Natural gas pipeline station security monitoring system and method

Abstract

The invention discloses a natural gas pipeline station security and protection monitoring system and method, and relates to the technical field of natural gas pipe networks, the system comprises a gas sensor, at least one environmental sensor, a data acquisition unit and a data processing unit, the system also comprises a storage unit, the storage unit stores a dynamic compensation model in advance, and the dynamic compensation model is a dynamic compensation model; the dynamic compensation model is used for representing a nonlinear mapping relationship between the response deviation of the gas sensor and the environmental parameters monitored by the at least one environmental sensor. The storage unit storing the dynamic compensation model is arranged, the data acquisition unit is combined with original gas concentration data and environmental parameter data synchronously acquired in real time, and the data processing unit calls the model to calculate the dynamic compensation value and correct the original data, so that environmental change can be dynamically adapted in real time; the problem that fixed compensation cannot deal with nonlinear drift in a complex environment is solved.

Description

Technical Field

[0001] This invention relates to the field of natural gas pipeline technology, specifically to a natural gas pipeline station security monitoring system and method. Background Technology

[0002] In natural gas pipeline station security systems, the sensitivity of gas leak detection sensors (such as infrared and catalytic combustion type) is easily affected by environmental factors, with fluctuations in ambient temperature, humidity, and gas pressure being the main sources of interference. Specifically, high temperatures may cause response delays in catalytic combustion sensors, while high humidity conditions may cause optical path attenuation in infrared sensors, leading to a drift in the detection threshold. The correlation between these environmental factors and sensor performance is a fundamental characteristic that needs to be considered during the operation of station security systems.

[0003] However, existing systems have the following problems when dealing with the above issues: 1. Using a fixed temperature compensation coefficient or periodic manual calibration makes it difficult to adapt to environmental changes in real time. Fixed compensation formulas cannot cover nonlinear drift under complex environments (such as large day-night temperature differences and sudden humidity changes). 2. There is a lack of multi-sensor collaboration. Temperature and humidity sensors and gas sensors often work independently and no effective parameter correlation model has been established. As a result, temperature and humidity data cannot be used in compensation calculations, which affects the accuracy of detection. 3. The self-calibration process relies on manual operation, and periodic calibration requires shutdown, which affects the continuity of monitoring and creates monitoring blind spots. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a natural gas pipeline station security monitoring system and method. It solves the problems of gas leak detection sensor sensitivity being easily affected by environmental temperature, humidity, and gas pressure fluctuations in existing natural gas pipeline station security systems, and existing systems using fixed compensation or periodic manual calibration which are difficult to adapt to nonlinear drift in complex environments. Furthermore, the lack of multi-sensor collaboration and reliance on manual shutdown for self-calibration result in monitoring blind spots.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a natural gas pipeline station security monitoring system, comprising a gas sensor, at least one environmental sensor, a data acquisition unit, and a data processing unit; the system further includes: A storage unit, wherein a dynamic compensation model is pre-stored in the storage unit, the dynamic compensation model being used to characterize the nonlinear mapping relationship between the response deviation of the gas sensor and the environmental parameters monitored by the at least one environmental sensor; The data acquisition unit is configured to synchronously acquire the raw gas concentration data output by the gas sensor and the real-time environmental parameter data output by the at least one environmental sensor in real time. The data processing unit is configured as follows: Receive the raw gas concentration data and the real-time environmental parameter data; The dynamic compensation model in the storage unit is invoked, and the real-time environmental parameter data is transmitted to the dynamic compensation model as an input variable. Based on the dynamic compensation model, a dynamic compensation value corresponding to the real-time environmental parameter data is calculated. The original gas concentration data is corrected using the dynamic compensation value to generate calibrated final gas concentration data.

[0006] Furthermore, the at least one environmental sensor includes at least one of a temperature sensor, a humidity sensor, and an atmospheric pressure sensor; the real-time environmental parameter data includes at least one of real-time temperature data, real-time humidity data, and real-time atmospheric pressure data.

[0007] Furthermore, the data processing unit is further configured to construct a multidimensional environmental parameter vector from the real-time temperature data, real-time humidity data, and real-time atmospheric pressure data, and use the multidimensional environmental parameter vector as the input variable of the dynamic compensation model.

[0008] Furthermore, the dynamic compensation model is a regression model built based on machine learning algorithms, including neural network models, support vector machine models, gradient boosting decision tree models, or random forest models.

[0009] Furthermore, the system also includes a model training module, which is configured to: Under multiple preset calibration environment conditions, acquire multiple sets of historical environmental parameter data collected by the data acquisition unit, each corresponding to one of the multiple calibration environment conditions; Under the multiple calibration environmental conditions, multiple sets of original calibration concentration data, which are measured by the gas sensor and correspond to the multiple sets of historical environmental parameter data, are acquired simultaneously, along with the corresponding standard gas concentration benchmark values. Based on the multiple sets of original calibration concentration data and the standard gas concentration benchmark value, multiple sets of historical response deviation data are calculated; The dynamic compensation model is trained offline or updated online using the multiple sets of historical environmental parameter data as training input and the multiple sets of historical response deviation data as training output.

[0010] This invention further proposes a method for security monitoring of natural gas pipeline stations, comprising the following steps: A. Real-time synchronous acquisition of raw gas concentration data output by gas sensors and real-time environmental parameter data output by at least one environmental sensor; B. Provide a pre-trained dynamic compensation model, which constructs a nonlinear mapping relationship between the response deviation of the gas sensor and the environmental parameters monitored by the at least one environmental sensor; C. Input the real-time environmental parameter data into the dynamic compensation model, and obtain the dynamic compensation value corresponding to the real-time environmental parameter data through the calculation of the dynamic compensation model; D. Based on the dynamic compensation value, perform a calibration operation on the original gas concentration data to generate and output the compensated final gas concentration data.

[0011] Furthermore, the real-time environmental parameter data includes at least one or any combination of real-time temperature data obtained from a temperature sensor, real-time humidity data obtained from a humidity sensor, and real-time atmospheric pressure data obtained from an atmospheric pressure sensor.

[0012] Further, step C specifically includes: The real-time temperature data, real-time humidity data, and real-time atmospheric pressure data are combined into a multi-dimensional environmental feature vector. The multidimensional environmental feature vector is used as the overall input of the dynamic compensation model to calculate the dynamic compensation value.

[0013] Furthermore, the dynamic compensation model is a model trained by a machine learning algorithm, which is selected from neural networks, support vector machines, gradient boosting decision trees, or random forest algorithms.

[0014] Furthermore, prior to step A, a training step for generating the dynamic compensation model is included, the training step comprising: Under various preset combinations of calibration environment parameters, the raw calibration concentration data output by the gas sensor are acquired respectively, and the corresponding standard concentration reference values ​​are recorded. Calculate the response deviation data of the original calibration concentration data relative to the standard concentration reference value under various combinations of the calibration environment parameters; The calibration environment parameter combination is used as the input sample, and the response deviation data is used as the output label to perform supervised learning training on the dynamic compensation model until the model converges.

[0015] Beneficial effects This invention, by setting up a storage unit containing a dynamic compensation model, combines the raw gas concentration data and environmental parameter data collected in real time by the data acquisition unit. The data processing unit calls the model to calculate the dynamic compensation value and correct the original data, enabling real-time dynamic adaptation to environmental changes and solving the problem that fixed compensation cannot cope with nonlinear drift in complex environments. It constructs multiple environmental parameters into a multi-dimensional vector input model, establishes parameter correlations, and overcomes the limitations of lack of multi-sensor collaboration. The model can be trained offline or updated online, eliminating the need for manual periodic shutdown calibration, avoiding monitoring blind spots, and comprehensively improving the accuracy, real-time performance, and continuity of security monitoring at natural gas pipeline stations. Attached Figure Description

[0016] Figure 1 This is a system structure diagram of the present invention; Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 This invention provides a natural gas pipeline station security monitoring system, including a gas sensor, at least one environmental sensor, a data acquisition unit, and a data processing unit. The system also includes: The storage unit contains a dynamic compensation model that is pre-stored. The dynamic compensation model is used to characterize the nonlinear mapping relationship between the response deviation of the gas sensor and the environmental parameters monitored by at least one environmental sensor. The data acquisition unit is configured to synchronously acquire raw gas concentration data output by a gas sensor and real-time environmental parameter data output by at least one environmental sensor in real time. The data processing unit is configured as follows: Receive raw gas concentration data and real-time environmental parameter data; The dynamic compensation model in the storage unit is invoked, and real-time environmental parameter data is transmitted to the dynamic compensation model as input variables. Based on the dynamic compensation model, dynamic compensation values ​​corresponding to real-time environmental parameter data are calculated. The original gas concentration data is corrected using dynamic compensation values ​​to generate calibrated final gas concentration data.

[0019] Specifically, in this natural gas pipeline station security monitoring system, infrared gas sensors can be used to detect the concentration of the main components in natural gas; environmental sensors can be installed in different areas of the station to ensure comprehensive acquisition of environmental parameters. The data acquisition unit uses a high-speed data acquisition card with a sampling frequency of 10Hz to achieve real-time synchronous acquisition of raw gas concentration data and real-time environmental parameter data. The storage unit uses a large-capacity solid-state drive to stably store the dynamic compensation model and various acquired data. The data processing unit uses a high-performance industrial control computer. Upon receiving the raw gas concentration data and real-time environmental parameter data, it immediately retrieves the dynamic compensation model from the storage unit and inputs the real-time environmental parameter data into the model. For example, when the real-time environmental parameters are a temperature of 25℃ and a humidity of 60%, the model calculates the corresponding dynamic compensation value, which is then used to correct the raw gas concentration data. The calibrated final gas concentration data more accurately reflects the actual gas concentration, effectively reducing detection errors caused by environmental factors and improving the reliability of station security monitoring.

[0020] Furthermore, at least one environmental sensor includes at least one of a temperature sensor, a humidity sensor, and an atmospheric pressure sensor; the real-time environmental parameter data includes at least one of real-time temperature data, real-time humidity data, and real-time atmospheric pressure data.

[0021] Specifically, the temperature sensor can be a platinum resistance temperature sensor, with a measurement range of -40℃ to 120℃, which can accurately acquire real-time temperature data of the station; the humidity sensor is a capacitive humidity sensor, with a measurement range of 0 to 100%RH, which can accurately collect real-time humidity data; the atmospheric pressure sensor is a piezoresistive atmospheric pressure sensor, with a measurement range of 30kPa to 110kPa, used to acquire real-time atmospheric pressure data. These sensors can be installed near the gas sensors to ensure that the measured environmental parameters are consistent with the environment in which the gas sensors are located. For example, in the valve group area of ​​the station, temperature, humidity, and atmospheric pressure sensors are installed simultaneously, and they output data such as 28℃, 55%RH, and 101kPa in real time, respectively. By acquiring these multiple real-time environmental parameter data, more comprehensive input information can be provided for the dynamic compensation model, which helps to improve the calculation accuracy of dynamic compensation values, thereby improving the accuracy of the final gas concentration data.

[0022] Furthermore, the data processing unit is further configured to construct a multidimensional environmental parameter vector from real-time temperature data, real-time humidity data, and real-time atmospheric pressure data, and use the multidimensional environmental parameter vector as the input variable of the dynamic compensation model.

[0023] Specifically, after receiving real-time temperature, humidity, and atmospheric pressure data, the data processing unit combines them into a multi-dimensional environmental parameter vector according to a preset order. For example, when the real-time temperature is 30℃, the real-time humidity is 65%RH, and the real-time atmospheric pressure is 100kPa, the constructed multi-dimensional environmental parameter vector is [30, 65, 100]. This multi-dimensional environmental parameter vector is used as the input variable of the dynamic compensation model, enabling the model to comprehensively consider the synergistic effects of multiple environmental factors. Compared to inputting a single environmental parameter, this method can more comprehensively reflect the impact of the environment on the gas sensor, thus making the compensation value calculated by the dynamic compensation model more consistent with the actual situation and improving the correction effect on the original gas concentration data.

[0024] Furthermore, the dynamic compensation model is a regression model built based on machine learning algorithms, including neural network models, support vector machine models, gradient boosting decision tree models, or random forest models.

[0025] Specifically, a dynamic compensation model is constructed using a neural network model as an example. This neural network model employs a three-layer structure. The number of neurons in the input layer corresponds to the dimension of the environmental parameters; for example, when the environmental parameters are temperature, humidity, and atmospheric pressure, the number of neurons in the input layer is 3. The hidden layer has 10 neurons, using the ReLU activation function. The output layer has 1 neuron, corresponding to the dynamic compensation value. The model's loss function uses mean squared error, and the formula is: ,in For the sample size, This is the actual response deviation value. The values ​​represent the predicted response bias (i.e., dynamic compensation values), all in %LEL (lower explosion limit percentage). The model is trained using a backpropagation algorithm, continuously adjusting the weights and biases between neurons in each layer until the loss function converges to a preset threshold. Employing this neural network model as a dynamic compensation model effectively fits the complex nonlinear relationship between environmental parameters and the gas sensor response bias. Compared to traditional fixed compensation formulas, it can calculate dynamic compensation values ​​more accurately, improving the accuracy of gas concentration detection.

[0026] Furthermore, the system also includes a model training module, which is configured as follows: Under multiple preset calibration environment conditions, acquire multiple sets of historical environmental parameter data collected by the data acquisition unit, each corresponding to a different calibration environment condition; Under multiple calibration environmental conditions, multiple sets of original calibration concentration data, which are measured by gas sensors and correspond to multiple sets of historical environmental parameter data, as well as the corresponding standard gas concentration reference values, are acquired simultaneously. Based on multiple sets of original calibration concentration data and standard gas concentration benchmark values, multiple sets of historical response deviation data were calculated. The dynamic compensation model is trained offline or updated online by using multiple sets of historical environmental parameter data as training input and multiple sets of historical response deviation data as training output.

[0027] Specifically, the model training module can be integrated into the data processing unit. Multiple preset calibration environmental conditions can be implemented through an environmental simulation chamber. For example, the temperature range can be set to -20℃ to 50℃, the humidity range to 20%RH to 90%RH, and the atmospheric pressure range to 80kPa to 110kPa, within which 100 different combinations of calibration environmental conditions can be defined. Under each calibration environmental condition, the data acquisition unit collects corresponding historical environmental parameter data, while the gas sensor measures the original calibration concentration data. A standard gas analyzer is then used to obtain the standard gas concentration baseline value for that environment. The historical response deviation data is calculated as follows: Historical response deviation data = (Original calibration concentration data - Standard gas concentration baseline value) / Standard gas concentration baseline value × 100%, in %. For example, when the original calibration concentration data is 5% LEL and the standard gas concentration baseline value is 4% LEL, the historical response deviation data is 25%. These historical environmental parameter data and historical response deviation data are used as training samples to train the dynamic compensation model offline. After the model is put into use, newly collected data that meets the calibration conditions can be used periodically as incremental samples to update the model online. The dynamic compensation model trained and updated in this way can better adapt to different environmental conditions, improve the model's generalization ability, and thus ensure accurate prediction of gas sensor response deviation.

[0028] Please see Figure 2 The present invention further proposes a method for security monitoring of natural gas pipeline stations, characterized by comprising the following steps: A. Real-time synchronous acquisition of raw gas concentration data output by gas sensors and real-time environmental parameter data output by at least one environmental sensor; B. Provide a pre-trained dynamic compensation model that constructs a nonlinear mapping relationship between the response deviation of the gas sensor and the environmental parameters monitored by at least one environmental sensor; C. Input the real-time environmental parameter data into the dynamic compensation model, and obtain the dynamic compensation value corresponding to the real-time environmental parameter data through the calculation of the dynamic compensation model; D. Based on the dynamic compensation value, perform a calibration operation on the original gas concentration data to generate and output the compensated final gas concentration data.

[0029] Specifically, in step A, a synchronous acquisition controller is used to control the gas sensor and environmental sensor, with an acquisition cycle set to 1 second to ensure that the original gas concentration data and real-time environmental parameter data are consistent in time. The dynamic compensation model in step B is trained using a large number of samples, for example, using environmental parameters from the station over the past year and corresponding gas sensor response deviation data. In step C, after inputting the real-time environmental parameter data into the dynamic compensation model, the model quickly obtains the dynamic compensation value through its internal computational mechanism, such as forward propagation calculation of a neural network. In step D, the calibration operation uses the original gas concentration data plus the dynamic compensation value. For example, if the original gas concentration data is 6% LEL and the dynamic compensation value is -0.5% LEL, then the final compensated gas concentration data is 5.5% LEL, which is then output to the station's monitoring center via the data transmission module. This method can calibrate the gas concentration data in real time according to environmental changes, avoiding the limitations of fixed compensation methods and improving the real-time performance and accuracy of security monitoring.

[0030] Furthermore, the real-time environmental parameter data includes at least one or any combination of real-time temperature data obtained from a temperature sensor, real-time humidity data obtained from a humidity sensor, and real-time atmospheric pressure data obtained from an atmospheric pressure sensor.

[0031] Specifically, the temperature sensor collects real-time temperature data every second, in degrees Celsius (e.g., 26°C); the humidity sensor also collects data every second, in terms of relative humidity percentage (e.g., 58%); and the atmospheric pressure sensor collects data every second, in terms of kilopascals (kPa) (e.g., 102 kPa). In practical applications, appropriate parameter combinations can be selected based on the environmental characteristics of the station. For example, in stations with minimal humidity variation, only real-time temperature and atmospheric pressure data can be used for compensation calculations. By acquiring these different combinations of real-time environmental parameters, the system can flexibly adapt to different station conditions, ensuring compensation effectiveness while reducing unnecessary data processing and improving the operational efficiency of the monitoring method.

[0032] Furthermore, step C specifically includes: Real-time temperature data, real-time humidity data, and real-time atmospheric pressure data are combined into a multi-dimensional environmental feature vector. The multidimensional environmental feature vector is used as the overall input to the dynamic compensation model to calculate the dynamic compensation value.

[0033] Specifically, in step C, real-time temperature data, real-time humidity data, and real-time atmospheric pressure data are combined in the order of temperature, humidity, and atmospheric pressure to form a multi-dimensional environmental feature vector. For example, when the real-time temperature is 29℃, the real-time humidity is 62%, and the real-time atmospheric pressure is 100kPa, the combined multi-dimensional environmental feature vector is [29, 62, 100]. This vector is then input into the dynamic compensation model. The model comprehensively processes each element in the vector, such as mapping it to a high-dimensional space using the kernel function in the support vector machine model, to obtain the dynamic compensation value. This method of using the multi-dimensional environmental feature vector as a whole input fully considers the correlation between various environmental parameters, enabling the dynamic compensation model to more comprehensively capture the influence of the environment on the gas sensor, thereby calculating a more accurate dynamic compensation value and improving the calibration effect on the original gas concentration data.

[0034] Furthermore, the dynamic compensation model is a model trained by a machine learning algorithm, which is selected from neural networks, support vector machines, gradient boosting decision trees, or random forest algorithms.

[0035] Specifically, a dynamic compensation model is constructed using the Support Vector Machine (SVM) algorithm as an example. The goal of this model is to find an optimal hyperplane that maximizes the distance from environmental parameter data to the hyperplane, thereby achieving accurate prediction of response bias. Its mathematical model can be expressed as: ,in The input environment parameter vector, For Lagrange multipliers, This represents the response bias value for the training samples (in %LEL). For the kernel function, the radial basis function is chosen here. , For kernel function parameters, This is the bias term. It is determined through learning from the training samples. , and The dynamic compensation model trained using the support vector machine algorithm exhibits good generalization ability when processing small sample data. It can effectively cope with various complex environmental conditions that may occur in the station, accurately calculate the dynamic compensation value, and improve the accuracy of gas concentration monitoring.

[0036] Furthermore, prior to step A, a training step for generating the dynamic compensation model is included, which includes: Under various preset combinations of calibration environment parameters, the raw calibration concentration data output by the gas sensor were acquired, and the corresponding standard concentration reference values ​​were recorded. Calculate the response deviation data of the original calibration concentration data relative to the standard concentration reference value under various combinations of calibration environmental parameters; The dynamic compensation model is trained under supervised learning by using the combination of calibrated environmental parameters as input samples and the response deviation data as output labels until the model converges.

[0037] Specifically, in the training process, various preset combinations of calibration environmental parameters can be set using environmental control equipment, such as 20 different combinations of temperature, humidity, and atmospheric pressure. Under each combination, the gas sensor operates continuously for 10 minutes, acquiring raw calibration concentration data every 10 seconds, and recording the corresponding standard concentration baseline value using a high-precision standard gas detection device. The formula for calculating the response deviation data is: Response Deviation Data = (Raw Calibration Concentration Data - Standard Concentration Baseline Value), with units consistent with both the raw calibration concentration data and the standard concentration baseline value, all expressed as %LEL. For example, if the raw calibration concentration data is 8%LEL and the standard concentration baseline value is 7.5%LEL, then the response deviation data is 0.5%LEL. The calibration environmental parameter combinations and the corresponding response deviation data form a training dataset, which is used to supervise the training of the dynamic compensation model. For example, gradient descent algorithms are used to adjust the model parameters. When the model's prediction error is below 0.1%LEL, the model is considered converged. The dynamically compensated model generated through this training accurately reflects the relationship between environmental parameters and the gas sensor's response deviation, providing a reliable basis for subsequent gas concentration data calibration and ensuring the effectiveness of the monitoring method.

[0038] In summary, this invention, by setting up a storage unit containing a dynamic compensation model, combined with the real-time synchronous acquisition of raw gas concentration data and environmental parameter data by the data acquisition unit, and the data processing unit calling the model to calculate dynamic compensation values ​​and correct the original data, can dynamically adapt to environmental changes in real time, solving the problem that fixed compensation cannot cope with nonlinear drift in complex environments. It constructs a multi-dimensional vector input model for various environmental parameters, establishing parameter correlations and overcoming the limitations of lack of multi-sensor collaboration. The model can be trained offline or updated online, eliminating the need for regular manual shutdowns for calibration, avoiding monitoring blind spots, and comprehensively improving the accuracy, real-time performance, and continuity of security monitoring at natural gas pipeline stations.

[0039] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0040] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A natural gas pipeline station security monitoring system, comprising a gas sensor, at least one environmental sensor, a data acquisition unit, and a data processing unit, characterized in that, The system also includes: A storage unit, wherein a dynamic compensation model is pre-stored in the storage unit, the dynamic compensation model being used to characterize the nonlinear mapping relationship between the response deviation of the gas sensor and the environmental parameters monitored by the at least one environmental sensor; The data acquisition unit is configured to synchronously acquire the raw gas concentration data output by the gas sensor and the real-time environmental parameter data output by the at least one environmental sensor in real time. The data processing unit is configured as follows: Receive the raw gas concentration data and the real-time environmental parameter data; The dynamic compensation model in the storage unit is invoked, and the real-time environmental parameter data is transmitted to the dynamic compensation model as an input variable. Based on the dynamic compensation model, a dynamic compensation value corresponding to the real-time environmental parameter data is calculated. The original gas concentration data is corrected using the dynamic compensation value to generate calibrated final gas concentration data.

2. The natural gas pipeline station security monitoring system according to claim 1, characterized in that, The at least one environmental sensor includes at least one of a temperature sensor, a humidity sensor, and an atmospheric pressure sensor; the real-time environmental parameter data includes at least one of real-time temperature data, real-time humidity data, and real-time atmospheric pressure data.

3. The natural gas pipeline station security monitoring system according to claim 2, characterized in that, The data processing unit is further configured to construct a multidimensional environmental parameter vector from the real-time temperature data, real-time humidity data, and real-time atmospheric pressure data, and to use the multidimensional environmental parameter vector as the input variable of the dynamic compensation model.

4. The natural gas pipeline station security monitoring system according to claim 1, characterized in that, The dynamic compensation model is a regression model built based on machine learning algorithms, including neural network models, support vector machine models, gradient boosting decision tree models, or random forest models.

5. A natural gas pipeline station security monitoring system according to claim 1, characterized in that, The system also includes a model training module, which is configured as follows: Under multiple preset calibration environment conditions, acquire multiple sets of historical environmental parameter data collected by the data acquisition unit, each corresponding to one of the multiple calibration environment conditions; Under the multiple calibration environmental conditions, multiple sets of original calibration concentration data, which are measured by the gas sensor and correspond to the multiple sets of historical environmental parameter data, are acquired simultaneously, along with the corresponding standard gas concentration benchmark values. Based on the multiple sets of original calibration concentration data and the standard gas concentration benchmark value, multiple sets of historical response deviation data are calculated; The dynamic compensation model is trained offline or updated online using the multiple sets of historical environmental parameter data as training input and the multiple sets of historical response deviation data as training output.

6. A method for security monitoring of a natural gas pipeline station, used to implement the security monitoring system for a natural gas pipeline station as described in any one of claims 1-5, characterized in that, Includes the following steps: A. Real-time synchronous acquisition of raw gas concentration data output by gas sensors and real-time environmental parameter data output by at least one environmental sensor; B. Provide a pre-trained dynamic compensation model, which constructs a nonlinear mapping relationship between the response deviation of the gas sensor and the environmental parameters monitored by the at least one environmental sensor; C. Input the real-time environmental parameter data into the dynamic compensation model, and obtain the dynamic compensation value corresponding to the real-time environmental parameter data through the calculation of the dynamic compensation model; D. Based on the dynamic compensation value, perform a calibration operation on the original gas concentration data to generate and output the compensated final gas concentration data.

7. A method for security monitoring of a natural gas pipeline station according to claim 6, characterized in that, The real-time environmental parameter data includes at least one or any combination of real-time temperature data obtained from a temperature sensor, real-time humidity data obtained from a humidity sensor, and real-time atmospheric pressure data obtained from an atmospheric pressure sensor.

8. A method for security monitoring of a natural gas pipeline station according to claim 7, characterized in that, Step C specifically includes: The real-time temperature data, real-time humidity data, and real-time atmospheric pressure data are combined into a multi-dimensional environmental feature vector. The multidimensional environmental feature vector is used as the overall input of the dynamic compensation model to calculate the dynamic compensation value.

9. A method for security monitoring of a natural gas pipeline station according to claim 6, characterized in that, The dynamic compensation model is a model trained by a machine learning algorithm, which is selected from neural networks, support vector machines, gradient boosting decision trees, or random forest algorithms.

10. A method for security monitoring of a natural gas pipeline station according to claim 6, characterized in that, Prior to step A, a training step for generating the dynamic compensation model is included, the training step comprising: Under various preset combinations of calibration environment parameters, the raw calibration concentration data output by the gas sensor are acquired respectively, and the corresponding standard concentration reference values ​​are recorded. Calculate the response deviation data of the original calibration concentration data relative to the standard concentration reference value under various combinations of the calibration environment parameters; The calibration environment parameter combination is used as the input sample, and the response deviation data is used as the output label to perform supervised learning training on the dynamic compensation model until the model converges.

Images