CN-121993746-A - Intelligent early warning system and method for leakage risk of large-diameter pipeline

CN121993746ACN 121993746 ACN121993746 ACN 121993746ACN-121993746-A

Abstract

The invention discloses an intelligent early warning system and method for leakage risk of a large-diameter pipeline, which relate to the technical field of pipeline risk early warning, and are used for collecting upstream/downstream operation data of the pipeline, constructing a boundary self-adaptive model, constructing a multidimensional feature map of internal state change of the pipeline, analyzing a dynamic response mode of the multidimensional feature map under upstream/downstream change conditions, comparing the dynamic response of the current multidimensional feature map with the predicted dynamic response of the multidimensional feature map, analyzing whether abnormal dynamics exists or not based on the comparison result, and generating intelligent early warning prompt and recommended action if the abnormal dynamics exists. The early warning system not only effectively solves the core problems of high false alarm rate, large missing report rate, difficult early leakage identification, inaccurate early warning response and the like in the prior art, but also greatly improves the accuracy, instantaneity and intellectualization level of the large-pipe-diameter pipeline leakage risk early warning system through multi-dimensional, multi-modal and multi-factor intelligent fusion analysis, and provides powerful technical guarantee for the safe operation of a pipe network.

Inventors

SUN TIANYU
CHEN LIJIN
LIU ZHIGANG
MAO HUI
CHEN LILI
WANG QUNBIAO
YUE XIEHUI

Assignees

宁波城市供水水质监测站有限公司

Dates

Publication Date: 20260508
Application Date: 20260106

Claims (10)

1. The intelligent early warning method for the leakage risk of the large-diameter pipeline is characterized by comprising the following steps of: collecting upstream/downstream operation data of a pipeline in real time, and constructing a boundary self-adaptive model; Constructing a multidimensional feature map of the internal state change of the pipeline, and analyzing a dynamic response mode of the multidimensional feature map under the upstream/downstream fluctuation condition; Comparing the current multi-dimensional feature map dynamic response with the predicted multi-dimensional feature map dynamic response, analyzing whether abnormal dynamics exist based on the comparison result, and if so, generating intelligent early warning prompt and recommending actions.
2. The intelligent early warning method for leakage risk of large-diameter pipeline according to claim 1, which is characterized in that: comparing the current multi-dimensional feature map dynamic response with the predicted multi-dimensional feature map dynamic response, analyzing whether abnormal dynamics exist or not based on the comparison result, and comprising the following steps: Synchronous acquisition of true values during operation And predicted value Comparing, calculating the node prediction error ; Weighting and aggregating the difference indexes of all nodes, and calculating the abnormality degree of the overall graph structure ; When (when) In the time-course of which the first and second contact surfaces, The dynamic abnormality threshold is determined to be a dynamic abnormality response.
3. The intelligent early warning method for leakage risk of large-diameter pipeline according to claim 2, characterized in that the difference indexes of all nodes are weighted and aggregated, and the degree of abnormality of the overall graph structure is calculated : , wherein, Representing the total number of nodes, Represent the first The weight of the individual nodes is determined, Representing node characteristic difference values.
4. The intelligent early warning method for leakage risk of large-diameter pipeline according to claim 3, characterized in that real values are synchronously collected in the operation process And predicted value Comparing, calculating the node prediction error : In which, in the process, Representing the predicted value of parameter i at time t +1, Representing the true value of parameter i at time t + 1.
5. The intelligent early warning method for leakage risk of large-diameter pipeline according to claim 4, wherein the method is characterized by analyzing the dynamic response mode of the multidimensional feature map under the upstream/downstream fluctuation condition, and comprises the following steps: Each graph structure The input graph neural network has the expression: , wherein, Representing the adjacency matrix after adding the self-join, The node degree matrix is represented as a matrix of degrees, Represent the first The node representation of the layer is such that, The weight matrix is represented by a matrix of weights, Representing an activation function; For each time step Respectively, embedding vector sequences into output nodes, and embedding and representing graphs at each moment As a time series input, input into the LSTM network, model the evolution trend of the node states in the time dimension.
6. The intelligent early warning method for leakage risk of large-diameter pipeline according to claim 5, wherein the method is characterized by constructing a multidimensional feature map of the internal state change of the pipeline and comprises the following steps: Collecting running state information in the pipeline, and dividing the pipeline into a plurality of continuous monitoring unit sections; the features of each node are expressed as vectors Formal representation, construction of graph structure , Representing a collection of nodes, each node being a physical segment, Representing a set of edges; at each sampling time window Generating a graph structure Obtaining a time sequence diagram structure according to the node matrix and the edge matrix T represents the length of the duration window.
7. The intelligent early warning method for leakage risk of the large-diameter pipeline according to claim 6, wherein the intelligent early warning method is characterized by collecting the running state information of the interior of the pipeline, including instantaneous pressure, pressure change gradient, acoustic wave propagation delay, echo characteristics, attenuation signals, strain distribution, temperature gradient and vibration modes; Dividing the pipeline into a plurality of continuous monitoring unit sections, wherein each unit section is used as a node The node binding characteristics comprise average pressure value, pressure gradient change rate, echo delay, signal-to-noise ratio decreasing amplitude, strain peak value and vibration abnormal frequency in the current time window of the unit; the features of each node are expressed as vectors The form represents: , wherein, The pressure average value is represented as, The rate of change of the pressure is indicated, Representing the reflection amplitude or attenuation coefficient of the sound wave, The propagation delay of the sound wave, Indicating that the optical fiber should be changed by a constant value, Indicating the frequency of the high-frequency vibration of the optical fiber.
8. The intelligent early warning method for leakage risk of large-diameter pipeline according to claim 7, wherein the method is characterized by collecting upstream/downstream operation data of the pipeline in real time and constructing a boundary self-adaptive model, and comprises the following steps: Collecting upstream operation data of a pipeline, wherein the upstream operation data comprise inlet pressure, water flow rate and temperature fluctuation, and collecting downstream operation data of the pipeline, and the downstream operation data comprise outlet pressure, water flow rate and temperature fluctuation; The collected historical operation data are sorted, a structured historical operation data sample set is constructed, the historical operation data are classified and marked based on operation time periods, seasonal changes and water source fluctuation dimensions, and historical operation tracks under a plurality of operation scenes are established; extracting and analyzing fluctuation characteristics of operation indexes in historical operation data, and dynamically calculating the mean value, variance, variation coefficient and change rate of each parameter by utilizing a sliding time window; and establishing a boundary self-adaptive model based on the historical fluctuation characteristics through a K-Means clustering algorithm.
9. The intelligent early warning method for leakage risk of large-diameter pipelines is characterized in that a boundary self-adaptive model is established based on historical fluctuation characteristics through a K-Means clustering algorithm, and the intelligent early warning method comprises the following steps: constructing a feature vector of each piece of historical data: , Is the mean value of the two values, As a function of the variance of the values, As the coefficient of variation, the number of the variations, Is the rate of change; Collecting feature vectors at a plurality of moments to form a training set: , wherein, Is the number of feature samples; Dividing historical feature samples into Optimizing objective functions for each class of operating states In which, in the process, Represent the first The feature vector of each sample is used to determine, Represent the first The number of cluster categories is chosen, Represent the first The center of the feature of the class, Representing euclidean distances between vectors; For each category Calculating the upper and lower boundaries: In which, in the process, Represent the first The first sample is at The values in the dimensions of the individual features, Represent the first Class 1 The maximum value of the dimension is set, Represent the first Class 1 Dimensional minima.
10. The intelligent early warning system for the leakage risk of the large-diameter pipeline is used for realizing the early warning method according to any one of claims 1-9 and is characterized by comprising a model construction module, a response prediction module and an abnormality analysis module; The model construction module is used for acquiring the upstream/downstream operation data of the pipeline in real time by utilizing the sensor and constructing a boundary self-adaptive model based on the historical operation data; The response prediction module is used for constructing a multidimensional feature map of the internal state change of the pipeline through technologies such as pressure sensors, acoustic wave detection, optical fiber monitoring and the like, analyzing the dynamic response mode of the multidimensional feature map under the upstream/downstream fluctuation condition by using a map neural network and an LSTM algorithm; And the anomaly analysis module is used for comparing the current multi-dimensional feature map dynamic response with the predicted multi-dimensional feature map dynamic response, analyzing whether the anomaly dynamic exists based on the comparison result, and generating intelligent early warning prompt and recommended action by combining the anomaly level, the spatial position and the environmental information if the anomaly exists.

Description

Intelligent early warning system and method for leakage risk of large-diameter pipeline Technical Field The invention relates to the technical field of pipeline risk early warning, in particular to an intelligent early warning system and method for leakage risk of a large-diameter pipeline. Background With the continuous acceleration of the urban process and the rapid development of industrial infrastructure, various pipeline systems (such as urban water supply pipelines, industrial raw material conveying pipelines, petroleum and natural gas long-distance conveying pipelines and the like) are widely paved on the ground or the ground, especially large-diameter pipelines, and once leakage accidents occur, serious resource waste and economic loss can be caused, and safety accidents such as environmental pollution, ground collapse and even casualties can be possibly caused. The prior art has the following defects: 1. In the aspect of monitoring the internal state of a pipeline, single data dimension is mostly adopted for analysis, multi-dimensional and multi-mode sensing and joint analysis on a complex dynamic process in the pipeline are lacked, the influence of leakage or pipeline defects on the internal fluid state is difficult to comprehensively reflect, and the limitation makes early tiny leakage difficult to accurately identify and early warning response is lagged; 2. The traditional leakage early warning system generally lacks multi-factor comprehensive evaluation on abnormal events, often depends on a single index to judge leakage risk, so that the early warning information lacks rich hierarchical division and space positioning, and specific treatment suggestions and optimized response paths are difficult to provide for operation and maintenance personnel; 3. The traditional pipeline leakage detection method mostly depends on single type sensor data, or abnormal judgment is carried out based on an empirical threshold, complex variability of a pipeline operation environment and dynamic floating of working conditions are difficult to deal with, so that the early warning system is often insufficient in sensitivity to normal working condition change, a large number of false positives or false negatives are easy to occur, and the response efficiency and the pipe network safety guarantee capability of operation and maintenance personnel are affected. Based on the method, the intelligent early warning system and the intelligent early warning method for the leakage risk of the large-diameter pipeline are provided, the core problems of high false alarm rate, difficult early leakage identification, inaccurate early warning response and the like in the prior art are effectively solved, the accuracy, the instantaneity and the intelligent level of the intelligent early warning system for the leakage risk of the large-diameter pipeline are greatly improved through multi-dimensional, multi-modal and multi-factor intelligent fusion analysis, and a powerful technical guarantee is provided for the safe operation of a pipeline network. Disclosure of Invention The invention aims to provide an intelligent early warning system and method for leakage risk of a large-diameter pipeline, and aims to solve the defects in the background technology. In order to achieve the purpose, the invention provides the technical scheme that the intelligent early warning method for the leakage risk of the large-diameter pipeline comprises the following steps: collecting upstream/downstream operation data of a pipeline in real time, and constructing a boundary self-adaptive model; Constructing a multidimensional feature map of the internal state change of the pipeline, and analyzing a dynamic response mode of the multidimensional feature map under the upstream/downstream fluctuation condition; Comparing the current multi-dimensional feature map dynamic response with the predicted multi-dimensional feature map dynamic response, analyzing whether abnormal dynamics exist based on the comparison result, and if so, generating intelligent early warning prompt and recommending actions. In a preferred embodiment, comparing the current multi-dimensional feature map dynamic response with the predicted multi-dimensional feature map dynamic response, analyzing whether there is abnormal dynamic based on the comparison result, comprising the steps of: Synchronous acquisition of true values during operation And predicted valueComparing, calculating the node prediction error; Weighting and aggregating the difference indexes of all nodes, and calculating the abnormality degree of the overall graph structure; When (when)In the time-course of which the first and second contact surfaces,The dynamic abnormality threshold is determined to be a dynamic abnormality response. In a preferred embodiment, the difference indexes of all nodes are weighted and aggregated to calculate the degree of abnormality of the overall graph structure:, wherein,Representing the total num