Low-computing-power edge computing data collection terminal for transmission equipment and monitoring method thereof

Abstract

The application discloses a low-computing-power edge computing data acquisition terminal for a transmission device and a monitoring method thereof, adopts modular design, and comprises a sensor assembly, a low-computing-power edge computing core, a wireless communication module, a power management module and a protective shell. The application completes core processing and feature extraction of vibration and temperature signals on the spot of the transmission device. Compared with a scheme in which original data needs to be uploaded to a cloud end for analysis, the closed loop delay of the method is reduced from a second level to a millisecond level. The real-time edge intelligence can effectively capture early fault features such as impact and peak that occur instantaneously during the operation of the transmission device (such as a gear box), thereby issuing a pre-warning at the fault germination stage, striving for a key time window for predictive maintenance and operation scheduling decision of the smart grid, avoiding expansion of the fault or unplanned shutdown due to response lag, and significantly improving the real-time performance and early warning capability of state perception.

Description

Technical Field

[0001] This invention relates to the field of smart grid condition monitoring and fault diagnosis technology, specifically to a low-computing-power edge computing data acquisition terminal for transmission equipment and its monitoring method. Background Technology

[0002] With the advancement of new power system construction, the integration of a high proportion of renewable energy into the power grid has made system operation increasingly complex, placing higher demands on the reliability and real-time status monitoring of key equipment. As the core mechanical component of energy conversion, the health status of transmission equipment directly affects the stable operation of the entire power production unit. Statistics show that in the wind power industry, downtime costs caused by gearbox failures alone can account for 35% of total operation and maintenance expenditures. Therefore, real-time and accurate status monitoring and early fault warning of transmission equipment are crucial to ensuring the safe, stable, and economical operation of the smart grid.

[0003] Currently, technical solutions for power equipment condition monitoring are mainly developing along two paths:

[0004] The first category is monitoring solutions based on centralized cloud computing. These solutions typically deploy sensors on-site to collect data, and then upload the raw data or data after simple preprocessing to a remote cloud platform or data center via communication networks (such as 4G / 5G, fiber optics) for processing and analysis. For example, publicly available technical documents mention practices in scenarios such as steel rolling production lines and power transmission lines, where massive amounts of raw data, such as vibration, temperature, or traveling wave signals, are directly sent to a cloud platform.

[0005] Although this architecture can utilize the powerful centralized computing power of the cloud to run complex diagnostic models, its inherent defects are becoming increasingly prominent under the real-time requirements of smart grids: (1) Poor real-time performance and delayed response: Data needs to be transmitted through a long path of "terminal-communication network-cloud center", and the network delay is usually in the millisecond to second range. For sudden and transient faults in transmission equipment (such as gear tooth breakage, impact signals generated in the early stage of bearing breakage), this delay may cause the local fault to evolve into equipment shutdown or even system-level accident by the time the cloud completes the analysis, which cannot meet the real-time diagnostic requirements of smart grids for "millisecond-level response". (2) Huge pressure on communication and computing resources: The vibration signals generated by transmission equipment (such as gearboxes) have high frequency and large data volume. Full upload will form a "data flood", which will greatly occupy the communication bandwidth and bring a heavy storage and computing burden to the cloud server. Some studies have pointed out that more than 70% of the data in power business needs to be processed on the edge side, and the full upload mode has caused serious waste of resources. (3) High risk of data security and reliability: Sensitive equipment operation data is transmitted over long distances in public networks, which poses a risk of interception and leakage. Meanwhile, the continuity of monitoring is highly dependent on the stability of the network; once the network fluctuates or is interrupted, the entire monitoring system will fail.

[0006] The second category is edge intelligence-based monitoring solutions. To overcome the drawbacks of cloud-based solutions, edge computing technology has been introduced into the power Internet of Things (IoT). By deploying edge intelligent terminals or gateways with computing capabilities close to the data source, data processing, feature extraction, and even preliminary diagnosis can be completed locally, with only key features or diagnostic results uploaded, thereby greatly alleviating network pressure and improving response speed. Existing technologies already include edge intelligent devices for transmission line fault monitoring, employing a high-performance architecture of "ARM main controller + FPGA coprocessor," capable of microsecond-level traveling wave signal processing. However, these high-performance edge computing solutions for smart grids still have the following prominent problems when applied to the monitoring of a large number of widely used transmission equipment:

[0007] Redundancy in computing power and excessive cost: Existing solutions mostly use high-performance processors (such as the ARM Cortex-A series) or FPGAs, designed to handle complex tasks such as image recognition and high-frequency traveling wave analysis, with computing power reaching several TOPS (trillion operations per second). However, for routine condition monitoring and trend early warning of vibration and temperature in transmission equipment, there is significant redundancy in computing power, resulting in high cost and power consumption of individual terminal hardware, making it difficult to support low-cost, large-scale deployment on thousands of transmission devices.

[0008] Lack of specialization and complex models: Current research on power edge computing largely focuses on specific monitoring of power transmission and transformation equipment (such as conductors and insulators) or general platform architectures, lacking specialized and lightweight edge diagnostic models for the coupled analysis characteristics of vibration and temperature signals in transmission equipment. Directly deploying large-scale AI models designed for the cloud to the edge, even after compression, still places significant demands on terminal computing power and power consumption.

[0009] Insufficient environmental adaptability and integration: The working environment of transmission equipment (especially outdoor wind turbine gearboxes) is harsh (high and low temperatures, high humidity, strong electromagnetic interference). Some existing edge computing devices have not fully considered the high reliability integration with the transmission equipment itself, long-term stable operation under extreme conditions, and the matching of the life cycle with the primary equipment.

[0010] In summary, current cloud-based centralized computing models that rely on full data uploads cannot meet the real-time status monitoring needs of smart grids for transmission equipment. Existing high-performance edge computing solutions, due to their high cost, redundant computing power, and poor specialization, are difficult to implement cost-effectively on a large scale in the field of transmission equipment monitoring. Therefore, there is an urgent need for edge data acquisition terminals with precise computing power configurations, low cost, and high specialization, capable of deeply integrating vibration and temperature information to perform reliable real-time status assessment and early warning at the transmission equipment site, thereby filling the gap in the current smart grid equipment status monitoring system in this specific area.

[0011] To address this, we propose a new low-computing-power edge computing data acquisition terminal and its monitoring method for transmission equipment. Summary of the Invention

[0012] The purpose of this invention is to provide a low-computing-power edge computing data acquisition terminal and its monitoring method for transmission equipment. The innovative system architecture and lightweight algorithm design effectively solve the defects of existing solutions mentioned in the background art in terms of real-time performance, resource consumption, cost and applicability.

[0013] To achieve the above objectives, the present invention provides the following technical solution: a low-computing-power edge computing data acquisition terminal for transmission equipment, comprising a sensor assembly, a low-computing-power edge computing core, a wireless communication module, a power management module, and a protective housing;

[0014] The sensor components, low-power edge computing core, wireless communication module, and power management module are all integrated on the internal circuit board and encapsulated within the protective housing.

[0015] The sensor assembly is used to physically contact the monitored transmission equipment to acquire raw terminal data of the operating status in real time.

[0016] The low-power edge computing core is electrically connected to the sensor components and the wireless communication module, respectively, and is used to read sensor data, run lightweight signal processing and diagnostic algorithms, and control the communication module to transmit data according to the diagnostic results.

[0017] The power management module is used to provide a stable operating voltage for each component of the terminal and to manage the power supply mode of the terminal.

[0018] The protective housing is used to provide mechanical fixation and environmental protection.

[0019] Preferably, the sensor assembly includes a vibration sensor and a temperature sensor;

[0020] The vibration sensor is an integrated triaxial MEMS accelerometer. The vibration sensor is fixedly installed on the gearbox bearing seat or housing surface of the transmission equipment by high-strength epoxy resin adhesive or magnetic base, and is used to acquire vibration acceleration signals in the X, Y and Z directions.

[0021] The temperature sensor is a digital temperature sensor, which is tightly attached and fixed to the outer ring of the bearing or the surface of the lubrication oil passage of the transmission equipment by thermally conductive silicone or metal clamps, and is used to directly convert the physical quantity of temperature into a digital signal.

[0022] Preferably, the low-computing-power edge computing core includes a core processor and a storage unit;

[0023] The core processor is a 32-bit microcontroller (MCU) based on an ARM Cortex-M4 core or an equivalent RISC-V core. The MCU receives sensor data through a digital interface and runs lightweight algorithms and manages power and sleep modes internally.

[0024] The storage unit includes a Flash memory integrated inside the MCU and an external extended memory, wherein the external extended memory is an SRAM or a ferroelectric memory (FRAM); the external extended memory is connected to the MCU via an I2C bus and is used to temporarily cache high-frequency acquired vibration waveform data blocks as well as store characteristic value history records and event logs.

[0025] Preferably, the wireless communication module adopts a low-power wide area network communication module based on 4GCat.1, NB-IoT or LoRa technology, and the communication module is connected to the low-computing-power edge computing core through UART serial port and GPIO control pin.

[0026] Preferably, the power management module includes a wide-voltage input DC-DC buck circuit, a lithium battery charging management circuit, and multiple low-dropout linear regulators (LDOs); the input voltage range of the wide-voltage input DC-DC buck circuit is 9-36VDC; the power management module is configured to automatically switch to built-in battery power supply when the external power supply is interrupted, and cooperate with the sleep mode management of the low-computing-power edge computing core to achieve microampere-level standby power consumption, wherein the built-in battery is a lithium thionyl chloride battery.

[0027] Preferably, the protective housing has an IP67 protection rating, and is provided with a waterproof aviation plug interface for the sensor signal line, a power input interface, and a status indicator window. The bottom of the housing is provided with a guide rail slot or threaded mounting hole.

[0028] The present invention also provides a monitoring method for a low-computing-power edge computing data acquisition terminal for transmission equipment, comprising the following steps:

[0029] Step S1: Data Synchronization Acquisition and Buffering. The terminal's MCU simultaneously triggers the vibration sensor and temperature sensor to acquire data according to a preset fixed sampling period; the vibration signal is continuously acquired and buffered in blocks, while the current temperature value is read; wherein, the sampling frequency of the vibration signal is set to 2kHz to 10kHz, and 1024 points are continuously acquired each time to form a data block; the sampling frequency of the temperature signal is 1Hz;

[0030] Step S2: Signal preprocessing and noise reduction. The MCU runs a preprocessing algorithm on the buffered vibration data block and applies a digital bandpass filter to filter out power frequency interference and high-frequency noise. The digital bandpass filter is an IIR filter with a cutoff frequency of 50Hz and 1500Hz, used to filter out 50Hz power frequency interference and high-frequency noise above 1500Hz.

[0031] Step S3: Multidimensional feature extraction. The MCU performs time-domain and frequency-domain feature calculations on the preprocessed data block, and calculates temperature-domain features in combination with temperature data to form a feature vector characterizing the current operating status of the equipment.

[0032] Step S4: Local abnormal status judgment and early warning. The MCU compares the feature vector with the built-in threshold or lightweight model to judge the current device status. If abnormal, a local early warning is triggered and a status label is marked.

[0033] Step S5: Data encapsulation and intelligent upload. The MCU encapsulates the feature vector and status tag into a data packet. Based on the judgment result of step S4, it selects a normal upload strategy or an event-triggered upload strategy to send the data through the wireless communication module.

[0034] Preferably, in step S3, the feature vector includes time-domain features, frequency-domain features, and temperature-domain features; the time-domain features include the root mean square (RMS) value and kurtosis of the vibration acceleration; the frequency-domain features are obtained by performing a 256-point or 512-point fixed-point fast Fourier transform (FFT) on the data block to extract the amplitude at the inherent characteristic frequency of the transmission equipment; the temperature-domain features are the difference between the current temperature value and the historical average, i.e., the rate of temperature change.

[0035] Preferably, in step S4, the local abnormal state judgment includes: setting static alarm thresholds for RMS, kurtosis, characteristic frequency amplitude, and temperature change rate, or using a pre-trained lightweight decision tree or anomaly detection model for comprehensive scoring; if any feature value exceeds the threshold or the model output is abnormal, it is judged as an abnormal state, triggering an early warning flag and marking the state label as "high priority event".

[0036] Preferably, in step S5, the normal upload strategy is: when the device is normal, data packets are uploaded every 5-10 minutes; the event-triggered upload strategy is: when an anomaly is detected, upload is started immediately, and the subsequent upload interval is shortened to 30 seconds to achieve intensive tracking.

[0037] Compared with the prior art, the beneficial effects of the present invention are:

[0038] 1. This invention achieves "on-site identification" of fault characteristics by completing the core processing and feature extraction of vibration and temperature signals at the transmission equipment site. Compared to solutions that require uploading raw data to the cloud for analysis, this method reduces the closed-loop delay of "data acquisition - feature analysis - anomaly judgment" from seconds to milliseconds. This real-time edge intelligence can effectively capture early fault characteristics such as instantaneous impacts and spikes that occur during the operation of transmission equipment (such as gearboxes), thereby issuing early warnings at the fault initiation stage. This provides a critical time window for predictive maintenance and operation scheduling decisions of the smart grid, avoiding fault expansion or unplanned downtime due to delayed response, and significantly improving the real-time performance and early warning capabilities of state awareness.

[0039] 2. This invention performs lightweight signal preprocessing (such as filtering) and extraction of key feature values ​​(such as vibration RMS value, kurtosis, characteristic frequency amplitude, and temperature change rate) instead of simply transmitting the raw data stream. This process compresses the amount of data that needs to be uploaded by one to two orders of magnitude, fundamentally solving the "data flood" problem caused by high-frequency sampling in traditional solutions. This significantly reduces the bandwidth pressure on smart grid communication networks (especially public or private wireless networks), while also reducing the resource overhead of cloud data centers for storing and preprocessing massive amounts of raw data. This allows cloud computing power to focus more on advanced analysis and model optimization across devices and sites, greatly optimizing communication bandwidth and cloud computing resource usage.

[0040] 3. This invention addresses the issues of redundant computing power and excessive cost in existing high-performance edge computing solutions for monitoring transmission equipment. It employs a low-power microcontroller (MCU) with precise task matching as the core computing unit, along with a lightweight dedicated diagnostic algorithm. This design significantly reduces terminal hardware cost and power consumption compared to solutions using high-performance processors or FPGAs. The low power consumption also allows it to adapt to diverse power supply environments (such as solar power + batteries). This reduction in cost and power consumption overcomes the economic and engineering barriers to large-scale deployment, making it possible to widely implement cost-effective intelligent monitoring across the massive number of transmission equipment nodes in smart grids.

[0041] 4. This invention offloads core diagnostic functions to an independent terminal on the device side, reducing reliance on continuous, high-quality network connections. Even during network outages or when cloud services are unavailable, the terminal can still operate independently and perform local diagnostics and alarms, ensuring the continuity of monitoring functions and improving the overall robustness of the system. Simultaneously, sensitive raw device data remains locally transmitted; only anonymized feature results are uploaded, effectively reducing the risk of data interception and leakage during transmission, meeting the data security requirements of the power system. Furthermore, the terminal employs industrial-grade hardware and a compact structural design, enabling it to better adapt to harsh operating conditions such as high temperature, high humidity, strong vibration, and electromagnetic interference in transmission equipment, ensuring long-term operational stability.

[0042] 5. The standardized feature data and event information output by the terminal of this invention are designed with full consideration of compatibility with existing smart grid operation and maintenance platforms (such as SCADA and production management systems). This "plug-and-play" data interface enables the real-time health status of equipment to be seamlessly and efficiently integrated into the power grid information flow, directly providing underlying data support for advanced applications such as load forecasting, maintenance plan optimization, operation mode adjustment, and even the ancillary service market. This truly integrates equipment status awareness into the smart grid's scheduling and decision-making closed loop, improving the precision and intelligence of power grid operation and promoting the deep integration of monitoring data with advanced power grid applications. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of the data acquisition terminal system architecture and working scenario of the present invention;

[0044] Figure 2 This is a block diagram of the data acquisition terminal structure of the present invention;

[0045] Figure 3 This is a schematic diagram of the terminal and transmission device installation of the present invention;

[0046] Figure 4 This is a flowchart of the terminal operation and data processing of the present invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] This involves a low-computing-power data acquisition terminal that integrates real-time vibration and temperature data acquisition, edge computing, and intelligent diagnosis, and is deployed on transmission equipment in power generation and substation processes (such as wind turbine gearboxes and pump gearboxes).

[0049] Example 1

[0050] Please see Figure 1 - Figure 4 This invention provides a low-computing-power edge computing data acquisition terminal for transmission equipment. The data acquisition terminal is an embedded hardware device integrating data acquisition, edge computing, and wireless communication. Its core design concept is to achieve low cost, low power consumption, and high reliability by selecting a low-computing-power core processor and optimizing the structural design, while meeting the monitoring requirements of the transmission equipment. The terminal adopts a modular design, including sensor components, a low-computing-power edge computing core, a wireless communication module, a power management module, and a protective shell. The sensor components, low-computing-power edge computing core, wireless communication module, and power management module are all integrated on an internal circuit board. Each component is electrically connected and mechanically fixed through the internal circuit board and is encapsulated within the protective shell.

[0051] 1. The sensor assembly is the data sensing unit of the terminal, used to physically contact the transmission equipment being monitored and acquire raw terminal data of the operating status in real time. The sensor assembly includes a vibration sensor and a temperature sensor.

[0052] The vibration sensor employs an integrated triaxial MEMS (Micro-Electro-Mechanical Systems) accelerometer (such as the ADXL345 or an equivalent chip; this invention selects a MEMS sensor instead of a traditional piezoelectric sensor primarily due to its small size, low power consumption, cost-effectiveness, and ease of integration with circuit boards). The vibration sensor is fixedly installed at key vibration monitoring points of the transmission equipment, such as gearbox bearing housings or the surface of the housing, using high-strength epoxy resin adhesive or a magnetic base, to acquire vibration acceleration in the X, Y, and Z directions and convert it into analog or digital electrical signals.

[0053] The temperature sensor is a digital temperature sensor (such as model DS18B20 or PT100 with signal conditioning circuit). The temperature sensor is tightly attached to the key temperature rise monitoring point of the transmission equipment by thermally conductive silicone or metal clamp, such as near the outer ring of the bearing or the surface of the lubrication oil circuit. Its function is to directly convert the physical quantity of temperature into a digital signal, reducing the signal conditioning link and conforming to the low power consumption design principle.

[0054] 2. The low-computing-power edge computing core is the control, computing, and decision-making center of the terminal. It follows the design principle of "computing power matching" rather than "computing power redundancy". It is electrically connected to the sensor components and the wireless communication module respectively. It is used to read sensor data, run lightweight signal processing and diagnostic algorithms, and control the communication module to transmit data according to the diagnostic results. The low-computing-power edge computing core includes a core processor and a storage unit.

[0055] The core processor is a 32-bit microcontroller (MCU) based on an ARM Cortex-M4 core or an equivalent RISC-V core, such as the STM32F4 or GD32VF103 series. It is located at the center of the circuit board as the main control chip, either via an integrated circuit socket or surface-mount soldering. Its main functions include: receiving and reading sensor data via an SPI / I2C digital interface, running lightweight signal processing and diagnostic algorithms internally, controlling indicator lights and communication modules based on the results, and managing the power and sleep modes of the entire terminal.

[0056] The storage unit includes an integrated Flash memory (for storing program code) within the MCU and an externally extended SRAM or FRAM (ferroelectric memory, such as FM24CL16B). The external extended memory is connected to the MCU via an I2C bus and is used to temporarily cache high-frequency acquired vibration waveform data blocks and store long-term characteristic value history records and event logs to ensure that data is not lost in extreme cases.

[0057] 3. The wireless communication module serves as a bridge for interaction between the terminal and the upper-layer system. It adopts a low-power wide-area network communication module based on 4GCat.1, NB-IoT, or LoRa technology (such as Quectel EC200S, which supports LTE networks, has wide coverage, and low power consumption). The communication module is directly soldered onto the main circuit board through UART serial port and GPIO control pins, and is connected to the low-computing-power edge computing core. It is responsible for sending the processed feature data and early warning messages to the remote smart grid operation and maintenance platform or gateway according to the set protocol format.

[0058] 4. The power management module provides stable operating voltage for all components of the terminal and manages the terminal's power supply mode. It includes a wide-input DC-DC step-down circuit (input voltage range 9-36VDC, adaptable to common 12V or 24V power supplies in industrial environments), a lithium battery charging management circuit, and multiple low-dropout linear regulators (LDOs) to reduce power ripple interference with high-precision sampling. All power chips are located in the power area of ​​the circuit board, with input terminals connected to external power lines via wiring terminals. This module not only converts unstable voltages in industrial environments into internal stable voltages, but also performs trickle charging maintenance on the battery during normal power supply. When external power is interrupted, it automatically switches to the built-in lithium thionyl chloride battery to ensure critical data preservation and alarm information transmission. Combined with the sleep mode management of the low-power edge computing core, it maintains only the real-time clock and critical data preservation, reducing the overall standby current to below 50 microamps, achieving microamp-level standby power consumption for the terminal.

[0059] 5. The protective housing is made of metal or engineering plastic and is used to provide mechanical fixation and environmental protection. The protection level is IP67, which is dustproof and waterproof. The housing is equipped with a waterproof aviation plug interface for sensor signal lines, a power input interface and a status indicator window. The bottom of the housing is equipped with a guide rail slot (which can be directly attached to the guide rail in the power distribution cabinet) or a threaded mounting hole (which can be fixed to the equipment body with screws through the countersunk hole at the bottom) to facilitate on-site installation.

[0060] In summary, this invention, through the design concept of "computing power matching" and a modular hardware architecture, successfully realizes a low-cost, low-power, and highly reliable edge computing data acquisition terminal suitable for smart grid transmission equipment, solving the problems of high cost and difficult deployment of traditional monitoring equipment.

[0061] Example 2

[0062] This invention also provides a monitoring method for a low-computing-power edge computing data acquisition terminal for transmission equipment, comprising the following steps:

[0063] Step S1: Data synchronous acquisition and caching. The terminal's MCU triggers the vibration sensor and temperature sensor to acquire data simultaneously according to a preset fixed sampling period (e.g., every 10 milliseconds). That is, the vibration signal is continuously acquired and cached in blocks, while the current temperature value is read. The synchronous acquisition ensures the time alignment of the vibration and temperature data, which facilitates subsequent correlation analysis and provides a time window for subsequent frequency domain analysis.

[0064] The process conditions are set such that the sampling frequency of the vibration signal is set to 2kHz to 10kHz, and 1024 points are continuously collected each time to form a data block, in order to meet the capture requirements of the common fault characteristic frequency of transmission equipment (usually below 1kHz), while avoiding data redundancy caused by excessive sampling; the sampling frequency of the temperature signal is 1Hz, and one current temperature value is read each time.

[0065] Step S2: Signal preprocessing and noise reduction. The MCU runs a preprocessing algorithm (digital signal processing) on ​​the buffered vibration data block. A digital bandpass filter (preferably an IIR filter) with cutoff frequencies of 50Hz and 1500Hz is applied to the time-domain vibration signal to filter out 50Hz power frequency interference and high-frequency noise above 1500Hz, improving the signal-to-noise ratio and laying the foundation for accurate feature extraction. All calculations are performed internally by the MCU, without relying on external resources.

[0066] Step S3: Multidimensional feature extraction. The MCU performs time-domain and frequency-domain feature calculations on the preprocessed data block, and calculates temperature-domain features in combination with temperature data to form a feature vector characterizing the current operating status of the equipment.

[0067] The time-domain feature calculation includes calculating the root mean square (RMS) value of vibration acceleration to reflect vibration energy and calculating kurtosis to capture impact fault signals; the frequency-domain feature calculation involves performing a 256-point or 512-point fixed-point Fast Fourier Transform (FFT) on the data block to extract the amplitude at frequencies inherent to the transmission equipment (such as gear meshing frequency); the temperature-domain feature calculation refers to calculating the difference between the current temperature value and the historical average value one minute ago, i.e., the rate of temperature change; this step is the core of edge computing, condensing the massive original waveform data into fewer than 10 key feature values, achieving extreme data compression, and directly extracting sensitive indicators for fault diagnosis, significantly reducing the subsequent transmission load;

[0068] Step S4: Local abnormal status judgment and early warning. The MCU compares the feature vector with the built-in static alarm threshold (which can be remotely configured and updated) or the lightweight decision tree / anomaly detection model to determine the current device status. If an anomaly is detected, a local early warning is triggered and a status label is marked.

[0069] Specifically, local anomaly judgment includes: setting static alarm thresholds for RMS, kurtosis, feature frequency amplitude, and temperature change rate, or using a pre-trained lightweight decision tree or anomaly detection model, inputting the feature vector for comprehensive scoring; if any feature value exceeds the threshold or the model output is abnormal, it is judged as an abnormal equipment state, triggering a local warning sign—indicator light flashing, and labeling the current feature vector and anomaly status as a "high-priority event"; this process achieves millisecond-level on-site fault identification;

[0070] Step S5: Data encapsulation and intelligent upload. The MCU encapsulates the feature vector, device identifier / ID, timestamp, and status tag into a JSON data packet. Based on the judgment result of step S4, it selects a normal upload strategy or an event-triggered upload strategy to send the data through the wireless communication module.

[0071] The normal upload strategy is as follows: when the device is working properly, data packets are uploaded every 5-10 minutes to maintain online heart rate and health baseline, thus saving energy.

[0072] The event-triggered upload strategy is as follows: when an anomaly is detected, the communication module is immediately activated to upload and send data, and the subsequent upload interval is shortened to 30 seconds, thereby enabling intensive tracking and real-time monitoring of the fault development process.

[0073] This step enables intelligent data upload strategy. Regular low-frequency uploads maintain device online status and establish a health baseline, while event-triggered high-frequency uploads ensure the real-time nature and completeness of fault information, maximizing energy and bandwidth savings while maintaining monitoring effectiveness.

[0074] By combining the above product structure and working method, this invention constructs a complete closed loop from physical perception to intelligent decision-making, realizing economical, efficient and reliable real-time status monitoring of smart grid transmission equipment.

[0075] In summary, this invention, through precise configuration of computing power at the edge and lightweight algorithm, effectively balances and resolves multiple contradictions between real-time performance, economy, reliability, and integration in the specific scenario of transmission equipment monitoring, providing a practical and feasible technical solution for building an efficient, economical, and secure comprehensive equipment status perception network for smart grids.

[0076] Application Examples

[0077] This embodiment is applied to the monitoring of transformer cooling fans in smart grids. The terminal uses an STM32F4 series MCU as its core.

[0078] First, proceed to step S1, where the MCU configures a timer to trigger the ADXL345 vibration sensor and DS18B20 temperature sensor with a period of 10ms. The vibration sampling rate is set to 5kHz, and 1024 points are collected each time and stored in the external FRAM buffer; the temperature is read once per second.

[0079] Next, step S2 is executed. The MCU calls the IIR bandpass filter function on the 1024 vibration data points, sets the passband to 50Hz-1500Hz, and filters out power grid frequency interference.

[0080] Then, in step S3, the MCU calculates the RMS value and kurtosis of the filtered data; at the same time, it calls the 512-point fixed-point FFT library function to calculate the spectrum, extract the amplitude corresponding to the fan bearing rotation frequency, and calculate the average temperature difference in the most recent minute.

[0081] In step S4, the MCU compares the above feature value with a preset threshold (RMS threshold is set to 0.5g). If the RMS value suddenly increases to 0.8g and the kurtosis increases, the MCU determines that the bearing may have premature wear, immediately sets the abnormality flag, and lights up the red LED.

[0082] Finally, in step S5, the MCU sends a JSON packet containing information such as RMS, kurtosis, and temperature via the 4G module. Since this is determined to be an anomaly, the MCU switches its sending strategy from the default once every 10 minutes to once every 30 seconds, continuously sending alerts to the operations and maintenance platform until the monitoring data returns to normal.

[0083] The above are merely preferred embodiments of the present invention and are 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 low-computing-power edge computing data acquisition terminal for transmission equipment, characterized in that, It includes sensor components, a low-power edge computing core, a wireless communication module, a power management module, and a protective housing; The sensor components, low-power edge computing core, wireless communication module, and power management module are all integrated on the internal circuit board and encapsulated within the protective housing. The sensor assembly is used to physically contact the monitored transmission equipment to acquire raw terminal data of the operating status in real time. The low-power edge computing core is electrically connected to the sensor components and the wireless communication module, respectively, and is used to read sensor data, run lightweight signal processing and diagnostic algorithms, and control the communication module to transmit data according to the diagnostic results. The power management module is used to provide a stable operating voltage for each component of the terminal and to manage the power supply mode of the terminal. The protective housing is used to provide mechanical fixation and environmental protection.

2. The low-computing-power edge computing data acquisition terminal for transmission equipment according to claim 1, characterized in that, The sensor assembly includes a vibration sensor and a temperature sensor; The vibration sensor is an integrated triaxial MEMS accelerometer. The vibration sensor is fixedly installed on the gearbox bearing seat or housing surface of the transmission equipment by high-strength epoxy resin adhesive or magnetic base, and is used to acquire vibration acceleration signals in the X, Y and Z directions. The temperature sensor is a digital temperature sensor, which is tightly attached and fixed to the outer ring of the bearing or the surface of the lubrication oil passage of the transmission equipment by thermally conductive silicone or metal clamps, and is used to directly convert the physical quantity of temperature into a digital signal.

3. The low-computing-power edge computing data acquisition terminal for transmission equipment according to claim 1, characterized in that, The low-computing-power edge computing core includes a core processor and a storage unit; The core processor is a 32-bit microcontroller (MCU) based on an ARM Cortex-M4 core or an equivalent RISC-V core. The MCU receives sensor data through a digital interface and runs lightweight algorithms and manages power and sleep modes internally. The storage unit includes a Flash memory integrated inside the MCU and an external extended memory, wherein the external extended memory is an SRAM or a ferroelectric memory (FRAM); the external extended memory is connected to the MCU via an I2C bus and is used to temporarily cache high-frequency acquired vibration waveform data blocks as well as store characteristic value history records and event logs.

4. The low-computing-power edge computing data acquisition terminal for transmission equipment according to claim 1, characterized in that, The wireless communication module adopts a low-power wide area network communication module based on 4GCat.1, NB-IoT or LoRa technology. The communication module is connected to the low-computing-power edge computing core through UART serial port and GPIO control pin.

5. The low-computing-power edge computing data acquisition terminal for transmission equipment according to claim 1, characterized in that, The power management module includes a wide voltage input DC-DC buck circuit, a lithium battery charging management circuit, and multiple low dropout linear regulators (LDOs). The input voltage range of the wide voltage input DC-DC buck circuit is 9-36VDC; The power management module is configured to automatically switch to built-in battery power when the external power supply is interrupted, and in conjunction with the sleep mode management of the low computing power edge computing core, achieve microampere-level standby power consumption. The built-in battery is a lithium thionyl chloride battery.

6. The low-computing-power edge computing data acquisition terminal for transmission equipment according to claim 1, characterized in that, The protective housing has an IP67 protection rating and is equipped with a waterproof aviation plug interface for the sensor signal line, a power input interface, and a status indicator window. The bottom of the housing is equipped with a guide rail slot or threaded mounting holes.

7. The monitoring method for a low-computing-power edge computing data acquisition terminal for transmission equipment according to any one of claims 1-6, characterized in that, Includes the following steps: Step S1: Data Synchronization Acquisition and Buffering. The terminal's MCU simultaneously triggers the vibration sensor and temperature sensor to acquire data according to a preset fixed sampling period; the vibration signal is continuously acquired and buffered in blocks, while the current temperature value is read; wherein, the sampling frequency of the vibration signal is set to 2kHz to 10kHz, and 1024 points are continuously acquired each time to form a data block; the sampling frequency of the temperature signal is 1Hz; Step S2: Signal preprocessing and noise reduction. The MCU runs a preprocessing algorithm on the buffered vibration data block and applies a digital bandpass filter to filter out power frequency interference and high-frequency noise. The digital bandpass filter is an IIR filter with a cutoff frequency of 50Hz and 1500Hz, used to filter out 50Hz power frequency interference and high-frequency noise above 1500Hz. Step S3: Multidimensional feature extraction. The MCU performs time-domain and frequency-domain feature calculations on the preprocessed data block, and calculates temperature-domain features in combination with temperature data to form a feature vector characterizing the current operating status of the equipment. Step S4: Local abnormal status judgment and early warning. The MCU compares the feature vector with the built-in threshold or lightweight model to judge the current device status. If abnormal, a local early warning is triggered and a status label is marked. Step S5: Data encapsulation and intelligent upload. The MCU encapsulates the feature vector and status tag into a data packet. Based on the judgment result of step S4, it selects a normal upload strategy or an event-triggered upload strategy to send the data through the wireless communication module.

8. The monitoring method for a low-computing-power edge computing data acquisition terminal for transmission equipment according to claim 7, characterized in that, In step S3, the feature vector includes time-domain features, frequency-domain features, and temperature-domain features; The time-domain features include the root mean square (RMS) value of the vibration acceleration and the kurtosis; The frequency domain feature is obtained by performing a 256-point or 512-point fixed-point fast Fourier transform (FFT) on the data block to extract the amplitude at the inherent characteristic frequency of the transmission equipment. The temperature range characteristic is the difference between the current temperature value and the historical average, i.e., the rate of temperature change.

9. The monitoring method for a low-computing-power edge computing data acquisition terminal for transmission equipment according to claim 7, characterized in that, In step S4, the local abnormal state judgment includes: setting static alarm thresholds for RMS, kurtosis, feature frequency amplitude, and temperature change rate, or using a pre-trained lightweight decision tree or anomaly detection model for comprehensive scoring; if any feature value exceeds the threshold or the model output is abnormal, it is judged as an abnormal state, triggering an early warning flag and marking the state label as "high priority event".

10. The monitoring method for a low-computing-power edge computing data acquisition terminal for transmission equipment according to claim 7, characterized in that, In step S5, the normal upload strategy is as follows: when the device is working properly, data packets are uploaded every 5-10 minutes. The event-triggered upload strategy is as follows: when an anomaly is detected, the upload is started immediately, and the subsequent upload interval is shortened to 30 seconds to achieve intensive tracking.

Images