Intelligent fault location method and system for multi-branch power distribution lines

Abstract

The present invention provides an intelligent fault location method and system applicable to multi-branch power distribution lines, belonging to the technical field of distribution network line fault location technology. The method obtains the phase voltage and phase current waveform information at the head end of the primary branch line, as well as the effective current value collected by the smart meter at the end of the secondary branch line; based on the effective current value collected by the smart meter at the end of the secondary branch line, the fault line is determined; using a pre-trained distribution network fault distance measurement model, the obtained phase voltage and phase current waveform information at the head end of the primary branch line is processed to complete the fault point location distance measurement; the fault line judgment result and the fault point location distance measurement are combined to complete the fault location of the faulty multi-branch power distribution line. The present invention realizes the precise location of the fault point in the case of multi-branch power distribution lines, has strong universality, and is trained in combination with the actual distribution network topology diagram. The trained location network is extremely accurate and reliable, and has certain development prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network line fault location, and in particular to an intelligent fault location method and system applicable to multi-branch distribution lines. Background Art

[0002] The reliability of distribution networks is significantly affected by the duration of faults. Accurate and rapid fault location is key to faster fault clearance and reduced downtime, thereby improving system reliability. However, due to the multiple branches within a distribution network, its topology is complex. Distribution network equipment is relatively dispersed, making it impractical to install high-precision measurement devices at the end of each branch. Therefore, fault location in distribution networks remains a challenging problem.

[0003] Fault location techniques can be divided into three categories: impedance-based, traveling wave-based, and machine learning (ML)-based algorithms. Impedance-based methods use voltage and current measurements to estimate the fault location. These methods are easy to implement but are susceptible to multi-branch, metering errors, and system size. With the increasing demand for positioning accuracy, traveling wave-based methods are becoming more and more widely used in transmission systems. However, few studies have used traveling wave-based methods for fault location in distribution networks. The main reason is that traveling waves are prone to reflection and refraction when transmitted in multi-branch distribution networks, and the reflection and refraction behavior is complex. The detection of traveling wave heads and the precise fault location are very difficult. Some researchers have used microphasor measurement devices to improve positioning accuracy. These methods are less affected by fault type and resistance, but they require high-quality sampling rates, which are not common in distribution networks.

[0004] Currently, a large number of smart meters and intelligent electronic devices have been deployed in distribution networks. It is essential to fully leverage the power of intelligent measurement technology to improve the safe operation of the power industry. Supported by industrial informatization, distribution networks have accumulated a vast amount of sample data. These conditions provide opportunities for machine learning-based methods, which have emerged in the wake of the surge in artificial intelligence. Previous studies have used simulated data from various fault scenarios to train machine learning-based models, achieving superior localization results. This is because traditional fault localization methods make certain system assumptions about the distribution network, for example, neglecting to some extent the impact of fault type, resistance, and line branches on localization results. In this paper, these system assumptions are referred to as the "uncontrollable factors" of the physical model. Machine learning-based methods, on the other hand, account for these "uncontrollable factors" through hyperparameters, resulting in fitting results that are closer to the actual physical model. Summary of the Invention

[0005] The object of the present invention is to provide an intelligent fault location method and system applicable to multi-branch distribution lines, which fully considers the influence of measurement errors and line parameter errors and is more accurate in fault location of multi-branch distribution lines, so as to solve at least one of the technical problems existing in the above-mentioned background technology.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] On the one hand, the present invention provides an intelligent fault location method applicable to multi-branch distribution lines, including:

[0008] Using a pre-constructed distribution network model, performing fault simulation on the distribution network model to obtain the phase voltage and phase current waveform information at the head end of the primary branch line; obtaining the effective value of the current collected by the smart meter at the end of the secondary branch line;

[0009] [[ID=??]]Using a pre-trained distribution network fault location model to process the obtained phase voltage and phase current waveform information at the head end of the primary branch line to complete the fault point location ranging;

[0010] Based on the effective value of the current collected by the smart meter at the end of the secondary branch line, judge the faulty line;

[0011] Integrating the faulty line judgment result and the fault point location ranging to complete the fault location of the multi-branch distribution line.

[0012] Preferably, based on the effective value of the current collected by the smart meter at the end of the secondary branch line, judging the faulty line includes: collecting all currents I , k(p-1) , ,

[0014] , kp ,

[0013] (1 ≤ i ≤ n) of the smart meters at the ends of the secondary branches connected to the feeder to form a vector I o ; the vector I e represents the set of branches with only small currents, and its value is less than the threshold γ; 0 < p < n; N and O respectively represent the nodes on the main branch and the far-end nodes of the secondary branches. If p = q, it means that only one secondary branch has a fault, and the faulty line is N kp O kp ; when p ≠ q, it means that the main branch N k(p-1) N kp has a fault, which may cause multiple secondary branches to lose power, and the faulty line can be judged.

[0013] Preferably, the distribution network fault location model is composed of an autoencoder module and a long short-term memory network module in parallel; the autoencoder module is used for dimensionality reduction of waveform information; the long short-term memory network module is used to take the extracted time-related feature sequence as input and perform multiplication operations on the cells in different time periods. <000009Preferably, t AEs are connected in parallel to form an autoencoder module layer; the value of t is determined by the length of the input waveform; during the training of the autoencoder module layer, the labeled output of the t-th autoencoder has the same dimension as its input vector, and the loss function of the t-th autoencoder is:

[0015]

[0016] in Represents the dataset S t The reconstructed sample of the input sample in W ti represents the weight of the t-th autoencoder; H t is the number of neurons in the hidden layer; λ, α are regularization parameters for adjusting the weight of the penalty term; KL(ρ||ρ ti ) is the Kullback-Leibler divergence, ρ ti represents the average activation function value of hidden layer neuron i; ρ is the sparsity parameter.

[0017] Preferably, the LSTM module multiplies cells from different time periods, so the output or error at the previous time step is the same as the output at the next time step; the loss function used to train the LSTM network is the mean squared error:

[0018] Among them, S is the total number of training samples, is the estimation result given by the LSTM model, y i represents the true value of sample i.

[0019] Preferably, when training the regression distribution network fault location model, a loss function is defined to consider the influence of the autoencoder module and the long short-term memory network module: L = L PAE +L LSTM ;

[0020] Grid search and cross-validation methods are used to determine the hyperparameters, and the Sigmoid function is used as the activation function of the fully connected layer in the long short-term memory network module.

[0021] In a second aspect, the present invention provides an intelligent fault location system applicable to multi-branch power distribution lines, comprising:

[0022] The acquisition module is used to use the pre-built distribution network model to perform fault simulation on the distribution network model, obtain the phase voltage and phase current waveform information at the head end of the primary branch line, and obtain the effective value of the current collected by the smart meter at the end of the secondary branch line;

[0023] A judgment module, used to judge the fault line based on the effective value of the current collected by the smart meter at the end of the secondary branch line;

[0024] The distance measurement module is used to process the phase voltage and phase current waveform information obtained at the head end of the primary branch line using a pre-trained distribution network fault distance measurement model to complete the distance measurement of the fault point;

[0025] The positioning module is used to integrate the fault line judgment results and fault point location distance measurement to complete the fault positioning of the faulty multi-branch power distribution line.

[0026] In a third aspect, the present invention provides a non-transitory computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the intelligent fault location method applicable to multi-branch power distribution lines as described above is implemented.

[0027] In a fourth aspect, the present invention provides a computer program product, comprising a computer program, wherein when the computer program is run on one or more processors, the computer program is configured to implement the intelligent fault location method applicable to multi-branch power distribution lines as described above.

[0028] In a fifth aspect, the present invention provides an electronic device comprising: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes instructions for implementing the intelligent fault location method applicable to multi-branch power distribution lines as described above.

[0029] The beneficial effects of the present invention are as follows: by training a large number of high-dimensional samples, it is possible to achieve accurate distance measurement of fault points in multi-branch power distribution systems; it achieves accurate positioning of fault points in multi-branch power distribution lines and realizes its application in actual engineering projects; it has strong universality, and combined with actual distribution network topology training, the trained positioning network has extremely high accuracy and strong reliability, and has certain development prospects.

[0030] Additional advantages of the present invention will be more clearly given in the following description or learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0032] Figure 1 This is a flow chart of an intelligent fault location method applicable to multi-branch power distribution lines according to an embodiment of the present invention.

[0033] Figure 2 This is a schematic diagram of data collection according to an embodiment of the present invention.

[0034] Figure 3 This is a schematic diagram of a distribution network model according to an embodiment of the present invention.

[0035] Figure 4 This is a structural diagram of the fault location model constructed according to an embodiment of the present invention.

[0036] Figure 5 Schematic diagram of the distribution of the error of the AB-G fault according to the fault location according to an embodiment of the present invention.

[0037] Figure 6 Schematic diagram of the relationship between grounding resistance and fault phase angle according to an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention and are not to be construed as limiting the present invention.

[0039] Those skilled in the art will understand that unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which this invention belongs.

[0040] It should also be understood that terms, such as those defined in commonly used dictionaries, should be understood to have a meaning consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless as defined herein.

[0041] Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a," "an," "said," and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the stated features, integers, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, and / or groups thereof.

[0042] In the description of this specification, reference to the terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and integrate different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless otherwise inconsistent.

[0043] To facilitate understanding of the present invention, the present invention is further explained below with reference to specific embodiments in conjunction with the accompanying drawings. However, the specific embodiments do not constitute a limitation on the embodiments of the present invention.

[0044] Those skilled in the art should understand that the drawings are merely schematic diagrams of embodiments, and the components in the drawings are not necessarily necessary for implementing the present invention.

[0045] Example 1

[0046] In this embodiment 1, an intelligent fault location system applicable to multi-branch power distribution lines is first provided, including:

[0047] The acquisition module is used to use the pre-built distribution network model to perform fault simulation on the distribution network model, obtain the phase voltage and phase current waveform information at the head end of the primary branch line, and obtain the effective value of the current collected by the smart meter at the end of the secondary branch line;

[0048] The distance measurement module is used to process the phase voltage and phase current waveform information obtained at the head end of the primary branch line using a pre-trained distribution network fault distance measurement model to complete the distance measurement of the fault point;

[0049] A judgment module, used to judge the fault line based on the effective value of the current collected by the smart meter at the end of the secondary branch line;

[0050] The positioning module is used to integrate the fault line judgment results and fault point location measurement to complete the fault positioning of the faulty multi-branch power distribution line.

[0051] In this embodiment 1, the above-mentioned system is used to implement an intelligent fault location method applicable to multi-branch distribution lines, including:

[0052] Using the pre-built distribution network model, perform fault simulation on the distribution network model to obtain the phase voltage and phase current waveform information at the head end of the primary branch line; obtain the effective current value collected by the smart meter at the end of the secondary branch line;

[0053] Using a pre-trained fault location ranging model for distribution networks, process the phase voltage and phase current waveform information obtained at the head of the primary branch line to complete the fault location ranging for the fault point;

[0054] Based on the effective current values collected by the smart meters at the ends of the secondary branch lines, determine the faulty line;

[0055] Integrate the judgment result of the faulty line and the fault point location ranging to complete the fault location for the multi-branch distribution line with faults.

[0056] Based on the effective current values collected by the smart meters at the ends of the secondary branch lines, determining the faulty line includes: collecting all the currents I ki (1 ≤ i ≤ n) of the smart meters at the ends of the secondary branches connected to the feeder to form a vector I o ; The vector I e represents the set of branches with only small currents, whose values are less than the threshold γ; 0 < p < n; N and O respectively represent the nodes on the main branch and the distal nodes of the secondary branches. If p = q, it means that only one secondary branch has a fault, and the faulty line is N kp O kp ; When p ≠ q, it means that the main branch N k(p-1) N kp has a fault, which may cause multiple secondary branches to lose power, and thus the faulty line can be determined.

[0057] The fault location ranging model for distribution networks consists of an autoencoder module and a long short-term memory network module connected in parallel; the autoencoder module is used for dimensionality reduction of waveform information; the long short-term memory network module is used to perform multiplication operations on cells in different time periods with the extracted time-related feature sequences as input.

[0058] Connect t AEs in parallel to form an autoencoder module layer; the value of t is determined by the length of the input waveform; during the training process of the autoencoder module layer, the labeled output of the t-th autoencoder has the same dimension as its input vector, and the loss function of the t-th autoencoder is:

[0059]

[0060] where represents the reconstructed sample of the input sample in the dataset S t ; Wti represents the weight of the t-th autoencoder; H t is the number of neurons in the hidden layer; λ and α are regularization parameters for adjusting the penalty term weights; KL(ρ||ρ ti ) is the Kullback-Leibler divergence, ρ ti represents the average activation function value of the i-th neuron in the hidden layer; ρ is the sparsity parameter.

[0061] The LSTM module multiplies cells from different time periods, so the output or error at the previous time step is the same as the output at the next time step; the loss function used to train the LSTM network is the mean squared error:

[0062] Among them, S is the total number of training samples, is the estimation result given by the LSTM model, y i represents the true value of sample i.

[0063] When training the regression distribution network fault location model, the loss function is defined to consider the influence of the autoencoder module and the long short-term memory network module: L = L PAE +L LSTM ;

[0064] Grid search and cross-validation methods are used to determine the hyperparameters, and the Sigmoid function is used as the activation function of the fully connected layer in the long short-term memory network module.

[0065] Example 2

[0066] like Figures 1 to 4 As shown, in this embodiment 2, an intelligent fault location method applicable to multi-branch power distribution lines is provided to solve the problem that the fault location problem of multi-branch power distribution lines in the prior art does not fully consider the influence of measurement errors and line parameter errors.

[0067] In view of this, the present invention provides an intelligent fault location method applicable to multi-branch power distribution lines. The method comprises the following steps:

[0068] S1. Build a distribution network model, perform fault simulation on the distribution network model, and obtain the phase voltage and phase current waveform information at the head end of the primary branch line;

[0069] S2. Build a fault location deep learning model using historical and simulated fault samples, and use the trained deep learning model as the distribution network fault location model;

[0070] S3. Collect the effective current value collected by the smart meter at the end of the secondary branch line and determine the fault line through logical reasoning;

[0071] S4. Collect the phase voltage and phase current fault waveforms at the head end of the first-level branch line and input them into the distribution network fault location model to complete the accurate distance measurement of the fault point location.

[0072] S5. Comprehensive fault line identification and fault point accurate distance measurement are used to complete the fault location of the faulty multi-branch power distribution line.

[0073] Step S2 specifically includes:

[0074] After obtaining the waveform information of phase voltage and phase current, the data of the cycle before and three cycles after the fault are taken for simulation training with the fault location to obtain the distribution network fault location model;

[0075] Optionally, the deep learning model consists of a parallel autoencoder (PAE) and a long short-term memory network (LSTM) model.

[0076] 1) PAE module: To preserve the time-dependent characteristics of the original data, several autoencoders (e.g., t autoencoders) are connected in parallel to form a PAE layer. The value of t is determined by the length of the input waveform. The PAE layer primarily reduces the dimensionality of waveform information. It originally consisted of two symmetrical parts: an encoder and a decoder, which were trained together to achieve unsupervised feature learning. The fault localization model only uses the encoder layer.

[0077] During training, the labeled output of the t-th autoencoder has the same dimension as its input vector, and the loss function of the t-th autoencoder is:

[0078]

[0079] in Represents the dataset S t The reconstructed sample of the input sample in W. ti H represents the weight of the t-th autoencoder. t is the number of neurons in the hidden layer. λ, α are regularization parameters that adjust the weight of the penalty term.

[0080] KL(ρ||ρ ti ) is the Kullback-Leibler divergence, which is used to constrain the sparsity of hidden layer neurons to keep it at a small value. KL divergence is a standard function that measures the difference between two contribution values ​​and is often used to train autoencoders. It can be expressed as:

[0081]

[0082] Where ρ ti represents the average activation function value of hidden layer neuron i. ρ is the sparsity parameter. By setting ρ to a small value, ρ ti Can be kept close to zero.

[0083] 2) LSTM Module: The extracted time-related feature sequence is used as the input of the LSTM network. This architecture consists of LSTM cells with self-connections. It allows the value flowing into the cell (forward pass) or gradient (backward pass) to be saved and then retrieved at the required time step. The LSTM module multiplies the cells at different time periods, so the output or error of the previous time step is the same as that of the next time step. The loss function used to train the LSTM network is the mean square error (MSE), as shown in the equation,

[0084]

[0085] where S is the total number of training samples in the LSTM, is the estimation result given by the LSTM model, and y i represents the true value of sample i.

[0086] When training the regression PAE-LSTM model, a loss function is defined to consider the influence of the PAE and LSTM modules, as shown in the equation,

[0087] L = L PAE + L LSTM

[0088] This framework uses the methods of grid search and cross-validation to determine the hyperparameters and uses the Sigmoid function as the activation function of the fully connected layer in the LSTM.

[0089] Step S3 Fault line selection judges the fault line through the effective value of the current at the end of the secondary branch, and the specific description is as follows:

[0090] Collect all the currents I ki (1 ≤ i ≤ n) of the SMs connected to the feeder to form a vector I o . The vector I e represents the set of branches with only small currents, and its value is less than the threshold γ. Here, 0 < p < n. N and O represent the nodes on the main branch and the far-end nodes of the secondary branches respectively. If p = q, it means that only one secondary branch has a fault, and the fault line is N kp O<( kp . When p ≠ q, it means that the main branch N k(p-1) N kp has a fault, which may cause multiple secondary branches to lose power. Thus, the fault line can be judged. <(

[0091] Step S4 Precise fault location ranging is to input the phase voltage and current of the primary branch collected after the fault into the deep learning ranging model constructed by historical data and simulation data, and the ranging model directly outputs the fault point by using the non-linear fitting ability of deep learning.

[0092] Step S5 integrates the fault line and fault point obtained in S3 and S4 to complete the precise distance measurement of the multi-branch distribution lines.

[0093] The performance of this method is verified by modeling a typical tree-structured distribution network. Figure 3 The network shown was modeled in MATLAB / Simulink and consists of a single main branch and three secondary branches. The distribution network voltage is 10 kV. Four faults were simulated: a single-phase ground fault (AG), a phase-to-phase ground fault (AB-g), a phase-to-phase short-circuit fault (AB), and a three-phase ground fault (ABC-G). The waveform sampling frequency was 10 kHz. Various fault parameters, such as fault distance, ground resistance, and phase angle, were set, as shown in Table 1.

[0094] Table 1 Fault sample parameters

[0095]

[0096] In this embodiment, in order to evaluate the performance of the proposed location method, the recognition rate L of the fault line is used. right and the error E of fault location err_L The definitions of these two indicators are as follows:

[0097] L right =N right / N total

[0098] E err_L =(L pre -L act ) / L total

[0099] Among them, N right is the number of correct results of fault line selection, N total is the number of all test samples. pre To calculate the fault distance to the measuring device, L act is the actual fault distance, L total is the distance between nodes, such as N 11 N 12 Table 2 shows the positioning results of the trained PAE-LSTM model in the simulated source domain. The fault line was correctly identified for all test data. The fault location error was also very low, mostly less than 2%, or approximately 60 meters.

[0100] Table 2 Test results of different types of faults

[0101]

[0102] Since the average error and maximum error of AB-G fault are the largest among all fault types, the distribution of the error of AB-G fault with fault location is taken as an example, as shown in Figure 5 As shown in the figure, we discuss the effectiveness of the proposed method. Figure 5 It can be seen that the positioning error, Eerr_L, increases with the distance from the measuring device. The error for closely spaced segments, such as Lines 1, 2, and 3, is smaller than that for the more distant Lines 6 and 7. Within the same segment, faults with relatively large relative positions (such as 80% and 90%) produce larger errors than those with relatively small relative positions. Clearly, the performance of the PAE-LSTM model degrades as the distance from the fault increases. However, the worst-case scenario is acceptable, as the error remains below 3%.

[0103] In this embodiment, the influence of ground resistance and fault phase angle is also studied. Figure 6 As shown in the figure, the error for different fault phase angles varies little. The average error at 90° is larger, while the average error at 30° is the smallest. The maximum error difference is only 0.0285%. It also illustrates how the average positioning error varies with ground resistance. Although normalization is used to reduce the impact of ground resistance, the average error increases slightly as the ground resistance increases. When the ground resistance reaches 50Ω, the average error is less than 2%, which is acceptable. Based on the above analysis, the proposed PAE-LSTM-based fault location network can correctly identify the fault line segment and accurately locate the fault point with low positioning error. Although the positioning performance decreases with increasing fault distance and ground resistance, the positioning error is less than 3% in all simulated scenarios.

[0104] Verification has shown that this method accurately locates fault points in multi-branch distribution lines. By training on a large number of high-dimensional samples, it can accurately measure the distance to fault points in multi-branch distribution systems. This method has been applied in practical projects. This method is highly universal and can be trained on actual distribution network topologies. The trained positioning network has extremely high accuracy and reliability, and has promising future development prospects.

[0105] Example 3

[0106] This embodiment 3 provides a non-transitory computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the intelligent fault location method applicable to multi-branch distribution lines as described above is implemented. The method includes:

[0107] Using the pre-built distribution network model, perform fault simulation on the distribution network model to obtain the phase voltage and phase current waveform information at the head end of the primary branch line; obtain the effective current value collected by the smart meter at the end of the secondary branch line;

[0108] Using the pre-trained distribution network fault location model, the acquired phase voltage and phase current waveform information at the head end of the primary branch line is processed to complete the fault point location.

[0109] Determine the fault line based on the effective current value collected by the smart meter at the end of the secondary branch line;

[0110] The fault line judgment results and fault point location are integrated to complete the fault location of the faulty multi-branch power distribution line.

[0111] Example 4

[0112] This embodiment 4 provides a computer program product, including a computer program. When the computer program is executed on one or more processors, the computer program is used to implement the above-mentioned intelligent fault location method applicable to multi-branch distribution lines. The method includes:

[0113] Using the pre-built distribution network model, perform fault simulation on the distribution network model to obtain the phase voltage and phase current waveform information at the head end of the primary branch line; obtain the effective current value collected by the smart meter at the end of the secondary branch line;

[0114] Using the pre-trained distribution network fault location model, the acquired phase voltage and phase current waveform information at the head end of the primary branch line is processed to complete the fault point location.

[0115] Determine the fault line based on the effective current value collected by the smart meter at the end of the secondary branch line;

[0116] The fault line judgment results and fault point location are integrated to complete the fault location of the faulty multi-branch power distribution line.

[0117] Example 5

[0118] This embodiment 5 provides an electronic device, including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory to cause the electronic device to execute instructions for implementing the above-mentioned intelligent fault location method applicable to multi-branch power distribution lines. The method includes:

[0119] Using the pre-built distribution network model, perform fault simulation on the distribution network model to obtain the phase voltage and phase current waveform information at the head end of the primary branch line; obtain the effective current value collected by the smart meter at the end of the secondary branch line;

[0120] Using the pre-trained distribution network fault location model, the acquired phase voltage and phase current waveform information at the head end of the primary branch line is processed to complete the fault point location.

[0121] Determine the fault line based on the effective current value collected by the smart meter at the end of the secondary branch line;

[0122] The fault line judgment results and fault point location are integrated to complete the fault location of the faulty multi-branch power distribution line.

[0123] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, 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 magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0124] 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 flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts 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, 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 flowcharts and / or block diagrams. 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.

[0125] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work 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 The function specified in one or more boxes.

[0126] These computer program instructions can also be loaded onto a computer or other programmable data processing device, and a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide the functions for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0127] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solutions disclosed in the present invention without the need for creative work should be included in the scope of protection of the present invention.

Claims

1. An intelligent fault location method applicable to multi-branch power distribution lines, characterized in that: include: Using the pre-built distribution network model, the fault simulation is performed on the distribution network model to obtain the phase voltage and phase current waveform information at the head end of the primary branch line; Obtain the effective value of the current collected by the smart meter at the end of the secondary branch line; Based on the effective value of the current collected by the smart meter at the end of the secondary branch line, the faulty line is judged; it includes: collecting all the currents \(I_i\) of the smart meters at the ends of the secondary branches connected to the feeder, where \(1\leq i\leq n\), to form a vector \(\mathbf{I}\). ki , where \(1\leq i\leq n\), to form a vector \(\mathbf{I}\). o The vector \(\mathbf{I}_s\) e represents the set of branches with only small currents, and its value is less than the threshold \(\gamma\); \(0 < p < n\); \(N\) and \(O\) respectively represent the nodes on the main branch and the distal nodes of the secondary branches. If \(p = q\), it means that only one secondary branch has a fault, and the faulty line is \(N\). kp O kp When \(p\neq q\), it means that the main branch \(N\) k(p-1) N kp has a fault, which may cause multiple secondary branches to lose power, and thus the faulty line can be judged. Using a pre-trained distribution network fault location model, the acquired phase voltage and phase current waveform information at the head end of the primary branch line is processed to complete the fault point location. The distribution network fault location model includes an autoencoder module and a long short-term memory network module in parallel. The autoencoder module is used to reduce the dimension of the waveform information. The long short-term memory network module is used to multiply cells in different time periods using the extracted time-related feature sequence as input. T AEs are connected in parallel to form an autoencoder module layer. The value of t is determined by the length of the input waveform. During the training process of the autoencoder module layer, the labeled output of the tth autoencoder has the same dimension as its input vector, and the loss function of the tth autoencoder is: in, Represents the dataset S t The reconstructed sample of the input sample in W ti represents the weight of the t-th autoencoder; H t is the number of neurons in the hidden layer; λ, α are regularization parameters for adjusting the weight of the penalty term; KL(ρ||ρ ti ) is the Kullback-Leibler divergence, ρ ti represents the average activation function value of hidden layer neuron i; ρ is the sparsity parameter; The fault line judgment results and fault point location are integrated to complete the fault location of the faulty multi-branch power distribution line.

2. The intelligent fault location method applicable to multi-branch power distribution lines according to claim 1, characterized in that: The LSTM module multiplies cells from different time periods, so the output or error at the previous time step is the same as the output at the next time step; the loss function used to train the LSTM network is the mean squared error: Among them, S is the total number of training samples, is the estimation result given by the LSTM model, y i represents the true value of sample i.

3. The intelligent fault location method applicable to multi-branch power distribution lines according to claim 2, characterized in that: When training the regression distribution network fault location model, the loss function is defined to consider the influence of the autoencoder module and the long short-term memory network module: L = L PAE +L LSTM ; Grid search and cross-validation methods are used to determine the hyperparameters, and the Sigmoid function is used as the activation function of the fully connected layer in the long short-term memory network module.

4. An intelligent fault location system suitable for multi-branch power distribution lines, characterized in that: include: An acquisition module is used to use a pre-built distribution network model to perform fault simulation on the distribution network model and obtain the phase voltage and phase current waveform information at the head end of the primary branch line; Obtain the effective value of the current collected by the smart meter at the end of the secondary branch line; A judgment module, which is used to judge the faulty line based on the effective value of the current collected by the smart meter at the end of the secondary branch line; it includes: collecting all the currents I of the smart meters at the ends of the secondary branches connected to the feeder ki , where 1 ≤ i ≤ n, to form a vector I o ; the vector I e represents the set of branches with only small currents, and its value is less than the threshold γ; 0 < p < n; N and O respectively represent the nodes on the main branch and the far-end nodes of the secondary branches. If p = q, it means that only one secondary branch has a fault, and the faulty line is N kp O kp ; when p ≠ q, it means that the main branch N k(p-1) N kp has a fault, which may cause multiple secondary branches to lose power, and thus the faulty line can be judged; The ranging module is used to process the phase voltage and phase current waveform information obtained at the head end of the first-level branch line using a pre-trained distribution network fault ranging model to complete the fault point location ranging. The distribution network fault ranging model includes an autoencoder module and a long short-term memory network module connected in parallel. The autoencoder module is used to reduce the dimension of the waveform information. The long short-term memory network module is used to multiply cells in different time periods using the extracted time-related feature sequence as input. T AEs are connected in parallel to form an autoencoder module layer. The value of t is determined by the length of the input waveform. During the training process of the autoencoder module layer, the labeled output of the t-th autoencoder has the same dimension as its input vector, and the loss function of the t-th autoencoder is: in, Represents the dataset S t The reconstructed sample of the input sample in W ti represents the weight of the t-th autoencoder; H t is the number of neurons in the hidden layer; λ, α are regularization parameters for adjusting the weight of the penalty term; KL(ρ||ρ ti ) is the Kullback-Leibler divergence, ρ ti represents the average activation function value of hidden layer neuron i; ρ is the sparsity parameter; The positioning module is used to integrate the fault line judgment results and fault point location measurement to complete the fault positioning of the faulty multi-branch power distribution line.

5. A non-transitory computer-readable storage medium, characterized in that The non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the intelligent fault location method applicable to multi-branch power distribution lines according to any one of claims 1 to 3 is implemented.

6. A computer program product, characterized in that The invention comprises a computer program, which is used to implement the intelligent fault location method applicable to multi-branch power distribution lines as claimed in any one of claims 1 to 3 when running on one or more processors.

7. An electronic device, characterized in that: include: A processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to execute instructions for implementing the intelligent fault location method for multi-branch power distribution lines according to any one of claims 1 to 3.

Images