Machine learning remote self-adaptive water supply control working method and system

Through machine learning remote adaptive water supply control method, a mapping model of water supply flow and pressure is established, the head energy loss is calculated, and the output energy sample node is generated according to the law of conservation of energy, which solves the problem of unstable convergence speed of the GDPSO algorithm and achieves more efficient Water supply system control and energy efficiency optimization.

CN118885901AActive Publication Date: 2024-11-01CHONGQING INST OF MECHANICAL & ELECTRICAL ENG

Patent Information

Application Number
CN202410949930.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-16
Publication Date
2024-11-01
Estimated Expiration
2044-07-16

AI Technical Summary

Technical Problem

In the existing water supply control system, the convergence rate of the GDPSO algorithm is unstable, making it difficult to achieve real-time water use status and effective control of water supply network data, resulting in low water supply allocation efficiency.

Method used

Using machine learning remote adaptive water supply control method, by obtaining the historical period data of the water supply system, a mapping model of water supply flow and pressure is established, the nonlinear input vector is transformed into a high-dimensional space, the head energy loss is calculated, and the output energy is generated according to the law of conservation of energy Sample nodes realize stable control of the water supply system.

Benefits of technology

It improves the accuracy and control accuracy of water supply pressure prediction, simplifies model complexity, optimizes system energy efficiency, reduces unnecessary energy consumption, and improves the stability and reliability of the water supply system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118885901A_ABST
    Figure CN118885901A_ABST
Patent Text Reader

Abstract

The invention provides a machine learning remote self-adaptive water supply control working method and system, and the method comprises the following steps: S1, obtaining the water supply flow data of a historical period of a water supply system, and building a mapping model of the water supply flow and the water supply pressure; and S2, the nonlinear input vectors of the mapping models of the two adjacent nodes are sequentially converted into linear functions in a high-dimensional space, energy mean value parameters are calculated according to the energy loss of the two water heads, the method can effectively prevent the algorithm from falling into local optimal calculation, and the influence of global deviation is prevented. And S3, evaluating the difference between the current water supply state and the target state through iterative calculation, and generating output energy sample nodes for the mean value parameters of the water supply nodes of the identified water supply system according to the law of conservation of energy.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of intelligent water supply, and in particular to a machine learning remote adaptive water supply control working method and system. Background Art

[0002] In the process of urbanization construction, the effective allocation of residential water is a livelihood project related to thousands of households. The water supply entity needs to obtain real-time water use status and the use and consumption data of the water supply network, and effectively manage and control it through remote terminals. This requires comprehensive data analysis of water body loss, water use time, and water consumption. In CN113723708A, the Gaussian particle swarm perturbation optimization algorithm is used to perform optimization operations to predict water consumption, and then allocate water supply equipment for water supply control. Among them, the convergence speed of the GDPSO algorithm will be affected by multiple factors such as initial parameter settings and problem characteristics, resulting in unstable convergence speed. This urgently requires technical personnel in this field to solve the corresponding technical problems. Summary of the invention

[0003] The present invention aims to at least solve the technical problems existing in the prior art, and in particular innovatively proposes a machine learning remote adaptive water supply control working method.

[0004] In order to achieve the above-mentioned object of the present invention, the present invention provides a machine learning remote adaptive water supply control working method, which is characterized by comprising:

[0005] S1, obtaining water supply flow data of the historical period of the water supply system and establishing a mapping model of water supply flow and water supply pressure;

[0006] S2, sequentially transforms the nonlinear input vectors of the mapping model of two adjacent nodes into linear functions in a high-dimensional space, and calculates the energy mean parameter based on the two head energy losses;

[0007] S3, through iterative calculation, evaluates the gap between the current water supply state and the target state, and generates output energy sample nodes for the identified water supply system based on the mean parameters of the water supply nodes according to the law of conservation of energy.

[0008] Preferably, the above technical solution is that S1 comprises:

[0009] S1-1, based on the water supply flow data of m nodes in the water supply system, form the morning water supply flow data vector Q mor =[q1,q2,...,q m ] T , noon water supply flow data vector Q aft =[q1′,q2′,...,q′ m ] T , night water supply flow data vector Qeve =[q1″,q2″,...,q′ m ′] T , then the mean column vector of water supply flow data is,

[0010]

[0011] The water supply pressure data is obtained at the corresponding water supply flow node, forming a water supply pressure data vector P = [p1, p2, ..., p m ].

[0012] Preferably, the above technical solution, S1 further comprises:

[0013] S1-2, the water supply system acquisition node establishes a mapping model of the water supply flow and water supply pressure throughout the day, and establishes a training subset by performing relationship calculations; according to the flow matrix FM (Flow Matrix,) of the water supply system, the conservation formula related to the flow and pressure of all nodes in the water supply system is established

[0014]

[0015] Among them, f1 is the quantitative function of the theoretical pressure adjustment value ΔP, p out is the output pressure of the water supply system, p client is the user-side output pressure, Δq m is the output flow regulation value of each node in the water supply system, ΔQ is the column vector of the output flow regulation values ​​of all nodes in the water supply system, the superscript T is the transpose, P m-1 is the actual pressure value of the m-1th node, P m is the actual pressure value of the mth node, P m-1 -P m is the actual pressure difference between the two nodes, ρ is the density of the flowing medium, v m is the flow velocity at the mth node, v m-1 is the flow velocity at the m-1th node, g is the gravitational acceleration of the flowing medium, h is the height difference between the m and m-1 nodes, λ is the friction coefficient, d is the pipe radius, L is the pipe length between the m and m-1 nodes, is the average flow velocity between nodes m and m-1, is the actual flow matrix of all nodes m, and Q is the mean column vector of water supply flow data. The flow distribution of the nodes is balanced with the sum of the expected regulation values. It is a universal quantifier.

[0016] In the above technical solution, preferably, S2 comprises:

[0017] According to the conservation formula, the nonlinear input vectors of the mapping model of two adjacent nodes are transformed into linear functions in high-dimensional space in turn to form a training subset of pressure. For the pressure mapping function P′→φ(P′), φ(P′) is a high-dimensional linear function of the input vector P′. The training subset is U, and the training subset U is to find x pressure optimal samples according to each time period.

[0018] In the conservation formula of each node flow, the corresponding prediction model Q′ is established.

[0019] Where Q k is the leakage vector in the process of traffic output for each node,

[0020] The training subset for each node flow is V, and the training subset V is to find y flow optimization samples according to each time period.

[0021] The sample probability is calculated based on the training subsets of U and V, so as to optimize the pressure and flow of the corresponding nodes.

[0022] Preferably, the above technical solution, S2 further includes:

[0023] In the water supply system, the pressure and flow rate from node m-1 to node m need to achieve theoretical energy conservation. In the corresponding pressure optimization period sample x, the corresponding pressure values ​​of nodes m-1 and m are obtained. In the corresponding flow optimization period sample y, the corresponding flow values ​​of nodes m-1 and m are obtained.

[0024]

[0025] Among them, h m and h m-1 are the water supply heights of node m and node m-1 respectively, is the pressure value of node m-1 in sample x during the pressure optimization period, is the pressure value of node m in sample x during the pressure optimization period, is the flow value of the m-1 node in sample y during the flow optimization period, is the flow value of node m in sample y during the flow optimization period, is the water head energy loss of the m-1 node in the water supply system pipeline friction, is the head energy loss of the friction node m in the water supply system pipe.

[0026] In the above technical solution, S3 preferably includes:

[0027] Calculate the energy mean parameter w for the two head energy losses,

[0028] σ is the regulation weight. According to the mean parameters obtained in different time periods, the energy regulation output is determined to meet the stability of the pressure and flow of the water supply system.

[0029] By calculating the energy mean parameter w for every two adjacent nodes and evaluating the gap between the current water supply state and the target state through iterative calculation, the identified water supply system generates output energy sample nodes based on the mean parameter of the water supply node according to the law of conservation of energy.

[0030] The present invention also discloses a machine learning remote adaptive water supply control working system, comprising:

[0031] A data acquisition module, which acquires the water supply flow data of the historical period of the water supply system and establishes a mapping model of the water supply flow and water supply pressure;

[0032] The mean value calculation module sequentially transforms the nonlinear input vectors of the mapping model of two adjacent nodes into linear functions in a high-dimensional space, and calculates the energy mean parameter according to the two head energy losses;

[0033] The water supply output module evaluates the gap between the current water supply state and the target state through iterative calculation, and generates output energy sample nodes for the mean parameters of the water supply nodes of the identified water supply system according to the law of conservation of energy.

[0034] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0035] By establishing a mapping model between water supply flow and water supply pressure, the required water supply pressure under different water supply flow rates can be predicted more accurately, thereby improving the control accuracy; by converting the nonlinear relationship into a linear function in a high-dimensional space, the complexity of the model is simplified, allowing the control algorithm to respond more quickly to changes in the water supply system. By calculating the energy mean parameter, the energy loss of the water head in the water supply system is quantified, thereby optimizing the energy efficiency of the system.

[0036] Generating output energy sample nodes according to the law of conservation of energy helps to achieve efficient use of energy and reduce unnecessary energy consumption during system control. This control strategy helps to reduce system fluctuations and improve the reliability and stability of water supply. The data analysis, model building and iterative optimization process involved in this sequence of steps reflects the intelligent characteristics of the water supply system control strategy.

[0037] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0039] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0040] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0041] like Figure 1 As shown, the present invention discloses a machine learning remote adaptive water supply control working method, comprising the following steps:

[0042] S1, obtaining water supply flow data of the historical period of the water supply system and establishing a mapping model of water supply flow and water supply pressure;

[0043] S2, transforms the nonlinear input vectors of the mapping model of two adjacent nodes into linear functions in high-dimensional space in turn, and calculates the energy mean parameter according to the two head energy losses. This method can effectively prevent the algorithm from falling into local optimal calculation and prevent the influence of global deviation.

[0044] S3, through iterative calculation, evaluates the gap between the current water supply state and the target state, and generates output energy sample nodes for the identified water supply system based on the mean parameters of the water supply nodes according to the law of conservation of energy.

[0045] This strategy combines data analysis, nonlinear transformations, physical laws, and iterative optimization to help improve the stability and efficiency of water supply systems.

[0046] Preferably, the above technical solution is that S1 comprises:

[0047] S1-1, based on the water supply flow data of m nodes in the water supply system, form the morning water supply flow data vector Q mor =[q1,q2,...,q m ] T , noon water supply flow data vector Q aft =[q1′,q2′,...,q′ m ] T , night water supply flow data vector Q eve =[q1″,q2″,...,q′ m ′] T , then the mean column vector of water supply flow data is,

[0048]

[0049] The water supply pressure data is obtained at the corresponding water supply flow node, forming a water supply pressure data vector P = [p1, p2, ..., p m ], in order to make the data obtained by the water supply system more accurate, the water supply flow data and water supply pressure data in the morning, afternoon and evening are collected.

[0050] S1-2, the water supply system acquisition node establishes a mapping model of the water supply flow and water supply pressure throughout the day, and establishes a training subset by performing relationship calculations; according to the flow matrix FM (Flow Matrix,) of the water supply system, the conservation formula related to the flow and pressure of all nodes in the water supply system is established

[0051]

[0052] Among them, f1 is the quantitative function of the theoretical pressure adjustment value ΔP, p out is the output pressure of the water supply system, p client is the user-side output pressure, Δq m is the output flow regulation value of each node in the water supply system, ΔQ is the column vector of the output flow regulation values ​​of all nodes in the water supply system, and the superscript T is the transpose. The loss caused by leakage during the operation of the water supply system is also one of the factors, but it is not considered in this application. m-1 is the actual pressure value of the m-1th node, P m is the actual pressure value of the mth node, P m-1 -P m is the actual pressure difference between the two nodes, ρ is the density of the flowing medium, v m is the flow velocity at the mth node, v m-1 is the flow velocity at the m-1th node, g is the gravitational acceleration of the flowing medium, h is the height difference between the m and m-1 nodes, λ is the friction coefficient, d is the pipe radius, L is the pipe length between the m and m-1 nodes, is the average flow velocity between nodes m and m-1, is the actual flow matrix of all nodes m, and Q is the mean column vector of water supply flow data. The flow distribution of the nodes is balanced with the sum of the expected regulation values. It is a universal quantifier.

[0053] The flow medium is water or other liquid.

[0054] In the above technical solution, preferably, S2 comprises:

[0055] According to the conservation formula, the nonlinear input vectors of the mapping model of two adjacent nodes are transformed into linear functions in high-dimensional space in turn to form a training subset of pressure. For the pressure mapping function P′→φ(P′), φ(P′) is a high-dimensional linear function of the input vector P′. The training subset is U, and the training subset U is to find x pressure optimal samples according to each time period.

[0056] In the conservation formula of each node flow, the corresponding prediction model Q′ is established.

[0057] Where Q k is the leakage vector in the process of traffic output for each node,

[0058] The training subset for each node flow is V, and the training subset V is to find y flow optimization samples according to each time period.

[0059] The sample probability is calculated based on the training subsets of U and V, so as to optimize the pressure and flow of the corresponding nodes;

[0060] In the water supply system, the pressure and flow rate from node m-1 to node m need to achieve theoretical energy conservation. In the corresponding pressure optimization period sample x, the corresponding pressure values ​​of nodes m-1 and m are obtained. In the corresponding flow optimization period sample y, the corresponding flow values ​​of nodes m-1 and m are obtained.

[0061]

[0062] Among them, h m and h m-1 are the water supply heights of node m and node m-1 respectively, is the pressure value of node m-1 in sample x during the pressure optimization period, is the pressure value of node m in sample x during the pressure optimization period, is the flow value of the m-1 node in sample y during the flow optimization period, is the flow value of node m in sample y during the flow optimization period, is the head energy loss of the friction node m-1 in the water supply system pipeline, is the head energy loss of the friction node m in the water supply system pipe.

[0063] Calculate the energy mean parameter w for the two head energy losses,

[0064] σ is the regulation weight. According to the mean parameters obtained in different time periods, the energy regulation output is determined to meet the stability of the pressure and flow of the water supply system.

[0065] By calculating the energy mean parameter w for every two adjacent nodes and evaluating the gap between the current water supply state and the target state through iterative calculation, the identified water supply system generates output energy sample nodes for the mean parameter of the water supply node according to the law of conservation of energy. This method is applied to a water supply network and its performance is tested. The results show that this method can better allocate the water supply control system and improve the efficiency and accuracy of pressure and flow detection, thereby applying better pressure and flow at the corresponding nodes to meet the user experience of the end user.

[0066] Through real-time monitoring and data analysis, combined with prediction models and the law of conservation of energy, the pressure and flow in the water supply system can be controlled more accurately. This helps to reduce system fluctuations and improve the stability and reliability of water supply. Based on the energy conservation formula and the calculation of head energy loss, the pressure and flow of each node in the water supply system can be accurately adjusted, thereby reducing unnecessary energy consumption. By optimizing the adjustment weights and energy mean parameters, the energy efficiency of the system can be improved. This strategy finds the preferred samples of pressure and flow at different time periods, and performs optimal calculations based on these samples, so that the system can adaptively respond to water demand and changes in external conditions at different time periods. The large amount of data collected and analyzed provides a valuable data basis for the long-term planning and management of the water supply system.

[0067] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A machine learning remote adaptive water supply control working method, characterized in that: include: S1, obtaining water supply flow data of the historical period of the water supply system and establishing a mapping model of water supply flow and water supply pressure; S2, sequentially transforms the nonlinear input vectors of the mapping model of two adjacent nodes into linear functions in a high-dimensional space, and calculates the energy mean parameter based on the two head energy losses; S3, through iterative calculation, evaluates the gap between the current water supply state and the target state, and generates output energy sample nodes for the identified water supply system based on the mean parameters of the water supply nodes according to the law of conservation of energy.

2. The machine learning remote adaptive water supply control working method according to claim 1 is characterized in that: The S1 includes: S1-1, based on the water supply flow data of m nodes in the water supply system, form the morning water supply flow data vector Q mor =[q1,q2,...,q m ] T , noon water supply flow data vector Q aft =[q1′,q2′,...,q′ m ] T , night water supply flow data vector Q eve =[q1″,q2″,...,q′ m ′] T , then the mean column vector of water supply flow data is, The water supply pressure data is obtained at the corresponding water supply flow node, forming a water supply pressure data vector P = [p1, p2, ..., p m ].

3. The machine learning remote adaptive water supply control working method according to claim 2 is characterized in that: The S1 further comprises: S1-2, the water supply system acquisition node establishes a mapping model of the water supply flow and water supply pressure throughout the day, and establishes a training subset by performing relationship calculations; according to the flow matrix FM (Flow Matrix,) of the water supply system, the conservation formula related to the flow and pressure of all nodes in the water supply system is established Among them, f1 is the quantitative function of the theoretical pressure adjustment value ΔP, p out is the output pressure of the water supply system, p client is the user-side output pressure, Δq m is the output flow regulation value of each node in the water supply system, ΔQ is the column vector of the output flow regulation values ​​of all nodes in the water supply system, the superscript T is the transpose, P m-1 is the actual pressure value of the m-1th node, P m is the actual pressure value of the mth node, P m-1 -P m is the actual pressure difference between the two nodes, ρ is the density of the flowing medium, v m is the flow velocity at the mth node, v m-1 is the flow velocity at the m-1th node, g is the gravitational acceleration of the flowing medium, h is the height difference between the m and m-1 nodes, λ is the friction coefficient, d is the pipe radius, L is the pipe length between the m and m-1 nodes, is the average flow velocity between nodes m and m-1, is the actual flow matrix of all nodes m, and Q is the mean column vector of water supply flow data. The flow distribution of the nodes is balanced with the sum of the expected regulation values. It is a universal quantifier.

4. The machine learning remote adaptive water supply control working method according to claim 1 is characterized in that: The S2 includes: According to the conservation formula, the nonlinear input vectors of the mapping model of two adjacent nodes are transformed into linear functions in high-dimensional space in turn to form a training subset of pressure. For the pressure mapping function P′→φ(P′), φ(P′) is a high-dimensional linear function of the input vector P′. The training subset is U, and x pressure samples are found in each time period. In the conservation formula of each node flow, the corresponding prediction model Q′ is established. Where Q k is the leakage vector in the process of traffic output for each node, The training subset for each node flow is V, and the training subset V is to find y flow samples according to each time period. The sample probability is calculated based on the training subsets of U and V, so as to optimize the pressure and flow of the corresponding nodes.

5. The machine learning remote adaptive water supply control working method according to claim 4 is characterized in that: The S2 further includes: In the water supply system, the pressure and flow rate from node m-1 to node m need to achieve theoretical energy conservation. In the corresponding pressure optimization period sample x, the corresponding pressure values ​​of nodes m-1 and m are obtained. In the corresponding flow optimization period sample y, the corresponding flow values ​​of nodes m-1 and m are obtained. Among them, h m and h m-1 are the water supply heights of node m and node m-1 respectively, is the pressure value of node m-1 in sample x during the pressure optimization period, is the pressure value of node m in sample x during the pressure optimization period, is the flow value of the m-1 node in sample y during the flow optimization period, is the flow value of node m in sample y during the flow optimization period, is the water head energy loss of the m-1 node in the water supply system pipeline friction, is the head energy loss of the friction node m in the water supply system pipe.

6. The machine learning remote adaptive water supply control working method according to claim 1 is characterized in that: The S3 includes: Calculate the energy mean parameter w for the two head energy losses, σ is the regulation weight. According to the mean parameters obtained in different time periods, the energy regulation output is determined to meet the stability of the pressure and flow of the water supply system. By calculating the energy mean parameter w for every two adjacent nodes and evaluating the gap between the current water supply state and the target state through iterative calculation, the identified water supply system generates output energy sample nodes based on the mean parameter of the water supply node according to the law of conservation of energy.

7. A machine learning remote adaptive water supply control working system, characterized in that: include: A data acquisition module, which acquires the water supply flow data of the historical period of the water supply system and establishes a mapping model of the water supply flow and water supply pressure; The mean value calculation module sequentially transforms the nonlinear input vectors of the mapping model of two adjacent nodes into linear functions in a high-dimensional space, and calculates the energy mean parameter according to the two head energy losses; The water supply output module evaluates the gap between the current water supply state and the target state through iterative calculation, and generates output energy sample nodes for the mean parameters of the water supply nodes of the identified water supply system according to the law of conservation of energy.

Citation Information

Patent Citations

  • Online mapping model parameter estimation method and system for heat supply system

    CN112150307A

  • Urban daily water consumption prediction method based on machine learning

    CN113723708A

  • Pipeline pressure prediction method based on characteristic line and visual platform

    CN117610338A

  • Method for selecting section of water leakage suspicion by water network analysis and water supply integrated management operating system with function thereof

    KR101105192B1

  • Systems and methods for water distribution network leakage detection and / or localization

    US20240084561A1

Cited By

  • Water supply control method and system based on self-adaptive adjustment of fluid flow

    CN119597071A

  • Water supply control method and system based on fluid flow adaptive regulation

    CN119597071B