Fault prediction method and system based on probabilistic graph network model

Abstract

The invention discloses a fault prediction method and system based on a probabilistic graph network model, and belongs to the technical field of power distribution system operation control. The invention provides a fault prediction method based on a probabilistic graph network model. The method comprises the following steps: establishing a probabilistic graph network model for the topology of a power distribution network; based on the probabilistic graph network model, learning fusion data of the power distribution network, and determining early warning hidden nodes of the power distribution network; and based on the early warning hidden node, predicting the fault of the power distribution network node. According to the method, data characteristics can be comprehensively understood by learning the power grid fusion data, so that fault nodes can be determined to predict faults.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution system operation control, and more specifically, to a fault prediction method and system based on a probability graph network model. Background Art

[0002] With the rapid growth of electricity demand and the accelerated transformation of energy structure, new power systems are being built rapidly. A large number of new loads dominated by power electronic inverters are continuously connected to the power grid, which breaks the original hierarchical structure of the power grid, diversifies the system operation mode, and makes it difficult to enumerate fault scenarios. As the "nerve endings" of the source, grid, load and storage of the new power system, the distribution network directly faces power users and plays an important role in ensuring power supply, supporting economic and social development, and serving to improve people's livelihood. At present, traditional distribution network data monitoring and analysis methods still have problems such as incomplete data resources, inaccurate fault analysis and difficult evaluation of operating status for supply and service command and coordination and fault emergency repair services. It is urgent to use cutting-edge information technology to improve the effective perception, monitoring, analysis and judgment capabilities of the distribution network, thereby improving the operating efficiency and stability of the distribution network.

[0003] As a cutting-edge information technology, digital twin technology can accurately map the physical distribution network in the virtual space, realize real-time two-way exchange of data, and enable the distribution network digital twin to fully reflect the life cycle of its corresponding physical entity. It can realize dynamic perception of business scenarios based on the integration of underlying data information, and effectively support typical practical business applications such as large-scale state deduction of distribution networks, analysis of the impact range of sudden faults, analysis and location of fault points, emergency repair route planning, and power outage restoration time prediction.

[0004] Data feature learning methods can analyze the operating status of distribution networks and predict future operating trends by extracting features from business data. At present, the amount of business data in power grids is huge and diverse. When faced with complex, multi-factor, and multi-level system failures in new power systems, the computational efficiency and prediction accuracy of the training model are greatly challenged. Summary of the invention

[0005] In view of the above problems, the present invention proposes a fault prediction method based on a probabilistic graph network model, comprising:

[0006] Establish a probabilistic graph network model for the topology of the distribution network;

[0007] Based on the probabilistic graph network model, the distribution network fusion data is learned to determine the early warning hidden nodes of the distribution network;

[0008] Based on the early warning hidden node, the fault of the distribution network node is predicted.

[0009] Optionally, a probabilistic graph network model is established for the topology of the distribution network, including:

[0010] Acquire a physical model of the distribution network topology, generate an electrical network model based on the physical model, number each node in the electrical network model, simplify the connection line of each node through the node numbering, so as to generate a distribution network topology graphical model;

[0011] Introducing the remote signaling state of the distribution network switch device node into the electrical network model, determining the open / closed state of the real-time switch of the distribution network through the remote signaling state, deleting the switch device node in the open state and the dynamic topological connection with the switch device node in the open state, and creating the switch device node in the closed state and the dynamic topological connection with the static topological connection node of the switch device node in the closed state, so as to generate a directed acyclic graph;

[0012] The directed acyclic graph is a probabilistic graph network model.

[0013] Optionally, based on the probabilistic graph network model, learning the distribution network fusion data to determine the early warning hidden nodes of the distribution network includes:

[0014] Divide the distribution network node data information into internal monitoring data and external model data, and extract key attributes of nodes in the internal monitoring data and external model data through principal component analysis, generate distribution network fusion data according to the key attributes of the nodes, and encode the distribution network fusion data;

[0015] Based on the probabilistic graph network model, the encoded distribution network fusion data is learned to set each possible fault in the distribution network as a node of the early warning layer, and the model parameters of the probabilistic graph network model are estimated by the maximum likelihood estimation method to determine the early warning hidden nodes.

[0016] Optionally, predicting the failure of the distribution network node based on the early warning hidden node includes:

[0017] The early warning hidden node is introduced into the probability graph network model. Based on the probability graph network model and the early warning hidden node, an expectation maximization EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node. If the parameter iteration value corresponding to the specific situation of the distribution network node is less than the parameter preset threshold, convergence is reached, the iterative process is terminated, and the node corresponding to the probability value corresponding to the specific situation greater than the probability preset threshold is taken as the fault node, and the fault is predicted based on the fault node.

[0018] Optionally, according to the early warning hidden node, an EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node, including:

[0019] According to the early warning hidden node, the EM algorithm is used for iterative calculation to determine the specific situation of the distribution network node. The calculation formula is as follows:

[0020]

[0021] Among them, q(θ t+1 |θ t ) is the expected log-likelihood function based on the distribution network nodes, m is the warning hidden node, n is the fault node, P(Z|θ t ) is the node state probability at the current stage, P(S,Z|θ t+1 ) is the failure probability of the next stage, Z is the node information, θ t The model parameters of the current stage, S is the node state category, θ t+1 is the model parameter of the next stage, i is the early warning hidden node in the range of [1, m];

[0022] Calculate q(θ t+1 |θ t ) takes its maximum value, θ t+1 The value is calculated as follows:

[0023]

[0024] in, is the updated parameter status information, k is the node of the distribution network, l is the warning hidden node in the range of [1, m], Z l To warn of hidden node information;

[0025] |θ t+1 -θ t |Iterate over the values ​​as parameters.

[0026] On the other hand, the present invention also proposes a fault prediction system based on a probabilistic graph network model, comprising:

[0027] A modeling unit, for establishing a probabilistic graphical network model for the topology of the distribution network;

[0028] A learning unit, used to learn the distribution network fusion data based on the probabilistic graph network model and determine the early warning hidden nodes of the distribution network;

[0029] A prediction unit is used to predict the failure of the distribution network node based on the early warning hidden node.

[0030] Optionally, a probabilistic graph network model is established for the topology of the distribution network, including:

[0031] Acquire a physical model of the distribution network topology, generate an electrical network model based on the physical model, number each node in the electrical network model, simplify the connection line of each node through the node numbering, so as to generate a distribution network topology graphical model;

[0032] Introducing the remote signaling state of the distribution network switch device node into the electrical network model, determining the open / closed state of the real-time switch of the distribution network through the remote signaling state, deleting the switch device node in the open state and the dynamic topological connection with the switch device node in the open state, and creating the switch device node in the closed state and the dynamic topological connection with the static topological connection node of the switch device node in the closed state, so as to generate a directed acyclic graph;

[0033] The directed acyclic graph is a probabilistic graph network model.

[0034] Optionally, based on the probabilistic graph network model, learning the distribution network fusion data to determine the early warning hidden nodes of the distribution network includes:

[0035] Divide the distribution network node data information into internal monitoring data and external model data, and extract key attributes of nodes in the internal monitoring data and external model data through principal component analysis, generate distribution network fusion data according to the key attributes of the nodes, and encode the distribution network fusion data;

[0036] Based on the probabilistic graph network model, the encoded distribution network fusion data is learned to set each possible fault in the distribution network as a node of the early warning layer, and the model parameters of the probabilistic graph network model are estimated by the maximum likelihood estimation method to determine the early warning hidden nodes.

[0037] Optionally, predicting the failure of the distribution network node based on the early warning hidden node includes:

[0038] The early warning hidden node is introduced into the probability graph network model. Based on the probability graph network model and the early warning hidden node, an expectation maximization EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node. If the parameter iteration value corresponding to the specific situation of the distribution network node is less than the parameter preset threshold, convergence is reached, the iterative process is terminated, and the node corresponding to the probability value corresponding to the specific situation greater than the probability preset threshold is taken as the fault node, and the fault is predicted based on the fault node.

[0039] Optionally, according to the early warning hidden node, an EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node, including:

[0040] According to the early warning hidden node, the EM algorithm is used for iterative calculation to determine the specific situation of the distribution network node. The calculation formula is as follows:

[0041]

[0042] Among them, q(θ t+1 |θ t ) is the expected log-likelihood function based on the distribution network nodes, m is the warning hidden node, n is the fault node, P(Z|θ t ) is the node state probability at the current stage, P(S,Z|θ t+1 ) is the failure probability of the next stage, Z is the node information, θ t The model parameters of the current stage, S is the node state category, θ t+1 is the model parameter of the next stage, i is the early warning hidden node in the range of [1, m];

[0043] Calculate q(θ t+1 |θ t ) takes its maximum value, θ t+1 The value is calculated as follows:

[0044]

[0045] in, is the updated parameter status information, k is the node of the distribution network, l is the warning hidden node in the range of [1, m], Z l To warn of hidden node information;

[0046] |θ t+1 -θ t |Iterate over the values ​​as parameters.

[0047] In yet another aspect, the present invention further provides a computing device, comprising: one or more processors;

[0048] a processor for executing one or more programs;

[0049] When the one or more programs are executed by the one or more processors, the above-described method is implemented.

[0050] In yet another aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed, the method described above is implemented.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] The present invention provides a fault prediction method based on a probability graph network model, comprising: establishing a probability graph network model for the topology of a distribution network; based on the probability graph network model, learning the distribution network fusion data to determine the early warning hidden nodes of the distribution network; based on the early warning hidden nodes, predicting the faults of the distribution network nodes. The present invention can fully understand the data characteristics by learning the power grid fusion data, so that the fault nodes can be determined to predict the faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 is a flow chart of the method of the present invention;

[0054] Figure 2 It is a principle diagram of an embodiment of the method of the present invention;

[0055] Figure 3 It is a structural diagram of the system of the present invention. DETAILED DESCRIPTION

[0056] Now, exemplary embodiments of the present invention are described with reference to the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely and to fully convey the scope of the present invention to those skilled in the art. The terms used in the exemplary embodiments shown in the accompanying drawings are not intended to limit the present invention. In the accompanying drawings, the same units / elements are marked with the same reference numerals.

[0057] Unless otherwise specified, the terms (including technical terms) used herein have the commonly understood meanings to those skilled in the art. In addition, it is understood that the terms defined in commonly used dictionaries should be understood to have the same meanings as those in the context of the relevant fields, and should not be understood as idealized or overly formal meanings.

[0058] Embodiment 1:

[0059] The present invention proposes a fault prediction method based on a probabilistic graph network model, such as Figure 1 As shown, including:

[0060] Step 1: Establish a probabilistic graph network model for the topology of the distribution network;

[0061] Step 2: Based on the probabilistic graph network model, the distribution network fusion data is learned to determine the early warning hidden nodes of the distribution network;

[0062] Step 3: Based on the early warning hidden node, predict the failure of the distribution network node.

[0063] Among them, a probabilistic graph network model is established for the topology of the distribution network, including:

[0064] Acquire a physical model of the distribution network topology, generate an electrical network model based on the physical model, number each node in the electrical network model, simplify the connection line of each node through the node numbering, so as to generate a distribution network topology graphical model;

[0065] Introducing the remote signaling state of the distribution network switch device node into the electrical network model, determining the open / closed state of the real-time switch of the distribution network through the remote signaling state, deleting the switch device node in the open state and the dynamic topological connection with the switch device node in the open state, and creating the switch device node in the closed state and the dynamic topological connection with the static topological connection node of the switch device node in the closed state, so as to generate a directed acyclic graph;

[0066] The directed acyclic graph is a probabilistic graph network model.

[0067] Wherein, based on the probabilistic graph network model, the distribution network fusion data is learned to determine the early warning hidden nodes of the distribution network, including:

[0068] Divide the distribution network node data information into internal monitoring data and external model data, and extract key attributes of nodes in the internal monitoring data and external model data through principal component analysis, generate distribution network fusion data according to the key attributes of the nodes, and encode the distribution network fusion data;

[0069] Based on the probabilistic graph network model, the encoded distribution network fusion data is learned to set each possible fault in the distribution network as a node of the early warning layer, and the model parameters of the probabilistic graph network model are estimated by the maximum likelihood estimation method to determine the early warning hidden nodes.

[0070] Wherein, based on the early warning hidden node, predicting the fault of the distribution network node includes:

[0071] The early warning hidden node is introduced into the probability graph network model. Based on the probability graph network model and the early warning hidden node, an expectation maximization EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node. If the parameter iteration value corresponding to the specific situation of the distribution network node is less than the parameter preset threshold, convergence is reached, the iterative process is terminated, and the node corresponding to the probability value corresponding to the specific situation greater than the probability preset threshold is taken as the fault node, and the fault is predicted based on the fault node.

[0072] According to the early warning hidden node, an EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node, including:

[0073] According to the early warning hidden node, the EM algorithm is used for iterative calculation to determine the specific situation of the distribution network node. The calculation formula is as follows:

[0074]

[0075] Among them, q(θ t+1 |θ t ) is the expected log-likelihood function based on the distribution network nodes, m is the warning hidden node, n is the fault node, P(Z|θ t ) is the node state probability at the current stage, P(S,Z|θ t+1 ) is the failure probability of the next stage, Z is the node information, θ t The model parameters of the current stage, S is the node state category, θ t+1 is the model parameter of the next stage, i is the early warning hidden node in the range of [1, m];

[0076] Calculate q(θ t+1 |θ t ) takes its maximum value, θ t+1 The value is calculated as follows:

[0077]

[0078] in, is the updated parameter status information, k is the node of the distribution network, l is the warning hidden node in the range of [1, m], Z l To warn of hidden node information;

[0079] |θ t+1 -θ t |Iterate over the values ​​as parameters.

[0080] The present invention is further described below with reference to specific cases:

[0081] Specific examples include Figure 2 As shown, including:

[0082] The specific steps are as follows:

[0083] Distribution network topology modeling:

[0084] The number of device nodes in the distribution network usually reaches tens of millions, including tens of thousands of nodes and branches, which makes the multi-node analysis of the power grid topology very complicated. It is necessary to abstract the actual distribution system according to the topological structure characteristics of the distribution network to achieve unified management of the distribution nodes. The specific steps are as follows:

[0085] 1. Generate an electrical network model from the physical model of the distribution network, ignore the properties of the electrical component nodes, number each node in the distribution network, and simplify the connection lines in the nodes to obtain the corresponding distribution network topology graphic model.

[0086] 2. Introduce the remote signaling status of the real-time switch of the distribution system, determine the open and close states by monitoring the real-time remote signaling status of the switch, delete the dynamic topological connection between the switch device node in the open state and its topological connection node, and create a dynamic topological connection between the switch device node in the closed state and its static topological connection node.

[0087] 3. The power grid topology model is represented as a directed acyclic graph G(V,E), where V = {1,2,...,N} represents the set of device nodes, and E = { <V i ,V j >,..., <V k ,V l >} represents the set of running lines. The adjacency matrix of the topological nodes of the distribution network is constructed according to the distribution network information and stored in the graph database.

[0088] Distribution network fusion feature learning:

[0089] At present, most feature learning methods mainly use the characteristics of the nodes themselves to learn the properties of the nodes, but this learning method ignores the topological dependence of the nodes to a certain extent, resulting in unsatisfactory algorithm effects. The present invention constructs a probabilistic graph model to learn the representation of the global nodes of the digital twin of the distribution network in high-dimensional situations.

[0090] 1. The directed network G(V,E) is used as the model input, and the node data information is divided into two categories: internal monitoring data and external model data. The principal component analysis method is used to extract the key attributes of the nodes, generate the attribute feature vector of the distribution network node, and encode the feature vector of each distribution network node.

[0091] 2. Introduce warning hidden node Z in the network i , each possible fault in the distribution network is set as a node in the early warning layer, and the model parameters are estimated by the maximum likelihood estimation method. The specific formula is as follows:

[0092]

[0093] Among them, S i is the node data information in the topological network, par(S i )) is the same as S i All random variables associated with the node. The value of the model parameter θ is obtained by maximizing the above formula:

[0094] θ ijk=argmaxL(θ|S i = k, par(S i )=j)

[0095] Among them, k represents the code of this node, j represents the code of the parent node. After the parameters of the early warning layer model are determined, the value of L at each node is calculated and compared with the set threshold to screen the early warning hidden node Z. i , and pay attention to the early warning hidden nodes.

[0096] 3. Through the early warning hidden node Z i The information is used to predict the faulty node. Starting from each hidden node in the early warning layer, the EM algorithm is used based on the iterative idea to evaluate the specific situation of the faulty node. The specific steps are as follows:

[0097]

[0098] Calculate q(θ t+1 |θ t ) takes the maximum value of the parameter θ t+1 , we can get the following formula:

[0099]

[0100] During the model operation, the parameter threshold ε is set to judge. When |θ t+1 -θ t When |<ε, the requirement is considered to have been met, the algorithm will terminate the iteration process, and pay attention to the nodes whose node failure probability exceeds the set probability threshold μ, thereby completing the fault prediction of the distribution network nodes and reducing the power outage time and power supply risk caused by the fault.

[0101] The above technical solution mainly includes:

[0102] The distribution network topology model is generated from the distribution network physical model, and each node in the distribution network is numbered. At the same time, the remote signaling status of the real-time switch of the distribution system is introduced. The opening and closing states are determined by monitoring the real-time remote signaling status of the switch. The dynamic topological connection between the switch device node in the opening state and its topological connection node is deleted, and the dynamic topological connection between the switch device node in the closing state and its static topological connection node is established.

[0103] The data types of the distribution network digital twin device nodes are classified into two parts: internal monitoring data and external model data. According to the data usage, the data are divided into three parts: training set, test set and validation set. The training set is used to train and build the model, the test set is used for actual prediction of the model, and the validation set is used to update and adjust the weights of the model.

[0104] Feature extraction is performed on the node data of distribution network equipment, and the principal component analysis method is used to reduce the dimension of node features to obtain key feature information vectors. These processed data points are used as model input information for subsequent processes.

[0105] The probability graph model is used to calculate the probability of failure of the early warning hidden node in the network. Each possible failure of the distribution network is set as a node in the early warning layer, and the relevant early warning layer nodes are found and represented by weight coefficients. The probability of node failure is compared with the threshold, and the nodes with high probability are selected.

[0106] By predicting the failure of the nodes selected by the early warning layer within the subsequent time range and learning the characteristic information of the early warning nodes, it is predicted whether the nodes will fail in the future.

[0107] Relevant operators analyze the predicted fault information and inspect the on-site situation to ensure the efficiency of fault handling.

[0108] The present invention can effectively improve the computational efficiency of the algorithm: compared with the existing feature learning methods, the present invention can more efficiently form a distribution network topology graph model for the changes and updates of the distribution network topology nodes through the dynamic topological connection of the nodes. At the same time, the introduction of the early warning layer can avoid the subsequent calculation of other irrelevant hidden nodes, effectively improve the computational efficiency and fault location efficiency of the algorithm model, and realize efficient calculation of the model.

[0109] The present invention can improve the prediction accuracy of the algorithm: compared with the existing feature learning methods, the present invention extracts features from the internal monitoring data and external model data of the distribution network digital twin to form node data information, avoids overfitting caused by high-dimensional information, and combines the probabilistic graphical model with the distribution scenario to analyze the data information and dependencies within and between the distribution network equipment nodes, so as to have a more comprehensive understanding of the data characteristics, thereby converging to the optimal solution more quickly and effectively improving the prediction performance of the algorithm.

[0110] Embodiment 2:

[0111] The present invention also proposes a fault prediction system 200 based on a probability graph network model, such as Figure 3 As shown, including:

[0112] A modeling unit 201, used to establish a probabilistic graph network model for the topology of the distribution network;

[0113] A learning unit 202, configured to learn the distribution network fusion data based on the probabilistic graph network model and determine the early warning hidden nodes of the distribution network;

[0114] The prediction unit 203 is used to predict the fault of the distribution network node based on the early warning hidden node.

[0115] Among them, a probabilistic graph network model is established for the topology of the distribution network, including:

[0116] Acquire a physical model of the distribution network topology, generate an electrical network model based on the physical model, number each node in the electrical network model, simplify the connection line of each node through the node numbering, so as to generate a distribution network topology graphical model;

[0117] Introducing the remote signaling state of the distribution network switch device node into the electrical network model, determining the open / closed state of the real-time switch of the distribution network through the remote signaling state, deleting the switch device node in the open state and the dynamic topological connection with the switch device node in the open state, and creating the switch device node in the closed state and the dynamic topological connection with the static topological connection node of the switch device node in the closed state, so as to generate a directed acyclic graph;

[0118] The directed acyclic graph is a probabilistic graph network model.

[0119] Wherein, based on the probabilistic graph network model, the distribution network fusion data is learned to determine the early warning hidden nodes of the distribution network, including:

[0120] Divide the distribution network node data information into internal monitoring data and external model data, and extract key attributes of nodes in the internal monitoring data and external model data through principal component analysis, generate distribution network fusion data according to the key attributes of the nodes, and encode the distribution network fusion data;

[0121] Based on the probabilistic graph network model, the encoded distribution network fusion data is learned to set each possible fault in the distribution network as a node of the early warning layer, and the model parameters of the probabilistic graph network model are estimated by the maximum likelihood estimation method to determine the early warning hidden nodes.

[0122] Wherein, based on the early warning hidden node, predicting the fault of the distribution network node includes:

[0123] The early warning hidden node is introduced into the probability graph network model. Based on the probability graph network model and the early warning hidden node, an expectation maximization EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node. If the parameter iteration value corresponding to the specific situation of the distribution network node is less than the parameter preset threshold, convergence is reached, the iterative process is terminated, and the node corresponding to the probability value corresponding to the specific situation greater than the probability preset threshold is taken as the fault node, and the fault is predicted based on the fault node.

[0124] According to the early warning hidden node, an EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node, including:

[0125] According to the early warning hidden node, the EM algorithm is used for iterative calculation to determine the specific situation of the distribution network node. The calculation formula is as follows:

[0126]

[0127] Among them, q(θ t+1 |θ t ) is the expected log-likelihood function based on the distribution network nodes, m is the warning hidden node, n is the fault node, P(Z|θ t ) is the node state probability at the current stage, P(S,Z|θ t+1 ) is the failure probability of the next stage, Z is the node information, θ t The model parameters of the current stage, S is the node state category, θ t+1 is the model parameter of the next stage, i is the early warning hidden node in the range of [1, m];

[0128] Calculate q(θ t+1 |θ t ) takes its maximum value, θ t+1 The value is calculated as follows:

[0129]

[0130] in, is the updated parameter status information, k is the node of the distribution network, l is the warning hidden node in the range of [1, m], Z l To warn of hidden node information;

[0131] |θ t+1 -θ t |Iterate over the values ​​as parameters.

[0132] The present invention can fully understand the data characteristics by learning the power grid fusion data, so that the fault node can be determined to predict the fault.

[0133] Embodiment 3:

[0134] Based on the same inventive concept, the present invention also provides a computer device, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding functions, so as to implement the steps of the method in the above embodiment.

[0135] Embodiment 4:

[0136] Based on the same inventive concept, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both a built-in storage medium in a computer device and an extended storage medium supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of the method in the above embodiment.

[0137] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0138] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0139] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0140] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0141] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0142] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A fault prediction method based on a probabilistic graph network model, characterized in that: include: Establish a probabilistic graph network model for the topology of the distribution network; Based on the probabilistic graph network model, the distribution network fusion data is learned to determine the early warning hidden nodes of the distribution network; Based on the early warning hidden node, the fault of the distribution network node is predicted.

2. The fault prediction method according to claim 1, characterized in that: The method of establishing a probabilistic graph network model for the topology of the distribution network includes: Acquire a physical model of the distribution network topology, generate an electrical network model based on the physical model, number each node in the electrical network model, simplify the connection line of each node through the node numbering, so as to generate a distribution network topology graphical model; Introducing the remote signaling state of the distribution network switch device node into the electrical network model, determining the open / closed state of the real-time switch of the distribution network through the remote signaling state, deleting the switch device node in the open state and the dynamic topological connection with the switch device node in the open state, and creating the switch device node in the closed state and the dynamic topological connection with the static topological connection node of the switch device node in the closed state, so as to generate a directed acyclic graph; The directed acyclic graph is a probabilistic graph network model.

3. The fault prediction method according to claim 1, characterized in that: The method of learning the distribution network fusion data based on the probabilistic graph network model and determining the early warning hidden nodes of the distribution network includes: Divide the distribution network node data information into internal monitoring data and external model data, and extract key attributes of nodes in the internal monitoring data and external model data through principal component analysis, generate distribution network fusion data according to the key attributes of the nodes, and encode the distribution network fusion data; Based on the probabilistic graph network model, the encoded distribution network fusion data is learned to set each possible fault in the distribution network as a node of the early warning layer, and the model parameters of the probabilistic graph network model are estimated by the maximum likelihood estimation method to determine the early warning hidden nodes.

4. The fault prediction method according to claim 1, characterized in that: The predicting of the failure of the distribution network node based on the early warning hidden node includes: The early warning hidden node is introduced into the probability graph network model. Based on the probability graph network model and the early warning hidden node, an expectation maximization EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node. If the parameter iteration value corresponding to the specific situation of the distribution network node is less than the parameter preset threshold, convergence is reached, the iterative process is terminated, and the node corresponding to the probability value corresponding to the specific situation greater than the probability preset threshold is taken as the fault node, and the fault is predicted based on the fault node.

5. The method according to claim 4, characterized in that According to the early warning hidden node, an EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node, including: According to the early warning hidden node, the EM algorithm is used for iterative calculation to determine the specific situation of the distribution network node. The calculation formula is as follows: Among them, q(θ t+1 |θ t ) is the expected log-likelihood function based on the distribution network nodes, m is the warning hidden node, n is the fault node, P(Z|θ t ) is the node state probability at the current stage, P(S,Z|θ t+1 ) is the failure probability of the next stage, Z is the node information, θ t The model parameters of the current stage, S is the node state category, θ t+1 is the model parameter of the next stage, i is the early warning hidden node in the range of [1, m]; Calculate q(θ t+1 |θ t ) takes its maximum value, θ t+1 The value is calculated as follows: in, is the updated parameter status information, k is the node of the distribution network, l is the warning hidden node in the range of [1, m], Z l To warn of hidden node information; |θ t+1 -θ t |Iterate over the values ​​as parameters.

6. A fault prediction system based on a probabilistic graph network model, characterized in that: include: A modeling unit, for establishing a probabilistic graphical network model for the topology of the distribution network; A learning unit, used to learn the distribution network fusion data based on the probabilistic graph network model and determine the early warning hidden nodes of the distribution network; A prediction unit is used to predict the failure of the distribution network node based on the early warning hidden node.

7. The fault prediction system according to claim 6, characterized in that: The method of establishing a probabilistic graph network model for the topology of the distribution network includes: Acquire a physical model of the distribution network topology, generate an electrical network model based on the physical model, number each node in the electrical network model, simplify the connection line of each node through the node numbering, so as to generate a distribution network topology graphical model; Introducing the remote signaling state of the distribution network switch device node into the electrical network model, determining the open / closed state of the real-time switch of the distribution network through the remote signaling state, deleting the switch device node in the open state and the dynamic topological connection with the switch device node in the open state, and creating the switch device node in the closed state and the dynamic topological connection with the static topological connection node of the switch device node in the closed state, so as to generate a directed acyclic graph; The directed acyclic graph is a probabilistic graph network model.

8. The fault prediction system according to claim 6, characterized in that: The method of learning the distribution network fusion data based on the probabilistic graph network model and determining the early warning hidden nodes of the distribution network includes: Divide the distribution network node data information into internal monitoring data and external model data, and extract key attributes of nodes in the internal monitoring data and external model data through principal component analysis, generate distribution network fusion data according to the key attributes of the nodes, and encode the distribution network fusion data; Based on the probabilistic graph network model, the encoded distribution network fusion data is learned to set each possible fault in the distribution network as a node of the early warning layer, and the model parameters of the probabilistic graph network model are estimated by the maximum likelihood estimation method to determine the early warning hidden nodes.

9. The fault prediction system according to claim 6, characterized in that: The predicting of the failure of the distribution network node based on the early warning hidden node includes: The early warning hidden node is introduced into the probability graph network model. Based on the probability graph network model and the early warning hidden node, an expectation maximization EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node. If the parameter iteration value corresponding to the specific situation of the distribution network node is less than the parameter preset threshold, convergence is reached, the iterative process is terminated, and the node corresponding to the probability value corresponding to the specific situation greater than the probability preset threshold is taken as the fault node, and the fault is predicted based on the fault node.

10. The system according to claim 9, characterized in that According to the early warning hidden node, an EM algorithm is used to perform iterative calculation to determine the parameter iteration value corresponding to the specific situation of the distribution network node, including: According to the early warning hidden node, the EM algorithm is used for iterative calculation to determine the specific situation of the distribution network node. The calculation formula is as follows: Among them, q(θ t+1 |θ t ) is the expected log-likelihood function based on the distribution network nodes, m is the warning hidden node, n is the fault node, P(Z|θ t ) is the node state probability at the current stage, P(S,Z|θ t+1 ) is the failure probability of the next stage, Z is the node information, θ t The model parameters of the current stage, S is the node state category, θ t+1 is the model parameter of the next stage, i is the early warning hidden node in the range of [1, m]; Calculate q(θ t+1 |θ t ) takes its maximum value, θ t+1 The value is calculated as follows: in, is the updated parameter status information, k is the node of the distribution network, l is the warning hidden node in the range of [1, m], Z l To warn of hidden node information; |θ t+1 -θ t |Iterate over the values ​​as parameters.

11. A computer device, characterized in that: include: one or more processors; a processor for executing one or more programs; When the one or more programs are executed by the one or more processors, the method according to any one of claims 1 to 5 is implemented.

12. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, the method according to any one of claims 1 to 5 is implemented.

Images