A method and system for monitoring natural gas pipeline pressure based on AI edge computing

Abstract

The present invention belongs to the technical field of natural gas pipeline monitoring. A method and system for monitoring natural gas pipeline pressure based on AI edge computing are provided. The method comprises: constructing an AI edge prediction model; fitting the first pressure monitoring data group of each pressure monitoring point in the natural gas monitoring area into a pressure change curve; using the first AI edge prediction model to evaluate the leakage probability of each first pressure drop curve segment in the pressure change curve, and obtaining a plurality of second pressure drop curve segments; receiving the working status data of each gas terminal associated with the corresponding pressure monitoring point, and using the second AI edge prediction model to perform a secondary evaluation on each second pressure drop curve segment based on each working status data, and obtaining the leakage probability of each second pressure drop curve segment; when there is at least one second pressure drop curve segment with a leakage probability higher than a probability threshold, it is determined that there is a leakage and an alarm is issued. The present invention can accurately identify minor natural gas leaks.

Description

Technical Field

[0001] The present invention relates to the field of natural gas pipeline monitoring technology, and in particular to a method and system for monitoring natural gas pipeline pressure based on AI edge computing. Background Art

[0002] In urban pipe networks or long-distance pipelines, terminals with pressure monitoring are usually installed at certain intervals. When the pipeline pressure rises or drops to the preset alarm threshold, the pressure monitoring terminal will upload the alarm information. At present, the practice of only alarming when the set threshold is reached has the problem of no alarm or late alarm for leaks. It is mainly reflected in the following aspects: when a natural gas pipeline is dug up and leaks, there may not be a pressure monitoring point at the leak point. At this time, the pressure at the remote pressure monitoring point will not immediately drop to the alarm threshold. Depending on the amount of leakage, it may take several minutes to dozens of minutes, or even due to the continuous gas supply at the back end, it will not drop to the alarm threshold, resulting in the problem of no alarm or late alarm for leaks.

[0003] Therefore, how to accurately monitor leakage when the pressure monitoring points are insufficiently covered and there is a slight leak in the natural gas pipeline is a technical problem that needs to be solved urgently. Summary of the invention

[0004] In order to solve at least one of the above technical problems, the present invention specifically provides a method, system, electronic device, computer storage medium and computer program product for monitoring natural gas pipeline pressure based on AI edge computing.

[0005] The present invention provides a method for monitoring natural gas pipeline pressure based on AI edge computing, comprising the following steps:

[0006] Acquire and build a first AI edge prediction model and a second AI edge prediction model based on a federated learning mechanism;

[0007] Receiving a first pressure monitoring data group of each pressure monitoring point in a natural gas monitoring area, and obtaining a pressure change curve by fitting according to the first pressure monitoring data group;

[0008] identifying a plurality of first pressure drop curve segments in the pressure change curve, evaluating a first leakage probability of each of the first pressure drop curve segments using the first AI edge prediction model, and determining the first pressure drop curve segment having a first leakage probability higher than a probability threshold as a second pressure drop curve segment;

[0009] Receiving working status data of each gas-using terminal associated with the corresponding pressure monitoring point, performing a secondary evaluation on each second pressure drop curve segment based on each working status data using the second AI edge prediction model, and obtaining a second leakage probability of each second pressure drop curve segment;

[0010] When there is at least one second pressure drop curve segment with a second leakage probability higher than a probability threshold, it is confirmed that there is leakage, and corresponding alarm information is output.

[0011] Further, the identifying a plurality of first pressure drop curve segments in the pressure change curve includes:

[0012] Each pressure drop section in the pressure variation curve is identified, and the slope of each pressure drop section is calculated, and a pressure drop section whose slope is higher than a slope threshold is determined as a first pressure drop curve section.

[0013] Furthermore, the obtaining and constructing of the first AI edge prediction model based on the federated learning mechanism includes:

[0014] Publish the model training task in the blockchain, and receive the training data set and model parameters fed back by each blockchain participating node in response to the model training task;

[0015] Filtering target model parameters from the model parameters, and using the target model parameters to update the model parameters of the first AI edge prediction model;

[0016] A target training data set is obtained by screening each of the training data sets according to the target model parameters, and the first AI edge prediction model with updated model parameters is trained using the target training data set.

[0017] Furthermore, the target model parameters are obtained by screening the model parameters, including:

[0018] The prediction record of the corresponding model is parsed from the feedback message of each blockchain participating node, and the number of accurate identification times of minor natural gas leaks by the model of the blockchain participating node is determined based on the prediction record, and the model parameters of the model with the highest accurate identification times are screened as the target model parameters.

[0019] Furthermore, the step of obtaining a target training data set by screening each of the training data sets according to the target model parameters includes:

[0020] Determining a training data set corresponding to the target model parameters, and other training data sets;

[0021] Randomly selecting a number of training data corresponding to a first ratio from the training data set corresponding to the target model parameter, and randomly selecting a number of training data corresponding to a second ratio from other training data sets; wherein the first ratio is smaller than the second ratio;

[0022] All the selected training data are combined into the target training data set.

[0023] Further, using the second AI edge prediction model to perform a secondary evaluation on each of the second pressure drop curve segments based on each of the working status data to obtain a second leakage probability of each of the second pressure drop curve segments, including:

[0024] Performing load evaluation according to each of the working status data to obtain a sufficient load value, wherein the sufficient load value represents a ratio of a real-time gas consumption rate of each gas-using terminal associated with the pressure monitoring point to a theoretical maximum gas consumption rate;

[0025] The load sufficiency value, the distance value between each gas terminal and the associated pressure monitoring point, and each second pressure drop curve segment are input into the second AI edge prediction model, and the second AI edge prediction model outputs the second leakage probability of each second pressure drop curve segment.

[0026] The present invention provides a system for monitoring natural gas pipeline pressure based on AI edge computing, including a training module, a fitting module, a first recognition module, a second recognition module, and an early warning module;

[0027] The training module is used to acquire and construct the first AI edge prediction model and the second AI edge prediction model based on the federated learning mechanism;

[0028] The fitting module is used to receive a first pressure monitoring data group of each pressure monitoring point in the natural gas monitoring area, and obtain a pressure change curve by fitting according to the first pressure monitoring data group;

[0029] The first identification module is used to identify a plurality of first pressure drop curve segments in the pressure change curve, evaluate the first leakage probability of each of the first pressure drop curve segments using the first AI edge prediction model, and determine the first pressure drop curve segment whose first leakage probability is higher than a probability threshold as a second pressure drop curve segment;

[0030] The second identification module is used to receive the working status data of each gas terminal associated with the corresponding pressure monitoring point, and use the second AI edge prediction model to perform a secondary evaluation on each second pressure drop curve segment based on each working status data to obtain a second leakage probability of each second pressure drop curve segment;

[0031] The early warning module is used to confirm that there is a leakage and output corresponding alarm information when there is at least one second pressure drop curve segment with a second leakage probability higher than a probability threshold.

[0032] The present invention provides an electronic device, comprising: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute any of the methods described in the preceding items.

[0033] The present invention provides a computer storage medium, on which a computer program is stored. When the computer program is run by a processor, any of the above methods is executed.

[0034] The present invention uses an AI edge prediction model to perform two leakage assessments on the abnormal pressure drop section in the pressure change curve of the natural gas pipeline, which can greatly improve the accuracy and timeliness of leakage identification when there is a slight leak in the natural gas pipeline and the leakage point is far away from the pressure monitoring point. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0036] Figure 1 It is a flow chart of a method for monitoring natural gas pipeline pressure based on AI edge computing disclosed in an embodiment of the present invention.

[0037] Figure 2 It is a schematic diagram of a pressure variation curve disclosed in an embodiment of the present invention.

[0038] Figure 3 It is a structural schematic diagram of a system for monitoring natural gas pipeline pressure based on AI edge computing disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0040] See also Figure 1 As shown in the flowchart, an embodiment of the present invention provides a method for monitoring natural gas pipeline pressure based on AI edge computing, including the following steps:

[0041] Acquire and build a first AI edge prediction model and a second AI edge prediction model based on a federated learning mechanism;

[0042] Receiving a first pressure monitoring data group of each pressure monitoring point in a natural gas monitoring area, and obtaining a pressure change curve by fitting according to the first pressure monitoring data group;

[0043] identifying a plurality of first pressure drop curve segments in the pressure change curve, evaluating a first leakage probability of each of the first pressure drop curve segments using the first AI edge prediction model, and determining the first pressure drop curve segment having a first leakage probability higher than a probability threshold as a second pressure drop curve segment;

[0044] Receiving working status data of each gas-using terminal associated with the corresponding pressure monitoring point, performing a secondary evaluation on each second pressure drop curve segment based on each working status data using the second AI edge prediction model, and obtaining a second leakage probability of each second pressure drop curve segment;

[0045] When there is at least one second pressure drop curve segment with a second leakage probability higher than a probability threshold, it is confirmed that there is leakage, and corresponding alarm information is output.

[0046] The present invention pre-establishes an AI edge prediction model group, namely a first AI edge prediction model and a second AI edge prediction model, based on a federated learning mechanism. Both models are arranged in an edge server. At least one AI edge prediction model group can be arranged in a single natural gas monitoring area, and monitoring data and model data can be shared between edge servers.

[0047] In the actual leakage monitoring process of the natural gas pipeline, the first pressure monitoring data group of each pressure monitoring point in the natural gas monitoring area is fitted into a pressure change curve, such as Figure 2 As shown. The pressure change curve of each pressure monitoring point will contain multiple abnormal pressure drop sections, namely the first pressure drop curve sections. The slope and recovery law of the pressure rise and fall in the natural gas pipeline are different under normal conditions and in leakage conditions. The first AI edge prediction model trained using the corresponding training data has the ability to distinguish between normal and abnormal leakage, and can evaluate the first leakage probability of each first pressure drop curve section, and confirm the first pressure drop curve section with a first leakage probability higher than the probability threshold as the second pressure drop curve section.

[0048] At the same time, the pressure drop in the natural gas pipeline may also be caused by changes in the working status of the surrounding gas terminals. For example, when the gas terminal of a large gas user is turned on, the gas terminals of multiple users are turned on together. In this regard, the present invention further pre-constructs a second AI edge prediction model, which can perform a secondary evaluation on each second pressure drop curve segment based on the working status data of each user terminal associated with the surrounding pressure monitoring point. If the second leakage probability of the second pressure drop curve segment is higher than the probability threshold, it means that there is a natural gas leakage near the pressure monitoring point. At this time, the corresponding alarm information is output to instruct the processing personnel to check and deal with it in time.

[0049] The present invention uses an AI edge prediction model to perform two leakage assessments on the abnormal pressure drop section in the pressure change curve of the natural gas pipeline, which can greatly improve the accuracy and timeliness of leakage identification when there is a slight leak in the natural gas pipeline and the leakage point is far away from the pressure monitoring point.

[0050] Reference Figure 2 As shown, a, b, and c are three key points of trend change. When using the conventional leakage monitoring method, only when the pressure in the natural gas pipeline drops from point a to point c, that is, below the alarm lower limit, will it be determined that there is a leakage and an alarm will be issued. However, the solution of the present invention can accurately identify the leakage and issue an alarm when the pressure in the natural gas pipeline drops from point a to point b, which can save a lot of disposal time for the disposal personnel.

[0051] Further, the identifying a plurality of first pressure drop curve segments in the pressure change curve includes:

[0052] Each pressure drop section in the pressure variation curve is identified, and the slope of each pressure drop section is calculated, and a pressure drop section whose slope is higher than a slope threshold is determined as a first pressure drop curve section.

[0053] In this embodiment, the pressure in the natural gas pipeline fluctuates to a certain extent, but the fluctuation is relatively stable. When there is a natural gas leak, the fluctuation of the pressure change curve will be larger. Therefore, the present invention selects the first pressure drop curve segment where there may be a leak based on the slope of each pressure drop segment.

[0054] The slope of the pressure drop section may be the slope of a line connecting the first and last points of the pressure drop section, or may be the average value of the slopes of each subsection of the pressure drop section, which is not limited in the present invention.

[0055] In addition, the pressure change curve in the pipeline will also change with the change of seasons. At this time, the AI ​​algorithm can be used to collect pressure data in different temperature environments and learn the changes in the pressure curve over a long period of time, so as to determine multiple slope thresholds corresponding to different temperature segments. This can improve the judgment accuracy of the first pressure drop curve segment and reduce the false alarm rate.

[0056] Furthermore, the obtaining and constructing of the first AI edge prediction model based on the federated learning mechanism includes:

[0057] Publish the model training task in the blockchain, and receive the training data set and model parameters fed back by each blockchain participating node in response to the model training task;

[0058] Filtering target model parameters from the model parameters, and using the target model parameters to update the model parameters of the first AI edge prediction model;

[0059] A target training data set is obtained by screening each of the training data sets according to the target model parameters, and the first AI edge prediction model with updated model parameters is trained using the target training data set.

[0060] In this embodiment, federated learning is a distributed machine learning technology that allows multiple participants to jointly train a machine learning model while protecting data privacy. However, traditional federated learning performs data training on local devices and only shares model parameters or gradient information, not raw data. This approach is insufficient for training. The present invention improves the traditional federated learning mechanism, that is, the participants share not only the training data set but also the model parameters.

[0061] This node first publishes the model training task on the blockchain. The model training task contains the corresponding data requirements. The nodes that meet the requirements feedback the training data set and model parameters to this node. The nodes that meet the requirements can be the stored model with the ability to identify natural gas micro-leakage (for example, there is a successful recognition record). This node first screens the received model parameters to obtain the target model parameters, and then screens the target training data set from all the training data sets based on the target model parameters. The target training data set can be used to fully train the first AI edge prediction model.

[0062] Furthermore, the target model parameters are obtained by screening the model parameters, including:

[0063] The prediction record of the corresponding model is parsed from the feedback message of each blockchain participating node, and the number of accurate identification times of minor natural gas leaks by the model of the blockchain participating node is determined based on the prediction record, and the model parameters of the model with the highest accurate identification times are screened as the target model parameters.

[0064] In this embodiment, each blockchain participating node also stores corresponding models, which can identify and analyze obvious natural gas leaks and minor natural gas leaks, so that the number of accurate identifications can be statistically obtained based on the accuracy of the identification and analysis results. When the number of accurate identifications of the corresponding model is higher, its model parameters are preferably determined as the "initial" model parameters of the first AI edge prediction model of this node.

[0065] Furthermore, the step of obtaining a target training data set by screening each of the training data sets according to the target model parameters includes:

[0066] Determining a training data set corresponding to the target model parameters, and other training data sets;

[0067] Randomly selecting a number of training data corresponding to a first ratio from the training data set corresponding to the target model parameter, and randomly selecting a number of training data corresponding to a second ratio from other training data sets; wherein the first ratio is smaller than the second ratio;

[0068] All the selected training data are combined into the target training data set.

[0069] In this embodiment, after selecting the target model parameters, the priority of the training data set of the blockchain participating node corresponding to the target model parameters is set to be lower, while the priority of the training data sets of other blockchain participating nodes is increased. The reason for this setting is that the target model parameters are obtained based on the earlier training data set of the corresponding blockchain participating node, and the current training data set of the blockchain participating node (i.e., the training data set fed back to this node) has a high similarity with the earlier training data set. The optimization effect of using the current training data set to train the first AI edge prediction model of this node is limited, while using the training data sets of other blockchain participating nodes is likely to achieve better training effects, especially the recognition accuracy of more types of minor natural gas leaks can be obtained.

[0070] Among them, the total amount of training data in the target training data set can be a fixed value N (for example, 1000), and the first ratio and the second ratio are respectively configured for the blockchain participating nodes corresponding to the target model parameters and the other blockchain participating nodes (3), for example, the first ratio is 10% and the second ratio is 30%. Correspondingly, 0.1N training data are randomly selected from the training data set corresponding to the target model parameters, and 0.3N training data are randomly selected from the training data set corresponding to each other blockchain participating node.

[0071] Further, using the second AI edge prediction model to perform a secondary evaluation on each of the second pressure drop curve segments based on each of the working status data to obtain a second leakage probability of each of the second pressure drop curve segments, including:

[0072] Performing load evaluation according to each of the working status data to obtain a sufficient load value, wherein the sufficient load value represents a ratio of a real-time gas consumption rate of each gas-using terminal associated with the pressure monitoring point to a theoretical maximum gas consumption rate;

[0073] The load sufficiency value, the distance value between each gas terminal and the associated pressure monitoring point, and each second pressure drop curve segment are input into the second AI edge prediction model, and the second AI edge prediction model outputs the second leakage probability of each second pressure drop curve segment.

[0074] In this embodiment, the second AI edge prediction model is used to analyze the impact of fluctuations in gas consumption data of each gas-consuming terminal associated with the pressure monitoring point on the drop in gas pressure in the natural gas pipeline. It mainly includes two parameters, namely, the load sufficiency value and the distance value between each gas-consuming terminal and the associated pressure monitoring point.

[0075] The load sufficiency value indicates the ratio of the real-time gas consumption rate of each gas terminal associated with the pressure monitoring point to the theoretical maximum gas consumption rate, that is, the ratio of the real-time gas consumption rate of the associated gas terminals that have been opened to the total gas consumption rate when opened to the maximum extent. The larger the ratio, the higher the actual gas consumption rate, because the probability of the pressure drop in the natural gas pipeline caused by the sudden opening of the gas terminal is low; and the smaller the ratio, the lower the actual gas consumption rate, because there are still many gas terminals that have not been opened or the gas consumption rate is low, and the probability of the pressure drop in the natural gas pipeline caused by the sudden opening of the gas terminal or the increase in the gas consumption rate is high.

[0076] At the same time, the distance between the gas terminal and the associated pressure monitoring point will also affect the pressure drop rate in the pipeline at the pressure monitoring point. Generally speaking, the farther the gas terminal is from the pressure monitoring point, the later the pressure monitoring point will detect a drop in pressure or a significant drop in pressure after the gas terminal is turned on or the gas consumption rate is increased; and the closer the gas terminal is to the pressure monitoring point, the earlier the pressure monitoring point will detect a drop in pressure or a significant drop in pressure after the gas terminal is turned on or the gas consumption rate is increased.

[0077] Therefore, the present invention inputs the above two parameters together with each second pressure drop curve segment into the second AI edge prediction model, and the second AI edge prediction model performs a more accurate analysis of the second leakage probability of each second pressure drop curve segment.

[0078] It should be noted that the gas terminal can send its own location and real-time gas consumption rate data directly or indirectly to the corresponding edge computing server, and the specific details are not repeated here.

[0079] In addition, the second AI edge prediction model can be trained using the federated learning mechanism like the first AI edge prediction model, the only difference being the training data. The second AI edge prediction model mainly includes the load sufficiency value, the second pressure drop curve segment, the above distance value and the corresponding leakage label. After training, the second AI edge prediction model can analyze the load sufficiency value, the second pressure drop curve segment, and the influence factors corresponding to the above distance value, and then analyze the curve characteristics of each second pressure drop curve segment based on the influence factor to obtain the corresponding second leakage probability.

[0080] like Figure 3 As shown in the structural schematic diagram, the embodiment of the present invention also discloses a system for monitoring natural gas pipeline pressure based on AI edge computing, including a training module, a fitting module, a first recognition module, a second recognition module, and an early warning module;

[0081] The training module is used to acquire and construct the first AI edge prediction model and the second AI edge prediction model based on the federated learning mechanism;

[0082] The fitting module is used to receive a first pressure monitoring data group of each pressure monitoring point in the natural gas monitoring area, and obtain a pressure change curve by fitting according to the first pressure monitoring data group;

[0083] The first identification module is used to identify a plurality of first pressure drop curve segments in the pressure change curve, evaluate the first leakage probability of each of the first pressure drop curve segments using the first AI edge prediction model, and determine the first pressure drop curve segment whose first leakage probability is higher than a probability threshold as a second pressure drop curve segment;

[0084] The second identification module is used to receive the working status data of each gas terminal associated with the corresponding pressure monitoring point, and use the second AI edge prediction model to perform a secondary evaluation on each second pressure drop curve segment based on each working status data to obtain a second leakage probability of each second pressure drop curve segment;

[0085] The early warning module is used to confirm that there is a leakage and output corresponding alarm information when there is at least one second pressure drop curve segment with a second leakage probability higher than a probability threshold.

[0086] An embodiment of the present invention further discloses an electronic device, comprising: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the method described in the above embodiment.

[0087] It should be noted that the above processor may generally be referred to herein as a “data processing circuit.” The data processing circuit may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof.

[0088] The embodiment of the present invention further discloses a computer storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described in the above embodiment is executed.

[0089] The above storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store program codes.

[0090] It should be noted that the above-mentioned sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above-mentioned specific embodiments of the present application are described. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0091] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and equipment embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0092] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.

[0093] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for monitoring natural gas pipeline pressure based on AI edge computing, characterized in that: The steps include: Acquire and build a first AI edge prediction model and a second AI edge prediction model based on a federated learning mechanism; Receiving a first pressure monitoring data group of each pressure monitoring point in a natural gas monitoring area, and obtaining a pressure change curve by fitting according to the first pressure monitoring data group; identifying a plurality of first pressure drop curve segments in the pressure change curve, evaluating a first leakage probability of each of the first pressure drop curve segments using the first AI edge prediction model, and determining the first pressure drop curve segment having a first leakage probability higher than a probability threshold as a second pressure drop curve segment; Receiving working status data of each gas-using terminal associated with the corresponding pressure monitoring point, performing a secondary evaluation on each second pressure drop curve segment based on each working status data using the second AI edge prediction model, and obtaining a second leakage probability of each second pressure drop curve segment; When there is at least one second pressure drop curve segment whose second leakage probability is higher than the probability threshold, confirming that there is leakage and outputting corresponding alarm information; Acquire and build the first AI edge prediction model based on the federated learning mechanism, including: Publish the model training task in the blockchain, and receive the training data set and model parameters fed back by each blockchain participating node in response to the model training task; Filtering target model parameters from the model parameters, and using the target model parameters to update the model parameters of the first AI edge prediction model; A target training data set is obtained by screening each of the training data sets according to the target model parameters, and the first AI edge prediction model with updated model parameters is trained using the target training data set.

2. According to claim 1, a method for monitoring natural gas pipeline pressure based on AI edge computing is characterized in that: Identifying a plurality of first pressure drop curve segments in the pressure change curve comprises: Each pressure drop section in the pressure variation curve is identified, and the slope of each pressure drop section is calculated, and a pressure drop section whose slope is higher than a slope threshold is determined as a first pressure drop curve section.

3. According to claim 1, a method for monitoring natural gas pipeline pressure based on AI edge computing is characterized in that: Target model parameters are obtained by screening the model parameters, including: The prediction record of the corresponding model is parsed from the feedback message of each blockchain participating node, and the number of accurate identification times of minor natural gas leaks by the model of the blockchain participating node is determined based on the prediction record, and the model parameters of the model with the highest accurate identification times are screened as the target model parameters.

4. According to claim 3, a method for monitoring natural gas pipeline pressure based on AI edge computing is characterized in that: Filtering the target training data set from each of the training data sets according to the target model parameters comprises: Determining a training data set corresponding to the target model parameters, and other training data sets; Randomly selecting a number of training data corresponding to a first ratio from the training data set corresponding to the target model parameter, and randomly selecting a number of training data corresponding to a second ratio from other training data sets; wherein the first ratio is smaller than the second ratio; All the selected training data are combined into the target training data set.

5. According to claim 1, a method for monitoring natural gas pipeline pressure based on AI edge computing is characterized in that: Using the second AI edge prediction model to perform a secondary evaluation on each of the second pressure drop curve segments based on each of the working status data to obtain a second leakage probability of each of the second pressure drop curve segments, including: Performing load evaluation according to each of the working status data to obtain a sufficient load value, wherein the sufficient load value represents a ratio of a real-time gas consumption rate of each gas-using terminal associated with the pressure monitoring point to a theoretical maximum gas consumption rate; The load sufficiency value, the distance value between each gas terminal and the associated pressure monitoring point, and each second pressure drop curve segment are input into the second AI edge prediction model, and the second AI edge prediction model outputs the second leakage probability of each second pressure drop curve segment.

6. A system for monitoring natural gas pipeline pressure based on AI edge computing, the system is based on the method according to any one of claims 1 to 5, characterized in that: It includes a training module, a fitting module, a first recognition module, a second recognition module, and an early warning module; The training module is used to acquire and construct the first AI edge prediction model and the second AI edge prediction model based on the federated learning mechanism; The fitting module is used to receive a first pressure monitoring data group of each pressure monitoring point in the natural gas monitoring area, and obtain a pressure change curve by fitting according to the first pressure monitoring data group; The first identification module is used to identify a plurality of first pressure drop curve segments in the pressure change curve, evaluate the first leakage probability of each of the first pressure drop curve segments using the first AI edge prediction model, and determine the first pressure drop curve segment whose first leakage probability is higher than a probability threshold as a second pressure drop curve segment; The second identification module is used to receive the working status data of each gas terminal associated with the corresponding pressure monitoring point, and use the second AI edge prediction model to perform a secondary evaluation on each second pressure drop curve segment based on each working status data to obtain a second leakage probability of each second pressure drop curve segment; The early warning module is used to confirm that there is a leakage and output corresponding alarm information when there is at least one second pressure drop curve segment with a second leakage probability higher than a probability threshold.

7. An electronic device comprising: A memory storing executable program code; A processor coupled to the memory; characterized in that: the processor calls the executable program code stored in the memory to execute the method according to any one of claims 1-5.

8. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is executed.

9. A computer program product, characterized in that: When the computer program in the computer program product is executed, the method according to any one of claims 1 to 5 is implemented.

Images