Intelligent gas leakage monitoring method and system based on big data analysis

Abstract

The invention discloses an intelligent gas leakage monitoring method and system based on big data analysis, and belongs to the technical field of gas leakage monitoring, and the method comprises the steps: collecting the pressure data and the concentration data of gas in a pipeline in real time, and obtaining the current pressure data and the current concentration data; calculating a current state value according to the current pressure data and the current concentration data; when the current state value is greater than a state threshold value, collecting gas pipeline sound wave data; inputting the collected sound wave data into a pre-trained sound wave prediction model for judgment, and outputting an abnormal probability value; when the abnormal probability value is greater than a probability threshold value, scheduling the unmanned aerial vehicle to carry out video shooting along the pipeline to obtain accident video data; and the leakage condition is judged and the leakage position is positioned through the accident video data. According to the invention, a multi-level anomaly detection mechanism is adopted, so that the accuracy and reliability of gas leakage detection are greatly improved, and the occurrence of false alarms and missing alarms is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of gas leakage monitoring, and in particular to a gas leakage intelligent monitoring method and system based on big data analysis. Background Art

[0002] With the rapid development of the gas industry and the acceleration of urbanization, gas has become an indispensable energy source in modern urban life, and the safety of its supply and use has always been the focus of social attention. However, due to the aging, damage or human factors of gas pipelines, gas leakage accidents occur from time to time, which not only poses a serious threat to people's lives and property safety, but also poses a challenge to the safety and stability of urban operations. Therefore, the development of an efficient and accurate gas leakage monitoring technology is of great significance for timely detection and treatment of gas leakage problems and ensuring the safety of gas supply and use.

[0003] At present, with the rapid development of big data analysis and machine learning technology, these advanced technologies have provided new solutions for gas leak monitoring. Big data analysis technology can collect and process massive data from gas pipeline systems in real time, including gas pressure, concentration, temperature, humidity and other parameters. Through in-depth mining and analysis of these data, potential gas leak risks can be revealed. Machine learning technology can automatically identify and classify complex data features through training models, thereby improving the accuracy and robustness of the monitoring system.

[0004] However, most of the gas leak monitoring systems on the market currently only use single or limited monitoring methods, lack a comprehensive monitoring and early warning mechanism, and have no way to comprehensively analyze data, resulting in false alarms and missed alarms, reducing the accuracy and robustness of detection. Summary of the invention

[0005] To solve the above problems, the present invention provides a gas leakage intelligent monitoring method and system based on big data analysis. By real-time collection of gas pressure data and concentration data in the pipeline, big data analysis technology is used to perform state numerical analysis, and advanced technologies such as acoustic wave prediction models and drone video shooting are combined to achieve multi-level, high-precision monitoring and early warning of gas leaks.

[0006] The above objectives can be achieved through the following solutions: A gas leakage intelligent monitoring method based on big data analysis comprises: collecting gas pressure data and concentration data in a pipeline in real time to obtain current pressure data and current concentration data; calculating a current state value according to the current pressure data and the current concentration data; when the current state value is greater than a state threshold, collecting gas pipeline acoustic wave data; inputting the collected acoustic wave data into a pre-trained acoustic wave prediction model for judgment, and outputting an abnormal probability value; when the abnormal probability value is greater than the probability threshold, dispatching a drone to shoot video along the pipeline to obtain accident video data; judging the leakage situation and locating the leakage position through the accident video data.

[0007] Optionally, the calculation of the current state value according to the pressure data and the concentration data includes: collecting historical pressure data and historical concentration data of the gas in the pipeline; obtaining the average value and standard deviation of the historical pressure data according to the historical pressure data; obtaining the average value and standard deviation of the historical concentration data according to the historical concentration data; establishing a state function for characterizing the relationship between the state of the gas in the pipeline and the gas pressure and concentration according to the average value and standard deviation of the historical pressure data, the average value and standard deviation of the historical concentration data, the current pressure data and the current concentration data; inputting the current pressure data and the current concentration data into the state function to obtain the current state value, and the current state value of the gas in the pipeline ,have , In the formula, The weight coefficient representing the gas pressure in the pipeline, represents the current pressure data of the gas in the pipeline, Represents the average value of the historical pressure data of the gas in the pipeline, represents the standard deviation of the historical pressure data of the gas in the pipeline, The weight coefficient representing the gas concentration in the pipeline, Indicates the current concentration data of the gas in the pipeline, Represents the average value of the historical concentration data of the gas in the pipeline, Represents the standard deviation of the historical concentration data of the gas in the pipeline.

[0008] Optionally, when the state value is greater than a state threshold, collecting the gas pipeline acoustic wave data includes: collecting the pressure data and concentration data of the gas in the pipeline of each gas leakage accident to obtain an accident data set; substituting the pressure data and concentration data in the accident data set into the state function to obtain an accident state value set; setting a state threshold according to the accident state value set, and judging whether the current state value is greater than the state threshold; if so, collecting the gas pipeline acoustic wave data; if not, periodically collecting the gas pressure data and concentration data in the pipeline to calculate the current state value, and re-judging the size of the current state value and the state threshold.

[0009] Optionally, the collected sound wave data is input into a pre-trained sound wave prediction model for judgment, and the output abnormal probability value includes: collecting the sound wave data of pipeline leakage in each gas leakage accident and the sound wave data under normal conditions, and preprocessing them to obtain a historical sound wave data set; establishing a convolutional neural network model including an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the output layer uses a softmax function as an activation function; using the historical sound wave data set to train the convolutional neural network model to obtain a stable sound wave prediction model; inputting the collected sound wave data into the stable sound wave prediction model to obtain the abnormal probability value.

[0010] Optionally, the method further includes: when the current state value is greater than the state threshold and the abnormal probability value is less than or equal to the probability threshold, optimizing and updating the state threshold according to the current pressure data and the current concentration data.

[0011] Optionally, the method also includes: setting a flight path for the drone along the pipeline; when the abnormal probability value is greater than the probability threshold, automatically dispatching the drone to shoot video according to the flight path to obtain the accident video data; converting the accident video data frame by frame into a picture data set; using a spectral analysis algorithm to search for pictures with gas leakage characteristics from the picture data set to obtain accident picture data; using a spectral analysis algorithm to extract characteristic spectral information of the leakage point from the accident picture data; analyzing the leakage situation based on the characteristic spectral information; when a gas leakage is confirmed, locating the leakage position according to the accident picture data and issuing an alarm.

[0012] Optionally, the method also includes: when no image of gas leakage characteristics is found or it is confirmed that there is no leakage, optimizing and updating the state threshold according to the current pressure data and the current concentration data; optimizing the stable sound wave prediction model according to the collected gas pipeline sound wave data, and updating the probability threshold.

[0013] Optionally, locating the leakage position according to the accident picture data and issuing an alarm includes: dispatching a drone to shoot video according to the flight path to obtain standard video data; aligning the accident video data with the standard video data; determining the position of the accident picture data in the standard video data according to the position of the accident picture data in the accident video data to obtain position information; determining the leakage position according to the position information, and issuing an alarm.

[0014] Optionally, the method also includes: periodically scheduling the drone to take video according to the flight path and determine the leakage situation; if a leakage is confirmed, optimizing and updating the state threshold according to the pressure data and concentration data of the gas in the current pipeline; after updating the state threshold, collecting the current gas pipeline sound wave data and calculating the current abnormal probability value; if the current abnormal probability value is greater than the probability threshold, keeping the parameters of the stable sound wave prediction model and the probability threshold unchanged; if the current abnormal probability value is less than or equal to the probability threshold, optimizing the stable sound wave prediction model according to the collected current gas pipeline sound wave data, and updating the probability threshold.

[0015] Based on the same inventive concept, the present invention also provides a gas leakage intelligent monitoring system based on big data analysis, the system comprising: a pressure sensor, used to collect gas pressure data in the pipeline in real time to obtain current pressure data; a concentration sensor, used to collect gas concentration data in the pipeline in real time to obtain current concentration data; a state value analysis module, used to calculate the current state value according to the current pressure data and the current concentration data; the state value analysis module is also used to judge the size of the current state value and the state threshold; an acoustic wave data acquisition module, used to collect gas pipeline acoustic wave data when the current state value is greater than the state threshold; an abnormal probability analysis module, used to input the collected acoustic wave data into a pre-trained acoustic wave prediction model for judgment, and output an abnormal probability value; the abnormal probability analysis module is also used to judge the size of the abnormal probability value and the probability threshold; an accident video data acquisition module, used to use a drone to shoot a video along the pipeline to obtain accident video data when the abnormal probability value is greater than the probability threshold; a leakage analysis module, used to judge the leakage situation and locate the leakage position through the accident video data; a dynamic optimization module, used to optimize and update the state threshold, the acoustic wave prediction model and the probability threshold.

[0016] Compared with the prior art, the present invention has the following advantages: 1. The present invention can timely detect potential gas leakage by collecting gas pressure data and concentration data in the pipeline in real time; once an abnormal state is detected, the system will immediately trigger the subsequent detection process, including acoustic wave data collection and video shooting, so as to achieve rapid response and early warning of gas leakage; 2. The present invention adopts a multi-level anomaly detection mechanism, including state numerical analysis, acoustic wave prediction model judgment and spectral analysis confirmation; this multi-level detection mechanism can greatly improve the accuracy and reliability of gas leak detection and reduce the occurrence of false alarms and missed alarms; 3. The present invention realizes intelligent and automatic gas leak monitoring by introducing technologies such as big data analysis and machine learning. The system can automatically analyze data, judge abnormalities and take corresponding measures, which greatly reduces the workload of manual monitoring and improves monitoring efficiency. 4. The present invention has dynamic optimization and self-adaptation capabilities. The system can continuously optimize parameters such as state thresholds, acoustic wave prediction models and probability thresholds based on actual data and feedback to improve the accuracy and reliability of the system. At the same time, the system can also adapt to different pipeline characteristics and operating environments, and has strong versatility and flexibility.

[0017] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 It is a flow chart of a gas leakage intelligent monitoring method based on big data analysis according to an embodiment of the present invention.

[0020] Figure 2 It is an execution flow chart of a gas leak intelligent monitoring method based on big data analysis according to an embodiment of the present invention.

[0021] Figure 3 It is a schematic diagram of a process of periodically detecting gas leakage according to an embodiment of the present invention.

[0022] Figure 4 It is a structural schematic diagram of a gas leak intelligent monitoring system based on big data analysis according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] Reference Figure 1 An embodiment of the present invention proposes an intelligent gas leakage monitoring method based on big data analysis, which realizes multi-level, high-precision monitoring and early warning of gas leakage by collecting gas pressure data and concentration data in the pipeline in real time, using big data analysis technology to perform state numerical analysis, and combining advanced technologies such as acoustic wave prediction models and drone video shooting.

[0025] The method of this embodiment specifically includes: Collect gas pressure data and concentration data in the pipeline in real time to obtain current pressure data and current concentration data; Specifically, the pressure sensor and concentration sensor installed on the gas pipeline are used to collect the gas pressure data and concentration data in the pipeline in real time; these data are the basis for subsequent analysis.

[0026] Calculating a current state value according to the current pressure data and the current concentration data; When the current state value is greater than the state threshold, collecting gas pipeline acoustic wave data; Specifically, the current state value is compared with the preset state threshold. If the current state value is greater than the state threshold, it means that there may be a risk of gas leakage. At this time, the acoustic wave data collection process will be triggered. After confirming the existence of potential risks, the acoustic wave data of the gas pipeline will be collected. These data contain acoustic wave signals caused by gas leakage.

[0027] The collected sound wave data is input into the pre-trained sound wave prediction model for judgment, and the abnormal probability value is output; Specifically, the collected sound wave data is input into the sound wave prediction model, and the model outputs an abnormal probability value, which represents the possibility of abnormal signals in the sound wave data.

[0028] When the abnormal probability value is greater than the probability threshold, dispatching a drone to shoot video along the pipeline to obtain accident video data; Specifically, if the abnormal probability value is greater than the preset probability threshold, it means that there is indeed an abnormal signal in the sound wave data. At this time, the drone video shooting process will be triggered; thereby automatically dispatching the drone to shoot video along the pipeline to obtain accident video data, which contains intuitive image information of the leakage site.

[0029] The leakage situation is determined and the leakage position is located through the accident video data.

[0030] Specifically, through the accident video data, image processing technologies such as spectral analysis algorithms can be used to determine the leakage situation and locate the leakage position. This process can include converting the video data frame by frame into an image data set, and using algorithms to find images with gas leakage characteristics. Once a gas leak is confirmed, an alarm will be issued immediately and the leakage location information will be provided. At the same time, parameters such as state thresholds, acoustic wave prediction models and probability thresholds will be optimized based on current data and feedback to improve the accuracy and reliability of subsequent monitoring.

[0031] For example, Figure 2 As shown in the figure, it is assumed that the real-time data collected by the pressure sensor and concentration sensor of a certain section of gas pipeline are 1.2MPa and 500ppm respectively (assuming that these values ​​are abnormal relative to historical data); these data are input into the state function, and the current state value is calculated to be 0.85 (assuming that the state threshold is 0.8); because the current state value is greater than the state threshold, the acoustic wave data collection process is triggered; the acoustic wave prediction model determines that the probability of abnormality of the collected acoustic wave data is 0.95 (assuming that the probability threshold is 0.9); therefore, the drone is dispatched to take video shots, and the existence of gas leakage is confirmed through video data, and the leakage position is located; subsequently, an alarm is issued, and relevant parameters are optimized to meet subsequent monitoring needs.

[0032] Optionally, calculating the current state value according to the pressure data and the concentration data includes: Collect historical pressure data and historical concentration data of gas in pipelines; According to the historical pressure data, obtaining an average value and a standard deviation of the historical pressure data; According to the historical concentration data, an average value and a standard deviation of the historical concentration data are obtained; Establishing a state function for characterizing the relationship between the state of the gas in the pipeline and the gas pressure and concentration according to the average value and standard deviation of the historical pressure data, the average value and standard deviation of the historical concentration data, the current pressure data and the current concentration data; The current pressure data and the current concentration data are input into the state function to obtain the current state value. For the current state value of the gas in the pipeline ,have , In the formula, The weight coefficient representing the gas pressure in the pipeline, represents the current pressure data of the gas in the pipeline, Represents the average value of the historical pressure data of the gas in the pipeline, represents the standard deviation of the historical pressure data of the gas in the pipeline, The weight coefficient representing the gas concentration in the pipeline, Indicates the current concentration data of the gas in the pipeline, Represents the average value of the historical concentration data of the gas in the pipeline, Represents the standard deviation of the historical concentration data of the gas in the pipeline.

[0033] Specifically, when calculating the current state value, the weight coefficient and The selection of is very important, as they reflect the relative importance of gas pressure and concentration in evaluating pipeline status; usually, these weight coefficients are determined based on practical experience and data analysis to ensure the accuracy and reliability of the status function.

[0034] For example, firstly, the historical pressure data and historical concentration data of the gas in the pipeline are collected. These data are the records of the pressure and concentration of the gas in the pipeline in the past period of time, which are used to calculate the average value and standard deviation, and to subsequently establish the state function; based on the collected historical pressure data, the average value of the historical pressure data is calculated ( ) and standard deviation ( ); Similarly, based on the historical concentration data, calculate the average value of the historical concentration data ( ) and standard deviation ( Next, using the mean and standard deviation of historical data and current pressure data ( ) and current concentration data ( ), establish a state function for characterizing the relationship between the gas state in the pipeline and the gas pressure and concentration; input the current pressure data and current concentration data collected in real time into the state function; obtain the current state value through calculation ; This value reflects the current state of the gas in the pipeline. By comparing this value with the preset state threshold, it is determined whether it is necessary to further collect acoustic wave data for judgment, or to dispatch a drone to take video to confirm the leak. This process realizes multi-level, high-precision monitoring and early warning of gas leaks, providing strong support for the timely detection and handling of gas leaks.

[0035] Optionally, when the state value is greater than a state threshold, collecting gas pipeline acoustic wave data includes: Collect the pressure data and concentration data of the gas in the pipeline of each gas leakage accident to obtain the accident data set; For example, in past gas leakage accidents, the pressure data and concentration data of the gas in the pipelines of multiple accidents were recorded; these data were organized into an accident data set for subsequent analysis and modeling.

[0036] Substituting the pressure data and concentration data in the accident data set into the state function to obtain an accident state value set; Exemplarily, a state function is established using the collected historical pressure data and historical concentration data (including the mean and standard deviation) to characterize the relationship between the gas state in the pipeline and the gas pressure and concentration; then the pressure data and concentration data in the accident data set are substituted into the state function to obtain a series of accident state values, which are used to set state thresholds so as to determine whether the current state is abnormal in subsequent monitoring.

[0037] Setting a state threshold according to the accident state value set, and determining whether the current state value is greater than the state threshold; For example, in daily monitoring, the gas pressure data and concentration data in the pipeline are collected in real time, and these data are input into the state function to calculate the current state value; when it is detected that the current state value is greater than the preset state threshold, this means that there may be a risk of gas leakage; at this time, the acoustic wave data collection will be automatically triggered to start collecting the acoustic wave data of the gas pipeline.

[0038] If yes, then collect gas pipeline acoustic wave data; Exemplarily, the collected sound wave data is input into a pre-trained sound wave prediction model; the model uses convolutional neural networks and other technologies to process and analyze the sound wave data to determine whether there are abnormal characteristics of gas leakage; if the abnormal probability value output by the sound wave prediction model is greater than a preset probability threshold, further action will be taken, such as dispatching a drone to shoot video.

[0039] If not, the gas pressure data and concentration data in the pipeline are periodically collected to calculate the current state value, and the magnitude of the current state value and the state threshold is re-judged.

[0040] For example, if the current state value does not exceed the state threshold, the gas pressure data and concentration data in the pipeline will continue to be collected periodically, and the current state value will be recalculated for monitoring; at the same time, according to the actual monitoring data and feedback results, the state threshold, acoustic wave prediction model, probability threshold and other parameters will be continuously optimized to improve the accuracy and reliability of monitoring.

[0041] Optionally, the step of inputting the collected sound wave data into a pre-trained sound wave prediction model for judgment and outputting an abnormal probability value comprises: Collect the acoustic wave data of pipeline leakage in each gas leakage accident and the acoustic wave data under normal conditions, and pre-process them to obtain a historical acoustic wave data set; For example, a large amount of acoustic wave data generated by pipeline leaks in gas leakage accidents and acoustic wave data in normal conditions (i.e., when there is no leakage) are first collected; these data come from gas pipelines at different times, locations, and environmental conditions; in order to ensure the accuracy and generalization ability of the model, these acoustic wave data need to be preprocessed, including denoising, filtering, normalization and other steps to eliminate background noise and interference and improve data quality; after preprocessing, a historical acoustic wave data set containing leakage and normal state acoustic wave data is obtained.

[0042] Establishing a convolutional neural network model including an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the output layer uses a softmax function as an activation function; Exemplarily, a convolutional neural network model including an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer is established; the input layer is responsible for receiving the preprocessed sound wave data as input; the convolutional layer extracts feature information in the sound wave data through convolution operations; the pooling layer downsamples the features output by the convolutional layer to reduce the data dimension and the amount of calculation; the fully connected layer maps the features output by the pooling layer to the final classification result; the output layer uses the softmax function as the activation function to convert the output of the fully connected layer into a probability distribution, that is, the probability values ​​of abnormal (leakage) and normal states.

[0043] Using the historical sound wave data set to train the convolutional neural network model to obtain a stable sound wave prediction model; Exemplarily, a convolutional neural network model is trained using a historical sound wave dataset; during the training process, the model's parameters (such as weights and biases) are continuously adjusted to minimize the error between the predicted results and the true labels; through multiple iterations and optimizations, a stable sound wave prediction model is obtained that can accurately distinguish between leakage and normal state sound wave data.

[0044] The collected sound wave data is input into the stable sound wave prediction model to obtain the abnormal probability value.

[0045] For example, in the actual monitoring process, when an abnormal gas state in the pipeline is detected (such as abnormal pressure or concentration), the acoustic wave data of the gas pipeline will be automatically collected; the collected acoustic wave data will be pre-processed and input into a stable acoustic wave prediction model; the model processes and analyzes the input acoustic wave data and outputs the probability value of the abnormal (leakage) state.

[0046] For example, if the abnormal probability value output by the acoustic wave prediction model is greater than a preset probability threshold (such as 0.8 or higher), it is considered that there is a risk of gas leakage, and corresponding response measures are automatically triggered, such as dispatching a drone to shoot a video, sending an alarm to notify relevant personnel, etc.; if the abnormal probability value is lower than the probability threshold, the current state is considered normal, and the gas pressure and concentration data in the pipeline continue to be periodically collected for monitoring. At the same time, the drone inspection cycle can also be set according to the calculated abnormal probability value.

[0047] Alternatively, if Figure 2 As shown, the method also includes: When the current state value is greater than the state threshold and the abnormal probability value is less than or equal to the probability threshold, the state threshold is optimized and updated according to the current pressure data and the current concentration data.

[0048] Exemplarily, the gas pressure data and concentration data in the pipeline are collected in real time, and the current state value is calculated based on these data; when it is detected that the current state value is greater than a preset state threshold, it is considered that there may be a risk of gas leakage, and the acoustic wave data collection is automatically triggered; after pre-processing, the collected acoustic wave data is input into a pre-trained acoustic wave prediction model to process and analyze the acoustic wave data, and output a probability value of the abnormal (leakage) state; if the abnormal probability value output by the acoustic wave prediction model is less than or equal to a preset probability threshold (for example, the probability threshold is set to 0.7), this means that although the current state value exceeds the state threshold, the acoustic wave data does not show obvious leakage characteristics; in this case, an alarm will not be triggered immediately or other countermeasures will be taken, but the state threshold will be optimized and updated based on the current pressure data and concentration data, as well as historical data; the optimization process may include recalculating the mean and standard deviation of the historical data, adjusting the weight coefficient and , or use other statistical methods to determine a more reasonable status threshold; the updated status threshold will be used for subsequent monitoring and judgment to improve the accuracy and adaptability of the system; after the status threshold is optimized and updated, the gas pressure data and concentration data in the pipeline will continue to be collected periodically, and the current status value will be recalculated; if the new current status value still exceeds the updated status threshold, and the abnormal probability value output by the acoustic wave prediction model also exceeds the probability threshold, an alarm will be triggered and corresponding countermeasures will be taken; thereby improving the flexibility and accuracy of the system, reducing false alarms and missed alarms, and better ensuring the safe operation of the gas pipeline.

[0049] Optionally, the method further comprises: Setting up a flight path for the drone along the pipeline; For example, first, according to the layout of the gas pipeline and the surrounding environment, a series of UAV flight paths are planned along the pipeline. These paths ensure that the UAV can fully cover the pipeline area while avoiding collisions with obstacles (such as trees, buildings, etc.); the setting of the flight path also takes into account factors such as the UAV's endurance, flight speed, and shooting angle to ensure that high-quality video data is captured.

[0050] When the abnormal probability value is greater than the probability threshold, the drone is automatically dispatched to shoot video according to the flight path to obtain the accident video data; Exemplarily, the gas pressure, concentration and other parameters in the pipeline are collected in real time, and the abnormal probability value is calculated through a pre-trained acoustic wave prediction model; when the abnormal probability value exceeds a preset probability threshold (for example, the probability threshold is set to 0.8), it is considered that there is a risk of gas leakage, and the dispatch instruction of the drone is automatically triggered; after receiving the instruction, the drone quickly arrives at the suspected leakage area according to the preset flight path, and starts video shooting to obtain accident video data.

[0051] Converting the accident video data frame by frame into a picture data set; Using a spectral analysis algorithm to search for images showing gas leakage characteristics from the image data set to obtain accident image data; Extracting characteristic spectrum information of the leakage point from the accident image data using a spectrum analysis algorithm; Exemplarily, the accident video data is converted frame by frame into an image data set for subsequent spectral analysis; the image data set is analyzed frame by frame using a spectral analysis algorithm to find images showing gas leakage characteristics (such as color change, gas diffusion, etc.), which are marked as accident image data; then, characteristic spectral information of the leakage point is extracted from the accident image data, which includes the composition, concentration, and diffusion range of the leaked gas.

[0052] Analyze leakage according to the characteristic spectrum information; For example, based on the extracted characteristic spectral information, an in-depth analysis of the leakage situation is carried out; this includes determining the type of leakage (such as small leakage, large leakage, etc.), assessing the severity of the leakage, and predicting the possible consequences of the leakage; at the same time, the spatial information in the accident image data (such as GPS coordinates, shooting angles, etc.) is used, combined with the flight path and shooting parameters of the drone, to accurately locate the position of the leakage point.

[0053] When a gas leak is confirmed, the leak location is located according to the accident image data and an alarm is issued.

[0054] For example, when a gas leak is confirmed, the alarm mechanism is immediately triggered and a leak alarm message is sent to relevant departments and personnel. The alarm message includes information such as the location of the leak, the type of leak, the severity of the leak, and recommended emergency response measures. After receiving the alarm, relevant departments and personnel can take prompt action, such as closing the gas valve near the leak, evacuating surrounding personnel, and initiating emergency plans, so as to minimize the losses and risks caused by the leak. By combining the advantages of drone technology, spectral analysis algorithms, and data processing technology, rapid and accurate monitoring and positioning of gas leaks are achieved, providing a strong guarantee for the safe operation of gas pipelines.

[0055] Alternatively, if Figure 2 As shown, the method also includes: When no image of gas leakage characteristics is found or it is confirmed that there is no leakage, the state threshold is optimized and updated according to the current pressure data and the current concentration data; For example, after the video data taken by the drone is spectrally analyzed, if no images with gas leakage characteristics are found, or after comprehensive analysis it is confirmed that no leakage has occurred, it will enter an optimization stage; in this stage, the current gas pipeline pressure data and concentration data will be reviewed and analyzed, which reflect the parameter changes of the pipeline under normal operating conditions; based on these data, statistical learning methods (such as Bayesian optimization, gradient descent, etc.) will be used to optimize and update the status threshold; the new status threshold will better reflect the actual operating status of the pipeline and reduce the possibility of false alarms and missed alarms.

[0056] The stable acoustic wave prediction model is optimized according to the collected gas pipeline acoustic wave data, and the probability threshold is updated.

[0057] For example, after the video data taken by the drone is spectrally analyzed, if no images with gas leakage characteristics are found, or after comprehensive analysis it is confirmed that no leakage has occurred, the current gas pipeline acoustic wave data will be reviewed and analyzed, and the acoustic wave prediction model and probability threshold will be optimized and updated based on these and historical data; at the same time, the acoustic wave data of the gas pipeline will continue to be collected, and these data include acoustic wave signals generated by gas flow inside the pipeline, equipment vibration, etc.; these acoustic wave data are input into the pre-trained acoustic wave prediction model, and the model will output the probability value of the abnormal (leakage) state based on the input data; over time, a large amount of acoustic wave data and corresponding labels (i.e., whether a leak has occurred) will be accumulated, and these data can be used to further train and optimize the acoustic wave prediction model to improve its prediction accuracy; during the optimization process, the probability threshold will also be adjusted based on the new data and the performance of the model; the new probability threshold will better balance the sensitivity and specificity of the system to ensure that true leaks are detected while reducing false alarms.

[0058] Illustratively, these two optimization processes (updating status thresholds and optimizing acoustic wave prediction models and probability thresholds) are important components of continuous optimization and iteration; these optimization processes can be triggered periodically or according to preset conditions to ensure that they always remain in the best state; in addition, parameter changes and performance before and after each optimization are recorded for subsequent analysis and evaluation; these optimization processes not only improve the accuracy and reliability of the system, but also enhance its self-learning and adaptability, providing a more solid guarantee for the safe operation of the gas pipeline.

[0059] Optionally, locating the leakage position according to the accident image data and issuing an alarm includes: Dispatching the drone to shoot video according to the flight path to obtain standard video data; For example, the flight path of the drone is planned according to the layout of the gas pipeline and the surrounding environment to ensure that the drone can fully cover the pipeline area; the drone is dispatched to shoot video according to the planned flight path to obtain standard video data containing scenes along the pipeline; these video data will serve as a benchmark for subsequent analysis.

[0060] Aligning the accident video data with the standard video data; Exemplarily, key frames and feature points, such as edges, corners, textures, etc. in the image, are extracted from the standard video data and the accident video data; the feature points in the accident video data are matched with the feature points in the standard video data using a feature matching algorithm (such as SIFT, SURF, etc.) to find the correspondence between the two; based on the matching results of the feature points, the time axis and image coordinate axis of the accident video data and the standard video data are aligned to ensure that the two are consistent in time and space.

[0061] According to the position of the accident picture data in the accident video data, determine the position of the accident picture data in the standard video data to obtain position information; Exemplarily, in the accident video data, pictures containing gas leakage characteristics (accident picture data) are marked. These pictures usually have specific color changes, gas diffusion and other characteristics; according to the position of the accident picture data in the accident video data (such as timestamp and image coordinates), the corresponding positions of these pictures in the standard video data are determined using the results of video alignment; combined with the flight path and shooting parameters of the drone (such as altitude, angle, speed, etc.), as well as the geographic information in the standard video data (such as GPS coordinates), the precise coordinates of the leakage location are calculated.

[0062] The leakage location is determined according to the location information, and an alarm is issued.

[0063] For example, based on the location information, the specific location of the leakage point is determined and marked on the map; key information such as the leakage location information, leakage type, and severity are notified to relevant departments and personnel via text messages, emails, telephone calls, etc.; after receiving the alarm, relevant departments and personnel will quickly activate the emergency plan based on the leakage location and severity, and take necessary measures (such as closing the gas valve, evacuating surrounding personnel, etc.) to ensure safety.

[0064] Alternatively, if Figure 3 As shown, the method also includes: Periodically dispatching drones to take video according to the flight path and determine the leakage situation; For example, according to the layout of the pipeline and the surrounding environment, the flight path of the drone is planned to ensure that the pipeline area can be fully covered; at set time intervals (such as daily, weekly, etc.), the drone is dispatched to take video according to the planned flight path, and the gas pressure data and concentration data in the pipeline are collected simultaneously; through comprehensive analysis of video analysis, pressure data and concentration data, it is determined whether there is a leak in the pipeline. For example, video analysis can detect whether there is abnormal gas diffusion around the pipeline, and pressure data and concentration data can reflect changes in the state of gas in the pipeline.

[0065] If leakage is confirmed, the state threshold is optimized and updated according to the pressure data and concentration data of the gas in the current pipeline; For example, if a leak is found through periodic drone inspections, the state threshold optimization and update stage is entered to collect the pressure data and concentration data of the gas in the current pipeline. These data will be used to optimize the state threshold; the state threshold is optimized based on the collected data to more accurately reflect the actual operating status of the pipeline; the optimized state threshold is updated to the monitoring system for subsequent use.

[0066] After updating the state threshold, collecting the current gas pipeline acoustic wave data, and calculating the current abnormal probability value; Exemplarily, after updating the status threshold, it is necessary to further confirm whether the acoustic wave prediction model and the probability threshold are accurate; collect the acoustic wave data of the current gas pipeline, which will be used to evaluate the performance of the acoustic wave prediction model; input the collected acoustic wave data into a stable acoustic wave prediction model to calculate the current abnormal probability value.

[0067] If the current abnormal probability value is greater than the probability threshold, the parameters of the stable sound wave prediction model and the probability threshold are kept unchanged; If the current abnormal probability value is less than or equal to the probability threshold, the stable sound wave prediction model is optimized according to the current gas pipeline sound wave data collected, and the probability threshold is updated.

[0068] For example, if the current abnormal probability value is greater than the probability threshold, it means that the performance of the acoustic wave prediction model is good and can accurately identify leakage signals. At this time, keep the parameters and probability threshold of the acoustic wave prediction model unchanged; if the current abnormal probability value is less than or equal to the probability threshold, it means that the performance of the acoustic wave prediction model needs to be improved. At this time, enter the model optimization and probability threshold update stage; optimize the acoustic wave prediction model according to the collected acoustic wave data to improve its accuracy in identifying leakage signals; adjust and update the probability threshold according to the performance of the optimized acoustic wave prediction model to match the performance of the model.

[0069] For example, suppose that during a periodic drone inspection, the drone captured a video clip of a suspected leak and confirmed the existence of the leak through comprehensive analysis; then, the pressure data and concentration data of the gas in the current pipeline were collected, and the state threshold was optimized and updated; then, the acoustic wave data of the current gas pipeline was collected, and the probability value of the abnormality was calculated; after judgment, it was found that the current abnormal probability value was less than the probability threshold, so it was decided to optimize the acoustic wave prediction model and update the probability threshold; through this series of optimization and update operations, the accuracy and sensitivity of leak monitoring are improved, providing a strong guarantee for the safe operation of the pipeline.

[0070] Based on the same inventive concept, Figure 4 As shown, the present invention also provides a gas leakage intelligent monitoring system based on big data analysis, the system comprising: The pressure sensor is used to collect the gas pressure data in the pipeline in real time and obtain the current pressure data; Concentration sensor, used to collect gas concentration data in the pipeline in real time and obtain current concentration data; A state value analysis module, used to calculate the current state value according to the current pressure data and the current concentration data; the state value analysis module is also used to determine the magnitude of the current state value and the state threshold; An acoustic wave data collection module, used for collecting the acoustic wave data of the gas pipeline when the current state value is greater than the state threshold; An abnormal probability analysis module is used to input the collected sound wave data into a pre-trained sound wave prediction model for judgment and output an abnormal probability value; the abnormal probability analysis module is also used to judge the size of the abnormal probability value and the probability threshold; An accident video data acquisition module is used to obtain accident video data by using a drone to shoot video along the pipeline when the abnormal probability value is greater than a probability threshold; A leakage analysis module, used to determine the leakage situation and locate the leakage position through the accident video data; A dynamic optimization module is used to optimize and update the state threshold, the sound wave prediction model and the probability threshold.

[0071] It should be noted that the electrical connection between the above-mentioned units does not necessarily mean the connection between the lines. The indirect connection method can be applied to the embodiments of the present invention as long as the purpose of the present invention is achieved. The above is only an exemplary embodiment of the present invention and cannot be used to limit the scope of the present invention.

[0072] That is, any equivalent changes and modifications made according to the teachings of the present invention are still within the scope of the present invention. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present invention. This application is intended to cover any variation, use or adaptive change of the present invention, which follows the general principles of the present invention and includes common knowledge or customary technical means in the art that are not described in the present invention.

Claims

1. A gas leak intelligent monitoring method based on big data analysis, characterized in that: The method comprises: Collect gas pressure data and concentration data in the pipeline in real time to obtain current pressure data and current concentration data; Calculating a current state value according to the current pressure data and the current concentration data; When the current state value is greater than the state threshold, collecting gas pipeline acoustic wave data; The collected sound wave data is input into the pre-trained sound wave prediction model for judgment, and the abnormal probability value is output; When the abnormal probability value is greater than the probability threshold, dispatching a drone to shoot video along the pipeline to obtain accident video data; The leakage situation is determined and the leakage position is located through the accident video data.

2. According to claim 1, a gas leak intelligent monitoring method based on big data analysis is characterized in that: Calculating the current state value according to the pressure data and the concentration data includes: Collect historical pressure data and historical concentration data of gas in pipelines; According to the historical pressure data, obtaining an average value and a standard deviation of the historical pressure data; According to the historical concentration data, an average value and a standard deviation of the historical concentration data are obtained; Establishing a state function for characterizing the relationship between the state of the gas in the pipeline and the gas pressure and concentration according to the average value and standard deviation of the historical pressure data, the average value and standard deviation of the historical concentration data, the current pressure data and the current concentration data; The current pressure data and the current concentration data are input into the state function to obtain the current state value. For the current state value of the gas in the pipeline ,have ， In the formula, The weight coefficient representing the gas pressure in the pipeline, represents the current pressure data of the gas in the pipeline, Represents the average value of the historical pressure data of the gas in the pipeline, represents the standard deviation of the historical pressure data of the gas in the pipeline, The weight coefficient representing the gas concentration in the pipeline, Indicates the current concentration data of the gas in the pipeline, Represents the average value of the historical concentration data of the gas in the pipeline, Represents the standard deviation of the historical concentration data of the gas in the pipeline.

3. The intelligent gas leakage monitoring method based on big data analysis according to claim 2 is characterized in that: The method comprises: Collect the pressure data and concentration data of the gas in the pipeline of each gas leakage accident to obtain the accident data set; Substituting the pressure data and concentration data in the accident data set into the state function to obtain an accident state value set; Setting a state threshold according to the accident state value set, and determining whether the current state value is greater than the state threshold; If yes, then collect gas pipeline acoustic wave data; If not, the gas pressure data and concentration data in the pipeline are periodically collected to calculate the current state value, and the magnitude of the current state value and the state threshold is re-judged.

4. According to claim 1, a gas leak intelligent monitoring method based on big data analysis is characterized in that: The collected sound wave data is input into a pre-trained sound wave prediction model for judgment, and the output abnormal probability value includes: Collect the acoustic wave data of pipeline leakage in each gas leakage accident and the acoustic wave data under normal conditions, and pre-process them to obtain a historical acoustic wave data set; Establishing a convolutional neural network model including an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the output layer uses a softmax function as an activation function; Using the historical sound wave data set to train the convolutional neural network model to obtain a stable sound wave prediction model; The collected sound wave data is input into the stable sound wave prediction model to obtain the abnormal probability value.

5. The intelligent gas leakage monitoring method based on big data analysis according to claim 1 is characterized in that: The method further comprises: When the current state value is greater than the state threshold and the abnormal probability value is less than or equal to the probability threshold, the state threshold is optimized and updated according to the current pressure data and the current concentration data.

6. The intelligent gas leakage monitoring method based on big data analysis according to claim 4 is characterized in that: The method further comprises: Setting up a flight path for the drone along the pipeline; When the abnormal probability value is greater than the probability threshold, the drone is automatically dispatched to shoot video according to the flight path to obtain the accident video data; Converting the accident video data frame by frame into a picture data set; Using a spectral analysis algorithm to search for images showing gas leakage characteristics from the image data set to obtain accident image data; Extracting characteristic spectrum information of the leakage point from the accident image data using a spectrum analysis algorithm; Analyze leakage according to the characteristic spectrum information; When a gas leak is confirmed, the leak location is located according to the accident image data and an alarm is issued.

7. The method for intelligent gas leakage monitoring based on big data analysis according to claim 6 is characterized in that: The method further comprises: When no image of gas leakage characteristics is found or it is confirmed that there is no leakage, the state threshold is optimized and updated according to the current pressure data and the current concentration data; The stable acoustic wave prediction model is optimized according to the collected gas pipeline acoustic wave data, and the probability threshold is updated.

8. The method for intelligent gas leakage monitoring based on big data analysis according to claim 6 is characterized in that: The locating the leakage position according to the accident image data and issuing an alarm comprises: Dispatching the drone to shoot video according to the flight path to obtain standard video data; Aligning the accident video data with the standard video data; According to the position of the accident picture data in the accident video data, determine the position of the accident picture data in the standard video data to obtain position information; The leakage location is determined according to the location information, and an alarm is issued.

9. The intelligent gas leakage monitoring method based on big data analysis according to claim 6 is characterized in that: The method further comprises: Periodically dispatching drones to take video according to the flight path and determine the leakage situation; If leakage is confirmed, the state threshold is optimized and updated according to the pressure data and concentration data of the gas in the current pipeline; After updating the state threshold, collecting the current gas pipeline acoustic wave data, and calculating the current abnormal probability value; If the current abnormal probability value is greater than the probability threshold, the parameters of the stable sound wave prediction model and the probability threshold are kept unchanged; If the current abnormal probability value is less than or equal to the probability threshold, the stable sound wave prediction model is optimized according to the current gas pipeline sound wave data collected, and the probability threshold is updated.

10. A gas leak intelligent monitoring system based on big data analysis, characterized in that: The system comprises: The pressure sensor is used to collect the gas pressure data in the pipeline in real time and obtain the current pressure data; Concentration sensor, used to collect gas concentration data in the pipeline in real time and obtain current concentration data; A state value analysis module, used to calculate the current state value according to the current pressure data and the current concentration data; the state value analysis module is also used to determine the magnitude of the current state value and the state threshold; An acoustic wave data collection module, used for collecting the acoustic wave data of the gas pipeline when the current state value is greater than the state threshold; An abnormal probability analysis module is used to input the collected sound wave data into a pre-trained sound wave prediction model for judgment and output an abnormal probability value; the abnormal probability analysis module is also used to judge the size of the abnormal probability value and the probability threshold; An accident video data acquisition module is used to obtain accident video data by using a drone to shoot video along the pipeline when the abnormal probability value is greater than a probability threshold; A leakage analysis module, used to determine the leakage situation and locate the leakage position through the accident video data; A dynamic optimization module is used to optimize and update the state threshold, the sound wave prediction model and the probability threshold.

Images