A remote monitoring method of a power device, a server, a product and a medium

Abstract

A remote monitoring method of power equipment, a server, a product and a medium, relate to the field of safety production monitoring. The method comprises the following steps: synchronously acquiring real-time running temperature, load current and multi-dimensional meteorological parameters of the environment of the power equipment; inputting the load current and multi-dimensional meteorological parameters into a preset normal state regression prediction model to output a theoretical temperature value under the current working condition; calculating a temperature residual value between the real-time running temperature and the theoretical temperature value; if the real-time running temperature is less than or equal to a preset static alarm threshold, and it is determined that the temperature residual value is always greater than a preset residual alarm threshold within a time window, a fault uploading instruction is generated; then the data uploading frequency is adjusted to an alarm uploading frequency, and the preset historical period running data and the real-time data are packaged and uploaded. The application excludes the interference of external cooling effect, effectively captures the latent fault hidden by the static threshold, and improves the accuracy of power equipment monitoring under complex weather.

Description

Technical Field

[0001] This application relates to the field of safety production monitoring, and in particular to a remote monitoring method, server, product and medium for power equipment. Background Technology

[0002] With the application of power Internet of Things (IoT) technology, the status monitoring of outdoor power equipment typically involves deploying sensors such as temperature and current at key nodes of the equipment, and then transmitting the collected raw time-series data back to the cloud master station for processing via wireless network. However, this mode of continuous full data transmission consumes network bandwidth resources and generates communication traffic when there are a large number of devices. Furthermore, in areas with unstable communication signals, the increased data transmission latency affects the timeliness of data acquisition from the cloud.

[0003] To optimize network bandwidth usage and improve data transmission efficiency, related technologies typically employ edge gateway data filtering schemes based on static thresholds. This involves setting a pre-defined temperature alarm value for the device on the local gateway. When the real-time temperature value collected by the sensor exceeds this alarm value, the relevant data segment is deemed abnormal and uploaded. Data below the alarm value is filtered locally or sampled at a lower frequency. This approach reduces the amount of unnecessary data uploaded under relatively constant ambient temperature conditions, thereby saving storage resources and communication costs.

[0004] However, in outdoor scenarios with variable weather conditions, when power equipment encounters environmental disturbances such as sudden rainfall or strong winds, the cooling effect brought by the external environment may offset the temperature rise caused by early latent faults inside the equipment (such as increased contact resistance). As a result, the combined temperature value is still within the preset normal threshold range, causing the edge gateway to judge data containing fault characteristics as normal and intercept it based on static logic. This increases the risk of the system missing early potential problems in complex environments. Summary of the Invention

[0005] This application provides a method, server, product, and medium for remote monitoring of power equipment, which can improve the accuracy of equipment monitoring systems in detecting latent faults under complex weather conditions.

[0006] In a first aspect, this application provides a remote monitoring method for power equipment, applied to an edge computing server of an equipment monitoring system. The method includes: synchronously acquiring the real-time operating temperature, load current, and multi-dimensional meteorological parameters of the environment in which the power equipment is located; inputting the load current and multi-dimensional meteorological parameters into a preset normal state regression prediction model, and outputting the theoretical temperature value of the power equipment under the current operating conditions; calculating the difference between the real-time operating temperature and the theoretical temperature value to obtain a temperature residual value; if the real-time operating temperature is less than or equal to a preset static alarm threshold, and the temperature residual value is always greater than the preset residual alarm threshold within a preset judgment time window, then generating a fault upload command; in response to the fault upload command, adjusting the data reporting frequency for sending data to the equipment monitoring system to an alarm reporting frequency, wherein the alarm reporting frequency is higher than a preset daily reporting frequency; encapsulating the operating data within a preset historical period before the generation of the fault upload command with the currently collected real-time data, and sending it to the equipment monitoring system according to the alarm reporting frequency.

[0007] By adopting the above technical solution, the edge computing server first inputs the synchronously acquired load current and multi-dimensional meteorological parameters into a pre-set normal state regression prediction model, decoupling the external environmental heat dissipation effect of the power equipment from the internal load heating effect. Next, the edge computing server calculates the temperature residual between the real-time operating temperature and the theoretical temperature value. This residual directly reflects the abnormal temperature rise component inside the equipment that is masked by environmental cooling effects such as rainwater, thus identifying hidden fault characteristics even before the real-time operating temperature reaches the static alarm threshold. Finally, the edge computing server adjusts the data reporting frequency to the alarm reporting frequency and encapsulates historical data for tracing back to the equipment monitoring system. In summary, this solution eliminates the interference of cooling effects caused by external multi-dimensional meteorological parameters, captures hidden abnormal states of equipment masked by static alarm thresholds, and improves the accuracy of the equipment monitoring system in detecting latent faults under complex meteorological conditions.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, after calculating the difference between the real-time operating temperature and the theoretical temperature value to obtain the temperature residual value, the method further includes: within a preset judgment time window, if the change in the real-time operating temperature is greater than a preset fluctuation threshold, and the temperature residual value is less than or equal to a preset residual alarm threshold, then the data reporting frequency is maintained at a preset daily reporting frequency.

[0009] By adopting the above technical solution, the edge computing server introduces a dual verification mechanism for temperature fluctuations to distinguish between normal operating fluctuations and actual faults. When a drastic change in real-time operating temperature is detected, the edge computing server does not directly determine it as an anomaly, but further checks the temperature residual value from the same period. If the temperature residual value remains within a safe range, it indicates that the drastic temperature fluctuation fully conforms to the theoretical predictions made by the regression model based on sudden load changes or meteorological changes, and is a reasonable physical response rather than an internal fault. Therefore, the preset daily reporting frequency is maintained, and no alarm is triggered. In summary, this solution utilizes the model's predictive capabilities to filter out false alarms caused by sudden environmental or load changes, reducing the occupation of communication channels by invalid data.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, after the step of calculating the difference between the real-time operating temperature and the theoretical temperature value to obtain the temperature residual value, the method further includes: if the real-time operating temperature is greater than a preset static alarm threshold, then generating a fault upload command.

[0011] By adopting the above technical solution, considering that in certain extreme runaway scenarios, the absolute temperature of the equipment may rapidly rise to the material's tolerance limit, the edge computing server immediately generates a fault upload command when the real-time operating temperature exceeds the static alarm threshold, ensuring an immediate response to the risk of high-temperature melting or fire. In summary, this solution ensures that while the algorithm model handles complex operating conditions, it still possesses the ultimate protection capability to prevent equipment damage due to absolute overheating.

[0012] In conjunction with some embodiments of the first aspect, in some embodiments, the method further includes: within a preset evaluation period, acquiring multiple continuously calculated temperature residual values ​​to form a reference residual sequence, wherein the time span of the preset evaluation period is greater than a preset judgment time window; calculating the fitting slope of the reference residual sequence as a function of time; if the fitting slope is greater than zero, and each temperature residual value in the reference residual sequence is less than or equal to the residual alarm threshold, then triggering the baseline offset diagnosis step.

[0013] By employing the above technical solution, the edge computing server accumulates a temperature residual sequence within a preset evaluation period and calculates its fitting slope, thereby keenly detecting subtle signs of slow deterioration in equipment thermal performance that have not yet triggered instantaneous alarms. When a continuously rising positive slope is detected in the residuals, the edge computing server triggers the baseline offset diagnostic step during the "sub-healthy" stage before a fault occurs, rather than waiting for the fault to fully develop. In summary, this solution enables predictive maintenance of power equipment status, and can detect gradual hidden dangers such as contact point oxidation or decreased heat dissipation capacity in advance.

[0014] In conjunction with some embodiments of the first aspect, in some embodiments, the step of encapsulating the operating data within a preset historical period before the generation of the fault upload command with the currently collected real-time data specifically includes: in response to the fault upload command, performing a retrospective comparison of multiple consecutive historical temperature residual values ​​in the reference residual sequence in reverse chronological order, identifying the preceding inflection point where the historical temperature residual value changes from a stable fluctuation state to a continuous upward trend; taking the timestamp corresponding to the preceding inflection point as the start time, and taking the current time of generating the fault upload command as the end time; extracting the operating data and real-time data within the dynamic period consisting of the start time and the end time and encapsulating them.

[0015] By employing the above technical solution, the edge computing server first identifies the preceding inflection point in the reference residual sequence, where the fluctuation transitions from steady-state to monotonically increasing, using a time-reverse tracing algorithm. This inflection point marks the physical start time of the fault in the time series. Next, the edge computing server constructs a dynamic time period for data extraction, starting from this inflection point and ending at the alarm time, ensuring that the uploaded data packets completely cover the entire lifecycle of the fault, from its inception to its development and eventual outbreak. In summary, this solution abandons the method of mechanically extracting historical data of fixed durations, providing high-quality diagnostic samples containing complete fault characteristics, and improving the accuracy and efficiency of root cause analysis of faults by the cloud-based main station.

[0016] In conjunction with some embodiments of the first aspect, in some embodiments, triggering a baseline offset diagnostic step specifically includes: acquiring multiple historical load currents corresponding to each temperature residual value within a preset evaluation period; calculating the square value of each historical load current and constructing a heating reference sequence from multiple square values ​​according to time sequence; calculating the correlation coefficient between the reference residual sequence and the heating reference sequence; if the correlation coefficient is greater than a preset correlation threshold, determining that there is an abnormal contact resistance inside the power equipment and generating a fault upload command; if the correlation coefficient is less than or equal to the preset correlation threshold, determining that there is heat dissipation aging on the surface of the power equipment and performing model parameter calibration operation on the normal state regression prediction model.

[0017] By adopting the above technical solution, the edge computing server achieves automatic classification of the root causes of temperature anomalies. The edge computing server calculates the correlation coefficient between the temperature residual sequence and the square value of historical load current (i.e., heat generation reference). If the correlation coefficient is high, the edge computing server confirms that the temperature rise is proportional to the square of the current, and determines it to be an electrical fault with excessive contact resistance; if the correlation coefficient is low, it is determined to be an external heat dissipation aging that is not related to current. In summary, this solution utilizes the computing power of the edge side to deeply mine the physical correlation between data, and integrates the thermodynamic mechanism of Joule's law (heat generation is proportional to the square of the current), giving the purely data-driven model interpretable physical meaning, thereby realizing a penetrating diagnosis from the characterization (temperature change) to the root cause (electrical contact degradation and physical material aging).

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, a model parameter calibration operation is performed on the normal regression prediction model, specifically including: calculating the arithmetic mean of each temperature residual value contained in the reference residual sequence; and storing the calculated arithmetic mean as a temperature compensation parameter in the normal regression prediction model.

[0019] By adopting the above technical solution, after confirming that the heat dissipation aging is not due to failure, the edge computing server first calculates the arithmetic mean of the temperature residual sequence to quantify the systematic temperature deviation caused by aging. The edge computing server then uses this average as a compensation parameter to update the normal regression prediction model, so that the theoretical benchmark output by the model is re-aligned with the current actual physical state of the device. Finally, this solution eliminates false alarms caused by accumulated errors due to natural aging of the device, improving the accuracy and robustness of the monitoring system during long-term operation.

[0020] In a second aspect, this application provides an edge computing server for a device monitoring system. The edge computing server includes: one or more processors and a memory; the memory is coupled to one or more processors and is used to store computer program code, which includes computer instructions. The one or more processors invoke the computer instructions to cause the edge computing server to perform the method described in the first aspect and any possible implementation thereof.

[0021] Thirdly, this application provides a computer-readable storage medium storing computer instructions that, when executed on an edge computing server, cause the edge computing server to perform the method described in the first aspect and any possible implementation thereof.

[0022] Fourthly, this application provides a computer program product, including a computer program or instructions that, when run on an edge computing server, cause the edge computing server to perform the method described in the first aspect and any possible implementation thereof.

[0023] Understandably, the edge computing server provided in the second aspect, the computer-readable storage medium provided in the third aspect, and the computer program product provided in the fourth aspect are all used to execute the methods provided in the embodiments of this application. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods, and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0025] 1. Because a normal regression prediction model is used to calculate the temperature residual value, and a fault upload command is generated when the real-time operating temperature is lower than the static alarm threshold but the temperature residual value continues to exceed the standard, the edge computing server can isolate environmental interference and identify abnormal temperature rise inside the equipment that is masked by external cooling effects such as strong winds or rain. This effectively solves the technical problem that relying solely on fixed threshold monitoring in related technologies can easily lead to missed detection of latent thermal faults under complex weather conditions. As a result, it achieves accurate capture of hidden faults and adaptive adjustment of data reporting frequency, taking into account both monitoring timeliness and network resource efficiency.

[0026] 2. By constructing a reference residual sequence and calculating its fitting slope within a preset evaluation period, and triggering a baseline offset diagnosis step when the fitting slope is greater than zero and the temperature residual value does not exceed the standard, the edge computing server can keenly capture subtle signs of slow deterioration of equipment thermal performance from long-term time-series data that have not yet triggered instantaneous alarms. This effectively solves the problem in related technologies that only focus on transient threshold exceedances and lack the ability to predict trends of early gradual hidden dangers, thereby realizing early warning and predictive maintenance of the "sub-healthy" state of power equipment.

[0027] 3. By using the correlation coefficient between the calculated reference residual sequence and the heating reference sequence constructed based on historical load current, and distinguishing between abnormal contact resistance and heat dissipation aging, the edge computing server can automatically identify whether the temperature rise is caused by load-driven electrical faults or non-urgent physical aging based on the physical mechanism of Joule's law. This effectively solves the problem that related technologies have difficulty distinguishing the nature of faults, resulting in frequent false alarms of non-critical aging drift. This enables the automated classification of fault root causes and the adaptive adjustment of monitoring benchmarks through model parameter calibration operations. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of a scenario for a remote monitoring method for power equipment according to an embodiment of this application;

[0029] Figure 2 This is a flowchart illustrating a remote monitoring method for power equipment according to an embodiment of this application;

[0030] Figure 3 This is another flowchart illustrating a remote monitoring method for power equipment in an embodiment of this application;

[0031] Figure 4 This is a schematic diagram of the physical device structure of an edge computing server in an embodiment of this application. Detailed Implementation

[0032] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification of this application, the singular expressions “a,” “an,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0033] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0034] This application provides a method for remote monitoring of power equipment, which is described below in conjunction with... Figure 1 This application describes the application scenarios of the embodiments. Figure 1 This is a schematic diagram illustrating a remote monitoring method for power equipment according to an embodiment of this application.

[0035] In related technologies, edge gateway data filtering schemes based on static thresholds can be used to optimize network bandwidth usage and reduce the amount of unnecessary data uploaded.

[0036] The remote monitoring method for power equipment in this application involves inputting load current and multidimensional meteorological parameters into a preset normal state regression prediction model and calculating the temperature residual between the real-time operating temperature and the output theoretical temperature value. This enables the automatic generation of a fault upload command and adjustment of the data reporting frequency to the alarm reporting frequency when the real-time operating temperature is less than or equal to a preset static alarm threshold but the temperature residual value is continuously greater than a preset residual alarm threshold. This not only eliminates the interference of external cooling effects caused by multidimensional meteorological parameters but also identifies hidden abnormal states of equipment that are masked by the static alarm threshold.

[0037] It is evident that the remote monitoring method for power equipment in this application embodiment can effectively solve the technical problem that relying solely on fixed threshold monitoring in related technologies can easily lead to missed detection of latent thermal faults under complex weather conditions, thereby improving the effectiveness and accuracy of detecting hidden faults in power equipment.

[0038] To facilitate understanding, the method provided in this implementation will be described in detail below, using the above scenario as an example. Please refer to [link / reference]. Figure 2 This is a flowchart illustrating a remote monitoring method for power equipment in an embodiment of this application.

[0039] S201. Synchronously acquire real-time operating temperature, load current, and multi-dimensional meteorological parameters of the environment in which the power equipment is located;

[0040] In this context, "electrical equipment" refers to key physical node equipment in a power network used for power transmission, distribution, or energy conversion, such as outdoor switchgear, transformers, or high-voltage transmission joints. Real-time operating temperature refers to the physical temperature reading measured on the surface of the electrical equipment or its key heat-generating components at that moment, such as a contactor contact temperature of 45℃. Load current indicates the actual operating current flowing through the main circuit of the electrical equipment, reflecting the intensity of the Joule heat source generated by impedance within the equipment, such as a current operating current of 200A. Multidimensional meteorological parameters represent a collection of environmental condition data reflecting the microclimate and heat dissipation conditions of the local area where the electrical equipment is located, such as ambient temperature of 20℃, ambient humidity of 80%, ambient wind speed of 5 m / s, and instantaneous rainfall.

[0041] Edge computing servers perform routine local monitoring and real-time data acquisition from the front end. Specifically, the edge computing server establishes communication connections with current sensors, temperature sensors, and environmental meteorological monitoring stations deployed around the node under test via wired fieldbus or wireless short-range communication interfaces. During data reception, the edge computing server's built-in clock synchronization mechanism (such as the NTP protocol) timestamps the data streams transmitted from multiple different types of sensors to ensure that the acquired load current, multi-dimensional meteorological parameters, and real-time operating temperature occur at the same sampling time.

[0042] S202. Input the load current and multi-dimensional meteorological parameters into the preset normal state regression prediction model, and output the theoretical temperature value of the power equipment under the current operating conditions.

[0043] The pre-set normal state regression prediction model refers to a machine learning or mathematical mechanism model that has been pre-trained in the cloud and deployed locally on the edge. It is trained based on a large amount of historical operating data under fault-free conditions and is used to accurately characterize the dynamic thermodynamic relationship between "internal heat generation and external heat dissipation" of power equipment. Examples include locally deployed random forest regression models or multiple linear compensation models. The current operating condition refers to the comprehensive service environment of the power equipment at the current moment, which combines the internal electrical load and external meteorological and physical feedback, such as a high-load operating state accompanied by cooling from heavy rain and strong winds. The theoretical temperature value represents the expected temperature response level that the power equipment should exhibit when it is in a completely healthy (i.e., without hidden dangers such as excessive contact resistance) and in an absolutely normal state, under the combined influence of environmental heat dissipation caused by multi-dimensional meteorological parameters and internal heat generation caused by load current.

[0044] After successfully collecting and synchronizing multi-source heterogeneous data, the edge computing server needs to perform a baseline mapping of the absolute health status of the devices. Specifically, the edge computing server extracts load current values ​​and multi-dimensional meteorological parameter matrices at the same timestamp from the synchronously aggregated data frames as input feature vectors, and calls the normal state regression prediction model deployed in the memory of the edge computing server for forward inference calculation.

[0045] During inference, the normal state regression prediction model automatically assesses the base temperature rise caused by the current load and introduces negative temperature compensation weights caused by sudden changes in wind speed and rainfall. This dynamically inversely derives a theoretical temperature value that closely matches the current complex environmental heat exchange background. This theoretical temperature value replaces the traditional fixed reference cold junction temperature, providing a dynamic and condition-adaptive reference benchmark for fault identification.

[0046] Furthermore, the training process of the normal regression prediction model includes:

[0047] First, a massive amount of sample data from the power equipment during its historical normal operation cycle is acquired. The sample set includes synchronously recorded load current, multi-dimensional meteorological parameters, and corresponding actual operating temperature markings. The sample set is then preprocessed, with outliers removed using median filtering or Gaussian filtering, and features of different dimensions such as current, wind speed, and rainfall normalized to eliminate dimensional influences. Next, a regression algorithm (such as random forest regression or deep neural networks) is used to establish a nonlinear mapping logic. Load current and multi-dimensional meteorological parameters are used as model inputs, and actual operating temperature is used as the prediction target for iterative training. The model's internal weights are optimized by minimizing the mean squared error loss function, enabling it to accurately simulate the thermal balance of the equipment under healthy conditions. After training, an independent validation set is used to evaluate the model's prediction bias distribution, thereby determining a residual alarm threshold that balances sensitivity and robustness. The generated lightweight model parameters are then pre-placed in an edge computing server.

[0048] S203. Calculate the difference between the real-time operating temperature and the theoretical temperature value to obtain the temperature residual value;

[0049] The temperature residual value is used to represent the deviation or net error between the actual temperature measured by the sensing device and the expected normal temperature inferred by the model. For example, if the actual operating temperature is 42℃ and the theoretical temperature value calculated by the model is 38℃, then the temperature residual value is 4℃.

[0050] After the edge computing server successfully outputs the theoretical temperature value, it needs to quantify the degree to which the device's current operating characteristics deviate from the healthy baseline. Specifically, the edge computing server uses its embedded arithmetic logic unit to extract the synchronized real-time operating temperature and simultaneously retrieves the theoretical temperature value of the same frequency predicted by the model in S202. The edge computing server performs a difference operation on the two, and the resulting temperature residual value is stored in the edge computing server's local register or short-term cache queue.

[0051] S204. If the real-time operating temperature is less than or equal to the preset static alarm threshold, and the temperature residual value is always greater than the preset residual alarm threshold within the preset judgment time window, then a fault upload command is generated.

[0052] The preset static alarm threshold refers to the upper limit of temperature set according to the rated operating parameters, safety specifications and historical operating experience of the power equipment. The function of this parameter is to provide the first basic fuse interception at the edge to prevent fire caused by seriously exceeding the tolerance limit. Its setting scheme usually refers to the maximum allowable operating temperature provided by the equipment manufacturer, combined with the percentile value (such as the 95th percentile value) of the historical temperature statistics on site, and determined after leaving a certain safety margin.

[0053] The preset judgment time window refers to the time interval used to observe whether the abnormal temperature deviation has the persistence and stability in the time dimension. The function of this parameter is to filter out transient calculation glitches caused by the meteorological sensor being momentarily blocked by leaves, wind gusts, eddies, or electromagnetic pulse interference, so as to avoid frequent false alarms. Its setting scheme is usually combined with the thermal capacitance hysteresis characteristics and thermal conduction time constant of the power equipment, for example, it is set to 5 consecutive sampling cycles or 3 minutes.

[0054] The preset residual alarm threshold refers to the upper limit of the maximum reasonable white noise or inherent temperature measurement error between the actual temperature drift of the device and the theoretical temperature allowed by the system. Its setting scheme is usually based on the maximum fitting residual boundary range of the machine learning prediction model on the validation dataset (such as the 3-Sigma principle) and the inherent accuracy of the sensor, for example, it is set to 3.5℃.

[0055] The fault upload command is a high-level interrupt and control signaling message automatically generated by the edge computing server operating system through logical judgment. This signaling message is used to wake up the subsequent high-priority alarm communication scheduling mechanism.

[0056] This step is executed when hidden faults are masked by drastic fluctuations in outdoor weather conditions (such as a sudden drop in temperature caused by heavy rain), and the edge computing server enters the deep fault diagnosis logic after completing the residual sequence calculation. Specifically, the edge computing server first compares the real-time operating temperature with the preset static alarm threshold. If it does not exceed the threshold, it means that according to the traditional static filtering mechanism, this feature would be mistakenly identified as safe "normal data" and blocked.

[0057] Based on this, the edge computing server opens a preset judgment time window and initiates a sliding window monitoring mechanism to scan and review multiple consecutive temperature residual values ​​entering the time window. It determines whether the temperature residual value at each sampling moment within the window is greater than a preset residual alarm threshold. If the temperature residual value consistently exceeds the residual alarm threshold throughout the entire judgment time window, the edge computing server determines that the device has an internal latent fault masked by the external cooling effect and generates a fault upload command.

[0058] S205. In response to the fault upload command, the data reporting frequency for sending data to the equipment monitoring system is adjusted to the alarm reporting frequency, which is higher than the preset daily reporting frequency.

[0059] The data reporting frequency refers to the frequency or time pulse interval at which the edge computing server sends valid data packets to the cloud via the network link, such as 10 seconds / time.

[0060] Alarm reporting frequency is used to indicate the high-density tracking sampling transmission rate used when the device is triggered into a suspected fault state. It is designed to supplement the long-sequence diagnostic map in the cloud with rich local detail changes, such as sending data to the background once per second.

[0061] The preset daily reporting frequency refers to the lower data reporting frequency used when the device is running normally and no abnormalities are detected. The setting scheme is usually based on the minimum keep-alive heartbeat interval of the IoT control flow (such as the MQTT protocol) and combined with the battery power consumption limit. For example, it is set to send a data packet to the cloud once every 5 minutes.

[0062] The execution of this scenario involves an edge computing server diagnosing and capturing early signs of a deteriorating problem, thereby intercepting the fault upload command. This requires a background coordination mechanism to obtain a rich pool of diagnostic samples. Specifically, upon receiving the fault upload command, the edge computing server's main control thread immediately intervenes in its own communication service daemon (such as the communication interface board process of the edge gateway). The edge computing server directly interrupts the timer interrupt of the preset daily reporting frequency set in the network protocol stack, resetting and refreshing the transmission interval parameter to the alarm reporting frequency. This enables adaptive switching between two communication states: low-frequency sleep mode during normal operation and high-frequency concurrent transmission during abnormal events.

[0063] S206. The operating data within the preset historical period before the fault upload command is generated is encapsulated with the currently collected real-time data and sent to the equipment monitoring system according to the alarm reporting frequency.

[0064] This step specifically includes:

[0065] The system packages the operational data within a preset historical period before the fault upload command is generated into a traceability data package and uploads it to the equipment monitoring system; and adjusts the data reporting frequency for sending real-time data to the equipment monitoring system from the preset daily reporting frequency to the alarm reporting frequency.

[0066] The preset historical time period refers to the length of time that traces back from the moment the fault upload command is generated. The operational data within this period contains information on the evolution of the fault from its inception to its detection. The setting scheme for this time period is determined based on the latency period of typical power equipment faults and the local cache capacity of the edge computing server. It is usually set to several hours to several days, which is necessary to cover the early characteristics of the fault while avoiding excessive data volume that could lead to transmission delays.

[0067] Operational data refers to a complete time-series dataset collected and cached by the edge computing server within a preset historical period, including real-time operating temperature, load current, multi-dimensional meteorological parameters, theoretical temperature values, and temperature residual values.

[0068] Once the frequency scheduling switch is successfully completed and the edge computing server has the conditions for broadband channel transmission, it needs to report the original event details and corresponding source information. Specifically, the edge computing server first retrieves and extracts historical continuous operation data covering this preset historical time period from its local cache unit. Subsequently, the edge computing server uses a data stream processor to encapsulate this previously unreported "forward source information" and the "real-time data" that floods in after the alarm, forming a complete data packet containing the entire fault evolution process. During the encapsulation process, time tags and data type identifiers are respectively marked for historical data and real-time data to facilitate time-series reconstruction and segmented analysis by the device monitoring system. Finally, the edge computing server transparently transmits the alarm to the cloud-based device monitoring system with high priority at the alarm reporting frequency through 4G / 5G or other wireless link channels, thereby completely completing the reporting of hidden fault closed-loop alarms and long-term multi-dimensional graph data.

[0069] It should be noted that historical continuous operation data only needs to be packaged and uploaded once (or in sub-packages) as a traceability archive, while subsequent increases in frequency (alarm reporting frequency) only send real-time data.

[0070] In this embodiment, since the load current and the multidimensional meteorological parameters are input into a preset normal state regression prediction model to output the theoretical temperature value, and a fault upload command is generated when the real-time operating temperature is less than or equal to a preset static alarm threshold and the temperature residual value is greater than a preset residual alarm threshold, it is possible to identify internal abnormal temperature rise by removing external cooling effect interference. This effectively solves the problem in related technologies where relying solely on fixed threshold monitoring leads to the easy omission of latent faults under complex weather conditions, thereby achieving accurate capture of hidden faults and adaptive adjustment of data reporting frequency.

[0071] In the above embodiments, the real-time capture and alarm data reporting of hidden faults under complex meteorological conditions are mainly achieved by the transient joint determination of real-time running temperature and temperature residual value.

[0072] In practical applications, the deterioration of the thermal performance of power equipment is often a gradual characteristic over a long period of time, and the equipment monitoring system needs to further analyze the physical root causes of abnormal temperature rise in order to accurately distinguish between acute contact resistance abnormalities and long-term external heat dissipation aging.

[0073] Therefore, in conjunction with the above embodiments, the method provided in this embodiment will be described in further detail below. Please refer to... Figure 3 This is another flowchart illustrating a remote monitoring method for power equipment in an embodiment of this application.

[0074] S301. Synchronously acquire real-time operating temperature, load current, and multi-dimensional meteorological parameters of the environment in which the power equipment is located;

[0075] S302. Input the load current and multidimensional meteorological parameters into the preset normal state regression prediction model, and output the theoretical temperature value of the power equipment under the current operating conditions.

[0076] S303. Calculate the difference between the real-time operating temperature and the theoretical temperature value to obtain the temperature residual value;

[0077] Steps S301 to S303 are similar to those described in the above embodiments as steps S201 to S203, and will not be repeated here.

[0078] S304. Within the preset evaluation period, multiple temperature residual values ​​obtained from continuous calculation are used to form a reference residual sequence. The time span of the preset evaluation period is greater than the preset judgment time window.

[0079] This step is performed after the edge computing server completes the calculation of a single temperature residual value, and when a data foundation for long-term trend analysis needs to be built. Specifically, the edge computing server starts the local time-series data management module, traces back the time length corresponding to the preset evaluation period from the current moment, and extracts multiple temperature residual values ​​calculated continuously within that time period from the local cache unit.

[0080] The preset evaluation period refers to the time span during which the edge computing server observes the long-term evolution trend of the temperature residual of power equipment. The setting scheme is usually determined by combining the typical aging cycle of power equipment (such as the oxidation cycle of contact points and the dust accumulation cycle of heat sinks) and the local cache capacity of the edge computing server, for example, it is set to several days to several weeks.

[0081] The edge computing server organizes these discrete temperature residual values ​​into an ordered reference residual sequence according to the sampling timestamps, and establishes a time index for the sequence to facilitate subsequent trend fitting and abnormal pattern recognition.

[0082] The reference residual sequence represents a time-series data set consisting of multiple temperature residual values ​​arranged in chronological order within a preset evaluation period, used to reflect the long-term trajectory of equipment temperature deviation from the theoretical benchmark.

[0083] The reference residual sequence is stored in a circular buffer on the edge computing server and is dynamically updated as new data is generated, always maintaining a data window that covers the most recent complete preset evaluation period.

[0084] It should be noted that the preset evaluation period has a longer time span than the preset judgment time window. This design enables the edge computing server to identify slowly accumulating baseline offset phenomena over a longer time scale, while capturing sudden fault characteristics over a shorter time scale, thus achieving collaborative monitoring across multiple time scales.

[0085] S305. Calculate the fitting slope of the reference residual sequence as a function of time;

[0086] The fitting slope is used to represent the overall trend direction and rate of change of the reference residual sequence over time, and is obtained by performing linear regression or least squares fitting on the time series data.

[0087] This step is performed after the edge computing server successfully constructs the reference residual sequence, when it is necessary to quantify the long-term evolution of the sequence. Specifically, the edge computing server calls its built-in numerical analysis algorithm library to perform linear regression fitting calculations on the reference residual sequence. The edge computing server converts the timestamps corresponding to each temperature residual value in the sequence into relative time variables (e.g., the relative number of seconds with the start time of the preset evaluation period as zero), constructing a set of coordinate points for time-residual values. Subsequently, the edge computing server uses the least squares method or gradient descent algorithm to calculate the slope parameter of the best-fit line, which is the fitting slope.

[0088] The edge computing server compares the calculated fitting slope with the zero value to determine whether the reference residual sequence shows a continuous upward trend, providing quantitative criteria for subsequent baseline shift diagnosis.

[0089] S306. If the fitting slope is greater than zero, and all temperature residual values ​​in the reference residual sequence are less than or equal to the residual alarm threshold, then the baseline offset diagnosis step is triggered, specifically including:

[0090] Obtain multiple historical load currents corresponding to each temperature residual value within a preset evaluation period;

[0091] Calculate the square value of each historical load current, and construct a heating reference sequence from multiple square values ​​according to time sequence;

[0092] Calculate the correlation coefficient between the reference residual sequence and the pyrogenic reference sequence;

[0093] If the correlation coefficient is greater than the preset correlation threshold, it is determined that there is an abnormal contact resistance inside the power equipment, and a fault upload command is generated.

[0094] If the correlation coefficient is less than or equal to the preset correlation threshold, it is determined that the power equipment's exterior is experiencing heat dissipation aging, and the normal regression prediction model undergoes model parameter calibration, specifically including:

[0095] Calculate the arithmetic mean of the individual temperature residuals contained in the reference residual sequence;

[0096] The calculated arithmetic mean is stored as a temperature compensation parameter in the normal regression prediction model.

[0097] Among them, the baseline offset diagnostic step sequence refers to a set of deep physical attribution analysis processes initiated by the edge computing server for special operating conditions where the temperature residual rises slowly over a long period of time but has not yet triggered an immediate alarm. It is used to distinguish between faulty heating (such as increased contact resistance) and non-faulty baseline drift (such as heat dissipation aging).

[0098] The preset correlation threshold is a critical value used to determine whether there is a significant causal relationship between the increase in temperature residual and the square of load current. Its setting scheme is usually based on the statistical analysis of a large number of historical failure cases, combined with the significance test of the correlation coefficient (such as the critical correlation coefficient corresponding to p<0.05), for example, it is set to 0.7.

[0099] Model parameter calibration refers to the process by which the edge computing server systematically corrects the deviations in the output of the normal regression prediction model. This is achieved by introducing temperature compensation parameters to adjust the model baseline, enabling the model to adapt to long-term changes in the heat dissipation conditions of the equipment.

[0100] This step is executed when the edge computing server detects that the reference residual sequence shows a slow upward trend (fitting slope is greater than zero), but all temperature residual values ​​do not exceed the residual alarm threshold, i.e., the device is in a "sub-healthy" state rather than an acute fault state.

[0101] Specifically, the edge computing server first extracts multiple historical load current values ​​from its local cache, each corresponding to a temperature residual value collected at a specific time within the preset evaluation period. Based on Joule's law (Q=I²Rt), the edge computing server squares each historical load current to obtain multiple squared values, and then organizes these squared values ​​into a heating reference sequence according to their original sampling time order. Subsequently, the edge computing server calls the statistical analysis module to calculate the correlation coefficient between the reference residual sequence and the heating reference sequence.

[0102] The correlation coefficient is used to represent the degree of linear correlation between the reference residual sequence and the pyrogenic reference sequence. It is usually calculated using the Pearson correlation coefficient or the Spearman rank correlation coefficient, with a value range of [-1, 1]. The closer the absolute value is to 1, the stronger the correlation.

[0103] The edge computing server compares the calculated correlation coefficient with a preset association threshold:

[0104] If the correlation coefficient is greater than the preset correlation threshold, it indicates that the increase in temperature residual is strongly positively correlated with the square of the load current. According to the physical mechanism of Joule's law, the edge computing server determines that there is an abnormal contact resistance inside the power equipment. This abnormality will amplify the heating effect as the current increases, and has the potential risk of deteriorating into thermal breakdown. Therefore, the edge computing server immediately generates a fault upload command and starts the high-frequency alarm reporting process.

[0105] If the correlation coefficient is less than or equal to the preset correlation threshold, it indicates that the increase in temperature residual is not significantly related to the change in load current. The edge computing server determines that there is heat dissipation aging on the surface of the power equipment. This phenomenon causes the actual temperature of the equipment to be systematically higher under various loads, but it does not belong to the hardware failure that requires emergency handling.

[0106] In response to thermal aging, the edge computing server initiates a model parameter calibration operation: it calculates the arithmetic mean of each temperature residual value contained in the reference residual sequence, which represents the average temperature rise caused by thermal aging; the edge computing server uses this arithmetic mean as a temperature compensation parameter and writes it into the parameter storage area of ​​the normal regression prediction model, so that the model automatically adds this compensation amount when subsequently calculating the theoretical temperature value, thereby using the new state after thermal aging as the updated health benchmark, avoiding frequent false alarms caused by natural aging, and realizing the adaptive evolution of the edge-side model.

[0107] S307. Within the preset judgment time window, if the change in real-time operating temperature is greater than the preset fluctuation threshold and the temperature residual value is less than or equal to the preset residual alarm threshold, then the data reporting frequency is maintained at the preset daily reporting frequency.

[0108] Among them, the preset fluctuation threshold is a critical value used to determine whether the real-time operating temperature fluctuation belongs to the normal operating condition change. The function of this parameter is to distinguish between reasonable temperature drastic fluctuations caused by sudden load changes or severe weather conditions (such as gusts and rain) and abnormal temperature rises caused by actual faults, so as to avoid false alarms triggered by normal operating condition fluctuations. Its setting scheme is usually based on the maximum allowable temperature change rate of power equipment under rated load change conditions, combined with the maximum temperature gradient that may be caused by meteorological changes (such as rain and temperature drop), and determined after leaving a certain margin.

[0109] After the edge computing server completes the temperature residual value calculation, it needs to make a reasonable judgment on drastic fluctuations in real-time operating temperature that occur in a short period of time, in order to avoid misjudging normal operating condition changes as faults when this step is performed. Specifically, the edge computing server initiates a preset judgment time window monitoring mechanism, extracting multiple real-time operating temperature values ​​continuously collected within this time window from the local cache unit. The edge computing server traverses this temperature sequence, identifies the maximum and minimum values, and calculates the difference between the two to obtain the magnitude of the real-time operating temperature change.

[0110] The edge computing server compares the magnitude of the change with a preset fluctuation threshold, and at the same time checks whether the temperature residual value within the judgment time window is less than or equal to the preset residual alarm threshold.

[0111] If the change exceeds the preset fluctuation threshold, it indicates a drastic fluctuation in the real-time operating temperature. However, if the temperature residual value remains within a safe range, it means the normal regression prediction model has successfully explained the fluctuation (i.e., the theoretical temperature value calculated by the model synchronously follows the actual temperature change). The edge computing server determines that the fluctuation is a normal operating condition response caused by predictable factors such as sudden load changes or drastic weather changes, rather than abnormal heating caused by internal equipment failure. In this case, the edge computing server maintains the data reporting frequency at the preset daily reporting frequency and does not trigger the alarm mechanism, thereby effectively avoiding false alarms and saving communication bandwidth resources.

[0112] S308. If the real-time operating temperature is greater than the preset static alarm threshold, a fault upload command will be generated.

[0113] Specifically, after obtaining the real-time operating temperature of the power equipment in step S301, if the real-time operating temperature is greater than the preset static alarm threshold, it means that the equipment has exceeded the upper limit of the maximum operating temperature allowed by the design. Regardless of the temperature residual value or other auxiliary criteria, the equipment is in a high-risk overheating state, and there is a serious risk of insulation breakdown, material melting or fire.

[0114] In this situation, the edge computing server immediately and unconditionally generates a fault upload command and initiates the highest priority emergency alarm process to ensure that the cloud device monitoring system and maintenance personnel can be informed of the dangerous status of the device as soon as possible and take emergency measures.

[0115] S309. After obtaining the temperature residual value, if the real-time operating temperature is less than or equal to the preset static alarm threshold, and the temperature residual value is always greater than the preset residual alarm threshold within the preset judgment time window, then a fault upload command is generated.

[0116] S310. In response to the fault upload command, the data reporting frequency for sending data to the equipment monitoring system is adjusted to the alarm reporting frequency, which is higher than the preset daily reporting frequency.

[0117] Steps S309 and S310 are similar to those described in steps S204 and S205 in the above embodiments, and will not be repeated here.

[0118] In actual deployment, the edge computing server's logical judgment engine contains multiple parallel diagnostic threads. The absolute high-temperature fuse protection in step S308 has the highest interrupt priority; steps S304 to S306 constitute a low-frequency, long-term sub-health diagnostic task; while steps S307 and S309 constitute a real-time residual diagnostic task for the current sampling period. Each task, upon meeting its respective triggering conditions, independently outputs a corresponding fault upload command or calibration operation to the event bus.

[0119] S311. In response to the fault upload command, the system performs a retrospective comparison of multiple consecutive historical temperature residual values ​​in the reference residual sequence in reverse chronological order to identify the inflection point before the historical temperature residual value changes from a stable fluctuation state to a continuous upward trend.

[0120] Among them, the stable fluctuation state refers to the operating state in which the temperature residual value fluctuates slightly and randomly around a certain benchmark level without obvious unidirectional trend. It is usually manifested as the first difference of the residual value (the difference between two adjacent points) being randomly distributed around zero. For example, the residual value fluctuates irregularly between 2℃ and 4℃.

[0121] This step is executed when the edge computing server receives a fault upload command and needs to accurately pinpoint the start time of the fault evolution in order to provide the device monitoring system with complete data on the entire fault development process. Specifically, in response to the fault upload command, the edge computing server initiates the historical data intelligent backtracking analysis module. The edge computing server retrieves a reference residual sequence from its local cache unit and, starting from the current moment when the fault upload command was generated, extracts multiple consecutive historical temperature residual values ​​from the sequence in reverse chronological order.

[0122] Edge computing servers perform retrospective comparison and analysis on these historical temperature residual values, specifically using a sliding window difference algorithm or a change point detection algorithm: calculating the first-order difference between adjacent historical temperature residual values ​​and determining the sign and magnitude of the difference; or performing linear fitting on the residual values ​​within a local time period and calculating the local fitting slope.

[0123] Edge computing servers identify the turning point where the first-order difference changes from random alternating positive and negative values ​​to continuous positive values, or where the local fitting slope changes from near zero to significantly greater than zero, thus pinpointing the precursor inflection point where historical temperature residual values ​​transition from a stable fluctuation state to a continuous upward trend. This precursor inflection point represents the time node where historical temperature residual values ​​change from a stable fluctuation state to a continuous upward trend. This point marks the beginning of the fault evolution process, signifying the transition of the equipment from normal operation to the nascent stage. The timestamp corresponding to this precursor inflection point is marked by the edge computing server as the starting point of the fault evolution.

[0124] S312. Take the timestamp corresponding to the previous inflection point as the start time and the current time when the fault upload command is generated as the end time.

[0125] This step is performed after the edge computing server successfully identifies the initial inflection point, when it is necessary to clearly define the time range for extracting data throughout the entire fault process. Specifically, the edge computing server extracts the timestamp corresponding to the inflection point from the initial inflection point identification result and stores and marks it as the start time. Simultaneously, the edge computing server obtains the current system timestamp that generated the fault upload command and records it as the end time. The edge computing server determines the length of the dynamic time period by calculating the time difference between the end time and the start time. This dynamic time period precisely covers the entire time span from the onset of abnormal symptoms in the device (initial inflection point) to the clear detection of the fault and the triggering of an alarm (current time). Compared to traditional fixed preset historical time periods, the dynamic time period can adaptively adjust according to the actual evolution speed of different faults, avoiding bandwidth waste caused by extracting too much irrelevant data while ensuring that no key fault evolution characteristics are missed.

[0126] S313. Extract the running data and real-time data within the dynamic time period consisting of the start time and the end time, encapsulate them, and send them to the equipment monitoring system according to the alarm reporting frequency.

[0127] After the edge computing server has determined the start and end times of the dynamic time period, this step is performed when it is necessary to report complete data of the entire fault process to the device monitoring system. Specifically, based on the dynamic time period formed by the start and end times, the edge computing server extracts all operational data within that time range from the local cache unit, including a complete multi-dimensional time-series dataset such as real-time operating temperature, load current, multi-dimensional meteorological parameters, theoretical temperature values, and temperature residual values.

[0128] The edge computing server integrates the extracted operational data within dynamic time periods with real-time data continuously collected according to the alarm reporting frequency after generating fault upload commands. The edge computing server calls the data encapsulation module to add metadata such as timestamps (identifying the precise time of data collection), data type identifiers (distinguishing between historical backtracking data and real-time alarm data), and device identifiers (indicating the power equipment number from which the data originates) to historical operational data and real-time data. It then encapsulates the data frames according to the communication protocol format required by the equipment monitoring system (such as MQTT, CoAP, or a custom protocol) to form structured data packets.

[0129] After encapsulation, the edge computing server, through the communication module, sends the encapsulated data packet to the device monitoring system with high priority via 4G / 5G or other wireless communication links, according to the adjusted alarm reporting frequency.

[0130] It should be noted that historical continuous operation data only needs to be packaged and uploaded once (or in sub-packages) as a traceability archive, while subsequent increases in frequency (alarm reporting frequency) only send real-time data.

[0131] In this embodiment, the baseline offset diagnostic step is triggered by calculating the fitting slope of the reference residual sequence over time and calculating the correlation coefficient between it and the heating reference sequence, thereby determining whether there is an abnormal contact resistance or performing model parameter calibration on the normal state regression prediction model. Therefore, it can capture early subtle signs of equipment deterioration and automatically identify the physical root cause of temperature rise, solving the problem of frequent false alarms due to difficulty in distinguishing the nature of the fault, and thus realizing the automated classification of fault root causes and predictive maintenance of long-term gradual hidden dangers.

[0132] The edge computing server in the embodiments of this invention is described below from the perspective of hardware processing. Please refer to [link / reference needed]. Figure 4 This is a schematic diagram of the physical device structure of an edge computing server in an embodiment of this application.

[0133] It should be noted that, Figure 4 The structure of the edge computing server shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0134] like Figure 4 As shown, the edge computing server includes a CPU 401, which can perform various appropriate actions and processes based on a program stored in the read-only memory ROM 402 or a program loaded from the storage section 408 into the random access memory RAM 403, such as performing the methods described in the above embodiments. The RAM 403 also stores various programs and data required for system operation. The CPU 401, ROM 402, and RAM 403 are interconnected via a bus 404. An I / O interface 405 is also connected to the bus 404.

[0135] The following components are connected to I / O interface 405: input section 406 including audio input devices, push-button switches, etc.; output section 407 including a liquid crystal display (LCD) and audio output devices, indicator lights, etc.; storage section 408 including a hard disk, etc.; and communication section 409 including a network interface card such as a LAN (Local Area Network) card, modem, etc. Communication section 409 performs communication processing via a network such as the Internet. Drive 410 is also connected to I / O interface 405 as needed. Removable media 411, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 410 as needed so that computer programs read from them can be installed into storage section 408 as needed.

[0136] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program / instructions carried on a computer-readable medium, the computer program / instructions containing computer program / instructions for performing the methods shown in the flowcharts. In such embodiments, the computer program / instructions can be downloaded and installed from a network via communication section 409, and / or installed from removable medium 411. When the computer program is executed by CPU 401, it performs the various functions defined in the present invention.

[0137] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0138] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. Each block in a flowchart or block diagram may represent a module, program segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those shown in the drawings.

[0139] Specifically, the edge computing server in this embodiment includes a processor and a memory. The memory stores a computer program, and when the computer program is executed by the processor, it implements a remote monitoring method for power equipment provided in the above embodiment.

[0140] In another aspect, the present invention also provides a computer-readable storage medium, which may be included in the edge computing server described in the above embodiments; or it may exist independently and not incorporated into the edge computing server. The storage medium carries one or more computer programs that, when executed by a processor of the edge computing server, cause the edge computing server to implement a remote monitoring method for power equipment provided in the above embodiments.

[0141] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0142] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as meaning "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as meaning "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

[0143] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for remote monitoring of power equipment, characterized in that, An edge computing server applied to a device monitoring system, the method comprising: Simultaneously acquire real-time operating temperature, load current, and multi-dimensional meteorological parameters of the environment in which the power equipment is located; The load current and the multidimensional meteorological parameters are input into a preset normal state regression prediction model, and the theoretical temperature value of the power equipment under the current operating conditions is output. Calculate the difference between the real-time operating temperature and the theoretical temperature value to obtain the temperature residual value; If the real-time operating temperature is less than or equal to the preset static alarm threshold, and the temperature residual value is always greater than the preset residual alarm threshold within the preset judgment time window, then a fault upload command is generated. In response to the fault upload command, the data reporting frequency for sending data to the device monitoring system is adjusted to the alarm reporting frequency, which is higher than the preset daily reporting frequency. The running data within a preset historical period prior to generating the fault upload command is encapsulated with the currently collected real-time data and sent to the device monitoring system according to the alarm reporting frequency.

2. The method according to claim 1, characterized in that, After the step of calculating the difference between the real-time operating temperature and the theoretical temperature value to obtain the temperature residual value, the method further includes: If, within a preset judgment time window, the change in the real-time operating temperature exceeds a preset fluctuation threshold, and the temperature residual value is less than or equal to a preset residual alarm threshold, then the data reporting frequency is maintained at the preset daily reporting frequency.

3. The method according to claim 1, characterized in that, After the step of calculating the difference between the real-time operating temperature and the theoretical temperature value to obtain the temperature residual value, the method further includes: If the real-time operating temperature is greater than the preset static alarm threshold, a fault upload command is generated.

4. The method according to claim 1, characterized in that, The method further includes: Within a preset evaluation period, multiple temperature residual values ​​obtained through continuous calculation are used to form a reference residual sequence, wherein the time span of the preset evaluation period is greater than the preset judgment time window. Calculate the fitting slope of the reference residual sequence as a function of time; If the fitting slope is greater than zero, and each temperature residual value in the reference residual sequence is less than or equal to the residual alarm threshold, then the baseline offset diagnostic step is triggered.

5. The method according to claim 4, characterized in that, The step of encapsulating the operational data from the preset historical period prior to generating the fault upload instruction with the currently collected real-time data specifically includes: In response to the fault upload command, multiple consecutive historical temperature residual values ​​in the reference residual sequence are compared in reverse chronological order to identify the inflection point where the historical temperature residual value changes from a stable fluctuation state to a continuous upward trend. The timestamp corresponding to the preceding inflection point is taken as the start time, and the current time when the fault upload instruction is generated is taken as the end time. The running data and real-time data within the dynamic time period formed by the start time and the end time are extracted and encapsulated.

6. The method according to claim 4, characterized in that, The triggering baseline offset diagnostic steps specifically include: Obtain multiple historical load currents corresponding to each of the temperature residual values ​​within the preset evaluation period; Calculate the square value of each of the historical load currents, and construct a heating reference sequence from multiple square values ​​according to time sequence; Calculate the correlation coefficient between the reference residual sequence and the pyrogenic reference sequence; If the correlation coefficient is greater than the preset correlation threshold, it is determined that there is an abnormal contact resistance inside the power equipment, and the fault upload instruction is generated. If the correlation coefficient is less than or equal to the preset correlation threshold, it is determined that the power equipment has heat dissipation aging on its surface, and the model parameters of the normal state regression prediction model are calibrated.

7. The method according to claim 6, characterized in that, The model parameter calibration operation for the normal regression prediction model specifically includes: Calculate the arithmetic mean of the individual temperature residuals contained in the reference residual sequence; The calculated arithmetic mean is stored as a temperature compensation parameter in the normal state regression prediction model.

8. An edge computing server for an equipment monitoring system, characterized in that, The edge computing server includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the edge computing server to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium storing computer instructions, characterized in that, When the computer instructions are executed on the edge computing server, the edge computing server causes the edge computing server to perform the method as described in any one of claims 1-7.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are run on the edge computing server, the edge computing server performs the method as described in any one of claims 1-7.

Images