Power grid edge monitoring terminal regulation device and method combining energy active sensing

Abstract

This invention relates to the field of intelligent monitoring technology for power systems, and discloses a control device and method for a power grid edge monitoring terminal combined with active energy sensing. The device includes: a micro-meteorological sensing module for real-time acquisition of meteorological parameters of the environment where the monitoring terminal is located; an energy prediction module for predicting expected energy gains within a future time window using a time-series prediction model; an energy assessment module for calculating an energy availability score; a mode decision module for dynamically switching the operating mode of the monitoring terminal based on a comparison of the energy availability score with a preset threshold; a defect detection module for real-time defect detection of acquired images or video streams; and a verification decision module for determining whether to trigger a second power consumption mode for defect verification when the defect detection module detects a suspected defect in a first power consumption mode. This invention can significantly reduce the false negative rate of dynamic defects without increasing hardware costs or battery capacity.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology for power systems, specifically to a control device and method for a power grid edge monitoring terminal that combines active energy sensing. Background Technology

[0002] With the continuous expansion of the power grid and the accelerated advancement of new power systems, the safe and stable operation of power infrastructure such as transmission, transformation, and distribution is crucial as a core component of power grid operation. Due to the wide distribution of power facilities and the complex and diverse terrain, traditional manual inspection methods face numerous problems such as low efficiency, incomplete coverage, and delayed response. To address this challenge, intelligent monitoring networks have been deployed on a large scale in the transmission, transformation, and distribution sector. Among these, fixed, visual online monitoring terminals installed at poles, substations, and distribution rooms are the core equipment for achieving all-weather, all-time power equipment status awareness. These devices typically transmit monitoring data and perform preliminary analysis via 4G / 5G private networks or fiber optic communication.

[0003] Because monitoring terminals in the power transmission and distribution sector are mostly located in remote areas or enclosed spaces, the difficulty of obtaining power is a key factor restricting the long-term stable operation of these terminals. Existing monitoring terminals generally use solar photovoltaic panels and batteries for power supply, with some high-voltage line scenarios using inductive power. During operation, mainstream equipment typically employs timed image capture or simple trigger-based operating modes, transmitting the collected equipment images or video data back to the cloud or performing local edge computing analysis. Although existing online intelligent monitoring networks have improved the efficiency of power equipment inspection to some extent, in actual operation, due to the randomness of energy acquisition and the complexity of dynamic defect detection, the following problems urgently need to be addressed:

[0004] (1) Energy management strategy is passive and lagging. At present, monitoring terminal equipment generally lacks energy planning mechanism. The energy management strategy is usually based on passive feedback of the current battery voltage. The equipment cannot sense future weather changes, such as solar radiation intensity. As a result, when there is sufficient sunlight, the overflow energy is not fully utilized for high-frequency sampling and in-depth analysis. When there is a sudden continuous rainy weather, the equipment will frequently disconnect due to power depletion, which cannot guarantee the continuity and reliability of monitoring data.

[0005] (2) Static sensing mode is difficult to capture dynamic defects. In order to save energy as much as possible, existing monitoring terminals mostly adopt static image capture mode. However, many high-risk hazards faced by power equipment have significant dynamic characteristics, such as conductor galloping in transmission lines, abnormal fluctuations in transformer oil levels, changes in heating of distribution cabinet equipment, foreign objects hanging, and external damage to lines. Static capture is prone to missed detection or misjudgment of such dynamic defects, while if full-time video recording is enabled, its energy consumption is far greater than that of static capture, which will cause the battery to be depleted quickly. Therefore, how to dynamically adjust the acquisition mode according to the defect characteristics while ensuring detection accuracy is a challenge currently faced by intelligent monitoring terminals.

[0006] (3) Lack of adaptive scheduling capability based on environmental awareness. Existing monitoring terminals cannot intelligently adjust their working mode according to external environmental factors such as light intensity and weather conditions. Regardless of whether the lighting is excellent or the energy is scarce, the equipment executes a fixed sampling frequency and lacks an intelligent scheduling and switching mechanism. In addition, existing monitoring terminals often only focus on the energy status at a single moment and fail to optimize the energy collection benefits and consumption globally, resulting in low energy utilization and limited effective working time.

[0007] (4) Lack of comprehensive assessment of edge-side intelligent reasoning and energy-coordinated optimization. Currently, intelligent algorithms are deployed on monitoring terminal devices, enabling local real-time detection and identification of defects. However, the operation of these algorithms consumes computing resources and energy. Currently, there is a lack of comprehensive consideration of energy consumption in various stages such as image acquisition and reasoning, and data transmission, and a lack of a decision-making mechanism that coordinates energy constraints with detection benefits.

[0008] In summary, existing online monitoring terminals for power transmission channels urgently need a method that can combine weather forecasts for proactive energy management and intelligently switch between video and image acquisition modes while ensuring energy continuity, in order to solve the problems of missed detection of dynamic hidden dangers and defects and easy equipment offline. Summary of the Invention

[0009] This invention aims to solve the aforementioned problems in the prior art and provides a control device and method for a power grid edge monitoring terminal that combines active energy sensing. Specifically, this invention addresses the following technical problems:

[0010] (1) Solve the problem that existing power monitoring terminal devices rely solely on the current battery voltage for passive energy management, which cannot predict future weather conditions and thus lead to energy overflow waste / power depletion and offline operation. At the same time, solve the problem of insufficient energy planning of monitoring devices in scenarios such as substations and power transmission channels.

[0011] (2) It solves the problem that the traditional static capture mode of the monitoring device cannot effectively identify dynamic defects such as conductor galloping and external construction, while avoiding the rapid depletion of batteries caused by full-time video recording.

[0012] (3) To solve the problem that existing monitoring terminals lack end-side intelligent reasoning and energy consumption collaborative optimization, establish a quantitative relationship between energy loss and monitoring benefits in each link such as image acquisition, algorithm reasoning, and data transmission, and realize intelligent decision-making under energy constraints.

[0013] This invention provides a power grid edge monitoring terminal control device that combines active energy sensing. This device introduces micro-meteorological forecasting into the edge computing terminal, constructs an active management framework based on energy revenue expectations, and realizes intelligent scheduling of operations such as power image acquisition, real-time end-side reasoning, and dynamic defect verification.

[0014] The core innovation of this technical solution lies in the following: Under normal working mode, the monitoring terminal captures visible light images at a preset fixed frequency. The lightweight defect detection algorithm deployed on the edge performs inference analysis on the acquired images in real time. When a suspected defect is detected and further confirmation and verification are required, the system will comprehensively evaluate the current energy state, expected energy gain and energy consumption of the verification operation, and then dynamically decide whether to adjust the camera angle or switch the shooting mode from image mode to video mode, so as to achieve the optimal detection strategy under energy constraints.

[0015] To achieve the above objectives, the present invention adopts the following technical solution:

[0016] A grid edge monitoring terminal control device combining active energy sensing, comprising:

[0017] The micro-meteorological sensing module is used to collect meteorological parameters of the environment where the monitoring terminal is located in real time;

[0018] An energy prediction module, deployed at the edge, is used to predict expected energy gains within future time windows based on the time series of the meteorological parameters using a time-series prediction model.

[0019] The energy assessment module is used to calculate an energy availability score based on the current remaining energy storage capacity and the expected energy gain.

[0020] The mode decision module is used to dynamically switch the working mode of the monitoring terminal based on the comparison result of the energy availability score and the preset threshold. The working mode includes at least a first power consumption mode and a second power consumption mode, wherein the energy consumption per unit time of the second power consumption mode is higher than that of the first power consumption mode.

[0021] The defect detection module, deployed at the edge, is used to perform real-time defect detection on the acquired images or video streams;

[0022] The review decision module is used to comprehensively evaluate the currently available remaining energy, the energy consumption required for the review operation, the detection benefit of the suspected defect, and the detection confidence level when the defect detection module detects a suspected defect in the first power consumption mode, and determine whether to trigger the second power consumption mode for defect review.

[0023] As an optimization, the meteorological parameters include at least one of solar irradiance, ambient temperature, ambient humidity, atmospheric pressure, and cloud cover.

[0024] The time-series prediction model is a long short-term memory network or a gated recurrent unit;

[0025] The expected energy gain is calculated based on predicted solar irradiance combined with a photovoltaic panel conversion model to obtain the expected power generation within the future time window. .

[0026] As an optimization, the energy assessment module calculates an energy availability score. The method is as follows:

[0027] ;

[0028] in, This represents the current remaining battery percentage. For the expected power generation, Rated peak power, This is a weighting coefficient used to adjust the proportion of current electricity and expected power generation in the decision-making process.

[0029] As an optimization, the mode decision module adopts an asymmetric threshold hysteresis switching mechanism:

[0030] Set high threshold and low threshold ,and ;

[0031] when When switching to the second power consumption mode, Assess energy availability.

[0032] when When the time comes, switch to the first power consumption mode;

[0033] when At that time, the current work mode will remain unchanged.

[0034] As an optimization, the review decision module specifically includes:

[0035] The energy consumption quantification submodule is used to calculate the energy consumption of the monitoring terminal under a predetermined operation, which includes at least: image acquisition energy consumption. Defect detection inference energy consumption Standby power consumption Gimbal motor energy consumption Video capture energy consumption One or more of the following;

[0036] The remaining energy calculation submodule is used to calculate the dynamically available remaining energy based on the current battery capacity, expected power generation, basic energy consumption, and safety minimum energy threshold. ;

[0037] The benefit assessment submodule is used to quantify the detection benefit value based on the risk level of the suspected defect. ;

[0038] The decision execution submodule is used to calculate the energy efficiency ratio. ,in Total energy consumption required to perform the review operation;

[0039] The decision execution submodule triggers a second power consumption mode for defect review when the following conditions are met simultaneously:

[0040] Condition A: ;

[0041] Condition B: , This is a preset energy efficiency ratio threshold;

[0042] Condition C: The detection confidence level is lower than the preset confidence threshold.

[0043] As an optimization, the first power consumption mode is an image acquisition mode, and the second power consumption mode is a video acquisition mode. In the image acquisition mode, the monitoring terminal captures visible light images at a preset fixed frequency and runs a lightweight defect detection algorithm. In the video acquisition mode, the monitoring terminal enables video stream acquisition and runs a video analysis algorithm to identify potential defects with temporal dynamic characteristics.

[0044] As an optimization, the monitoring terminal is applicable to self-powered IoT devices, including power transmission channel monitoring terminals, substation monitoring terminals, power distribution room monitoring terminals, forest fire prevention monitoring terminals, hydrogeological disaster monitoring terminals, inspection drones or inspection robots.

[0045] This invention also discloses a grid edge monitoring terminal control method combining active energy sensing, applied to the aforementioned grid edge monitoring terminal control device combining active energy sensing, comprising the following steps:

[0046] Meteorological parameters of the environment where the monitoring terminal is located are collected in real time. Based on the time series of the meteorological parameters, the expected energy gain in the future time window is predicted by a time series prediction model deployed on the edge side.

[0047] Calculate the energy availability score based on the current remaining energy storage capacity and the expected energy gain;

[0048] Based on the comparison result between the energy availability score and the preset threshold, the working mode of the monitoring terminal is dynamically switched, and the working mode includes at least a first power consumption mode and a second power consumption mode.

[0049] In the first power consumption mode, real-time defect detection is performed on the acquired images;

[0050] When a suspected defect is detected, the system comprehensively evaluates the currently available remaining energy, the energy consumption required for the review operation, the detection benefit of the suspected defect, and the detection confidence level to determine whether to trigger the second power consumption mode for defect review.

[0051] As an optimization, the specific process for determining whether to trigger the second power consumption mode for defect review is as follows:

[0052] A refined energy loss model is established for the monitoring terminal in image acquisition, defect detection inference, standby maintenance, gimbal rotation, and video acquisition operations.

[0053] Calculate the dynamic available remaining energy ,in This represents the current available battery power. To predict the total expected power generation within the forecast window, Based on energy consumption, To ensure a safe minimum energy threshold;

[0054] Quantify the detection benefit value based on the risk level of suspected defects. ;

[0055] Obtain the detection confidence level of the defect detection module for suspected defects;

[0056] Calculate the energy efficiency ratio ,in Total energy consumption required for the verification operation;

[0057] The second power consumption mode is triggered for defect review when all three of the following conditions are met:

[0058] ;

[0059] , This is a preset energy efficiency ratio threshold;

[0060] The detection confidence level is lower than a preset confidence threshold.

[0061] As an optimization, the specific process of dynamically switching the working mode of the monitoring terminal based on the comparison result of the energy availability score and the preset threshold is as follows:

[0062] Set high threshold and low threshold ,and ;

[0063] when When switching to the second power consumption mode, Assess energy availability.

[0064] when When the time comes, switch to the first power consumption mode;

[0065] when At that time, the current work mode will remain unchanged.

[0066] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0067] (1) Realize active energy management of online monitoring devices for power transmission channels and solve the problem of easy disconnection of passive power monitoring equipment. This invention introduces an edge-side ultra-short-term meteorological forecast model, which can detect changes in solar irradiance in the future time window in advance, so as to make full use of the overflow energy for high-frequency operation when sufficient sunlight is predicted, and reduce the frequency in advance to ensure power supply when insufficient sunlight is predicted.

[0068] (2) Solving the problem of high missed detection rate of dynamic hidden dangers and further improving the efficiency of channel inspection. Traditional static capture mode is difficult to accurately detect dynamic defects such as conductor galloping, foreign object drifting, and insulator flashover. The present invention designs an energy availability score that incorporates future power generation expectations into the decision-making factors, comprehensively considers the energy consumption threshold and energy efficiency ratio of the review operation, and establishes a dynamic mode switching mechanism based on energy perception, which can capture dynamic hidden dangers with time-series characteristics in video mode. Without increasing hardware costs and battery capacity, it can significantly reduce the missed detection rate of dynamic defects.

[0069] (3) Design a refined energy loss model to improve the accuracy of edge decision-making. This invention incorporates the energy consumption of edge camera image acquisition, edge inference, and device standby into the energy consumption calculation of the prediction time window, providing an accurate basis for the dynamic mode switching of edge intelligent devices and preventing decision-making errors caused by incorrect energy assessment.

[0070] (4) The algorithm model is adapted to edge heterogeneous computing power, low-cost deployment, and suitable for large-scale engineering promotion. The technical method of this invention has low computational complexity and can be directly deployed on the embedded microcontroller (MCU) or heterogeneous AI chip of the monitoring terminal with limited computing power resources. It does not need to rely on cloud computing power support and can still work in environments without network or with weak network, thus possessing engineering practicality and promotion value.

[0071] (5) Achieve energy-business synergy optimization. By introducing detection benefits and energy efficiency ratio, this invention has for the first time realized intelligent decision-making in the field of power monitoring terminals, which is to "use energy on the most valuable detection tasks" and maximize the comprehensive benefits of defect detection under limited energy constraints. Attached Figure Description

[0072] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0073] Figure 1 This is a flowchart of the monitoring terminal device of the present invention;

[0074] Figure 2 This is a schematic diagram of the module connection of the monitoring terminal device described in this invention. Detailed Implementation

[0075] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of this invention are only for explaining this invention and are not intended to limit this invention.

[0076] Example 1 provides a grid edge monitoring terminal control device that combines active energy sensing. For example... Figure 1 , Figure 2 As shown, the device mainly includes: a micro-meteorological sensing module, an energy prediction module, an energy assessment module, a model decision module, a defect detection module, and a verification decision module.

[0077] Next, we will introduce the composition of each module and the functions it will implement.

[0078] In some embodiments, the micro-meteorological sensing module is integrated inside or outside the monitoring terminal to collect meteorological parameters of the environment in which the monitoring terminal is located in real time. In this embodiment, the module includes a solar irradiance sensor, a temperature sensor, a humidity sensor, an atmospheric pressure sensor, and a cloud cover sensor, which are used to collect total solar irradiance. (Unit: W / m²), Ambient Temperature (Unit: °C), Relative Humidity (Unit: %) Atmospheric pressure (Unit: hPa) and cloud cover (Unit: %). The above sensors continuously collect and store historical data with a sampling period of 15 minutes, providing a data foundation for the energy prediction module.

[0079] In some embodiments, the energy prediction module is deployed on the edge computing unit (such as an embedded MCU or a low-power AI chip) of the monitoring terminal to predict the expected energy gain within a future time window based on the time series of meteorological parameters through a time-series prediction model.

[0080] In this embodiment, the time series prediction model employs a Long Short-Term Memory (LSTM) network. Specifically, it takes data collected from the past n time points (in this embodiment, the past 4 hours, with each point occurring every 15 minutes, i.e., n=16) to form the input sequence. ,in The sequence, after being min-max normalized, is input into the LSTM network. The LSTM network controls the information flow through three gate structures: a forget gate, an input gate, and an output gate, maintaining its memory of the changing trend of solar irradiance. The output of the LSTM at the last time step... After linear transformation and inverse normalization by the fully connected layer, the future is obtained. Predicted solar irradiance at time In this embodiment, the prediction time window Set to 15 minutes.

[0081] Specifically, the sequence The input is fed into an LSTM network, where information flow is controlled by three gate structures (forget gate, input gate, and output gate), thereby maintaining a memory of the trend of solar irradiance changes in long sequences for prediction. The sequence's... The specific calculation steps for each time step are as follows:

[0082] The Forget Gate calculates the cell state from the previous time step. What irrelevant information is discarded:

[0083] ;

[0084] in It is the Sigmoid activation function. This is the weight matrix. For bias terms, This is the output of the hidden layer from the previous time step.

[0085] The input gate calculates which new information needs to be updated to the current cell state:

[0086] ;

[0087] ;

[0088] Cell State updates combine the results of the forgetting gate and the input gate to update the current long-term memory state. :

[0089] ;

[0090] The output gate calculates the output value at the current time step based on the current cell state and input.

[0091] ;

[0092] ;

[0093] in, Here, is the Sigmoid activation function, and tanh is the hyperbolic tangent activation function. , , , This is the weight matrix. , , , For bias terms, This is the output of the hidden layer from the previous time step.

[0094] The output of the last time step of the LSTM network Input to a fully connected layer, and obtain the future through a linear transformation. Normalized solar irradiance prediction at time The actual predicted irradiance is obtained after inverse normalization. :

[0095] .

[0096] Subsequently, the energy prediction module, combined with the photovoltaic panel conversion model, calculates the expected power generation:

[0097] ;

[0098] in, The photoelectric conversion efficiency of a photovoltaic panel. This refers to the effective area of ​​the photovoltaic panel. Taking a monitoring terminal for a 500kV mountain transmission channel as an example, it is equipped with a monocrystalline silicon photovoltaic panel with a rated peak power of 100W. Take 0.18, Taking a radius of 0.55 m², when the predicted irradiance is 800 W / m², .

[0099] This invention, by deploying the aforementioned lightweight prediction model at the edge, enables the terminal to independently assess future energy gains without relying on the cloud, maintaining intelligent energy sensing and scheduling even in the event of communication interruptions. For indoor scenarios where solar power cannot be utilized, this prediction module can serve as a reference for subsequent energy consumption planning, combining weather forecast data to determine the impact of external temperature changes on device power consumption, indirectly optimizing energy consumption strategies.

[0100] In some embodiments, the energy assessment module is used to calculate an energy availability score based on the current remaining energy storage capacity and expected energy returns. This embodiment employs a linear weighted method, fusing the battery's current static charge (SOC) with the expected power generation based on weather forecasts to calculate the energy availability score. :

[0101] ;

[0102] in, This indicates the current remaining battery percentage, which can be read directly through the Battery Management System (BMS). This indicates the rated peak power of the solar photovoltaic panel. Hyperparameters This is a weighting coefficient used to adjust whether the current power generation or the expected power generation is emphasized. This coefficient can be flexibly adjusted according to the application scenario: for scenarios with stable lighting conditions, it can be appropriately reduced. For scenes with drastic fluctuations in lighting, the intensity can be appropriately increased. For indoor scenarios where solar power generation is not possible, the energy availability score is calculated based solely on the current remaining battery power, while also allowing for a larger safety margin to ensure that the device can complete necessary data transmission and alarm operations before the energy is depleted.

[0103] In some embodiments, the mode decision module is used to dynamically switch the operating mode of the monitoring terminal based on the comparison result of the energy availability score and a preset threshold. In this embodiment, the operating modes include a first power consumption mode (image acquisition mode) and a second power consumption mode (video acquisition mode), wherein the energy consumption per unit time of the video acquisition mode is higher than that of the image acquisition mode.

[0104] To avoid frequent switching between high and low power consumption modes due to minor fluctuations in meteorological parameters, this embodiment employs an asymmetric threshold hysteresis switching mechanism:

[0105] Set high threshold and low threshold ,and ;

[0106] when When necessary, switch to video capture mode;

[0107] when When the time comes, switch to image acquisition mode;

[0108] when At that time, the current work mode will remain unchanged.

[0109] In other words, the monitoring terminal shooting mode switching strategy designed in this invention is as follows:

[0110] Video mode: When energy availability score Greater than the set threshold When activated, the camera enters video stream acquisition mode, runs video analysis algorithms, and analyzes moving targets in the video stream in real time, effectively identifying potential hazards with time-series dynamic characteristics such as conductor galloping, hanging foreign objects, and external line damage. This mode fully utilizes overflow energy to maximize monitoring efficiency;

[0111] Image pattern: When energy availability score Less than the set threshold When activated, the camera begins capturing low-frequency static images and runs only a lightweight target detection model for routine inspections of static defects.

[0112] In this step, to prevent the occurrence of To address the issue of frequent start-stop switching between video and image modes caused by minute fluctuations near a critical point, an asymmetric threshold was designed. In this invention, the following settings are made: =0.7, .when At that time, the current work mode will remain unchanged.

[0113] The aforementioned asymmetric threshold design can effectively avoid frequent mode switching caused by factors such as short-term cloud cover and instantaneous fluctuations in illumination, thereby reducing system overhead.

[0114] In some embodiments, the defect detection module is deployed at the edge to perform real-time defect detection on the acquired images or video streams.

[0115] In image acquisition mode, the monitoring terminal captures visible light images at a preset fixed frequency (once every 5 minutes in this embodiment). The defect detection module runs a lightweight target detection algorithm (such as YOLOv5s or Mobile Net-SSD) to identify static defects such as insulator damage, hardware corrosion, and bird nests.

[0116] In video acquisition mode, the monitoring terminal starts video stream acquisition (in this embodiment, it is set to a 10-second short video), and the defect detection module runs video analysis algorithms (such as Slow Fast or TSM) to perform real-time analysis of potential hazards with time-series dynamic characteristics, such as conductor galloping, foreign object hanging, and external line damage.

[0117] In some embodiments, the review decision module is one of the core innovations of this invention. When the defect detection module detects a suspected defect that needs further confirmation in image acquisition mode, this module comprehensively evaluates the currently available remaining energy, the energy consumption required for the review operation, the detection benefit of the suspected defect, and the detection confidence level to determine whether to trigger the video acquisition mode for defect review.

[0118] Specifically, the review decision module includes the following sub-modules:

[0119] (1) Energy consumption quantification submodule

[0120] This submodule is used to calculate the energy consumption of the monitoring terminal under predetermined operating conditions. In normal operating mode (typically image mode), the monitoring terminal captures visible light images at a preset fixed frequency. A lightweight defect detection algorithm deployed on the terminal side performs real-time inference and analysis on the acquired images. Within the prediction time window... Internally, under normal operating mode, the equipment's basic energy consumption The calculation mainly includes the following parts: Specifically, it includes:

[0121] Image acquisition energy consumption The energy consumption of a single image capture by a camera, including sensor startup, image acquisition, and image encoding, is usually a fixed value and can be expressed as: ,in For camera operating power, This represents the duration of a single data collection session.

[0122] Defect detection inference energy consumption The energy consumption of a single inference iteration of a lightweight defect detection algorithm model (such as YOLO) is related to the algorithm complexity, input image resolution, and the energy efficiency of the processing chip. The energy consumption of a single inference iteration of the lightweight detection model used in this invention is expressed as follows: ,in This refers to the power consumption of the edge AI chip during inference. This refers to the time consumed in a single reasoning operation.

[0123] Standby power consumption The device is within the predicted time window. The total static power consumption during non-working periods includes communication module heartbeat maintenance, sensor data caching, BMS monitoring, etc. This part of the energy consumption is a fixed expense in the energy budget.

[0124] Gimbal motor energy consumption Energy consumption required to adjust the camera's viewing angle.

[0125] Video capture power consumption Energy consumption for data acquisition after switching to video mode.

[0126] (2) Remaining Energy Calculation Submodule

[0127] This submodule is used to calculate the dynamically available remaining energy. First, the prediction time window is calculated. Basic energy consumption within 15 minutes :

[0128] ;

[0129] Where N represents the number of regular image captures within the time window. In this embodiment, image capture is performed every 5 minutes in image capture mode, so N=3 within 15 minutes.

[0130] Then, assuming the equipment can safely continue operating and basic system inspections are performed, the dynamic available remaining energy for high-energy-consuming operations (including video capture, gimbal rotation, etc.) is calculated:

[0131] ;

[0132] in: This represents the current actual usable battery capacity. Assuming a battery capacity of 20Ah, a system nominal voltage of 12V, a total energy of 240Wh, and a current SOC of 50%, Wh.

[0133] This represents the expected total power generation within the forecast window.

[0134] This is the minimum safety energy threshold set to ensure that the device is not turned off or offline.

[0135] (3) Benefit Evaluation Submodule

[0136] This submodule is used to quantify the detection benefit value based on the risk level of suspected defects. For example, the mapping relationship between defect risk levels and benefit weights can be pre-established as shown in the table below:

[0137] Defect types Risk level Detection benefit value Insulator damage / flashover High risk 1.8 Wires dancing High risk 1.6 Foreign objects hanging Medium risk 1.2 External damage to the power line (construction) High risk 1.8 Bird's Nest Low risk 0.6 Metal fittings corrosion Low risk 0.4

[0138] (4) Decision Execution Submodule

[0139] When the detection algorithm deployed on the edge identifies a suspected defect that requires a review operation, this submodule will evaluate the additional energy consumption required for the review:

[0140] ;

[0141] in Energy consumption for end-side inference analysis of continuous video streams.

[0142] Then calculate the energy efficiency ratio:

[0143] ;

[0144] This indicator reflects the defect detection benefit obtained per unit of energy consumption and serves as the basis for the system to make intelligent judgments and trade-offs when energy is limited.

[0145] For example, for insulator damage ( ), .

[0146] When the system is in image mode, if the edge-deployed algorithm detects a suspected defect requiring further confirmation, the decision execution submodule determines whether the following three conditions are met simultaneously. If they are met, a high-energy-consuming video review is triggered:

[0147] Condition A: The dynamically available remaining energy can fully cover the requirements of the verification operation, that is... (Sufficient energy);

[0148] Condition B: The energy efficiency ratio of the review operation is higher than the set threshold, i.e. , The preset energy efficiency ratio threshold is set to 0.5 in this embodiment;

[0149] Condition C: (Suspected Defect Judgment) The detection confidence level is lower than the preset confidence level threshold (in this embodiment, it is set to 0.85, that is, suspected defects with a confidence level between 0.5 and 0.85 need to be reviewed and confirmed; if the confidence level is higher than 0.85, it is reported directly, and if it is lower than 0.5, it is judged as a false alarm).

[0150] The design principle of the above three-condition AND logic is as follows: Condition A ensures that the equipment has sufficient energy to perform the review operation without losing connection; Condition B ensures that the review operation has a sufficiently high "energy efficiency ratio," that is, the detection benefit obtained per unit of energy consumption exceeds a preset threshold; Condition C ensures that only suspected defects that the model is "unsure" need to be reviewed. If the confidence level is already high, it can be reported directly; if the confidence level is too low, it is likely a false alarm and it is not worthwhile to consume energy for review. All three conditions are indispensable and together constitute the optimal review decision mechanism under energy constraints.

[0151] When all three conditions are met, the system dynamically adjusts the pan-tilt unit to observe the suspected area at a fixed point and switches to video acquisition mode for continuous confirmation. After verification, the system automatically switches back to image acquisition mode and resumes low-power operation.

[0152] In summary, existing technologies passively adjust the operating mode based solely on the current remaining battery power. This invention proposes an energy management method that considers both the current battery power and the expected future energy. By calculating and predicting future power generation through edge-side calculations, this method combines the predicted power generation with the current battery power to determine whether to activate a high-power video sensing mode. In addition to power transmission channel monitoring, this technology can be applied to the energy management of self-powered IoT devices, including forest fire monitoring terminals and hydrological / geological disaster monitoring equipment.

[0153] In situations where edge resources are limited, this invention establishes a comprehensive energy consumption model for actions such as image acquisition, edge AI inference, and gimbal driving. When a suspected defect is detected, the energy efficiency ratio of the detection benefit to the additional energy expenditure is calculated to determine whether to perform video review. This technology is applicable to and can be extended to various resource-constrained edge embodied intelligent devices, such as inspection drones and inspection robots.

[0154] This invention comprehensively considers the current remaining power and the future predicted power generation, and proposes an energy availability scoring formula with low computing power consumption. Compared with other complex integral calculations or reinforcement learning models, it has lower computational complexity, is suitable for real-time operation on low-cost, low-power MCUs, and can sensitively reflect environmental changes.

[0155] Taking the intelligent energy dispatching process of a monitoring terminal for a 500kV mountain power transmission channel as an example, the terminal is equipped with a monocrystalline silicon photovoltaic panel with a rated peak power of 100W, and a lithium battery pack with a capacity of 20Ah, a nominal system voltage of 12V, and a total energy of 240Wh. The core processor of the terminal adopts a low-power edge AI chip, and the energy scoring weight coefficient is... =0.6, mode switching threshold =0.7, Low safety minimum energy Battery capacity (24Wh), energy efficiency ratio threshold .

[0156] When there is sufficient light, the terminal's micro-weather sensor detects good ambient light, and the BMS system reads the remaining battery power. The edge computing module calls the LSTM model, inputs the historical sequence of the past 4 hours, and predicts the average power generation in the next 15 minutes. Calculate the energy availability score. ,because When the system determines that the energy is sufficient, the camera terminal automatically switches to video mode, starts high frame rate video stream acquisition, and loads the corresponding video analysis algorithm. In this mode, dynamic hazards such as wire swaying and foreign objects hanging are identified in real time, and alarm video clips are immediately sent to the cloud.

[0157] After several hours of operation, the battery power was depleted. The LSTM predicts that the light intensity will decrease in the next 15 minutes, with an expected power of [missing value]. At this point, calculation Although it was lower at this time But still higher This triggers the anti-shake mechanism, allowing the terminal to maintain its current video mode and avoid the overhead of frequent device startup and shutdown and mode switching caused by short-term cloud cover.

[0158] When the battery level drops to The LSTM predicts that future lighting conditions will worsen (such as evening or rainy weather), and the expected power is [value missing]. ,at this time lower than The terminal immediately switches back to low-power image mode and loads a lightweight target detection algorithm.

[0159] During image mode operation, the lightweight AI model on the edge detected a suspected "damaged insulator skirt" in a static image capture, which is considered a medium-to-high risk hazard. At this point, the review decision mechanism was triggered.

[0160] The system calculates the current actual available energy. Expected power generation within the forecast window Basic energy consumption Deducting safety energy Then, the dynamically available remaining energy was calculated. The system continues to evaluate and adjust the gimbal positioning and record a 10-second video for verification of energy consumption. Because insulator damage is a high-risk hazard, the quantified value of its detection benefits is determined to be: The calculated energy efficiency ratio .because ,and While satisfying the constraints, the system schedules the PTZ to be aimed at the suspected area, briefly pulls up the video stream for accurate verification and confirmation, uploads the results to the cloud, and then returns to image mode.

[0161] Example 2 discloses an energy regulation method combining active energy sensing and grid edge monitoring terminal regulation, applied to the active energy sensing and regulation device of the grid edge intelligent monitoring terminal described in Example 1, including the following steps:

[0162] Meteorological parameters of the environment where the monitoring terminal is located are collected in real time. Based on the time series of the meteorological parameters, the expected energy gain in the future time window is predicted by a time series prediction model deployed on the edge side.

[0163] Calculate the energy availability score based on the current remaining energy storage capacity and the expected energy gain;

[0164] Based on the comparison result between the energy availability score and the preset threshold, the working mode of the monitoring terminal is dynamically switched, and the working mode includes at least a first power consumption mode and a second power consumption mode.

[0165] In the first power consumption mode, real-time defect detection is performed on the acquired images;

[0166] When a suspected defect is detected, the system comprehensively evaluates the currently available remaining energy, the energy consumption required for the review operation, the detection benefit of the suspected defect, and the detection confidence level to determine whether to trigger the second power consumption mode for defect review.

[0167] In some embodiments, the specific process for determining whether to trigger the second power consumption mode for defect review is as follows:

[0168] A refined energy loss model is established for the monitoring terminal in image acquisition, defect detection inference, standby maintenance, gimbal rotation, and video acquisition operations.

[0169] Calculate the dynamic available remaining energy ,in This represents the current available battery power. To predict the total expected power generation within the forecast window, Based on energy consumption, To ensure a safe minimum energy threshold;

[0170] Quantify the detection benefit value based on the risk level of suspected defects. ;

[0171] Obtain the detection confidence level of the defect detection module for suspected defects;

[0172] Calculate the energy efficiency ratio ,in Total energy consumption required for the verification operation;

[0173] The second power consumption mode is triggered for defect review when all three of the following conditions are met:

[0174] ;

[0175] , This is a preset energy efficiency ratio threshold;

[0176] The detection confidence level is lower than a preset confidence threshold.

[0177] In some embodiments, the specific process of dynamically switching the working mode of the monitoring terminal based on the comparison result of the energy availability score and the preset threshold is as follows:

[0178] Set high threshold and low threshold ,and ;

[0179] when When switching to the second power consumption mode, Assess energy availability.

[0180] when When the time comes, switch to the first power consumption mode;

[0181] when At that time, the current work mode will remain unchanged.

[0182] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A power grid edge monitoring terminal regulation device combined with active energy perception, characterized in that, include: The micro-meteorological sensing module is used to collect meteorological parameters of the environment where the monitoring terminal is located in real time; An energy prediction module, deployed at the edge, is used to predict expected energy gains within future time windows based on the time series of the meteorological parameters using a time-series prediction model. The energy assessment module is used to calculate an energy availability score based on the current remaining energy storage capacity and the expected energy gain. The mode decision module is used to dynamically switch the working mode of the monitoring terminal based on the comparison result of the energy availability score and the preset threshold. The working mode includes at least a first power consumption mode and a second power consumption mode, wherein the energy consumption per unit time of the second power consumption mode is higher than that of the first power consumption mode. The defect detection module, deployed at the edge, is used to perform real-time defect detection on the acquired images or video streams; The review decision module is used to comprehensively evaluate the currently available remaining energy, the energy consumption required for the review operation, the detection benefit of the suspected defect, and the detection confidence level when the defect detection module detects a suspected defect in the first power consumption mode, and determine whether to trigger the second power consumption mode for defect review. 2.The power grid edge monitoring terminal regulation device combined with active energy perception of claim 1, wherein, The meteorological parameters include at least one of solar irradiance, ambient temperature, ambient humidity, atmospheric pressure, and cloud cover. The time-series prediction model is a long short-term memory network or a gated recurrent unit; The expected energy yield is calculated based on predicted solar irradiance combined with a photovoltaic panel conversion model to calculate the expected power generation within a future time window .

3. The grid edge monitoring terminal control device combining active energy sensing according to claim 1, characterized in that, The energy assessment module calculates an energy availability score in the following manner: ； wherein, is the current percentage of remaining power, is the expected power generation, is the rated peak power, is a weight coefficient for adjusting the proportion of the current power and the expected power generation in the decision-making. 4.The power grid edge monitoring terminal regulation device combined with active energy perception of claim 1, wherein, The mode decision module employs an asymmetric threshold hysteresis switching mechanism: Set high threshold and low threshold ,and ; when When switching to the second power consumption mode, Assess energy availability. when When the time comes, switch to the first power consumption mode; when At that time, the current work mode will remain unchanged.

5. The grid edge monitoring terminal control device combining active energy sensing according to claim 1, characterized in that, The review decision module specifically includes: The energy consumption quantification submodule is used to calculate the energy consumption of the monitoring terminal under a predetermined operation, which includes at least: image acquisition energy consumption. Defect detection inference energy consumption Standby power consumption Gimbal motor energy consumption Video capture energy consumption One or more of the following; The remaining energy calculation submodule is used to calculate the dynamically available remaining energy based on the current battery capacity, expected power generation, basic energy consumption, and safety minimum energy threshold. ; The benefit assessment submodule is used to quantify the detection benefit value based on the risk level of the suspected defect. ; The decision execution submodule is used to calculate the energy efficiency ratio. ,in Total energy consumption required to perform the review operation; The decision execution submodule triggers a second power consumption mode for defect review when the following conditions are met simultaneously: Condition A: ; Condition B: , This is a preset energy efficiency ratio threshold; Condition C: The detection confidence level is lower than the preset confidence threshold.

6. The grid edge monitoring terminal control device combining active energy sensing according to claim 1, characterized in that, The first power consumption mode is an image acquisition mode, and the second power consumption mode is a video acquisition mode. In the image acquisition mode, the monitoring terminal captures visible light images at a preset fixed frequency and runs a lightweight defect detection algorithm. In the video acquisition mode, the monitoring terminal enables video stream acquisition and runs a video analysis algorithm to identify potential defects with temporal dynamic characteristics.

7. The grid edge monitoring terminal control device combining active energy sensing according to claim 1, characterized in that, The monitoring terminal is applicable to self-powered IoT devices, including power transmission channel monitoring terminals, substation monitoring terminals, power distribution room monitoring terminals, forest fire prevention monitoring terminals, hydrogeological disaster monitoring terminals, inspection drones or inspection robots.

8. A method for controlling a power grid edge monitoring terminal combining active energy sensing, applied to the power grid edge monitoring terminal control device combining active energy sensing as described in any one of claims 1 to 7, characterized in that, Includes the following steps: Meteorological parameters of the environment where the monitoring terminal is located are collected in real time. Based on the time series of the meteorological parameters, the expected energy gain in the future time window is predicted by a time series prediction model deployed on the edge side. Calculate the energy availability score based on the current remaining energy storage capacity and the expected energy gain; Based on the comparison result between the energy availability score and the preset threshold, the working mode of the monitoring terminal is dynamically switched, and the working mode includes at least a first power consumption mode and a second power consumption mode. In the first power consumption mode, real-time defect detection is performed on the acquired images; When a suspected defect is detected, the system comprehensively evaluates the currently available remaining energy, the energy consumption required for the review operation, the detection benefit of the suspected defect, and the detection confidence level to determine whether to trigger the second power consumption mode for defect review.

9. The grid edge monitoring terminal control method combining active energy sensing according to claim 8, characterized in that, The specific process for determining whether to trigger the second power consumption mode for defect review is as follows: A refined energy loss model is established for the monitoring terminal during image acquisition, defect detection inference, standby maintenance, gimbal rotation, and video acquisition operations. Calculate the dynamic available remaining energy ,in This represents the current available battery power. To predict the total expected power generation within the forecast window, Based on energy consumption, To ensure a safe minimum energy threshold; Quantify the detection benefit value based on the risk level of suspected defects. ; Obtain the detection confidence level of the defect detection module for suspected defects; Calculate the energy efficiency ratio ,in Total energy consumption required for the verification operation; The second power consumption mode is triggered for defect review when all three of the following conditions are met: ； , This is a preset energy efficiency ratio threshold; The detection confidence level is lower than a preset confidence threshold.

10. A method for regulating a power grid edge monitoring terminal combining active energy sensing according to claim 8, characterized in that, The specific process of dynamically switching the working mode of the monitoring terminal based on the comparison result of the energy availability score and the preset threshold is as follows: Set high threshold and low threshold ,and ; when When switching to the second power consumption mode, Assess energy availability. when When the time comes, switch to the first power consumption mode; when At that time, the current work mode will remain unchanged.

Images