A monitoring and management method and system based on wind farms

Abstract

This invention discloses a monitoring and management method and system for wind farms, relating to the field of computer platform load balancing technology. The method includes installing sensors to collect wind farm operation data; cleaning and filtering the data; extracting state features using time-frequency domain analysis techniques; establishing a baseline state model of the equipment during normal operation to monitor the operation of wind farm equipment; and predicting the remaining service life of key components using machine learning algorithms. The method described in this invention achieves full-time and spatial monitoring and intelligent analysis of wind farm operation data, improving equipment operating efficiency and the overall reliability of the wind farm. This invention belongs to the field of wind power generation system monitoring and management, providing an efficient and stable technical means for realizing intelligent wind farm operation.

Description

Technical Field

[0001] This invention relates to the field of wind power management technology, specifically to a monitoring and management method and system based on wind farms. Background Technology

[0002] With the deepening utilization of renewable energy globally, wind power, as a clean and sustainable energy form, has developed rapidly worldwide. Modern wind farms are expanding in scale, from single wind turbine units to large-scale centralized wind farms, and even forming multi-regional interconnected wind power networks. To achieve efficient and safe wind farm operation, wind farm monitoring and management technologies have been extensively researched and applied. Traditional wind farm monitoring mainly relies on SCADA (Supervisory Control and Data Acquisition) systems, which collect wind turbine operating data for status monitoring and basic fault alarms. However, with the increasing scale and complexity of wind farms, traditional SCADA systems have gradually revealed their insufficient capabilities in data integration, fault prediction, and intelligent scheduling. In recent years, a new generation of monitoring systems combining IoT, AI, and big data technologies has emerged, aiming to achieve intelligent, real-time, and refined management of wind farm operation.

[0003] Despite significant research and technological applications in wind farm monitoring and management, current technologies still have several shortcomings. First, traditional monitoring methods often rely on fixed-rule threshold alarms, failing to predict potential failures of critical components and leading to sudden downtime and high maintenance costs. Second, existing monitoring systems largely depend on centralized data processing architectures, which have limited processing efficiency and struggle to meet real-time requirements when dealing with large-scale wind farms. Furthermore, scheduling and management issues in the collaborative operation of cross-regional wind farms remain unresolved, particularly regarding the coordinated control of grid fluctuations and wind power instability. Finally, current technologies lack comprehensive perception and optimization of the operational status of various devices within a wind farm, hindering intelligent lifecycle management from individual devices to the entire wind farm. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is to address the problems of insufficient wind power equipment fault prediction, limited real-time monitoring capabilities, and insufficient cross-regional collaborative management capabilities in the existing technology.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a monitoring and management method based on wind farms, including...

[0007] Install sensors to collect wind farm operation data;

[0008] The data is cleaned and filtered, and state features are extracted using time-frequency domain analysis techniques.

[0009] Establish a baseline status model for the equipment during normal operation to monitor the operation of wind farm equipment;

[0010] By combining machine learning algorithms, the remaining service life of key components can be predicted.

[0011] As a preferred embodiment of the wind farm-based monitoring and management method described in this invention, the installation of sensors includes: determining the equipment components and operating parameters to be monitored; installing vibration, temperature, and current / voltage sensors on key components such as gearbox bearings, generator stators, and wind turbine blades; ensuring accurate sensor placement; reducing external noise interference; and using a SCADA system to collect sensor data in real time, with the sampling frequency set to above 1 kHz to capture high-frequency vibration signals.

[0012] The data acquisition formula uses sensors to collect data from the fan and the environment, including wind speed. Power generation

[0013] ,

[0014] in, Indicates conversion efficiency. Indicates air density, This represents the swept area of ​​the blades; the data sampling period formula is expressed as:

[0015] ,

[0016] in, This indicates the sensor sampling frequency.

[0017] As a preferred embodiment of the wind farm-based monitoring and management method described in this invention, the data cleaning and filtering process includes: removing redundant data with the same timestamp or minimal data value variation from frequently collected duplicate data packets; performing signal filtering by setting a cutoff frequency, selecting a frequency higher than the signal bandwidth based on the equipment's operating frequency range, applying a filter to retain low-frequency information while suppressing high-frequency noise, and using a low-pass filter for filtering.

[0018] ,

[0019] in, Represents the filter transfer function. Indicates the cutoff frequency;

[0020] The signal is decomposed using discrete wavelet transform. A suitable mother wavelet is selected to perform multi-level decomposition of the signal. High-frequency components are thresholded, and the decomposed signal is reconstructed to generate a denoised signal.

[0021] As a preferred embodiment of the wind farm-based monitoring and management method described in this invention, the step of extracting state features using time-frequency domain analysis technology includes converting the time-domain signal to the frequency domain, analyzing the intensity of different frequency components in the signal to identify possible faults, and performing a Fast Fourier Transform (FFT).

[0022] The formula for feature extraction mean is expressed as:

[0023] ,

[0024] The variance formula is expressed as:

[0025] ,

[0026] Power spectral density (PSD):

[0027] ,

[0028] in, The short-time Fourier transform represents the spectrum of a signal.

[0029] ,

[0030] in, Represents the window function. It represents a time signal.

[0031] As a preferred embodiment of the wind farm-based monitoring and management method of the present invention, the establishment of the equipment baseline state model includes: constructing a baseline model, assuming a normal distribution or an adaptive baseline;

[0032] Perform anomaly detection, assuming the normal data distribution is available. Weighted sum representation of Gaussian distributions:

[0033] ,

[0034] in, Indicates the first The weights of the Gaussian components satisfy the following conditions: , Represents the mean vector. Represent the covariance matrix;

[0035] Construct an autoencoder, which maps input features to a latent variable space:

[0036] ,

[0037] in, Represents the encoder function. Indicates encoder weights, This represents the encoder bias; the decoder maps the latent variables back to the original space:

[0038] ,

[0039] in, Represents the decoder function. Indicates the decoder weights. This represents the decoder bias; the loss function is the reconstruction error.

[0040] ,

[0041] in, Represents the model reconstruction loss. This represents the original input data. This represents the reconstructed output data; the model is trained using normal operating data with the goal of minimizing the reconstruction error loss function.

[0042] Real-time data Input the baseline model, perform comparative detection, and the probabilistic model detection is represented as follows:

[0043] ,

[0044] Anomaly detection, Indicates the reconstruction error detection threshold; when When this occurs, the data is considered abnormal; the reconstruction error detection formula is expressed as:

[0045] ,

[0046] in, This represents the reconstructed output data of real-time data; anomaly detection is performed when the reconstruction error exceeds a set threshold, indicating that the data is abnormal.

[0047] As a preferred embodiment of the wind farm-based monitoring and management method described in this invention, the monitoring of the wind farm equipment operation includes optimizing the maintenance plan based on the monitoring results, and the dynamic programming optimization total cost formula is expressed as:

[0048] ,

[0049] in, Indicates maintenance costs. Indicates downtime losses;

[0050] Maintenance priorities, using risk assessment methods, risk scoring:

[0051] ,

[0052] in, Indicates the probability of failure. This indicates that the malfunction affects costs.

[0053] Using swarm intelligence algorithms and genetic algorithms to optimize the maintenance order: the objective function is to minimize... The constraint is to satisfy the maintainability of the fan.

[0054] As a preferred embodiment of the wind farm-based monitoring and management method of the present invention, the prediction of the remaining useful life of key components includes remaining useful life prediction (RUL), which is calculated based on a linear degradation model.

[0055] ,

[0056] in, Indicates current health status. This indicates the set fault threshold. The degradation rate can be represented by fitting historical data;

[0057] A machine learning model is constructed using historical operating data of wind farm equipment, including health status characteristics, operating conditions, environmental factors, and actual failure times. A Long Short-Term Memory (LSTM) network is trained to predict health status trends. The state update formula is expressed as:

[0058] ,

[0059] in, Indicates the current memory state. Indicates forgetting and input gate, outputting predicted value:

[0060] ,

[0061] in, When the predicted remaining useful life (RUL) is lower than the preset threshold, an alarm signal is triggered to prompt the maintenance team to perform maintenance in advance; different levels of remaining useful life correspond to different maintenance strategies.

[0062] As a preferred embodiment of the wind farm-based monitoring and management system described in this invention, the data acquisition unit integrates data inputs from multiple sensors;

[0063] The communication module supports both wireless and wired transmission, ensuring real-time transmission to the central control system or cloud platform.

[0064] The data processing and analysis module processes the collected data, extracts features, and performs status monitoring and anomaly detection.

[0065] The fault prediction module predicts faults based on machine learning algorithms, formulates optimized maintenance plans, dynamically plans and calculates the optimal maintenance strategy to obtain the real-time cost of different optimal maintenance strategies.

[0066] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement a monitoring and management method based on a wind farm.

[0067] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a wind farm-based monitoring and management method.

[0068] The beneficial effects of this invention are as follows: The wind farm-based monitoring and management method provided by this invention offers an innovative solution integrating IoT sensing technology, distributed data processing architecture, and intelligent prediction algorithms. This invention enables real-time and spatial monitoring and intelligent analysis of wind farm operation data, improving equipment operating efficiency and overall wind farm reliability, and providing an efficient and stable technical means for achieving intelligent wind farm operation. Attached Figure Description

[0069] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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.

[0070] Figure 1 The first embodiment of the present invention provides an overall flowchart of a monitoring and management method based on a wind farm. Detailed Implementation

[0071] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0072] Example 1, referring to Figure 1 As an embodiment of the present invention, a monitoring and management method based on a wind farm is provided, comprising:

[0073] S1: Install sensors to collect wind farm operation data.

[0074] Furthermore, the installation of sensors includes determining the equipment components and operating parameters that need to be monitored, installing vibration, temperature, and current / voltage sensors on key components such as gearbox bearings, generator stators, and wind turbine blades; ensuring accurate sensor placement, reducing external noise interference, and using a SCADA system to collect sensor data in real time, with the sampling frequency set to above 1 kHz to capture high-frequency vibration signals;

[0075] The data acquisition formula uses sensors to collect data from the fan and the environment, including wind speed. Power generation

[0076] ,

[0077] in, Indicates conversion efficiency. Indicates air density, This represents the swept area of ​​the blades; the data sampling period formula is expressed as:

[0078] ,

[0079] in, This indicates the sensor sampling frequency.

[0080] The data cleaning and filtering process includes: removing redundant data with the same timestamp or minimal data value variation from frequently collected duplicate data packets; performing signal filtering by setting a cutoff frequency, selecting a frequency higher than the signal bandwidth based on the device's operating frequency range, applying a filter to retain low-frequency information while suppressing high-frequency noise, and using a low-pass filter for filtering.

[0081] ,

[0082] in, Represents the filter transfer function. Indicates the cutoff frequency;

[0083] It should be noted that the discrete wavelet transform is used to decompose the signal. A suitable mother wavelet is selected to perform multi-level decomposition of the signal, the high-frequency components are thresholded, and the decomposed signal is reconstructed to generate a denoised signal.

[0084] S2: Clean and filter the data, and extract state features using time-frequency domain analysis techniques.

[0085] Extracting state features using time-frequency domain analysis techniques includes converting the time-domain signal to the frequency domain, analyzing the intensity of different frequency components in the signal to identify possible faults, and performing a Fast Fourier Transform (FFT).

[0086] The formula for feature extraction mean is expressed as:

[0087] ,

[0088] The variance formula is expressed as:

[0089] ,

[0090] Power spectral density (PSD):

[0091] ,

[0092] in, The short-time Fourier transform represents the spectrum of a signal.

[0093] ,

[0094] in, Represents the window function. It represents a time signal.

[0095] S3: Establish a baseline state model for the equipment during normal operation to monitor the operation of wind farm equipment.

[0096] Collect operational data during normal equipment operation as input for baseline status. Long-term data acquisition: Select periods of stable equipment operation under stable load and environmental conditions to collect time-series data of key parameters (such as vibration, temperature, and current). Preprocessing: Clean and filter the collected data to remove noise, missing values, and outliers to ensure data quality. Feature extraction: Extract features in the time domain, frequency domain, and time-frequency domain, such as mean, peak value, and spectral amplitude.

[0097] Reconstruction error, for real-time data Calculate its probability in the Gaussian mixture model. .

[0098] If the probability value is below a certain threshold (e.g.) If so, it is considered abnormal.

[0099] Real-time data is reconstructed using an autoencoder, and the reconstruction error is calculated. If the error exceeds the threshold (e.g., the 95% confidence interval), it is considered abnormal.

[0100] By combining multiple features, a health score (e.g., a normalized value between 0 and 1) is calculated, and a baseline model is constructed by setting a score threshold, using either a normal distribution assumption or an adaptive baseline. Anomaly detection is then performed, assuming that the data distribution of the normal state is available. Weighted sum representation of Gaussian distributions:

[0101] ,

[0102] in, Indicates the first The weights of the Gaussian components satisfy the following conditions: , Represents the mean vector. Represent the covariance matrix. The model parameters are estimated using the Expectation-Maximization (EM) algorithm:

[0103] ,

[0104] ,

[0105] Construct an autoencoder, which maps input features to a latent variable space:

[0106] ,

[0107] in, Represents the encoder function. Indicates encoder weights, This represents the encoder bias; the decoder maps the latent variables back to the original space:

[0108] ,

[0109] in, Represents the decoder function. Indicates the decoder weights. This represents the decoder bias; the loss function is the reconstruction error.

[0110] ,

[0111] in, Represents the model reconstruction loss. This represents the original input data. This represents the reconstructed output data. The model is trained using normal operating data, with the goal of minimizing the reconstruction error loss function.

[0112] Real-time data Input the baseline model, perform comparative detection, and the probabilistic model detection is represented as follows:

[0113] ,

[0114] Anomaly detection, This represents the threshold for detecting reconstruction errors.

[0115] Model training using normal operating data The goal of training the model is to minimize the following loss function: GMM minimizes the negative log-likelihood function:

[0116] ,

[0117] Autoencoder, minimizing reconstruction error:

[0118] ,

[0119] when When this occurs, the data is considered abnormal. The reconstruction error detection formula is expressed as:

[0120] ,

[0121] in, This represents the reconstructed output data of real-time data; it performs anomaly detection, considering the data as abnormal when the reconstruction error exceeds a set threshold.

[0122] The monitoring of wind farm equipment operation includes optimizing the maintenance plan based on the monitoring results. The dynamic programming optimization total cost formula is expressed as follows:

[0123] ,

[0124] in, Indicates maintenance costs. This indicates losses due to downtime.

[0125] Maintenance priorities, using risk assessment methods, risk scoring:

[0126] ,

[0127] in, Indicates the probability of failure. This indicates that the malfunction affects costs.

[0128] Using swarm intelligence algorithms and genetic algorithms to optimize the maintenance order: the objective function is to minimize... The constraint is to satisfy the maintainability of the fan.

[0129] It should be noted that rolling window updates use the latest normal operating data to retrain the model in a sliding window manner. Incremental training is performed on the existing model, gradually updating the model parameters without requiring a complete rebuild. The model's detection performance is periodically validated to ensure that the updated model accurately reflects the latest normal state.

[0130] S4: Combine machine learning algorithms to predict the remaining service life of key components.

[0131] The predicted remaining useful life of critical components includes Remaining Useful Life (RUL) prediction, which is calculated based on a linear degradation model.

[0132] ,

[0133] in, Indicates current health status. This indicates the set fault threshold. The degradation rate can be represented by fitting historical data.

[0134] A machine learning model is constructed using historical operating data of wind farm equipment, including health status characteristics, operating conditions, environmental factors, and actual failure times. A Long Short-Term Memory (LSTM) network is trained to predict health status trends. The state update formula is expressed as:

[0135]

[0136] in, Indicates the current memory state. Indicates forgetting and input gate, outputting predicted value:

[0137] ,

[0138] in, When the predicted remaining useful life (RUL) is lower than the preset threshold, an alarm signal is triggered to prompt the maintenance team to perform maintenance in advance; different levels of remaining useful life correspond to different maintenance strategies.

[0139] It should be noted that real-time operational data input: The device's real-time operational data is input into the trained model to calculate the current health factors. and the predicted remaining useful life Regularly update the model, using newly collected data for incremental training to ensure prediction accuracy. When the equipment operating environment changes (such as load fluctuations or extreme weather), incorporate environmental factors to correct the prediction results.

[0140] Example 2 is an embodiment of the present invention, which provides a monitoring and management method based on wind farms. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0141] To verify the effectiveness of a wind farm-based monitoring and management method and system, 10 wind turbines in a wind farm were selected as test subjects. Vibration and temperature sensors were installed on key components such as gearbox bearings and generator stators. Data acquisition was performed using a SCADA system with a sampling frequency set to 1 kHz to capture high-frequency signal characteristics. The data was transmitted in real time to the central monitoring system via a fiber optic network.

[0142] First, data such as wind speed, power generation, and component vibration frequency are recorded using sensors.

[0143] During the data cleaning stage, a low-pass filter is used to remove high-frequency noise.

[0144] The cutoff frequency was set to 500 Hz. Subsequently, a Fast Fourier Transform (FFT) was performed on the vibration signal to extract the main frequency domain features. To capture the time-frequency characteristics of the non-stationary signal, a Short-Time Fourier Transform (STFT) was also employed.

[0145] The time-frequency resolution is optimized using the Hanning window function. Next, a baseline state model of the equipment is constructed based on normal operation data, employing the normal distribution assumption and an anomaly scoring function. Anomaly detection is combined with Support Vector Machine (SVM) to identify abnormal states of key components.

[0146] Finally, the remaining lifetime (RUL) of critical components is predicted using a linear degradation model via a long short-term memory (LSTM) network.

[0147] Table 1 Experimental Data

[0148] Fan number Wind speed (m / s) Power generation capacity (kW) Vibration frequency (Hz) Temperature (°C) Abnormal scoring Remaining life expectancy prediction (days) Fan 1 12.3 2500 35.6 55 1.2 320 Fan 2 11.8 2400 36.1 56 1.5 290 Fan 3 13.2 2600 38.3 57 2.1 250 Fan 4 10.5 2300 34.7 54 1.0 340 Fan 5 12.8 2550 37.4 58 2.5 210 Fan 6 11.6 2450 35.8 55 1.4 310 Fan 7 14.0 2700 39.5 59 2.8 180

[0149] As can be seen from the table data, the wind turbine data in this embodiment exhibits significant differences. Regarding anomaly scoring, wind turbines 1 and 8 have lower anomaly scores (1.2 and 0.9), indicating that their operating status is close to the baseline and the risk of anomalies is low; while wind turbines 5 and 7 have higher scores (2.5 and 2.8), suggesting a need for close monitoring. This demonstrates that the anomaly scoring function can effectively quantify the degree to which the wind turbine status deviates from the baseline.

[0150] Example 3 is an embodiment of the present invention, which provides a monitoring and management system based on a wind farm, wherein the data acquisition unit integrates data input from multiple sensors.

[0151] The communication module supports wireless transmission such as LoRa and 5G, or wired transmission such as fiber optics, ensuring real-time transmission to the central control system or cloud platform.

[0152] The data processing and analysis module processes the collected data, extracts features, and performs status monitoring and anomaly detection.

[0153] The fault prediction module predicts faults based on machine learning algorithms, formulates optimized maintenance plans, dynamically plans and calculates the optimal maintenance strategy to obtain the real-time cost of different optimal maintenance strategies.

[0154] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0155] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0156] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0157] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0158] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A monitoring management method based on a wind farm, characterized by, Comprise: Install sensors to collect wind farm operation data; Clean and filter the data, and extract state characteristics using time-frequency domain analysis techniques; Establish a baseline state model of the equipment during normal operation to monitor the operation of the wind farm equipment; Combine machine learning algorithms to predict the remaining useful life of key components; The use of time-frequency domain analysis techniques to extract state characteristics includes converting time domain signals to frequency domain, analyzing the intensity of different frequency components in the signal to identify possible faults, and performing fast Fourier transform (FFT); The feature extraction feature mean formula is expressed as: , The variance formula is expressed as: , wherein denotes the signal variance; Power spectral density (PSD): , wherein denotes the short-time Fourier transform of the signal , wherein denotes a window function, denotes a time signal; The establishment of the baseline state model of the equipment includes the construction of the baseline model, the normal distribution hypothesis or the adaptive baseline; Anomaly detection is performed, assuming that the data distribution of the normal state is available is represented as a weighted sum of Gaussian distributions: , wherein, denotes the weight of the th Gaussian distribution, satisfying , denotes the mean vector, denotes the covariance matrix; Build an autoencoder that maps input features to a latent variable space: , where, represents an encoder function, represents an encoder weight, represents an encoder bias; the decoder maps the latent variable back to the original space: , wherein, denotes the decoder function, denotes the decoder weight, denotes the decoder bias; the loss function is the reconstruction error: , wherein, represents a model reconstruction loss, represents original input data, represents reconstructed output data; the model is trained using normal operation data, the goal being to minimize the reconstruction error loss function; Real-time data Input baseline model, perform contrast detection, probability model detection represented as: , abnormality determination, denotes the reconstruction error detection threshold value; when the data is considered abnormal; the reconstruction error detection formula is expressed as: , wherein, reconstructed output data representing real-time data; performing abnormality determination when the reconstruction error exceeds a set threshold value, the data is considered abnormal; The monitoring of the operation of the wind farm equipment includes optimizing the maintenance plan according to the monitoring situation, and the dynamic planning optimization total cost formula is expressed as: , wherein, represents maintenance costs, represents downtime losses; Maintenance priority, using risk assessment methods, risk score: , wherein, represents the probability of failure, represents the cost of failure impact; Using swarm intelligence algorithm, genetic algorithm optimization maintenance sequence: objective function for minimizing , constraint conditions for meeting the maintainability of the fan; The prediction of the remaining useful life of the key components includes the remaining useful life prediction (RUL) based on the linear degradation model to calculate the remaining useful life: , wherein, represents the current health state, represents the set failure threshold, represents the degradation rate, fitted from historical data; Build a machine learning model using the operational history data of the wind farm equipment, which contains health status characteristics, operating conditions, environmental factors, and actual failure time, train a long short-term memory network (LSTM), and predict the health status trend. The state update formula is expressed as: , wherein, denotes the current memory state, denotes the forget and input gates; , wherein, represents the predicted output of the current time step; when the predicted remaining useful life RUL is lower than a preset threshold, an alarm signal is triggered to prompt the operation and maintenance team to overhaul in advance; different levels of remaining useful life correspond to different maintenance strategies.

2. The wind farm based monitoring management method of claim 1, wherein: The installation of sensors includes determining the equipment components and operating parameters that need to be monitored, installing vibration, temperature and current voltage sensors on key components such as gearbox bearings, generator stators and fan blades; Ensure the accuracy of the sensor position, reduce external noise interference, use SCADA system to collect sensor data in real time, and set the sampling frequency to 1 kHz or higher to capture high-frequency vibration signals; Data acquisition formula, through the sensor to collect fan and environmental data, wind speed , power generation , , wherein, represents the conversion efficiency, represents the air density, represents the blade swept area; the data sampling period formula is represented as: , wherein, denotes the sensor sampling frequency.

3. The wind farm based monitoring management method of claim 2, wherein: The data cleaning and filtering includes removing redundant data with the same timestamp or very small data value changes for frequently collected repeated data packets; Signal filtering is performed, the cutoff frequency is set, the frequency higher than the signal bandwidth is selected according to the operating frequency range of the equipment, and the filter is applied to retain low-frequency information while suppressing high-frequency noise. Low-pass filter is used for filtering: , wherein represents a transfer function of a low-pass filter, represents a cut-off frequency; Discrete wavelet transform is used to decompose the signal, suitable mother wavelet is selected for multi-layer decomposition of the signal, high-frequency components are thresholded, and the decomposed signal is reconstructed to generate the denoised signal.

4. A system employing the monitoring and management method based on a wind farm according to any one of claims 1 to 3, characterized in that, Comprise: Data acquisition unit, integrated with multiple sensor data inputs; Communication module, supporting wireless and wired transmission, ensuring real-time transmission to central control system or cloud platform; Data processing and analysis module, processing, feature extraction and state monitoring and anomaly detection of collected data; Fault prediction module, based on machine learning algorithms to predict faults and develop optimized maintenance plans, dynamically plan optimal maintenance strategies to obtain optimal real-time costs of different maintenance strategies. 5.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-4 when the computer program is executed by the processor. The processor executes the computer program to realize the steps of the wind farm-based monitoring management method of any one of claims 1 to 3.

6. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program, when executed by a processor, implements the steps of the wind farm based monitoring management method according to any one of claims 1 to 3.

Images