A power management method for a distributed fault location device for a transmission line

Abstract

The present invention relates to a power management method suitable for a distributed fault locating device for a transmission line, comprising the following steps: dividing a plurality of power supply modes; dynamically and adaptively adjusting the current load threshold interval and the light intensity threshold interval of each power supply mode; realizing mode switching based on a hysteresis decision mechanism and a linear transition mechanism; and implementing a dynamic load reduction strategy based on power demand and the remaining capacity of the battery module when both the current load level and the light intensity reach the lowest power supply mode. By structuring an intelligent power management method, the problem of dynamic optimization of multi-source power supply of the device under complex working conditions is accurately solved. The system collects key parameters such as current, light, and battery status in real time, and automatically adjusts the power supply strategy according to the operating characteristics of the traveling wave fault locating device, cleverly realizing efficient energy distribution under different load and light conditions. It avoids the prominent problems of power switching lag, energy waste, and insufficient battery life in extreme environments in traditional power management solutions.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply technology for transmission lines, and in particular to a power supply management method applicable to a distributed fault location device for transmission lines. Background Art

[0002] Transmission lines are the critical arteries of the power system, and their stability directly impacts the reliability of energy delivery. With the expansion of ultra-high voltage (UHV) networks and the widespread integration of distributed energy resources, fault location and monitoring of transmission lines are placing higher demands on power management systems. Distributed fault location devices must operate continuously in complex environments, and their power management systems must efficiently coordinate multiple energy sources, such as solar energy and current transformers, while dynamically adapting to load changes and environmental fluctuations.

[0003] Existing power management systems typically use fixed, pre-set thresholds and lack the ability to adaptively adjust parameters. This can easily lead to reduced power supply efficiency in complex and changing power transmission environments, impacting device performance and reliability. Furthermore, current mode switching mechanisms are inflexible and lack a hysteresis decision-making mechanism. This simplistic switching approach can lead to system instability and shock during mode switching, reducing system reliability and service life. Finally, existing systems also lack dynamic load shedding, failing to adjust in real time based on power demand and battery capacity. This can lead to energy waste and impact battery efficiency and lifespan.

[0004] In summary, existing technologies have obvious defects in adaptive adjustment, mode switching and dynamic load reduction, which makes it difficult to meet the needs of modern power transmission systems for efficient, reliable and intelligent power management. Solutions are urgently needed to improve performance. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a power management method applicable to a distributed fault location device for a power transmission line, so as to overcome the deficiencies in the above-mentioned prior art.

[0006] The present invention solves the above technical problems with the following technical solution: A power management method applicable to a distributed fault location device for a power transmission line comprises the following steps:

[0007] Step S01: Dividing multiple power supply modes based on the current load threshold interval and the light intensity threshold interval, wherein the multiple power supply modes are composed of one, two or three of the CT induction power module, the solar power module and the battery module;

[0008] Step S02: Real-time cyclic collection includes the following data:

[0009] CT induction power module output current and output voltage ;

[0010] Light intensity of solar power module , output voltage and output current ;

[0011] Remaining capacity of the battery module , terminal voltage and charge / discharge current ;

[0012] Operating voltage of system load , working current and load power ;

[0013] Current power supply mode;

[0014] Step S03: Based on the real-time data collected in step S02, combined with a historical database storing historical operating condition threshold parameters and typical scene characteristics, a threshold adaptive adjustment algorithm driven by Bayesian optimization and meta-learning is used to dynamically and adaptively adjust the current load threshold range and light intensity threshold range for each power supply mode;

[0015] Step S04: When the output current of the CT sensing power module is collected in real time Falling into the current load threshold range, the light intensity of the solar power module collected in real time When the light intensity falls within the threshold range, mode switching is achieved based on the hysteresis decision mechanism and linear transition mechanism;

[0016] Step S05: When the current load level and light intensity both reach the lowest power supply mode, the power supply is switched to the lowest power supply mode according to the power demand and the remaining capacity of the battery module. Implement dynamic load shedding strategies.

[0017] The beneficial effects of the present invention are as follows: the present invention is specially designed for a distributed traveling wave fault locating device for power transmission, and accurately solves the problem of dynamic optimization of multi-source power supply of the device under complex working conditions by constructing an intelligent power management method. The system collects key parameters such as current, light, and battery status in real time, and automatically adjusts the power supply strategy according to the operating characteristics of the traveling wave fault locating device, cleverly realizing efficient energy distribution under different load and light conditions. This innovative solution effectively avoids prominent problems such as power switching lag, energy waste, and insufficient battery life in extreme environments in traditional power management solutions. Its significant advantage is that it provides the distributed traveling wave fault locating device for power transmission with adaptive adjustment capabilities of multi-source collaborative power supply, greatly improving energy utilization efficiency and adaptability to extreme environments, while reducing system maintenance costs, effectively ensuring the long-term stable operation of the device, and providing solid energy support for reliable monitoring of transmission lines.

[0018] On the basis of the above technical solution, the present invention can also be improved as follows.

[0019] Furthermore, the power supply mode in step S01 is divided into a high load mode, a medium load mode, and a low load mode according to the current load threshold interval; the medium load mode is divided into a medium load high light mode and a medium load low light mode according to the light intensity threshold interval; and the low load mode is divided into a low load high light mode and a low load low light mode according to the light intensity threshold interval;

[0020] High-load mode includes: CT induction power module as main power supply, solar power module as supplementary power supply, and excess energy is used for floating charge of battery module;

[0021] The medium load and high sunlight mode includes: the solar power module is the main power supply, the CT induction power module is the supplementary power supply, and the excess energy is used for floating charge of the battery module;

[0022] The medium-load and low-light mode includes: the CT induction power module provides the main power supply, the solar power module provides the supplementary power supply, and the excess energy is used for floating charge of the battery module;

[0023] Low-load and high-light modes include: battery module power supply, solar power module power supply;

[0024] Low load and low light mode includes: battery module powered, and based on the remaining capacity of the battery module Achieve dynamic load shedding.

[0025] Furthermore, step S05 specifically includes the following steps:

[0026] Step S51: Determine whether a dynamic load reduction strategy needs to be implemented based on power:

[0027] like , the system operates normally;

[0028] like , the system dynamically reduces load;

[0029] Step S511: Dynamically adjust computing power:

[0030] Adjusted hashrate ,and ;

[0031] Where, is the rated computing power, To ensure the minimum functional computing power, is the maximum output power, is the total power consumption of the current system;

[0032] Dynamically adjust CPU frequency based on computing task priority;

[0033] Step S512: Adaptive adjustment of communication rate:

[0034] Adjusted baud rate ;

[0035] Where, is the rated baud rate;

[0036] Use an intermittent transmission mode of sending status data every 10 minutes to extend the wake-up cycle;

[0037] Extended wake-up period , , In the formula, is the basic cycle;

[0038] Step S513: Based on the dynamic adjustment of computing power in step S511 and the adaptive adjustment of communication rate in step S512, a response mechanism is made;

[0039] Step S52: Pass Determine whether a dynamic load reduction strategy needs to be implemented:

[0040] like , the system operates normally;

[0041] like , the system uses the method in step S51 to perform dynamic load reduction.

[0042] Furthermore, the priority order of the computing power tasks in step S511 is:

[0043] Priority 1, task type is fault detection or battery status detection, and load shedding policy is not allowed;

[0044] Priority 2, task type is edge computing, and the load reduction strategy is to reduce the batch size;

[0045] Priority 3, task type is data caching and preprocessing, and the load reduction strategy is to extend the processing cycle;

[0046] Priority 4, task type is non-real-time data transmission, and load reduction strategy is to pause or compress transmission.

[0047] Furthermore, the response mechanism in step S513 is:

[0048] Phase 1, The computing power is adjusted to: reduce the frequency, retain 80% computing power; the communication is adjusted to: reduce the baud rate by 40%;

[0049] Phase 2, ; Hashrate adjustment: shut down non-core tasks and retain 50% of the hashrate; communication adjustment: intermittent transmission;

[0050] Phase 3, The computing power is adjusted to: deep sleep, only retaining the real-time clock; the communication is adjusted to: stop communication, and only wake up for detection every 30 minutes.

[0051] Furthermore, the threshold adaptive adjustment algorithm driven by Bayesian optimization and meta-learning specifically includes:

[0052] Step S31: Input the output current of the CT sensing power module , the light intensity of the solar power module , the remaining capacity of the battery module And the historical 24-hour operation data, using the dynamic time warping (DTW) algorithm to calculate the matching degree Sim between the current working condition and the historical scenario;

[0053] Step S32: If the matching degree Sim with a certain historical scene is ≥90%, the meta-learning rapid response is triggered, and the historical optimization threshold parameters corresponding to the historical scene are directly called to quickly switch the power supply mode; if the matching degree Sim of all historical scenes is <90%, the Bayesian optimization search is triggered; the threshold is iteratively optimized in the parameter space, and the new optimization result is stored in the historical scene database.

[0054] Furthermore, the Sim calculation method in step S31 is as follows:

[0055] Step S311: Calculate the Euclidean distance between the real-time sequence and each time point in the reference template to construct a two-dimensional distance matrix;

[0056] Step S312: Find the optimal curved path through dynamic programming to minimize the cumulative distance, obtain the minimum path distance D, and normalize D to the matching degree Sim. The normalization formula is as follows;

[0057] ;

[0058] Where max_distance is the maximum possible distance;

[0059] Step S313: Calculate 、 、 The matching degree of each dimension is weighted averaged according to the weight distribution to obtain the comprehensive matching degree Sim.

[0060] Furthermore, the triggering of the Bayesian optimization search in step S32 specifically includes:

[0061] Construct the evaluation function that guides the Bayesian optimization search:

[0062] ;

[0063] Where, is the threshold parameter to be optimized; is the total power consumption of the current system; is the total input power; This is a battery health penalty item, which increases the penalty when the battery capacity is lower than the float charge value; is the mode switching frequency penalty term, which is linked to the threshold interval of the hysteresis decision; is the weight coefficient, which is dynamically adjusted according to the power supply mode. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is the overall flow chart of the present invention;

[0065] Figure 2 This is a flow chart of the threshold adaptive adjustment algorithm driven by Bayesian optimization and meta-learning of the present invention. DETAILED DESCRIPTION

[0066] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0067] like Figures 1 and 2 As shown, in embodiment 1, a power management method applicable to a distributed fault location device for a transmission line includes the following steps:

[0068] Step S01: Dividing multiple power supply modes based on the current load threshold interval and the light intensity threshold interval, wherein the multiple power supply modes are composed of one, two or three of the CT induction power module, the solar power module and the battery module;

[0069] Step S02: Real-time cyclic collection includes the following data:

[0070] CT induction power module output current and output voltage ;

[0071] Light intensity of solar power module , output voltage and output current ;

[0072] Remaining capacity of the battery module , terminal voltage and charge / discharge current ;

[0073] Operating voltage of system load , working current and load power ;

[0074] Current power supply mode;

[0075] Step S03: Based on the real-time data collected in step S02, combined with a historical database storing historical operating condition threshold parameters and typical scene characteristics, a threshold adaptive adjustment algorithm driven by Bayesian optimization and meta-learning is used to dynamically and adaptively adjust the current load threshold range and light intensity threshold range for each power supply mode;

[0076] Step S04: When the output current of the CT sensing power module is collected in real time Falling into the current load threshold range, the light intensity of the solar power module collected in real time When the light intensity falls within the threshold range, mode switching is achieved based on the hysteresis decision mechanism and linear transition mechanism;

[0077] Step S05: When the current load level and light intensity both reach the lowest power supply mode, the power supply is switched to the lowest power supply mode according to the power demand and the remaining capacity of the battery module. Implement dynamic load shedding strategies.

[0078] The present invention is designed specifically for a distributed traveling wave fault locating device for power transmission. It accurately solves the problem of dynamic optimization of multi-source power supply for the device under complex working conditions by constructing an intelligent power management method. The system collects key parameters such as current, light, and battery status in real time, and automatically adjusts the power supply strategy based on the operating characteristics of the traveling wave fault locating device, cleverly achieving efficient energy distribution under different load and light conditions. This innovative solution effectively avoids prominent problems in traditional power management solutions such as power switching lag, energy waste, and insufficient battery life in extreme environments. Its significant advantage is that it provides the distributed traveling wave fault locating device for power transmission with adaptive adjustment capabilities for multi-source collaborative power supply, greatly improving energy utilization efficiency and adaptability to extreme environments, while reducing system maintenance costs, effectively ensuring the long-term stable operation of the device, and providing solid energy support for reliable monitoring of transmission lines.

[0079] Example 2: This example is a further improvement on Example 1, and its details are as follows:

[0080] The power supply mode in step S01 is divided into high load mode, medium load mode and low load mode according to the current load threshold interval; the medium load mode is divided into medium load high light mode and medium load low light mode according to the light intensity threshold interval; the low load mode is divided into low load high light mode and low load low light mode according to the light intensity threshold interval;

[0081] High-load mode includes: CT induction power module as main power supply, solar power module as supplementary power supply, and excess energy is used for floating charge of battery module;

[0082] The medium load and high sunlight mode includes: the solar power module is the main power supply, the CT induction power module is the supplementary power supply, and the excess energy is used for floating charge of the battery module;

[0083] The medium-load and low-light mode includes: the CT induction power module provides the main power supply, the solar power module provides the supplementary power supply, and the excess energy is used for floating charge of the battery module;

[0084] Low-load and high-light modes include: battery module power supply, solar power module power supply;

[0085] Low load and low light mode includes: battery module powered, and based on the remaining capacity of the battery module Achieve dynamic load shedding.

[0086] Example 3: This example is a further improvement on Example 2, and its details are as follows:

[0087] Step S05 specifically includes the following steps:

[0088] Step S51: Determine whether a dynamic load reduction strategy needs to be implemented based on power:

[0089] like , the system operates normally;

[0090] like , the system dynamically reduces load;

[0091] Step S511: Dynamically adjust computing power:

[0092] Adjusted hashrate ,and ;

[0093] Where, is the rated computing power, To ensure the minimum functional computing power, is the maximum output power, is the total power consumption of the current system;

[0094] Dynamically adjust CPU frequency based on computing task priority;

[0095] Step S512: Adaptive adjustment of communication rate:

[0096] Adjusted baud rate ;

[0097] Where, is the rated baud rate;

[0098] Use an intermittent transmission mode of sending status data every 10 minutes to extend the wake-up cycle;

[0099] Extended wake-up period , , In the formula, is the basic cycle;

[0100] Step S513: Based on the dynamic adjustment of computing power in step S511 and the adaptive adjustment of communication rate in step S512, a response mechanism is made;

[0101] Step S52: Pass Determine whether a dynamic load reduction strategy needs to be implemented:

[0102] like , the system operates normally;

[0103] like , the system uses the method in step S51 to perform dynamic load reduction.

[0104] Example 4: This example is a further improvement on Example 3, and its details are as follows:

[0105] The priority order of computing tasks in step S511 is:

[0106] Priority 1, task type is fault detection or battery status detection, and load shedding policy is not allowed;

[0107] Priority 2, task type is edge computing, and the load reduction strategy is to reduce the batch size;

[0108] Priority 3, task type is data caching and preprocessing, and the load reduction strategy is to extend the processing cycle;

[0109] Priority 4, task type is non-real-time data transmission, and load reduction strategy is to pause or compress transmission.

[0110] Example 5: This example is a further improvement on Example 3, and its details are as follows:

[0111] The response mechanism in step S513 is:

[0112] Phase 1, The computing power is adjusted to: reduce the frequency, retain 80% computing power; the communication is adjusted to: reduce the baud rate by 40%;

[0113] Phase 2, ; Hashrate adjustment: shut down non-core tasks and retain 50% of the hashrate; communication adjustment: intermittent transmission;

[0114] Phase 3, The computing power is adjusted to: deep sleep, only retaining the real-time clock; the communication is adjusted to: stop communication, and only wake up for detection every 30 minutes.

[0115] Example 6: This example is a further improvement on Example 1, and its details are as follows:

[0116] The Bayesian optimization-meta-learning algorithm specifically includes:

[0117] Step S31: Input the output current of the CT sensing power module , the light intensity of the solar power module , the remaining capacity of the battery module As well as historical 24-hour operating data, the dynamic time warping (DTW) algorithm is used to calculate the matching degree Sim (range 0~1) between the current working condition and the historical scenario;

[0118] Step S32: If the matching degree Sim with a certain historical scene is ≥90%, meta-learning rapid response is triggered, and the historical optimization threshold parameters corresponding to the historical scene are directly called to quickly switch the power supply mode to achieve millisecond-level response; if the matching degree Sim of all historical scenes is <90%, Bayesian optimization search is triggered; the present invention takes "minimizing comprehensive energy consumption + maximizing battery health" as the optimization goal, iteratively optimizes the threshold in the parameter space, and stores the new optimization results in the historical scene database.

[0119] Example 7: This example is a further improvement on Example 6, and its details are as follows:

[0120] The Sim calculation method in step S31 is as follows:

[0121] Step S311: Calculate the Euclidean distance between the real-time sequence and each time point in the reference template to construct a two-dimensional distance matrix;

[0122] 1. Real-time sequence

[0123] The real-time sequence is a time series composed of continuous monitoring data collected at fixed time intervals during the current operation phase of the system, covering various operating parameters related to threshold adjustment.

[0124] (1) Current sequence: real-time sampling value of load current over a period of time , It is The current value at a certain moment is used to reflect the load change dynamics.

[0125] (2) Light sequence: real-time sampling sequence of light intensity in photovoltaic power supply scenarios , reflecting the temporal fluctuations of lighting conditions.

[0126] (3) Battery status sequence: Battery , voltage and other parameters of the timing data , assisting in determining battery charging and discharging requirements and mode switching basis.

[0127] These sequences are continuously updated to accurately depict the current operating conditions of the system.

[0128] 2. Reference Template

[0129] The reference template is a time series data template of working conditions with typical characteristics that is selected from the historical scenario database. It is used to compare with the real-time sequence to determine which historical scenario the current working condition is similar to.

[0130] The typical working condition template is based on historical operating data, and performs cluster analysis on different working conditions (such as stable working conditions such as high load and high light, low load and low light, or transient working conditions such as sudden load change and sudden light change), and extracts representative time series data. For example, in the historical high load stable operation scene, a continuous data segment with stable current in high range and sufficient light is selected and organized into , Serves as a reference template for corresponding working conditions.

[0131] Optimal Threshold Association Template: The historical time series data corresponding to the optimal threshold configuration after Bayesian optimization convergence is used as a reference template. For example, if the threshold is adjusted to the optimal value during a high-load scenario, and the system exhibits good energy efficiency and stability, the corresponding time series data, such as current and light intensity, can serve as a reference template for this high-load scenario. The threshold strategy can be reused when encountering similar real-time sequences in the future.

[0132] The reference template stores the characteristic laws of historical working conditions and is the basis for meta-learning to achieve "guiding current optimization based on historical experience".

[0133] Step S312: Find the optimal curved path through dynamic programming to minimize the cumulative distance, obtain the minimum path distance D, and normalize D to the matching degree Sim. The normalization formula is as follows;

[0134] ;

[0135] Where max_distance is the maximum possible distance;

[0136] Step S313: Calculate 、 、 The matching degree of each dimension is weighted averaged according to the weight distribution to obtain the comprehensive matching degree Sim.

[0137] Example 8: This example is a further improvement on Example 6, and its details are as follows:

[0138] Triggering the Bayesian optimization search in step S32 specifically includes:

[0139] Construct the evaluation function that guides the Bayesian optimization search:

[0140] ;

[0141] Where, is the threshold parameter to be optimized; is the total power consumption of the current system; is the total input power; This is a battery health penalty item, which increases the penalty when the battery capacity is lower than the float charge value; is the mode switching frequency penalty term, which is linked to the threshold interval of the hysteresis decision; is the weight coefficient, which is dynamically adjusted according to the power supply mode.

[0142] The specific embodiments of the present invention are as follows:

[0143] The detailed steps are as follows:

[0144] Step 1: When the intelligent power management system boots up, the hardware module activates the power path, completes self-tests of the CT induction power module, solar power module, and battery module, and reads the battery module's initial status. Simultaneously, the software module sets initial threshold parameter values ​​and activates the scene recognition module. Based on this, a hybrid Bayesian optimization and meta-learning algorithm begins initialization, creating a historical database that stores historical operating condition threshold parameters and typical scene characteristics. After completing this initialization process, the system enters the real-time data acquisition loop.

[0145] Step 2: Real-time acquisition of the output current of the CT induction power module and output voltage ; Light intensity of solar power module , output voltage and output current ;Remaining capacity of the battery module , terminal voltage and charge / discharge current ; Operating voltage of system load and operating current ; Current power supply mode.

[0146] Step 3: Based on the real-time data collected in Step 2, the threshold adaptive adjustment algorithm driven by Bayesian optimization and meta-learning is used to achieve the following: high load current threshold interval , medium load current threshold range , medium load high light threshold range , low load and high light threshold range , low light threshold And dynamic adaptive adjustment of battery float charge cut-off capacity.

[0147] Step 4: The high load switching threshold interval is The system realizes mode switching through hysteresis decision and linear transition mechanism. If the power supply is too high and lasts for more than 30 seconds, the system enters high-load mode. At this time, the system uses the CT induction power module as the main power source for the device, the solar power module as a supplementary power source, and uses the excess energy to charge the battery module. The details are as follows:

[0148] If the load lasts for more than 30 seconds, the system is in high load mode. The CT induction power module has sufficient capacity and is optimized to achieve high efficiency. , so it takes on 90% of the load power first, and the remaining 10% is supplemented by the solar module. The initial float charge cutoff capacity is set to 95%. The battery is then charged according to the float charge cutoff capacity optimized by the Bayesian Optimization-Meta-Learning algorithm in "Step 3."

[0149] CT induction power module main output power , solar modules supplement the power supply output power , battery float charge .

[0150] At the same time, in order to avoid high-frequency switching, the system realizes mode switching through hysteresis decision and linear transition mechanism. The initial value of is set to [45A, 55A], and the current threshold interval is then updated in real time by the Bayesian optimization-meta-learning algorithm in "Step 3". If the load lasts for more than 30 seconds, it will enter high load mode. If the high load mode lasts for more than 30 seconds, the system will exit the high load mode. When , the proportion of CT induction power modules changes linearly, and the power supply mode transitions to medium load mode. Subsequently, the specific mode is further distinguished by the lighting conditions.

[0151] Step 5: The medium load switching threshold range is , the medium load and high light threshold interval is The system realizes mode switching through hysteresis decision and linear transition mechanism. 、 If the power supply lasts for more than 60 seconds, the system enters the medium load and high light mode. At this time, the system uses the solar power module as the main power supply for the device, the CT induction power module as a supplementary power supply, and uses the excess energy to charge the battery module. The specific details are as follows:

[0152] when and During this time, the system is in medium-load, high-insolation mode. Using the MPPT algorithm, the system uses the solar power module as the primary power source, supplemented by the CT induction power module, and charges the battery module according to the float charge cutoff capacity optimized by the Bayesian optimization-meta-learning algorithm in Step 3.

[0153] Output power , CT induction power module supplements power supply output power , battery float charge .

[0154] At the same time, in order to avoid high-frequency switching, the system uses a hysteresis mechanism and a linear transition mechanism to achieve mode switching and set the high light threshold range to The initial value of is set to [90lux, 110lux], and the light threshold interval is then updated in real time by the Bayesian optimization-meta-learning algorithm in "Step 3". 、 If the light level lasts for more than 60 seconds, it will enter the medium-load high-light mode. 、 If the light level lasts for more than 60 seconds, the system will exit the medium-load high-light mode. When the power supply mode is low, the proportion of solar power modules changes linearly, and the power supply mode transitions to low light mode. The specific mode is further distinguished by the current conditions.

[0155] Step 6: The medium load switching threshold range is , the low light threshold is The system realizes mode switching through hysteresis decision and linear transition mechanism. 、 If the power supply lasts for more than 180 seconds, the system enters the medium load low light mode. At this time, the system uses the CT induction power module as the main power supply, the solar power module as the supplementary power supply, and the excess energy is used to charge the battery module. The details are as follows:

[0156] when 、 If the power supply lasts for more than 180 seconds, the system is in the medium load and low light mode, and the solar power module and the CT induction power module are used to supply power. , CT induction power module output power .

[0157] At the same time, to ensure stable switching, the system implements mode switching through hysteresis decision and linear transition mechanism. The initial value of is set to [5A, 15A], and the current threshold interval is then updated in real time by the Bayesian optimization-meta-learning algorithm in "Step 3". 、 If the light level lasts for more than 180 seconds, it will enter the medium-load low-light mode. 、 If the light level remains low for more than 180 seconds, the system will exit the medium-load low-light mode. When , the proportion of CT induction power modules changes linearly, and the power supply mode transitions to medium load mode. Subsequently, the specific mode is further distinguished by the lighting conditions.

[0158] Step 7: Low load threshold is , the low load and high light switching threshold interval is The system realizes mode switching through hysteresis decision and linear transition mechanism. 、 If the power is kept on for more than 300 seconds, the system will enter the low-load high-light mode. At this time, the system uses the solar power module and the battery module to power the device. The details are as follows:

[0159] when 、 When the power supply lasts for more than 300 seconds, the system is in a low-load and high-light mode. At this time, the system adopts a combined power supply mode, but the solar power module is the main power supply. , battery module output power .

[0160] At the same time, in order to avoid system misjudgment, the system realizes mode switching through hysteresis decision and linear transition mechanism. The initial value of is set to [90lux, 110lux], and the light threshold interval is then updated in real time by the Bayesian optimization-meta-learning algorithm in "Step 3". 、 If the state lasts for more than 300 seconds, it will enter low-load and high-light mode; 、 If the state lasts for more than 300 seconds, the system will exit the low-load high-light mode. When the power supply mode is low, the proportion of solar power modules changes linearly, and the power supply mode transitions to low light mode. The specific mode is further distinguished by the current conditions.

[0161] Step 8: Low load threshold is , the low light threshold is The system realizes mode switching through hysteresis decision and linear transition mechanism. 、 If the power is off for more than 600 seconds, the system enters low-load and low-light mode. At this point, the battery module is used as the power supply for the device, and dynamic load reduction is implemented based on power demand and battery module capacity. The details are as follows:

[0162] when 、 When the system lasts for more than 600 seconds, it is in low-load and low-light mode. At this time, the battery module is used as the sole power supply mode, and dynamic load reduction is implemented according to power demand and battery module capacity.

[0163] At the same time, to avoid system misjudgment, the system implements mode switching through hysteresis decision and linear transition mechanism. The initial threshold values ​​of low load current and low light are set to 10A and 100lux respectively. The current threshold and light threshold are then updated in real time by the Bayesian optimization-meta-learning algorithm in "Step 3". When the system detects the low load current and low light threshold in "Step 2", the system will automatically switch between the two modes. 、 If the state lasts for more than 600 seconds, it will enter low-load and low-light mode; or If the low-load and low-light mode lasts for more than 600 seconds, the system will exit the low-load and low-light mode and switch to other power supply modes.

[0164] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A power management method for a distributed fault location device for a transmission line, characterized in that: The steps include: Step S01: Dividing a plurality of power supply modes based on a current load threshold interval and a light intensity threshold interval, wherein the plurality of power supply modes are selected from one, two or three of a CT induction power module, a solar power module and a battery module; Step S02: Real-time cyclic collection includes the following data: CT induction power module output current and output voltage ; Light intensity of solar power module , output voltage and output current ; Remaining capacity of the battery module , terminal voltage and charge / discharge current ; Operating voltage of system load , working current and load power ; Current power supply mode; Step S03: Based on the real-time data collected in step S02, combined with a historical database storing historical operating condition threshold parameters and typical scene features, dynamically and adaptively adjust the current load threshold interval and the light intensity threshold interval of each power supply mode through a threshold adaptive adjustment algorithm driven by Bayesian optimization and meta-learning; Step S04: When the output current of the CT sensing power module is collected in real time Falling within the current load threshold range, the light intensity of the solar power module collected When the light intensity falls within the threshold range, mode switching is achieved based on a hysteresis decision mechanism and a linear transition mechanism; Step S05: When the current load level and light intensity both reach the lowest power supply mode, the battery module is switched to the lowest power supply mode according to the power demand and the remaining capacity of the battery module. Implement dynamic load shedding strategies.

2. A power management method applicable to a distributed fault location device for a transmission line according to claim 1, characterized in that: The power supply mode in step S01 is divided into a high load mode, a medium load mode and a low load mode according to the current load threshold interval; the medium load mode is divided into a medium load high light mode and a medium load low light mode according to the light intensity threshold interval; The low-load mode is divided into a low-load high-light mode and a low-load low-light mode according to the light intensity threshold interval; The high-load mode includes: the CT induction power module is the main power supply, the solar power module is the supplementary power supply, and the excess energy is used for floating charging of the battery module; The medium load and high light mode includes: the solar power module is the main power supply, the CT induction power module is the supplementary power supply, and the excess energy is used for floating charge of the battery module; The medium load and low light mode includes: the CT induction power module is the main power supply, the solar power module is the supplementary power supply, and the excess energy is used for floating charging of the battery module; The low-load and high-light mode includes: power supply by battery module and power supply by solar power module; The low load and low light mode includes: power supply by battery module and based on the remaining capacity of the battery module Achieve dynamic load shedding.

3. The power management method for a distributed fault location device for a power transmission line according to claim 2, characterized in that: The step S05 specifically includes the following steps: Step S51: Determine whether a dynamic load reduction strategy needs to be implemented based on power: like , the system operates normally; like , the system dynamically reduces load; Step S511: Dynamically adjust computing power: Adjusted hashrate ,and ; Where, is the rated computing power, To ensure the minimum functional computing power, is the maximum output power, is the total power consumption of the current system; Dynamically adjust CPU frequency based on computing task priority; Step S512: Adaptive adjustment of communication rate: Adjusted baud rate ; Where, is the rated baud rate; Use an intermittent transmission mode of sending status data every 10 minutes to extend the wake-up cycle; Extended wake-up period , , In the formula, is the basic cycle; Step S513: Based on the dynamic adjustment of computing power in step S511 and the adaptive adjustment of communication rate in step S512, a response mechanism is made; Step S52: Pass Determine whether a dynamic load reduction strategy needs to be implemented: like , the system operates normally; like The system uses the method in step S51 to perform dynamic load reduction.

4. A power management method applicable to a distributed fault location device for a power transmission line according to claim 3, characterized in that: The priority order of the computing power tasks in step S511 is: Priority 1, task type is fault detection or battery status detection, and load shedding policy is not allowed; Priority 2, task type is edge computing, and the load reduction strategy is to reduce the batch size; Priority 3, task type is data caching and preprocessing, and the load reduction strategy is to extend the processing cycle; Priority 4, task type is non-real-time data transmission, and load reduction strategy is to pause or compress transmission.

5. The power management method for a distributed fault location device for a power transmission line according to claim 3, characterized in that: The response mechanism in step S513 is: Phase 1, The computing power is adjusted to: reduce the frequency, retain 80% computing power; the communication is adjusted to: reduce the baud rate by 40%; Phase 2, ; Hashrate adjustment: shut down non-core tasks and retain 50% of the hashrate; communication adjustment: intermittent transmission; Phase 3, The computing power is adjusted to: deep sleep, only retaining the real-time clock; the communication is adjusted to: stop communication, and only wake up for detection every 30 minutes.

6. The power management method for a distributed fault location device for a power transmission line according to claim 1, characterized in that: The Bayesian optimization-meta-learning driven threshold adaptive adjustment algorithm specifically includes: Step S31: Input the output current of the CT sensing power module , the light intensity of the solar power module , the remaining capacity of the battery module And the historical 24-hour operation data, using the dynamic time warping (DTW) algorithm to calculate the matching degree Sim between the current working condition and the historical scenario; Step S32: If the matching degree Sim with a certain historical scene is ≥90%, the meta-learning rapid response is triggered, and the historical optimization threshold parameters corresponding to the historical scene are directly called to quickly switch the power supply mode; if the matching degree Sim of all historical scenes is <90%, the Bayesian optimization search is triggered; the threshold is iteratively optimized in the parameter space, and the new optimization result is stored in the historical scene database.

7. The power management method for a distributed fault location device for a power transmission line according to claim 6, characterized in that: The Sim calculation method in step S31 is as follows: Step S311: Calculate the Euclidean distance between the real-time sequence and each time point in the reference template to construct a two-dimensional distance matrix; Step S312: Find the optimal curved path through dynamic programming to minimize the cumulative distance, obtain the minimum path distance D, and normalize D to the matching degree Sim. The normalization formula is as follows; ； Where max_distance is the maximum possible distance; Step S313: Calculate 、 、 The matching degree of each dimension is weighted averaged according to the weight distribution to obtain the comprehensive matching degree Sim.

8. The power management method for a distributed fault location device for a power transmission line according to claim 6, characterized in that: The triggering of the Bayesian optimization search in step S32 specifically includes: Construct the evaluation function that guides the Bayesian optimization search: ； Where, is the threshold parameter to be optimized; is the total power consumption of the current system; is the total input power; This is a battery health penalty item, which increases the penalty when the battery capacity is lower than the float charge value; is the mode switching frequency penalty term, which is linked to the threshold interval of the hysteresis decision; is the weight coefficient, which is dynamically adjusted according to the power supply mode.

Images