Photovoltaic wind power equipment intelligent early warning management method based on edge calculation and equipment thereof

Abstract

The invention provides a photovoltaic wind power equipment intelligent early warning management method based on edge calculation and related equipment thereof, and the method comprises the steps: collecting the operation parameter data of target equipment in real time according to a preset collection frequency; preprocessing the operation parameter data to obtain target operation parameter data; based on a multi-dimensional collaborative analysis algorithm, calculating the target operation parameter data to obtain abnormal data corresponding to the target equipment; and generating corresponding early warning information based on a preset strategy and the abnormal data, and executing corresponding protection measures. Through data preprocessing and analysis on the edge computing node, real-time monitoring and quick response of the running state of the photovoltaic and wind power equipment can be realized, data transmission delay and central node computing load are reduced, and the accuracy of anomaly recognition and the early warning response speed are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent early warning, and in particular to an intelligent early warning management method, device, electronic device and storage medium for photovoltaic and wind power equipment based on edge computing. Background Technology

[0002] With the rapid development of new energy technologies, photovoltaic (PV) and wind power generation systems have become important components of the current energy structure. Due to the large number of PV modules and wind turbines, their wide distribution, and complex operating environments, their operational status is affected by multiple factors such as sunlight, wind speed, temperature, and humidity. Real-time monitoring and health management of equipment operating conditions have become crucial for ensuring power generation efficiency and equipment safety. In recent years, edge computing technology has been introduced into energy operation and maintenance scenarios. By executing computing tasks at edge nodes close to the data source, localized data analysis and response decisions are achieved, effectively reducing data transmission latency and the computing load on central nodes, providing a new technological path for intelligent operation and maintenance of PV and wind power equipment.

[0003] Existing early warning methods mostly only perform simple data collection and threshold judgment, failing to achieve deep feature extraction and intelligent computing at the edge, resulting in anomaly identification relying on the central server and experiencing response delays. Most methods use fixed thresholds or static models, unable to self-calibrate parameters based on historical operating status, environmental changes, and equipment aging, leading to frequent false alarms and missed alarms. Existing technologies often monitor individual devices (such as a single photovoltaic panel or wind turbine) independently, without establishing collaborative relationships between devices. For example, a failure in one module of a photovoltaic array may cause current overload in adjacent modules, and a grid connection failure in a wind turbine may cause voltage fluctuations throughout the wind farm, but existing solutions cannot identify this "fault propagation" trend. Furthermore, they fail to integrate external environmental data (such as sunlight, wind speed, and precipitation) with equipment operating parameters for comprehensive analysis. For instance, under heavy rain conditions, settlement of the wind turbine tower foundation may cause vertical deviation of the unit, but traditional solutions only monitor internal unit parameters and cannot combine precipitation data for early warning.

[0004] This "isolated equipment monitoring" model cannot effectively cope with the cascading failures of multiple equipment in photovoltaic and wind power plants. In some cases, the failure to detect overload of adjacent components has led to the spread of the fault from a single component to multiple components, resulting in long-term shutdowns and huge economic losses.

[0005] Therefore, existing intelligent early warning management methods for photovoltaic and wind power equipment based on edge computing have the problem of being unable to achieve multi-dimensional data fusion and cross-device collaborative analysis, and are difficult to identify and block cascading failures in a timely manner. Summary of the Invention

[0006] This invention provides an intelligent early warning management method for photovoltaic and wind power equipment based on edge computing, in order to solve the problems of existing intelligent early warning management methods for photovoltaic and wind power equipment based on edge computing, which cannot eliminate baseline drift and background fluctuations and have difficulty in accurately identifying the real gas anomaly waveforms formed by tunneling disturbances.

[0007] In a first aspect, the present invention provides an intelligent early warning management method for photovoltaic and wind power equipment based on edge computing, the method comprising the following steps: The operating parameter data of the target device are collected in real time according to the preset collection frequency. The operating parameter data is preprocessed to obtain the target operating parameter data; Based on a multi-dimensional collaborative analysis algorithm, the target operating parameter data is calculated to obtain the abnormal data corresponding to the target device; Based on the preset strategy and the abnormal data, corresponding early warning information is generated, and corresponding protection measures are implemented.

[0008] Optionally, the step of collecting the target device's operating parameter data in real time according to a preset collection frequency includes: The preset acquisition frequency is determined based on the operating efficiency of the target device; Based on the preset acquisition frequency, the photovoltaic device's voltage data, current data, surface temperature data, and inverter output power are acquired through the photovoltaic device's sensors. Based on the preset acquisition frequency, lubrication data, wear data, and pitch data of the rotor equipment are collected through wind power equipment sensors. Based on the preset acquisition frequency, environmental sensors are used to collect data on light intensity, wind speed, wind direction, and precipitation in the environment where the target device is located.

[0009] Optionally, the preprocessing of the operating parameter data to obtain the target operating parameter data includes: The operating parameter data is repeatedly filtered to obtain the first processed data; The first processed data is subjected to anomaly removal processing to obtain the second processed data; Feature extraction is performed on the second processed data to generate the target operating parameter data.

[0010] Optionally, the step of calculating the target operating parameter data based on a multi-dimensional collaborative analysis algorithm to obtain the abnormal data corresponding to the target device includes: Based on the correlation between the target operating parameter data, a feature model is established, which is used to reflect the characteristics of the target equipment's operating status. Based on the feature model, the target operating parameter data are calculated to obtain the feature deviation values ​​of the target device under each operating state; The characteristic deviation value is compared with the preset deviation threshold to determine the abnormal data of the target device under each operating state.

[0011] Optionally, the step of calculating the target operating parameter data based on the feature model to obtain the feature deviation value of the target device under each operating state includes: Obtain historical operating baseline data of the target device under various operating states; The target operating parameter data is compared with the historical operating baseline data to determine the feature difference value; Based on the aforementioned feature difference values, the feature deviation values ​​of the target device under each operating state are calculated.

[0012] Optionally, the abnormal data includes the number of anomalies, the magnitude of the deviation, and the duration. The step of generating corresponding early warning information and executing corresponding protective measures based on a preset strategy and the abnormal data includes: Based on the number of anomalies, the magnitude of the deviation, and the duration, the corresponding warning level is determined; Based on the aforementioned warning level, a warning message is generated that includes the device identifier, fault type, warning level, and maintenance recommendations. Based on the warning information, corresponding protection measures are implemented, including shutdown protection and speed reduction protection.

[0013] Optionally, after generating corresponding early warning information based on the preset strategy and the abnormal data, and executing corresponding protective measures, the method further includes: Obtain abnormal data samples of the target device during its historical operating cycle and their corresponding early warning information; Based on the abnormal data samples and early warning information, the feature model is incrementally trained, and the parameters and deviation threshold of the feature model are updated to obtain the updated feature model. When the updated feature model meets the preset convergence condition during training, the current feature model is replaced, and the updated feature model is used for the calculation of subsequent input target running parameter data.

[0014] Secondly, the present invention also provides an intelligent early warning management device for photovoltaic and wind power equipment based on edge computing, the intelligent early warning management device for photovoltaic and wind power equipment based on edge computing includes: The first acquisition module is used to acquire the operating parameter data of the target device in real time according to the preset acquisition frequency; The first preprocessing module is used to preprocess the operating parameter data to obtain the target operating parameter data; The first calculation module is used to calculate the target operating parameter data based on a multi-dimensional collaborative analysis algorithm to obtain the abnormal data corresponding to the target device; The first generation and execution module is used to generate corresponding early warning information and execute corresponding protection measures based on the preset strategy and the abnormal data.

[0015] Thirdly, the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in the intelligent early warning management method for photovoltaic and wind power equipment based on edge computing provided by the present invention.

[0016] Fourthly, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in the intelligent early warning management method for photovoltaic and wind power equipment based on edge computing provided by the invention.

[0017] This invention collects operational parameter data of target equipment in real time according to a preset acquisition frequency; preprocesses the operational parameter data to obtain target operational parameter data; calculates the target operational parameter data based on a multi-dimensional collaborative analysis algorithm to obtain abnormal data corresponding to the target equipment; and generates corresponding early warning information and executes corresponding protective measures based on a preset strategy and the abnormal data. By performing data preprocessing and analysis on edge computing nodes, real-time monitoring and rapid response to the operating status of photovoltaic and wind power equipment can be achieved, reducing data transmission latency and the computing load on central nodes, and improving the accuracy of anomaly identification and the speed of early warning response. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of an intelligent early warning management method for photovoltaic and wind power equipment based on edge computing, provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of another intelligent early warning management device for photovoltaic and wind power equipment based on edge computing provided in an embodiment of the present invention; Figure 3This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] 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.

[0021] like Figure 1 As shown, Figure 1 This is a flowchart of an intelligent early warning management method for photovoltaic and wind power equipment based on edge computing, provided by an embodiment of the present invention. The method includes the following steps: 101. Collect the operating parameter data of the target device in real time according to the preset acquisition frequency.

[0022] In this embodiment of the invention, the above-mentioned intelligent early warning management method for photovoltaic and wind power equipment based on edge computing can be applied to an intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing. The intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing has functions such as early warning data processing, early warning data transmission and reception, and early warning data memory storage. It can be built based on a server or server cluster. The server or server cluster can be an electronic device with early warning data processing capabilities.

[0023] The aforementioned preset acquisition frequency can be a data acquisition cycle predefined by the edge computing-based intelligent early warning management platform for photovoltaic and wind power equipment based on the target equipment's operating characteristics, load status, and environmental change patterns. The target equipment may include, but is not limited to, photovoltaic equipment, wind power equipment, and environmental sensors. These target devices are interconnected through edge nodes, supporting cross-type data coupling analysis such as "photovoltaic-photovoltaic," "wind turbine-wind turbine," and "photovoltaic-wind turbine."

[0024] For example, the sampling frequency of photovoltaic equipment is set to 10 seconds per sampling, used to collect data on voltage, current, module surface temperature, and inverter output power; the sampling frequency of wind power equipment is set to 5 seconds per sampling, used to collect data on lubricating oil temperature, vibration acceleration, pitch angle, and rotational speed; and the sampling frequency of environmental sensors is set to 30 seconds per sampling, used to collect data on light intensity, wind speed, wind direction, humidity, and precipitation. It is understood that the above preset sampling frequencies can be dynamically adjusted according to the equipment's operating efficiency. When the platform detects that the equipment's operation deviates from the baseline state, it automatically shortens the sampling period to improve the sensitivity of abnormal response.

[0025] The aforementioned operating parameters can be multi-source real-time data collected from the target equipment, such as voltage, current, power, oil temperature, vibration frequency, wind speed, light intensity, and humidity. For example, in a wind farm, data sources include tower vibration sensors, blade angle sensors, and generator current detectors; in a photovoltaic array, data sources include module temperature sensors and inverter monitoring modules.

[0026] 102. Preprocess the operating parameter data to obtain the target operating parameter data.

[0027] In this embodiment of the invention, the above-mentioned intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing can clean, filter and normalize the raw data at the edge computing node. Specifically, it can perform steps such as duplicate data filtering and anomaly removal (removing sensor jitter data), signal smoothing and outlier correction, unit conversion and time sequence alignment, data standardization and feature extraction (such as extracting feature indicators such as power fluctuation and spectrum amplitude), thereby reducing data redundancy and noise, and making subsequent algorithm analysis more efficient and reliable.

[0028] The aforementioned target operating parameter data can refer to a standardized input dataset obtained after preprocessing, possessing a unified timestamp, unified sampling dimension, and stable feature distribution. In multi-device scenarios, target operating parameter data from different devices can be synchronized in time and mapped in features through edge nodes, thereby forming a joint data structure that supports cross-device analysis.

[0029] 103. Based on a multi-dimensional collaborative analysis algorithm, the target operating parameter data is calculated to obtain the abnormal data corresponding to the target device.

[0030] In this embodiment of the invention, the aforementioned multi-dimensional collaborative analysis algorithm can employ a multimodal feature fusion network or an LSTM-based time-series analysis model to extract the dynamic coupling relationships between the operating parameters of various devices. This is used to construct feature models between devices and between devices and the environment, reflecting the corresponding correlations. For example, when the current of a photovoltaic module decreases, the aforementioned multi-dimensional collaborative analysis algorithm can analyze the load changes of neighboring modules and the output changes of the inverter; in a wind farm, it can calculate indicators such as wind speed gradient and power phase difference between wind turbines to identify potential cascading fault propagation trends. It is understood that the aforementioned multi-dimensional collaborative analysis algorithm can comprehensively consider temporal, spatial, and physical correlation characteristics to achieve joint modeling of multi-source data.

[0031] In one possible embodiment, the aforementioned edge computing-based intelligent early warning management platform for photovoltaic and wind power equipment can input target operating parameter data into the aforementioned multi-dimensional collaborative analysis algorithm to perform operations such as feature extraction, deviation assessment, and threshold comparison, thereby calculating the characteristic deviation values ​​of each target device under different operating states, comparing them with preset deviation thresholds, and outputting the abnormal probability or risk level.

[0032] The aforementioned abnormal data may include, but is not limited to, the characteristic deviation values, abnormal duration, deviation magnitude, and impact range of the target equipment under various operating states. For example, when the current of a component in the photovoltaic array deviates from the average value by more than 15% and continues to exceed the sampling period threshold, the system marks the component as abnormal and analyzes its impact on adjacent components, forming an abnormal diffusion chain data.

[0033] 104. Based on preset strategies and abnormal data, generate corresponding early warning information and execute corresponding protection measures.

[0034] In this embodiment of the invention, the aforementioned preset strategy can be used to define the response rules of the edge computing-based intelligent early warning management platform for photovoltaic and wind power equipment under different anomaly levels. For example, when the number of anomalies exceeds a set threshold, the deviation is large, or the duration is too long, the edge computing-based intelligent early warning management platform for photovoltaic and wind power equipment triggers an early warning of the corresponding level according to the strategy. The aforementioned preset strategy includes a priority allocation mechanism to ensure that critical equipment (such as the main inverter and main wind turbine) responds first in the event of anomalies.

[0035] In this embodiment, the aforementioned intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing can automatically generate structured early warning information messages according to the abnormal data and policy rules. Specifically, through operations such as data summary generation, early warning level determination, equipment identification association, and push channel selection, early warning information can be initially generated at the edge node and simultaneously uploaded to the cloud platform for fusion analysis.

[0036] The aforementioned warning information may include, but is not limited to, fields such as equipment identification, fault type, abnormal parameters, warning level, recommended measures, and timestamp. For example: "Equipment ID: PV-12, Abnormal type: Component current too low, Level: Level II, Recommendation: Check the busbar and detect hot spots." It can be understood that for wind turbine abnormalities, the warning may indicate "Blade pitch failure, servo angle adjustment or blade reset required."

[0037] The aforementioned protective measures may include, but are not limited to, automatic shutdown, reduced-speed operation, branch isolation, voltage balance adjustment, and wind turbine power reduction. For example, when excessive vibration of the wind turbine tower is detected and the wind speed continues to rise, the aforementioned edge computing-based intelligent early warning management platform for photovoltaic and wind power equipment executes "reduction protection"; when overheating of the photovoltaic inverter is detected, "shutdown protection" is executed and a maintenance work order is issued. This embodiment can also implement "anti-spreading" measures in advance on affected adjacent equipment through an inter-equipment coordination mechanism, such as pre-adjusting power distribution and reducing grid connection voltage, to avoid the expansion of cascading faults.

[0038] In one possible embodiment, the aforementioned edge computing-based intelligent early warning management platform for photovoltaic and wind power equipment acquires operating parameter data such as voltage, current, temperature, and wind speed according to a preset acquisition frequency. After preprocessing by edge nodes, it forms unified target operating parameter data. Through multi-dimensional collaborative analysis algorithms, it calculates the correlation characteristics between equipment and equipment and between equipment and environment, identifies abnormal data and potential fault propagation trends, generates graded early warning information according to preset strategies, and executes protective measures such as shutdown, speed reduction, or linkage to prevent propagation.

[0039] By using the above methods and steps, cross-domain collaborative monitoring and dynamic risk warning of photovoltaic and wind power equipment can be achieved at the edge, improving the accuracy and response speed of cascading failures and reducing operation and maintenance costs.

[0040] In this embodiment of the invention, operating parameter data of the target device is collected in real time according to a preset acquisition frequency; the operating parameter data is preprocessed to obtain target operating parameter data; based on a multi-dimensional collaborative analysis algorithm, the target operating parameter data is calculated to obtain abnormal data corresponding to the target device; based on a preset strategy and the abnormal data, corresponding early warning information is generated, and corresponding protection measures are executed. By performing data preprocessing and analysis on edge computing nodes, real-time monitoring and rapid response to the operating status of photovoltaic and wind power equipment can be achieved, reducing data transmission latency and the computing load of the central node, and improving the accuracy of anomaly identification and the speed of early warning response.

[0041] Optionally, in the step of collecting the operating parameter data of the target equipment in real time according to the preset acquisition frequency, the preset acquisition frequency can also be determined based on the operating efficiency of the target equipment; based on the preset acquisition frequency, the voltage data, current data, surface temperature data and inverter output power of the photovoltaic equipment can be collected through the photovoltaic equipment sensors; based on the preset acquisition frequency, the lubrication data, wear data and pitch data of the rotor equipment can be collected through the wind power equipment sensors; based on the preset acquisition frequency, the light intensity, wind speed, wind direction and precipitation data of the environment where the target equipment is located can be collected through the environmental sensors.

[0042] In this embodiment of the invention, the above-mentioned intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing determines the sampling frequency based on the operating efficiency of the target equipment. For example, when fluctuations in the power generation efficiency of the photovoltaic array or deviations in the power of the wind turbine are detected, the sampling frequency is automatically increased to obtain data with higher timeliness. When the target equipment is operating stably, the sampling frequency is reduced to reduce data redundancy and energy consumption.

[0043] Specifically, layered data acquisition can be achieved using different types of sensors: Photovoltaic equipment sensors: used to collect real-time data on module voltage, current, surface temperature, and inverter output power to comprehensively reflect the power generation efficiency and thermal stability of a single photovoltaic panel and its branches; Wind power equipment sensors: used to collect data on rotor equipment lubricating oil temperature, bearing wear condition and pitch angle to assess the mechanical health and power output stability of the wind turbine; Environmental sensors: used to collect information on light intensity, wind speed, wind direction and precipitation, and synchronize them with equipment operation data in time, providing support for subsequent equipment-environment correlation analysis.

[0044] In edge computing nodes, by collecting data from the above multi-dimensional sources, synchronous acquisition and unified time-series calibration of heterogeneous data from multiple sources are achieved, forming a real-time dataset covering three-dimensional elements of electrical, mechanical and meteorological aspects.

[0045] By employing the above methods and steps, the sampling density can be increased when the equipment load is abnormal or the environment changes abruptly, thereby capturing potential fault signals in advance. During the stable operation phase of the equipment, the sampling frequency can be reduced, reducing network transmission and storage pressure, improving the real-time performance and energy efficiency of monitoring, providing a high-quality data foundation for subsequent feature extraction and anomaly identification, and enhancing the response speed and reliability of intelligent early warning for photovoltaic and wind power equipment.

[0046] Optionally, the step of preprocessing the operating parameter data to obtain the target operating parameter data may further include performing repeated filtering on the operating parameter data to obtain first processed data; performing anomaly removal on the first processed data to obtain second processed data; and performing feature extraction on the second processed data to generate the target operating parameter data.

[0047] In this embodiment of the invention, the aforementioned intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing can repeatedly filter operating parameter data according to timestamps and sampling sequences. Specifically, when a sensor generates multiple repetitive data frames with extremely short time intervals under high-frequency sampling, the aforementioned intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing retains only one valid sampled data, deleting repetitive frames and invalid values ​​generated by the sensor offline, thereby obtaining the first processed data. For example, when a photovoltaic module voltage sensor samples 10 times per second, only one valid reading is retained, and the remaining 9 are automatically discarded, reducing communication and storage burden.

[0048] The second processed data can be obtained by using the above-mentioned edge computing-based photovoltaic and wind power equipment intelligent early warning management platform to perform anomaly removal based on sliding window and median filtering algorithm on the first processed data, so as to identify and delete instantaneous abnormal values ​​caused by external interference, including but not limited to voltage spikes caused by lightning strikes and electromagnetic interference, and short-term temperature fluctuations caused by wind and sand blockage.

[0049] In this embodiment, the aforementioned intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing can generate target operating parameter data by performing multi-dimensional feature extraction on the second processed data at the edge computing node. Specifically, it can extract feature parameters including but not limited to the current fluctuation amplitude of photovoltaic equipment, the component temperature rise rate, the vibration spectrum peak value of wind turbine unit, the pitch angle change rate, and environmental parameters such as wind speed gradient and light intensity change rate. These parameters can be used to reflect the operating health status of a single device and can also provide input support for subsequent collaborative analysis between "device-device" and "device-environment".

[0050] Optionally, in the step of calculating the target operating parameter data based on the multi-dimensional collaborative analysis algorithm to obtain the abnormal data corresponding to the target device, the method further includes establishing a feature model based on the correlation between the target operating parameter data. The feature model is used to reflect the characteristics of the target device's operating state. Based on the feature model, the target operating parameter data is calculated to obtain the feature deviation value of the target device in each operating state. The feature deviation value is compared with a preset deviation threshold to determine the abnormal data of the target device in each operating state.

[0051] In this embodiment of the invention, the aforementioned correlation can refer to the mutual influence between the photovoltaic equipment, wind power equipment, and environmental factors at the operational parameter level. In this embodiment, the aforementioned intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing can determine the degree of dynamic coupling between variables by calculating the Pearson correlation coefficient and mutual information index of data such as voltage, current, temperature, wind speed, and vibration. For example, an increase in the surface temperature of a photovoltaic module is usually accompanied by a decrease in output current, while changes in wind speed may affect the wind turbine speed and output power.

[0052] In one possible embodiment, the aforementioned intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing can construct a mathematical model reflecting the operating status of the target equipment within edge computing nodes. Specifically, the feature model can be a fusion of random forest and Long Short-Term Memory (LSTM) network structures, capable of simultaneously capturing static features, such as power and temperature distribution, and dynamic temporal features, such as current variation trends. By inputting historical operating data and environmental data, the feature model can automatically learn multi-dimensional feature patterns under normal conditions, forming a set of benchmark feature expressions for assessing the health status of the equipment.

[0053] In another possible embodiment, the aforementioned edge computing-based intelligent early warning management platform for photovoltaic and wind power equipment uses the established feature model to perform inference calculations on the target operating parameter data collected in real time. Specifically, it can send the newly input data stream into the feature model, calculate the difference between the model output and the historical benchmark output, and obtain the degree of deviation of each parameter relative to the health state through matrix operations and time-series residual analysis, and finally obtain the feature deviation value.

[0054] The aforementioned characteristic deviation values ​​refer to the differences between the characteristic indicators of the target equipment under its current operating state and its normal characteristics. These characteristic deviation values ​​can be used to reflect the degree of abnormality in the operating state, such as photovoltaic module power fluctuations exceeding the normal value by 5%, or wind turbine vibration frequencies deviating from the average by 0.8Hz. The platform can combine multiple characteristic deviation values ​​into a deviation vector to simultaneously reflect the comprehensive health level of the equipment across its electrical, mechanical, and environmental dimensions.

[0055] The aforementioned preset deviation threshold can refer to a reference upper limit or range set based on historical data statistics and empirical parameters. In this embodiment, different types of equipment correspond to different thresholds; for example, the vibration threshold for wind turbine bearings is ±0.8 mm / s, and the deviation threshold for photovoltaic current is ±5%. It is understood that the aforementioned preset deviation threshold can be dynamically adjusted according to the equipment lifespan curve and seasonal environmental changes to ensure both sensitivity and stability in anomaly identification.

[0056] In another possible embodiment, the aforementioned edge computing-based intelligent early warning management platform for photovoltaic and wind power equipment can compare the aforementioned feature deviation values ​​with preset deviation thresholds item by item to determine whether the target equipment is abnormal. When the deviation value is below the threshold, the equipment is determined to be operating normally; when it approaches the threshold, it is marked as an early warning state; and when it exceeds the threshold, it is determined to be an abnormal state.

[0057] Through the aforementioned collaborative analysis mechanism, abnormal states of the target equipment can be identified, environmental disturbances can be distinguished from mechanical failures, and cascading failure trends can be predicted in advance through multi-dimensional feature deviation vectors.

[0058] Optionally, in the step of calculating the target operating parameter data based on the feature model to obtain the feature deviation value of the target device under each operating state, the method further includes obtaining the historical operating benchmark data of the target device under each operating state; performing feature comparison calculation between the target operating parameter data and the historical operating benchmark data to determine the feature difference value; and calculating the feature deviation value of the target device under each operating state based on the feature difference value.

[0059] In this embodiment of the invention, the aforementioned historical operating benchmark data may refer to the set of standard operating parameters accumulated by the aforementioned edge computing-based intelligent early warning management platform for photovoltaic and wind power equipment during the monitoring process for the aforementioned target equipment, such as various photovoltaic and wind power equipment under different operating conditions (e.g., rated power, half load, low wind speed, high irradiance), including but not limited to steady-state voltage, current, temperature, vibration, oil temperature and environmental coupling data, and after statistical screening and mean normalization processing, to represent the characteristic mode of the equipment in a healthy state.

[0060] In one possible embodiment, the aforementioned edge computing-based intelligent early warning management platform for photovoltaic and wind power equipment can compare the target operating parameter data collected in real time with the corresponding historical operating benchmark data on a dimension-by-dimensional basis. Specifically, by performing Euclidean distance calculation and cosine similarity evaluation on each feature dimension (such as current fluctuation rate, vibration frequency, temperature rise gradient, and wind speed change rate), the degree of deviation between the current operating state and the benchmark state can be determined.

[0061] For example, when the main peak of the wind turbine impeller vibration spectrum shifts and the oil temperature rises, the aforementioned intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing can identify it as an initial signal of mechanical wear through comparative calculation; while when the photovoltaic module temperature rises and the current drops simultaneously, it is identified as a potential hot spot trend.

[0062] The aforementioned feature difference values ​​refer to the differences between the current operating state and the historical baseline state of each feature parameter obtained through the above feature comparison calculations. Specifically, the aforementioned intelligent early warning management platform for photovoltaic and wind power equipment based on edge computing can represent the above difference results in vector form and assign weights according to feature dimensions, such as electrical parameters with a weight of 0.5, mechanical parameters with a weight of 0.3, and environmental parameters with a weight of 0.2, to comprehensively reflect the overall operational deviation of the equipment. It can be understood that the larger the aforementioned feature difference values, the more obvious the deviation of the equipment from normal operating characteristics, and the higher the potential risk level.

[0063] Optionally, in the step of generating corresponding warning information based on a preset policy and abnormal data and executing corresponding protection measures, it further includes determining a corresponding warning level based on the number of anomalies, deviation range, and duration; generating warning information including device identification, fault type, warning level, and maintenance suggestions based on the warning level; and executing corresponding protection measures based on the warning information, where the protection measures include shutdown protection and speed reduction protection.

[0064] In an embodiment of the present invention, the above abnormal data may include, but is not limited to, the number of anomalies, deviation range, and duration corresponding to the target device in the current operating state.

[0065] In a possible embodiment, when the intelligent warning management platform for photovoltaic and wind power equipment based on edge computing executes the step of generating warning information based on a preset policy and executing protection measures, it obtains abnormal data from the previous analysis stage and determines the warning level according to the indicators corresponding to the abnormal data. For example, when the number of anomalies ≥ 3, the deviation range exceeds 15%, and the duration exceeds 5 minutes, the system classifies it as a level I warning; When the number of anomalies is 1 - 2 and the deviation range is between 5% - 15%, it is classified as a level II warning; When the anomaly is only a short-term fluctuation or the deviation range < 5%, it is classified as a level III prompt warning.

[0066] Based on the determination result, the platform automatically generates structured warning information, the content of which includes device identification, fault type, warning level, risk cause, and maintenance suggestions. For example: "Device ID: WT - 05, Fault type: High vibration of the gearbox, Warning level: Level I, Suggestion: Immediately shut down and replace the lubricating oil filter element model XX." When the warning information is generated, the platform executes corresponding protection measures through the edge control interface: For a level I warning, trigger shutdown protection and directly issue power-off and emergency pitch commands through the control interface; For a level II warning, execute speed reduction protection and adjust the fan speed or cut the output power of the photovoltaic array; For a level III warning, only generate a visual prompt and synchronize it to the operation and maintenance personnel's terminal for observation.

[0067] When the platform executes protection actions, it can achieve a millisecond-level linkage reaction to ensure that key devices enter a safe state before the abnormal situation spreads. At the same time, the system uploads the warning information to the cloud log module for subsequent statistics and model re-training.

[0068] Optionally, after generating corresponding early warning information and implementing corresponding protection measures based on preset strategies and abnormal data, the steps further include obtaining abnormal data samples of the target device in the historical operating cycle and their corresponding early warning information; incrementally training the feature model based on the abnormal data samples and early warning information, and updating the parameters and deviation threshold of the feature model to obtain the updated feature model; when the updated feature model meets the preset convergence condition during the training process, replacing the current feature model, and using the updated feature model for subsequent input target operating parameter data calculation.

[0069] In this embodiment of the invention, the aforementioned abnormal data samples can be data segments with "abnormal tags" collected and organized within a selected time window before and after the occurrence of an early warning. This includes, but is not limited to, the original / preprocessed characteristics of the target equipment, such as voltage, current, vibration spectrum, temperature rise gradient, wind speed change rate, etc., and the corresponding early warning information, such as equipment ID, fault type, early warning level, handling result, and time consumption.

[0070] Understandably, to ensure the usability of the aforementioned abnormal data samples, deduplication, timestamp alignment, and missing value processing can be performed on these samples, and they can be managed by binning according to the scenario (sunny / cloudy, low / high wind, season). This can reflect both single-point failures and retain the correlation characteristics between devices and between devices and the environment, which can be used to learn "chain failure" patterns.

[0071] In this embodiment, the existing feature model can be fine-tuned and updated using newly added abnormal data samples without retraining the model. Specifically, the base layer can be frozen, and only the high-level time series head or classification head can be trained with a small learning rate. At the same time, replay samples can be introduced to avoid forgetting. During the training process, the deviation threshold can also be synchronously calibrated based on sliding window statistics or Bayesian updates, so that the model weights and decision boundaries converge together.

[0072] The aforementioned preset convergence conditions can be used to determine whether incremental training has reached a set of quantifiable thresholds that qualify the model for replacement and deployment. These include, but are not limited to, a stable decrease in validation set loss, key business metrics being better than the baseline, and threshold update magnitude and online metric fluctuations entering a safe range. If the conditions are met, the new model is considered stable and reliable, and can be released in a phased manner to gradually replace the old model; if the conditions are not met, fine-tuning continues or the model is rolled back to the optimal checkpoint.

[0073] In one possible embodiment, the aforementioned edge computing-based intelligent early warning management platform for photovoltaic and wind power equipment can collect multi-source parameters of photovoltaic, wind power, and environment in real time according to a preset collection frequency. After repeated filtering, anomaly removal, and feature extraction are completed at the edge, unified target operating parameter data is obtained. Then, based on a multi-dimensional collaborative analysis algorithm, a device-device / device-environment correlation feature model is constructed. The input data is calculated to obtain feature deviation values, which are compared with preset deviation thresholds to generate abnormal data. Subsequently, early warning information containing equipment identification, fault type, and maintenance suggestions is generated hierarchically according to a preset strategy, and protective measures such as shutdown or speed reduction are implemented in conjunction. Afterwards, incremental training is performed using abnormal data samples and early warning records to dynamically update the model and thresholds to form a self-learning closed loop.

[0074] By employing the above methods and steps, we can achieve low-latency, high-quality data acquisition, early identification of cross-device cascading faults, rapid linkage of graded handling, and continuous model adaptation, thereby improving early warning accuracy and response speed, reducing false alarms / missed alarms and operation and maintenance costs, and ensuring the safe and stable operation of the power plant.

[0075] like Figure 2 As shown, this embodiment of the invention also provides a photovoltaic and wind power equipment intelligent early warning management device 200 based on edge computing. This edge computing-based photovoltaic and wind power equipment intelligent early warning management device 200 includes: The first acquisition module 201 is used to acquire the operating parameter data of the target device in real time according to the preset acquisition frequency; The first preprocessing module 202 is used to preprocess the operating parameter data to obtain target operating parameter data; The first calculation module 203 is used to calculate the target operating parameter data based on a multi-dimensional collaborative analysis algorithm to obtain the abnormal data corresponding to the target device; The first generation and execution module 204 is used to generate corresponding early warning information and execute corresponding protection measures based on the preset strategy and the abnormal data.

[0076] Optionally, the first acquisition module 201 mentioned above includes: The first acquisition submodule is used to determine the preset acquisition frequency based on the operating efficiency of the target device; The second acquisition submodule is used to acquire voltage data, current data, surface temperature data, and inverter output power of the photovoltaic device through the photovoltaic device sensor based on the preset acquisition frequency. The third acquisition submodule is used to acquire lubrication data, wear data and pitch data of the rotor equipment through the wind power equipment sensors based on the preset acquisition frequency. The fourth acquisition submodule is used to acquire data on light intensity, wind speed, wind direction, and precipitation in the environment where the target device is located, based on the preset acquisition frequency and through environmental sensors.

[0077] Optionally, the first preprocessing module 202 mentioned above includes: The first processing submodule is used to repeatedly filter the running parameter data to obtain first processed data. The second processing submodule is used to perform anomaly removal processing on the first processed data to obtain the second processed data; The third processing submodule is used to extract features from the second processed data and generate the target operating parameter data.

[0078] Optionally, the first calculation module 203 mentioned above includes: The first submodule is used to establish a feature model based on the correlation between the target operating parameter data. The feature model is used to reflect the features of the target device's operating status. The first calculation submodule is used to calculate the target operating parameter data based on the feature model to obtain the feature deviation value of the target device under each operating state; The first determining submodule is used to compare the feature deviation value with the preset deviation threshold to determine the abnormal data of the target device under each operating state.

[0079] Optionally, the first calculation submodule mentioned above includes: The first acquisition unit is used to acquire historical operating baseline data of the target device under various operating states; The first calculation unit is used to perform feature comparison calculations between the target operating parameter data and the historical operating benchmark data to determine the feature difference value. The second calculation unit is used to calculate the characteristic deviation value of the target device under each operating state based on the characteristic difference value.

[0080] Optionally, the first calculation module 203 mentioned above includes: The second determining submodule is used to determine the corresponding warning level based on the number of anomalies, the magnitude of the deviation, and the duration. A generation submodule is used to generate warning information, including device identifier, fault type, warning level, and maintenance suggestions, based on the warning level. The execution submodule is used to execute corresponding protection measures based on the warning information, including shutdown protection and speed reduction protection.

[0081] Optionally, the above-mentioned device further includes: The second acquisition module is used to acquire abnormal data samples of the target device during its historical operating cycle and their corresponding early warning information. The training module is used to incrementally train the feature model based on the abnormal data samples and early warning information, and update the parameters and deviation threshold of the feature model to obtain the updated feature model. The update module is used to replace the current feature model when the updated feature model meets the preset convergence condition during training, and to use the updated feature model for the calculation of subsequent input target running parameter data.

[0082] like Figure 3 As shown, this embodiment of the invention also provides an electronic device 300, including a processor, which can execute any of the above-mentioned edge computing-based intelligent early warning management methods for photovoltaic and wind power equipment.

[0083] Specifically, it includes a processor 301 and a memory 302, as well as a computer program stored in the memory 302 and capable of running on the processor 301, which executes an edge computing-based intelligent early warning management method for photovoltaic and wind power equipment, wherein: The processor 301 executes the calculator program for the intelligent early warning management method of photovoltaic and wind power equipment based on edge computing, stored in the memory 302, and performs the following steps: The operating parameter data of the target device are collected in real time according to the preset collection frequency. The operating parameter data is preprocessed to obtain the target operating parameter data; Based on a multi-dimensional collaborative analysis algorithm, the target operating parameter data is calculated to obtain the abnormal data corresponding to the target device; Based on the preset strategy and the abnormal data, corresponding early warning information is generated, and corresponding protection measures are implemented.

[0084] Optionally, the processor 301 performs the step of collecting operating parameter data of the target device in real time according to a preset collection frequency, including: The preset acquisition frequency is determined based on the operating efficiency of the target device; Based on the preset acquisition frequency, the photovoltaic device's voltage data, current data, surface temperature data, and inverter output power are acquired through the photovoltaic device's sensors. Based on the preset acquisition frequency, lubrication data, wear data, and pitch data of the rotor equipment are collected through wind power equipment sensors. Based on the preset acquisition frequency, environmental sensors are used to collect data on light intensity, wind speed, wind direction, and precipitation in the environment where the target device is located.

[0085] Optionally, the processor 301 performs preprocessing on the operating parameter data to obtain target operating parameter data, including: The operating parameter data is repeatedly filtered to obtain the first processed data; The first processed data is subjected to anomaly removal processing to obtain the second processed data; Feature extraction is performed on the second processed data to generate the target operating parameter data.

[0086] Optionally, the processor 301 executes the multi-dimensional collaborative analysis algorithm to calculate the target operating parameter data and obtain the abnormal data corresponding to the target device, including: Based on the correlation between the target operating parameter data, a feature model is established, which is used to reflect the characteristics of the target equipment's operating status. Based on the feature model, the target operating parameter data are calculated to obtain the feature deviation values ​​of the target device under each operating state; The characteristic deviation value is compared with the preset deviation threshold to determine the abnormal data of the target device under each operating state.

[0087] Optionally, the processor 301 executes the calculation of the target operating parameter data based on the feature model to obtain the feature deviation values ​​of the target device under each operating state, including: Obtain historical operating baseline data of the target device under various operating states; The target operating parameter data is compared with the historical operating baseline data to determine the feature difference value; Based on the aforementioned feature difference values, the feature deviation values ​​of the target device under each operating state are calculated.

[0088] Optionally, the processor 301 executes the abnormal data, including the number of abnormalities, the deviation magnitude, and the duration. Based on a preset strategy and the abnormal data, it generates corresponding early warning information and executes corresponding protective measures, including: Based on the number of anomalies, the magnitude of the deviation, and the duration, the corresponding warning level is determined; Based on the aforementioned warning level, a warning message is generated that includes the device identifier, fault type, warning level, and maintenance recommendations. Based on the warning information, corresponding protection measures are implemented, including shutdown protection and speed reduction protection.

[0089] Optionally, after the processor 301 executes the method to generate corresponding early warning information based on the preset strategy and the abnormal data, and then executes corresponding protection measures, the method further includes: Obtain abnormal data samples of the target device during its historical operating cycle and their corresponding early warning information; Based on the abnormal data samples and early warning information, the feature model is incrementally trained, and the parameters and deviation threshold of the feature model are updated to obtain the updated feature model. When the updated feature model meets the preset convergence condition during training, the current feature model is replaced, and the updated feature model is used for the calculation of subsequent input target running parameter data.

[0090] This invention also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the various processes of the edge computing-based intelligent early warning management method for photovoltaic and wind power equipment or the edge computing-based intelligent early warning management method for photovoltaic and wind power equipment provided in this invention, and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0091] Those skilled in the art will understand that implementing all or part of the processes in the above embodiments can be done by a computer program instructing related hardware, and can be stored in a computer-readable storage medium. When executed, the program can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0092] The above description discloses only preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for intelligent early warning management of photovoltaic and wind power equipment based on edge computing, characterized in that, include: The operating parameter data of the target device are collected in real time according to the preset collection frequency. The operating parameter data is preprocessed to obtain the target operating parameter data; Based on a multi-dimensional collaborative analysis algorithm, the target operating parameter data is calculated to obtain the abnormal data corresponding to the target device; Based on the preset strategy and the abnormal data, corresponding early warning information is generated, and corresponding protection measures are implemented.

2. The intelligent early warning management method for photovoltaic and wind power equipment based on edge computing as described in claim 1, characterized in that, The step of collecting operating parameter data of the target device in real time according to a preset collection frequency includes: The preset acquisition frequency is determined based on the operating efficiency of the target device; Based on the preset acquisition frequency, the photovoltaic device's voltage data, current data, surface temperature data, and inverter output power are acquired through the photovoltaic device's sensors. Based on the preset acquisition frequency, lubrication data, wear data, and pitch data of the rotor equipment are collected through wind power equipment sensors. Based on the preset acquisition frequency, environmental sensors are used to collect data on light intensity, wind speed, wind direction, and precipitation in the environment where the target device is located.

3. The intelligent early warning management method for photovoltaic and wind power equipment based on edge computing as described in claim 1, characterized in that, The preprocessing of the operating parameter data to obtain the target operating parameter data includes: The operating parameter data is repeatedly filtered to obtain the first processed data; The first processed data is subjected to anomaly removal processing to obtain the second processed data; Feature extraction is performed on the second processed data to generate the target operating parameter data.

4. The intelligent early warning management method for photovoltaic and wind power equipment based on edge computing as described in claim 1, characterized in that, The multi-dimensional collaborative analysis algorithm calculates the target operating parameter data to obtain the abnormal data corresponding to the target device, including: Based on the correlation between the target operating parameter data, a feature model is established, which is used to reflect the characteristics of the target equipment's operating status. Based on the feature model, the target operating parameter data are calculated to obtain the feature deviation values ​​of the target device under each operating state; The characteristic deviation value is compared with the preset deviation threshold to determine the abnormal data of the target device under each operating state.

5. The intelligent early warning management method for photovoltaic and wind power equipment based on edge computing as described in claim 4, characterized in that, The step of calculating the target operating parameter data based on the feature model to obtain the feature deviation values ​​of the target device under each operating state includes: Obtain historical operating baseline data of the target device under various operating states; The target operating parameter data is compared with the historical operating baseline data to determine the feature difference value; Based on the aforementioned feature difference values, the feature deviation values ​​of the target device under each operating state are calculated.

6. The intelligent early warning management method for photovoltaic and wind power equipment based on edge computing as described in claim 1, characterized in that, The abnormal data includes the number of anomalies, the magnitude of the deviation, and the duration. Based on the preset strategy and the abnormal data, corresponding early warning information is generated, and corresponding protective measures are executed, including: Based on the number of anomalies, the magnitude of the deviation, and the duration, the corresponding warning level is determined; Based on the aforementioned warning level, a warning message is generated that includes the device identifier, fault type, warning level, and maintenance recommendations. Based on the warning information, corresponding protection measures are implemented, including shutdown protection and speed reduction protection.

7. The intelligent early warning management method for photovoltaic and wind power equipment based on edge computing as described in claim 6, characterized in that, After generating corresponding early warning information based on the preset strategy and the abnormal data, and executing corresponding protective measures, the method further includes: Obtain abnormal data samples of the target device during its historical operating cycle and their corresponding early warning information; Based on the abnormal data samples and early warning information, the feature model is incrementally trained, and the parameters and deviation threshold of the feature model are updated to obtain the updated feature model. When the updated feature model meets the preset convergence condition during training, the current feature model is replaced, and the updated feature model is used for the calculation of subsequent input target running parameter data.

8. A smart early warning management device for photovoltaic and wind power equipment based on edge computing, characterized in that, include: The first acquisition module is used to acquire the operating parameter data of the target device in real time according to the preset acquisition frequency; The first preprocessing module is used to preprocess the operating parameter data to obtain the target operating parameter data; The first calculation module is used to calculate the target operating parameter data based on a multi-dimensional collaborative analysis algorithm to obtain the abnormal data corresponding to the target device; The first generation and execution module is used to generate corresponding early warning information and execute corresponding protection measures based on the preset strategy and the abnormal data.

9. An electronic device, characterized in that, include: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the edge computing-based intelligent early warning management method for photovoltaic and wind power equipment as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the intelligent early warning management method for photovoltaic and wind power equipment based on edge computing as described in any one of claims 1 to 7.

Images