Intelligent diagnosis method for pressure anomaly of directly buried steam pipeline

Abstract

The application discloses a direct-buried steam pipeline pressure anomaly intelligent diagnosis method, comprising the following steps: collecting original data of a plurality of pressure monitoring nodes, and generating a standardized pressure data sequence through time synchronization and interpolation completion; then adopting double-path parallel processing, wherein one path applies a Hampel algorithm to identify local abnormal fluctuations and form candidate abnormal points, and the other path retains original pressure data; based on the candidate abnormal points, data of a corresponding time period is extracted, and a Bayesian online variable point detection method is used to infer variable point posterior probability; by analyzing the probability change trend, an abnormal event point is identified, and an abnormal propagation path is constructed in combination with a topological structure and a flow direction of the steam pipeline, so that a source position and a first abnormal occurrence time are finally determined, and a complete abnormal diagnosis result is output. The application realizes early identification and tracing of potential faults of the steam pipeline.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial process monitoring and fault diagnosis, and particularly relates to an intelligent diagnosis method for pressure abnormality of a directly buried steam pipeline. BACKGROUND

[0002] With the continuous expansion of the scale and the continuous rise of the operation complexity of urban heating systems, the steam pipeline is prone to abnormal events such as leakage, blockage or pipe explosion in a long-term high-temperature and high-pressure environment, which seriously affects the heating efficiency and operation safety. At present, the state monitoring of the directly buried steam pipeline mainly relies on pressure sensors arranged at key nodes, and abnormality recognition is realized by combining a fixed threshold alarm mechanism or a statistical model based on overall trend analysis.

[0003] However, the prior art has obvious deficiencies in dealing with local sudden abnormality and weak signal fluctuation. On the one hand, the traditional abnormality detection method lacks sensitivity to local abnormal fluctuation, and is prone to miss short-term weak but precursory abnormal signs, resulting in delayed diagnosis or misjudgment; on the other hand, the existing variable point detection method relies on static prior information, and it is difficult to accurately identify the abnormal propagation path and its source in the pressure data, and cannot effectively realize joint reasoning and accurate positioning between multiple points of data.

[0004] Therefore, how to provide an intelligent diagnosis method for pressure abnormality of a directly buried steam pipeline is a problem to be solved by those skilled in the art. SUMMARY

[0005] One object of the present application is to provide an intelligent diagnosis method for pressure abnormality of a directly buried steam pipeline. The present application performs double-path processing on a standardized pressure data sequence, constructs a dynamic prior probability structure in combination with candidate abnormal points, and identifies the abnormal propagation path and the abnormal source location on the basis of the enhanced variable point posterior probability, and has the advantages of high diagnosis accuracy, fast response speed and strong adaptability.

[0006] The intelligent diagnosis method for pressure abnormality of a directly buried steam pipeline according to an embodiment of the present application comprises the following steps:

[0007] Collecting original pressure data of a plurality of pressure monitoring nodes, and obtaining steam pipeline topology structure information and steam flow direction information;

[0008] Performing time synchronization and missing interpolation processing on the original pressure data to generate a standardized pressure data sequence;

[0009] Performing double-path processing on the standardized pressure data sequence, wherein the first path adopts an adaptive sliding window Hampel algorithm to detect local abnormal fluctuation and generate a candidate abnormal point index set, and the second path retains the standardized pressure data sequence;

[0010] Based on the candidate abnormal point index set, the standardized pressure data of the corresponding time period is extracted, when the cumulative amplitude of the change point probability exceeds the preset threshold, the Bayesian online change point detection method is used for inference, and the change point posterior probability sequence is obtained;

[0011] According to the continuous probability rising section of the change point posterior probability sequence, the intersection processing is performed with the candidate abnormal point index set, and the abnormal event point set is generated;

[0012] The abnormal event point set is mapped to the standardized pressure data sequence of each pressure monitoring node, the change point posterior probability difference value of adjacent pressure monitoring nodes in the same time period is calculated, and the abnormal propagation path is constructed in combination with the steam pipeline topology structure information and the steam flow direction information;

[0013] According to the change point posterior probability difference value direction and time sequence in the abnormal propagation path, the abnormal source pressure monitoring node and the first abnormal occurrence time are determined, and the abnormal diagnosis result is output.

[0014] Optionally, the steam pipeline topology structure information and the steam flow direction information include:

[0015] The steam pipeline topology structure information includes pressure monitoring node identification information, spatial coordinate information of the pressure monitoring node, connection relationship information between the pressure monitoring nodes, and physical length information of each connected pipe section, the connection relationship information indicates whether there is a pipeline connection between adjacent pressure monitoring nodes, and the physical length information indicates the axial distance of the connected pipe section;

[0016] The steam flow direction information includes steam source point identification, preset steam flow direction information in each pipe section, and a node inter-flow relationship matrix, the flow direction information indicates the transmission path of steam in the pipeline network from upstream to downstream, and the flow direction relationship matrix is a directed relationship matrix, each element of which indicates whether the steam flow direction between the corresponding node pair exists and its directionality.

[0017] Optionally, the generation of the standardized pressure data sequence includes:

[0018] Time synchronization processing is performed on the original pressure data collected by the plurality of pressure monitoring nodes, a unified time index set is constructed, a unified sampling period is set as a time reference, the original sampling time of each pressure monitoring node is aligned with the unified sampling period, and a preliminary synchronized pressure data sequence is generated;

[0019] In the synchronized pressure data sequence, the time points not mapped by the original pressure data are identified, and a missing data point set is generated;

[0020] For each missing data point in the set of missing data points, find its adjacent previous valid pressure data point and next valid pressure data point in the synchronized pressure data sequence of the corresponding pressure monitoring node, if the time interval between the two valid pressure data points is less than a preset interpolation threshold, and the pressure difference between the two valid pressure data points is less than a preset fluctuation threshold, then perform interpolation completion processing on the missing data point according to the trend between the adjacent valid pressure data points, and write the completion result into the synchronized pressure data sequence;

[0021] Perform consistency check on the synchronized pressure data sequence after interpolation completion, delete the local mutation data points caused by interpolation completion in the synchronized pressure data sequence, and obtain the verified pressure data sequence;

[0022] Reorder the verified pressure data sequence according to the set of uniform time indexes, and combine it into a uniform format with each pressure monitoring node as a dimension to form a standardized pressure data sequence.

[0023] Optionally, the double-path processing of the standardized pressure data sequence includes:

[0024] The standardized pressure data sequence is input into the first path and the second path for parallel processing, respectively;

[0025] In the first path, the standardized pressure data sequence of each pressure monitoring node is processed by the Hampel algorithm based on an adaptive sliding window, which is used to detect local abnormal fluctuations in the standardized pressure data sequence;

[0026] A sliding window structure is constructed, the length and step size of the sliding window are set, and the data subsequence corresponding to the sliding window is extracted along the set of uniform time indexes;

[0027] In each sliding window, the median of the corresponding data subsequence is calculated, and the median absolute deviation of the corresponding data subsequence is calculated based on the absolute deviation value of each data point from the median;

[0028] Determine the data point corresponding to the center position of the sliding window, and judge whether the deviation between the data point and the median is greater than the product of the median absolute deviation and a set multiple threshold, if the determination condition is met, the corresponding data point is identified as a local abnormal fluctuation, and the corresponding data point is marked as a candidate abnormal point;

[0029] Record the position index of all data points marked as candidate abnormal points in the standardized pressure data sequence, and aggregate to form a set of candidate abnormal point indexes;

[0030] In the second path, the standardized pressure data sequence of each pressure monitoring node is written into a data buffer structure in time sequence;

[0031] The candidate abnormal point index set formed in the first path is passed to the second path, and according to the time index in the candidate abnormal point index set, a standardized pressure data sub-segment corresponding to the time index in the second path is extracted.

[0032] Optionally, the generation of the sequence of change point posterior probabilities comprises:

[0033] A time index corresponding to each candidate abnormal point in the candidate abnormal point index set is obtained, and a standardized pressure data symmetrically before and after the corresponding time index is extracted from a standardized pressure data sequence corresponding to a pressure monitoring node to which the corresponding candidate abnormal point belongs according to a preset fixed window length, to form a standardized pressure data sub-segment;

[0034] The standardized pressure data sub-segment is passed as input to the Bayesian online change point detection method, and the Bayesian online change point detection method outputs a sequence of change point posterior probabilities corresponding to each time point in the standardized pressure data sub-segment based on the pressure values of the time points in the standardized pressure data sub-segment;

[0035] The sequence of change point posterior probabilities is analyzed to identify a change in amplitude between a maximum probability value and a minimum probability value in the sequence of change point posterior probabilities, and the change in amplitude is taken as a change point probability cumulative amplitude value;

[0036] The change point probability cumulative amplitude value is compared with a preset change point probability amplitude determination threshold value, and if the change point probability cumulative amplitude value exceeds the change point probability amplitude determination threshold value, the time index corresponding to the candidate abnormal point is confirmed as an effective abnormal trigger point.

[0037] Optionally, the Bayesian online change point detection method comprises:

[0038] A time index corresponding to each candidate abnormal point in the candidate abnormal point index set is obtained, and a standardized pressure data symmetrically before and after the corresponding time index is extracted from a standardized pressure data sequence corresponding to a pressure monitoring node to which the corresponding candidate abnormal point belongs according to a preset fixed window length, to form a standardized pressure data sub-segment;

[0039] For the standardized pressure data sub-segment formed, a dynamic prior probability structure containing candidate abnormal point information is constructed, and each time index in the standardized pressure data sub-segment is sequentially traversed to determine whether each time index belongs to the candidate abnormal point index set, if it does, the change point prior probability of the time index is set to an enhanced multiple of the base prior probability, and if it does not, the change point prior probability of the corresponding time index is maintained as the default value of the base prior probability;

[0040] The standardized pressure data sub-segment containing the dynamic prior probability structure is input into a Bayesian online change point detection method, a recursive-based joint posterior inference operation is performed, and a change point posterior probability sequence corresponding to each time index in the standardized pressure data sub-segment is output;

[0041] The output change point posterior probability sequence is enhanced, each time index in the change point posterior probability sequence corresponding to a candidate abnormal point index set is identified, a local symmetric response window is constructed around the identified time index according to a preset length, and a confidence enhancement operation is performed on the change point posterior probability value at each time index in the response window, including weighted smoothing and trend interpolation processing, to generate an enhanced change point posterior probability sequence.

[0042] Optionally, the generation of the abnormal event point set comprises:

[0043] The enhanced change point posterior probability sequence corresponding to each pressure monitoring node is obtained, and the change point posterior probability sequences are scanned time index by time index according to a unified time index set;

[0044] In the change point posterior probability sequence of each pressure monitoring node, a time period in which the change point posterior probability value continuously rises with the time index is identified, which is defined as a continuous probability rising section. The continuous probability rising section needs to meet the following conditions: the change point posterior probability values corresponding to adjacent time indexes monotonically increase or fluctuate within a limited range, and the difference between the change point posterior probability values at the beginning and end of the corresponding time period is not less than a set minimum amplitude threshold, and the time span is not less than a set minimum duration threshold;

[0045] All time indexes that meet the conditions are extracted and aggregated to form a suspected abnormal time index set;

[0046] The candidate abnormal point index set is obtained, the intersection processing based on the unified time index set is performed on the suspected abnormal time index set and the candidate abnormal point index set, and each time index in the intersection is extracted;

[0047] Each time index in the intersection is bound with the pressure monitoring node identifier to which it belongs to form a pressure monitoring node-time index binary tuple, and all binary tuples are aggregated to generate an abnormal event point set.

[0048] Optionally, the construction of the abnormal propagation path comprises:

[0049] Each pressure monitoring node-time index binary tuple in the abnormal event point set is obtained, and the corresponding change point posterior probability value in the enhanced change point posterior probability sequence based on the time index is extracted to form an abnormal feature set containing the abnormal event point and the change point posterior probability;

[0050] According to the steam pipeline topology information, a set of directly adjacent pressure monitoring nodes of each abnormal event node is determined, and for each pair of adjacent pressure monitoring nodes having a connection relationship, enhanced change point posterior probability values at the same time index are extracted under a unified time index set;

[0051] Based on the steam flow direction information, a set of downstream adjacent pressure monitoring nodes of the pressure monitoring node corresponding to each abnormal event point is obtained, and a difference value calculation is performed on the change point posterior probability values of the abnormal event point and its downstream adjacent nodes in the unified time index set, to obtain a change point posterior probability difference value of each node pair at the corresponding time index;

[0052] Node pairs with a change point posterior probability difference value greater than a preset threshold are screened out, and whether the difference value direction is consistent with the steam flow direction is confirmed according to the steam flow direction information, and if consistent, the corresponding node pair is defined as an abnormal propagation edge;

[0053] All abnormal propagation edges are path-spliced in the order of time indexes, and an abnormal propagation path composed of multiple abnormal event points in the propagation order is constructed in combination with the spatial coordinate information of each pressure monitoring node in the steam pipeline topology information, and an abnormal propagation path set is formed by collecting all paths.

[0054] Optionally, the output of the abnormal diagnosis result includes:

[0055] Each abnormal propagation path in the abnormal propagation path set is obtained, the abnormal trigger frequency of the pressure monitoring node corresponding to the path starting point under the unified time index set is counted, and the time index at which the node is earliest identified as an abnormal event point is recorded to form a starting node candidate set and a corresponding time index set;

[0056] In the starting node candidate set, a path starting point is screened out, in which the change point posterior probability difference value direction in the abnormal propagation path is always consistent with the steam flow direction, and the change point posterior probability values of the pressure monitoring nodes on the path present a monotonically decreasing trend, to form a suspected abnormal source node set;

[0057] Each node in the suspected abnormal source node set is sorted according to the first appearance time index of the node in the abnormal event point set, and the suspected abnormal source node with the earliest time index is extracted as an abnormal source pressure monitoring node;

[0058] The abnormal source pressure monitoring node and the corresponding first appearance time index are defined as the abnormal source pressure monitoring node and the first abnormal occurrence time respectively, and are output together with the corresponding abnormal propagation path to form an abnormal diagnosis result.

[0059] The beneficial effects of the present application are:

[0060] (1) The intelligent diagnosis mechanism for identifying local pressure fluctuation and variable point anomaly in the direct-buried steam pipeline in real time is constructed by combining the Hampel algorithm of the adaptive sliding window with the Bayesian online variable point detection method, which can effectively capture early abnormal signs and improve the response ability of the system to complex dynamic pressure anomalies.

[0061] (2) The detection sensitivity and accuracy of abnormal signals are enhanced, false positives and false negatives are reduced, and the diagnosis stability and repeatability are improved by introducing a dynamic prior enhancement mechanism based on candidate abnormal points into the Bayesian online variable point detection method and implementing confidence enhancement processing on the variable point posterior probability sequence.

[0062] (3) The abnormal propagation path is constructed by fusing the steam pipeline topological structure information and the steam flow direction information, and the abnormal source node and the first abnormal occurrence time are identified by combining the variable point posterior probability difference direction and the time sequence, which enhances the physical rationality of abnormal positioning and the result interpretability, and is suitable for intelligent operation and maintenance requirements of complex multi-node steam pipe network. BRIEF DESCRIPTION OF DRAWINGS

[0063] The accompanying drawings are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application, and do not constitute a limitation of the application. In the drawings:

[0064] Fig. 1 The flowchart of the intelligent diagnosis method for pressure anomalies in the direct-buried steam pipeline proposed in the present application;

[0065] Fig. 2 The processing flowchart of the Bayesian online variable point detection method with dynamic prior enhancement introduced in the present application. DETAILED DESCRIPTION

[0066] The application will now be described in further detail with reference to the drawings. These drawings are simplified schematic diagrams that only schematically show the basic structure of the application, and therefore only show the components related to the application.

[0067] Reference Figs. 1-2 The intelligent diagnosis method for pressure anomalies in the direct-buried steam pipeline comprises the following steps:

[0068] Collecting raw pressure data of multiple pressure monitoring nodes and obtaining steam pipeline topological structure information and steam flow direction information;

[0069] Performing time synchronization and missing interpolation processing on the raw pressure data to generate a standardized pressure data sequence;

[0070] The double-path processing is performed on the standardized pressure data sequence, wherein the first path adopts the Hampel algorithm of the adaptive sliding window to detect local abnormal fluctuations and generate a candidate abnormal point index set, and the second path retains the standardized pressure data sequence;

[0071] The standardized pressure data of the corresponding time period is extracted based on the candidate abnormal point index set, and when the cumulative amplitude of the change point probability exceeds a preset threshold, a Bayesian online change point detection method is used for inference to obtain a change point posterior probability sequence;

[0072] According to the continuous probability rising section of the change point posterior probability sequence, the intersection processing is performed with the candidate abnormal point index set to generate an abnormal event point set;

[0073] The abnormal event point set is mapped to the standardized pressure data sequence of each pressure monitoring node, the change point posterior probability difference value of adjacent pressure monitoring nodes in the same time period is calculated, and the abnormal propagation path is constructed in combination with the steam pipeline topology structure information and the steam flow direction information;

[0074] According to the direction and time sequence of the change point posterior probability difference value in the abnormal propagation path, the abnormal source pressure monitoring node and the first abnormal occurrence time are determined, and an abnormal diagnosis result is output.

[0075] In the embodiment, the steam pipeline topology structure information and the steam flow direction information include:

[0076] The steam pipeline topology structure information includes pressure monitoring node identification information, spatial coordinate information of the pressure monitoring node, connection relationship information between the pressure monitoring nodes, and physical length information of each connected pipe section, the connection relationship information indicates whether there is a pipeline connection between adjacent pressure monitoring nodes, and the physical length information indicates the axial distance of the connected pipe section;

[0077] The steam flow direction information includes steam source point identification, preset steam flow direction information in each pipe section, and a node inter-flow relationship matrix, the flow direction information indicates the transmission path of steam in the pipeline network from upstream to downstream, and the flow direction relationship matrix is a directed relationship matrix, each element of which indicates whether there is a steam flow direction between the corresponding node pair and its directionality;

[0078] The steam pipeline topology structure information and the steam flow direction information constitute a static structure data set for representing the structural connection mode and flow transmission relationship of steam in the pipeline network.

[0079] In the embodiment, the generation of the standardized pressure data sequence includes:

[0080] Perform time synchronization processing on the original pressure data collected by the plurality of pressure monitoring nodes, construct a unified time index set, set a unified sampling period as a time reference, align the original sampling time of each pressure monitoring node with the unified sampling period, and generate a preliminary synchronized pressure data sequence;

[0081] Identify time points in the synchronized pressure data sequence that are not mapped by the original pressure data, and generate a set of missing data points;

[0082] For each missing data point in the set of missing data points, find its adjacent previous valid pressure data point and next valid pressure data point in the synchronized pressure data sequence of the corresponding pressure monitoring node. If the time interval between the two valid pressure data points is less than a preset interpolation threshold, and the pressure difference between the two valid pressure data points is less than a preset fluctuation threshold, perform interpolation completion processing on the missing data point according to the trend between the adjacent valid pressure data points, and write the completion result to the synchronized pressure data sequence;

[0083] Perform consistency check on the synchronized pressure data sequence after interpolation completion, delete local abrupt data points caused by interpolation completion in the synchronized pressure data sequence, and obtain the verified pressure data sequence;

[0084] Reorder the verified pressure data sequence according to the unified time index set, and combine it into a unified format with each pressure monitoring node as a dimension to form a standardized pressure data sequence.

[0085] In this embodiment, the double-path processing of the standardized pressure data sequence includes:

[0086] The standardized pressure data sequence is input into the first path and the second path for parallel processing;

[0087] In the first path, the standardized pressure data sequence of each pressure monitoring node is processed by the Hampel algorithm based on an adaptive sliding window, which is used to detect local abnormal fluctuations in the standardized pressure data sequence;

[0088] A sliding window structure is constructed, the sliding window length and sliding step are set, and the data subsequence corresponding to the sliding window is extracted along the unified time index set;

[0089] In each sliding window, the median of the corresponding data subsequence is calculated, and the median absolute deviation of the corresponding data subsequence is calculated based on the absolute deviation value of each data point from the median;

[0090] Determine the data point corresponding to the center position of the sliding window, and judge whether the deviation between the data point and the median is greater than the product of the median absolute deviation and a set multiple threshold. If the determination condition is met, the corresponding data point is identified as a local abnormal fluctuation, and the corresponding data point is marked as a candidate abnormal point.

[0091] Record the position index of all data points marked as candidate abnormal points in the standardized pressure data sequence, and aggregate to form a candidate abnormal point index set;

[0092] In the second path, the standardized pressure data sequence of each pressure monitoring node is written into the data buffer structure in time sequence;

[0093] The candidate abnormal point index set formed in the first path is transmitted to the second path, and according to the time index in the candidate abnormal point index set, the corresponding standardized pressure data sub-segment in the second path is extracted.

[0094] In this embodiment, the generation of the change point posterior probability sequence includes:

[0095] Obtain the time index corresponding to each candidate abnormal point in the candidate abnormal point index set, and extract the standardized pressure data symmetrically before and after the corresponding time index in the standardized pressure data sequence corresponding to the pressure monitoring node to which the corresponding candidate abnormal point belongs according to a preset fixed window length, to form a standardized pressure data sub-segment;

[0096] The standardized pressure data sub-segment is transmitted as input to the Bayesian online change point detection method, and the Bayesian online change point detection method outputs a change point posterior probability sequence corresponding to each time point based on the pressure values of each time point in the standardized pressure data sub-segment;

[0097] Analyze the change point posterior probability sequence, identify the amplitude change between the maximum probability value and the minimum probability value in the change point posterior probability sequence, and take the amplitude change as the change point probability cumulative amplitude;

[0098] Compare the change point probability cumulative amplitude with a preset change point probability amplitude determination threshold, and if the change point probability cumulative amplitude exceeds the change point probability amplitude determination threshold, the time index corresponding to the candidate abnormal point is confirmed as an effective abnormal trigger point.

[0099] In this embodiment, the Bayesian online change point detection method includes:

[0100] Obtain the time index corresponding to each candidate abnormal point in the candidate abnormal point index set, and extract the standardized pressure data symmetrically before and after the corresponding time index in the standardized pressure data sequence corresponding to the pressure monitoring node to which the corresponding candidate abnormal point belongs according to a preset fixed window length, to form a standardized pressure data sub-segment;

[0101] For the composed standardized pressure data sub-section, a dynamic prior probability structure containing candidate abnormal point information is constructed, each time index in the standardized pressure data sub-section is traversed in turn, it is judged whether each time index belongs to the candidate abnormal point index set, if it belongs, the change point prior probability of the time index is set to the enhanced multiple of the basic prior probability, if it does not belong, the change point prior probability of the corresponding time index is kept as the default value of the basic prior probability;

[0102] The standardized pressure data sub-section containing the dynamic prior probability structure is input into the Bayesian online change point detection method, a recursive joint posterior inference operation is performed, and a change point posterior probability sequence corresponding to each time index in the standardized pressure data sub-section is output;

[0103] The output change point posterior probability sequence is enhanced, each time index in the change point posterior probability sequence corresponding to the candidate abnormal point index set is identified, and a local symmetric response window is constructed according to the preset length, centered on the identified time index, the change point posterior probability value at each time index is processed in the response window, including weighted smoothing and trend interpolation processing, to generate an enhanced change point posterior probability sequence;

[0104] The dynamic prior probability structure containing the candidate abnormal point information is constructed, including establishing a prior probability value mapping relationship corresponding to each time index in the standardized pressure data sub-section as an index dimension, the structure is composed of a prior probability vector, each element of which corresponds to a time index in the standardized pressure data sub-section and records the prior probability value of the change point occurrence corresponding to the index, the prior probability value is set according to the candidate abnormal point index set, if a time index belongs to the candidate abnormal point index set, its corresponding prior probability value is increased by a fixed enhancement proportion from the basic prior probability, if it does not belong, it remains the basic prior probability, the basic prior probability is obtained according to the default crisis function in the Bayesian online change point detection method, the dynamic prior probability structure as a whole shows a linear array structure with the same length as the standardized pressure data sub-section;

[0105] The performing a recursive joint posterior inference operation refers to recursively calculating the posterior probability of a current time being a change point according to a time sequence processing manner of a Bayesian online change point detection method under the condition of a standardized pressure data subsegment and a corresponding dynamic prior probability structure, and specific operations include: initializing a change point occurrence probability at a time index of zero; for each subsequent time index, calculating a joint posterior probability of the time index being a change point position based on a change point posterior distribution at a previous time, a standardized pressure value at the current time index, and a corresponding prior probability value; and updating and normalizing the joint posterior probability at all time indexes in the range of the standardized pressure data subsegment to finally obtain a change point posterior probability sequence corresponding to each time index in the standardized pressure data subsegment.

[0106] In the embodiment, the generation of the abnormal event point set includes:

[0107] The enhanced change point posterior probability sequence corresponding to each pressure monitoring node is obtained, and the change point posterior probability sequences are scanned according to the uniform time index set;

[0108] In the change point posterior probability sequence of each pressure monitoring node, a time period in which the change point posterior probability value continuously rises with the time index is identified, and is defined as a continuous probability rising section. The continuous probability rising section needs to meet the following conditions: the change point posterior probability values corresponding to adjacent time indexes monotonically increase or fluctuate and rise in a limited range, and the difference between the change point posterior probability values at the beginning and end of the corresponding time period is not less than a set minimum amplitude threshold, and the time span is not less than a set minimum duration threshold.

[0109] All time indexes meeting the conditions are extracted, and are summarized to form a suspected abnormal time index set;

[0110] The candidate abnormal point index set is obtained, and the intersection processing based on the uniform time index set is performed on the suspected abnormal time index set and the candidate abnormal point index set, and each time index in the intersection is extracted;

[0111] Each time index in the intersection is bound with the pressure monitoring node identifier to which the time index belongs, a pressure monitoring node-time index binary tuple is formed, and all binary tuples are summarized to generate an abnormal event point set;

[0112] The minimum increase threshold is a determination criterion for determining whether the probability rising trend in the change point posterior probability sequence is significant, which represents the minimum cumulative growth amplitude between the posterior probability values in the continuous time index. If the posterior probability value is monotonically increasing in a continuous time index interval, and the probability difference between the beginning and the end exceeds the minimum increase threshold, it is considered that there is a significant probability rising trend. The minimum duration threshold is a determination criterion for limiting the shortest time length required to maintain the rising trend, which represents the minimum length of the continuous time index interval corresponding to the above-mentioned probability rising trend. Only when the continuous rising section of the posterior probability satisfies both the minimum increase threshold and the minimum duration threshold, it is identified as an effective continuous probability rising section.

[0113] In the embodiment, the construction of the abnormal propagation path includes:

[0114] Each pressure monitoring node-time index pair in the abnormal event point set is obtained, and the corresponding change point posterior probability value in the enhanced change point posterior probability sequence is extracted based on the time index to form an abnormal feature set containing the abnormal event point and the change point posterior probability;

[0115] According to the steam pipeline topology structure information, a set of directly adjacent pressure monitoring nodes of each abnormal event node is determined, and for each pair of adjacent pressure monitoring nodes having a connection relationship, the enhanced change point posterior probability values thereof at the same time index are extracted under the unified time index set;

[0116] Based on the steam flow direction information, a set of downstream adjacent pressure monitoring nodes of each pressure monitoring node corresponding to the abnormal event point is obtained, and the change point posterior probability values of the abnormal event point and its downstream adjacent nodes are subjected to difference calculation in the unified time index set to obtain the change point posterior probability difference values of each node pair at the corresponding time index;

[0117] Node pairs with a change point posterior probability difference greater than a preset threshold are screened out, and whether the difference direction is consistent with the steam flow direction is confirmed according to the steam flow direction information. If they are consistent, the corresponding node pair is defined as an abnormal propagation edge;

[0118] All abnormal propagation edges are spliced in the order of time index, and the abnormal propagation path composed of multiple abnormal event points in the propagation order is constructed in combination with the spatial coordinate information of each pressure monitoring node in the steam pipeline topology structure information, and all paths are aggregated to form an abnormal propagation path set.

[0119] In the embodiment, the output of the abnormal diagnosis result includes:

[0120] For each abnormal propagation path in the set of abnormal propagation paths, the frequency of abnormal triggering of the pressure monitoring node corresponding to the starting point of the path under the set of uniform time indexes is counted, and the time index at which the node is earliest identified as an abnormal event point is recorded, forming a set of starting node candidates and a corresponding set of time indexes;

[0121] In the set of starting node candidates, the starting points of the paths in which the difference in the posterior probability of the change point in the abnormal propagation path is always consistent with the direction of steam flow, and the posterior probability values of the change point of each pressure monitoring node on the path present a monotonically decreasing trend, are screened out to form a set of suspected abnormal source nodes.

[0122] For each node in the set of suspected abnormal source nodes, the nodes are sorted according to the first occurrence time index of the node in the set of abnormal event points, and the suspected abnormal source node with the earliest time index is extracted as the abnormal source pressure monitoring node.

[0123] The abnormal source pressure monitoring node and its corresponding first occurrence time index are defined as the abnormal source pressure monitoring node and the first abnormal occurrence time, respectively, and are output together with the corresponding abnormal propagation path to form an abnormal diagnosis result.

[0124] Example 1: In order to verify the feasibility of the present application in implementation, the present application is applied to a city energy group to deploy and verify the application of an intelligent pressure abnormal diagnosis system for a buried steam main pipeline with a total length of about 13.7 kilometers in its jurisdiction. The steam pipeline connects multiple heat exchange stations, and the service area includes residential areas, office areas and industrial parks, with high overall heat load and complex structure, and potential problems such as steam leakage, aging of insulation layer or fatigue of welding points exist in year-round operation. Since the pipeline is completely buried underground, traditional inspection methods rely on manual and periodic temperature measurement points, and there are problems such as response delay, inaccurate fault location, and inability to identify abnormalities in real time.

[0125] A total of 23 high-precision pressure monitoring nodes are installed along the pipeline, with a sampling period of 10 seconds, and the data is transmitted in real time to the dispatching master station through the LoRa wireless network. The data processing server is deployed in a local IDC machine room, equipped with an industrial computing platform with a 16-core CPU and 64 GB of memory, and runs the diagnostic system software proposed in the present application.

[0126] In the data preprocessing stage, the system first performs uniform time alignment processing on the original data collected by the 23 pressure monitoring nodes. All node data is supplemented based on the set of uniform time indexes. For missing data points, the system uses local linear trend interpolation method to complete the missing data according to the interval and pressure difference between adjacent valid data points. The completed data is subjected to consistency verification and local mutation elimination, and finally generates a standardized pressure data sequence as the input for subsequent analysis.

[0127] In the diagnostic analysis phase, the standardized pressure data sequence is sent into the dual-path analysis module in parallel. The first path uses the Hampel algorithm with adaptive sliding window to perform local abnormal fluctuation detection on the sequence of each node. The system sets the sliding window length to 20 sampling points, the sliding step to 5 sampling points, and uses the median and the median absolute deviation for judgment. The center point with a deviation greater than 3 times the MAD is marked as a candidate abnormal point. The second path synchronously caches all node data and extracts the data sub-section corresponding to the time period according to the candidate abnormal point index detected by the first path.

[0128] The extracted data sub-section is input into the Bayesian online change point detection model. When the model is executed, the system constructs a dynamic prior structure based on the candidate abnormal point, enhances the prior probability at the candidate index, and simultaneously performs a recursive joint posterior probability update operation. The output result is the change point posterior probability value at each time point. The system further enhances the probability sequence, performs weighted smoothing and trend interpolation within the local symmetric window centered on the candidate point, and generates the enhanced change point posterior probability sequence.

[0129] According to the change point posterior probability sequence, the system identifies multiple continuous probability rising sections. The system sets the minimum amplitude threshold to 0.15 and the minimum duration threshold to 120 seconds, extracts the suspected abnormal time index that meets the conditions, and performs intersection processing with the candidate abnormal point index set to finally identify the abnormal event point set. Each abnormal event point is composed of a monitoring node identifier and a time index, accurately reflecting the location and time of abnormal occurrence.

[0130] The system calculates the change point probability difference between abnormal event points based on the actual topological structure and steam flow direction of the steam pipeline, constructs the abnormal propagation path, and matches the nodes with smaller posterior probabilities downstream from the abnormal event point. The system confirms whether the path direction conforms to the steam flow direction, retains only the abnormal propagation paths that meet the conditions, and finally constructs multiple abnormal propagation path subgraphs.

[0131] The system automatically sorts and summarizes all abnormal propagation paths, identifies the first abnormal node in the propagation path as the abnormal source pressure monitoring node, records the time of the first abnormal occurrence, and forms a structured abnormal diagnosis report.

[0132] In this embodiment, the abnormal fluctuation is first detected at node D-21, the initial posterior probability value rapidly rises to 0.78, and then nodes D-20 and D-19 successively detect similar features after 4 minutes and 6 minutes. The system successfully constructs an abnormal path propagating from D-21 to D-19 and finally confirms that node D-21 is the abnormal source point. The following is a data diagram of this abnormal event:

[0133] Table 1 Part of abnormal propagation path node variable point posterior probability value table

[0134]

[0135] To verify the actual effect of the method of the present application, we selected typical leakage events recorded by the energy company in the past two years for backtracking test, deployed the same nodes in the middle of a total of 6 test pipe sections and reproduced verification with the diagnosis system, and compared with the traditional temperature sensing anomaly detection method.

[0136] Table 2 Comparison table of detection performance of new and old methods in typical anomaly detection scenarios

[0137]

[0138] In Table 1, the D-21 node first showed a significant jump in variable point posterior probability at 14:35:20, with a value as high as 0.78, much higher than the anomaly judgment threshold (usually set around 0.5), indicating that it is the earliest point of local pressure fluctuation. Subsequently, the downstream nodes D-20 and D-19 showed an increase in variable point probability after about 4 minutes and 6 minutes, respectively, with values of 0.69 and 0.56, respectively, showing a decreasing trend, consistent with the propagation characteristics of the steam anomaly in the actual physical flow direction. The probability values of the next D-18 and D-17 nodes further decreased to 0.34 and 0.22, respectively, indicating that the abnormal influence gradually attenuated, further verifying the rationality and time continuity of the abnormal propagation path constructed by the system.

[0139] In Table 2, the system performed significantly better than the traditional method in the 6 typical anomaly type scenarios. The average delay of the traditional temperature sensing detection method was 33.83 minutes, while the present application was only 5.5 minutes, with an advance response time of more than 85%. In terms of spatial positioning, the average error of the traditional method was 161.5 meters, while the present application was only 27.5 meters, with an error reduction of more than 80%, greatly improving the precision positioning ability of the abnormal source. These results fully demonstrate that the present application has significant advantages in extracting the time and spatial characteristics of abnormal occurrence by combining the Hampel algorithm and the Bayesian online variable point detection strategy, and is particularly suitable for online monitoring and precise diagnosis of complex buried steam pipe network systems.

[0140] In summary, the present embodiment successfully deployed and verified the effectiveness of the present application in a real urban steam pipe network environment, not only achieving rapid identification and propagation path tracking of abnormal events, but also having good universality and engineering landing ability. In the future, this method is expected to be popularized to other urban steam systems, heat pipe networks and high-risk industrial pipeline anomaly diagnosis.

[0141] The above merely describes preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art, according to the technical solution and inventive concept of the present application, makes equivalent replacement or change within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for intelligent diagnosis of abnormal pressure in directly buried steam pipelines, characterized in that: include: Raw pressure data from multiple pressure monitoring nodes were collected, and information on the topology of the steam pipeline and the direction of steam flow was obtained. Time synchronization and missing interpolation are performed on the raw pressure data to generate a standardized pressure data sequence; A dual-path processing is performed on the standardized stress data sequence. The first path uses the Hampel algorithm with an adaptive sliding window to detect local abnormal fluctuations and generate a set of candidate anomaly indexes. The second path preserves the standardized stress data sequence. Based on the candidate outlier index set, standardized stress data for the corresponding time period is extracted. When the cumulative amplitude of the change point probability exceeds the preset threshold, the Bayesian online change point detection method is used for inference to obtain the change point posterior probability sequence. Based on the continuous probability increasing segment of the posterior probability sequence of the change point, the intersection with the candidate outlier index set is processed to generate an outlier event point set. The set of abnormal event points is mapped to the standardized pressure data sequence of each pressure monitoring node, the difference in the posterior probability of change points of adjacent pressure monitoring nodes in the same time period is calculated, and the abnormal propagation path is constructed by combining the steam pipeline topology information and steam flow direction information. Based on the direction and time sequence of the posterior probability difference of the change points in the abnormal propagation path, the pressure monitoring node of the abnormal source and the time of its first abnormal occurrence are determined, and the abnormal diagnosis results are output. The dual-path processing of the standardized pressure data sequence includes: The standardized pressure data sequence is input into the first path and the second path respectively for parallel processing; In the first path, the standardized pressure data sequence of each pressure monitoring node is processed by the Hampel algorithm based on an adaptive sliding window to detect local abnormal fluctuations in the standardized pressure data sequence. Construct a sliding window structure, set the sliding window length and sliding step size, and extract the data subsequences corresponding to the sliding window sequentially along a unified time index set; Within each sliding window, the median of the corresponding data subsequence is calculated, and the absolute deviation of the median of the corresponding data subsequence is obtained based on the absolute deviation of each data point from the median. Determine the data point corresponding to the center position of the sliding window, and determine whether the deviation between it and the median is greater than the product of the absolute deviation of the median and a set multiple threshold. If the judgment condition is met, the corresponding data point is identified as having local abnormal fluctuations and is marked as a candidate abnormal point. The location index records of all data points marked as candidate outliers in the standardized stress data sequence are recorded and summarized to form a candidate outlier index set; In the second path, the standardized pressure data sequence of each pressure monitoring node is written into the data buffer structure in chronological order; The candidate outlier index set formed in the first path is passed to the second path. Based on the time index in the candidate outlier index set, the corresponding standardized stress data segment in the second path is extracted.

2. The intelligent diagnostic method for abnormal pressure in directly buried steam pipelines according to claim 1, characterized in that, Steam pipeline topology information and steam flow direction information include: The steam pipeline topology information includes pressure monitoring node identification information, spatial coordinate information of pressure monitoring nodes, connection relationship information between pressure monitoring nodes, and physical length information of each connecting pipe segment. The connection relationship information indicates whether there is a pipeline connection between adjacent pressure monitoring nodes, and the physical length information indicates the axial distance between the connecting pipe segments. The steam flow direction information includes steam source identification, preset steam flow direction information in each pipe section, and flow direction relationship matrix between nodes. The flow direction information represents the transmission path of steam from upstream to downstream in the pipeline network. The flow direction relationship matrix is ​​a directed relationship matrix, and each element of it represents whether the steam flow direction exists and its directionality between the corresponding node pairs.

3. The intelligent diagnostic method for abnormal pressure in directly buried steam pipelines according to claim 1, characterized in that, The generation of standardized stress data sequences includes: Time synchronization processing is performed on the raw pressure data collected from multiple pressure monitoring nodes, a unified time index set is constructed, a unified sampling period is set as the time base, the raw sampling time of each pressure monitoring node is aligned with the unified sampling period, and a preliminary synchronized pressure data sequence is generated. Identify time points in the synchronous pressure data sequence that are not mapped by the original pressure data, and generate a set of missing data points; For each missing data point in the set of missing data points, find its adjacent previous and next effective pressure data points in the synchronous pressure data sequence of the corresponding pressure monitoring node. If the time interval between the two effective pressure data points is less than the preset interpolation threshold and the pressure difference between the two effective pressure data points is less than the preset fluctuation threshold, then perform interpolation completion processing on the missing data point according to the trend between the adjacent effective pressure data points, and write the completion result into the synchronous pressure data sequence. A consistency check is performed on the interpolated and completed synchronous pressure data sequence, and local abrupt data points caused by interpolation are deleted to obtain the verified pressure data sequence. The verified pressure data sequence is reordered according to a unified time index set and combined into a unified format with each pressure monitoring node as the dimension, forming a standardized pressure data sequence.

4. The intelligent diagnostic method for abnormal pressure in directly buried steam pipelines according to claim 1, characterized in that, The generation of the posterior probability sequence at the change point includes: Obtain the time index corresponding to each candidate anomaly in the candidate anomaly index set. With the corresponding time index as the center, extract the standardized pressure data symmetrical before and after the corresponding time index in the standardized pressure data sequence corresponding to the pressure monitoring node to which the corresponding candidate anomaly belongs, according to the preset fixed window length, and form a standardized pressure data segment. The standardized pressure data segment is passed as input to the Bayesian online change point detection method. The Bayesian online change point detection method outputs a change point posterior probability sequence corresponding to each time point based on the pressure value at each time point in the standardized pressure data segment. The posterior probability sequence of the change point is analyzed to identify the amplitude change between the maximum and minimum probability values ​​in the posterior probability sequence of the change point, and the amplitude change is used as the cumulative amplitude of the change point probability. The cumulative amplitude of the change point probability is compared with a preset change point probability amplitude judgment threshold. If the cumulative amplitude of the change point probability exceeds the change point probability amplitude judgment threshold, the time index corresponding to the candidate anomaly point is confirmed as a valid anomaly trigger point.

5. The intelligent diagnostic method for abnormal pressure in directly buried steam pipelines according to claim 1, characterized in that, Bayesian online change point detection methods include: Obtain the time index corresponding to each candidate anomaly in the candidate anomaly index set, and extract the standardized pressure data symmetrical before and after the corresponding time index in the standardized pressure data sequence corresponding to the pressure monitoring node associated with the corresponding candidate anomaly, with the time index as the center and according to the preset fixed time window length, to form a standardized pressure data segment. For the standardized stress data segments, a dynamic prior probability structure containing candidate outlier information is constructed. Each time index in the standardized stress data segment is traversed sequentially to determine whether each time index belongs to the candidate outlier index set. If it does, the change point prior probability of the time index is set to the enhanced value of the basic prior probability. If it does not, the change point prior probability of the corresponding time index is kept at the default value of the basic prior probability. The standardized stress data segment containing the dynamic prior probability structure is input into the Bayesian online change point detection method, and a recursive joint posterior inference operation is performed to output the change point posterior probability sequence that corresponds one-to-one with each time index in the standardized stress data segment. The output change point posterior probability sequence is enhanced by identifying each time index of the candidate outlier index set in the change point posterior probability sequence. A local symmetric response window is constructed with the identified time index as the center and a preset length. Within the response window, confidence enhancement operations are performed on the change point posterior probability values ​​at each time index, including weighted smoothing and trend interpolation, to generate the enhanced change point posterior probability sequence.

6. The intelligent diagnostic method for abnormal pressure in directly buried steam pipelines according to claim 1, characterized in that, The generation of the set of abnormal event points includes: Obtain the enhanced variable point posterior probability sequence corresponding to each pressure monitoring node, and perform time-by-time index scanning on the variable point posterior probability sequence based on the unified time index set; In the posterior probability sequence of each pressure monitoring node, the time period in which the posterior probability value of the change point continuously increases with the time index is identified and defined as the continuous probability increasing segment. The continuous probability increasing segment must meet the following conditions: the posterior probability value of the change point corresponding to adjacent time indices shows a monotonically increasing or fluctuating increase within a limited range, and the difference between the posterior probability of the change point at the beginning and end of the corresponding time period is not less than the set minimum increase threshold, and the time span is not less than the set minimum duration threshold. Extract all time indices that meet the conditions and summarize them to form a set of suspected abnormal time indices; Obtain the candidate anomaly point index set, perform intersection processing based on the unified time index set on the suspected anomaly time index set and the candidate anomaly point index set, and extract each time index in the intersection; Each time index in the intersection is bound to the identifier of its respective pressure monitoring node to form a pressure monitoring node-time index tuple, and all tuples are aggregated to generate a set of abnormal event points.

7. The intelligent diagnostic method for abnormal pressure in directly buried steam pipelines according to claim 1, characterized in that, The construction of the anomaly propagation path includes: Obtain the pressure monitoring node-time index tuple in the abnormal event point set, and extract the corresponding change point posterior probability value in the enhanced change point posterior probability sequence based on the time index to form an abnormal feature set containing abnormal event points and change point posterior probabilities. Based on the topology information of the steam pipeline, the set of direct adjacent pressure monitoring nodes for each abnormal event node is determined, and for each pair of adjacent pressure monitoring nodes with a connection relationship, the enhanced posterior probability value of the change point at the same time index is extracted under a unified time index set. Based on the steam flow direction information, the set of downstream adjacent pressure monitoring nodes corresponding to each abnormal event point is obtained. In the unified time index set, the difference calculation is performed on the change point posterior probability value between the abnormal event point and its downstream adjacent nodes to obtain the change point posterior probability difference of each node at the corresponding time index. Node pairs with a difference in posterior probability greater than a preset threshold are selected, and the direction of the difference is confirmed to be consistent with the direction of steam flow based on the steam flow direction information. If they are consistent, the corresponding node pairs are defined as abnormal propagation edges. All abnormal propagation edges are concatenated according to their time index, and combined with the spatial coordinate information of each pressure monitoring node in the steam pipeline topology information, an abnormal propagation path composed of multiple abnormal event points in the propagation order is constructed, and all paths are summarized to form an abnormal propagation path set.

8. The intelligent diagnostic method for abnormal pressure in directly buried steam pipelines according to claim 1, characterized in that, The output of abnormal diagnostic results includes: Obtain each abnormal propagation path in the abnormal propagation path set, count the frequency of abnormal triggering of the pressure monitoring node corresponding to the starting point of the path under the unified time index set, and record the time index of the earliest time that the node was identified as an abnormal event point, forming a candidate set of starting nodes and its corresponding time index set. In the candidate set of starting nodes, the path starting points in the abnormal propagation path are selected such that the direction of the difference in the posterior probability of the change point is always consistent with the direction of steam flow, and the posterior probability of the change point of each pressure monitoring node on the path shows a monotonically decreasing trend, thus forming a set of suspected abnormal source nodes. For each node in the set of suspected anomaly source nodes, sort them according to the first appearance time index in the set of anomaly event points, and extract the suspected anomaly source node with the earliest time index as the anomaly source pressure monitoring node. The abnormal source pressure monitoring node and its corresponding first occurrence time index are defined as the abnormal source pressure monitoring node and the first abnormal occurrence time, respectively, and output together with the corresponding abnormal propagation path to form the abnormal diagnosis result.

Images