Natural gas pipeline network state parameter prediction method, device, equipment and medium

Abstract

The invention relates to the field of natural gas pipeline networks, and discloses a natural gas pipeline network state parameter prediction method and device, equipment and a medium, which can obtain state parameter data and a station sequence of a plurality of target stations in a natural gas pipeline network system, and obtain a constructed pipeline network adjacency matrix. Constructing a parameter value matrix and a correlation matrix according to the state parameter data and the station order of the plurality of target stations, extracting first feature data and second feature data from the parameter value matrix based on the pipe network adjacency matrix and the correlation matrix, and performing fusion to obtain feature fusion data, and inputting the feature fusion data into the trained state parameter prediction model for parameter prediction to obtain a parameter value matrix of the target prediction time point. According to the method, the parameter values of the multiple target stations in the natural gas pipeline network system at the target prediction time point can be predicted based on the machine learning model, and the accuracy of the determined parameter values is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of natural gas pipeline networks, and in particular to a method, device, equipment and medium for predicting state parameters of a natural gas pipeline network. Background Art

[0002] The natural gas pipeline network system can be used to transport natural gas from the collection site to consumption sites such as cities and industrial users. It consists of multiple stations and pipelines.

[0003] The natural gas pipeline network system is a complex system. Relevant technologies can collect parameter values ​​of certain status parameters in the natural gas pipeline network system and conduct status assessment of related equipment and pipelines based on the parameter values ​​of the status parameters.

[0004] However, in the process of collecting parameter values ​​of state parameters, related technologies are easily affected by factors such as the external environment and equipment structure failures, which may cause noise in the collected data and thus reduce the accuracy of state assessment.

[0005] Therefore, the relevant technology urgently needs a method that can effectively determine the parameter values ​​of state parameters in the natural gas pipeline network system. Summary of the invention

[0006] The present invention provides a natural gas pipeline network state parameter prediction method, device, equipment and medium, which are used to determine the parameter value of the natural gas pipeline network by performing parameter prediction through a state parameter model, thereby solving the defect of noise in the parameter value determined by data collection in the related art.

[0007] In a first aspect, the present invention provides a method for predicting state parameters of a natural gas pipeline network, comprising: Acquire state parameter data and site order of multiple target sites in the natural gas pipeline network system, and acquire a constructed pipeline network adjacency matrix; wherein the state parameter data of each target site includes parameter values ​​of the target state parameter in the target site at N consecutive time points, N is an integer greater than 1, and the pipeline network adjacency matrix includes nodes corresponding to each target site and edges between two nodes, and the edges are the lengths of pipelines between two target sites; Based on the site order of each of the target sites and the parameter value in each of the state parameter data, construct a parameter value matrix; and based on the site order of each of the target sites and the parameter value in each of the state parameter data, construct a relevance matrix; Extracting features from the parameter value matrix according to the pipe network adjacency matrix and the correlation matrix respectively to obtain first feature data and second feature data; fusing the first feature data and the second feature data to obtain feature fusion data; The feature fusion data is input into a trained state parameter prediction model to predict parameter values, and a prediction parameter value matrix corresponding to a target prediction time point is obtained; wherein the target prediction time point includes at least one time point after the N consecutive time points.

[0008] Optionally, constructing a parameter value matrix based on the site order of each target site and the parameter value in each state parameter data includes: For any of the target sites, the parameter values ​​of the target state parameters at N consecutive time points in the state parameter data of the target site are arranged in chronological order to obtain a row sequence corresponding to the target site; The row sequence corresponding to each target site is used as a row of data in the matrix to be constructed, and the row of data corresponding to each target site is arranged according to the site order of each target site to construct the corresponding parameter value matrix.

[0009] Optionally, constructing a correlation matrix based on the site order of each target site and the parameter value in each state parameter data includes: Determining the correlation between the target state parameters of each of the target sites based on the parameter value in the state parameter data of each of the target sites; Determine the correlation between the target state parameters of the same target site as 1; The correlation matrix is ​​constructed according to the correlation between the target state parameters of the same target site, the correlation between the target state parameters of two target sites, and the site order.

[0010] Optionally, the extracting features of the parameter value matrix according to the pipe network adjacency matrix and the correlation matrix to obtain first feature data and second feature data respectively includes: Inputting the pipe network adjacency matrix and the parameter value matrix into a trained first feature extraction model, so that the first feature extraction model performs feature extraction in the parameter value matrix based on the pipe network adjacency matrix to obtain the first feature data; and inputting the correlation matrix and the parameter value matrix into a trained second feature extraction model, so that the second feature extraction model performs feature extraction in the parameter value matrix based on the correlation matrix to obtain the second feature data; The fusing the first feature data and the second feature data to obtain feature fusion data includes: The first feature data and the second feature data are input into a trained fusion model to perform feature fusion to obtain the feature fusion data.

[0011] Optionally, the first feature extraction model is:

[0012] in, is the first characteristic data, is the first trained weight matrix, is the activation function, is the parameter value matrix, is the pipe network adjacency matrix; The second feature extraction model is:

[0013] in, is the second characteristic data, is the trained second weight matrix, is the parameter value matrix; The fusion model is:

[0014] in, Fusion data for the features, is the trained third weight matrix, is the transpose symbol.

[0015] Optionally, the state parameter prediction model includes a time series prediction model and a linear mapping model; The step of inputting the feature fusion data into the trained state parameter prediction model to predict parameter values, and obtaining a prediction parameter value matrix corresponding to the target prediction time point, includes: Inputting the feature fusion data into the time series prediction model to perform time series prediction to obtain feature prediction data; wherein the time series prediction model is obtained by training a pre-trained time series prediction model based on the sample feature fusion data; The feature prediction data is input into the linear mapping model for linear mapping to obtain the prediction parameter value matrix; wherein the linear mapping model is trained based on the sample feature prediction data and a pre-trained linear mapping model.

[0016] Optionally, the target state parameter is pressure or flow.

[0017] In a second aspect, the present invention provides a natural gas pipeline network state parameter prediction device, comprising: An acquisition unit is used to acquire state parameter data and site order of multiple target sites in a natural gas pipeline network system, and to acquire a constructed pipeline network adjacency matrix; wherein the state parameter data of each target site includes parameter values ​​of the target state parameter in the target site at N consecutive time points, N is an integer greater than 1, and the pipeline network adjacency matrix includes nodes corresponding to each target site and edges between two of the nodes, wherein the edges are the lengths of pipelines between two target sites; A first construction unit, configured to construct a parameter value matrix based on the site order of each of the target sites and the parameter value in each of the state parameter data; A second construction unit is used to construct a correlation matrix based on the site order of each of the target sites and the parameter value in each of the state parameter data; A feature extraction unit, used to extract features from the parameter value matrix according to the pipe network adjacency matrix and the correlation matrix, respectively, to obtain first feature data and second feature data; A feature fusion unit, used for fusing the first feature data and the second feature data to obtain feature fusion data; A parameter prediction unit is used to input the feature fusion data into a trained state parameter prediction model to predict parameter values, and obtain a prediction parameter value matrix corresponding to a target prediction time point; wherein the target prediction time point includes at least one time point after the N consecutive time points.

[0018] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the natural gas pipeline network state parameter prediction method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0019] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the natural gas pipeline network state parameter prediction method of the first aspect or any corresponding embodiment thereof.

[0020] The natural gas pipeline network state parameter prediction method, device, equipment and medium provided by the present invention can obtain the state parameter data and site order of multiple target sites in the natural gas pipeline network system, and obtain the constructed pipeline adjacency matrix, and construct a parameter value matrix and a correlation matrix according to the state parameter data and site order of multiple target sites in the natural gas pipeline network system, extract the first feature data and the second feature data from the parameter value matrix based on the pipeline adjacency matrix and the correlation matrix, fuse the first feature data and the second feature data to obtain feature fusion data, input the feature fusion data into the trained state parameter prediction model to perform parameter prediction, and obtain the parameter value matrix at the target prediction time point. The present invention can realize the prediction of the parameter values ​​of multiple target sites in the natural gas pipeline network system at the target prediction time point based on the machine learning model, without determining the parameter values ​​of the state parameters through data collection, avoiding the noise that may exist in the determined parameter values ​​due to the influence of factors such as the external environment and equipment structure failure, and effectively improving the accuracy of the determined parameter values. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 A flow chart of a natural gas pipeline network state parameter prediction method provided by an embodiment of the present invention; Figure 2 A flowchart of another natural gas pipeline network state parameter prediction method provided by an embodiment of the present invention; Figure 3 A schematic diagram of the structure of a natural gas pipeline network state parameter prediction device provided by an embodiment of the present invention; Figure 4 A schematic diagram of the structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

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

[0024] Combine the following Figure 1-Figure 2 The natural gas pipeline network state parameter prediction method of the present invention is described.

[0025] like Figure 1 As shown, this embodiment proposes a first natural gas pipeline network state parameter prediction method, which may include the following steps: S101. Obtain state parameter data and site order of multiple target sites in the natural gas pipeline network system, and obtain a constructed pipeline network adjacency matrix; wherein the state parameter data of each target site includes parameter values ​​of the target state parameters in the target site at N consecutive time points, N is an integer greater than 1, and the pipeline network adjacency matrix includes nodes corresponding to each target site and edges between two nodes, and an edge is the length of the pipeline between the two target sites.

[0026] The target site may be a site in the natural gas pipeline network system.

[0027] Optionally, in this embodiment, a technician may select a target site from all sites included in the natural gas pipeline network system.

[0028] Optionally, in this embodiment, each site in the natural gas pipeline network system may be directly determined as a target site. In this case, the multiple target sites may be all sites in the natural gas pipeline network system.

[0029] Specifically, the site order may be the ranking of each target site among all target sites assigned in this embodiment, such as 1, 2, or 3.

[0030] The target state parameter may be a state parameter in the natural gas pipeline network system.

[0031] Optionally, the target state parameter may be pressure or flow.

[0032] Specifically, a time point may be half a day, a day or multiple days, or 20 hours. This embodiment does not limit the specific duration of the time point.

[0033] Specifically, the status parameter data of any target site include the parameter values ​​of the target status parameters in the target site at N consecutive time points. For example, the status parameter data of target site A include the parameter values ​​of the target status parameters in target site A on the 1st, 2nd and 3rd of this month, and the status parameter data of target site B include the parameter values ​​of the target status parameters in target site B on the 1st, 2nd and 3rd of this month.

[0034] Optionally, each parameter value in the state parameter data may be a value obtained by normalizing the original parameter value. Of course, each parameter value in the state parameter data may also be an original parameter value.

[0035] It should be noted that, after determining each target site in the natural gas pipeline network system, this embodiment can use each target site as a node in the adjacency matrix to be constructed, and use the pipeline length between each target site as the edge between the corresponding two nodes in the adjacency matrix to be constructed, and construct the corresponding adjacency matrix, i.e., the pipeline network adjacency matrix, based on all nodes and edges.

[0036] S102: construct a parameter value matrix based on the site order of each target site and the parameter value in each state parameter data.

[0037] Specifically, this embodiment may construct a parameter value matrix according to the site order of all target sites and all parameter values ​​included in the state parameter data of each target site.

[0038] Optional, such as Figure 2 As shown, in other natural gas network state parameter prediction methods proposed in this embodiment, step S102 may include steps S1021 and S1022. Among them: S1021. For any target site, the parameter values ​​of the target state parameters in the state parameter data of the target site at N consecutive time points are arranged in chronological order to obtain a row sequence corresponding to the target site.

[0039] S1022: taking the row sequence corresponding to each target site as a row of data in the matrix to be constructed, and arranging the row of data corresponding to each target site according to the site order of each target site to construct a corresponding parameter value matrix.

[0040] Specifically, for any target site, this embodiment can arrange the parameter values ​​of the target state parameter at N consecutive time points in the state parameter data of the target site based on the time sequence to obtain the row sequence corresponding to the target site. For example, when the state parameter data of the target site includes the parameter value A1 of the target state parameter on the 1st of this month, the parameter value A2 on the 2nd of this month, and the parameter value A3 on the 3rd of this month, this embodiment can arrange A1, A2 and A3 in the time sequence to obtain the row sequence [A1, A2, A3] corresponding to the target site.

[0041] After obtaining the row sequence corresponding to each target site, this embodiment can construct a parameter value matrix based on the row sequences and site order of all target sites. Specifically, this embodiment can use the row sequence corresponding to the target site with a site order of 1 as the row data of the first row in the parameter value matrix to be constructed, and use the row sequence corresponding to the target site with a site order of 2 as the row data of the second row in the parameter value matrix to be constructed, until all row sequences are arranged, and a constructed matrix, i.e., a parameter value matrix, is obtained.

[0042] It can be understood that the total number of parameter values ​​in the status parameter data of each target site is N. At this time, when the total number of target sites is P, the dimension of the parameter value matrix is ​​P×N, that is, the number of rows of the parameter value matrix is ​​equal to P, and the number of columns is equal to N.

[0043] S103: construct a correlation matrix based on the site order of each target site and the parameter value in each state parameter data.

[0044] Specifically, this embodiment can construct a correlation matrix between target state parameters of all target sites based on the site order of each target site and based on all parameter values ​​included in the state parameter data of each target site.

[0045] Optionally, step S103 may include: The correlation between the target state parameters of the same target site is determined to be 1; Determine the correlation between the target state parameters of each target site based on the parameter value in the state parameter data of each target site; A correlation matrix is ​​constructed according to the correlation between the target state parameters of the same target site, the correlation between the target state parameters of each target site and the site order.

[0046] It should be noted that the Pearson correlation coefficient analysis method can be used as a statistical method to evaluate the correlation between two indicator variables, and the calculated correlation can be between -1 and 1.

[0047] Specifically, this embodiment may determine the correlation between target state parameters of different target sites based on the Pearson correlation coefficient analysis method and the parameter value included in the state parameter data of each target site.

[0048] Specifically, after determining the correlations between target state parameters of different target sites and the correlations between target state parameters of the same site, this embodiment can arrange all the determined correlations according to the site order of the target sites to construct a correlation matrix.

[0049] It can be understood that when the total number of target sites is P, the dimension of the correlation matrix is ​​P×P. The first correlation in the first row of the correlation matrix is ​​the correlation between the target state parameters in the same target site with a site order of 1, the second correlation in the first row is the correlation between the target state parameters in the target sites with a site order of 2 and a site order of 1, the third correlation in the first row is the correlation between the target state parameters in the target sites with a site order of 3 and a site order of 1, the first correlation in the second row is the correlation between the target state parameters in the target sites with a site order of 1 and a site order of 2, and so on.

[0050] S104 , extracting features from the parameter value matrix according to the pipe network adjacency matrix and the correlation matrix to obtain first feature data and second feature data.

[0051] In this embodiment, feature extraction can be performed in the parameter value matrix based on the pipe network adjacency matrix and the correlation matrix, respectively, to obtain first feature data and second feature data of the parameter value matrix.

[0052] Specifically, the first characteristic data is the first characteristic data extracted from the parameter value matrix based on the pipe network adjacency matrix in this embodiment. The second characteristic data is the second characteristic data extracted from the parameter value matrix based on the correlation matrix in this embodiment.

[0053] S105: Fuse the first feature data and the second feature data to obtain feature fusion data.

[0054] Specifically, this embodiment may fuse the first feature data and the second feature data extracted from the parameter value matrix to obtain feature fusion data.

[0055] S106. Input the feature fusion data into the trained state parameter prediction model to predict parameter values, and obtain a prediction parameter value matrix corresponding to a target prediction time point; wherein the target prediction time point includes at least one time point after N consecutive time points.

[0056] The target prediction time point may be after the above N consecutive time points and adjacent to the above N consecutive time points. The target prediction time point may include at least one time point. For example, when the time point is one day and the above N consecutive time points are from the 1st to the 10th of this month, the target prediction time point may be the 11th of this month or from the 11th to the 20th of this month.

[0057] Specifically, the state parameter prediction model may be based on feature fusion data to predict the parameter value of each target site at a target prediction time point.

[0058] The parameter value matrix at the target prediction time point may include the parameter value of each target site at the target prediction time point.

[0059] It should be noted that, in this embodiment, multiple sample feature fusion data may be used to train the pre-trained state parameter prediction model until a state parameter prediction model having model performance that meets the requirements is obtained, that is, a trained state parameter prediction model is obtained.

[0060] It can be understood that when the parameter values ​​contained in the state parameter data of the target site are values ​​obtained by normalizing the original parameter values, each element in the prediction parameter value matrix output by the state parameter prediction model is a normalized value. This embodiment can perform denormalization processing on each element in the prediction parameter value matrix, that is, it can obtain parameter values ​​that can be used to directly evaluate the operating status of the natural gas pipeline network.

[0061] The natural gas pipeline network state parameter prediction method proposed in this embodiment can obtain the state parameter data and site order of multiple target sites in the natural gas pipeline network system, as well as obtain the constructed pipeline adjacency matrix, construct a parameter value matrix and a correlation matrix based on the state parameter data and site order of multiple target sites in the natural gas pipeline network system, extract the first feature data and the second feature data from the parameter value matrix based on the pipeline adjacency matrix and the correlation matrix, fuse the first feature data and the second feature data to obtain feature fusion data, input the feature fusion data into the trained state parameter prediction model for parameter prediction, and obtain the parameter value matrix at the target prediction time point. This embodiment can realize the prediction of the parameter values ​​of multiple target sites in the natural gas pipeline network system at the target prediction time point based on the machine learning model, without determining the parameter values ​​of the state parameters through data collection, avoiding the noise that may exist in the determined parameter values ​​due to the influence of factors such as the external environment and equipment structure failure, and effectively improving the accuracy of the determined parameter values.

[0062] Related technologies can reduce the impact of external interference factors on the unit's working state parameters by predicting state parameters, and timely monitor the changing trend of state parameter data during operation. Such prediction methods can help operators warn of potential faults, implement timely maintenance and repair measures, and ensure the safe operation of the pipeline network. At the same time, optimization decisions based on prediction results can maximize the operating efficiency of the pipeline network, reduce energy consumption and costs, and achieve sustainable development. Therefore, the rapid prediction of state parameters is of great significance to the safety and optimization of the natural gas pipeline network.

[0063] At present, facing the complex data generated by the daily operation of industrial systems, related technologies have also put forward higher requirements for industrial computing and analysis. Big data analysis methods have gradually become a research hotspot in various industries and fields. Among them, statistical methods and machine learning methods can be mainly used to complete the prediction work.

[0064] In the training process of the neural network, the relevant technology is essentially performing function fitting, that is, after inputting a set of data, a set of associated target output data is obtained. Once the neural network successfully fits the data, a function that can reveal the relationship between the input and output data is formed. When the weights and thresholds of the network are initialized, the network can be trained for function approximation (nonlinear regression), pattern association or pattern classification. In the training stage, the network performance function is minimized by iteratively adjusting the weights and deviations of the network, so that the trained network can provide accurate output for untrained input data. However, the relevant technology fails to consider the relationship between the various features in the original data, and it is difficult for the model to learn the laws and correlations between the features, which in turn affects the accuracy of the prediction of the state parameters of the natural gas pipeline network. In addition, due to the complexity of the natural gas pipeline network system, there is a complex coupling relationship between the state parameters of different sites. The machine learning methods in the relevant technology are difficult to capture this potential coupling relationship and it is difficult to achieve accurate prediction.

[0065] based on Figure 1 This embodiment proposes a second method for predicting natural gas pipeline network state parameters. In this method, step S104 may include: The pipeline network adjacency matrix and the parameter value matrix are input into the trained first feature extraction model, so that the first feature extraction model performs feature extraction in the parameter value matrix based on the pipeline network adjacency matrix to obtain first feature data; and the correlation matrix and the parameter value matrix are input into the trained second feature extraction model, so that the second feature extraction model performs feature extraction in the parameter value matrix based on the correlation matrix to obtain second feature data.

[0066] Step S105 may include: The first feature data and the second feature data are input into the trained fusion model for feature fusion to obtain feature fusion data.

[0067] Optionally, the first feature extraction model is:

[0068] in, is the first feature data, is the first trained weight matrix, is the activation function, is the parameter value matrix, is the pipe network adjacency matrix; The second feature extraction model is:

[0069] in, is the second feature data, is the trained second weight matrix, is the parameter value matrix; The fusion model is:

[0070] in, For feature fusion data, is the trained third weight matrix, is the transpose symbol.

[0071] In this method, the state parameter prediction model includes a time series prediction model and a linear mapping model. At this time, step S106 may include: Inputting the feature fusion data into the time series prediction model for time series prediction to obtain feature prediction data; wherein the time series prediction model is obtained by training a pre-trained time series prediction model based on the sample feature fusion data; The feature prediction data is input into the linear mapping model for linear mapping to obtain a prediction parameter value matrix; wherein the linear mapping model is trained based on the sample feature prediction data and the pre-trained linear mapping model.

[0072] Specifically, in the process of model training, this embodiment can use each site in the natural gas pipeline network system as a sample site, and construct a pipeline network adjacency matrix according to the topological structure of the natural gas pipeline network system. The pipeline network adjacency matrix can be static and unchanged. The historical state parameter data of the sample site is obtained as the sample state parameter data, and the sample parameter value matrix and the sample correlation matrix are constructed. It can be understood that when the sample state parameter data changes, the sample parameter value matrix and the sample correlation matrix will change accordingly.

[0073] In this embodiment, the pipe network adjacency matrix and the sample parameter value matrix can be input into the first feature extraction model to be trained, so that the first feature extraction model to be trained can extract features from the sample parameter value matrix based on the pipe network adjacency matrix, and obtain the first sample feature data. The sample correlation matrix and the sample parameter value matrix can be input into the second feature extraction model to be trained, so that the second feature extraction model to be trained can extract features from the sample parameter value matrix based on the sample correlation matrix, and obtain the second sample feature data. The first sample feature data and the second sample feature data are input into the fusion model to be trained for feature fusion, and the sample feature fusion data is obtained.

[0074] This embodiment can input the sample feature fusion data into a pre-trained time series prediction model for time series prediction, and obtain the sample feature prediction data output by the pre-trained time series prediction model. The sample feature prediction data is input into a pre-trained linear mapping model for linear mapping, and a prediction parameter value matrix output by the pre-trained linear mapping model is obtained. This embodiment can update the model parameters in the pre-trained linear mapping model and the pre-trained time series prediction model based on the gap between the prediction parameter value matrix and the corresponding true parameter value matrix, as well as the weight matrices in the first feature extraction model to be trained, the second feature extraction model, and the fusion model to be trained. , and Updates are performed until the trained first feature extraction model, second feature extraction model, fusion model, time series prediction model and linear mapping model are obtained.

[0075] This embodiment can be used to predict the state parameters of the natural gas pipeline network by mining the complex coupling relationships between the various stations in the gas pipeline network and integrating and analyzing these complex coupling relationships. At the same time, the impact of the time dimension is fully considered.

[0076] The natural gas pipeline network state parameter prediction method proposed in this embodiment can capture the complex coupling relationship between different sites of the natural gas pipeline network. First, according to the pipeline network topology, the first site coupling relationship and the pipeline network adjacency matrix are generated. Secondly, according to the Pearson correlation analysis, the second site coupling relationship, namely the correlation matrix, can be generated. After that, the two coupling relationships are fused, that is, feature extraction is performed in the parameter value matrix based on the pipeline network adjacency matrix and the correlation matrix respectively, and the extracted first feature data and the second feature data are fused to obtain feature fusion data, and the feature fusion data is input into the state parameter prediction model for parameter prediction, and the state parameter prediction model is used to analyze the impact of different moments on future state parameters to achieve state parameter prediction.

[0077] like Figure 3 As shown, this embodiment provides a natural gas pipeline network state parameter prediction device, including: The acquisition unit 301 is used to acquire the state parameter data and the station order of multiple target stations in the natural gas pipeline network system, and to acquire the constructed pipeline network adjacency matrix; wherein the state parameter data of each target station includes the parameter values ​​of the target state parameter in the target station at N consecutive time points, N is an integer greater than 1, and the pipeline network adjacency matrix includes the nodes corresponding to each target station and the edges between the two nodes, and the edges are the pipeline lengths between the two target stations; A first construction unit 302 is used to construct a parameter value matrix based on the site order of each target site and the parameter value in each state parameter data; A second construction unit 303 is used to construct a correlation matrix based on the site order of each target site and the parameter value in each state parameter data; A feature extraction unit 304 is used to extract features from the parameter value matrix according to the pipe network adjacency matrix and the correlation matrix to obtain first feature data and second feature data; A feature fusion unit 305 is used to fuse the first feature data and the second feature data to obtain feature fusion data; The parameter prediction unit 306 is used to input the feature fusion data into the trained state parameter prediction model to predict the parameter value, and obtain the prediction parameter value matrix corresponding to the target prediction time point; wherein the target prediction time point includes at least one time point after N consecutive time points.

[0078] It should be noted that the processing of the acquisition unit 301, the first construction unit 302, the second construction unit 303, the feature extraction unit 304, the feature fusion unit 305 and the parameter prediction unit 306 and the beneficial effects thereof can be respectively referred to in Figure 1 Steps S101 to S106 in the above are not described in detail.

[0079] Optionally, the first building unit 302 is further used for: For any target site, the parameter values ​​of the target state parameters at N consecutive time points in the state parameter data of the target site are arranged in chronological order to obtain a row sequence corresponding to the target site; The row sequence corresponding to each target site is used as a row of data in the matrix to be constructed, and the row of data corresponding to each target site is arranged according to the site order of each target site to construct a corresponding parameter value matrix.

[0080] Optionally, the second construction unit 303 is further used for: Determine the correlation between the target state parameters of each target site based on the parameter value in the state parameter data of each target site; The correlation between the target state parameters of the same target site is determined to be 1; A correlation matrix is ​​constructed according to the correlation between the target state parameters of the same target site, the correlation between the target state parameters of each target site and the site order.

[0081] Optionally, the feature extraction unit 304 is further configured to: Inputting the pipe network adjacency matrix and the parameter value matrix into the trained first feature extraction model, so that the first feature extraction model extracts features from the parameter value matrix based on the pipe network adjacency matrix to obtain first feature data; and inputting the correlation matrix and the parameter value matrix into the trained second feature extraction model, so that the second feature extraction model extracts features from the parameter value matrix based on the correlation matrix to obtain second feature data; The feature fusion unit 305 is further used for: The first feature data and the second feature data are input into the trained fusion model for feature fusion to obtain feature fusion data.

[0082] Optionally, the first feature extraction model is:

[0083] in, is the first feature data, is the first trained weight matrix, is the activation function, is the parameter value matrix, is the pipe network adjacency matrix; The second feature extraction model is:

[0084] in, is the second feature data, is the trained second weight matrix, is the parameter value matrix; The fusion model is:

[0085] in, For feature fusion data, is the trained third weight matrix, is the transpose symbol.

[0086] Optionally, the state parameter prediction model includes a time series prediction model and a linear mapping model; The parameter prediction unit 306 is further configured to: Inputting the feature fusion data into the time series prediction model for time series prediction to obtain feature prediction data; wherein the time series prediction model is obtained by training a pre-trained time series prediction model based on the sample feature fusion data; The feature prediction data is input into the linear mapping model for linear mapping to obtain a prediction parameter value matrix; wherein the linear mapping model is trained based on the sample feature prediction data and the pre-trained linear mapping model.

[0087] Optionally, the target state parameter is pressure or flow.

[0088] The natural gas pipeline network state parameter prediction device proposed in this embodiment can obtain the state parameter data and site order of multiple target sites in the natural gas pipeline network system, as well as obtain the constructed pipeline adjacency matrix, construct a parameter value matrix and a correlation matrix according to the state parameter data and site order of multiple target sites in the natural gas pipeline network system, extract the first feature data and the second feature data from the parameter value matrix based on the pipeline adjacency matrix and the correlation matrix, fuse the first feature data and the second feature data to obtain feature fusion data, input the feature fusion data into the trained state parameter prediction model for parameter prediction, and obtain the parameter value matrix at the target prediction time point. This embodiment can realize the prediction of the parameter values ​​of multiple target sites in the natural gas pipeline network system at the target prediction time point based on the machine learning model, without determining the parameter values ​​of the state parameters through data collection, avoiding the noise that may exist in the determined parameter values ​​due to the influence of factors such as the external environment and equipment structure failure, and effectively improving the accuracy of the determined parameter values.

[0089] The natural gas pipeline network state parameter prediction device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0090] The embodiment of the present invention also provides a computer device having the above Figure 3 The natural gas pipeline network state parameter prediction device shown.

[0091] See also Figure 4 , a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses for communication, and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed in the computer device, including instructions stored in or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 4 A processor 10 is taken as an example.

[0092] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0093] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0094] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function. The data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage devices. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0095] The memory 20 may include a volatile memory, such as a random access memory. The memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive. The memory 20 may also include a combination of the above-mentioned types of memory.

[0096] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0097] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0098] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting state parameters of a natural gas pipeline network, characterized in that: include: Acquire state parameter data and site order of multiple target sites in the natural gas pipeline network system, and acquire a constructed pipeline network adjacency matrix; wherein the state parameter data of each target site includes parameter values ​​of the target state parameter in the target site at N consecutive time points, N is an integer greater than 1, and the pipeline network adjacency matrix includes nodes corresponding to each target site and edges between two nodes, and the edges are the lengths of pipelines between two target sites; Based on the site order of each of the target sites and the parameter value in each of the state parameter data, construct a parameter value matrix; and based on the site order of each of the target sites and the parameter value in each of the state parameter data, construct a relevance matrix; Extracting features from the parameter value matrix according to the pipe network adjacency matrix and the correlation matrix respectively to obtain first feature data and second feature data; fusing the first feature data and the second feature data to obtain feature fusion data; The feature fusion data is input into a trained state parameter prediction model to predict parameter values, and a prediction parameter value matrix corresponding to a target prediction time point is obtained; wherein the target prediction time point includes at least one time point after the N consecutive time points.

2. The method according to claim 1, characterized in that: The step of constructing a parameter value matrix based on the site order of each target site and the parameter value in each state parameter data comprises: For any of the target sites, the parameter values ​​of the target state parameters at N consecutive time points in the state parameter data of the target site are arranged in chronological order to obtain a row sequence corresponding to the target site; The row sequence corresponding to each target site is used as a row of data in the matrix to be constructed, and the row of data corresponding to each target site is arranged according to the site order of each target site to construct the corresponding parameter value matrix.

3. The method according to claim 1, characterized in that The step of constructing a correlation matrix based on the site order of each target site and the parameter value in each state parameter data comprises: Determining the correlation between the target state parameters of each of the target sites based on the parameter value in the state parameter data of each of the target sites; Determine the correlation between the target state parameters of the same target site as 1; The correlation matrix is ​​constructed according to the correlation between the target state parameters of the same target site, the correlation between the target state parameters of two target sites, and the site order.

4. The method according to claim 1, characterized in that: The extracting features of the parameter value matrix according to the pipe network adjacency matrix and the correlation matrix to obtain first feature data and second feature data includes: Inputting the pipe network adjacency matrix and the parameter value matrix into a trained first feature extraction model, so that the first feature extraction model performs feature extraction in the parameter value matrix based on the pipe network adjacency matrix to obtain the first feature data; and inputting the correlation matrix and the parameter value matrix into a trained second feature extraction model, so that the second feature extraction model performs feature extraction in the parameter value matrix based on the correlation matrix to obtain the second feature data; The fusing the first feature data and the second feature data to obtain feature fusion data includes: The first feature data and the second feature data are input into a trained fusion model to perform feature fusion to obtain the feature fusion data.

5. The method according to claim 4, characterized in that The first feature extraction model is: in, is the first characteristic data, is the first trained weight matrix, is the activation function, is the parameter value matrix, is the pipe network adjacency matrix; The second feature extraction model is: in, is the second characteristic data, is the trained second weight matrix, is the parameter value matrix; The fusion model is: in, Fusion data for the features, is the trained third weight matrix, is the transpose symbol.

6. The method according to claim 1, characterized in that The state parameter prediction model includes a time series prediction model and a linear mapping model; The step of inputting the feature fusion data into the trained state parameter prediction model to predict parameter values, and obtaining a prediction parameter value matrix corresponding to the target prediction time point, includes: Inputting the feature fusion data into the time series prediction model to perform time series prediction to obtain feature prediction data; wherein the time series prediction model is obtained by training a pre-trained time series prediction model based on the sample feature fusion data; The feature prediction data is input into the linear mapping model for linear mapping to obtain the prediction parameter value matrix; wherein the linear mapping model is trained based on the sample feature prediction data and a pre-trained linear mapping model.

7. The method according to any one of claims 1 to 6, characterized in that The target state parameter is pressure or flow.

8. A natural gas pipeline network state parameter prediction device, characterized in that: include: An acquisition unit is used to acquire state parameter data and site order of multiple target sites in a natural gas pipeline network system, and to acquire a constructed pipeline network adjacency matrix; wherein the state parameter data of each target site includes parameter values ​​of the target state parameter in the target site at N consecutive time points, N is an integer greater than 1, and the pipeline network adjacency matrix includes nodes corresponding to each target site and edges between two of the nodes, wherein the edges are the lengths of pipelines between two target sites; A first construction unit, configured to construct a parameter value matrix based on the site order of each of the target sites and the parameter value in each of the state parameter data; A second construction unit is used to construct a correlation matrix based on the site order of each of the target sites and the parameter value in each of the state parameter data; A feature extraction unit, used to extract features from the parameter value matrix according to the pipe network adjacency matrix and the correlation matrix, respectively, to obtain first feature data and second feature data; A feature fusion unit, used for fusing the first feature data and the second feature data to obtain feature fusion data; A parameter prediction unit is used to input the feature fusion data into a trained state parameter prediction model to predict parameter values, and obtain a prediction parameter value matrix corresponding to a target prediction time point; wherein the target prediction time point includes at least one time point after the N consecutive time points.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the natural gas pipeline network state parameter prediction method according to any one of claims 1 to 7 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the natural gas pipeline network state parameter prediction method according to any one of claims 1 to 7.

Images