Low-power power transformer fault monitoring system and method based on edge cloud cooperation

Abstract

The application relates to the technical field of power equipment monitoring, and discloses a low-power-consumption power transformer fault monitoring system and method based on edge cloud cooperation. The system comprises a three-phase electric measurement module, a temperature measurement module, an edge processing module and a cloud server. The edge processing module monitors voltage and current in real time, divides the system state into four levels of normal, normal early warning, fault early warning and actual fault according to preset multistage threshold values, and performs graded response to realize low-power-consumption data uploading; the cloud server responds to a fault early warning request to perform fault diagnosis, periodically calculates an aging state according to monitoring data of a normal state of the transformer, issues a threshold coefficient to adaptively adjust an edge end monitoring threshold value according to an aging degree, and realizes intelligent feedback that the more serious the aging is, the more timely the early warning is. The application realizes real-time, efficient and low-power-consumption monitoring of transformer faults by combining edge graded response with cloud intelligent feedback, and is suitable for the scene of power transformers.

Description

Technical Field

[0001] This invention relates to the field of power equipment monitoring technology, specifically to a low-power power transformer fault monitoring system and method based on edge cloud collaboration. Background Technology

[0002] Power transformers are critical equipment in power systems, and their operating status directly affects the stability and reliability of the power grid. Traditional transformer monitoring systems are mostly three-phase meters, primarily monitoring electrical parameters such as voltage, current, and power. These systems have the following limitations: Limited monitoring parameters: Traditional three-phase meters typically only collect electrical parameters, which is insufficient to fully reflect the transformer's operating status and cannot effectively detect faults such as three-phase imbalance, abnormal temperature, or equipment aging.

[0003] 2. Poor real-time performance and reliance on manual inspection: Traditional systems rely heavily on regular manual inspections, which makes it difficult to quickly detect sudden faults such as overload and short circuits. Furthermore, the inability to upload monitoring data to the cloud in real time leads to response delays and fails to meet the needs of immediate protection.

[0004] 3. Non-adaptive data upload leads to an imbalance between power consumption and efficiency: Traditional systems typically use fixed-frequency data acquisition and upload, resulting in data redundancy and wasted power consumption under normal conditions, or insufficient data to support timely analysis under abnormal conditions, making it impossible to achieve an optimized balance of low power consumption.

[0005] 4. Limited communication capabilities: Traditional three-phase meters typically lack or have only limited communication functions, lack cloud data processing support, and are difficult to achieve remote real-time monitoring and complex fault analysis.

[0006] 5. Low level of intelligence: Traditional systems lack edge cloud collaboration technology support, making it difficult to classify and respond to faults or dynamically adjust thresholds, resulting in low monitoring efficiency and an inability to achieve efficient fault prevention and aging trend prediction. Summary of the Invention

[0007] The present invention aims to provide a low-power power transformer fault monitoring system and method based on edge cloud collaboration, so as to solve the problems of current monitoring systems such as single parameters, poor real-time performance, imbalance between power consumption and efficiency, and low level of intelligence.

[0008] To achieve the above objectives, this invention provides a low-power power transformer fault monitoring system based on edge-cloud collaboration, comprising: a three-phase electrical measurement module, a temperature measurement module, an edge processing module, a communication module, a cloud server, a power management module, and a battery and relay module; the three-phase electrical measurement module is used to collect the three-phase voltage values, three-phase current values, three-phase voltage waveform data, and three-phase current waveform data of the transformer, and send the data to the edge processing module; the temperature measurement module is used to collect the temperature values ​​of preset temperature measurement points of the transformer, and send the data to the edge processing module; the edge processing module is used to monitor the three-phase voltage values, three-phase current values, and temperature values ​​of multiple temperature measurement points in real time, and classify the system status into normal state, normal warning state, fault warning state, and actual fault state according to preset multi-level thresholds; the multi-level thresholds include current thresholds and voltage thresholds, and each threshold is jointly determined by a reference threshold and a threshold adjustment coefficient k dynamically issued by the cloud; During normal warning, the current threshold is 110% × k of the rated current, and the voltage threshold is [90% + Δ × (1-k)] of the rated voltage; During fault warning, the current threshold is 120%×k of the rated current, and the voltage threshold is [70% + Δ×(1-k)] of the rated voltage; during actual faults, the current threshold is 130%×k of the rated current, and the voltage threshold is [60% + Δ×(1-k)] of the rated voltage; where Δ is the threshold offset coefficient. The edge processing module performs the following graded responses based on the current system state: Under normal conditions, the communication module maintains the lowest power consumption state and uploads data at the lowest upload frequency; under normal early warning conditions, the upload frequency is increased to the medium frequency level; under fault early warning conditions, the communication module is immediately woken up to upload data and requests the cloud server to immediately call the deep analysis module, while simultaneously increasing the upload frequency to the high frequency level; under actual fault conditions, the relay module is controlled to cut off the three-phase power supply of the transformer and switch the power management module to battery power. The cloud server and edge processing module form a two-way feedback loop: the cloud server responds to requests in the fault warning state, calls the deep analysis module in real time to perform fault diagnosis, and calculates the average value of the aging acceleration factor based on the preset temperature values ​​of the transformer's temperature measurement points. Then, based on the received monitoring data of the transformer's normal state, the transformer aging state is periodically calculated, and the threshold adjustment coefficient k is dynamically calculated according to the formula. k is then sent to the edge processing module to update the multi-level thresholds. The formula for calculating k is: in The average value of the aging acceleration factor; the cloud server is also used to monitor the gradual aging faults of the transformer based on the received data.

[0009] Based on the guiding principle of the IEEE Std C57.91 standard regarding the adoption of a conservative operating strategy during accelerated aging, this invention designs the linear adjustment coefficient. When k < 1 (accelerated aging), the multi-level thresholds are automatically tightened to achieve earlier warning; the cloud server is also used to monitor the gradual aging faults of the transformer based on the received data.

[0010] In a preferred embodiment of the present invention, Δ is set to 10%, i.e., 0.1. In other embodiments, Δ can be flexibly set to any value between 8% and 15% according to the actual degree of aging, specific application scenario or different transformer models, so as to further optimize the early warning sensitivity of different aging stages.

[0011] Specifically, the edge processing module executes the following tiered responses based on the current system state: 1) Normal state: When the voltage of each phase is higher than [90% + Δ×(1-k)] of the rated voltage and the current of each phase is lower than 110%×k of the rated current, the communication module maintains the lowest power consumption state and the edge processing module uploads data at the lowest upload frequency. 2) Normal warning state: When the voltage of any phase is lower than or equal to [90% + Δ×(1-k)] of the rated voltage, or the current of any phase is higher than or equal to 110%×k of the rated current, and the fault warning or actual fault threshold is not reached, the edge processing module will increase the upload frequency to the medium frequency level. 3) Fault warning status: When the voltage of any phase is lower than or equal to [70% + Δ×(1-k)] of the rated voltage, or the current of any phase is higher than or equal to 120%×k of the rated current, and the actual fault threshold is not reached, the edge processing module immediately wakes up the communication module to upload data and requests the cloud server to immediately call the deep analysis module, while increasing the upload frequency to a high frequency level. 4) Actual fault conditions: When the voltage of any phase is lower than or equal to [60% + Δ×(1-k)] of the rated voltage, or the current of any phase is higher than or equal to 130%×k of the rated current, the edge processing module controls the relay module to cut off the three-phase power supply of the transformer and switch the power management module to battery power supply.

[0012] Preferably, the edge processing module sends data to the cloud server via the MQTT protocol through the communication module.

[0013] Preferably, the relay module remains in the off state under actual fault conditions until external reset.

[0014] Preferably, the minimum upload frequency is 5 minutes, the medium frequency upload frequency is 2 minutes, and the high frequency upload frequency is 10 seconds.

[0015] Preferably, the deep analysis module is used to: 1) perform a fast Fourier transform on the transient fault waveform uploaded by the edge processing module at the moment the transformer enters the fault warning state, and calculate the content of the 2nd to 31st harmonics and the total harmonic distortion; 2) calculate the peak factor, asymmetry, total energy and other characteristics of the transient fault waveform to help identify and distinguish between external power grid transient interference and internal electrical faults of the transformer.

[0016] Preferably, the edge processing module uploads a single data packet containing voltage, current, temperature, waveform data, and trigger parameters in each state; the trigger parameter is a fault warning trigger flag, with a value of 0 or 1. When the transformer is in a fault warning state, it is set to 1, otherwise it is set to 0; when the trigger parameter is 1, the cloud server immediately calls the deep analysis module.

[0017] Preferably, the cloud server performs an aging status analysis on the accumulated data every 10 minutes under normal transformer conditions, and dynamically generates a threshold adjustment coefficient k based on the calculated value. This threshold coefficient is then distributed to adaptively adjust the edge monitoring threshold according to the degree of aging, achieving intelligent feedback where the more severe the aging, the more timely the warning.

[0018] To achieve the above objectives, the present invention also provides a low-power power transformer fault monitoring method based on edge cloud collaboration, comprising the following steps: Collect the three-phase voltage value, three-phase current value, three-phase voltage waveform data, three-phase current waveform data and preset temperature measurement point temperature value of the transformer; The edge processing module monitors the three-phase voltage and three-phase current values ​​in real time, and classifies the system status into normal state, normal warning state, fault warning state, and actual fault state based on the comparison results of the monitoring data and preset multi-level thresholds. The multi-level thresholds include current thresholds and voltage thresholds, and each threshold is determined by the reference threshold and the threshold adjustment coefficient k dynamically issued by the cloud. Under normal conditions, the communication module maintains a minimum power consumption state, and the edge processing module uploads data at the lowest upload frequency. Under normal early warning conditions, the edge processing module increases the upload frequency to the medium frequency level. Under fault early warning conditions, the edge processing module wakes up the communication module to upload data and requests the cloud server to immediately call the deep analysis module, while increasing the upload frequency to the high frequency level. Under actual fault conditions, the edge processing module controls the relay module to cut off the three-phase power supply of the transformer and switches the power management module to battery power. The cloud server responds to requests in the fault warning state by performing real-time FFT harmonic analysis and waveform time-domain feature identification to distinguish the fault type of the transformer in the fault warning state. The cloud server periodically calculates the transformer aging status based on the received monitoring data of the transformer's normal state, dynamically calculates the threshold adjustment coefficient k according to the following formula, and sends k to the edge processing module to update the multi-level thresholds. The calculation formula for K is: in, This invention designs a linear adjustment coefficient based on the IEEE Std C57.91 standard's guidance on adopting a conservative operating strategy during accelerated aging, where the aging acceleration factor is the average value. When k > 1 (accelerated aging), the thresholds are automatically tightened, enabling earlier warnings.

[0019] Preferably, when the cloud server calculates the aging acceleration factor (FAA), it uses the calculation formula in the IEEE Std C57.91 standard based on the top oil temperature and ambient temperature.

[0020] Preferably, in the normal warning state, the current threshold is 110% × k of the rated current, and the voltage threshold is [90% + Δ × (1-k)] of the rated voltage; in the fault warning state, the current threshold is 120% × k of the rated current, and the voltage threshold is [70% + Δ × (1-k)] of the rated voltage; in the actual fault state, the current threshold is 130% × k of the rated current, and the voltage threshold is [60% + Δ × (1-k)] of the rated voltage; where Δ is the threshold offset coefficient. The normal warning state is a Level 1 warning, the fault warning state is a Level 2 warning, and the actual fault state is a Level 3 warning.

[0021] Advantages of this solution: 1. Tiered response and low power consumption optimization: The edge processing module achieves tiered response through four levels of status (normal, normal warning, fault warning, and actual fault) and multi-level dynamic thresholds. In the normal state, it uploads at the lowest frequency. In the warning and fault states, it automatically increases the frequency and switches the trigger parameters, which significantly reduces the power consumption of the communication module, while ensuring real-time response to rapid faults, achieving low power operation and greatly reducing communication traffic and reliance on manual inspection.

[0022] 2. Dynamic threshold adaptation: The cloud-based system adapts to aging acceleration factors. The threshold adjustment coefficient k is dynamically calculated to realize an intelligent feedback mechanism that makes the threshold more stringent as aging becomes more severe. This conforms to the guiding principle of IEEE Std C57.91 that a conservative operation strategy should be adopted when aging is accelerated, which greatly improves the accuracy and preventiveness of monitoring.

[0023] 3. Simple and efficient cloud-based deep analysis: The cloud only needs to perform FFT calculations on the waveform data to determine harmonic content and THD, supplemented by time-domain characteristics such as peak factor, asymmetry, and total energy, to effectively distinguish between external power grid transient interference and internal transformer electrical faults. This requires minimal computation and is easy to deploy. Edge processing modules and cloud servers perform step-by-step monitoring for different types of faults. The cloud server and edge server form a complementary mechanism for rapid response, improving the comprehensiveness and efficiency of monitoring and enabling real-time, efficient, and low-power monitoring of transformer faults.

[0024] 4. Fault protection and continuous monitoring: In actual fault conditions, the relay immediately cuts off the three-phase power supply and remains locked, while switching to battery power supply to ensure that the system can continue to monitor and wait for external reset in the event of a fault, which greatly improves safety and reliability. Attached Figure Description

[0025] Figure 1 This is a structural diagram of the transformer fault monitoring system of the present invention.

[0026] Figure 2 This is a flowchart of the first type of fault detection and data upload process of the present invention.

[0027] Figure 3 This is a flowchart of the cloud server data stream processing of the present invention.

[0028] Figure 4 (a) is a graph showing the average current measurement results of the present invention at a 5-minute upload frequency.

[0029] Figure 4 (b) is a graph showing the average current measurement results of the present invention at a 2-minute upload frequency.

[0030] Figure 4 (c) is a graph showing the average current measurement results of the present invention at a 10s upload frequency. Detailed Implementation

[0031] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0032] Figure 1 A schematic diagram of an embodiment of the low-power power transformer fault monitoring system based on edge cloud collaboration of the present invention is shown. Figure 1 As shown, the system includes: a three-phase electrical measurement module, a temperature measurement module, an edge processing module, a communication module, a cloud server, a power management module, and a battery and relay module.

[0033] The three-phase electrical measurement module is configured to acquire three-phase voltage values, three-phase current values, three-phase voltage waveform data, and three-phase current waveform data from the transformer, and send the data to the edge processing module. This module includes three voltage transformers, three current transformers, and an RN8302 three-phase meter chip. The RN8302 chip receives analog signals of the three-phase voltage and current, converts them into digital signals via an analog-to-digital converter, and supports waveform acquisition.

[0034] The temperature measurement module is configured to acquire temperature values ​​from multiple measurement points on the transformer and send the data to the edge processing module. This module contains two PT100 temperature sensors, placed in the ambient environment and the top oil of the transformer, respectively, to measure the ambient temperature T2 and the top oil temperature T1. The PT100 temperature sensors are connected to the MAX31865 temperature acquisition and conversion module to convert the analog temperature signals into digital signals.

[0035] The edge processing module is specifically a microcontroller module, which can use an STM32 series chip. It is configured to monitor the three-phase voltage and three-phase current values ​​in real time, and classify the system status into normal state, normal warning state, fault warning state, and actual fault state according to preset multi-level thresholds. The multi-level thresholds are jointly determined by the base threshold and the threshold adjustment coefficient k dynamically issued by the cloud (see Table 1 for details).

[0036] The edge processing module uploads a single data packet containing voltage, current, temperature, waveform data, and trigger parameters (0 or 1) in each state; when the trigger parameter is 1, the cloud server immediately calls the deep analysis module.

[0037] Table 1. Multi-level threshold settings (with k as the adjustment coefficient) For ease of understanding, in the preferred embodiment of the present invention, Δ is set to 10%. The following uses a rated voltage of 220V as an example to give the actual values ​​and effects of the voltage threshold under different k values ​​(see Table 2).

[0038] Table 2 Examples of voltage thresholds under different aging levels (rated voltage 220V, Δ=10%) The communication module is configured to use the MQTT protocol and 4G technology to upload single data packets to the cloud server, ensuring reliable data transmission. In this embodiment, the communication module is configured to use 4G technology and a UART interface to upload data processed by the microcontroller module to the cloud server.

[0039] The cloud server and edge processing module form a two-way feedback loop: responding to requests in fault warning states, it calls the deep analysis module in real time for fault diagnosis; the deep analysis module is used to: 1) perform a fast Fourier transform on the transient fault waveform to calculate the content of the 2nd to 31st harmonics and the total harmonic distortion; 2) calculate the peak factor, asymmetry, total energy, and other characteristics of the transient fault waveform to help identify and distinguish between external power grid transient interference and internal transformer electrical faults. The cloud server also periodically calculates the transformer aging status based on the received monitoring data of the transformer's normal state (including temperature data), and dynamically calculates the threshold adjustment coefficient k according to the following formula: in The average value of the aging acceleration factor is used, and k is sent to the edge processing module to update the multi-level thresholds; under normal conditions, aging state analysis is performed on the accumulated data every 10 minutes.

[0040] The cloud server is also used to monitor the transformer's progressive aging faults based on the received data. Based on formulas in the IEEE C57.91 standard, the cloud server uses uploaded voltage and current waveform data to extract and calculate the total harmonic distortion (THD) and the amount of each harmonic of the waveform. Combined with the top oil temperature T1 and ambient temperature T2, it calculates the transformer's aging acceleration factor (FAA) and estimates its remaining lifespan.

[0041] First, go to "Calculate THD and Harmonics", then calculate the harmonic load factor (K-Factor) to integrate the current THD effects: (1) Where h is the harmonic order, I h,phase For the h-th order current amplitude, I 1,phase This is the fundamental current.

[0042] Then calculate the equivalent current: (2) And equivalent loading factor: (3) Among them I R This is the rated current.

[0043] Then, calculate the hot spot temperature. : (4) in: For ambient temperature, This refers to the top oil temperature; (Hot spot to top oil temperature difference) adopts: (5) parameter: The value is usually between 15-23°C, depending on the transformer design; y is the power index, which is 1.6-2.0 depending on the transformer cooling type. For example, the default value for a natural oil circulation, natural air cooling ONAN transformer is 1.6.

[0044] Then, the FAA is calculated. In this embodiment, the 6-degree rule of IEEE C57.91-2011 is used to ensure consistency with the standard empirical model. (6) Parameter: Reference hot spot temperature = 110℃.

[0045] Finally, update the equivalent aging time t_eq and calculate the remaining lifetime RL: (7) (8) The power management module uses AC220V power and outputs 5V DC power through voltage conversion, and supports switching to battery power in actual fault conditions. The relay module cuts off the three-phase power supply to the transformer in actual fault conditions and remains cut off until external reset.

[0046] Figure 2 and Figure 3 The flowchart of the low-power power transformer fault monitoring method based on edge cloud collaboration of the present invention is shown. The method is executed by the above system.

[0047] like Figure 2 As shown, the four-level state classification response process of the edge processing module is as follows: The system collects three-phase voltage, current and temperature data in real time to determine whether the normal state threshold is met; if it is met, a data packet containing trigger parameter 0 is uploaded at the lowest frequency of 5 minutes; if it enters the normal warning state, it is upgraded to the 2-minute medium frequency level; if it enters the fault warning state for the first time (rising edge), the communication module is immediately woken up, a complete data packet containing trigger parameter 1 is uploaded and a deep analysis request is made to the cloud, and the frequency is upgraded to the 10-second high frequency level; after triggering, it enters the suppression period. During the suppression period, even if the threshold is still exceeded, only a normal data packet containing trigger parameter 1 is uploaded, and the complete waveform data is not uploaded again; at this time, the cloud server only records the continuous alarm state after receiving trigger parameter 1, and does not call the deep analysis module again; if it is still in the fault warning state after the suppression period ends, the complete waveform data is uploaded again; if it enters the actual fault state, the three-phase power supply is immediately cut off and the battery power supply is switched, and the relay remains locked until external reset.

[0048] The specific process of data stream processing on the cloud server is as follows: Figure 3As shown, the cloud server is composed of Alibaba Cloud IoT Platform and Function Compute (FC 3.0) working together to achieve full lifecycle management and intelligent processing of data. The first stage is the data reception and intelligent routing layer. The IoT Platform is responsible for receiving MQTT data packets from the edge processing module, automatically parsing the JSON format messages according to the preset object model (TSL), and updating them in real time to the Device Shadow and Time Series Database (TSDB) for solidified storage. At the same time, the rule engine module performs real-time logical judgment on the reported data. When the trigger parameter is 1, it is determined to be an emergency trigger branch, and the data is directly pushed to the Function Compute FC module for immediate processing through the cloud product flow function in a synchronous call mode. When the trigger parameter is 0, it is determined to be a silent storage branch, and the data is only kept in the database for periodic calls. The second stage is the dual-drive computing and closed-loop control layer. The Function Compute (FC) module adopts two startup mechanisms: First, a timed trigger wakes up the FC script every 10 minutes, actively pulls accumulated historical data using Alibaba Cloud OpenAPI / SDK, executes the transformer aging assessment algorithm to calculate the health threshold adjustment coefficient k, and encapsulates the k value into an MQTT message through the SetDeviceProperty interface and sends it to the edge processing module to achieve dynamic threshold closed-loop feedback. Second, an event trigger directly receives emergency lightweight data packets. When the rule engine detects that the trigger parameter is 1, it is activated instantly. At the same time, it associates the corresponding waveform data packets from the device shadow, runs a deep analysis script to complete FFT harmonic analysis, peak factor, asymmetry, total energy, and other calculations. After diagnosis, the results are visualized and synchronously recorded in the fault log database for subsequent fault investigation.

[0049] To verify system performance, this invention also conducted a power consumption experiment, measuring the average current of the system at different upload frequencies using a FLUKE289. The results are as follows: Figure 4 As shown, measurements were taken for 15 minutes at each frequency. The experimental results are as follows: average current approximately 68.7 mA in normal state (5 minutes upload); average current approximately 73.5 mA in normal warning state (2 minutes upload); and average current approximately 96.1 mA in fault warning state (10 seconds upload). Calculations show that under typical operating conditions (normal / warning / fault state time ratio of 90:5:5), the weighted average current of this system is 70.31 mA. Compared to the baseline system with a fixed high-frequency upload (96.1 mA), this effectively reduces power consumption by approximately 26.8%. During long-term stable operation (normal state percentage > 99%), the energy-saving benefits will be further enhanced. The experimental results demonstrate that the graded response and trigger parameter mechanism significantly optimizes power consumption and is suitable for remote or low-power scenarios.

[0050] This solution acquires various data from the transformer through a three-phase electrical measurement module and a temperature measurement module, providing a comprehensive reflection of the transformer's operating status. An edge processing module enables four-level state-based response and trigger parameter control, while a cloud server provides in-depth analysis and dynamic threshold feedback. Lightweight, rapid fault monitoring is performed at the edge, while complex aging fault diagnosis is conducted in the cloud. This efficient complementary mechanism between the cloud and edge enhances the comprehensiveness, real-time performance, and low power consumption of the monitoring. The system has a simple structure, is easy to integrate, and is suitable for various power transformer fault monitoring scenarios, meeting the actual needs of power grids of different sizes.

[0051] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. Similarly, for the sake of brevity and to aid in understanding one or more aspects of the invention, in the description of exemplary embodiments of the invention above, various features of the embodiments are sometimes grouped together in a single embodiment, figure, or description thereof. The claims, which follow the detailed description, are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of the invention.

[0052] Those skilled in the art will understand that the modules in the device of the embodiment can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiment can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components, except that at least some of such features and / or processes or units are mutually exclusive.

[0053] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

Claims

1. A low-power power transformer fault monitoring system based on edge cloud collaboration, characterized in that, include: Three-phase electrical measurement module, temperature measurement module, edge processing module, communication module, cloud server, power management module, battery and relay module; The three-phase electrical measurement module is used to collect the three-phase voltage value, three-phase current value, three-phase voltage waveform data and three-phase current waveform data of the transformer, and send the data to the edge processing module; The temperature measurement module is used to collect the temperature values ​​of the preset temperature measurement points of the transformer and send the data to the edge processing module; The edge processing module is used to monitor the three-phase voltage and three-phase current values ​​in real time, and divide the system status into normal state, normal warning state, fault warning state and actual fault state according to the preset multi-level thresholds. The multi-level thresholds include current thresholds and voltage thresholds. Each threshold is determined by a base threshold and a threshold adjustment coefficient k dynamically issued from the cloud, wherein: Under normal warning conditions, the current threshold is 110% × k of the rated current, and the voltage threshold is [90% + Δ × (1-k)] of the rated voltage; In fault warning mode, the current threshold is 120%×k of the rated current, and the voltage threshold is [70% + Δ×(1-k)] of the rated voltage; In actual fault conditions, the current threshold is 130% × k of the rated current. The voltage threshold is [60% + Δ×(1-k)] of the rated voltage; where Δ is the threshold offset coefficient. The edge processing module performs the following graded responses based on the current system state: Under normal conditions, the communication module maintains the lowest power consumption state and uploads data at the lowest upload frequency; under normal early warning conditions, the upload frequency is increased to the medium frequency level; under fault early warning conditions, the communication module is immediately woken up to upload data and requests the cloud server to immediately call the deep analysis module, while simultaneously increasing the upload frequency to the high frequency level; under actual fault conditions, the relay module is controlled to cut off the three-phase power supply of the transformer and switch the power management module to battery power. The cloud server and edge processing module form a two-way feedback loop: the cloud server responds to requests in the fault warning state, calls the deep analysis module in real time to perform fault diagnosis, and calculates the average value of the aging acceleration factor based on the preset temperature values ​​of the transformer's temperature measurement points. Then, based on the received monitoring data of the transformer's normal state, the transformer aging state is periodically calculated, and the threshold adjustment coefficient k is dynamically calculated according to the formula. k is then sent to the edge processing module to update the multi-level thresholds. The formula for calculating k is: ； The cloud server is also used to monitor the gradual aging faults of the transformer based on the received data.

2. The low-power power transformer fault monitoring system based on edge cloud collaboration according to claim 1, characterized in that: The edge processing module sends data to the cloud server via the MQTT protocol through the communication module.

3. The low-power power transformer fault monitoring system based on edge cloud collaboration according to claim 1, characterized in that: The relay module remains in the off state under actual fault conditions until it is externally reset.

4. The low-power power transformer fault monitoring system based on edge cloud collaboration according to claim 1, characterized in that: The minimum upload frequency is 5 minutes, the medium frequency upload frequency is 2 minutes, and the high frequency upload frequency is 10 seconds.

5. The low-power power transformer fault monitoring system based on edge cloud collaboration according to claim 1, characterized in that: The depth analysis module is used for: 1) Perform a fast Fourier transform on the transient fault waveform uploaded by the edge processing module at the moment the transformer enters the fault warning state, and calculate the content of the 2nd to 31st harmonics and the total harmonic distortion. 2) Calculate the peak factor, asymmetry and total energy characteristics of the transient fault waveform to help identify and distinguish between external power grid transient interference and internal transformer electrical faults.

6. The low-power power transformer fault monitoring system based on edge cloud collaboration according to claim 1, characterized in that: The edge processing module uploads a single data packet containing voltage, current, temperature, waveform data, and trigger parameters in each state. The trigger parameter is a fault warning trigger flag with a value of 0 or 1. It is set to 1 when the transformer is in a fault warning state, and otherwise set to 0. When the trigger parameter is 1, the cloud server immediately calls the deep analysis module.

7. A low-power power transformer fault monitoring method based on edge cloud collaboration, characterized in that, Includes the following steps: Collect the three-phase voltage value, three-phase current value, three-phase voltage waveform data, three-phase current waveform data and preset temperature measurement point temperature value of the transformer; The edge processing module monitors the three-phase voltage and three-phase current values ​​in real time; and based on the comparison results of the monitoring data with preset multi-level thresholds, it divides the system status into normal status, normal warning status, fault warning status and actual fault status. The multi-level thresholds include current thresholds and voltage thresholds, and each threshold is determined by a base threshold and a threshold adjustment coefficient k dynamically issued by the cloud. When in normal operation, the communication module maintains a minimum power consumption state, and the edge processing module uploads data at the lowest upload frequency. When in a normal early warning state, the edge processing module increases the upload frequency to the medium frequency level; when in a fault early warning state, the edge processing module wakes up the communication module to upload data and requests the cloud server to immediately call the deep analysis module, while increasing the upload frequency to the high frequency level; when in an actual fault state, the edge processing module controls the relay module to cut off the three-phase power supply of the transformer and switches the power management module to battery power. The cloud server responds to requests in the fault warning state by performing real-time FFT harmonic analysis and waveform time-domain feature identification to distinguish the fault type of the transformer in the fault warning state. The cloud server periodically calculates the transformer's aging status based on the received monitoring data of the transformer's normal condition, and dynamically calculates the threshold adjustment coefficient k according to the following formula: in, The average value of the aging acceleration factor is k, and k is sent to the edge processing module to update the multi-level threshold.

8. The low-power power transformer fault monitoring method based on edge cloud collaboration according to claim 7, characterized in that: When calculating the aging acceleration factor (FAA) on the cloud server, the calculation formula in the IEEE StdC57.91 standard is used based on the top oil temperature and ambient temperature.

9. The low-power power transformer fault monitoring method based on edge cloud collaboration according to claim 7, characterized in that: Under normal warning conditions, the current threshold is 110% × k of the rated current, and the voltage threshold is [90% + Δ × (1-k)] of the rated voltage; In fault warning mode, the current threshold is 120%×k of the rated current, and the voltage threshold is [70% + Δ×(1-k)] of the rated voltage; In actual fault conditions, the current threshold is 130% × k of the rated current, and the voltage threshold is [60% + Δ × (1-k)] of the rated voltage; where Δ is the threshold offset coefficient.

10. The low-power power transformer fault monitoring system based on edge cloud collaboration according to claim 1, characterized in that: The cloud server performs an aging status analysis on the accumulated data every 10 minutes under normal transformer conditions, and calculates the average value of multiple aging acceleration factors (FAA). Dynamically generate the threshold adjustment coefficient k.

Images