, a minimum response flow table is constructed. The minimum response flow table is used to retrieve and obtain the minimum monitoring flow of all pipe sections in any monitoring point arrangement scheme in the area to be optimized. The specific steps for constructing the minimum response flow table are as follows: Step 2.1, perform a pipe burst simulation test on the pipe section to obtain a collection of monitoring points that issue alarms in the area where the pipe section is located; Step 2.2, according to the monitoring point set obtained in step 2.1, select all monitoring points whose pressure is less than the threshold pressure, and calculate the minimum monitoring flow set corresponding to all the monitoring points; Step 2.3, in the pipe burst simulation test, a non-alarm flow value is uniformly set for all monitoring points that have not issued an alarm, and the non-alarm flow value is the sum of the maximum pipe burst flow value of the pipe section and a constant; Step 2.4, according to the node numbers corresponding to all monitoring points, fill in the minimum flow value or non-alarm flow value corresponding to each monitoring point into the table in turn to obtain the minimum response flow sequence of the pipe section; Step 2.5, repeat steps 2.1-2.4 to obtain the minimum response flow sequence of all pipe sections, and sort the minimum response flow sequences of all pipe sections according to the size of the pipe section index number to obtain a minimum response flow table, wherein the row and column titles of the minimum response flow table are the node number and the pipe section index number respectively; Step 3: Set the performance evaluation functions of the three pipe network monitoring networks based on the pipe monitoring coverage, the non-uniformity of the pipe spatial partition and the average similarity between pipe segments. For a given pipe network monitoring point layout, the corresponding three performance evaluation function values ​​are calculated based on the pipe network data obtained in step 1 and the minimum response flow table constructed in step 2. The specific expressions are: Taking the maximization of pipe section monitoring coverage as the objective function, the monitoring point layout plan X is solved. Assuming that there are K sensors in the layout plan, the expression of the objective function is: Among them, F1 X represents the monitoring coverage of the pipe network, j represents the pipe section number of the pipe burst, L represents the total number of pipe sections, Q j represents the flow rate of the pipe section, l j Indicates the length of the pipe section, S j (t,X) represents the function of whether the pipe section j can be monitored by any monitoring point k in the given monitoring point arrangement X at time t. If it is monitored, it is 1, otherwise it is 0. j (t,k) represents the water pressure at monitoring point k when the maximum flow burst occurs in pipe section j at time t, p a (t,k) is the threshold pressure set for sensor k. j (t,k) <p a When (t, k), the monitoring point alarms, otherwise it does not alarm, α and β are weight coefficients; The objective function is to minimize the non-uniformity of the spatial partition of the pipe segment. The expression of the objective function is: Among them, F2 X Indicates the unevenness of the spatial partitioning of the pipe segment. represents the total length of the pipeline divided into the spatial partition where the monitoring point k is located, represents the average pipe length of the spatial partition where the K monitoring points are located, S pipe (t,k) represents the pipeline space partition; The objective function is to minimize the average similarity between pipe sections. The expression of the objective function is: Among them, F3 X represents the average similarity between pipe segments, y j,j′ represents the similarity between pipe segment j and pipe segment j′; Step 4: Set the number range of new monitoring points, iteratively calculate the monitoring point layout plan, perform multi-objective optimization on the iterative calculation results according to the three energy efficiency evaluation functions in step 3, and obtain the Pareto optimal solution set for determining the number and layout locations of the optimized monitoring points.

2. The method for optimizing the layout of monitoring points in a city water supply network according to claim 1, characterized in that: The threshold pressure in step 2 is set after statistical analysis of historical monitoring data. The threshold pressure is the difference between the pressure value under normal flow of the pipeline section and twice the standard deviation of the monitoring data in any time period.

3. The method for optimizing the layout of monitoring points in a city water supply network according to claim 1, characterized in that: The expression of the pipeline space partition is: S pipe (t,k)={j:f sensor (j,t)=k,j∈{1,…,L.}} Among them, q j is the simulated minimum burst flow of pipe segment j. The minimum response flow table is retrieved according to the index number of pipe segment j to obtain the minimum monitoring flow corresponding to the simulated minimum burst flow.

4. The method for optimizing the layout of monitoring points in a city water supply network according to claim 1, characterized in that: The similarity between the pipe segment j and the pipe segment j′ is obtained by the following steps: Step 3.1, there are L pipe sections in the pipe network, N is the total number of nodes, and the pressure value of node i under normal working conditions is P i The pressure value P when the pipe burst occurs i The difference between ′ i , we can get the residual r at each node: r=(r1,r2,…,r i ,…,r N ),i=1,2,…,N Step 3.2: Calculate the sensitivity matrix S′ based on the residual r and the burst flow rate of pipe section j: in, Among them, q j is the simulated minimum burst flow rate of pipe segment j, and the minimum response flow table is retrieved according to the index number of pipe segment j to obtain the minimum monitoring flow rate corresponding to the simulated minimum burst flow rate; Step 3.3: Normalize the sensitivity matrix S′: in, s j ' max represents the maximum value of the jth row in the S matrix, s j ' min Represents the minimum value of the jth row in the S matrix; Step 3.4: Calculate the similarity between pipe segment j and pipe segment j′: Among them, s j It represents the impact of pipe burst on the water pressure change of each monitoring point, which is composed of each row vector of the normalized sensitivity matrix S′, that is, s j =(s j1 ,s j2 ,…,s jN ).

5. The method for optimizing the layout of monitoring points in a city water supply network according to claim 1, characterized in that: The step 4 uses the multi-objective optimization algorithm Borg MOEA to perform optimization calculation, and the number of iterations is set between 5000 and 15000.

6. The method for optimizing the layout of monitoring points in a city water supply network according to claim 1, characterized in that: The Pareto optimal solution set of step 4 generates a corresponding set of monitoring point layout schemes based on the continuously increasing number of monitoring points, and multi-objective optimization is performed on each scheme in the set of monitoring point layout schemes to obtain the corresponding Pareto value. When the Pareto value no longer increases, the increase in the number of monitoring points is stopped, and the monitoring point layout scheme corresponding to the highest Pareto value is output as the final optimized layout scheme.