Gas pipe network simulation analysis system

Abstract

The invention discloses a gas pipe network simulation analysis system. Comprising a gas pipe network data acquisition unit, a pipe network topological structure building unit, a pipe network data value influence factor analysis unit, a gas pipe network data prediction model building unit, a gas pipe network data periodic prediction unit, a gas pipe network data regulation and control unit, a risk early warning unit and a gas pipe network data storage unit. According to the method, the simulation of the gas pipe network is realized by building the pipe network topological structure in the virtual environment; meanwhile, periodic prediction of the gas pipe network data is achieved through a time recurrent neural network model, the collected real data value on the gas pipe network is compared with the periodic prediction range value of the gas pipe network data, the risk of the gas pipe network is judged, and if the real data value exceeds the periodic prediction range value of the gas pipe network data, the risk of the gas pipe network is judged. And if yes, regulating and controlling the real data value of the gas pipe network data exceeding the preset threshold value by regulating valves on branches and branch nodes in each constituent part in the virtual pipe network.

Description

Technical Field

[0001] The invention belongs to the field of digital simulation systems, and in particular, relates to a gas pipe network simulation analysis system. Background Art

[0002] With the acceleration of urbanization, urban gas pipeline networks continue to extend, the operation scale is huge, and the underground pipelines are complex. The traditional manual management mode can no longer meet the needs of urban development. Digital gas pipeline network simulation technology analyzes and calculates various data of the pipeline network in real time online, grasps the operating conditions of the entire pipeline network, and promptly discovers and predicts possible risks. It is based on a high-precision pipeline system model and powerful algorithm computing power, and continuously analyzes and calculates various data of the pipeline network online to improve the accuracy and real-time performance of the analysis. Through the data fusion model, enterprises can comprehensively analyze and perceive the dynamic changes of the pipeline network, connect digital islands through the technical platform, coordinate internal and external data of the enterprise, and mobilize data resources. This technology can not only improve the safety and efficiency of gas management, but also help enterprises make refined decisions, optimize the structure and configuration of the pipeline network, and improve the safety and economy of the pipeline network.

[0003] A simulation method for a multi-level urban gas pipeline network is disclosed in a Chinese authorized patent with application number CN202310113798.X, which provides a first-level to a third-level system with decreasing operation permissions. Verify the user's operation rights. For the first-level permission, a city trunk model is established with a first-level system, and simulation analysis is performed to obtain simulation data of each trunk node. For the second-level permission, a second-level system is used to establish a jurisdiction model; the simulation data of the corresponding trunk nodes in the city trunk model is obtained, and it is used as a boundary condition for simulation analysis to obtain simulation data of the jurisdiction node. For the third-level permission, a third-level system is used to establish a management unit model; the simulation data of the corresponding jurisdiction node in the jurisdiction model is obtained, and it is used as a boundary condition for simulation analysis to obtain simulation data of the management unit node. In this way, nested hierarchical simulation and operation can be performed according to the hierarchical management architecture of the urban gas pipeline network, and the data of the pipeline network at all levels can be fully utilized across levels, and the overall control can be efficient and significantly reduced. The computing cost.

[0004] The defect of the above-mentioned patent is that although the existing gas pipeline network simulation system or method realizes the simulation of gas pipeline network node data for users with different operating permissions, it lacks the prediction of the load value of the relevant data of the gas pipeline network nodes, resulting in the existing gas pipeline network simulation system being unable to intuitively regulate the operating trend of the gas pipeline network at the next moment and provide effective guiding result prediction values ​​for risk warning. Summary of the invention

[0005] Aiming at the problem that the existing gas pipeline network simulation system lacks the prediction of the load value of the relevant data of the gas pipeline network nodes, the present invention provides a gas pipeline network simulation and analysis system.

[0006] In order to achieve the above technical objectives, the technical solution adopted by the present invention is as follows:

[0007] A gas pipeline network simulation analysis system, comprising a gas pipeline network data acquisition unit, a pipeline network topology structure building unit, a pipeline network data value influencing factor analysis unit, a gas pipeline network data prediction model building unit, a gas pipeline network data periodicity prediction unit, a gas pipeline network data control unit, a risk warning unit and a gas pipeline network data storage unit;

[0008] The gas pipeline network data acquisition unit is connected to the gas pipeline network data storage unit for communication, and collects corresponding gas pipeline network data through the SCADA monitoring system on the gas pipeline network;

[0009] The pipe network topology structure building unit is connected to the gas pipe network data acquisition unit for communication, and each component of the pipe network is built according to the relationship and location of the SCADA monitoring system acquisition equipment on the gas pipe network;

[0010] The pipeline network data value influencing factor analysis unit is connected to the gas pipeline network data acquisition unit for communication, and performs correlation analysis on each single type of gas pipeline network data collected to obtain data related to influencing factors that affect changes in gas pipeline network data;

[0011] The gas network data prediction model building unit is connected to the network data value influencing factor analysis unit in communication, a time recurrent neural network model (LSTM) is selected as the gas network data prediction model, and an input gate, a forget gate and an output gate of the time recurrent neural network model are determined;

[0012] The gas network data periodicity prediction unit is connected to the gas network data prediction model building unit for communication, and completes the prediction of the gas network data periodicity by adjusting the input gate value of the time recurrent neural network model within a preset period;

[0013] The risk warning unit is connected to the gas network data periodic prediction unit for communication, and determines whether the gas network data should be warned of risk by comparing the real data value within the collected gas network data period with the preset threshold value;

[0014] The gas network data control unit is connected to the risk warning unit through communication, and the gas network data is controlled when it exceeds the real data value of the preset threshold value by adjusting the valves on the branches and branch nodes in each component of the virtual network;

[0015] The gas pipeline network data storage unit is communicatively connected with the gas pipeline network data periodic prediction unit and is used for storing historical gas pipeline network data.

[0016] Furthermore, the components in the pipe network include branches, boundaries and branch points.

[0017] Furthermore, the gas pipeline network data includes methane gas concentration, pressure and flow, and a pipeline network monitoring point is set corresponding to each branch point in the pipeline network.

[0018] Furthermore, the detailed steps for correlation analysis of each single type of gas pipeline network data collected are as follows:

[0019] Firstly, a single data type in the gas pipeline network data is selected as the object to be analyzed;

[0020] Then, the gas pipe environment or weather data values ​​that may affect a single data type are selected as the objects of correlation analysis;

[0021] Then, the correlation analysis method is used to obtain the correlation coefficient between the two variables of the object to be judged and the object of correlation analysis;

[0022] According to the absolute value of the correlation coefficient, the correlation analysis objects with strong correlation are selected as the relevant data of the influencing factors affecting the changes in the gas pipeline network data; the correlation analysis objects with strong correlation are: the absolute value of the correlation coefficient in the interval [0.5, 0.8) is significantly correlated. The absolute value of the correlation coefficient in the interval [0.8, 1) is strongly correlated.

[0023] Finally, repeat the above steps until the correlation analysis of all different types of gas pipeline network data is completed.

[0024] Furthermore, the correlation analysis method corresponds to the calculation formula:

[0025]

[0026] Where: r x,y ——Correlation coefficient between variables x and y, n——The number of variables x and y, x i ——the i-th value of variable x, - the mean of variables x, y i ——the i-th value of variable y, ——the mean value of variable y;

[0027] The range of the correlation coefficient is between [-1,1]. The closer its absolute value is to 1, the stronger the correlation between the two is. If its absolute value is closer to 0, the weaker the correlation between the two is. The absolute value of the correlation coefficient in the interval [0, 0.3) indicates a very weak correlation. The absolute value of the correlation coefficient in the interval [0.3, 0.5) indicates a weak correlation. The absolute value of the correlation coefficient in the interval [0.5, 0.8) indicates a significant correlation. The absolute value of the correlation coefficient in the interval [0.8, 1) indicates a strong correlation.

[0028] Furthermore, the input parameters of the time-recurrent neural network model include the cell state C at the previous moment t-1 , hidden state ht-1 and the current input information h t ; The output parameter is the information C at the current moment t and the hidden state h t ;

[0029] The temporal recurrent neural network model includes a forget gate, an input gate, and an output gate;

[0030] Forget gate: determine the cell state C at the previous moment t-1 The information that needs to be retained and the information that needs to be discarded; t-1 and x t Enter the forget gate and pass it to the Sigmoid function; the specific calculation formula is:

[0031] f t =σ(W f h t-1 +U f x t +b f )

[0032] Where: σ——activation function Sigmoid, output value range [0, 1], control information propagation;

[0033] W f , U f ——The weight matrix of the forget gate;

[0034] b f ——Forget gate bias;

[0035] Input gate: used to determine the importance of the gas network data input at the current moment, and retain the important gas network data at the current moment and use it to update the cell state; the calculation formula is:

[0036] i t =σ(W i h t-1 +U i x t +b i )

[0037]

[0038] Where: tanh - activation function, output value range [-1, 1];

[0039] W i , U i ——Input gate weight matrix;

[0040] W C , U C ——Candidate layer weight matrix;

[0041] bi 、b C ——Input gate, candidate layer bias;

[0042] Output gate: confirm the current cell state C t How much information is lost to the value h of the hidden state at the next moment? t ; The calculation formula is:

[0043] O t =σ(W o h t-1 +U o x t +b t )

[0044] h t =tanh(C t )

[0045] Where: W O , U O ——The weight matrix of the output layer;

[0046] b t ——Output gate bias;

[0047] Cell state: used to update and store information; specific calculation formula:

[0048]

[0049] Furthermore, by adjusting the input gate value of the time recurrent neural network model within a preset period:

[0050] The historical gas network data collected within a preset period has multiple data values ​​as input gate values;

[0051] Sort multiple data values ​​and select the maximum and minimum values ​​as input gate values ​​of the time recurrent neural network model;

[0052] The maximum and minimum values ​​of the input gate of the time recurrent neural network model are input, and the maximum and minimum values ​​predicted by the output gate are obtained as the periodic preset thresholds of the gas network data.

[0053] Furthermore, the gas network data exceeds the preset threshold real data value and performs a control method:

[0054] S701, mapping the real data values ​​collected within the gas pipeline network data cycle to each component of the constructed pipeline network in a 1:1 ratio;

[0055] S702, adjusting the methane gas concentration, pressure and opening of the flow valves set on the branches and branch points;

[0056] S703, using the adjusted methane gas concentration, pressure and flow valve opening as the input gate value of the time recurrent neural network model to predict the gas pipeline network data at the next moment;

[0057] S704, judging whether the predicted value of the gas network data at the next moment is within the preset threshold; if it is within the preset threshold, proceeding to step S705; if it is not within the preset threshold, repeating the previous step S703;

[0058] S705, ending the adjustment process, and sending the corresponding methane gas concentration, pressure, and flow valve opening to the system administrator for gas network maintenance personnel to adjust and maintain parameters during network maintenance.

[0059] Furthermore, it also includes a gas pipeline network data preprocessing unit, which is communicatively connected to the gas pipeline network data acquisition unit and is used for preprocessing the acquired gas pipeline network data.

[0060] Furthermore, the preprocessing method of the collected gas pipeline network data includes:

[0061] Identification and handling of missing data;

[0062] Identification and treatment of outlier data.

[0063] Furthermore, the detailed steps for identifying and processing missing data are as follows:

[0064] By comparing the number of data fields of different data types, determine whether a single data type is missing;

[0065] If there are any missing values, the data interpolation method is used to fill the missing values, or the entire piece of other gas pipeline network data collected at the corresponding time of the missing data is deleted.

[0066] Further, the detailed steps for identifying and processing outlier data are:

[0067] Outlier data can be identified by checking whether the individual data under a single data type is out of the data range of the single data type or the collected data has a value of zero (except for zero temperature or zero natural gas usage, which are not identified as outlier data). Box plot analysis can also be used to identify outlier data:

[0068] Sort a set of data from small to large values. Find the first quartile Q of the set of data 1 , median M and third quartile Q 3 ;

[0069] Let IQR=Q 3 -Q 1, draw a rectangle with IQR (interquartile range) as the length of the rectangle, and the positions on both sides of the rectangle correspond to the first quartile Q of the group of data 1 and the third quartile Q 3 , draw a median line on the long side of the rectangle, corresponding to the median M of the group of data, this line is called the median line.

[0070] In Q 3 +1.5IQR and Q 1 Draw two straight lines at -1.5IQR to distinguish abnormal values ​​from normal values. The range between the two lines is called the inner limit. 3 +3IQR and Q 1 Draw two straight lines at -3IQR to distinguish between extreme outliers and mild outliers. 3 +3IQR and Q 1 The range outside the -3IQR is called the outer limit. Data outside the inner limit is considered an outlier, data between the inner and outer limits is considered mild outlier, and data outside the outer limit is considered extreme outlier.

[0071] The outlier data value is replaced by the data replacement method, or the entire gas pipeline network data collected at the corresponding time of the outlier data is deleted in its entirety.

[0072] Compared with the prior art, the present invention has the following beneficial effects:

[0073] The gas pipeline network simulation is realized by constructing the pipeline network topology structure in a virtual environment. At the same time, the time recurrent neural network model is used to realize the periodic prediction of the gas pipeline network data. The real data values ​​collected on the gas pipeline network are compared with the periodic prediction range values ​​of the gas pipeline network data to judge the risk of the gas pipeline network. If the real data value exceeds the periodic prediction range value of the gas pipeline network data, the valves on the branches and branch nodes in each component of the virtual pipeline network are adjusted to realize the regulation of the real data value of the gas pipeline network data exceeding the preset threshold. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 This is an overall structural block diagram of a gas pipe network simulation analysis system in an embodiment of the present invention;

[0075] Figure 2 A flowchart of the detailed steps of performing correlation analysis on gas network data in an embodiment of the present invention;

[0076] Figure 3 This is a flow chart of a method for regulating when gas network data exceeds a preset threshold real data value in an embodiment of the present invention.

[0077] Explanation of the marks in the figure: 10 - gas pipeline network data acquisition unit, 20 - pipeline network topology structure building unit, 30 - pipeline network data value influencing factor analysis unit, 40 - gas pipeline network data prediction model building unit, 50 - gas pipeline network data periodic prediction unit, 60 - gas pipeline network data control unit, 70 - risk warning unit, 80 - gas pipeline network data storage unit. DETAILED DESCRIPTION

[0078] In order to facilitate the understanding of those skilled in the art, the present invention is further described below in conjunction with embodiments and drawings. The contents mentioned in the implementation modes are not intended to limit the present invention.

[0079] like Figure 1 As shown, this embodiment provides a gas pipeline network simulation analysis system, including a gas pipeline network data acquisition unit 10, a pipeline network topology structure building unit 20, a pipeline network data value influencing factor analysis unit 30, a gas pipeline network data prediction model building unit 40, a gas pipeline network data periodicity prediction unit 50, a gas pipeline network data control unit 60 and a risk warning unit 70;

[0080] The gas network data collection unit 10 collects corresponding gas network data through the SCADA monitoring system on the gas network;

[0081] The pipe network topology structure building unit 20 builds each component of the pipe network according to the relationship and setting position of the SCADA monitoring system collection equipment on the gas pipe network;

[0082] The pipeline network data value influencing factor analysis unit 30 performs correlation analysis on each single type of gas pipeline network data collected to obtain relevant data on influencing factors affecting changes in gas pipeline network data;

[0083] The gas network data prediction model building unit 40 selects a time recurrent neural network model (LSTM) as the gas network data prediction model, and determines an input gate, a forget gate, and an output gate of the time recurrent neural network model;

[0084] The gas network data periodicity prediction unit 50 completes the prediction of the gas network data periodicity by adjusting the input gate value of the time recurrent neural network model within a preset period;

[0085] The risk warning unit 70 compares the real data value of the collected gas pipeline network data within the period with the preset threshold value to determine whether the gas pipeline network data should be warned of risk;

[0086] The gas network data control unit 60 controls the gas network data when it exceeds the real data value of the preset threshold value by adjusting the valves on the branches and branch nodes in each component of the virtual network.

[0087] The components of a pipe network include branches, boundaries, and branch points.

[0088] For each pipe segment or non-pipe element on each branch, if it is a pipe segment, mesh it according to the spatial step size and pipe segment length. Let N be a positive integer, and record the spatial step size. Then the spatial grid node on the pipe segment is x j =jΔx, j=0,1…,N. If it is a non-pipe element, first generate a node upstream and downstream of the element. After all the upstream and downstream nodes of all non-pipe elements are generated, if two non-pipe elements are adjacent, they are merged to generate upstream nodes or downstream nodes until there is only one node between any adjacent non-pipe elements.

[0089] For the boundaries, the completeness of the pipe network shows that the upstream or downstream of each boundary must be connected to one of the nodes obtained from the branch sequence, that is, no new nodes are generated when the boundary is divided into space.

[0090] The situation is similar for branch points. No new nodes are generated when the space is divided. Each branch point must be connected to one of the nodes obtained in the branch sequence.

[0091] The gas pipeline network data includes methane gas concentration, pressure and flow, and a pipeline network monitoring point is set up at each branch point in the pipeline network.

[0092] like Figure 2 As shown in the figure, the detailed steps of correlation analysis for each single type of gas pipeline network data collected are:

[0093] S301, firstly select a single data type in the gas pipeline network data as an object to be analyzed;

[0094] S302, then selecting the gas pipe environment or weather data value that may affect a single data type as a correlation analysis object;

[0095] S303, using a correlation analysis method to obtain the correlation coefficient between the two variables of the object to be judged and the correlation analysis object;

[0096] S304, selecting a correlation analysis object with a stronger correlation according to the absolute value of the correlation coefficient as the relevant data of the influencing factors affecting the change of the gas pipeline network data;

[0097] S305. Finally, repeat the above steps until the correlation analysis of all different types of gas pipeline network data is completed.

[0098] The corresponding calculation formula of the correlation analysis method is:

[0099]

[0100] Where: r x,y ——Correlation coefficient between variables x and y, n——The number of variables x and y, x i ——the i-th value of variable x, - the mean of variables x, y i ——the i-th value of variable y, ——the mean value of variable y;

[0101] The range of the correlation coefficient is between [-1, 1]. The closer its absolute value is to 1, the stronger the correlation between the two is. If its absolute value is closer to 0, the weaker the correlation between the two is. The absolute value of the correlation coefficient in the interval [0, 0.3) indicates a very weak correlation. The absolute value of the correlation coefficient in the interval [0.3, 0.5) indicates a weak correlation. The absolute value of the correlation coefficient in the interval [0.5, 0.8) indicates a significant correlation. The absolute value of the correlation coefficient in the interval [0.8, 1) indicates a strong correlation.

[0102] The input parameters of the time-recurrent neural network model include the cell state C at the previous moment. t-1 , hidden state h t-1 and the current input information h t ; The output parameter is the information C at the current moment t and the hidden state h t ;

[0103] Forget gate: determine the cell state C at the previous moment t-1 The information that needs to be retained and the information that needs to be discarded; t-1 and x t Enter the forget gate and pass it to the Sigmoid function; the specific calculation formula is:

[0104] f t =σ(W f h t-1 +U f x t +b f )

[0105] Where: σ——activation function Sigmoid, output value range [0, 1], control information propagation;

[0106] W f , U f ——The weight matrix of the forget gate;

[0107] b f ——Forget gate bias;

[0108] Input gate: used to determine the importance of the gas network data input at the current moment, and retain the important gas network data at the current moment and use it to update the cell state; the calculation formula is:

[0109] i t =σ(W i h t-1 +U i x t +b i )

[0110]

[0111] Where: tanh - activation function, output value range [-1, 1];

[0112] W i , U i ——Input gate weight matrix;

[0113] W C , U C ——Candidate layer weight matrix;

[0114] b i , b C ——Input gate, candidate layer bias;

[0115] h t-1 and X t Enter the input gate and pass it to the Sigmoid function. At the same time, h t-1 and x t Enter the candidate layer and pass it to the function tanh. If the number multiplied by the output information of the candidate layer is 0, then this part of the current input information needs to be discarded; if the number multiplied by the output information of the candidate layer is 1, then this part of the current input information is completely retained and the cell state C is updated. t ; If the number multiplied by the candidate layer output information is between (0, 1), then this part of the current input information is not completely retained in the new cell state;

[0116] Output gate: confirm the current cell state C t How much information is lost to the value h of the hidden state at the next moment? t ; The calculation formula is:

[0117] O t =σ(W o h t-1 +U o x t +b t )

[0118] h t =tanh(C t )

[0119] Where: W O , U O ——The weight matrix of the output layer;

[0120] b t ——Output gate bias;

[0121] h t-1 and x t Enter the output gate and pass it to the sigmoid function. At the same time, the updated cell state C t Pass it to the function tanh. Multiply the output values ​​on both sides point by point to get the hidden state h at the current moment t In the time recurrent neural network model, the hidden state h at the current moment t and cell state C t The cell state C at the previous moment t-1 , hidden state h t-1 and the current input information x t Joint decision making;

[0122] Cell state: used to update and store information; specific formula:

[0123]

[0124] First, the cell state C at the previous moment t-1 And the vector f output by the forget gate t Multiply point by point, if it is multiplied by a value f close to 0 t , then this information needs to be discarded if it is multiplied by a value f close to 1 t , then the information is fully preserved if it is multiplied by the value f t If the value is between (0, 1), the information can only be partially retained; finally, the remaining value of the previous cell state and the output value of the input gate are added point by point to update the cell state at the current moment.

[0125] By adjusting the input gate value of the time recurrent neural network model within the preset period:

[0126] The historical gas network data collected within a preset period has multiple data values ​​as input gate values;

[0127] Sort multiple data values ​​and select the maximum and minimum values ​​as input gate values ​​of the time recurrent neural network model;

[0128] The input gate of the time recurrent neural network model inputs the maximum and minimum values, and obtains the maximum and minimum values ​​predicted by the output gate as the periodic preset thresholds of the gas pipeline network data. The preset threshold setting of the gas pipeline network data in the next cycle is realized, which is convenient for judging the comparison between the real data value and the preset threshold within the gas pipeline network data cycle.

[0129] like Figure 3 As shown in the figure, the gas network data exceeds the preset threshold value of the real data value and the control method is as follows:

[0130] S701, mapping the real data values ​​collected within the gas pipeline network data cycle to each component of the constructed pipeline network in a 1:1 ratio;

[0131] S702, adjusting the methane gas concentration, pressure and opening of the flow valves set on the branches and branch points;

[0132] S703, using the adjusted methane gas concentration, pressure and flow valve opening as the input gate value of the time recurrent neural network model to predict the gas pipeline network data at the next moment;

[0133] S704, determine whether the predicted value of the gas pipeline network data at the next moment is within the preset threshold; if it is within the preset threshold, enter step S705; if it is not within the preset threshold, repeat the previous step S703; continue to adjust the methane gas concentration, pressure and flow valve opening set on the branch and branch point until the predicted value of the gas pipeline network data at the next moment is within the preset threshold, then end the process.

[0134] S705, ending the adjustment process, and sending the corresponding methane gas concentration, pressure, and flow valve opening to the system administrator for gas network maintenance personnel to adjust and maintain parameters during network maintenance.

[0135] It also includes a gas network data preprocessing unit, which is communicatively connected to the gas network data acquisition unit 10 and is used for preprocessing the acquired gas network data.

[0136] The preprocessing methods for the collected gas pipeline network data include:

[0137] Identification and handling of missing data;

[0138] Identification and treatment of outlier data.

[0139] Detailed steps for identifying and processing missing data:

[0140] By comparing the number of data fields of different data types, determine whether a single data type is missing;

[0141] If there are any missing values, the data interpolation method is used to fill the missing values, or the entire piece of other gas pipeline network data collected at the corresponding time of the missing data is deleted.

[0142] Detailed steps for identifying and processing outlier data:

[0143] Identify outliers by checking whether the individual data under a single data type is out of the data range of the single data type or the collected data has a value of zero (except for zero temperature or zero natural gas usage, which are not identified as outliers). You can also use the box plot analysis method to identify outliers:

[0144] Sort a set of data from small to large values. Find the first quartile Q of the set of data 1 , median M and third quartile Q 3 ;

[0145] Let IQR=Q 3 -Q 1 , draw a rectangle with IQR (interquartile range) as the length of the rectangle, and the positions on both sides of the rectangle correspond to the first quartile Q of the group of data 1 and the third quartile Q 3 , draw a median line on the long side of the rectangle, corresponding to the median M of the group of data, this line is called the median line.

[0146] In Q 3 +1.5IQR and Q 1 Draw two straight lines at -1.5IQR to distinguish abnormal values ​​from normal values. The range between the two lines is called the inner limit. 3 +3IQR and Q 1 Draw two straight lines at -3IQR to distinguish between extreme outliers and mild outliers. 3 +3IQR and Q 1 The range outside the -3IQR is called the outer limit. Data outside the inner limit is considered an outlier, data between the inner and outer limits is considered mild outlier, and data outside the outer limit is considered extreme outlier.

[0147] The outlier data value is replaced by the data replacement method, or the entire gas pipeline network data collected at the corresponding time of the outlier data is deleted in its entirety.

[0148] Compared with the prior art, the present invention has the following beneficial effects:

[0149] The gas pipeline network simulation is realized by constructing the pipeline network topology structure in a virtual environment. At the same time, the time recurrent neural network model is used to realize the periodic prediction of the gas pipeline network data. The real data values ​​collected on the gas pipeline network are compared with the periodic prediction range values ​​of the gas pipeline network data to judge the risk of the gas pipeline network. If the real data value exceeds the periodic prediction range value of the gas pipeline network data, the valves on the branches and branch nodes in each component of the virtual pipeline network are adjusted to realize the regulation of the real data value of the gas pipeline network data exceeding the preset threshold.

[0150] The above is a detailed introduction to a gas pipeline network simulation analysis system provided by the present application. The description of the specific embodiments is only used to help understand the method and its core idea of ​​the present application. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the claims of the present application.

Claims

1. A gas pipeline network simulation analysis system, characterized in that: It comprises a gas pipeline network data acquisition unit (10), a pipeline network topology structure building unit (20), a pipeline network data value influencing factor analysis unit (30), a gas pipeline network data prediction model building unit (40), a gas pipeline network data periodicity prediction unit (50), a gas pipeline network data control unit (60), a risk warning unit (70) and a gas pipeline network data storage unit (80); The gas pipe network data collection unit (10) is in communication connection with the gas pipe network data storage unit (80), and collects corresponding gas pipe network data through the SCADA monitoring system on the gas pipe network; The pipe network topology structure building unit (20) is in communication connection with the gas pipe network data acquisition unit (10), and builds various components in the pipe network according to the relationship and setting position of the SCADA monitoring system acquisition equipment on the gas pipe network; The pipeline data value influencing factor analysis unit (30) is connected to the gas pipeline data collection unit (10) for communication, and performs correlation analysis on each single type of gas pipeline data collected to obtain data related to influencing factors affecting changes in the gas pipeline data; The gas pipeline network data prediction model building unit (40) is in communication connection with the pipeline network data value influencing factor analysis unit (30), selects a time recurrent neural network model as the gas pipeline network data prediction model, and determines an input gate, a forget gate, and an output gate of the time recurrent neural network model; The gas pipeline network data periodicity prediction unit (50) is connected in communication with the gas pipeline network data prediction model building unit (40), and completes the prediction of the gas pipeline network data periodicity by adjusting the input gate value of the time recurrent neural network model within a preset period; The risk warning unit (70) is in communication connection with the gas pipeline network data periodicity prediction unit (50), and determines whether the gas pipeline network data should be subject to risk warning by comparing the real data value within the collected gas pipeline network data period with a preset threshold value; The gas pipe network data control unit (60) is in communication connection with the risk warning unit (70), and controls the gas pipe network data when it exceeds the real data value of a preset threshold value by adjusting the valves on the branches and branch nodes in each component of the virtual pipe network; The gas pipe network data storage unit (80) is communicatively connected to the gas pipe network data periodicity prediction unit (50) and is used to store historical gas pipe network data.

2. A gas pipeline network simulation analysis system according to claim 1, characterized in that: The components of a pipe network include branches, boundaries, and branch points.

3. A gas pipe network simulation analysis system according to claim 2, characterized in that: The gas pipeline network data includes methane gas concentration, pressure and flow, and a pipeline network monitoring point is set up at each branch point in the pipeline network.

4. A gas pipeline network simulation analysis system according to claim 3, characterized in that: Detailed steps for correlation analysis of each single type of gas pipeline network data collected: Firstly, a single data type in the gas pipeline network data is selected as the object to be analyzed; Then, the gas pipe environment or weather data values ​​that may affect a single data type are selected as the objects of correlation analysis; Then, the correlation analysis method is used to obtain the correlation coefficient between the two variables of the object to be judged and the object of correlation analysis; According to the absolute value of the correlation coefficient, the correlation analysis object with stronger correlation is selected as the relevant data of the influencing factors affecting the change of gas pipeline network data; Finally, repeat the above steps until the correlation analysis of all different types of gas pipeline network data is completed.

5. A gas pipe network simulation analysis system according to claim 4, characterized in that: The corresponding calculation formula of the correlation analysis method is: Where: r x，y ——Correlation coefficient between variables x and y, n——The number of variables x and y, x i ——the i-th value of variable x, ——The average value of variables x, y i ——the i-th value of variable y, ——the mean value of variable y; The value range of the correlation coefficient is between [-1, 1].

6. A gas pipe network simulation analysis system according to claim 5, characterized in that: The input parameters of the time-recurrent neural network model include the cell state C at the previous moment. t-1 , hidden state h t-1 and the current input information h t ; The output parameter is the information C at the current moment t and the hidden state h t ; The temporal recurrent neural network model includes a forget gate, an input gate, and an output gate; Forget gate: determine the cell state C at the previous moment t-1 The information that needs to be retained and the information that needs to be discarded; t-1 and x t Enter the forget gate and pass it to the Sigmoid function; the specific calculation formula is: f t =σ(W f h t-1 +U f x t +b f ) Where: σ——activation function Sigmoid, output value range [0, 1], control information propagation; W f , U f ——The weight matrix of the forget gate; b f ——Forget gate bias; Input gate: used to determine the importance of the gas network data input at the current moment, and retain the important gas network data at the current moment and use it to update the cell state; the calculation formula is: i t =σ(W i h t-1 +U i x t +b i ) Where: tanh - activation function, output value range [-1, 1]; W i , U i ——Input gate weight matrix; W C , U C ——Candidate layer weight matrix; b i , b C ——Input gate, candidate layer bias; Output gate: confirm the current cell state C t How much information is lost to the value h of the hidden state at the next moment? t ; The calculation formula is: ON t =σ(W0h t-1 +U o x t +b t ) h t =tanh(C t ) Where: W O , U O ——The weight matrix of the output layer; b t ——Output gate bias; Cell state: used to update and store information; specific calculation formula:

7. A gas pipe network simulation analysis system according to claim 6, characterized in that: By adjusting the input gate value of the time recurrent neural network model within the preset period: The historical gas network data collected within a preset period has multiple data values ​​as input gate values; Sort multiple data values ​​and select the maximum and minimum values ​​as input gate values ​​of the time recurrent neural network model; The maximum and minimum values ​​of the input gate of the time recurrent neural network model are input, and the maximum and minimum values ​​predicted by the output gate are obtained as the periodic preset thresholds of the gas network data.

8. A gas pipe network simulation analysis system according to claim 7, characterized in that: The gas network data exceeds the preset threshold value and the real data value is regulated: S701, mapping the real data values ​​collected within the gas pipeline network data cycle to each component of the constructed pipeline network in a 1:1 ratio; S702, adjusting the methane gas concentration, pressure and opening of the flow valves set on the branches and branch points; S703, using the adjusted methane gas concentration, pressure and flow valve opening as the input gate value of the time recurrent neural network model to predict the gas pipeline network data at the next moment; S704, determining whether the predicted value of the gas network data at the next moment is within a preset threshold; If it is within the preset threshold, proceed to step S705; if it is not within the preset threshold, repeat the previous step S703; S705, ending the adjustment process, and sending the corresponding methane gas concentration, pressure, and flow valve opening to the system administrator for gas network maintenance personnel to adjust and maintain parameters during network maintenance.

9. A gas pipe network simulation analysis system according to claim 8, characterized in that: It also includes a gas pipeline network data preprocessing unit, which is in communication connection with the gas pipeline network data acquisition unit (10) and is used to preprocess the acquired gas pipeline network data; The preprocessing methods for the collected gas pipeline network data include: Identification and handling of missing data: By comparing the number of data fields of different data types, determine whether a single data type is missing; If missing data occurs, data interpolation is used to fill the missing values, or the entire gas pipeline network data collected at the corresponding time of the missing data is deleted; Identification and processing of outlier data: Use box plot analysis to identify outliers: Sort a set of data from small to large values, and find the first quartile Q1, median M, and third quartile Q3 of the data; Let IQR = Q3-Q1, draw a rectangle with the IQR interquartile range as the length of the rectangle, the positions of the two sides of the rectangle correspond to the first quartile Q1 and the third quartile Q3 of the data set, and draw a midline on the long side of the rectangle, corresponding to the median M of the data set. This line is called the median line; Draw two straight lines at Q3+1.5IQR and Q1-1.5IQR respectively. The two straight lines are used to distinguish outliers from normal values. The range between the two lines is called the inner limit. Draw two straight lines at Q3+3IQR and Q1-3IQR to distinguish the boundary between extreme outliers and mild outliers. The range outside Q3+3IQR and Q1-3IQR is called the outer limit. Data outside the inner limit is considered an outlier, data between the inner and outer limits is mild outlier data, and data in the outer limit is extreme outlier data. Processing of outlier data: Use data replacement method to replace outlier data values, or delete the entire line of other gas pipeline network data collected at the corresponding time of the outlier data.

Images