Power distribution network protection method

Abstract

The invention discloses a power distribution network protection method in the field of power system protection, and the method comprises the following steps: S101, collecting the multi-source data of a power distribution network in real time, collected multi-source information such as electrical quantity data, distributed energy output data, meteorological data, equipment state monitoring data, node voltage, current waveform, topological structure, distributed power output state and environmental parameters in a power distribution network is transmitted to a protection device or a control center through a communication network. Through application of multi-source information fusion and self-adaptive protection criteria, the protection device can judge faults more accurately, maloperation and refusal operation are reduced, the accuracy of power distribution network protection is improved, an intelligent reclosing strategy and implementation of distributed cooperative protection are adopted, and the safety of power distribution network protection is improved. The secondary impact and islanding effect of reclosing on the system are effectively avoided, the stability of the system is improved, and the safe operation of the power distribution network is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of power system protection, and in particular to a distribution network protection method. Background Art

[0002] With the continuous expansion of distribution networks, the widespread integration of distributed energy resources (DGEs), and the increasing demand for power supply reliability from power users, traditional distribution network protection methods face numerous challenges. Traditional three-stage current protection, whose setting calculations primarily rely on the maximum and minimum operating modes of the line, is ill-suited to the volatile power flows of distribution networks after the integration of DGEs. Distributed energy output is intermittent and fluctuating. When a fault occurs near the DGE access point, the magnitude and direction of the fault current change, affecting the operating characteristics of traditional current protection, potentially leading to false tripping or failure to trip. While distance protection can accurately measure the distance from the fault point to the protection installation, inaccurate line parameters and the presence of numerous branch lines in distribution networks reduce its measurement accuracy. Especially in complex distribution network topologies, the presence of branch lines can distort the measured impedance, affecting the correct operation of distance protection. Reclosing circuit breakers are widely used in traditional distribution network protection as a means of improving power supply reliability. However, for some permanent faults, traditional reclosing circuit breakers can cause secondary shocks to the system, impacting system stability. Moreover, after distributed energy is connected, the coordination between the reclosing switch and the distributed power source becomes complicated. If the reclosing time is not selected properly, it may cause an islanding effect, threatening the safe operation of the power grid and equipment.

[0003] Furthermore, traditional distribution network protection methods mostly make decisions based on local information and lack the comprehensive utilization of global distribution network information. In the context of smart grids, distribution networks have a large number of monitoring devices and communication methods, but traditional protection methods fail to fully utilize these resources, resulting in inaccurate and inappropriate fault diagnosis and handling. Therefore, those skilled in the art have provided distribution network protection methods to address the issues raised in the background technology above. Summary of the Invention

[0004] The purpose of the present invention is to provide a distribution network protection method to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: The distribution network protection method includes the following steps: Step S101: Real-time data collection from multiple sources of the distribution network is performed, and the collected data, including electrical quantity data, distributed energy output data, meteorological data, equipment status monitoring data, voltage and current waveforms of each node, topology structure, distributed power output status, and environmental parameters, is transmitted to the protection device or control center via a communication network. Step S102: Use data mining and machine learning techniques to deeply analyze and process multi-source information, identify data features, extract distortion features from the current waveform based on a dynamic time warping algorithm, identify fault types and suspected fault areas, and classify voltage transient components based on an improved deep residual network to distinguish between internal and external faults. For example, cluster analysis algorithms are used to classify current and voltage data, identifying data features under normal operating conditions, fault conditions, and various abnormal operating conditions; neural network algorithms are used to learn distributed energy output data and meteorological data to predict output trends of distributed energy resources, providing a more accurate basis for protection decisions. Step S103: Combine the fault type, partition logic, and feature classification results to calculate dynamic protection settings, including overcurrent thresholds, directional blocking conditions, and action time limits. Based on the real-time operation and topology of the distribution network, the protection criteria are dynamically adjusted, introducing both fault component and comprehensive protection criteria. Blockchain technology is also used to record protection process data and generate an unalterable fault tracing log. For example, when distributed energy access causes changes in the distribution network flow, the current protection setting is adaptively adjusted by calculating the equivalent impedance and short-circuit current of the line in real time. Step S104: Before reclosing, use multi-source information to determine the nature of the fault and optimize the reclosing time. Consider the access situation of distributed energy resources and the stability requirements of the system, and adopt an adaptive reclosing time setting method. Based on the type, capacity, and access location of the distributed energy resources, combined with the transient stability calculation results of the system, the optimal reclosing time is determined to avoid the islanding effect and ensure the safe and stable operation of the system. Step S105: Set up multiple distributed protection units, and exchange information between each protection unit. At the same time, send protection instructions to the intelligent circuit breaker, distributed power supply controller and energy storage device to perform collaborative protection actions. When a fault occurs in a certain area, the protection unit in the fault area quickly collects local data and sends relevant information to the adjacent protection unit. Adjacent protection units conduct collaborative analysis and judgment based on the received information and their own monitoring data to jointly determine the fault range and fault type. Based on the results of distributed collaborative protection, each protection unit acts according to the pre-set action strategy to quickly isolate the fault and restore power supply to non-fault areas. For example, when a fault occurs in a certain line, the protection units on both sides of the fault line quickly exchange information through the communication network after detecting the fault, collaboratively determine the fault location, and then simultaneously trip the corresponding circuit breaker to isolate the fault line. Step S106: Use big data analysis and machine learning technology to establish a fault warning model, take preventive measures based on the warning results, update the topology database based on the protection action results, and iteratively optimize the protection logic matrix and constant parameters.

[0006] As a further solution of the present invention: the electrical quantity data in step S101 includes current, voltage, power and frequency data.

[0007] As a further solution of the present invention: the data mining and machine learning techniques in step S1012 include a cluster analysis algorithm and a neural network algorithm, and the optimization of the dynamic time warping algorithm includes: S1021. Introduce a sliding window mechanism to perform segmented matching on the current waveform; S1022. Adopting an adaptive threshold to eliminate noise interference and extract fundamental wave and harmonic distortion features; S1023. Establish a waveform similarity evaluation model based on the fault recording database to classify the fault type into short circuit, line break or ground fault.

[0008] As a further solution of the present invention: the improved deep residual network in step S102 further includes an input layer receiving the time-frequency domain joint features of the voltage transient component; the residual module embeds the attention mechanism and the output layer generates the in-zone / out-of-zone fault probability distribution through the Softmax function.

[0009] As a further solution of the present invention, when dynamically adjusting the protection criterion in step S103, the setting value of the current protection is adaptively adjusted by calculating the equivalent impedance and short-circuit current of the line in real time. The fault component criterion extracts the current and voltage fault components at the moment of fault occurrence and uses the amplitude, phase, and change rate of the fault components to perform fault judgment. The calculation method of the dynamic protection setting value includes the following steps: S1031. Establish a dynamic impedance model based on the fault current distribution and calculate the correction coefficient of the short-circuit current as the output of the distributed power source changes; S1032. Using a particle swarm optimization algorithm to find the Pareto optimal solution for the overcurrent threshold and the action time limit; S1033. Set the direction locking criterion to the phase difference threshold between the negative sequence voltage and the zero sequence current.

[0010] As a further solution of the present invention: when using multi-source information to determine the nature of the fault in step S104, the duration, waveform characteristics and equipment status monitoring data of the fault current are analyzed, and when optimizing the reclosing time, the type, capacity and access location of the distributed energy are considered, and the reclosing time is determined in combination with the transient stability calculation results of the system.

[0011] As a further solution of the present invention: in step S105, the distributed protection units exchange information with each other via a high-speed communication network.

[0012] As a further solution of the present invention, the fault warning model in step S106 is established by analyzing historical operating data and real-time monitoring data of the distribution network, and preventive measures are taken, including adjusting the output of distributed energy, optimizing load distribution, or adjusting the operation mode of the power grid. The iterative optimization method includes the following steps: S1061. Constructing a protection strategy evaluation model based on a reinforcement learning framework; S1062. Update network weights using historical fault handling results as a reward function; S1063. Dynamically adjust the weight coefficients in the protection partition logic matrix.

[0013] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention uses multi-source information fusion and adaptive protection criteria to enable protection devices to more accurately judge faults, reduce the occurrence of false operation and refusal to operate, and improve the accuracy of distribution network protection. The implementation of intelligent reclosing strategy and distributed collaborative protection effectively avoids the secondary impact of reclosing on the system and the occurrence of islanding effect, improves the stability of the system, and ensures the safe operation of the distribution network.

[0014] 2. Through the combination of distributed collaborative protection and fault warning and prevention mechanisms, the present invention can quickly isolate faults, restore power supply to non-fault areas, and prevent the occurrence of faults in advance, significantly improving the power supply reliability of the distribution network, reducing the duration and scope of power outages, and effectively responding to the challenges brought by distributed energy access and the complex topology of the distribution network, adapting to the development needs of smart grids, and providing technical support for the intelligent upgrade of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a step diagram of the present invention; DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0017] See also Figure 1 In an embodiment of the present invention, a distribution network protection method includes the following steps: Step S101: Real-time data collection from multiple sources of the distribution network is performed, and the collected data, including electrical quantity data, distributed energy output data, meteorological data, equipment status monitoring data, voltage and current waveforms of each node, topology structure, distributed power output status, and environmental parameters, is transmitted to the protection device or control center via a communication network. Step S102: Use data mining and machine learning techniques to deeply analyze and process multi-source information, identify data features, extract distortion features from the current waveform based on a dynamic time warping algorithm, identify fault types and suspected fault areas, and classify voltage transient components based on an improved deep residual network to distinguish between internal and external faults. For example, cluster analysis algorithms are used to classify current and voltage data, identifying data features under normal operating conditions, fault conditions, and various abnormal operating conditions; neural network algorithms are used to learn distributed energy output data and meteorological data to predict output trends of distributed energy resources, providing a more accurate basis for protection decisions. Step S103: Combine the fault type, partition logic, and feature classification results to calculate dynamic protection settings, including overcurrent thresholds, directional blocking conditions, and action time limits. Based on the real-time operation and topology of the distribution network, the protection criteria are dynamically adjusted, introducing both fault component and comprehensive protection criteria. Blockchain technology is also used to record protection process data and generate an unalterable fault tracing log. For example, when distributed energy access causes changes in the distribution network flow, the current protection setting is adaptively adjusted by calculating the equivalent impedance and short-circuit current of the line in real time. Step S104: Before reclosing, use multi-source information to determine the nature of the fault and optimize the reclosing time. Consider the access situation of distributed energy resources and the stability requirements of the system, and adopt an adaptive reclosing time setting method. Based on the type, capacity, and access location of the distributed energy resources, combined with the transient stability calculation results of the system, the optimal reclosing time is determined to avoid the islanding effect and ensure the safe and stable operation of the system. Step S105: Set up multiple distributed protection units, and exchange information between each protection unit. At the same time, send protection instructions to the intelligent circuit breaker, distributed power supply controller and energy storage device to perform collaborative protection actions. When a fault occurs in a certain area, the protection unit in the fault area quickly collects local data and sends relevant information to the adjacent protection unit. Adjacent protection units conduct collaborative analysis and judgment based on the received information and their own monitoring data to jointly determine the fault range and fault type. Based on the results of distributed collaborative protection, each protection unit acts according to the pre-set action strategy to quickly isolate the fault and restore power supply to non-fault areas. For example, when a fault occurs in a certain line, the protection units on both sides of the fault line quickly exchange information through the communication network after detecting the fault, collaboratively determine the fault location, and then simultaneously trip the corresponding circuit breaker to isolate the fault line. Step S106: Use big data analysis and machine learning technology to establish a fault warning model, take preventive measures based on the warning results, update the topology database based on the protection action results, and iteratively optimize the protection logic matrix and constant parameters.

[0018] The electrical quantity data in step S101 includes current, voltage, power and frequency data.

[0019] The data mining and machine learning techniques in step S1012 include cluster analysis algorithms and neural network algorithms, and the optimization of the dynamic time warping algorithm includes: S1021. Introduce a sliding window mechanism to perform segmented matching on the current waveform; S1022. Adopting an adaptive threshold to eliminate noise interference and extract fundamental wave and harmonic distortion features; S1023. Establish a waveform similarity evaluation model based on the fault recording database to classify the fault type into short circuit, line break or ground fault.

[0020] The improved deep residual network in step S102 further includes an input layer receiving the time-frequency domain joint features of the voltage transient component; a residual module embedding an attention mechanism and an output layer generating an in-zone / out-of-zone fault probability distribution through a Softmax function.

[0021] In step S103, when dynamically adjusting the protection criterion, the equivalent impedance and short-circuit current of the line are calculated in real time to adaptively adjust the setting value of the current protection. The fault component criterion extracts the current and voltage fault components at the moment of fault occurrence and uses the amplitude, phase, and change rate of the fault components to perform fault judgment. The calculation method of the dynamic protection setting value includes the following steps: S1031. Establish a dynamic impedance model based on the fault current distribution and calculate the correction coefficient of the short-circuit current as the output of the distributed power source changes; S1032. Using a particle swarm optimization algorithm to find the Pareto optimal solution for the overcurrent threshold and the action time limit; S1033. Set the direction locking criterion to the phase difference threshold between the negative sequence voltage and the zero sequence current.

[0022] Among them, when using multi-source information to determine the nature of the fault in step S104, the duration, waveform characteristics and equipment status monitoring data of the fault current are analyzed. When optimizing the reclosing time, the type, capacity and access location of the distributed energy are considered, and the reclosing time is determined in combination with the transient stability calculation results of the system.

[0023] In step S105, the distributed protection units exchange information with each other via a high-speed communication network.

[0024] The fault warning model in step S106 is established by analyzing historical operating data and real-time monitoring data of the distribution network. Preventive measures taken include adjusting the output of distributed energy, optimizing load distribution, or adjusting the grid operation mode. The iterative optimization method includes the following steps: S1061. Constructing a protection strategy evaluation model based on a reinforcement learning framework; S1062. Update network weights using historical fault handling results as a reward function; S1063. Dynamically adjust the weight coefficients in the protection partition logic matrix.

[0025] In this embodiment, through the application of multi-source information fusion and adaptive protection criteria, the protection device can more accurately judge the fault, reduce the occurrence of false operation and refusal to operate, and improve the accuracy of distribution network protection. The implementation of intelligent reclosing strategy and distributed collaborative protection effectively avoids the secondary impact of reclosing on the system and the occurrence of islanding effect, improves the stability of the system, and ensures the safe operation of the distribution network.

[0026] Through the combination of distributed collaborative protection and fault warning and prevention mechanisms, faults can be quickly isolated, power supply to non-fault areas can be restored, and faults can be prevented in advance, significantly improving the power supply reliability of the distribution network, reducing the duration and scope of power outages, and effectively responding to the challenges brought by distributed energy access and the complex topology of the distribution network, adapting to the development needs of the smart grid, and providing technical support for the intelligent upgrade of the distribution network.

[0027] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A distribution network protection method, characterized in that: The steps include: Step S101: Real-time data collection from multiple sources of the distribution network is performed, and the collected electrical quantity data, distributed energy output data, meteorological data, equipment status monitoring data, voltage and current waveforms of each node, topology structure, distributed power output status, and environmental parameters are transmitted to the protection device or control center via the communication network; Step S102: Use data mining and machine learning techniques to deeply analyze and process multi-source information, identify data features, extract distortion features from the current waveform based on a dynamic time warping algorithm, identify fault types and suspected fault areas, and classify voltage transient components based on an improved deep residual network to distinguish between internal and external faults. Step S103: Combine the fault type, partition logic, and feature classification results to calculate dynamic protection settings, including overcurrent thresholds, directional blocking conditions, and action time limits. Dynamically adjust protection criteria based on the real-time operation mode and topology of the distribution network, introduce fault component criteria and comprehensive protection criteria, and simultaneously record protection process data based on blockchain technology to generate an unalterable fault tracing log. Step S104: Before reclosing, use multi-source information to determine the nature of the fault and optimize the reclosing time; Step S105: multiple distributed protection units are set up, and information is exchanged between the protection units. Protection instructions are simultaneously issued to the intelligent circuit breaker, distributed power supply controller, and energy storage device to perform coordinated protection actions. Step S106: Use big data analysis and machine learning technology to establish a fault warning model, take preventive measures based on the warning results, update the topology database based on the protection action results, and iteratively optimize the protection logic matrix and constant parameters.

2. The distribution network protection method according to claim 1, characterized in that: The electrical quantity data in step S101 includes current, voltage, power and frequency data.

3. The distribution network protection method according to claim 1, characterized in that: The data mining and machine learning techniques in step S1012 include cluster analysis algorithms and neural network algorithms, and the optimization of the dynamic time warping algorithm includes: S1021. Introduce a sliding window mechanism to perform segmented matching on the current waveform; S1022. Adopting an adaptive threshold to eliminate noise interference and extract fundamental wave and harmonic distortion features; S1023. Establish a waveform similarity evaluation model based on the fault recording database to classify the fault type into short circuit, line break or ground fault.

4. The distribution network protection method according to claim 1, wherein: The improved deep residual network in step S102 further includes an input layer receiving a time-frequency domain joint feature of a voltage transient component; The residual module embeds the attention mechanism and the output layer generates the in-region / out-of-region fault probability distribution through the Softmax function.

5. The distribution network protection method according to claim 1, characterized in that: When dynamically adjusting the protection criterion in step S103, the effective impedance and short-circuit current of the line are calculated in real time to adaptively adjust the setting value of the current protection. The fault component criterion extracts the current and voltage fault components at the moment of fault occurrence and uses the amplitude, phase, and change rate of the fault components to perform fault judgment. The calculation method of the dynamic protection setting value includes the following steps: S1031. Establish a dynamic impedance model based on the fault current distribution and calculate the correction coefficient of the short-circuit current as the output of the distributed power source changes; S1032. Using a particle swarm optimization algorithm to find the Pareto optimal solution for the overcurrent threshold and the action time limit; S1033. Set the direction locking criterion to the phase difference threshold between the negative sequence voltage and the zero sequence current.

6. The distribution network protection method according to claim 1, characterized in that: In step S104, when multi-source information is used to determine the nature of the fault, the duration, waveform characteristics and equipment status monitoring data of the fault current are analyzed. When optimizing the reclosing time, the type, capacity and access location of the distributed energy are considered, and the reclosing time is determined in combination with the transient stability calculation results of the system.

7. The distribution network protection method according to claim 1, characterized in that: In step S105, the distributed protection units exchange information with each other via a high-speed communication network.

8. The distribution network protection method according to claim 1, characterized in that: In step S106, the fault warning model is established by analyzing the historical operation data and real-time monitoring data of the distribution network. Preventive measures taken include adjusting the output of distributed energy, optimizing load distribution, or adjusting the grid operation mode. The iterative optimization method includes the following steps: S1061. Constructing a protection strategy evaluation model based on a reinforcement learning framework; S1062. Update network weights using historical fault handling results as a reward function; S1063. Dynamically adjust the weight coefficients in the protection partition logic matrix.

Images