An artificial intelligence monitoring method and system based on multiple types of monitoring data

Abstract

The application relates to an artificial intelligence monitoring method and system based on multiple types of monitoring data, which combines multiple types of gas pipe network monitoring data for artificial intelligence monitoring through de-typing processing based on global standard values and numerical intervals, effectively utilizes various types of monitoring data, improves monitoring efficiency, and enables various types of data to jointly participate in subsequent artificial intelligence prediction without losing information.

Description

Technical Field

[0001] This invention belongs to the field of automated monitoring technology for gas pipeline networks, and particularly relates to an artificial intelligence monitoring method and system based on multi-type monitoring data. Background Technology

[0002] Currently, natural gas consumption is growing rapidly, with demand from industry and power generation surging. As the demand for gas continues to increase, the gas pipelines in cities are becoming increasingly dense and complex. This places higher demands on the safety and maintenance of gas pipelines, as well as on meeting user needs. Monitoring gas pipeline pressure is a common method for gas pipeline safety. By analyzing the pressure in different monitoring areas of the gas pipeline at different times, information such as whether there are any anomalies and the types of anomalies can be obtained. This information allows for appropriate maintenance operations to be carried out to ensure the safety of gas supply and the gas pipeline network.

[0003] However, with advancements in technology, monitoring methods are becoming increasingly diverse, resulting in a wider variety of monitoring data types. Furthermore, the methods for analyzing this data are also expanding, with artificial intelligence (AI) being a popular approach. Currently, when using AI for gas pipeline network monitoring, only one type of monitoring data is typically selected. While this single type of data is often crucial, the lack of supporting data from other types leads to the loss of much important information. Simply performing calculations or manually combining independent monitoring results from different types is not an organic or efficient way to utilize diverse information. How to organically combine AI detection methods with multiple types of monitoring data remains a problem. This invention organically combines various types of gas pipeline network monitoring data for AI monitoring, effectively utilizing different data types and improving monitoring efficiency. Summary of the Invention

[0004] To address the aforementioned problems in the prior art, this invention proposes an artificial intelligence monitoring method and system based on multi-type monitoring data, the method comprising:

[0005] Step S1: For each type of monitoring data, de-type it to obtain a de-typed sequence; Step S1 specifically includes the following steps:

[0006] Step S11: For each type of monitoring data, arrange them in descending order to form a monitoring data sequence corresponding to each type, where the monitoring data sequence corresponding to the k-th type is... , This refers to the s-th monitoring data element corresponding to the s-th monitoring node in the monitoring data sequence, which is the monitoring data value of the s-th monitoring node.

[0007] Step S12: Obtain the distance sequence of the monitoring data sequence ;in: ; This is the s-th distance element in the distance sequence; Where: NALL is the total number of monitoring nodes;

[0008] Step S13: Based on distance sequence Calculate detyped sequence ;in: ; This refers to the s-th detyped element in the detyped sequence; in: It is the base value of the k-th type;

[0009] Step S2: The monitoring node is encoded in two dimensions, and the two-dimensional encoding corresponds to the monitoring data position of the monitoring node in the feature matrix;

[0010] Step S3: Construct a corresponding feature matrix for each type of monitoring data k. The element (i, j) in the feature matrix corresponding to the k-th type of monitoring data is set to the two-dimensional encoding of the monitoring node (i, j) in the detype sequence. The corresponding element value in;

[0011] Step S4: For the characteristic matrix of type k Perform gradient transformation processing and obtain type k gradient transformation vectors. ;

[0012] Step S5: Dynamically set the weight vector of the neural network model; each element in the weight vector corresponds to a type.

[0013] Step S6: Input the gradient change vector as input and the weight vector as the weights of the input parameters into the neural network model and obtain the output vector;

[0014] Step S7: Find the monitoring status comparison table in the output vector to obtain the monitoring results.

[0015] Furthermore, the monitoring status comparison table is a preset comparison table.

[0016] Furthermore, the monitored data types include pressure, temperature, sound, flow rate, noise level, images, and human-assessed parameters.

[0017] Furthermore, the numbering is based on the traffic volume of the monitoring node.

[0018] Furthermore, the monitoring data is preprocessed before alignment.

[0019] An artificial intelligence monitoring system based on multi-type monitoring data is provided, the system being used to implement the artificial intelligence monitoring method based on multi-type monitoring data.

[0020] Furthermore, the system is a monitoring system based on gas pipeline network data from a big data platform.

[0021] A processor for running a program, wherein the program executes the artificial intelligence monitoring method based on multi-type monitoring data during runtime.

[0022] A computer-readable storage medium includes a program that, when run on a computer, causes the computer to perform the artificial intelligence monitoring method based on multi-type monitoring data.

[0023] An execution device includes a processor coupled to a memory storing program instructions. When the program instructions stored in the memory are executed by the processor, the artificial intelligence monitoring method based on multi-type monitoring data is implemented.

[0024] Based on the same inventive concept, the present invention also provides a processor for running a program, wherein the program executes the artificial intelligence monitoring method based on multi-type monitoring data during runtime.

[0025] The beneficial effects of this invention include:

[0026] (1) Based on the flux value, the monitoring nodes are encoded with QR codes, so that the relationship between nodes in the gas pipeline network structure is substituted into the subsequent calculation. By de-classifying based on global standard values ​​and numerical intervals, the differences of sample data within the same time range are maintained; so that various types of data can participate in subsequent artificial intelligence prediction without losing information. (2) Based on the feature matrix formed by the de-classified monitoring data, the two-dimensional matrix is ​​converted into a one-dimensional vector while retaining the significance of the data through gradient changes. The data analysis dimension is greatly increased, so that various types of gas pipeline network monitoring data can be organically combined for artificial intelligence monitoring. (3) The binary values ​​represented by the output vector are pre-corresponded to the meaningless monitoring results to form a reference table. The meaning of the binary values ​​themselves is formed through training, which expands the range of the monitoring results. Attached Figure Description

[0027] The accompanying drawings, which are provided to further illustrate the invention and form part of this application, are not intended to unduly limit the invention. In the drawings:

[0028] Figure 1 This is a schematic diagram of the artificial intelligence monitoring method based on multi-type monitoring data according to the present invention.

[0029] Figure 2 This is a schematic diagram illustrating the calculation method of the k-type gradient change vector in an embodiment of the present invention. Detailed Implementation

[0030] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. The illustrative embodiments and descriptions are only used to explain the present invention and are not intended to limit the present invention.

[0031] As attached Figure 1 As shown, this invention proposes an artificial intelligence monitoring method based on multi-type monitoring data, the method comprising the following steps:

[0032] Step S1: For each type of monitoring data, detype it to obtain a detyped sequence; that is, without losing the data content information related to the type, it makes it easier to make a final comprehensive consideration of various types of data without causing excessive training and loss of accuracy.

[0033] Preferably, the monitored data types include pressure, temperature, sound, flow rate, noise level, images, and human-assessed parameters;

[0034] Preferably, the monitoring data is preprocessed before alignment; the preprocessing includes one or more of the following: filling blank elements with default values, removing outliers, and performing normalization.

[0035] Step S1 specifically includes the following steps:

[0036] Step S11: For each type of monitoring data, arrange them in descending order to form a monitoring data sequence corresponding to each type, where the monitoring data sequence corresponding to the k-th type is... , This refers to the s-th monitoring data element corresponding to the s-th monitoring node in the monitoring data sequence, which is the monitoring data value of the s-th monitoring node, such as the pressure parameter value.

[0037] Step S12: Obtain the distance sequence of the monitoring data sequence ;in: ; This is the s-th distance element in the distance sequence; Where: NALL is the total number of monitoring nodes;

[0038] Step S13: Based on distance sequence Calculate detyped sequence ;in: ; This refers to the s-th detyped element in the detyped sequence; in: It is the baseline value of type k; this baseline value can be set to the historical mean; this baseline value only needs to remain unchanged during the use of the model, and when the model is retrained, this baseline value can be changed according to the distribution of the selected monitoring data;

[0039] Through the above-mentioned de-typing process, this invention enables various types of data to participate in subsequent artificial intelligence predictions without losing information; various types of data maintain the differences of sample values ​​over historical usage time through relative data values, while incorporating span information to maintain the differences of sample data within the same time range.

[0040] Step S2: Perform two-dimensional encoding on the monitoring nodes, where the two-dimensional encoding corresponds to the monitoring data position of the monitoring nodes in the feature matrix; specifically, this includes the following steps:

[0041] Step S21: Obtain the connectivity of each monitoring node. and average flow value And calculate the flux value for each monitoring node. Where: the connectivity number is the number of other monitoring nodes that are directly connected to the monitoring node in the gas pipeline topology graph, which is the number of connecting edges;

[0042] Step S23: Set the number of rows and columns for the two-dimensional code; specifically: set the number of rows and columns to be equal. ;

[0043] Step S24: Assign monitoring nodes according to... The data are encoded in a row-first, column-second manner from largest to smallest; if there are 24 monitoring nodes, then... The two-dimensional code for the largest is (1,1). The two-dimensional code for the smallest one is (5,4);

[0044] Step S3: Construct a corresponding feature matrix for each type of monitoring data k. The element (i, j) in the feature matrix corresponding to the k-th type of monitoring data is set to the two-dimensional encoding of the monitoring node (i, j) in the detype sequence. The corresponding element value in the sequence; for example: the type sequence. The 24th element is placed in position (5,4) in the feature matrix; at this time, the monitoring area covered by a gas pipeline network structure corresponds to multiple feature matrices;

[0045] Step S4: For the characteristic matrix of type k Perform gradient transformation processing and obtain type k gradient transformation vectors. Converting a two-dimensional matrix into a one-dimensional vector reduces the dimensionality of data analysis.

[0046] Step S4 specifically includes the following steps:

[0047] Step S41: Set the feature matrix Initialize a specific element;

[0048] Preferably, the initial specific element is a feature matrix. The first element in;

[0049] Step S42: From the feature matrix Specific elements Begin by calculating the gradient element of the next row corresponding to the direction of the maximum row gradient. ;in: These are the row and column numbers of a specific element; It is the characteristic matrix The element in row u and column v;

[0050] ;

[0051] The corresponding element;

[0052] Step S43: If there is a next adjacent column in the current row, start from the next row's gradient element. Start by calculating the gradient element of the next column corresponding to the gradient direction of the largest adjacent column. ; Take the next column of gradient elements as specific elements and return to step S42; If there is no next adjacent column, proceed to step S44;

[0053] ;

[0054] The corresponding element;

[0055] Step S44: Starting from a specific element, according to the direction of the maximum row gradient or the direction of the maximum column gradient, sequentially set the gradient elements of the next row and the next column to form the gradient change vector. At this point, the number of elements in the gradient change vector is 2N.

[0056] Alternatives: as attached Figure 2As shown, step S44 specifically involves: starting from a specific element, sequentially setting the next column of gradient elements according to the direction of the maximum column gradient to form a gradient change vector. At this point, the number of elements in the gradient change vector is N; this setting can further reduce the amount of input data.

[0057] Step S5: Dynamically set the weight vector of the neural network model; each element in the weight vector corresponds to a type.

[0058] Preferably, the dynamic setting of the weight vector of the neural network model specifically involves setting the weights based on the completeness of the collected monitoring data.

[0059] Preferably, the dynamic setting of the weight vector of the neural network model is specifically: the weight vector is set according to the importance of the monitored data type, and the more important the type, the higher the value of the corresponding element in the weight vector;

[0060] Replaceable: Set weight vectors based on the importance of the monitored data type and the monitoring frequency of the monitored data; the monitoring frequency and type are related to existing attributes. If the monitored data itself is not easily changed, it does not need to be monitored frequently. In addition, due to the limitations of monitoring conditions, some parameters cannot be monitored frequently, and therefore the monitoring frequency requirements cannot be met.

[0061] When weighted calculation is used, a single hidden layer of neurons can achieve better training and prediction results, thus greatly improving the training speed of the network.

[0062] Step S6: Input the gradient change vector as input and the weight vector as the weights of the input parameters into the neural network model and obtain the output vector;

[0063] The neural network model comprises an input layer, hidden layers, and an output layer. The input vector propagates forward to the neurons in the hidden layer, undergoes activation function operations, and then propagates the output information from the hidden layer neurons to the output nodes, ultimately producing an output vector. The input layer has K input layer neurons, where K is the number of monitored data types. Each input layer neuron receives an input vector of the corresponding type, and the input vector and weight vector are fully connected to each neuron in the hidden layer.

[0064] Preferably, the activation function is the Sigmoid function;

[0065] Preferably, the action function of the hidden layer neuron node is a radial basis function;

[0066] Preferably, the hidden layer is one or more layers;

[0067] Step S7: Obtain the monitoring results by searching the monitoring status lookup table in the output vector; specifically: the output vector is an N-ary output vector, and each element in the output vector corresponds to the output of an output neuron. That is, the number of output layer neurons is N, which is the same as the number of monitoring types; the binary values ​​of the output vector correspond to one or more monitoring results, thereby forming a lookup table, which is preset.

[0068] If the meaning of each element in the input vector is represented separately, so that each element corresponds to the monitoring result of the input vector data, then simple judgment is still required, such as voting or manual judgment, to determine the final monitoring result. This invention pre-assigns the binary value represented by the output vector to the meaningless monitoring result to form a lookup table. Through training, the meaning of the binary value itself is formed. On the one hand, this expands the range of monitoring results, and on the other hand, the neural network model organically combines the meaning of multiple types of data.

[0069] Preferably, the weight vector calculated in step S5 is the initial value of the weight of the input layer of the neural network model. During the training process, the network is trained by input and output sample sets, thereby correcting the network threshold in the neural network model. The current value of the weight vector is continuously corrected through the feedback mechanism so that the neural network model can realize the mapping relationship between the given input and output, that is, achieve the accuracy requirement and be able to make subsequent predictions.

[0070] A computer program (also referred to as a program, software, software application, script, or code) can be written in any form of programming language, including assembly or interpreted languages, declarative or procedural languages, and can be deployed in any form, including as a standalone program or as a module, component, subroutine, object, or other unit suitable for use in a computing environment. A computer program may, but does not necessarily, correspond to a file in a file system. A program can be stored as part of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to said program, or in multiple co-located files (e.g., a file storing one or more modules, subroutines, or code portions). A computer program can be deployed to execute on a single computer or on multiple computers located at a single site or distributed across multiple sites and interconnected by a communications network.

[0071] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0072] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0073] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0074] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0075] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. An artificial intelligence monitoring method for gas pipeline networks based on multi-type monitoring data, characterized in that, include: Step S1: For each type of monitoring data, detype it to obtain a detyped sequence; the monitoring data types include pressure, temperature, sound, flow rate, noise level, image and / or human evaluation parameters; Step S1 specifically includes the following steps: Step S11: For each type of monitoring data, arrange them in descending order to form a monitoring data sequence corresponding to each type, where the monitoring data sequence corresponding to the k-th type is SEQ. k =(seq k,s ), seq k,s This refers to the s-th monitoring data element corresponding to the s-th monitoring node in the monitoring data sequence, which is the monitoring data value of the s-th monitoring node. Step S12: Obtain the distance sequence SP of the monitoring data sequence k ; Among them: SP k =(sp k,s );sp k,s sp is the s-th distance element in the distance sequence; k,s =seq k,s -seq k,s+1 ;s=1~(NALL-1);sp k,NALL =seq k,NALL -seq k,1 Where: NALL is the total number of monitoring nodes; Step S13: Based on the distance sequence SP k Calculate the detyped sequence DT k Among them: DT k =(dt) k,s );dt k,s This refers to the s-th detyped element in the detyped sequence; Among them: STD k This is the baseline value of type k; the baseline value is set to the historical mean. Step S2: The monitoring nodes are encoded in two dimensions, and the two-dimensional encoding corresponds to the monitoring data position of the monitoring nodes in the feature matrix; Step S3: Construct a corresponding feature matrix MT for each type of monitoring data k. k The element (i, j) in the feature matrix corresponding to the k-th type of monitoring data is set to the two-dimensional encoding of the monitoring node (i, j) in the detyped sequence DT. k The corresponding element value in; Step S4: For the characteristic matrix MT of type k k Perform gradient transformation processing and obtain type k gradient transformation vector V k ; Step S5: Dynamically set the weight vector of the neural network model; each element in the weight vector corresponds to a type; the dynamic setting of the weight vector of the neural network model specifically means: setting the weight vector according to the importance of the monitored data type, the more important the type, the higher the value of the corresponding element in the weight vector; Step S6: Input the gradient change vector as input and the weight vector as the weights of the input parameters into the neural network model and obtain the output vector; Step S7: Find the monitoring status comparison table in the output vector to obtain the monitoring results.

2. The artificial intelligence monitoring method for gas pipeline networks based on multi-type monitoring data according to claim 1, characterized in that, The monitoring status comparison table is a pre-set comparison table.

3. The artificial intelligence monitoring method for gas pipeline networks based on multi-type monitoring data according to claim 2, characterized in that, The numbering is based on the traffic volume of the monitored nodes.

4. The artificial intelligence monitoring method for gas pipeline networks based on multi-type monitoring data according to claim 3, characterized in that, The monitoring data is preprocessed before alignment.

5. A processor, characterized in that, The processor is used to run a program, wherein the program executes the artificial intelligence monitoring method for gas pipeline networks based on multi-type monitoring data as described in any one of claims 1-4.

6. A computer-readable storage medium, characterized in that, The program, when run on a computer, causes the computer to perform the artificial intelligence monitoring method for gas pipeline networks based on multi-type monitoring data as described in any one of claims 1-4.

7. An execution device, characterized in that, The system includes a processor coupled to a memory, the memory storing program instructions, which, when executed by the processor, implement the gas pipeline network artificial intelligence monitoring method based on multi-type monitoring data as described in any one of claims 1-4.

Images