Natural gas energy metering and calorific value fluctuation early warning method and system

Abstract

The invention discloses a natural gas energy metering and calorific value fluctuation early warning method and system, and belongs to the technical field of natural gas intelligent monitoring, and the method comprises the steps: collecting temperature parameters, pressure parameters, flow parameters and gas component parameters of nodes in a natural gas pipeline network, and generating multi-source real-time data; performing coupling calculation on the multi-source real-time data to generate an instantaneous energy value; historical calorific value data and pipe network topological structure data are obtained, spatial-temporal feature extraction is carried out in combination with the instantaneous energy value, and calorific value fluctuation features are generated; performing dynamic mode recognition on the heat value fluctuation characteristics to generate a heat value fluctuation early warning signal; and performing multi-dimensional risk analysis on the calorific value fluctuation early warning signal to generate a calorific value risk grade regulation and control instruction. According to the method, multi-source data fusion is adopted, a dynamic mode recognition and risk quantification mapping mechanism is combined, and accurate metering of the natural gas energy value and early warning regulation and control of the heat value fluctuation risk level can be achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of natural gas intelligent monitoring, in particular to a natural gas energy metering and heat value fluctuation early warning method and system. BACKGROUND

[0002] Natural gas, as a major clean energy, is increasingly widely used in industrial production and civilian life. Unlike traditional volume metering, the core of the commercial value of natural gas lies in the energy, i.e. the heat value, it contains. Therefore, accurate energy metering of natural gas and real-time monitoring of the stability of its heat value are crucial for ensuring the fairness of energy trade, guiding the scientific scheduling of pipe networks and ensuring the safe and efficient operation of energy-using equipment. Monitoring and analyzing the heat value of natural gas is a key link in the quality control of natural gas.

[0003] In the prior art, in order to realize online monitoring of the heat value of natural gas, an online gas chromatograph is usually installed at a key node of a pipe network to periodically analyze the gas component parameters of natural gas and calculate the real-time heat value accordingly. At the same time, the flow parameter, pressure parameter and temperature parameter of natural gas are measured by a flow meter, a pressure transmitter and a temperature transmitter respectively, and the standard volume flow is obtained after independent working condition correction calculation. Finally, the calculated real-time heat value is multiplied by the standard volume flow to obtain the energy of natural gas. For the early warning of heat value fluctuation, a fixed high and low threshold value is generally set for the calculated heat value, and an alarm is triggered once the value exceeds the range.

[0004] However, the above prior art solution has obvious technical defects. First, the analysis of the online gas chromatograph has inherent time delay, which causes the calculated heat value to always lag behind the actual change of the natural gas components, making it difficult to capture instantaneous and severe heat value fluctuations. Second, the working condition correction of the volume flow and the component calculation of the heat value are two independent processes, and the separation of data sources and calculation logic can easily introduce and accumulate errors when the working condition of the pipe network changes rapidly. In addition, the early warning method based on fixed threshold value is too rigid and cannot adapt to the normal fluctuation range of the pipe network under different seasons and different loads, often leading to false alarms or insensitivity to real small abnormalities, and the effectiveness and reliability of the early warning are not high. SUMMARY

[0005] To solve the above problems, the present application provides a natural gas energy metering and heat value fluctuation early warning method and system, which adopts multi-source data fusion, combines dynamic pattern recognition and risk quantification mapping mechanism, and can realize accurate metering of the energy value of natural gas and early warning and regulation of the risk level of heat value fluctuation.

[0006] The above object can be achieved by the following solution: The method and system for natural gas energy metering and heat value fluctuation early warning comprises the following steps: collecting temperature parameters, pressure parameters, flow parameters and gas component parameters of nodes in a natural gas pipeline network to generate multi-source real-time data; performing coupling calculation on the multi-source real-time data to generate instantaneous energy values; obtaining historical heat value data and pipeline network topology data, combining the instantaneous energy values to perform space-time feature extraction to generate heat value fluctuation features; performing dynamic mode recognition on the heat value fluctuation features to generate heat value fluctuation early warning signals; and performing multi-dimensional risk analysis on the heat value fluctuation early warning signals to generate heat value risk level regulation instructions.

[0007] Optionally, the generating multi-source real-time data comprises: obtaining temperature parameters, pressure parameters, flow parameters and gas component parameters; appending time stamps to the temperature parameters, pressure parameters, flow parameters and gas component parameters and performing time alignment processing to generate multi-source real-time data.

[0008] Optionally, the coupling calculation on the multi-source real-time data to generate instantaneous energy values comprises: calculating a gas compression factor according to the temperature parameters, pressure parameters and gas component parameters in the multi-source real-time data; performing compensation calculation on the flow parameters in the multi-source real-time data by using the gas compression factor to obtain volume flow under standard conditions; calculating a current heat value according to the gas component parameters; and multiplying the volume flow under standard conditions and the current heat value to output an instantaneous energy value.

[0009] Optionally, the space-time feature extraction comprises: performing short-term fluctuation analysis and long-term trend analysis based on the instantaneous energy values and the historical heat value data to generate time features; performing spatial correlation analysis on the instantaneous energy values based on the pipeline network topology data to generate space features; and fusing the time features and the space features to generate heat value fluctuation features.

[0010] Optionally, the dynamic mode recognition on the heat value fluctuation features to generate heat value fluctuation early warning signals comprises: extracting a space-time correlation matrix of the heat value fluctuation features to obtain a multi-dimensional abnormal mode vector; performing dynamic threshold matching and mode evolution prediction on the multi-dimensional abnormal mode vector to generate heat value fluctuation early warning signals.

[0011] Optionally, the fusing the time features and the space features to generate heat value fluctuation features comprises: performing space-time graph convolution fusion on the time features and the space features to obtain a space-time correlation matrix; extracting a multi-scale mode vector from the space-time correlation matrix to generate heat value fluctuation features.

[0012] Optionally, the multi-dimensional risk analysis on the heat value fluctuation early warning signal to generate a heat value risk level regulation instruction comprises: analyzing abnormal mode information contained in the heat value fluctuation early warning signal to obtain a multi-dimensional heat value risk feature vector; and performing level quantization mapping on the multi-dimensional heat value risk feature vector to generate the heat value risk level regulation instruction.

[0013] Optionally, the level quantization mapping on the multi-dimensional heat value risk feature vector to generate a heat value risk level regulation instruction comprises: performing multi-scale feature fusion and entropy weight quantization on the multi-dimensional heat value risk feature vector to obtain a numerical risk assessment coefficient; and performing dynamic threshold interval matching and regulation strategy mapping on the risk assessment coefficient to generate the heat value risk level regulation instruction.

[0014] Optionally, the dynamic threshold matching and mode evolution prediction on the multi-dimensional abnormal mode vector to generate a heat value fluctuation early warning signal comprises: obtaining a multi-scale abnormality degree matching set by using the multi-dimensional abnormal mode vector; and performing spatio-temporal convolution evolution prediction on the multi-scale abnormality degree matching set to generate the heat value fluctuation early warning signal.

[0015] Based on the same inventive concept, the application further provides a natural gas energy metering and heat value fluctuation early warning system, which comprises: a data acquisition module for acquiring temperature parameters, pressure parameters, flow parameters and gas component parameters of nodes in a natural gas pipeline network to generate multi-source real-time data; an energy calculation module for coupling calculation on the multi-source real-time data to generate an instantaneous energy value; a feature extraction module for acquiring historical heat value data and pipeline network topology structure data, and performing spatio-temporal feature extraction in combination with the instantaneous energy value to generate heat value fluctuation features; an early warning module for performing dynamic mode recognition on the heat value fluctuation features to generate a heat value fluctuation early warning signal; and a risk analysis module for performing multi-dimensional risk analysis on the heat value fluctuation early warning signal to generate a heat value risk level regulation instruction.

[0016] Compared with the prior art, the application has the following advantages: The application significantly improves the accuracy of natural gas energy metering by constructing a dynamic feedback closed loop of metering and early warning. The temperature-pressure-component linkage correction model adopted by the application couples the traditional separate volume compensation and heat value calculation processes, synchronously processes multi-source real-time data, effectively avoids error accumulation caused by data lag and step-by-step calculation, especially under complex working conditions with frequent changes in gas source components, and ensures high fidelity of the instantaneous energy value, thereby providing a reliable data basis for fair trade settlement and fine management of the pipeline network.

[0017] The present application greatly enhances the reliability and intelligent level of heat value fluctuation early warning by introducing spatio-temporal feature extraction and dynamic threshold self-learning mechanism. The method not only analyzes the time trend of fluctuation, but also combines the spatial correlation diagnosis of the pipe network topology, and can accurately distinguish between systematic pipe network anomalies and local instrument interference. At the same time, its early warning threshold can be self-adjusted according to historical feedback and real-time working conditions, effectively adapting to different operating scenarios, significantly reducing the false positive rate and false negative rate, and making the early warning system more sensitive and accurate.

[0018] The present application realizes the robustness and adaptability of the overall performance of the system through the bidirectional cooperation between the metering module and the early warning module. The early warning system not only can find problems, but also can feed back the abnormal pattern information diagnosed to the energy metering module to dynamically adjust the core model parameters, so that the metering system can actively adapt to the changes of gas source quality. This self-correction and optimization mechanism forms an intelligent whole with synergistic effect, so that the system can still maintain efficient and stable operation performance when facing various unknown and unexpected conditions.

[0019] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application can be realized and obtained by the structure indicated in the specification, claims and drawings. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0021] Figure 1 is a flowchart of a natural gas energy metering and heat value fluctuation early warning method according to an embodiment of the present application.

[0022] Figure 2 is a heat value fluctuation time feature analysis diagram according to an embodiment of the present application.

[0023] Figure 3 is a dynamic pattern recognition diagram according to an embodiment of the present application.

[0024] Figure 4 is a dynamic threshold matching diagram according to an embodiment of the present application.

[0025] Figure 5 is a multi-dimensional risk analysis matrix diagram according to an embodiment of the present application.

[0026] Figure 6is a structural schematic diagram of a natural gas energy metering and heat value fluctuation early warning system according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below in connection with the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0028] Reference Figure 1 An embodiment of the present application proposes a natural gas energy metering and heat value fluctuation early warning method, which adopts multi-source data fusion, combines dynamic pattern recognition and risk quantification mapping mechanism, and can realize accurate metering of natural gas energy value and early warning regulation of heat value fluctuation risk level.

[0029] The method according to the embodiment specifically includes: Collecting temperature parameters, pressure parameters, flow parameters and gas component parameters of nodes in a natural gas pipeline network to generate multi-source real-time data; Performing coupling calculation on the multi-source real-time data to generate instantaneous energy value; Obtaining historical heat value data and pipeline network topology structure data, combining the instantaneous energy value to perform spatio-temporal feature extraction and generate heat value fluctuation feature; Performing dynamic pattern recognition on the heat value fluctuation feature to generate heat value fluctuation early warning signal; Performing multi-dimensional risk analysis on the heat value fluctuation early warning signal to generate heat value risk level regulation instruction.

[0030] Specifically, firstly, the temperature, pressure, flow and gas component and other multi-source heterogeneous data of nodes in the natural gas pipeline network are comprehensively collected, and time synchronization is performed, thereby providing a unified data basis for subsequent calculation. Then, the real-time data are coupled and calculated by using a physical model, the volume flow affected by the working condition is accurately converted into instantaneous energy value under standard state, and the essence of metering is upgraded. On this basis, the method further fuses historical data and pipeline network topology structure, and through spatio-temporal feature extraction technology, the fluctuation law of the instantaneous energy value is deeply mined from two dimensions of time and space, and heat value fluctuation features capable of fully representing the dynamic characteristics are generated. Subsequently, dynamic pattern recognition technology is used to analyze these complex features, to identify potential abnormal fluctuation patterns and generate early warning signals. Finally, the early warning signals are subjected to multi-dimensional risk analysis, the risk level is quantified, and the preset regulation strategy is automatically mapped, and finally the heat value risk level regulation instruction directly executable is generated, thereby forming a complete closed-loop management process.

[0031] Optionally, the generating the multi-source real-time data comprises: acquiring temperature parameters, pressure parameters, flow parameters and gas component parameters; appending time stamps to the temperature parameters, pressure parameters, flow parameters and gas component parameters, and performing time alignment processing to generate the multi-source real-time data.

[0032] Specifically, first, corresponding sensing and analyzing devices are deployed at key nodes of the natural gas pipeline network, such as stations, gate stations and main distribution points. Temperature sensors are used to acquire temperature parameters reflecting the state of the gas, pressure sensors are used to acquire pressure parameters, and metering devices such as ultrasonic flow meters or orifice flow meters are used to acquire flow parameters representing the gas delivery rate. Online gas chromatographs are used to analyze the component content of natural gas in real time, obtaining gas component parameters including the molar percentage of each component such as methane, ethane and propane. After data acquisition, a high-precision time stamp is appended to each independent data point uploaded from the sensors or analyzers. The time stamp is synchronized with the central server through the network time protocol, ensuring that the data of the entire pipeline network has a unified time reference. However, due to differences in sampling periods and data processing delays of different devices, for example, pressure parameters may be collected at a level of seconds, while gas component parameters may have an analysis period of several minutes, directly using these time-stamped data will result in the natural gas state described by each parameter not being truly synchronized at the same time. Therefore, time alignment processing must be performed. This processing sets a unified calculation time step and aligns all parameter sequences from different sources to this discrete time grid. For parameters with a sampling frequency higher than the calculation step, such as temperature and pressure, the average value or the last value in each step interval can be used for downsampling; for parameters with a sampling frequency lower than the calculation step, such as gas component parameters, forward filling or zero-order hold methods are used, that is, the last valid value of the previous period is continuously used until the new analysis result is available. After this processing, the original, time-discrete and asynchronous measurement values are integrated into a time series data set, where each time point corresponds to a complete and synchronized parameter set, which is the multi-source real-time data and lays the foundation for subsequent accurate coupling calculations.

[0033] Optionally, the coupling calculation on the multi-source real-time data to generate the instantaneous energy value comprises: calculating a gas compressibility factor according to the temperature parameters, pressure parameters and gas component parameters in the multi-source real-time data; performing compensation calculation on the flow parameters in the multi-source real-time data using the gas compressibility factor to obtain the volume flow under standard conditions; calculating a current calorific value according to the gas component parameters; The instantaneous energy value is output by multiplying the volumetric flow rate under the standard conditions by the current calorific value.

[0034] Specifically, firstly, based on the temperature, pressure, and gas composition parameters from the received multi-source real-time data, the gas compressibility factor is calculated according to the standardized thermodynamic equation of state. The gas compressibility factor quantifies the degree of deviation of real natural gas from the behavior of an ideal gas under specific temperature and pressure conditions. Secondly, using the gas compressibility factor calculated in the previous step, the flow rate parameters in the multi-source real-time data are compensated for, converting them from the actual operating conditions' volumetric flow rate to the recognized trade settlement benchmark, i.e., the volumetric flow rate under standard conditions. This conversion process follows the gas equation of state, and its calculation formula is as follows: , in, The volumetric flow rate under standard conditions is the result of this step's calculation. It is a flow parameter obtained from multi-source real-time data, that is, the volumetric flow rate measured under on-site operating conditions; and These are pressure and temperature parameters obtained from multi-source real-time data, respectively, where the temperature needs to be converted to an absolute temperature scale; and These are preset standard state pressure and temperature constants; That is, the result calculated in the previous step, under the operating conditions. The gas compressibility factor is then determined. Next, based on the measured gas component parameters from multi-source real-time data—namely, the mole fractions of each component gas in natural gas, such as methane and ethane—the current calorific value is calculated. The current calorific value, also known as the higher heating value, characterizes the heat released when a unit standard volume of natural gas is completely burned. The calculation method is based on the standard calorific values ​​of each pure component gas, obtained through a weighted average of the components. The calculation formula is: , in, The calculated current calorific value; This represents the mole fraction of the i-th component obtained from the gas composition parameters; It is the first The standard calorific value of the pure component is a physicochemical constant retrieved from a standard database. Finally, the volumetric flow rate under standard conditions calculated above is multiplied by the current calorific value to output the instantaneous energy value. This instantaneous energy value represents the total energy contained in the natural gas flowing through the metering point per unit time, and its calculation formula is as follows: , in, This is the final instantaneous energy value output; to compensate for the calculated volume flow at the standard state; to calculate the current heat value based on real-time components.

[0035] Optionally, the performing spatio-temporal feature extraction comprises: performing short-term fluctuation analysis and long-term trend analysis based on the instantaneous energy value and the historical heat value data to generate a time feature; performing spatial correlation analysis on the instantaneous energy value based on the pipe network topology data to generate a space feature; fusing the time feature and the space feature to generate a heat value fluctuation feature.

[0036] Specifically, first, in order to generate a time feature, short-term fluctuation analysis and long-term trend analysis are synchronously performed based on the instantaneous energy value calculated in the previous step and the stored historical heat value data. The heat value fluctuation time feature analysis is as shown in FIG. 2. The short-term fluctuation analysis focuses on the changes of the heat value in a short time, and the fluctuation intensity can be quantified by calculating the standard deviation of the instantaneous energy value in a sliding time window Figure 2 The calculation formula is: , In this formula, represents the instantaneous energy value at time point , which is generated by the energy calculation module; is the size of the set time window, for example, the last 60 data points; is the average value of the instantaneous energy values in these data points. The long-term trend analysis focuses on the macro trend of the heat value changes. The slope reflecting the long-term change trend can be obtained by linear regression modeling on the historical heat value data of a longer time span. Combining the standard deviation and the slope of the long-term change trend, the time feature reflecting the dynamic changes of the heat value is formed. At the same time, in order to generate a space feature, the pipe network topology data is used to perform spatial correlation analysis on the instantaneous energy values of each node in the pipe network. The pipe network topology data is a data describing the physical connection relationship between the monitoring nodes in the natural gas pipe network, which clearly shows which nodes are adjacent. The purpose of the spatial correlation analysis is to measure the correlation and difference degree of the energy value of any node and the energy values of its adjacent nodes. For example, for any node in the pipe network, its space feature can be obtained by calculating the weighted difference sum of its instantaneous energy value and the instantaneous energy values of all adjacent nodes, as shown in the formula: , wherein, and​ is the instantaneous energy value of node and any of its adjacent nodes ; is the set of all nodes directly connected to node defined in the pipe network topology data; is a weight coefficient, whose value can be set according to the physical parameters such as the length and diameter of the pipe between node and node , to represent the association strength between them. By performing this calculation for each node in the pipe network, a set of spatial features that fully reflect the distribution state of energy values in the entire pipe network space can be obtained. Finally, the time features generated in the foregoing are fused with the spatial features to generate the final heat value fluctuation features. This fusion process is realized through a spatio-temporal graph convolution technique. The specific operation is to regard the node data carrying the time features as signals defined on the pipe network topology graph, and then perform operations using a graph convolution network. The graph convolution operation can effectively aggregate the feature information of each node itself and its neighbor nodes, thereby capturing the dependence relationship in the spatial dimension. By applying this operation on the time series, the model can simultaneously learn the distribution patterns of the features in the space and the evolution law of these patterns over time. The output of the operation is a spatio-temporal association matrix, from which multi-scale pattern vectors are further extracted, and these vectors constitute the heat value fluctuation features that can comprehensively and deeply represent the heat value spatio-temporal dynamics.

[0037] Optionally, the dynamic pattern recognition on the heat value fluctuation features to generate a heat value fluctuation early warning signal comprises: extracting the spatio-temporal association matrix of the heat value fluctuation features to obtain a multi-dimensional abnormal pattern vector; performing dynamic threshold matching and pattern evolution prediction on the multi-dimensional abnormal pattern vector to generate a heat value fluctuation early warning signal.

[0038] Specifically, the dynamic pattern recognition is as shown in Figure 3 , which first needs to identify and extract the multi-dimensional abnormal pattern vector in the current state from the heat value fluctuation features in the form of a spatio-temporal association matrix generated in the previous link. This process is realized by comparing the current heat value fluctuation feature vector with a reference pattern vector representing the normal operation state. The reference pattern vector is a standard pattern obtained by learning a large amount of historical normal operation data. The deviation between the two can constitute the multi-dimensional abnormal pattern vector , and its calculation can be simplified as , wherein is a vector obtained in real time from the feature extraction module, is a reference vector stored in advance, Each dimension of quantifies the degree of deviation of the current heat value fluctuation state from the normal state in a certain aspect. Next, the method will perform dynamic threshold matching and pattern evolution prediction on the multi-dimensional anomaly pattern vector, as shown in the dynamic threshold matching. Figure 4 This process does not use a single fixed threshold, but uses a set of multi-scale thresholds dynamically adjusted according to different operating conditions, time or season, where each threshold represents a different level of abnormality. The components of the multi-dimensional anomaly pattern vector are compared with the set of dynamic thresholds to obtain a multi-scale anomaly degree matching set. This set clearly indicates the performance of the current anomaly pattern at different severity levels, avoiding false positives or false negatives due to changes in operating conditions. Finally, the obtained multi-scale anomaly degree matching set will be used for spatio-temporal convolution evolution prediction to generate the final heat value fluctuation warning signal. This prediction process does not analyze the anomaly matching result at the current time in isolation, but places it in a time series, combining the anomaly degree matching sets of the past few times, and inputs them into a pre-trained spatio-temporal convolution prediction model. The model has learned the laws of the development of different anomaly patterns in the pipe network over time and space by learning a large amount of historical anomaly evolution data. The model receives a sequence of recent multi-scale anomaly degree matching sets as input and outputs a prediction result of the anomaly pattern at one or more future time steps. If the prediction result shows that the anomaly is intensifying, spreading or persisting, a structured heat value fluctuation warning signal will be generated. The signal not only contains the warning level, but also contains information such as the location of the anomaly and the expected development trend.

[0039] Optionally, the fusing of the time feature and the space feature to generate a heat value fluctuation feature comprises: spatio-temporal graph convolution fusion of the time feature and the space feature to obtain a spatio-temporal correlation matrix; extracting a multi-scale pattern vector from the spatio-temporal correlation matrix to generate a heat value fluctuation feature.

[0040] Specifically, first, spatio-temporal graph convolution fusion is performed to obtain a spatio-temporal correlation matrix. To achieve this, the natural gas pipe network needs to be abstracted as a graph structure: ,

[0041] where represents the set of all monitoring nodes in the pipe network, and The graph structure is determined by the pipe network topology data. At each time point, each node has the temporal features and spatial features obtained from the previous steps, which together constitute the graph signal of the pipe network at that moment. The spatio-temporal graph convolution fusion model receives a time series of graph signals as input and performs convolution operations in both time and space dimensions. The core of the graph convolution operation in space is to update the feature representation of each node by a aggregation function that weights and integrates the node's own information and the information of all its neighboring nodes. A simplified single-layer graph convolution operation can be represented as: , In this formula, is the feature matrix of all input nodes, each row corresponds to a node and its features; is the normalized adjacency matrix calculated according to the pipe network topology, which defines the weight of information transmission between nodes; is a weight parameter matrix that the model needs to learn; is a nonlinear activation function. is the new feature matrix output after a round of spatial information fusion. The convolution operation in time is responsible for capturing the change pattern of the fused spatial features at consecutive time steps. By combining these two operations in a network architecture, the model can learn the complex rules of heat value fluctuation propagation and evolution in the pipe network, and the final output of the model is the spatio-temporal correlation matrix. After obtaining the spatio-temporal correlation matrix, this step extracts multi-scale pattern vectors from it to generate the final heat value fluctuation features. The spatio-temporal correlation matrix is a high-dimensional data structure that contains rich spatio-temporal dynamic information. In order to make it more suitable for subsequent pattern recognition tasks, it is necessary to extract key pattern information from it. Multi-scale means summarizing and extracting information from different levels and granularities. This can be achieved by applying different sizes of pooling windows or multiple convolution kernels to the spatio-temporal correlation matrix, for example, using smaller windows or convolution kernels can capture local, rapidly changing fluctuation patterns, while using larger windows or convolution kernels can capture global, slowly changing trend patterns. In this way, a set of pattern vectors can be obtained, each describing the characteristics of heat value fluctuations at a certain scale. Combining or concatenating these pattern vectors extracted from different scales forms the final output of the rich heat value fluctuation features.

[0042] Optionally, the multi-dimensional risk analysis of the heat value fluctuation early warning signal to generate a heat value risk level regulation instruction comprises: Analyzing the abnormal pattern information contained in the heat value fluctuation early warning signal to obtain a multi-dimensional heat value risk feature vector; The multi-dimensional calorific value risk feature vector is subjected to level quantification mapping to generate calorific value risk level control instructions.

[0043] Specifically, the first step is to analyze the abnormal pattern information contained in the calorific value fluctuation warning signal to obtain a multi-dimensional calorific value risk feature vector. The calorific value fluctuation warning signal is a structured data volume containing predictive information about future calorific value fluctuations. The analysis process involves extracting a series of key indicators directly related to risk assessment and combining them into a vector. For example, this multi-dimensional calorific value risk feature vector... This may include the following dimensions: the predicted fluctuation amplitude, the number of affected pipeline nodes, the expected duration of the fluctuation, the predicted propagation speed of the anomaly in the pipeline network, and the confidence score of the warning signal itself. A multi-dimensional risk analysis matrix is ​​as follows: Figure 5 As shown. Each dimension is a specific value parsed from the calorific value fluctuation early warning signal, collectively constituting a comprehensive quantitative description of potential risk events. After obtaining the multi-dimensional calorific value risk feature vector, it needs to be subjected to level quantification mapping to generate the final calorific value risk level control instruction. This mapping process consists of two steps. The first step is to perform multi-scale feature fusion and entropy weight quantification to obtain a single numerical risk assessment coefficient. Since the dimensions of each feature in the multi-dimensional calorific value risk feature vector are different, and their importance may not be consistent, the entropy weight method is used to objectively determine the weight of each dimension. The entropy weight method evaluates the information content of each feature by calculating its information entropy in historical data. The smaller the information entropy, the more information it provides, and the higher the weight should be assigned. Let the feature... The weight is The calculation formula is as follows: , in It is calculated based on historical data. Information entropy of each feature Representing the The information entropy value of each feature, This represents the total number of dimensions of the features. After obtaining the weights of all features, the multi-dimensional heat value risk feature vector is summed using a weighted average method. Integrate into a numerical risk assessment coefficient ,Right now: , in It is a vector The Middle the value of each feature after normalization. In the second step, the obtained risk assessment coefficient is matched with a dynamic threshold interval and mapped with a regulation strategy. The risk level corresponds to a numerical dynamic threshold interval. For example, the risk level can be divided into three levels: low, medium, and high, which correspond to different ranges of values of the risk assessment coefficient . The calculated risk assessment coefficient is matched with these threshold intervals to determine the risk level of the current early warning event. Then, according to the determined risk level, a corresponding regulation strategy is matched and selected from a set strategy library. For example, the "low" risk level may correspond to the instruction "suggest attention", the "medium" risk level corresponds to the instruction "notify the operation and maintenance personnel", and the "high" risk level corresponds to the instruction "start the emergency plan". The final generated specific operation instruction is the hot value risk level regulation instruction.

[0044] Optionally, the level quantization mapping of the multi-dimensional hot value risk feature vector to generate the hot value risk level regulation instruction comprises: multi-scale feature fusion and entropy weight quantization of the multi-dimensional hot value risk feature vector to obtain a numerical risk assessment coefficient; dynamic threshold interval matching and regulation strategy mapping of the risk assessment coefficient to generate the hot value risk level regulation instruction.

[0045] Specifically, first, the input multi-dimensional heat value risk feature vector is subjected to multi-scale feature fusion and entropy weight quantization. The feature vector here not only contains the risk intensity, range, trend and other dimensions at the current time, but also may contain statistical features extracted from different historical time scales, which together constitute multi-scale features. In order to objectively determine the importance of each dimension feature in the comprehensive risk assessment, an entropy weight quantization method is used to calculate the weight. This method is based on information entropy theory, and the weight is assigned by analyzing the dispersion degree of each dimension feature data in the historical sample. Specifically, the greater the numerical change of a feature dimension, the smaller the information entropy, indicating that the feature provides more effective information, and therefore should be given a higher weight. After calculating the weight of each risk feature dimension by this method, the component values in the multi-dimensional heat value risk feature vector are normalized, then multiplied by the corresponding weight and summed, to obtain a single, quantitative numerical risk assessment coefficient. After obtaining the numerical risk assessment coefficient, dynamic threshold interval matching and regulation strategy mapping are performed. The threshold here is not fixed but dynamically adjusted. It will be adjusted in real time according to the current operation mode of the pipeline network, seasonal demand changes, and even the expected stability of the upstream gas source and other macro background information, to adjust the threshold interval used to divide the risk level. For example, during the winter gas peak period, the tolerance for heat value fluctuations may be reduced, and the corresponding risk threshold at each level will be lowered, making it more sensitive. The calculated numerical risk assessment coefficient is compared with these dynamic threshold intervals to determine its risk level. Once the level is determined, the corresponding specific regulation strategy is retrieved and matched from the pre-set regulation strategy knowledge base, and finally the heat value risk level regulation instruction containing the specific risk level and operation suggestion is generated and output.

[0046] Optionally, the dynamic threshold matching and pattern evolution prediction of the multi-dimensional abnormal pattern vector to generate a heat value fluctuation early warning signal comprises: A multi-scale abnormality degree matching set is obtained using the multi-dimensional abnormal pattern vector; The multi-scale abnormality degree matching set is subjected to spatio-temporal convolution evolution prediction to generate a heat value fluctuation early warning signal.

[0047] Specifically, first, the multi-dimensional anomaly pattern vector generated in the previous step is used to obtain a multi-scale anomaly degree matching set through multi-scale analysis. Specifically, instead of making a single threshold judgment on the anomaly vector at the current time, the vector is examined from different time scales. For example, the instantaneous value of the vector is calculated, representing the instantaneous abnormal intensity; the moving average of the vector in a short time window is calculated to smooth the noise and reflect the recent average abnormal level; and the change rate of the vector in a longer time window is calculated to capture the accumulation or development trend of the anomaly. Comparing these indicators calculated at different scales with their respective dynamic thresholds, the matching results, whether they are Boolean trigger flags or continuous anomaly degree scores, together constitute a multi-scale anomaly degree matching set. This set can comprehensively depict whether the current abnormal state is a sudden spike, a sustained deviation, or a slow deterioration, providing more dynamic information than a single time point. After obtaining the multi-sequence multi-scale anomaly degree matching set, a spatiotemporal convolution evolution prediction is performed to generate the final heat value fluctuation warning signal. This step uses an advanced deep learning model, such as a spatiotemporal graph convolution network, which is specifically designed to handle data with complex spatiotemporal dependencies. In this model, the multi-scale anomaly degree matching set is used as input features, which are mapped to a graph structure representing the natural gas pipeline network. The model learns the mutual influence and propagation rules of abnormal patterns in the pipeline topology structure through spatial convolution layers; at the same time, it learns the evolution logic of these abnormal patterns over time from a multi-scale perspective through time convolution or recurrent layers. Through training on a large amount of historical data, the model can grasp the internal laws of evolution from the current abnormal state to the future state. In real-time operation, the model receives a sequence of continuous multi-scale anomaly degree matching sets for the recent period and predicts the possible values of each anomaly degree indicator in the set after one or more time steps. When one or more indicators in the predicted future values exceed the pre-set safety limit, a heat value fluctuation warning signal is generated. This signal not only indicates the occurrence of the warning, but more importantly, it contains a quantitative prediction of the future abnormal state, such as the type of anomaly, the expected intensity, and the possible time of occurrence.

[0048] Based on the same inventive concept, as Figure 6 shown, the present application also provides a natural gas energy metering and heat value fluctuation warning system, which comprises: a data acquisition module for acquiring temperature parameters, pressure parameters, flow parameters and gas component parameters of nodes in a natural gas pipeline network, and generating multi-source real-time data; an energy calculation module for coupling calculation on the multi-source real-time data to generate instantaneous energy values; a feature extraction module, configured to acquire historical heat value data and pipe network topology structure data, extract time-space features in combination with the instantaneous energy value, and generate heat value fluctuation features; a pre-warning module, configured to perform dynamic pattern recognition on the heat value fluctuation features, and generate heat value fluctuation pre-warning signals; a risk analysis module, configured to perform multi-dimensional risk analysis on the heat value fluctuation pre-warning signals, and generate heat value risk level regulation instructions.

[0049] To verify the feasibility of the application in implementation, the application is applied to the dispatching and monitoring center of a provincial natural gas pipe network company. The pipe network structure of the company is complex, and natural gas from multiple gas sources, including domestic onshore gas and imported liquefied natural gas (LNG), is received, resulting in frequent fluctuations in the heat value of natural gas in the pipe network. The traditional volumetric measurement method cannot accurately reflect the energy value of the transaction, and the heat value fluctuation pre-warning means is lagging, which is difficult to meet the needs of downstream high-precision industrial users and brings risks to the safe operation of the pipe network. The company hopes to use the application to realize accurate energy measurement of natural gas flow and effectively pre-warn and risk control of abnormal fluctuations of the heat value.

[0050] In this embodiment, the company deploys the system described in the application at key nodes of its pipe network, including the A city gate station and the B industrial park gas supply point downstream. The system collects temperature parameters, pressure parameters, flow parameters and gas component parameters of the nodes in the natural gas pipe network through temperature sensors, pressure sensors, ultrasonic flow meters and online gas chromatographs installed at the nodes, and generates multi-source real-time data through time stamp addition and time alignment processing. Subsequently, the system performs coupling calculation on the multi-source real-time data to generate an instantaneous energy value. In combination with historical heat value data and pipe network topology structure, time-space features are extracted to generate heat value fluctuation features. Through dynamic pattern recognition analysis of the heat value fluctuation features, heat value fluctuation pre-warning signals are generated, and finally multi-dimensional risk analysis is performed to generate heat value risk level regulation instructions.

[0051] To verify the effectiveness of the application, the system records data under various working conditions, including upstream gas source switching, pipe network pressure adjustment and downstream user load change scenarios.

[0052] At 10:00 on March 15, 2025, the system collected data from the A City Gate Station: temperature 20°C (293.15K), pressure 4.0MPa, volumetric flow rate 50000m³ / h, gas composition 92.5% methane, 4.5% ethane, 1.2% propane, etc. The system first calculated the gas compressibility factor under the current working conditions according to the AGA8-DC92 standard, which was 0.985. Then it used the compensation formula to convert the working condition flow rate to the volumetric flow rate at standard conditions. At the same time, according to the real-time composition, it calculated the current heating value to be 38.5MJ / m³. Finally, the instantaneous energy value at that moment was obtained by multiplying the two, which was 1975GJ / h. This process realizes the accurate measurement from volume to energy.

[0053] In the spatio-temporal feature extraction stage, the system analyzed the instantaneous energy value sequence of the A City Gate Station in the past week, and identified its daily periodic fluctuation rule (time feature). At the same time, based on the pipe network topology data, the system analyzed the spatial correlation between the A City Gate Station and the downstream B Industrial Park gas supply point, and found that there was a 45-minute lag positive correlation between the heating values of the two (spatial feature). Through spatio-temporal graph convolution fusion, the system generated a heating value fluctuation feature that could fully reflect the dynamics of the pipe network.

[0054] On April 5, 2025 at 14:30, due to the emergency switching of the upstream high-heating-value LNG gas source, the system captured the anomaly through dynamic pattern recognition. The spatio-temporal correlation matrix extracted by the system showed that the heating value feature of the A City Gate Station deviated significantly from its historical pattern, and the correlation with the surrounding nodes weakened, thus generating a "upstream gas source mutation type" multi-dimensional abnormal pattern vector. Through pattern evolution prediction, the system inferred that the heating value of the B Industrial Park gas supply point would exceed the upper limit of 5% specified in the contract after 2 hours. Based on this, the system generated a heating value fluctuation warning signal at 14:35.

[0055] After receiving the warning signal, the risk analysis module was started. The system analyzed the warning signal and obtained a multi-dimensional heating value risk feature vector with high risk intensity, wide impact range and obvious deterioration trend. Through entropy weight quantification, the numerical risk assessment coefficient was calculated to be 0.82. This coefficient value matches the "Ⅲ (severe)" risk in the dynamic threshold interval. The system immediately generated a heating value risk level control instruction at 14:36: "Risk level: severe. Reason: upstream gas source mutation. Suggestion: immediately contact the upstream to confirm the gas source composition, prepare to start the pipe network mixed transportation plan to stabilize the heating value fluctuation, and issue a gas quality warning to the downstream B Industrial Park." According to this instruction, the dispatch personnel completed the control operation before 15:00, effectively avoiding the impact on the downstream users.

[0056] From the data comparison, the energy metering and early warning method of the present application has remarkable advantages. In terms of energy metering, compared with the traditional volume metering, the metering accuracy of the present method is improved by about 3%, effectively avoiding transaction disputes caused by heat value fluctuations. In terms of early warning capability, the system realizes 2-hour early warning for the incident on April 5, 2025, while the traditional alarm system based on single-point threshold can only be triggered after the heat value exceeds the limit, basically without advance. By introducing dynamic pattern recognition and risk assessment, the early warning accuracy of the system reaches more than 97%, and the false alarm rate is reduced by more than 60% compared with the traditional method.

[0057] Table 1 Natural gas official node data acquisition and energy metering table

[0058] Table 2 Heat value fluctuation characteristic analysis and early warning signal generation table

[0059] Table 3 Heat value risk assessment and control instruction table

[0060] From the data of the above table 1 to table 3, it can be seen that the present application has excellent performance in the actual operation of the natural gas pipeline network. Table 1 clearly shows the system's ability to accurately calculate the instantaneous energy value from multiple physical parameters in real time. On April 5, 2025, at 14:30, the system accurately captured the significant increase in heat value caused by gas source switching, providing high-fidelity data input for subsequent early warning. Table 2 reflects the strong prediction and early warning capability of the present application. In the key event on April 5, the system not only identified the root cause of the "upstream gas source mutation type", but also gave a valuable early warning period of 2 hours and a quantitative fluctuation amplitude prediction, proving the advancement of spatiotemporal feature extraction and dynamic pattern recognition. Table 3 proves the system's closed-loop management capability from early warning to decision-making. The system can convert complex early warning signals into explicit risk levels and executable control instructions. According to the generated instructions, the measures taken by the dispatch center have achieved good results, effectively resolving potential operational risks and ensuring the safety of pipeline operation and the fairness of trade transfer.

[0061] It should be noted that the electrical connection between the above-mentioned units does not necessarily represent the direct connection of the line, and the indirect connection mode can also be applied to the embodiments of the present application as long as the purpose of the present application is achieved. The above-mentioned is only an exemplary embodiment of the present application, and cannot limit the scope of the present application.

[0062] intended to encompass any and all embodiments of the application with equivalents as would be ascertained by those skilled in the art to which the application pertains. Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the application being indicated by the following claims.

Claims

1. A natural gas energy metering and heating value fluctuation early warning method, characterized in that, The method comprises: Collecting temperature parameters, pressure parameters, flow parameters and gas component parameters of nodes in a natural gas pipeline network to generate multi-source real-time data; Performing coupling calculation on the multi-source real-time data to generate instantaneous energy values; Obtaining historical heat value data and pipeline network topology structure data, combining the instantaneous energy values to perform space-time feature extraction to generate heat value fluctuation features; Performing dynamic mode recognition on the heat value fluctuation features to generate heat value fluctuation early warning signals; Performing multi-dimensional risk analysis on the heat value fluctuation early warning signals to generate heat value risk level regulation instructions.

2. The natural gas energy metering and heating value fluctuation early warning method according to claim 1, characterized in that, The generation of multi-source real-time data comprises: Obtaining temperature parameters, pressure parameters, flow parameters and gas component parameters; Adding time stamps to the temperature parameters, pressure parameters, flow parameters and gas component parameters and performing time alignment processing to generate multi-source real-time data.

3. The natural gas energy metering and heating value fluctuation early warning method according to claim 1, characterized in that, The coupling calculation on the multi-source real-time data to generate instantaneous energy values comprises: According to the temperature parameters, pressure parameters and gas component parameters in the multi-source real-time data, a gas compression factor is calculated; Using the gas compression factor to perform compensation calculation on the flow parameters in the multi-source real-time data to obtain volume flow under standard state; According to the gas component parameters, a current heat value is calculated; The volume flow under standard state is multiplied by the current heat value to output an instantaneous energy value.

4. The natural gas energy metering and heating value fluctuation early warning method according to claim 1, characterized in that, The space-time feature extraction comprises: Based on the instantaneous energy values and the historical heat value data, short-term fluctuation analysis and long-term trend analysis are performed to generate time features; Based on the pipeline network topology structure data, spatial correlation analysis is performed on the instantaneous energy values to generate space features; The time features and the space features are fused to generate heat value fluctuation features.

5. The natural gas energy metering and heating value fluctuation warning method of claim 1, wherein, The dynamic mode recognition on the heat value fluctuation features to generate heat value fluctuation early warning signals comprises: A space-time correlation matrix of the heat value fluctuation features is extracted to obtain a multi-dimensional abnormal mode vector; Dynamic threshold matching and mode evolution prediction are performed on the multi-dimensional abnormal mode vector to generate heat value fluctuation early warning signals.

6. The natural gas energy metering and heating value fluctuation early warning method according to claim 4, characterized in that, The fusion of the time features and the space features to generate heat value fluctuation features comprises: The time features and the space features are subjected to space-time graph convolution fusion to obtain a space-time correlation matrix; A multi-scale mode vector is extracted from the space-time correlation matrix to generate heat value fluctuation features.

7. The natural gas energy metering and heating value fluctuation warning method of claim 1, wherein, The multi-dimensional risk analysis on the heat value fluctuation early warning signals to generate heat value risk level regulation instructions comprises: Abnormal mode information contained in the heat value fluctuation early warning signals is analyzed to obtain a multi-dimensional heat value risk feature vector; The multi-dimensional heat value risk feature vector is subjected to level quantization mapping to generate heat value risk level regulation instructions.

8. The natural gas energy metering and heating value fluctuation early warning method according to claim 7, characterized in that, The level quantization mapping of the multi-dimensional heat value risk feature vector to generate heat value risk level regulation instructions comprises: The multi-dimensional heat value risk feature vector is subjected to multi-scale feature fusion and entropy weight quantization to obtain a numerical risk evaluation coefficient; The risk evaluation coefficient is subjected to dynamic threshold interval matching and regulation strategy mapping to generate heat value risk level regulation instructions.

9. The natural gas energy metering and heating value fluctuation warning method of claim 5, wherein, The dynamic threshold matching and pattern evolution prediction on the multi-dimensional abnormal pattern vector generate a heat value fluctuation early warning signal, and the method comprises the steps of: Using the multi-dimensional abnormal pattern vector, a multi-scale abnormality degree matching set is obtained; The multi-scale abnormality degree matching set is subjected to spatio-temporal convolution evolution prediction to generate a heat value fluctuation early warning signal.

10. A natural gas energy metering and heating value fluctuation early warning system applied to the natural gas energy metering and heating value fluctuation early warning method of any one of claims 1-9, characterized in that, The system comprises: A data acquisition module for acquiring temperature parameters, pressure parameters, flow parameters and gas component parameters of nodes in a natural gas pipeline network, and generating multi-source real-time data; An energy calculation module for coupling calculation on the multi-source real-time data to generate an instantaneous energy value; A feature extraction module for obtaining historical heat value data and pipeline network topology structure data, and combining the instantaneous energy value to perform spatio-temporal feature extraction to generate heat value fluctuation features; An early warning module for performing dynamic pattern recognition on the heat value fluctuation features to generate a heat value fluctuation early warning signal; A risk analysis module for performing multi-dimensional risk analysis on the heat value fluctuation early warning signal to generate a heat value risk grade regulation instruction.

Images