A new energy wind and light power data monitoring system and method

Abstract

The present application belongs to the technical field of wind and light power generation power monitoring, and specifically relates to a new energy wind and light power data monitoring system and method. The present application extracts power fluctuation characteristics and equipment state correlation indexes through anomaly value detection and time series analysis, thereby improving the accuracy and efficiency of data processing. In the aspect of prediction, the present application generates a power prediction curve and a confidence interval by using dynamic comparison of historical operation data sets and current power fluctuation characteristics, combining with environmental parameter change trends, so that the system can identify potential operation abnormalities in advance and provide decision support time for operation and maintenance personnel. In addition, the present application also has corresponding fault diagnosis and optimization capabilities. When the prediction deviation degree exceeds a preset threshold, the present application will automatically trigger an early warning mechanism, locate the fault source through a fault diagnosis process, and finally correct and optimize the power prediction curve according to the fault diagnosis result, forming a closed-loop optimization mechanism to continuously improve the prediction accuracy and operation efficiency of the system.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wind and light power generation power monitoring, and particularly relates to a new energy wind and light power data monitoring system and method. BACKGROUND

[0002] With the continuous expansion of the scale of new energy wind and light power generation, higher requirements are put forward for the operation efficiency and safety of wind and light power stations. Traditional monitoring methods often rely on a single data source, which is difficult to accurately reflect the actual operation state of the wind and light power station. Therefore, a monitoring system based on multi-source data fusion is needed to realize accurate monitoring and prediction of the operation state of the wind and light power station, improve power generation efficiency, and ensure equipment safety.

[0003] In the prior art, although there are some monitoring schemes based on multi-source data, most of them only focus on data collection and lack deep analysis and prediction functions, which cannot diagnose faults in real time and optimize operation strategies, resulting in limited monitoring effect and difficulty in meeting the needs of efficient and safe operation of wind and light power stations. At the same time, it is also difficult to cope with complex and variable environmental factors, affecting the stability of power generation. Based on this, the present application provides a new energy wind and light power data monitoring method to solve the above problems. SUMMARY

[0004] The purpose of the present application is to provide a new energy wind and light power data monitoring system and method, which can realize accurate monitoring and prediction of the operation state of the wind and light power station through multi-source data fusion, improve power generation efficiency, and ensure equipment safety.

[0005] The technical scheme adopted by the present application is as follows:

[0006] A new energy wind and light power data monitoring method, comprising:

[0007] Obtaining multi-source feedback data in the wind and light power station through pre-deployed sensors, including inverter voltage and current monitoring values, environmental parameters, and equipment state parameters;

[0008] Detecting abnormal values of the multi-source feedback data, and extracting power fluctuation characteristics and equipment state correlation indicators based on time series analysis;

[0009] Comparing the power fluctuation characteristics with historical operation data sets dynamically, identifying abnormal fluctuation patterns, and generating a power prediction curve and a predicted confidence interval in combination with the change trend of the environmental parameters;

[0010] According to the prediction result of the power prediction curve, comparing the measured power in real time, and outputting the prediction deviation degree, and when the prediction deviation degree exceeds the preset deviation threshold, triggering the early warning mechanism and diagnosing the fault of the equipment;

[0011] Through the fault diagnosis result, the fault source is located, and the power prediction curve is corrected and optimized, forming a closed-loop optimization mechanism.

[0012] In a preferred scheme, when the multi-source feedback data in the wind and light field station is acquired by the pre-deployed sensor, the following steps are included:

[0013] Dual-channel sensors are deployed on the DC side and the AC side of the inverter to monitor voltage and current, as well as the output waveform of active power, respectively;

[0014] Sensors for collecting wind speed, light intensity, and temperature and humidity information are deployed at the hub center point of the photovoltaic array tilt plane and the fan;

[0015] The acceleration and vibration spectrum of the fan gearbox and generator are monitored by a vibration sensor array, and the surface temperature distribution of the photovoltaic module is collected in real time by an infrared thermal imager;

[0016] A multi-source data time synchronization mechanism is established to align the time scales of various sensors.

[0017] In a preferred scheme, after the multi-source feedback data is output, data verification and data cleaning are performed synchronously;

[0018] The steps during data verification include:

[0019] Sliding window technology is used to process the original collected data in segments, and the statistical characteristic values of the original collected data in each window are calculated;

[0020] When the statistical characteristic values of the continuous N windows exceed the preset range, data review is triggered;

[0021] If the power difference between the DC side and the AC side of the inverter exceeds the preset threshold, it is marked as an abnormal data point and is excluded, otherwise, the corresponding original collected data is retained;

[0022] The steps during data cleaning include:

[0023] Abnormal power data is interpolated and completed, and data repair is performed through the spatial correlation of adjacent sensor nodes;

[0024] Wavelet denoising is performed on the vibration spectrum data to eliminate high-frequency interference components in the vibration signals of the fan gearbox and the generator;

[0025] An association rule library of environmental parameters and equipment states is established, and when the light intensity exceeds the preset intensity threshold and the temperature of the photovoltaic module does not rise synchronously, the inverter output data in the corresponding period is excluded.

[0026] In a preferred scheme, the step of detecting abnormal values in the multi-source feedback data includes:

[0027] The voltage and current data are processed by statistical analysis to eliminate data points deviating from the normal range;

[0028] The output power sequence of the inverter is decomposed to output multiple characteristic components, including a fundamental component, a high-frequency oscillation component, and a random disturbance component;

[0029] When the energy entropy of the high-frequency oscillation component fluctuates within a predetermined threshold range, it is determined to be in a normal state, otherwise it is marked as abnormal;

[0030] The random disturbance component and the environmental parameters are subjected to correlation analysis, and according to a preset correlation coefficient threshold, it is determined whether the random disturbance is caused by environmental factors;

[0031] If the correlation coefficient is lower than the correlation coefficient threshold, it is determined that the random disturbance is caused by internal equipment failure, and the abnormal data points are marked and recorded;

[0032] If the correlation coefficient is higher than the correlation coefficient threshold, it is determined to be affected by environmental factors, and the environmental parameters are retained and recorded.

[0033] In a preferred scheme, the step of extracting power fluctuation characteristics and device state correlation indicators based on time series analysis includes:

[0034] The output power sequence of the inverter is subjected to time series decomposition to extract trend components, periodic components, and random components;

[0035] The average change rate of the trend component is calculated and recorded as a first characteristic parameter;

[0036] The main frequency amplitude of the periodic component is recorded as a second characteristic parameter;

[0037] The variance of the random component is recorded as a third characteristic parameter;

[0038] The vibration frequency spectrum energy of the gearbox is time-domain aligned with the second characteristic parameter of the inverter, and a correlation indicator is calculated;

[0039] According to the linear regression relationship between the temperature gradient of the photovoltaic module and the third characteristic parameter, a temperature sensitivity coefficient is output;

[0040] When the first characteristic parameter exceeds a preset decay threshold, it is determined to be a device aging trend, triggering a warning mechanism;

[0041] When the correlation indicator is lower than a preset safety threshold, it is determined that the gearbox is abnormally worn, and a maintenance process is immediately started;

[0042] When the temperature sensitivity coefficient deviates from the normal range, it is determined that the performance of the photovoltaic module is declining.

[0043] In a preferred solution, the step of dynamically comparing the power fluctuation characteristics with the historical operation data set, identifying abnormal fluctuation patterns, and generating a power prediction curve and a predicted confidence interval in combination with the change trend of the environmental parameters comprises:

[0044] According to the historical operation data set, a power fluctuation characteristic library under different working conditions is established, including normal fluctuation patterns, known fault patterns, and environmental mutation patterns;

[0045] The similarity of the current power fluctuation sequence to each type of pattern in the power fluctuation characteristic library is calculated, and when the similarity is higher than a preset matching threshold, the current device state is identified according to the pattern with the highest similarity;

[0046] If the similarity is lower than the matching threshold, it is determined as an unknown abnormal pattern, and the abnormal data is recorded;

[0047] The average change rate of the trend component is taken as the decay factor, the main frequency amplitude of the periodic component is taken as the fluctuation coefficient, the power prediction curve in the future demand period is output in combination with the change trend of the environmental parameters, and the prediction confidence interval is calculated based on the distribution characteristics of the historical prediction deviation and in combination with the fluctuation range of the current environmental parameters;

[0048] According to the upper and lower boundaries of the confidence interval, a warning threshold is set, the deviation amount of the power prediction curve from the actual power is monitored in real time, and if the deviation amount exceeds the warning threshold, an alarm mechanism is triggered immediately.

[0049] In a preferred solution, the step of comparing the prediction result of the power prediction curve with the measured power in real time and outputting a prediction deviation degree, and triggering a warning mechanism and diagnosing a fault of the device when the prediction deviation degree exceeds a preset deviation threshold comprises:

[0050] The root mean square error between the predicted power and the measured power is calculated through a sliding time window, and is recorded as the prediction deviation degree;

[0051] When the prediction deviation degree exceeds the deviation threshold for K consecutive sampling periods, a fault diagnosis mechanism is started, which comprises:

[0052] An abnormal section in the vibration frequency spectrum energy is extracted and matched with a gear box characteristic frequency spectrum library to identify a specific fault type;

[0053] A hot spot abnormal area of the temperature distribution of the photovoltaic module is located, and a fault module is located in combination with infrared thermal imager data;

[0054] A power difference mutation of the inverter DC side and AC side is detected, and the similarity of the power difference mutation to historical fault data is analyzed, and if the power difference mutation exceeds a preset fault threshold, the inverter is determined to be abnormal, and fault information is recorded.

[0055] In a preferred solution, the step of locating the fault source and optimizing the power prediction curve based on the fault diagnosis result comprises:

[0056] According to the matching result of the gearbox characteristic spectrum, a three-dimensional space coordinate positioning signal is generated, and the position of the fault source in the fan transmission chain is calculated through the phase difference of the vibration sensor array;

[0057] When the photovoltaic module has a hot spot anomaly, the temperature gradient distribution matrix of the infrared thermal imager is used to identify the hot spot position at the cell level;

[0058] The power burst sequence of the inverter DC side and AC side is extracted, and the correlation analysis is performed based on the historical time series data to dynamically generate the decay rate adjustment value of the trend component and the amplitude correction coefficient of the periodic component;

[0059] The decay rate adjustment value and the amplitude correction coefficient are fed back to the reconstruction of the prediction curve to regenerate the power prediction curve and update the upper and lower boundaries of the confidence interval;

[0060] The update of the confidence interval is based on the re-fitting of the new prediction curve and the historical deviation distribution;

[0061] By comparing the width of the confidence interval before and after the update, the convergence degree of the confidence interval is output, and when the convergence degree is lower than the preset threshold, the closed-loop optimization process is suspended, and the fault diagnosis result is pushed to the management end for review, and before the review is completed, the original confidence interval setting is maintained unchanged.

[0062] The application also provides a new energy wind and light power data monitoring system using the new energy wind and light power data monitoring method, comprising:

[0063] The data acquisition module is used to acquire multi-source feedback data in the wind and light field station through the pre-deployed sensor, including inverter voltage and current monitoring values, environmental parameters and device state parameters;

[0064] The feature extraction module is used to detect abnormal values of the multi-source feedback data, and extract power fluctuation features and device state correlation indicators based on time series analysis;

[0065] The prediction module is used to dynamically compare the power fluctuation features with the historical operation data set, identify abnormal fluctuation patterns, and generate a power prediction curve and a predicted confidence interval based on the change trend of the environmental parameters;

[0066] The diagnosis module is used to compare the predicted result of the power prediction curve with the measured power in real time, output the prediction deviation degree, and trigger the early warning mechanism and perform fault diagnosis on the device when the prediction deviation degree exceeds the preset deviation threshold;

[0067] An optimization module is configured to locate a fault source through the fault diagnosis result and correct and optimize the power prediction curve, thereby forming a closed-loop optimization mechanism.

[0068] An electronic device includes:

[0069] at least one processor;

[0070] and a memory connected to the at least one processor in communication;

[0071] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the new energy wind and light power data monitoring method.

[0072] The present application has the following technical effects:

[0073] The present application realizes comprehensive monitoring and deep analysis of the operation state of the wind and light station through multi-source data fusion technology, extracts power fluctuation characteristics and equipment state correlation indicators through anomaly value detection and time series analysis, thereby improving the accuracy and efficiency of data processing, in the prediction aspect, the dynamic comparison of historical operation data set and current power fluctuation characteristics is used, combined with the change trend of environmental parameters, to generate a power prediction curve and a confidence interval, so that the system can identify potential operation abnormalities in advance, and provide decision support time for operation and maintenance personnel, in addition, it also has corresponding fault diagnosis and optimization capability, when the prediction deviation degree exceeds the preset threshold, the early warning mechanism will be automatically triggered, and through the fault diagnosis process, the fault source is located, and finally according to the fault diagnosis result, the power prediction curve is corrected and optimized, forming a closed-loop optimization mechanism, continuously improving the prediction accuracy and operation efficiency of the system. BRIEF DESCRIPTION OF DRAWINGS

[0074] Figure 1 is a method flowchart of the present application;

[0075] Figure 2 is a system module schematic diagram of the present application;

[0076] Figure 3 is an electronic device structure schematic diagram of the present application. DETAILED DESCRIPTION

[0077] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0078] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. However, it will be appreciated that the present application can be practiced in a variety of ways beyond the specific details set forth herein, assuming that the fundamental underlying principles are maintained. It should also be appreciated that the description set forth herein is not intended as a limitation on the scope of the present application but is intended to provide a description of one or more implementations thereof.

[0079] Second, the "one embodiment" or "an embodiment" as referred to herein means a specific implementation that can include features, structures or characteristics that are not included in other implementations. The various embodiments described throughout this specification are not necessarily mutually exclusive, and the specific features, structures, or characteristics of one embodiment can be combined with or substituted for those of another without deviating from the scope of the present application.

[0080] Referring to Figure 1 As shown in the drawings, the present application provides a new energy wind and light power data monitoring method, comprising:

[0081] S1, acquiring multi-source feedback data in the wind and light field station through pre-deployed sensors, including inverter voltage and current monitoring values, environmental parameters and equipment state parameters;

[0082] In the step S1, when the wind and light power data monitoring is performed, the relevant data of the wind and light field station needs to be collected comprehensively at first, which specifically covers the inverter voltage and current monitoring values, so as to reflect the running state of the inverter and provide a basis for subsequent abnormal detection and analysis. The environmental parameters, such as wind speed, light intensity and temperature, can directly affect the power generation efficiency of the wind and light field station, and the equipment state parameters, such as vibration and acceleration data, can reveal the health condition of the equipment. When the multi-source feedback data in the wind and light field station is acquired through the pre-deployed sensors, the following steps are included:

[0083] Deploying double-channel sensors on the direct current side and the alternating current side of the inverter to monitor the voltage and current and the output waveform of the active power respectively;

[0084] Deploying sensors to collect wind speed, light intensity and temperature and humidity information at the hub center point of the photovoltaic array and the fan;

[0085] Monitoring the acceleration and vibration spectrum of the fan gear box and the generator through the vibration sensor array, and collecting the surface temperature distribution of the photovoltaic module in real time by the infrared thermal imager;

[0086] Establishing a multi-source data time synchronization mechanism to align the time scales of various sensors;

[0087] Specifically, when collecting feedback data of various types of sensors, for the inverter, double-channel sensors are arranged on the DC side and the AC side respectively, so as to capture the changes of voltage and current, and the output waveform of active power, which are used to evaluate the efficiency and health status of the inverter. Sensors are arranged at key positions of the photovoltaic array and the fan, so as to obtain environmental parameters such as wind speed, illumination intensity, temperature and humidity in real time, so as to determine the influence of the wind and light field station power generation efficiency and environmental changes on the equipment. At the same time, the application of vibration sensor array and infrared thermal imager can monitor the state of the fan gear box, generator and photovoltaic components, and timely find potential fault points. In addition, in order to ensure the accuracy and consistency of data, a multi-source data time synchronization mechanism is established, and the time scales of various sensors are aligned, so that subsequent data analysis and processing work can be carried out on a unified time reference.

[0088] Secondly, after the multi-source feedback data is output, data checking and data cleaning are synchronously performed;

[0089] The steps of data checking include:

[0090] The original collected data is processed by using the sliding window technology, and the statistical characteristic values of the original collected data in each window are calculated;

[0091] When the statistical characteristic values of the continuous N windows exceed the preset range, data review is triggered;

[0092] If the power difference of the inverter DC side and AC side exceeds the preset threshold, it is marked as an abnormal data point and is excluded, otherwise, the corresponding original collected data is retained;

[0093] The steps of data cleaning include:

[0094] The abnormal power data is interpolated and completed, and the data is repaired through the spatial correlation of adjacent sensor nodes;

[0095] The vibration frequency spectrum data is wavelet denoised to eliminate the high-frequency interference components in the fan gear box and the generator vibration signal;

[0096] The association rule library of environmental parameters and equipment state is established, and when the illumination intensity exceeds the preset intensity threshold and the photovoltaic component temperature does not rise synchronously, the inverter output data of the corresponding period is excluded;

[0097] In this embodiment, after the multi-source feedback data output, synchronous data checking and cleaning are performed to ensure data quality. First, the original collected data is processed in segments using the sliding window technique, and statistical characteristic values (such as mean and variance) are calculated to monitor the stability of the data. Once the statistical characteristic values of consecutive multiple windows exceed the normal range, the data review process is automatically triggered to ensure data accuracy. It should be noted that when the power difference between the DC side and the AC side of the inverter is abnormal, the corresponding abnormal data points are marked and removed to avoid interference with subsequent analysis. In the data cleaning stage, the focus is on repairing and improving the data. For missing or abnormal power data, interpolation methods are used for completion, and the spatial correlation of adjacent sensor nodes is used to further repair the data. At the same time, for vibration spectrum data, wavelet denoising technology is used to eliminate high-frequency interference in the fan gearbox and generator vibration signals, improving the signal-to-noise ratio of the data. In addition, a correlation rule library of environmental parameters and device status is established. When the light intensity is abnormal and the photovoltaic module temperature does not rise synchronously, the inverter output data for the corresponding period is automatically removed to ensure that the data reflects the true device status.

[0098] S2, abnormal value detection is performed on the multi-source feedback data, and power fluctuation characteristics and device status correlation indicators are extracted based on time series analysis;

[0099] In the step S2, after the multi-source feedback data collection is completed, the collected multi-source data is preprocessed to remove abnormal values and ensure data quality. Then, time series analysis method is used to determine the periodicity and trend characteristics of power fluctuation, and a correlation index system is constructed combining with device status parameters, providing data support for subsequent fault diagnosis and performance optimization. The step of abnormal value detection on the multi-source feedback data includes:

[0100] Statistical analysis is used to process voltage and current data to remove data points deviating from the normal range;

[0101] The output power sequence of the inverter is decomposed to output multiple characteristic components, including fundamental component, high-frequency oscillation component and random disturbance component;

[0102] When the energy entropy of the high-frequency oscillation component fluctuates within a predetermined threshold range, it is determined to be normal, otherwise it is marked as abnormal;

[0103] Correlation analysis is performed on the random disturbance component and environmental parameters, and according to a pre-set correlation coefficient threshold, it is determined whether the random disturbance is caused by environmental factors;

[0104] If the correlation coefficient is lower than the correlation coefficient threshold, it is determined that the random disturbance is caused by internal equipment failure, and the abnormal data points are marked and recorded;

[0105] If the correlation coefficient is higher than the correlation coefficient threshold, it is determined that the environmental factor has an impact, and the environmental parameters are retained and recorded;

[0106] Specifically, in the abnormal value detection, first, the voltage and current data need to be statistically analyzed to identify and eliminate abnormal data points that deviate significantly from the normal range to ensure the purity of the data set. Then, the output power sequence of the inverter is decomposed to determine the characteristic components, including the fundamental component, the high-frequency oscillation component, and the random disturbance component. The fundamental component reflects the basic characteristics of the inverter output, the high-frequency oscillation component reflects the high-frequency component in the dynamic response, and the random disturbance component is caused by various factors, including minor faults inside the device or random interference from the external environment. When the energy entropy of the high-frequency oscillation component fluctuates within a predetermined threshold range, it is considered as the normal state of the inverter. If the energy entropy exceeds the predetermined threshold range, it is marked as an abnormal state, which means there is some degree of fault or instability inside the inverter. Further, the correlation between the random disturbance component and the environmental parameters is analyzed to output the corresponding correlation coefficient, which is compared with the preset correlation coefficient threshold to determine whether the random disturbance is mainly caused by environmental factors. If the correlation coefficient is lower than the correlation coefficient threshold, it is determined that the random disturbance is caused by some fault inside the device, and the abnormal data points are marked and recorded for subsequent analysis. On the contrary, if the correlation coefficient is higher than the threshold, it is considered that the random disturbance is mainly caused by environmental factors, and the related environmental parameters are retained and recorded for consideration of their impact in subsequent analysis.

[0107] In addition, the steps of extracting power fluctuation features and device state correlation indicators based on time series analysis include:

[0108] Time series decomposition is performed on the inverter output power sequence to extract trend components, periodic components, and random components;

[0109] The average change rate of the trend component is calculated and recorded as the first feature parameter;

[0110] The main frequency amplitude of the periodic component is recorded as the second feature parameter;

[0111] The variance of the random component is recorded as the third feature parameter;

[0112] The vibration frequency spectrum energy of the gearbox is time-domain aligned with the second feature parameter of the inverter to calculate the correlation indicator;

[0113] According to the linear regression relationship between the temperature gradient of the photovoltaic module and the third feature parameter, the temperature sensitivity coefficient is output;

[0114] When the first feature parameter exceeds the preset decay threshold, it is determined that the device is aging, and the warning mechanism is triggered;

[0115] When the correlation index is lower than the preset safety threshold, it is determined that the gearbox is abnormally worn, and a maintenance process is immediately started;

[0116] When the temperature sensitivity coefficient deviates from the normal range, it is determined that the performance of the photovoltaic module is degraded;

[0117] In this embodiment, when extracting the power fluctuation feature and the equipment state correlation index, the inverter output power sequence is first subjected to time series decomposition to identify the trend component, the periodic component and the random component in the inverter output power sequence, the trend component, the periodic component and the random component respectively represent the long-term trend, the periodic fluctuation and the random disturbance of the power change over time, so that the running state of the inverter can be more comprehensively understood, the average change rate of the trend component is recorded as a first feature parameter, reflecting the long-term attenuation trend of the inverter output power, if the first feature parameter exceeds the preset attenuation threshold, it means that the equipment has an aging trend, at this time, the warning mechanism is triggered to take timely measures for maintenance or replacement, the main frequency amplitude of the periodic component is recorded as a second feature parameter, reflecting the periodic fluctuation feature of the inverter output power, by comparison with historical data, it can be judged whether the current periodic fluctuation is normal or not, after the vibration spectrum energy of the gearbox and the second feature parameter of the inverter are time-domain aligned, the correlation index calculated can reveal the running state correlation between the gearbox and the inverter, when the correlation index is lower than the preset safety threshold, it means that the gearbox may have abnormal wear, at this time, the maintenance process should be immediately started to avoid fault expansion, the variance of the random component is recorded as a third feature parameter, reflecting the random fluctuation degree of the inverter output power, the linear regression relationship between the temperature gradient of the photovoltaic module and the third feature parameter is used to output the temperature sensitivity coefficient, the temperature sensitivity coefficient can reflect the performance state of the photovoltaic module, when the temperature sensitivity coefficient deviates from the normal range, it may mean that the performance of the photovoltaic module is degraded, and needs to be repaired or replaced, based on this, real-time monitoring and early warning of the wind and light field station equipment can be realized, providing a strong guarantee for the safe and stable operation of the wind and light field station.

[0118] S3, dynamically compare the power fluctuation feature with the historical running data set, identify abnormal fluctuation patterns, generate a power prediction curve in combination with the change trend of the environmental parameters, and calculate a predicted confidence interval;

[0119] In the step S3, after the power fluctuation feature is output, the extracted power fluctuation feature is dynamically compared with the historical running data set, through this way, abnormal fluctuation patterns can be identified, at the same time, in combination with the change trend of the environmental parameters, a power prediction curve is generated, and a predicted confidence interval is calculated, wherein the step of dynamically comparing the power fluctuation feature with the historical running data set, identifying abnormal fluctuation patterns, generating a power prediction curve in combination with the change trend of the environmental parameters, and calculating a predicted confidence interval comprises:

[0120] Based on historical operating datasets, a power fluctuation feature library for different operating conditions is established, including normal fluctuation modes, known fault modes, and environmental change modes.

[0121] Calculate the similarity between the current power fluctuation sequence and various patterns in the power fluctuation feature library, and when the similarity is higher than the preset matching threshold, identify the current device status based on the pattern with the highest similarity.

[0122] If the similarity is lower than the matching threshold, it is determined to be an unknown abnormal pattern, and the abnormal data is recorded.

[0123] Using the average rate of change of the trend component as the decay factor and the main frequency amplitude of the periodic component as the fluctuation coefficient, combined with the trend of environmental parameter changes, the power prediction curve for the future demand period is output. Then, based on the distribution characteristics of historical prediction deviations and combined with the fluctuation range of current environmental parameters, the prediction confidence interval is calculated.

[0124] Based on the upper and lower boundaries of the confidence interval, a warning threshold is set, and the deviation between the power prediction curve and the actual power is monitored in real time. If the deviation exceeds the warning threshold, an alarm mechanism is immediately triggered.

[0125] Specifically, when outputting the power prediction curve, a power fluctuation feature library needs to be constructed based on historical data. This library covers various operating conditions, including normal fluctuation modes, known fault modes, and sudden environmental changes, serving as a benchmark for comparative analysis. Then, the similarity between the current power fluctuation sequence and various modes in the feature library is calculated, using metrics such as cosine similarity or Euclidean distance. By comparing the similarity with a preset matching threshold, the current state of the equipment can be identified. If the similarity is higher than the matching threshold, the equipment state is determined based on the mode with the highest similarity. Conversely, if the similarity is lower than the matching threshold, it is judged as an unknown abnormal mode, and the abnormal data is recorded for subsequent analysis. After determining the current power fluctuation mode, the average rate of change of the trend component and the dominant frequency amplitude of the periodic component are combined as the attenuation factor and fluctuation coefficient, respectively. The changing trends of environmental parameters are also considered. These factors are then integrated to output the power prediction curve for the future demand period. The expression for the power prediction curve is as follows:

[0126] ;

[0127] In the formula, express The predicted power value at time t. Indicates the initial reference power. Indicates the attenuation factor. Indicates the volatility coefficient. Indicates the environmental sensitivity coefficient. denotes the angular frequency, which is determined by the dominant frequency of the periodic component in the historical operation data set, wherein, , denotes the dominant frequency of the periodic fluctuation, denotes the phase shift, denotes the environmental parameter change amount, in addition, in order to evaluate the reliability of the prediction result, the confidence interval of the prediction is also calculated based on the distribution characteristics of the historical prediction deviation and combined with the fluctuation range of the current environmental parameter, and the calculation formula of the confidence interval is: , wherein, denotes the confidence interval, denotes the standard normal distribution quantile corresponding to the confidence level, denotes the standard deviation of the historical prediction error, denotes the uncertainty caused by the environmental parameter fluctuation, finally, according to the upper and lower boundaries of the confidence interval, the early warning threshold is set, the deviation amount between the power prediction curve and the actual power is monitored in real time, and once the deviation amount exceeds the early warning threshold, the alarm mechanism is triggered immediately, so as to ensure the safe and stable operation of the wind and light field station.

[0128] S4, according to the prediction result of the power prediction curve, real-time comparison is made with the measured power, and the prediction deviation degree is output, and when the prediction deviation degree exceeds the preset deviation threshold, the early warning mechanism is triggered, and the equipment is diagnosed for fault;

[0129] In the step S4, after the power prediction curve is output, the prediction result of the power prediction curve is compared and analyzed with the real-time monitored power data, so that the prediction deviation degree, that is, the difference between the predicted value and the actual value, can be output, and when the prediction deviation degree exceeds the preset deviation threshold, the early warning mechanism is automatically triggered, and the equipment is diagnosed for fault, wherein, the step of comparing the prediction result of the power prediction curve with the measured power in real time and outputting the prediction deviation degree, and triggering the early warning mechanism and diagnosing the equipment for fault when the prediction deviation degree exceeds the preset deviation threshold, comprises:

[0130] The root mean square error between the predicted power and the measured power is calculated through the sliding time window, and is recorded as the prediction deviation degree;

[0131] When the prediction deviation degree exceeds the deviation threshold in the continuous K sampling periods, the fault diagnosis mechanism is started, including:

[0132] Extract the abnormal section in the vibration spectrum energy, and match it with the gear box characteristic spectrum library to identify the specific fault type;

[0133] Locate the hot spot abnormal area of the temperature distribution of the photovoltaic module, and locate the fault component combined with the infrared thermal imager data;

[0134] Detecting the power difference mutation variable of the DC side and the AC side of the inverter, analyzing the similarity of the power difference mutation variable and the historical fault data, and determining that the inverter is abnormal when the power difference mutation variable exceeds the preset fault threshold, and recording the fault information;

[0135] Specifically, after the power prediction curve and the corresponding confidence interval are determined, the root mean square error between the predicted power and the measured power is calculated in real time through the sliding time window technology, and the root mean square error value is recorded as the prediction deviation, so as to intuitively reflect the difference between the prediction result and the actual operation condition, and provide corresponding basis for subsequent analysis. When the prediction deviation is stable and exceeds the preset deviation threshold in the continuous K sampling periods, the fault diagnosis mechanism is automatically started to identify and locate the potential equipment fault. In the fault diagnosis process, first, the abnormal section in the vibration frequency spectrum energy is extracted, and the abnormal section contains the key information of the internal fault of the equipment. Then, the abnormal section is matched and compared with the pre-established gearbox characteristic spectrum library. Through comparison and analysis, the specific fault type such as gear wear and bearing fault can be identified, which provides a clear direction for subsequent maintenance and replacement. Secondly, the hot spot abnormal area in the temperature distribution of the photovoltaic module is located. The hot spot abnormal area usually shows abnormal temperature rise or drop. Combined with the real-time monitoring data of the infrared thermal imager, the photovoltaic module with fault can be located, so that timely measures can be taken for repair or replacement, thereby avoiding the influence of fault expansion on the overall power generation efficiency. Then, the power difference mutation variable of the DC side and the AC side of the inverter is detected to reflect the working state of the inverter. By analyzing the similarity of the power difference mutation variable and the historical fault data, whether the inverter is abnormal can be determined. When the power difference mutation variable exceeds the preset fault threshold, it is determined that the inverter is abnormal, and the related fault information is recorded. Based on this, real-time monitoring and early warning of the wind and light field station equipment can be realized, and corresponding protection for the safe and stable operation of the wind and light field station can be provided.

[0136] S5、Through the fault diagnosis result, the fault source is located, and the power prediction curve is corrected and optimized to form a closed-loop optimization mechanism;

[0137] In the step S5, according to the result of fault diagnosis, the fault source is located, and in addition, the power prediction curve is corrected and optimized according to the diagnosis result, so as to form a closed-loop optimization mechanism, thereby the prediction accuracy can be improved, and potential problems of the equipment can be found and solved in time to ensure the stable operation of the new energy wind and light field station. The steps of locating the fault source through the fault diagnosis result and correcting and optimizing the power prediction curve include:

[0138] According to the matching result of the gearbox characteristic spectrum, a three-dimensional space coordinate positioning signal is generated, and the position of the fault source in the fan transmission chain is calculated through the phase difference of the vibration sensor array;

[0139] When the photovoltaic module has a hot spot anomaly, the temperature gradient distribution matrix of the infrared thermal imager is used to identify the hot spot position at the cell level.

[0140] The power burst sequence of the inverter DC side and AC side is extracted, the historical time series data is associated and analyzed, the decay rate adjustment value of the trend component and the amplitude correction coefficient of the periodic component are dynamically generated;

[0141] The decay rate adjustment value and the amplitude correction coefficient are fed back to the reconstruction of the prediction curve, the power prediction curve is regenerated, and the upper and lower boundaries of the confidence interval are updated;

[0142] The update of the confidence interval is based on the re-fitting of the new prediction curve and the historical deviation distribution;

[0143] By comparing the width of the confidence interval before and after the update, the convergence degree of the confidence interval is output, when the convergence degree is lower than the preset threshold, the closed-loop optimization process is suspended, and the fault diagnosis result is pushed to the management end for review, and before the review is completed, the original confidence interval setting is maintained unchanged;

[0144] Specifically, in locating the fault source, first, according to the matching result of the gearbox characteristic spectrum, the corresponding three-dimensional space coordinate positioning signal is generated, which can clearly indicate the specific position of the fault in the fan transmission chain, thereby guiding the maintenance personnel to quickly locate and repair the fault. For the hot spot anomaly of the photovoltaic module, the temperature gradient distribution matrix obtained by the infrared thermal imager is used to identify the hot spot position at the battery piece level through corresponding image processing algorithms such as edge detection, threshold segmentation, etc., which helps maintenance personnel quickly locate the fault component and take appropriate repair or replacement measures. At the same time, the power burst sequence of the inverter DC side and AC side is extracted, which contains key information of the inverter working state. Through correlation analysis based on historical time series data, the decay rate adjustment value of the trend component and the amplitude correction coefficient of the periodic component can be dynamically generated. The decay rate adjustment value = (the difference between the current trend component decay rate and the historical average decay rate) x dynamic adjustment coefficient, the dynamic adjustment coefficient is determined according to the power recovery rate after historical fault repair, the amplitude correction coefficient = (the difference between the current periodic component amplitude and the historical average amplitude / historical average amplitude) x amplitude adjustment factor, and the amplitude adjustment factor is set according to the recovery characteristics of the periodic fluctuation in the historical fault data. The decay rate adjustment value and the amplitude correction coefficient are fed back to the reconstruction process of the prediction curve in real time, so as to realize the correction and optimization of the power prediction curve. After correction and optimization, the power prediction curve is regenerated, and the upper and lower boundaries of the confidence interval are also updated simultaneously. The update of the confidence interval is based on the refitting result of the new prediction curve and the historical deviation distribution, which helps to improve the accuracy of prediction. In addition, it needs to be clear that by comparing the confidence interval width before and after the update, a confidence interval convergence index is output. When the convergence degree is lower than the preset threshold, the closed-loop optimization process is paused, and the fault diagnosis result is pushed to the management end for review. Before the review is completed, the original confidence interval setting is maintained unchanged to ensure the stability and reliability of the prediction result. After the review is passed, the closed-loop optimization is restarted to continuously monitor the prediction accuracy until the preset optimization target is reached.

[0145] Please refer to Figure 2 A new energy wind and light power data monitoring system using the new energy wind and light power data monitoring method described above, comprising:

[0146] A data acquisition module for acquiring multi-source feedback data in a wind and light field station through pre-deployed sensors, including inverter voltage and current monitoring values, environmental parameters, and device state parameters;

[0147] A feature extraction module for detecting outliers on multi-source feedback data and extracting power fluctuation features and device state correlation indicators based on time series analysis;

[0148] a prediction module configured to dynamically compare the power fluctuation feature with a historical operation data set, identify an abnormal fluctuation pattern, and generate a power prediction curve and a predicted confidence interval in combination with a change trend of an environmental parameter;

[0149] a diagnosis module configured to compare the prediction result of the power prediction curve with a measured power in real time, output a prediction deviation degree, and trigger an early warning mechanism and perform fault diagnosis on the device when the prediction deviation degree exceeds a preset deviation threshold;

[0150] an optimization module configured to locate a fault source based on the fault diagnosis result and correct and optimize the power prediction curve to form a closed-loop optimization mechanism;

[0151] In the above, the data acquisition module is responsible for collecting real-time multi-source feedback data fed back by various sensors in the wind-solar station. The multi-source feedback data is derived from a pre-deployed sensor network and covers inverter voltage and current monitoring values, environmental parameters (such as wind speed, wind direction, temperature, humidity, etc.), and device state parameters (such as vibration, acceleration, etc.). The feature extraction module is responsible for preprocessing the collected multi-source feedback data, including outlier detection and data cleaning, to ensure data quality. On this basis, time series analysis technology is used to extract power fluctuation features and device state correlation indicators, which can reflect the operation state and power output of the wind-solar station device and provide key information for subsequent prediction and diagnosis. The prediction module dynamically compares the extracted power fluctuation feature with a historical operation data set, identifies an abnormal fluctuation pattern, and generates a power prediction curve and a predicted confidence interval for a future period of time in combination with a change trend of an environmental parameter. The diagnosis module compares the prediction result of the power prediction curve with real-time monitored power data, outputs a prediction deviation degree, i.e., the difference between the predicted value and the actual value, and automatically triggers an early warning mechanism and starts a fault diagnosis process to locate a fault source and take corresponding repair measures when the prediction deviation degree exceeds a preset deviation threshold. The optimization module corrects and optimizes the power prediction curve based on the fault diagnosis result to form a closed-loop optimization mechanism, thereby improving the accuracy of prediction, discovering and solving potential problems of the device in a timely manner, and ensuring stable operation and efficient power generation of the new energy wind-solar station.

[0152] Please refer to Figure 3 An electronic device, the electronic device comprising:

[0153] at least one processor;

[0154] and a memory connected in communication with the at least one processor;

[0155] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the new energy wind and light power data monitoring method.

[0156] The processor of the electronic device can be a central processing unit (CPU), a graphics processing unit (GPU), or a digital signal processor (DSP), and the memory can be a random access memory (RAM), a read-only memory (ROM), or a solid state disk (SSD). The electronic device can further include an arithmetic logic unit (ALU) or a floating point unit (FPU) as an operation unit, a keyboard and a mouse as input devices, and a display and a printer as output devices, which work together to efficiently process and store a large amount of data and ensure stable system operation.

[0157] The above description is only the preferred embodiments of the present application, and it should be pointed out that for ordinary skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, which should be considered as the protection scope of the present application. The structures, devices and operation methods not specifically described and explained in the present application, such as without special description and limitation, are implemented according to the conventional means in the art.

Claims

1. A method for monitoring power data of new energy wind and solar power, characterized in that: The method comprises the following steps: Obtain multi-source feedback data in the wind-solar station through pre-deployed sensors, including inverter voltage and current monitoring values, environmental parameters, and equipment state parameters; Perform outlier detection on the multi-source feedback data, and extract power fluctuation characteristics and equipment state correlation indicators based on time series analysis; Dynamically compare the power fluctuation characteristics with historical operation data sets, identify abnormal fluctuation patterns, generate a power prediction curve based on the trend of environmental parameters, and predict the confidence interval; Compare the prediction results of the power prediction curve with the measured power in real time, output the prediction deviation, and trigger the early warning mechanism when the prediction deviation exceeds the preset deviation threshold, and perform fault diagnosis on the equipment; Locate the fault source through the fault diagnosis results, and optimize the power prediction curve to form a closed-loop optimization mechanism; The step of locating the fault source through the fault diagnosis results and optimizing the power prediction curve comprises the following steps: According to the matching result of the gear box characteristic spectrum, a three-dimensional space coordinate positioning signal is generated, and the position of the fault source in the fan transmission chain is calculated through the phase difference of the vibration sensor array; When there is a hot spot anomaly in the photovoltaic module, the temperature gradient distribution matrix of the infrared thermal imager is used to identify the hot spot position at the cell level; Extract the power burst sequence of the inverter DC side and AC side, perform correlation analysis based on historical time series data, dynamically generate the attenuation rate adjustment value of the trend component, and the amplitude correction coefficient of the periodic component; The attenuation rate adjustment value and the amplitude correction coefficient are fed back to the reconstruction of the prediction curve to regenerate the power prediction curve and update the upper and lower boundaries of the confidence interval; The update of the confidence interval is based on the re-fitting of the new prediction curve and the historical deviation distribution; By comparing the width of the confidence interval before and after the update, the convergence degree of the confidence interval is output, and when the convergence degree is lower than the preset threshold, the closed-loop optimization process is suspended, and the fault diagnosis results are pushed to the management end for review. Before the review is completed, the original confidence interval setting remains unchanged.

2. The new energy wind-solar power data monitoring method according to claim 1, characterized in that: When the multi-source feedback data is obtained through the pre-deployed sensors, the following steps are included: Deploy double-channel sensors on the DC side and AC side of the inverter to monitor voltage and current, as well as the output waveform of active power; Deploy sensors to collect wind speed, light intensity, and temperature and humidity information at the hub center point of the photovoltaic array and the fan; Monitor the acceleration and vibration spectrum of the fan gear box and generator through the vibration sensor array, and collect the surface temperature distribution of the photovoltaic module in real time using the infrared thermal imager; Establish a multi-source data time synchronization mechanism to align the time scales of various sensors.

3. The new energy wind-solar power data monitoring method according to claim 2, characterized in that: After the multi-source feedback data is output, data verification and data cleaning are performed synchronously; The steps of data verification include: Use sliding window technology to segment the original collected data and calculate the statistical characteristic values of the original collected data in each window; When the statistical characteristic values of N consecutive windows exceed the preset range, trigger data review; If the power difference between the DC side and AC side of the inverter exceeds the preset threshold, mark it as an abnormal data point and exclude it, otherwise, retain the corresponding original collected data; The steps of data cleaning include: interpolation and completion of abnormal power data, and data repair through spatial correlation of adjacent sensor nodes; wavelet denoising of vibration spectrum data to eliminate high-frequency interference components in the vibration signals of the fan gearbox and the generator; establishment of an association rule base of environmental parameters and device states, and elimination of inverter output data in the corresponding period when the light intensity exceeds the preset intensity threshold and the temperature of the photovoltaic module does not rise synchronously.

4. The new energy wind-solar power data monitoring method according to claim 1, characterized in that: The steps of abnormal value detection on the multi-source feedback data include: processing of voltage and current data by statistical analysis to eliminate data points deviating from the normal range; decomposition of the output power sequence of the inverter to output multiple characteristic components, including a fundamental component, a high-frequency oscillation component and a random disturbance component; determination of a normal state when the energy entropy of the high-frequency oscillation component fluctuates within a predetermined threshold range, otherwise, the state is marked as abnormal; correlation analysis on the random disturbance component and the environmental parameters, and determination of whether the random disturbance is caused by environmental factors according to a preset correlation coefficient threshold; determination of the random disturbance as being caused by internal device failure if the correlation coefficient is lower than the correlation coefficient threshold, and marking and recording of the abnormal data points; determination of the influence of environmental factors if the correlation coefficient is higher than the correlation coefficient threshold, and retention and recording of the environmental parameters.

5. The new energy wind-solar power data monitoring method according to claim 4, characterized in that: The steps of extracting power fluctuation characteristics and device state association indicators based on time series analysis include: time series decomposition of the inverter output power sequence to extract a trend component, a periodic component and a random component; calculation of the average change rate of the trend component and recording as a first characteristic parameter; recording of the main frequency amplitude of the periodic component as a second characteristic parameter; recording of the variance of the random component as a third characteristic parameter; time-domain alignment of the vibration spectrum energy of the gearbox and the second characteristic parameter of the inverter to calculate an association indicator; output of a temperature sensitivity coefficient according to the linear regression relationship between the temperature gradient of the photovoltaic module and the third characteristic parameter; determination of a device aging trend when the first characteristic parameter exceeds a preset attenuation threshold, triggering a warning mechanism; determination of abnormal wear of the gearbox when the association indicator is lower than a preset safety threshold, immediately starting a maintenance process; determination of a performance decline of the photovoltaic module when the temperature sensitivity coefficient deviates from the normal range.

6. The new energy wind-solar power data monitoring method according to claim 1, characterized in that: The steps of dynamically comparing the power fluctuation characteristics with the historical running data set to identify abnormal fluctuation patterns, generating a power prediction curve in combination with the change trend of the environmental parameters, and predicting the confidence interval include: establishment of a power fluctuation characteristic library under different working conditions according to the historical running data set, including normal fluctuation patterns, known fault patterns and environmental mutation patterns; calculation of the similarity of the current power fluctuation sequence to each type of pattern in the power fluctuation characteristic library, and identification of the current device state according to the pattern with the highest similarity when the similarity is higher than a preset matching threshold; determination of an unknown abnormal pattern if the similarity is lower than the matching threshold, and recording of the abnormal data. The average change rate of the trend component is taken as a decay factor, the amplitude of the main frequency of the periodic component is taken as a fluctuation coefficient, and the power prediction curve in the future demand period is output in combination with the change trend of the environmental parameters; and then, based on the distribution characteristics of the historical prediction deviation and in combination with the fluctuation range of the current environmental parameters, the prediction confidence interval is calculated; According to the upper and lower boundaries of the confidence interval, the early warning threshold is set, the deviation amount of the power prediction curve from the actual power is monitored in real time, and if the deviation amount exceeds the early warning threshold, an alarm mechanism is triggered immediately.

7. The new energy wind-solar power data monitoring method according to claim 1, characterized in that: The step of comparing the prediction result of the power prediction curve with the measured power in real time and outputting the prediction deviation degree, and triggering the early warning mechanism and performing fault diagnosis on the equipment when the prediction deviation degree exceeds the preset deviation threshold, includes: The root mean square error between the predicted power and the measured power is calculated through a sliding time window, and is recorded as the prediction deviation degree; When the prediction deviation degree exceeds the deviation threshold in the continuous K sampling periods, the fault diagnosis mechanism is started, including: Abnormal sections in the vibration spectrum energy are extracted and matched with a gear box characteristic spectrum library to identify the specific fault type; An abnormal area of a hot spot of a temperature distribution of a photovoltaic component is located, and a fault component is located in combination with infrared thermal imager data; A power difference mutation of the inverter DC side and AC side is detected, and the similarity between the power difference mutation and historical fault data is analyzed, and if the power difference mutation exceeds a preset fault threshold, it is determined that the inverter is abnormal, and fault information is recorded.

8. A new energy wind and light power data monitoring system, characterized in that: The new energy wind and light power data monitoring method of any one of claims 1 to 7 is used, including: A data acquisition module is configured to acquire multi-source feedback data in a wind and light station through pre-deployed sensors, including inverter voltage and current monitoring values, environmental parameters, and equipment state parameters; A feature extraction module is configured to detect abnormal values of the multi-source feedback data, and extract power fluctuation features and equipment state correlation indicators based on time series analysis; A prediction module is configured to compare the power fluctuation features with historical operation data sets dynamically, identify abnormal fluctuation patterns, generate a power prediction curve in combination with a change trend of the environmental parameters, and a prediction confidence interval; A diagnosis module is configured to compare a prediction result of the power prediction curve with measured power in real time, output a prediction deviation degree, and trigger an early warning mechanism and perform fault diagnosis on equipment when the prediction deviation degree exceeds a preset deviation threshold; An optimization module is configured to locate a fault source through the fault diagnosis result, and correct and optimize the power prediction curve to form a closed-loop optimization mechanism.

9. An electronic device, comprising: The electronic device includes: at least one processor; and a memory connected in communication with the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the new energy wind and light power data monitoring method of any one of claims 1 to 7. The electronic device includes: at least one processor; and a memory connected in communication with the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the new energy wind and light power data monitoring method of any one of claims 1 to 7.

Images