Heat supply data processing method and device based on artificial intelligence and storage medium

Abstract

The invention relates to the technical field of heat supply data processing, in particular to a heat supply data processing method and device based on artificial intelligence and a storage medium, and the method comprises the steps: collecting the environment temperature in a preset period, so as to judge the temperature change type; analyzing material aging parameters; heat supply data are collected in real time, and abnormal values are removed; constructing a pipe network coupling feature graph, and marking vertex attributes of vertexes of each graph according to vertex feature weights; extracting a residual error sequence of the heat supply data and the physical data of the map vertexes, and analyzing probability density characteristics of the residual error sequence and the material aging parameters to determine a coupling abnormal state; training an LSTM model based on the historical data to predict the vertex feature weight in the future preset period duration, and judging the pipeline stress type according to the variation amplitude of the predicted vertex feature weight; and updating the judgment method of the coupling abnormality type according to the pipeline stress type and the temperature change type. According to the invention, abnormal analysis processing of the heat supply data is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of heating data processing, and in particular to a heating data processing method, device and storage medium based on artificial intelligence. Background Art

[0002] With the rapid development of urban centralized heating and the increasing demand for energy conservation and emission reduction, heating data analysis and processing technology has emerged. This technology effectively monitors abnormal conditions in the heating system through real-time monitoring and analysis of a large amount of heating data, providing strong guarantees for the safe and efficient operation of the heating system.

[0003] China Patent Publication No.: CN117851892A discloses a data processing method, device and system for a smart heating system, including the confidence coefficient of attention processing for each heating fault characterization vector, which characterizes the contribution level of the smart heating sensor data group to the global fault identification view of the target heating IOT sensor data set. Therefore, each heating fault attention vector obtained by attention processing can comprehensively and richly reflect the detailed semantic features of the target heating IOT sensor data set, so that the global fault identification view obtained based on the fault identification of each heating fault attention vector is as accurate and reasonable as possible. The invention realizes fault identification by disassembling the data group, mining fault characterization vectors and attention mechanism, but does not realize the comprehensive analysis of heating data for the environment with sudden temperature changes and the introduction of material aging parameters and environmental temperature mutation rate. There are problems of low efficiency in the analysis and processing of heating data and inaccurate monitoring of abnormal heating data. Summary of the invention

[0004] The object of the present invention is to provide a heating data processing method based on artificial intelligence to solve at least one of the problems existing in the prior art.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] In one aspect, the present invention provides a heating data processing method based on artificial intelligence, comprising:

[0007] Collect the ambient temperature within a preset period to determine the type of temperature change;

[0008] Analyze material aging parameters based on physical data of pipe materials used in heating systems;

[0009] Collect heating data in real time and remove abnormal values;

[0010] Construct a pipe network coupling feature graph, and assign weights to each vertex in the pipe network coupling feature graph to obtain vertex feature weights, and mark the vertex attributes of each vertex according to the vertex feature weights;

[0011] The residual sequences of heating data and physical data at the vertices of the graph are extracted, and the probability density characteristics of the residual sequences and material aging parameters are analyzed to obtain the material-environment coupling parameters. The material-environment coupling parameters are compared and analyzed to determine the abnormal coupling state.

[0012] Furthermore, the LSTM model is trained based on historical data to predict the vertex feature weights within a preset period in the future, and the pipeline stress type is determined according to the change range of the predicted vertex feature weights;

[0013] Update the judgment method of coupling anomaly type according to pipeline stress type and temperature change type.

[0014] Furthermore, the ratio of the ambient temperature change at the start time of the preset period to the preset period duration is taken as the temperature change rate, and the temperature change rate is compared and analyzed to determine the type of temperature change. When the temperature change rate is greater than or equal to the preset temperature change threshold, the temperature change type is determined to be an extreme mutation; when the temperature change rate is less than the preset temperature change threshold, the temperature change type is determined to be a normal temperature change.

[0015] Furthermore, the metal fatigue coefficient is calculated based on the physical data and the acquisition times;

[0016] The changes in pipe wall thickness in each year are counted respectively, and the average value of the changes in pipe wall thickness in each year is calculated as the corrosion rate.

[0017] Further, the interquartile range of the heating data is calculated, and the calculation method of the interquartile range is to arrange the heating data collected within the preset period in order from small to large according to the numerical value, extract the numerical value of the heating data corresponding to the first quarter as the first quartile, extract the numerical value of the heating data corresponding to the third quarter as the second quartile, and take the difference between the second quartile and the first quartile as the interquartile range;

[0018] Eliminate the heating data with a value less than the first quartile minus 1.5 times the interquartile range or a value greater than the second quartile plus 1.5 times the interquartile range.

[0019] Furthermore, the pipe network nodes are taken as graph vertices, the pipes connected to the pipe network nodes are taken as edges, and the pipe temperature, metal fatigue coefficient and corrosion rate are taken as vertex attributes of the graph vertices, and the pipe flow rate and temperature change rate are taken as edge attributes;

[0020] Weights are assigned to metal fatigue coefficient, corrosion rate and temperature change rate to obtain vertex feature weights;

[0021] The vertex feature weights are compared and analyzed to determine the vertex attributes of the graph vertices. When the vertex feature weight is less than the preset weight threshold, the vertex attribute of the graph vertex is marked as a normal node. When the vertex feature weight is greater than or equal to the preset weight threshold, the vertex attribute of the graph vertex is marked as a fragile node.

[0022] Furthermore, the stress amplitude and the pipeline temperature of the graph vertices whose vertex attributes are fragile nodes are respectively subjected to regression analysis to obtain the regression equation of the stress amplitude versus the acquisition number and the regression equation of the pipeline temperature versus the acquisition number, and the difference between the actual value and the estimated value of the stress amplitude is calculated as the stress residual, and the difference between the actual value and the estimated value of the pipeline temperature is calculated as the temperature residual, and the stress parameters and the temperature residuals are matched one by one with the acquisition number to obtain the residual sequence;

[0023] The covariance matrix of the residual sequence and the metal fatigue coefficient is used as the material environment coupling parameter;

[0024] The material environment coupling parameters are compared and analyzed to determine the coupling abnormal state. If Cov(R=r1,F)<M, the coupling abnormal state is not analyzed. If Cov(R=r1,F)≥M, the coupling abnormal state is determined to be material stress coupling abnormality; if Cov(R=r2,F)<M, the coupling abnormal state is not analyzed. If Cov(R=r2,F)≥M, the coupling abnormal state is determined to be material temperature coupling abnormality; wherein Cov(R,F) represents the material environment coupling parameter, and M represents the coupling determination threshold.

[0025] Furthermore, the collected ambient temperature, physical data and heating data are used as historical data and input into the LSTM model to predict the vertex feature weights within the future preset cycle length, and the ratio of the average value of the vertex feature weights within the future preset cycle length to the vertex feature weights at the current time is calculated as the predicted increase value;

[0026] The predicted increase value is compared and analyzed. When the predicted increase value is less than the preset increase threshold, if the pressure residual is greater than or equal to the positive first pressure residual threshold, the pipeline stress type is determined to be Class I; if the pressure residual is less than or equal to the negative first pressure residual threshold, the pipeline stress type is determined to be Class II; if the pressure residual is greater than the negative first pressure residual threshold and less than the positive first pressure residual threshold, the pipeline stress type is determined to be Class III; when the predicted increase value is greater than or equal to the preset increase threshold, if the pressure residual is greater than or equal to the positive second pressure residual threshold, the pipeline stress type is determined to be Class I; if the pressure residual is less than or equal to the negative second pressure residual threshold, the pipeline stress type is determined to be Class II; if the pressure residual is greater than the negative second pressure residual threshold and less than the positive second pressure residual threshold, the pipeline stress type is determined to be Class III;

[0027] When the coupling anomaly type is a stress coupling anomaly, if the pipeline stress type is Class I and the metal fatigue coefficient is greater than or equal to the preset fatigue threshold, the coupling anomaly type is updated to a high stress anomaly, and a pipeline stress rupture warning is issued; if the pipeline stress type is Class II and Cov(R=r1,F)≥M1, the coupling anomaly type is updated to a low stress anomaly, and a pipeline leakage warning is issued; when the coupling anomaly type is a temperature coupling anomaly, if the temperature change type is extremely sudden, the coupling anomaly type is updated to a thermal efficiency anomaly, and a heating efficiency attenuation warning is issued, where M1 represents the dynamic coupling judgment threshold, which is equal to the ratio of the coupling judgment threshold to the dynamic adjustment parameter.

[0028] On the other hand, the present invention also provides a heating data processing device based on artificial intelligence, comprising:

[0029] The acquisition module is used to collect the ambient temperature within a preset period and collect heating data in real time;

[0030] A filtering module is used to remove abnormal values ​​in the heating data;

[0031] An analysis module for analyzing material aging parameters based on physical data of pipe materials used in the heating system;

[0032] A construction module is used to construct a pipe network coupling feature graph, and to assign weights to each vertex in the pipe network coupling feature graph to obtain vertex feature weights, and to mark vertex attributes of each vertex according to the vertex feature weights;

[0033] An analysis module is used to extract the residual sequence of heating data and physical data at the vertex of the graph, and analyze the probability density characteristics of the residual sequence and the material aging parameters to obtain the material environment coupling parameters, and compare and analyze the material environment coupling parameters to determine the abnormal coupling state;

[0034] The update module is used to train the LSTM model based on historical data to predict the vertex feature weights within a preset period in the future, and to determine the pipeline stress type based on the change amplitude of the predicted vertex feature weights; it is also used to update the determination method of the coupling anomaly type based on the pipeline stress type and temperature change type.

[0035] On the other hand, the present invention further provides a storage medium storing instructions, which, when executed on a computer, enables the computer to execute any of the above methods.

[0036] The beneficial effects of the present invention are as follows: by introducing material aging parameters and ambient temperature mutation rate as specific analysis dimensions and combining multimodal data fusion technology, the efficiency of heating data analysis and the accuracy of abnormal monitoring are significantly improved, providing strong technical support for the stable operation of the heating system. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0038] Figure 1 This is a flow chart of the heating data processing method based on artificial intelligence in this embodiment.

[0039] Figure 2 Flow chart of the material aging parameter analysis method of this embodiment.

[0040] Figure 3 Flow chart of the method for analyzing vertex feature weights of this embodiment.

[0041] Figure 4 Flow chart of the method for determining a coupling abnormal state in this embodiment.

[0042] Figure 5 This is a structural block diagram of the heating data processing device based on artificial intelligence in this embodiment. DETAILED DESCRIPTION

[0043] In order to more clearly illustrate the present invention, the present invention is further described below in conjunction with preferred embodiments and accompanying drawings. Similar components in the accompanying drawings are represented by the same reference numerals. It should be understood by those skilled in the art that the content specifically described below is illustrative rather than restrictive, and should not be used to limit the scope of protection of the present invention.

[0044] It should be noted that, although the terms first, second, third, etc. may be used to describe in the embodiments of the present application, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, without departing from the scope of the embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first.

[0045] See also Figure 1 As shown, it is a heating data processing method based on artificial intelligence in this embodiment, including:

[0046] Step S1, collecting the ambient temperature within a preset period to determine the type of temperature change, wherein the ambient temperature is collected by accessing a local meteorological website.

[0047] Specifically, the preset period in this embodiment is set to 1 hour. In this embodiment, the duration of the preset period is not specifically limited, and those skilled in the art can freely set it, such as 0.5 hours, 1.5 hours, 2 hours, etc.

[0048] Specifically, in step S1 described in this embodiment, the ratio of the ambient temperature change at the start time of the preset period to the preset period duration is taken as the temperature change rate, and the temperature change rate is compared and analyzed to determine the type of temperature change. When the temperature change rate is greater than or equal to the preset temperature change threshold, the temperature change type is determined to be an extreme mutation. When the temperature change rate is less than the preset temperature change threshold, the temperature change type is determined to be a normal temperature change. In this embodiment, the preset temperature change threshold is set to 5°C / h. In this embodiment, the value of the preset temperature change threshold is not specifically limited. It only needs to meet the judgment of the temperature change type. The value of the temperature change threshold should meet the requirement of being greater than or equal to 4°C / h and less than or equal to 8°C / h.

[0049] Specifically, in step S1 described in this embodiment, by collecting the ambient temperature in real time and calculating the temperature change rate, extreme temperature mutations can be quickly identified to avoid abnormalities in the heating system caused by drastic fluctuations in ambient temperature, thereby improving the response speed to temperature changes and ensuring stable operation under extreme weather conditions, thereby reducing the risk of decreased heating efficiency or equipment damage due to temperature mutations.

[0050] Specifically, this embodiment is aimed at the abnormal monitoring of old heating pipe network systems under extreme temperature change conditions. The mechanical performance degradation caused by the aging of the pipe network materials and the drastic fluctuations in the ambient temperature form a coupling effect, which can easily cause stress imbalance in the pipe network, local leakage or a sudden drop in thermal efficiency.

[0051] Please continue reading Figure 1 As shown, the heating data processing method based on artificial intelligence also includes:

[0052] Step S2, analyzing material aging parameters based on physical data of the pipe material used in the heating system, wherein the material aging parameters include metal fatigue coefficient and corrosion rate.

[0053] Specifically, in step S2 described in this embodiment, by analyzing the metal fatigue coefficient and corrosion rate, the aging degree of the pipeline material can be accurately assessed, and the potential risk of pipeline rupture or leakage can be warned in advance, so as to improve the monitoring accuracy of material aging problems, reduce sudden failures caused by material aging, extend the service life of the pipeline, and reduce maintenance costs.

[0054] See also Figure 2 As shown in FIG. 1 , it is a material aging parameter analysis method, including:

[0055] Step S201, collecting physical data of the pipe material used in the heating system, the physical data including pipe wall thickness, stress amplitude, material ultimate stress and material fatigue index, the material fatigue index is a constant describing the sensitivity of the material to cyclic load, in this embodiment, the material fatigue index is set to 3, the stress amplitude is the stress inside the pipe, which is acquired through sensor acquisition, the material ultimate stress and material fatigue index are acquired through user interactive input, the pipe wall thickness is acquired through sensor acquisition, and the acquisition frequency of the physical data is 1 time / minute.

[0056] Please continue reading Figure 2 As shown, it is a material aging parameter analysis method, which also includes:

[0057] Step S202, analyzing and calculating the metal fatigue coefficient based on the physical data and the number of acquisitions.

[0058] Specifically, in step S202 of this embodiment, the metal fatigue coefficient is calculated based on the physical data and the number of acquisitions. The expression of the metal fatigue coefficient is: Among them, F represents the metal fatigue coefficient, D i represents the stress amplitude, i represents the acquisition number, D max represents the ultimate stress of the material, and m represents the fatigue index of the material.

[0059] Please continue reading Figure 2 As shown, it is a material aging parameter analysis method, which also includes:

[0060] Step S203, counting the change in the pipe wall thickness in each year, and calculating the average value of the change in the pipe wall thickness in each year as the corrosion rate.

[0061] Please continue reading Figure 1 As shown, the heating data processing method based on artificial intelligence also includes:

[0062] Step S3, collecting heating data in real time and removing abnormal values, the heating data including pipeline temperature and pipeline flow.

[0063] Specifically, the sampling frequency of the heating data in this embodiment is 1 time / minute.

[0064] Specifically, in step S3 described in this embodiment, the interquartile range of the heating data is calculated. The method for calculating the interquartile range is to arrange the collected heating data within a preset period in order from small to large, extract the value of the heating data corresponding to the first quarter as the first quartile, extract the value of the heating data corresponding to the third quarter as the second quartile, and take the difference between the second quartile and the first quartile as the interquartile range.

[0065] Specifically, in step S3 described in this embodiment, heating data with a value less than the first quartile minus 1.5 times the interquartile range or a value greater than the second quartile plus 1.5 times the interquartile range is eliminated.

[0066] Specifically, in step S3 described in this embodiment, outliers are eliminated by the interquartile range method to ensure the accuracy and reliability of the heating data, avoid the interference of abnormal data on subsequent analysis, improve the efficiency of data processing, reduce misjudgments caused by data noise, and provide a high-quality data basis for subsequent anomaly detection.

[0067] Please continue reading Figure 1 As shown, the heating data processing method based on artificial intelligence also includes:

[0068] Step S4, constructing a pipe network coupling feature graph, and assigning weights to each vertex in the pipe network coupling feature graph to obtain vertex feature weights, and marking vertex attributes of each vertex according to the vertex feature weights.

[0069] Specifically, in step S4 described in this embodiment, by constructing a pipeline network coupling characteristic graph and assigning vertex weights, the vulnerability of each node in the pipeline network can be intuitively displayed, and high-risk nodes can be identified to improve the efficiency of analyzing the pipeline network structure, help operation and maintenance personnel quickly locate potential problem areas, optimize resource allocation, and reduce the occurrence of failures.

[0070] See also Figure 3 As shown, it is an analysis method of vertex feature weights, including:

[0071] Step S401, constructing a pipe network coupling characteristic diagram in combination with pipe network nodes, heating data and material aging parameters of the heating system, wherein the pipe network node is a node in the pipeline of the heating system that connects at least two pipelines.

[0072] Specifically, in step S401 described in this embodiment, the pipeline network nodes are used as graph vertices, the pipes connected to the pipeline network nodes are used as edges, the pipeline temperature, metal fatigue coefficient and corrosion rate are used as vertex attributes of the graph vertices, and the pipeline flow and temperature change rate are used as edge attributes.

[0073] Please continue reading Figure 3 As shown, it is a method for analyzing vertex feature weights, and also includes:

[0074] Step S402: weighting each vertex in the pipe network coupling feature graph to obtain vertex feature weights.

[0075] Specifically, in step S402 of this embodiment, weights are assigned to the metal fatigue coefficient, corrosion rate, and temperature change rate to obtain vertex feature weights. The expression of the vertex feature weights is: i=α×F+β×C+γ×|ΔT / Δt|, where W i represents the vertex feature weight, α represents the aging weight, β represents the corrosion weight, γ represents the mutation weight, α+β+γ=1, C represents the corrosion rate, and |ΔT / Δt| represents the temperature change rate. It can be understood that the values ​​of the aging weight, corrosion weight and mutation weight are not specifically limited in this embodiment, and those skilled in the art can freely set them as long as they meet the analysis of the vertex feature weight. The settings of the aging weight, corrosion weight and mutation weight should satisfy α>γ and β>γ. In this embodiment, the aging weight is set to α=0.6, the corrosion weight is set to β=0.3, and the mutation weight is set to γ=0.1.

[0076] Please continue reading Figure 3 As shown, it is a method for analyzing vertex feature weights, and also includes:

[0077] Step S403: marking the vertex attributes of each graph vertex according to the vertex feature weight.

[0078] Specifically, in step S403 described in this embodiment, the vertex feature weights are compared and analyzed to determine the vertex attributes of the graph vertices. When the vertex feature weight is less than the preset weight threshold, the vertex attributes of the graph vertices are marked as normal nodes. When the vertex feature weight is greater than or equal to the preset weight threshold, the vertex attributes of the graph vertices are marked as fragile nodes. In this embodiment, the preset weight threshold is set to 0.8. It can be understood that in this embodiment, the value of the preset weight threshold is not specifically limited. Those skilled in the art can set it freely, as long as the marking of the vertex attributes is satisfied. The setting of the preset weight threshold should satisfy the requirement of being greater than or equal to 0.6 and less than or equal to 0.9.

[0079] Please continue reading Figure 1 As shown, the heating data processing method based on artificial intelligence also includes:

[0080] Step S5, extracting the residual sequence of the heating data and physical data of the graph vertex, and analyzing the probability density characteristics of the residual sequence and the material aging parameters to obtain the material environment coupling parameters, and performing a comparison analysis on the material environment coupling parameters to determine the coupling abnormality state.

[0081] Specifically, in step S5 described in this embodiment, by analyzing the covariance matrix of the residual sequence and the material aging parameters, the abnormal coupling state of the material and the environmental factors can be accurately judged to improve the detection accuracy of the coupling anomaly, reduce the failures caused by the combined effects of material aging and environmental mutations, and enhance stability and reliability.

[0082] See also Figure 4 As shown, it is a method for determining the abnormal coupling state, including:

[0083] Step S501 , performing regression analysis on the stress amplitude and pipeline temperature of the graph vertices whose vertex attributes are fragile nodes, respectively, to extract the residual sequence of the heating data of the graph vertices.

[0084] Specifically, in step S501 described in this embodiment, regression analysis is performed on the stress amplitude and pipeline temperature of the graph vertices whose vertex attributes are fragile nodes, respectively, to obtain a regression equation of the stress amplitude versus the acquisition number and a regression equation of the pipeline temperature versus the acquisition number, and the difference between the actual value and the estimated value of the stress amplitude is calculated as the stress residual, and the difference between the actual value and the estimated value of the pipeline temperature is calculated as the temperature residual, and the stress parameters and the temperature residuals are matched one by one with the acquisition number to obtain a residual sequence.

[0085] Please continue reading Figure 4 As shown, it is a method for judging the abnormal coupling state, and also includes:

[0086] Step S502: Perform probability density characteristic analysis on the residual sequence and the metal fatigue coefficient to obtain material environment coupling parameters.

[0087] Specifically, in step S502 of this embodiment, the covariance matrix of the residual sequence and the metal fatigue coefficient is used as the material environment coupling parameter, and the expression of the material environment coupling parameter is: Among them, Cov(R,F) represents the material environment coupling parameter, R represents the residual data, R∈UR, UR={r1,r2}, UR represents the set of residual sequences, r1 represents the residual sequence of stress residuals, r2 represents the residual sequence of temperature residuals, R i represents the data in the residual sequence corresponding to each acquisition number, Avg() represents the average value of the data in brackets, F i Indicates the metal fatigue coefficient corresponding to each acquisition number.

[0088] Please continue reading Figure 4 As shown, it is a method for judging the abnormal coupling state, and also includes:

[0089] Step S503: Compare and analyze the material environment coupling parameters to determine the abnormal coupling state.

[0090] Specifically, in step S503 described in this embodiment, the material environment coupling parameters are compared and analyzed to determine the coupling abnormal state. If Cov(R=r1,F)<M, the coupling abnormal state is not analyzed. If Cov(R=r1,F)≥M, the coupling abnormal state is determined to be material stress coupling abnormality; if Cov(R=r2,F)<M, the coupling abnormal state is not analyzed. If Cov(R=r2,F)≥M, the coupling abnormal state is determined to be material temperature coupling abnormality; wherein M represents the coupling determination threshold, 0.4≤M≤0.6. It can be understood that the value of the coupling determination threshold is not specifically limited in this embodiment, and those skilled in the art can freely set it as long as it satisfies the analysis of the coupling abnormal state. The optimal value of the coupling determination threshold is: M=0.5.

[0091] Please continue reading Figure 1 As shown, the heating data processing method based on artificial intelligence also includes:

[0092] Step S6, training the LSTM model based on historical data to predict the vertex feature weights within a future preset period, and determining the pipeline stress type according to the change range of the predicted vertex feature weights.

[0093] Specifically, in step S6 described in this embodiment, the collected ambient temperature, physical data and heating data are used as historical data and input into the LSTM model to predict the vertex feature weights within a future preset cycle length, and the ratio of the predicted average value of the vertex feature weights within the future preset cycle length to the vertex feature weights at the current time is calculated as the predicted increase value.

[0094] Specifically, in step S6 described in this embodiment, the standard deviation of three times the stress amplitude is used as the first pressure residual threshold, and the product of the first residual threshold and the dynamic adjustment parameter is used as the second pressure residual threshold. In this embodiment, the dynamic adjustment parameter is set to 0.9, and those skilled in the art can freely set it as long as the analysis of the second pressure residual threshold is satisfied. The value of the dynamic adjustment parameter should be greater than or equal to 0.9 and less than 1.

[0095] Specifically, in step S6 described in this embodiment, the predicted increase value is compared and analyzed. When the predicted increase value is less than the preset increase threshold, if the pressure residual is greater than or equal to the positive first pressure residual threshold, the pipeline stress type is determined to be one type, if the pressure residual is less than or equal to the negative first pressure residual threshold, the pipeline stress type is determined to be two types, if the pressure residual is greater than the negative first pressure residual threshold and less than the positive first pressure residual threshold, the pipeline stress type is determined to be three types; when the predicted increase value is greater than or equal to the preset increase threshold, if the pressure residual is greater than or equal to the positive second pressure residual threshold, the pipeline stress type is determined to be one type, if the pressure residual is less than or equal to the negative second pressure residual threshold, the pipeline stress type is determined to be two types, if the pressure residual is greater than the negative second pressure residual threshold and less than the positive second pressure residual threshold, the pipeline stress type is determined to be three types. In this embodiment, the preset increase threshold is set to 1.2, and those skilled in the art can freely set it, as long as the analysis of the pipeline pressure type is satisfied, and the value of the preset increase threshold should be greater than or equal to 1.1 and less than or equal to 1.3.

[0096] Specifically, in step S6 of this embodiment, the change of vertex feature weights is predicted by the LSTM model, which can give early warning of changes in pipeline stress types and avoid pipeline rupture or leakage caused by excessive or insufficient stress, so as to improve the prediction ability of pipeline stress, reduce the occurrence of sudden failures, and optimize operating efficiency.

[0097] Please continue reading Figure 1 As shown, the heating data processing method based on artificial intelligence also includes:

[0098] Step S7, updating the method for determining the coupling abnormality type according to the pipeline stress type and the temperature change type.

[0099] Specifically, in step S7 described in this embodiment, when the coupling abnormality type is a stress coupling abnormality, if the pipeline stress type is Class I and the metal fatigue coefficient is greater than or equal to the preset fatigue threshold, the coupling abnormality type is updated to a high stress abnormality, and a pipeline stress rupture warning is issued; if the pipeline stress type is Class II and Cov(R=r1,F)≥M1, the coupling abnormality type is updated to a low stress abnormality, and a pipeline leakage warning is issued; when the coupling abnormality type is a temperature coupling abnormality, if the temperature change type is extremely sudden, the coupling abnormality type is updated to a thermal efficiency abnormality, and a heating efficiency attenuation warning is issued, wherein M1 represents a dynamic coupling determination threshold, which is equal to the ratio of the coupling determination threshold to the dynamic adjustment parameter.

[0100] Specifically, in step S7 described in this embodiment, by combining the pipeline stress type and the temperature change type, the method for determining the coupled abnormality type is dynamically updated, so that different types of abnormal situations can be more accurately identified to improve the flexibility of abnormality detection and ensure that the early warning strategy can be adjusted according to actual conditions to reduce false alarms and missed alarms.

[0101] See also Figure 5 As shown, this embodiment also provides a heating data processing device based on artificial intelligence, and the device includes:

[0102] The acquisition module 801 is used to acquire the ambient temperature within a preset period and to acquire heating data in real time.

[0103] The filtering module 802 is used to remove abnormal values ​​in the heating data, and the filtering module 802 is connected to the acquisition module 801 .

[0104] The analysis module 803 is used to analyze material aging parameters according to the physical data of the pipe material used in the heating system. The analysis module 803 is connected to the filtering module 802 .

[0105] The construction module 804 is used to construct a pipe network coupling feature graph, and to assign weights to each vertex in the pipe network coupling feature graph to obtain vertex feature weights, and to mark vertex attributes of each vertex according to the vertex feature weights. The construction module 804 is connected to the analysis module 803.

[0106] The analysis module 805 is used to extract the residual sequence of the heating data and physical data of the graph vertices, and analyze the probability density characteristics of the residual sequence and the material aging parameters to obtain the material environment coupling parameters, and compare and analyze the material environment coupling parameters to determine the abnormal coupling state. The analysis module 805 is connected to the construction module 804.

[0107] The update module 806 is used to train the LSTM model based on historical data to predict the vertex feature weights within a future preset period, and to determine the pipeline stress type based on the change amplitude of the predicted vertex feature weights; it is also used to update the determination method of the coupling anomaly type based on the pipeline stress type and the temperature change type. The update module 806 is connected to the analysis module 805.

[0108] The embodiment of the present application also provides a computer-readable storage medium storing instructions, which, when executed on a computer, enables the computer to execute the method described in the above method embodiment.

[0109] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not limitations on the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the embodiments here. All obvious changes or modifications derived from the technical solution of the present invention are still within the protection scope of the present invention.

Claims

1. A heating data processing method based on artificial intelligence, characterized in that: include: Collect the ambient temperature within a preset period to determine the type of temperature change; Analyze material aging parameters based on physical data of pipe materials used in heating systems; Collect heating data in real time and remove abnormal values; Construct a pipe network coupling feature graph, and assign weights to each vertex in the pipe network coupling feature graph to obtain vertex feature weights, and mark the vertex attributes of each vertex according to the vertex feature weights; The residual sequences of heating data and physical data at the vertices of the graph are extracted, and the probability density characteristics of the residual sequences and material aging parameters are analyzed to obtain the material-environment coupling parameters. The material-environment coupling parameters are compared and analyzed to determine the abnormal coupling state.

2. The heating data processing method based on artificial intelligence according to claim 1 is characterized in that: The method further comprises: The LSTM model is trained based on historical data to predict the vertex feature weights within a preset period in the future, and the pipeline stress type is determined based on the change range of the predicted vertex feature weights; Update the judgment method of coupling anomaly type according to pipeline stress type and temperature change type.

3. The heating data processing method based on artificial intelligence according to claim 2 is characterized in that: The ratio of the ambient temperature change at the start time of the preset cycle to the preset cycle duration is taken as the temperature change rate, and the temperature change rate is compared and analyzed to determine the type of temperature change. When the temperature change rate is greater than or equal to the preset temperature change threshold, the temperature change type is determined to be an extreme mutation. When the temperature change rate is less than the preset temperature change threshold, the temperature change type is determined to be a normal temperature change.

4. The heating data processing method based on artificial intelligence according to claim 3 is characterized in that: Analyze and calculate metal fatigue coefficient based on physical data and acquisition times; The changes in the pipe wall thickness in each year are counted respectively, and the average value of the changes in the pipe wall thickness in each year is calculated as the corrosion rate.

5. The heating data processing method based on artificial intelligence according to claim 4 is characterized in that: Calculate the interquartile range of the heating data, wherein the calculation method of the interquartile range is to arrange the heating data collected within a preset period in order from small to large according to the numerical value, extract the numerical value of the heating data corresponding to the first quarter as the first quartile, extract the numerical value of the heating data corresponding to the third quarter as the second quartile, and take the difference between the second quartile and the first quartile as the interquartile range; Eliminate the heating data with a value less than the first quartile minus 1.5 times the interquartile range or a value greater than the second quartile plus 1.5 times the interquartile range.

6. The heating data processing method based on artificial intelligence according to claim 5 is characterized in that: The network nodes are taken as graph vertices, the pipes connected to the network nodes are taken as edges, the pipe temperature, metal fatigue coefficient and corrosion rate are taken as vertex attributes of the graph vertices, and the pipe flow and temperature change rate are taken as edge attributes; Weights are assigned to metal fatigue coefficient, corrosion rate and temperature change rate to obtain vertex feature weights; The vertex feature weights are compared and analyzed to determine the vertex attributes of the graph vertices. When the vertex feature weight is less than the preset weight threshold, the vertex attribute of the graph vertex is marked as a normal node. When the vertex feature weight is greater than or equal to the preset weight threshold, the vertex attribute of the graph vertex is marked as a fragile node.

7. The heating data processing method based on artificial intelligence according to claim 6 is characterized in that: Regression analysis is performed on the stress amplitude and pipeline temperature of the graph vertices whose vertex attributes are fragile nodes, respectively, to obtain the regression equation of stress amplitude versus acquisition number and the regression equation of pipeline temperature versus acquisition number, and the difference between the actual value and the estimated value of stress amplitude is calculated as the stress residual, and the difference between the actual value and the estimated value of pipeline temperature is calculated as the temperature residual, and the stress parameters and temperature residuals are matched one by one with the acquisition number to obtain the residual sequence; The covariance matrix of the residual sequence and the metal fatigue coefficient is used as the material environment coupling parameter; The material environment coupling parameters are compared and analyzed to determine the coupling abnormal state. If Cov(R=r1,F)<M, the coupling abnormal state is not analyzed. If Cov(R=r1,F)≥M, the coupling abnormal state is determined to be material stress coupling abnormality; if Cov(R=r2,F)<M, the coupling abnormal state is not analyzed. If Cov(R=r2,F)≥M, the coupling abnormal state is determined to be material temperature coupling abnormality; wherein Cov(R,F) represents the material environment coupling parameter, and M represents the coupling determination threshold.

8. The heating data processing method based on artificial intelligence according to claim 2 is characterized in that: The collected ambient temperature, physical data and heating data are used as historical data and input into the LSTM model to predict the vertex feature weights within the future preset cycle length, and the ratio of the average value of the vertex feature weights within the future preset cycle length to the vertex feature weights at the current time is calculated as the predicted increase value; The predicted increase value is compared and analyzed. When the predicted increase value is less than the preset increase threshold, if the pressure residual is greater than or equal to the positive first pressure residual threshold, the pipeline stress type is determined to be Class I; if the pressure residual is less than or equal to the negative first pressure residual threshold, the pipeline stress type is determined to be Class II; if the pressure residual is greater than the negative first pressure residual threshold and less than the positive first pressure residual threshold, the pipeline stress type is determined to be Class III; when the predicted increase value is greater than or equal to the preset increase threshold, if the pressure residual is greater than or equal to the positive second pressure residual threshold, the pipeline stress type is determined to be Class I; if the pressure residual is less than or equal to the negative second pressure residual threshold, the pipeline stress type is determined to be Class II; if the pressure residual is greater than the negative second pressure residual threshold and less than the positive second pressure residual threshold, the pipeline stress type is determined to be Class III; When the coupling anomaly type is stress coupling anomaly, if the pipeline stress type is Class I and the metal fatigue coefficient is greater than or equal to the preset fatigue threshold, the coupling anomaly type is updated to high stress anomaly, and a pipeline stress rupture warning is issued; If the pipeline stress type is Class II and Cov(R=r1,F)≥M1, the coupling anomaly type is updated to low stress anomaly, and a pipeline leakage warning is issued; when the coupling anomaly type is temperature coupling anomaly, if the temperature change type is extremely sudden, the coupling anomaly type is updated to thermal efficiency anomaly, and a heating efficiency attenuation warning is issued, where M1 represents the dynamic coupling judgment threshold, which is equal to the ratio of the coupling judgment threshold to the dynamic adjustment parameter.

9. A heating data processing device based on artificial intelligence, applied to the method according to any one of claims 1 to 8, comprising: The acquisition module is used to collect the ambient temperature within a preset period and collect heating data in real time; A filtering module is used to remove abnormal values ​​in the heating data; An analysis module for analyzing material aging parameters based on physical data of pipe materials used in the heating system; A construction module is used to construct a pipe network coupling feature graph, and to assign weights to each vertex in the pipe network coupling feature graph to obtain vertex feature weights, and to mark vertex attributes of each vertex according to the vertex feature weights; An analysis module is used to extract the residual sequence of heating data and physical data at the vertex of the graph, and analyze the probability density characteristics of the residual sequence and the material aging parameters to obtain the material environment coupling parameters, and compare and analyze the material environment coupling parameters to determine the abnormal coupling state; The update module is used to train the LSTM model based on historical data to predict the vertex feature weights within a preset period in the future, and to determine the pipeline stress type based on the change amplitude of the predicted vertex feature weights; it is also used to update the determination method of the coupling anomaly type based on the pipeline stress type and temperature change type.

10. A storage medium, characterized in that: The device stores instructions which, when executed on a computer, enable the computer to execute the method according to any one of claims 1 to 8.

Images