A drainage monitoring and analysis system based on big data

Abstract

The present invention relates to the field of deep learning, and discloses a drainage monitoring and analysis system based on big data, comprising the following steps: collecting pipe network element information, the pipe network element information includes pipe network node information, pipe section information, underground engineering point information, and monitoring data information; obtaining a number one two-dimensional structure data from the collected pipe network element information, the number one two-dimensional structure data includes a data matrix and a relationship matrix; sorting the number one two-dimensional structure data to obtain a number one one-dimensional structure data, the number one one-dimensional structure data includes n data items sorted by time; inputting the number one one-dimensional structure data into a monitoring model, and outputting an equipment allocation strategy, the equipment allocation strategy includes the type of flow meter and valve type that need to be configured on all pipe network nodes. The present invention can realize the monitoring of pipe network drainage, and can also analyze the leakage information of pipe network nodes according to the drainage flow data.

Description

Technical Field

[0001] The present invention relates to the field of deep learning technology, and more specifically, to a drainage monitoring and analysis system based on big data. Background Art

[0002] With the acceleration of urbanization, urban drainage pipe network systems are becoming increasingly large and complex. In order to ensure the normal operation of urban drainage pipe networks and timely detect and handle pipe network failures, it is necessary to effectively monitor and manage drainage pipe networks. Traditional drainage pipe network monitoring methods mainly rely on manual inspections and fixed-point monitoring equipment, which have problems such as low monitoring efficiency and insufficient data utilization, and are difficult to meet the needs of modern drainage pipe network management.

[0003] In recent years, the development of emerging technologies such as the Internet of Things and big data has provided new solutions for drainage network monitoring. However, the existing drainage network monitoring system still has deficiencies in data collection, processing and analysis. For example, the monitoring data lacks effective structured processing, and the correlation and time series between data are not fully utilized, which affects the accuracy and efficiency of data analysis and makes it difficult to achieve maximum coverage of pipeline network monitoring within a limited cost budget. Summary of the invention

[0004] The present invention provides a drainage monitoring and analysis system based on big data. The system should be able to comprehensively collect information on pipe network elements, effectively structure the monitoring data, and use big data analysis technology to achieve real-time monitoring of the pipe network operation status, fault diagnosis and early warning functions. At the same time, the system can also optimize the equipment allocation strategy according to the characteristics of the pipe network and monitoring requirements, and achieve the maximum monitoring coverage at the lowest cost. This is of great significance for improving the operating efficiency of the drainage pipe network and reducing the risk of pipe network failures.

[0005] The present invention provides a drainage monitoring and analysis system based on big data, comprising the following modules:

[0006] Information collection module, collecting pipeline network element information, including pipeline network node information, pipeline section information, underground engineering point information, and monitoring data information;

[0007] The first structural module obtains No. 1 two-dimensional structural data based on the collected pipe network element information. The No. 1 two-dimensional structural data includes a data matrix and a relationship matrix. A unit of the data matrix represents the data of an independent object. The element of the Ith row and the Jth column of the relationship matrix represents the association between the Ith unit of the data matrix and the independent object represented by the Jth unit. If there is an association, the value of the element of the relationship matrix is ​​1, otherwise it is 0;

[0008] The second structuring module obtains the first one-dimensional structure data based on the sorting of the first two-dimensional structure data, wherein the first one-dimensional structure data includes n data items sorted by time, and the t-th data item represents the first two-dimensional structure data generated by the pipe network element information collected at the t-th moment;

[0009] The monitoring network layout module inputs the No. 1 one-dimensional structure data into the monitoring model. The monitoring model includes a first intermediate layer, a second intermediate layer, a first output layer, a third intermediate layer, and a second output layer. The first intermediate layer inputs the No. 1 two-dimensional structure data and outputs the first intermediate representation data to the second intermediate layer; the second intermediate layer outputs the second intermediate representation data to the first output layer and outputs an equipment allocation strategy. The equipment allocation strategy includes the types of flow meters and valves that need to be configured on all pipe network nodes;

[0010] The drainage monitoring module is used to measure the drainage flow of the pipeline network through the flow meter after the equipment allocation strategy is executed, and to find the leakage points of the pipeline network nodes through the data of the flow meter.

[0011] In a preferred embodiment, the independent objects in the No. 1 two-dimensional structural data include pipe network nodes, pipe sections, underground engineering points, and monitoring data.

[0012] In a preferred embodiment, the pipeline network element information is obtained by field surveying of the pipeline network, and the monitoring data is obtained from the flow meter; the pipeline network node information includes the location coordinates of the connection points and branch points in the pipeline network; the pipe section information includes the starting and ending pipeline points; the underground engineering point information includes the engineering point location coordinates and engineering description text; the monitoring data information includes the location coordinates and data text of the flow meter.

[0013] In a preferred embodiment, corresponding features are obtained after feature engineering is performed on the pipe network node information, pipe section information, project description text, and data text;

[0014] The location of the pipe network nodes is embedded in the coordinates as a feature engineering method to obtain the characteristics of the pipe network nodes;

[0015] The pipe segment includes the starting pipe network node. For multiple pipe network nodes, the feature engineering method of coordinate embedding is used to obtain the pipe network node features, and the obtained multiple pipe network node features are spliced ​​to obtain the pipe segment feature.

[0016] The engineering description text of the underground engineering and the data text of the monitoring data are used as a feature engineering method to obtain the underground engineering features and monitoring data features.

[0017] In a preferred embodiment, the relationship between the independent objects in the first two-dimensional structure data is as follows:

[0018] The association between a pipeline network node and a pipeline segment means that the pipeline network node belongs to the pipeline segment; the association between pipeline network nodes means that the pipeline network nodes belong to the same pipeline segment; the association between pipeline segments means that two pipeline segments are connected to each other; the association between an underground engineering point and a pipeline segment means that the pipeline segment is within the influence range of the underground engineering point; the association between monitoring data and a pipeline segment means that the pipeline segment is within the monitoring range of the monitoring sensor.

[0019] In a preferred embodiment, a device allocation strategy is represented as a partition matrix, where the first row and Lth column of the partition matrix represent the type of valve that needs to be deployed at the Lth pipeline node, and the second row and Lth column represent the type of flow meter that needs to be deployed at the Lth pipeline node.

[0020] In a preferred embodiment, the third intermediate layer is combined with the second output layer for independent pre-training, and the third intermediate layer is trained to have the ability to identify underground engineering points, so that the partitioning results can avoid underground engineering points and prevent the partition from being affected by underground engineering.

[0021] In a preferred embodiment, only the first output layer is used during monitoring model training, and the training steps include:

[0022] Step 101, initializing the parameters of the monitoring model;

[0023] Step 102: Observe the network element information at time e , the device allocation strategy executed at time e , Pipeline network element information at time e+1 , execute device allocation strategy Rewards ;

[0024] Step 103, then calculate the device allocation strategy error:

[0025] ;

[0026] in represents the device allocation strategy error at time e, represents the discount factor, , Represents the monitoring model input The maximum probability value in the first output vector output when Represents the monitoring model input The first output vector corresponds to the device allocation strategy The probability value of

[0027] Step 104, update the monitoring model, the updated formula is as follows:

[0028] ;

[0029] , represents the step size of deep learning, Indicates delivery update;

[0030] Step 105, iterate steps 102-104 until the monitoring model converges or the number of iterations reaches a set value, the default value of which is 100.

[0031] In a preferred embodiment, the reward function calculation formula is as follows:

[0032] ;

[0033] in, represents the expected partitioning cost, Indicates the device allocation policy The cost of execution, Indicates the total number of nodes in all partitions, represents the number of nodes affected by underground engineering, Indicates the device allocation policy The difference between the flow measurement error after execution and the flow measurement error before execution, Indicates the device allocation policy After execution, the flow meter data is used to find the ratio of the leakage points in the pipeline network nodes to the total number of leakage points.

[0034] The present invention provides a storage medium storing non-temporary computer-readable instructions. When the non-temporary computer-readable instructions are executed by a computer, a module in a drainage monitoring and analysis system based on big data as described above can be executed.

[0035] The beneficial effects of the present invention are:

[0036] By comprehensively collecting information on pipe network elements, including pipe network nodes, pipe sections, underground engineering points, etc., a more complete and detailed digital model of the pipe network can be constructed. This provides a rich information basis for subsequent data analysis and decision-making, and helps to improve the comprehensiveness and precision of drainage network monitoring.

[0037] By structuring the monitoring data and integrating the scattered monitoring data into a data matrix and relationship matrix with temporal and spatial correlation, the intrinsic connection between the data can be better revealed, and the readability and analyzability of the data can be improved. This not only improves the quality of the monitoring data, but also improves the efficiency of data mining and utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a module diagram of a drainage monitoring and analysis system based on big data of the present invention;

[0039] Figure 2 It is the desensitized pipe network node information data sample of the present invention;

[0040] Figure 3 is the desensitized pipe section information data sample of the present invention;

[0041] Figure 4 It is the desensitized underground engineering point information data sample of the present invention;

[0042] Figure 5 It is the desensitized pipe monitoring data information data sample of the present invention;

[0043] Figure 6 It is the update status of the average Q value during the model training process of the present invention. DETAILED DESCRIPTION

[0044] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that the discussion of these implementations is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the present specification. Various examples may omit, replace, or add various processes or components as needed. In addition, the features described in some examples may also be combined in other examples.

[0045] At least one embodiment of the present invention discloses a drainage monitoring and analysis system based on big data, such as Figure 1 As shown, it includes the following modules:

[0046] Information collection module, collecting pipeline network element information, including pipeline network node information, pipeline section information, underground engineering point information, and monitoring data information;

[0047] The network node information includes the location coordinates of the connection points and branch points in the network; the pipe section information includes the starting and ending network points; the underground engineering point information includes the engineering point location coordinates and engineering description text; the monitoring data information includes the location coordinates and data text of the flow meter; the network element information is obtained by field surveying and mapping of the network, and the monitoring data is obtained from the flow meter;

[0048] Information is collected at fixed time intervals starting from the beginning of planning the pipe network zoning.

[0049] A first structured module obtains No. 1 two-dimensional structure data based on the collected pipe network element information. The No. 1 two-dimensional structure data includes a data matrix and a relationship matrix. A unit of the data matrix represents data of an independent object. The independent object includes a pipe network node, a pipe section, an underground engineering point, and monitoring data. A unit of the data matrix representing a pipe network node only includes pipe network node information, a unit of the data matrix representing a pipe section only includes pipe sections, a unit of the data matrix representing an underground engineering point only includes description text of the underground engineering point, and a unit of the data matrix representing monitoring data only includes description text of the monitoring data;

[0050] The element in the Ith row and Jth column of the relationship matrix represents the association between the Ith unit of the data matrix and the independent object represented by the Jth unit. If there is an association, the value of this element of the relationship matrix is ​​1, otherwise it is 0;

[0051] The association between a network node and a pipe segment means that the network node belongs to the pipe segment;

[0052] The association between pipe network nodes means that the pipe network nodes belong to the same pipe segment;

[0053] There is a relationship between pipe segments: two pipe segments are connected to each other;

[0054] The association between underground engineering points and pipe sections means that: the pipe section is within the influence range of the underground engineering point;

[0055] The association between monitoring data and pipe sections means that: the pipe section is within the monitoring range of the monitoring sensor;

[0056] The second structuring module obtains the first one-dimensional structure data based on the sorting of the first two-dimensional structure data, wherein the first one-dimensional structure data includes n data items sorted by time, and the t-th data item represents the first two-dimensional structure data generated by the pipe network element information collected at the t-th moment;

[0057] The monitoring network layout module inputs the No. 1 one-dimensional structure data into the monitoring model, and the monitoring model includes a first intermediate layer, a second intermediate layer, a first output layer, a third intermediate layer, and a second output layer, wherein the first intermediate layer inputs the No. 1 two-dimensional structure data and outputs the first intermediate representation data to the second intermediate layer; the second intermediate layer outputs the second intermediate representation data to the first output layer, and outputs an equipment allocation strategy, which includes the types of flow meters and valves that need to be configured on all pipe network nodes;

[0058] The third intermediate layer inputs the No. 1 two-dimensional structure data, outputs the third intermediate representation data to the second output layer, and outputs the pipe network nodes whose flow meters or valve controls will be affected by the underground project;

[0059] The drainage monitoring module is used to measure the drainage flow of the pipeline network through the flow meter after the equipment allocation strategy is executed, and to find the leakage points of the pipeline network nodes through the data of the flow meter.

[0060] The method of discovering the leakage points of pipeline network nodes through flow meter data is to input the flow meter data into a trained neural network model to output the results of whether all pipeline network nodes are leaking.

[0061] In one embodiment of the present invention, the calculation formula of the first intermediate layer is as follows:

[0062] ;

[0063] in The first unit of the ith unit of the data matrix represents the data in the middle, represents the state weight parameter, and are the pipe network element information of independent objects represented by the i-th and j-th cells of the data matrix, represents the set of vertices directly connected to the i-th vertex, express The total number of vertices in , , tanh represents the hyperbolic tangent function, Indicates adjustable parameters; The default value of is 1 / M. Represents a transpose operation;

[0064] ;

[0065] in represents the first intermediate representation data of the output, The data matrix represents the data in the middle of the cell of the i-th cell of the data matrix, It means that the first unit intermediate representation data of the data matrix of all the data matrix units in the data matrix are concatenated, and N represents the set of all the data matrix units. is the weight parameter, is the bias parameter, Represents the sigmoid function.

[0066] In one embodiment of the present invention, the calculation formula of the second intermediate layer is as follows:

[0067] ;

[0068] ;

[0069] ;

[0070] ;

[0071] in, , , represents the first, second, and third weight parameters, , , represents the first, second and third bias parameters, represents the dot product, where , and Respectively represent the first, second, and third intermediate states, It represents the first intermediate representation data output when the tth data item is input into the first intermediate layer. and Respectively represent the t-th and t-1-th second intermediate representation data, n≥t≥1, n represents the total number of data items of the one-dimensional structure data, t=1 , tanh is the hyperbolic tangent function, Represents the sigmoid function.

[0072] In one embodiment of the present invention, the calculation formula of the third intermediate layer is as follows:

[0073] ;

[0074] in The third intermediate representation data representing the vth cell of the data matrix, is the set of cells of the data matrix that are associated with the vth cell of the data matrix, The U-th unit of the data matrix represents the pipe network element information of the independent object. is a normalizing constant, represents the weight parameter, represents the bias parameter, is the sigmoid function.

[0075] In one embodiment of the present invention, the calculation formula of the first output layer is as follows:

[0076] ;

[0077] in It represents the output of the first vector, whose cth component value represents the probability value of the cth device allocation strategy. The device allocation strategy with the largest probability value is selected as the output. The device allocation strategy group contains all executable device allocation strategies. represents the t-th second intermediate representation data, represents the set of all cells of the data matrix, n represents the total number of data items of the one-dimensional structured data, is the first output weight parameter, is the first output bias parameter, Represents a sigmoid function.

[0078] In one embodiment of the present invention, a device allocation strategy is represented as a partition matrix, where the first row and the Lth column of the partition matrix represent the type of valve that needs to be deployed at the Lth pipeline node, and the second row and the Lth column represent the type of flow meter that needs to be deployed at the Lth pipeline node.

[0079] In one embodiment of the present invention, an example of a device allocation strategy for a first output layer is provided as follows:

[0080] The device allocation strategy probability values ​​are obtained as follows:

[0081] [0.02, 0.01, 0.05, 0.03, 0.15, 0.04, 0.02, 0.08, 0.03, 0.02, 0.01, 0.03, 0.07, 0.02, 0.04, 0.03, 0.02, 0.01, 0.09, 0.02, 0.04, 0.02, 0.03, 0.02, 0.08, 0.01, 0.02, 0.03, 0.02, 0.01], where the maximum probability value is 0.15, corresponding to the 5th device allocation strategy, so the 5th device allocation strategy is selected as the output;

[0082] The partition matrix corresponding to the device allocation strategy is as follows:

[0083] Valve type: [4,2,5,1,3,2,4,5,1,3,2,4,3,5,2,1,4,3,2,5,1,3,4,2,5,3,1,4,2,3];

[0084] Among them, 4, 2, 5, 1, 3, 2, 4, 5, 1, 3, 2, 4, 3, 5, 2, 1, 4, 3, 2, 5, 1, 3, 4, 2, 5, 3, 1, 4, 2, 3 respectively represent the types of valves to be deployed at the pipeline network nodes N1, N2, ..., N30, and N1, N2, ..., N30 represent the IDs of the pipeline network nodes;

[0085] Flow meter type: [2,1,3,2,4,1,2,3,1,2,1,3,2,4,1,2,3,1,2,3,1,2,3,1,2,3,1,2,1,2];

[0086] Among them, 2, 1, 3, 2, 4, 1, 2, 3, 1, 2, 1, 3, 2, 4, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 1, 2 represent the types of flow meters to be deployed at the network nodes N1, N2, ..., N30 respectively, and N1, N2, ..., N30 represent the IDs of the network nodes;

[0087] The partition matrix indicates that advanced valves need to be deployed at important nodes; high-precision flow meters need to be deployed in pipe sections with large flow changes; and basic valves and flow meters need to be used at ordinary connection points.

[0088] In one embodiment of the present invention, the valve types include 5 categories, of which categories 1-2 are basic valves; category 3 represents intermediate valves; categories 4-5 represent advanced valves;

[0089] Category 1: Ordinary stop valves and manual ball valves, used for ordinary connection points that do not require frequent operation and have low flow control requirements;

[0090] Category 2: Manual butterfly valve and low-pressure gate valve, suitable for light control and basic flow management needs;

[0091] Category 3: Electric ball valve or electric butterfly valve with basic automatic function, suitable for nodes that require certain automatic control functions and have medium flow requirements;

[0092] Category 4: High-performance electric butterfly valves and high-pressure gate valves, suitable for key nodes, need to be reliable and durable, and able to withstand higher flow loads;

[0093] Category 5: Intelligent control valves, such as regulating valves or mobile cloud-controllable valves, are suitable for important nodes and require high-precision flow control and remote operation;

[0094] In one embodiment of the present invention, the flow meter types include 4 categories, wherein categories 1-2 represent basic flow meters; category 3 represents intermediate flow meters; categories 4-5 represent advanced flow meters;

[0095] Category 1: Mechanical flow meters, such as rotary piston flow meters, are used for general monitoring points with limited budgets and low accuracy requirements;

[0096] Category 2: Turbine or impeller flowmeter, suitable for daily monitoring with low to medium accuracy requirements;

[0097] Category 3: Electromagnetic flowmeter or ultrasonic flowmeter, suitable for monitoring medium flow changes in pipe networks, providing higher accuracy;

[0098] Category 4: High-precision ultrasonic flowmeter or Coriolis flowmeter, deployed at critical network nodes with severe flow fluctuations to ensure accurate data collection.

[0099] In one embodiment of the present invention, the calculation formula of the second output layer is as follows:

[0100] ;

[0101] in Represents the output of the second output vector, the cth component of the vector represents the probability that the cth pipeline network node is affected by the underground project, The third intermediate representation data representing the u-th element of the data matrix, is the second output weight parameter, is the second output bias parameter, Represents the sigmoid function.

[0102] In one embodiment of the present invention, the third intermediate layer is combined with the second output layer for independent pre-training, and the third intermediate layer is trained to have the ability to identify underground engineering points, so that the partitioning result can avoid underground engineering points and prevent the partition from being affected by underground engineering.

[0103] In one embodiment of the present invention, only the first output layer is used during monitoring model training, and the training steps include:

[0104] Step 101, initializing the parameters of the monitoring model;

[0105] Step 102: Observe the network element information at time e , the device allocation strategy executed at time e , Pipeline network element information at time e+1 , execute device allocation strategy Rewards ;

[0106] Step 103, then calculate the device allocation strategy error:

[0107] ;

[0108] in represents the device allocation strategy error at time e, represents the discount factor, , Represents the monitoring model input The maximum probability value in the first output vector output when Represents the monitoring model input The first output vector corresponds to the device allocation strategy The probability value of

[0109] ;

[0110] in, represents the expected partitioning cost, Indicates the device allocation policy The cost of execution, Indicates the total number of nodes in all partitions, represents the number of nodes affected by underground engineering, Indicates the device allocation policy The difference between the flow measurement error after execution and the flow measurement error before execution, Indicates the device allocation policy After execution, the flow meter data is used to find the ratio of the leakage points in the pipeline network nodes to the total number of leakage points.

[0111] Step 104, update the monitoring model, the updated formula is as follows:

[0112] ;

[0113] , represents the step size of deep learning, Indicates delivery update;

[0114] Step 105, iterate steps 102-104 until the monitoring model converges or the number of iterations reaches a set value, the default value of which is 100.

[0115] In one embodiment of the present invention, corresponding features are obtained after feature engineering is performed on the pipe network node information, pipe section information, project description text, and data text;

[0116] The location of the pipe network nodes is embedded in the coordinates as a feature engineering method to obtain the characteristics of the pipe network nodes;

[0117] The pipe segment includes the starting pipe network node. For multiple pipe network nodes, the feature engineering method of coordinate embedding is used to obtain the pipe network node features, and the obtained multiple pipe network node features are spliced ​​to obtain the pipe segment feature.

[0118] The engineering description text of underground engineering and the data text of monitoring data are used to obtain underground engineering features through the WordEmbedding algorithm as a feature engineering method.

[0119] In one embodiment of the present invention, an example of a specific application scenario of the aforementioned drainage monitoring and analysis system based on big data is provided:

[0120] This example focuses on the monitoring and analysis of urban drainage networks, aiming to use big data technology and machine learning methods to achieve intelligent management and fault warning of drainage networks. Urban drainage networks are important urban infrastructure, and their operation directly affects urban drainage, flood control, and environmental sanitation. Due to the large scale, complex structure, and underground concealment of drainage networks, it is difficult for humans to detect and handle network faults in a timely manner. Therefore, the probability of network failure is relatively high.

[0121] Data preparation

[0122] Pipeline network element information

[0123] Pipeline network node information: location coordinates of connection points and branch points in the pipeline network.

[0124] Pipe section information: starting and ending pipe network points.

[0125] Underground engineering point information: engineering point location coordinates, engineering description text.

[0126] Monitoring data information: flow meter location coordinates and data text.

[0127] Data Source

[0128] Field surveying of pipeline networks: obtaining spatial information such as pipeline network nodes, pipe sections, and underground engineering points.

[0129] Flow meter: obtain pipeline network flow monitoring data.

[0130] Data structuring

[0131] Two-dimensional structure data No. 1:

[0132] Data matrix: Each cell represents the data of an independent object (pipeline network node, pipe section, underground engineering point, monitoring data).

[0133] Relationship matrix: represents the relationship between independent objects.

[0134] No. 1 one-dimensional structure data: a No. 1 two-dimensional structure data sequence sorted by time, each data item represents the pipeline network element information at a moment.

[0135] Feature Engineering

[0136] Pipe network node features: Use coordinate embedding method.

[0137] Pipe segment features: Features of the start and end pipe network nodes.

[0138] Underground engineering features: Use the Word Embedding algorithm for engineering description text.

[0139] Monitoring data features: Use the Word Embedding algorithm on data text.

[0140] Data Annotation

[0141] Identification of underground engineering points:

[0142] Positive samples: Pipeline network nodes affected by underground projects.

[0143] Negative samples: Pipeline network nodes that are not affected by underground projects.

[0144] Marking method: Based on the underground engineering point information and the location of the pipeline network nodes, determine whether the node is within the project impact range.

[0145] A set of desensitized data samples such as Figure 2 , 3 , 4, and 5.

[0146] Model training (pre-training)

[0147] Input: Pipeline network node characteristics, underground engineering characteristics.

[0148] Output: The probability that the pipeline network nodes are affected by underground projects.

[0149] Training data: labeled pipeline network node samples.

[0150] Training method: supervised learning, adjusting the model parameters to maximize the recognition accuracy on the training set.

[0151] Model training (reinforcement learning)

[0152] Training Algorithms

[0153] Reinforcement Learning Framework:

[0154] Environment: Urban underground drainage network, including pipe network nodes, pipe sections, pipe network node monitoring data, etc.

[0155] Intelligent agent: drainage monitoring and analysis system, monitoring and analyzing the flow and health status of pipe network nodes.

[0156] Status: Pipeline network element information.

[0157] Action: Device allocation policy.

[0158] Rewards: Rewards are calculated based on factors such as expected cost, execution cost, number of nodes affected by underground projects, flow error, and leak detection rate.

[0159] Q-learning algorithm: It iteratively updates the monitoring model based on the observed state transitions and rewards, converging to the optimal strategy.

[0160] The reward function is designed as follows:

[0161] ;

[0162] in, represents the expected partitioning cost, Indicates the device allocation policy The cost of execution, Indicates the total number of nodes in all partitions, represents the number of nodes affected by underground engineering, Indicates the device allocation policy The difference between the flow measurement error after execution and the flow measurement error before execution, Indicates the device allocation policy After execution, the flow meter data is used to find the ratio of the leakage points in the pipeline network nodes to the total number of leakage points.

[0163] In a simulated environment, the environment and the agent are modeled, and actions are performed to train and verify the model. The training results are as follows: Figure 6 As shown in the figure, the horizontal axis represents the epoch of training, and the vertical axis represents the average Q value. The average Q value approaches a stable state at around 80 epochs.

[0164] Results Analysis

[0165] Through pre-training, the model can accurately determine whether the pipeline nodes are affected by underground projects, providing a basis for the reasonable layout of monitoring equipment. On this basis, the trained monitoring model can dynamically generate the optimal equipment allocation strategy based on the real-time status of the pipeline network. While ensuring the monitoring coverage, this strategy avoids the interference of underground projects to the greatest extent and improves the reliability of monitoring data. In addition, the system can also promptly detect pipeline leakage points based on flow meter data, providing decision support for emergency repairs and maintenance.

[0166] A storage medium stores non-temporary computer-readable instructions. When the non-temporary computer-readable instructions are executed by a computer, the module in the drainage monitoring and analysis system based on big data as described above can be executed.

[0167] The above describes an embodiment of the present invention, but this embodiment is not limited to the above-mentioned specific implementation mode. The above-mentioned specific implementation mode is merely illustrative and not restrictive. Under the guidance of this embodiment, ordinary technicians in this field can also make more forms of equivalent embodiments, all of which are within the protection of this embodiment.

Claims

1. A drainage monitoring and analysis system based on big data, characterized in that: Includes the following modules: Information collection module, collecting pipeline network element information, including pipeline network node information, pipeline section information, underground engineering point information, and monitoring data information; The first structural module obtains No. 1 two-dimensional structural data based on the collected pipe network element information. The No. 1 two-dimensional structural data includes a data matrix and a relationship matrix. A unit of the data matrix represents the data of an independent object. The element of the Ith row and the Jth column of the relationship matrix represents the association between the Ith unit of the data matrix and the independent object represented by the Jth unit. If there is an association, the value of the element of the relationship matrix is ​​1, otherwise it is 0; The second structuring module obtains the first one-dimensional structure data based on the sorting of the first two-dimensional structure data, wherein the first one-dimensional structure data includes n data items sorted by time, and the t-th data item represents the first two-dimensional structure data generated by the pipe network element information collected at the t-th moment; The monitoring network layout module inputs the No. 1 one-dimensional structure data into the monitoring model, the monitoring model includes a first intermediate layer, a second intermediate layer, a first output layer, a third intermediate layer, and a second output layer, wherein the first intermediate layer inputs the No. 1 two-dimensional structure data and outputs the first intermediate representation data to the second intermediate layer; The second intermediate layer outputs the second intermediate representation data to the first output layer, and outputs a device allocation strategy, which includes the types of flow meters and valves that need to be configured on all pipeline network nodes; Only the first output layer is used when monitoring the model training. The training steps include: Step 101, initializing the parameters of the monitoring model; Step 102: Observe the network element information at time e , the device allocation strategy executed at time e , Pipeline network element information at time e+1 , execute device allocation strategy Rewards ; Step 103, then calculate the device allocation strategy error: ； in represents the device allocation strategy error at time e, represents the discount factor, , Represents the monitoring model input The maximum probability value in the first output vector output when Represents the monitoring model input The first output vector corresponds to the device allocation strategy The probability value of The reward function calculation formula is as follows: ； in, represents the expected partitioning cost, Indicates the device allocation policy The cost of execution, Indicates the total number of nodes in all partitions, represents the number of nodes affected by underground engineering, Indicates the device allocation policy The difference between the flow measurement error after execution and the flow measurement error before execution. Leak indicates the device allocation strategy. After execution, the flow meter data is used to find the ratio of the leakage points of the pipeline network nodes to the total number of leakage points; Step 104, update the monitoring model, the updated formula is as follows: ； , represents the step size of deep learning, Indicates delivery update; Step 105, iterate steps 102-104 until the monitoring model converges or the number of iterations reaches a set value, the default value of which is 100; The drainage monitoring module is used to measure the drainage flow of the pipeline network through the flow meter after the equipment allocation strategy is executed, and to find the leakage points of the pipeline network nodes through the data of the flow meter.

2. A drainage monitoring and analysis system based on big data according to claim 1, characterized in that: The independent objects in the No. 1 two-dimensional structural data include pipeline network nodes, pipeline sections, underground engineering points, and monitoring data.

3. The drainage monitoring and analysis system based on big data according to claim 1 is characterized in that: The pipeline network element information is obtained by field surveying of the pipeline network, and the monitoring data is obtained from the flow meter; the pipeline network node information includes the location coordinates of the connection points and branch points in the pipeline network; the pipe section information includes the starting and ending pipeline points; the underground engineering point information includes the engineering point location coordinates and engineering description text; the monitoring data information includes the location coordinates and data text of the flow meter.

4. A drainage monitoring and analysis system based on big data according to claim 3, characterized in that: After performing feature engineering on the pipe network node information, pipe section information, project description text, and data text, corresponding features are obtained; The location of the pipe network nodes is embedded in the coordinates as a feature engineering method to obtain the characteristics of the pipe network nodes; The pipe segment includes the starting pipe network node. For multiple pipe network nodes, the feature engineering method of coordinate embedding is used to obtain the pipe network node features, and the obtained multiple pipe network node features are spliced ​​to obtain the pipe segment feature. The engineering description text of the underground engineering and the data text of the monitoring data are used as a feature engineering method to obtain the underground engineering features and monitoring data features.

5. The drainage monitoring and analysis system based on big data according to claim 2 is characterized in that: The relationship between independent objects is as follows: The association between a pipeline network node and a pipeline segment means that the pipeline network node belongs to the pipeline segment; the association between pipeline network nodes means that the pipeline network nodes belong to the same pipeline segment; the association between pipeline segments means that two pipeline segments are connected to each other; the association between an underground engineering point and a pipeline segment means that the pipeline segment is within the influence range of the underground engineering point; the association between monitoring data and a pipeline segment means that the pipeline segment is within the monitoring range of the monitoring sensor.

6. The drainage monitoring and analysis system based on big data according to claim 1 is characterized in that: A device allocation strategy is represented as a partition matrix. The first row and the Lth column of the partition matrix represent the type of valve that needs to be deployed at the Lth pipeline node, and the second row and the Lth column represent the type of flow meter that needs to be deployed at the Lth pipeline node.

7. The drainage monitoring and analysis system based on big data according to claim 1 is characterized in that: The third intermediate layer is combined with the second output layer for independent pre-training, and the third intermediate layer is trained to have the ability to identify underground engineering points, so that the partitioning result can avoid underground engineering points and prevent the partition from being affected by underground engineering.

8. A storage medium storing non-transitory computer-readable instructions, characterized in that: When the non-temporary computer-readable instructions are executed by a computer, they can execute a module in a drainage monitoring and analysis system based on big data as described in any one of claims 1-7.

Images