A method for multi-branch fault identification and localization in multi-energy systems based on intelligent algorithms

Abstract

This invention proposes a method for multi-branch fault identification and location in multi-energy systems based on intelligent algorithms. The method includes the following steps: S110, analyzing the impact of multi-energy coupled systems on traditional distribution network fault identification and location; S120, based on the results of S110, identifying the line fault types in the multi-energy coupled system, including: determining the fault area and identifying the faulty line, thereby narrowing the location range; S130, within the narrowed range, performing multi-branch fault location based on intelligent algorithms. This invention proposes a binary particle swarm optimization algorithm with adaptive characteristics, which can overcome the limitations of traditional methods and significantly improve the efficiency and accuracy of fault location.

Description

Technical Field

[0001] This invention belongs to the field of distribution network fault location, specifically relating to a method for fault identification and location of multi-branch systems based on intelligent algorithms. Background Technology

[0002] In recent years, with the rapid development of the national economy, people's demand for electricity has been increasing, and the construction of first-class distribution networks has been steadily advancing. The network structure of distribution systems is becoming increasingly complex, with numerous branch lines, interconnected overhead and cable lines, and diverse operating conditions, inevitably leading to various types of faults. Especially with the widespread application of multi-energy systems containing electricity, gas, heat, and other energy sources, the problem of fault identification and location in power grids has become more complex, making it difficult for existing fault diagnosis methods to achieve the expected results. Quickly and accurately identifying fault types and locating fault points will facilitate rapid fault elimination, shorten power outage time, and reduce the workload of manual line inspections. While existing methods have high fault tolerance, they lack flexibility. When the power grid topology changes, location models based on artificial intelligence algorithms are difficult to apply, and the convergence speed for different fault points is slow during the location process. Therefore, there is an urgent need for a multi-branch fault location method for distribution networks containing renewable energy based on intelligent algorithms, capable of accurately locating fault points at different locations and of different types, and able to solve problems such as slow convergence speed, susceptibility to local optima, poor fault tolerance, and low population quality. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention proposes a multi-branch fault identification and location method for multi-energy systems based on intelligent algorithms, belonging to the field of distribution network fault location. This method first analyzes the impact of multi-energy systems on traditional distribution network fault identification and location, then studies fault identification methods, and finally investigates a multi-branch fault location method based on intelligent algorithms.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] A method for fault identification and localization in a multi-branch system based on intelligent algorithms includes the following steps:

[0006] S110, Analyze the impact of multi-energy coupling systems on fault identification and location in traditional distribution networks;

[0007] S120, based on the results of S110, identifies the line fault type of the multi-energy coupling system, including: determining the fault area and identifying the fault line, thereby narrowing down the location range;

[0008] S130 performs multi-branch fault location based on intelligent algorithms within the narrowed range.

[0009] A computing device includes: at least one processor and a memory storing program instructions; when the program instructions are read and executed by the processor, the computing device performs the multi-branch fault identification and location method for multi-energy systems based on intelligent algorithms.

[0010] Beneficial effects:

[0011] This invention proposes a binary particle swarm optimization algorithm with adaptive characteristics, which can overcome the limitations of traditional methods and significantly improve the efficiency and accuracy of fault location. Attached Figure Description

[0012] Figure 1 This is an overall flowchart of a multi-branch fault identification and location method for a multi-energy system based on intelligent algorithms according to the present invention.

[0013] Figure 2 This is a flowchart illustrating the fault location process of the present invention. Detailed Implementation

[0014] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0015] This invention proposes a multi-branch fault identification and location method for multi-energy systems based on intelligent algorithms, belonging to the field of distribution network fault location. The method first analyzes the impact of multi-energy systems on traditional distribution network fault identification and location, then studies fault identification methods, and finally investigates a multi-branch fault location method based on intelligent algorithms.

[0016] Figure 1 This is an overall flowchart of a multi-branch fault identification and location method for a multi-energy system based on intelligent algorithms, according to the present invention. Figure 1 As shown, the method includes:

[0017] S110, Analyze the impact of multi-energy coupling systems on fault identification and location in traditional distribution networks; multi-energy coupling systems are coupling systems that integrate multiple energy forms such as electricity, gas, and heat;

[0018] S120, based on the results of S110, identifies the line fault type of the multi-energy coupling system, including: determining the fault area and identifying the fault line, thereby narrowing down the location range;

[0019] S130, within the narrowed range, performs multi-branch fault location based on intelligent algorithms. Fault location based on intelligent algorithms can solve problems such as slow convergence speed, susceptibility to getting trapped in local optima, poor fault tolerance, and low population quality in fault location.

[0020] Specifically, S110 may include:

[0021] Multi-energy coupling systems integrate multiple energy forms such as electricity, gas, and heat, transforming the distribution network from a traditional single-source structure to a multi-source structure. This may lead to a series of problems, including changes in short-circuit current, increased voltage levels, and degraded power quality. The integration of multi-energy systems into the distribution network primarily impacts fault diagnosis in the following ways:

[0022] S110-1, after the multi-energy coupling system is connected to the grid, it will affect the power flow of the system and may cause changes in the power flow on the line. The capacity and connection location of the various energy sources included in the multi-energy coupling system will directly affect the magnitude and direction of the short-circuit current after a fault.

[0023] S110-2, different energy types have significantly different fault characteristics, which also affect the level of short-circuit current.

[0024] S110-3, the power output of multiple energy sources is easily affected by external factors, exhibiting randomness and fluctuation.

[0025] S110-4. Power electronic devices in multi-energy coupled systems can generate harmonic problems in the distribution network, causing distortion of fault information and interfering with and affecting fault identification and location models based on transient signals.

[0026] In S110-5, in a multi-energy coupled system, electrical, pneumatic, and thermal subsystems influence each other through coupling elements. A failure in one subsystem may trigger a cascading response in other subsystems, causing fault characteristics to intertwine across multiple energy domains and increasing the difficulty of identification.

[0027] In S110-6, the dynamic response time scales of different energy networks in a multi-energy coupled system differ. Fault location needs to consider the propagation characteristics across time and space, making traditional single-network methods difficult to apply. Intelligent algorithms are required to reduce dimensionality and simplify the location process.

[0028] The results of S110 are useful for S120 in analyzing the fault characteristics of multi-energy coupled systems. Before identifying faulty lines, the multi-energy coupled system is analyzed as a combination of subnets to reduce the complexity of fault analysis.

[0029] S120, based on the results of S110, identifies the line fault type of the multi-energy coupling system, including: determining the fault area and identifying the faulty line, thereby narrowing down the location range. Specifically, it may include:

[0030] When a power grid fault occurs, the pre-processed fault characteristic signals from each measuring device are first transmitted to the decision center. Next, the fault area is determined to narrow down the fault search range. Finally, the actual faulty line is identified from the fault area.

[0031] S120-1, Identify the fault area.

[0032] The entire multi-energy coupled system will be divided into a set of subnets to train fault region detectors. The reason for this system partitioning is that the transmission lines in each subnet typically exhibit highly correlated fault characteristics. Therefore, analyzing the target system as a combination of subnets before identifying faulty lines reduces the complexity of fault analysis. Furthermore, the partitioning of the system facilitates the training of the classifier model by reducing the number of classifications and trainable parameters. Specifically, the process includes: dividing the multi-energy coupled system into subnets based on the fault characteristics of the transmission lines; training the fault region detectors using a dataset measured by the fault region detectors, which can employ synchronous phasor measurement units or similar measurement devices; and determining the fault regions of the multi-energy coupled system using the fault region detectors.

[0033] S120-2, Identify faulty circuits.

[0034] Once a fault area is identified using the fault area detector, a fault line detection classifier is used to identify the actual faulty line within that area. During training, the fault line detector is labeled with the transmission line label, while the fault area detector is labeled with the sub-region label. The process may include: after a fault area is identified by the fault area detector, the identification result is sent to the fault line detector corresponding to the subnet containing that fault area for further identification of the faulty line.

[0035] Considering that faults occurring on transmission lines at the subnet boundary may lead to misjudgments, the training dataset used to train the fault line detector should include transmission lines at the subnet boundary in addition to those within the fault area. This will enable the fault line detector to identify the fault characteristics of the boundary lines and avoid misjudgments caused by subnet partitioning.

[0036] S130, within the narrowed range, performs multi-branch fault location based on intelligent algorithms, which may specifically include:

[0037] S130-1, Design an improved method for binary particle swarm optimization algorithm.

[0038] Traditional binary particle swarm optimization (BSO) algorithms often encounter problems such as premature convergence and getting trapped in local optima. To address this issue, this invention proposes a novel optimization strategy that improves the search efficiency at different iteration stages by flexibly adjusting the behavior of binary particles. The particle swarm is updated according to a specific iterative formula, as follows:

[0039] (1)

[0040] In the formula: and These represent the velocity and position in the (k+1)th iteration; It is the position of the k-th iteration; It is the speed of the k-th iteration; Inertial weight; , It is the acceleration factor, and the value of the acceleration factor is usually greater than 0; It is the optimal position for the population; It is the optimal position for an individual; , It is a random constant within the interval [0, 1].

[0041] The expression for this method is:

[0042] (2)

[0043] In the formula: is the iteration position; r is a random number distributed between 0 and 1; R is the probability of mutation, which adopts a decreasing strategy, namely:

[0044] (3)

[0045] In the formula: T is the total number of iterations set for the improved algorithm; t is the number of iterations that have been completed so far.

[0046] In the improved binary particle swarm optimization algorithm, a combined mutation operation is performed to generate new particles after each particle's position is updated. During this process, it is first checked whether the current iteration count has reached 5% of the maximum iteration count T. If the iteration count has not exceeded this threshold (i.e., within 5%), each particle has a certain probability of mutation; however, when the iteration count exceeds 5%, each dimension of the particle's position will no longer mutate, and its mutation probability will drop to zero and remain constant. Specifically, this mutation probability R gradually decreases from 1 to 0 as the iteration count increases, and remains constant after reaching 0.

[0047] Traditional binary particle swarm optimization algorithms suffer from the problem of similar inertial weights. To address this issue, this invention proposes a method for setting dynamic weights:

[0048] (4)

[0049] In the formula: It is inertial weight. It is the minimum value of the inertia weight; It is the maximum value of the inertia weight; f, , These represent the fitness of all current particles, the average fitness, and the minimum fitness, respectively. The adaptive weight strategy for particles is an intelligent and highly dynamic parameter adjustment method that abandons the fixed, inertial weight settings of traditional particle swarm optimization algorithms. Instead, adaptive weights can automatically adjust according to changes in the current fitness of each particle. This flexibility not only makes the algorithm more intelligent but also more closely matches the needs of real-world applications.

[0050] S130-2, based on the binary particle swarm optimization algorithm, performs fault location in the distribution network.

[0051] In the process of fault location in distribution networks using an improved binary particle swarm optimization (PSO) algorithm, the distribution network system is first subdivided into multiple independent feeder sections, simplifying the complex network structure into smaller, more manageable units. Specifically, sectionalizing switches are used to divide long lines into multiple feeder sections. Feeder terminal units are installed at these sectionalizing switches to detect the status of the switch locations. By mapping the dimension of the particle swarm optimization to the number of feeder sections, an intuitive and efficient fault mapping mechanism is constructed. Under this mapping, if a section experiences a fault, it is marked as 1 in the corresponding dimension of the particle swarm optimization; conversely, if the section is operating normally, it is marked as 0.

[0052] In the fault location process of a radial distribution network, the information recorded by the feeder terminal unit is converted into binary code for processing. Specifically, normal data for a section is marked as 0, while abnormal data is marked as 1. For radial distribution networks equipped with multiple energy sources, the fault location coding mechanism is further refined: when the current detected in the section is in a normal state, the output result is coded as 0; if a fault current is detected and its flow direction is opposite to the preset positive direction, the output result is coded as -1; if the fault current flows in the same direction as the positive direction, the output result is coded as 1.

[0053] Based on the principle of minimizing the deviation between the actual transmitted fault information and the information corresponding to the actual state of each feeder section to be determined, a fitness function is constructed:

[0054] (5)

[0055] In the formula: j is the number of the sectionalizing switch; This refers to the data on the current distribution network status sent by the j-th feeder terminal unit; Let be the expected function value of the j-th switch; M is the total number of segments; It serves as a penalty function; It is a constant, P=0.5; It is a penalty item; The output of the fitness function is given, where j ranges from 1 to M. An output of 0 indicates that the segment is operating normally, and an output of 1 indicates that the segment has failed. The flowchart is as follows. Figure 2 As shown, the specific steps are as follows:

[0056] (1) Initialize the parameters and the velocities and positions of all particles;

[0057] (2) Calculate the fitness value of all particles and determine the optimal position and best fitness of all particles;

[0058] (3) Update the velocity and position information of all particles;

[0059] (4) Calculate the fitness value for each particle;

[0060] (5) Update the optimal position and fitness of all individual particles, the global optimal position and the global optimal fitness;

[0061] (6) Determine whether the maximum number of iterations has been reached. If not, return to step (3). If the maximum number of iterations has been reached, output the best fitness value and the best position information.

[0062] The present invention also provides a computing device, comprising: at least one processor and a memory storing program instructions; when the program instructions are read and executed by the processor, the computing device performs the aforementioned method for multi-branch fault identification and location of multi-energy systems based on intelligent algorithms.

[0063] The above implementation steps are provided merely for the purpose of describing the present invention and are not intended to limit the scope of the invention. The scope of the invention is defined by the appended claims. All equivalent substitutions and modifications made without departing from the spirit and principles of the invention should be covered within the scope of the invention.

Claims

1. A multi-energy system multi-branch fault identification and positioning method based on intelligent algorithm, characterized in that, Includes the following steps: S110, Analyze the impact of multi-energy coupling systems on fault identification and location in traditional distribution networks; S120, based on the results of S110, identifies the line fault type of the multi-energy coupling system, including: determining the fault area and identifying the fault line, thereby narrowing down the location range; S130 performs multi-branch fault location based on intelligent algorithms within the narrowed range.

2. The method of claim 1, wherein, S120 includes: S120-1, Determine the fault area: Divide the multi-energy coupled system into subnets based on the fault characteristics of the transmission lines; train the fault area detector using the dataset measured by the fault area detector, which employs a synchronous phasor measurement unit; determine the fault area of ​​the multi-energy coupled system through the fault area detector. S120-2, Identify faulty lines: Use a faulty line detection classifier to identify the actual faulty lines from the fault area.

3. The method of claim 2, wherein, S120-2 includes: after the fault area detector identifies the fault area, the identification result is sent to the fault line detector corresponding to the subnet where the fault area is located to identify the fault line.

4. The method of claim 3, wherein, When training the fault line detector, the training dataset used covers transmission lines not only within the fault area but also at the subnet boundary.

5. The method for multi-branch fault identification and location in a multi-energy system based on intelligent algorithms according to claim 3, characterized in that, Step S130 includes: S130-1, Design an improved method for the binary particle swarm optimization algorithm, where the particle swarm is updated according to a specific iterative formula, as follows: （1） In the formula: and These represent the velocity and position in the (k+1)th iteration; It is the position of the k-th iteration; It is the speed of the k-th iteration; Inertial weight; , It is the acceleration factor, and the value of the acceleration factor is usually greater than 0; It is the optimal position for the population; It is the optimal position for an individual; , It is a random constant within the interval [0, 1]. The expression for this method is: （2） In the formula: is the iteration position; r is a random number distributed between 0 and 1; R is the probability of mutation; S130-2, based on an improved binary particle swarm optimization algorithm, performs fault location in the distribution network.

6. The method for fault identification and location in a multi-branch system based on intelligent algorithms according to claim 5, characterized in that, The probability R of mutation is determined using a decreasing strategy: （3） In the formula: T is the total number of iterations set for the improved algorithm; t is the number of iterations that have been completed so far.

7. The method for fault identification and location in a multi-branch system based on intelligent algorithms according to claim 5, characterized in that, Regarding the improved binary particle swarm optimization algorithm, the method for setting dynamic weights is as follows: （4） In the formula: It is inertial weight; It is the minimum value of the inertia weight; It is the maximum value of the inertia weight; f, , These represent the fitness of all current particles, the average fitness, and the minimum fitness, respectively.

8. The method for fault identification and location of multi-branch systems based on intelligent algorithms according to claim 7, characterized in that, S130-2 includes: refining the distribution network system into multiple independent feeder sections; constructing a fault mapping mechanism by mapping the dimension of the particle swarm to the number of feeder sections: under this mapping, if a feeder section fails, it is marked as 1 in the corresponding dimension of the particle swarm; otherwise, if the feeder section is operating normally, it is marked as 0.

9. The method for fault identification and location in a multi-branch system based on intelligent algorithms according to claim 7, characterized in that, Also includes: Based on the principle of minimizing the deviation between the actual transmitted fault information and the information corresponding to the actual state of each feeder section to be determined, a fitness function is constructed: （5） In the formula: j is the number of the sectionalizing switch; This refers to the data on the current distribution network status sent by the j-th feeder terminal unit; Let be the expected function value of the j-th switch; M is the total number of segments; It serves as a penalty function; It is a constant, P=0.5; It is a penalty item; The output of the fitness function is in the range j=1-M. An output of 0 indicates that the segment is operating normally, and an output of 1 indicates that the segment has failed.

10. A computing device, characterized in that, include: At least one processor and a memory storing program instructions; When the program instructions are read and executed by the processor, the computing device performs the multi-branch fault identification and location method for multi-energy systems based on intelligent algorithms as described in any one of claims 1-9.

Images