A method and device for monitoring the performance of a fan

Abstract

The application discloses a wind turbine working performance monitoring method and device, and relates to the technical field of wind power generation, which comprises the following steps: collecting environmental data, equipment working state data and equipment electrical data, and performing pretreatment; performing multi-source data integration on the pretreated equipment working state data and equipment electrical data to obtain a data set; extracting an input feature vector from the data set; constructing a fault identification model to obtain equipment fault state data according to the input feature vector; constructing a health degree evaluation model, which is used for evaluating the running state of the equipment according to the fault state data; if the running state of the equipment is good, then the current power generation of the equipment is judged by combining the collected data, and the pitch angle is dynamically adjusted until the wind energy utilization rate of the equipment reaches a threshold value when the current power generation is lower than the average value of historical power generation; if the running state of the equipment has problems, then a sound-light alarm unit is used for early warning; and the application is used for solving the problem that the traditional wind turbine working performance is not accurately judged.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular to a method and device for monitoring the working performance of a wind turbine. Background Art

[0002] A wind turbine is a device that converts wind energy into electrical energy. It is mainly composed of blades, generators, mechanical components and electrical components. According to the different rotating axes, wind turbines are mainly divided into two categories: horizontal axis wind turbines and vertical axis wind turbines. Currently, horizontal axis wind turbines occupy a mainstream position in the market. In actual operation, in order to ensure operational safety and facilitate maintenance and optimization of the operating status of the generator set, it is necessary to use a work monitoring device to monitor the wind turbine in real time.

[0003] Current working monitoring devices generally use a threshold judgment method to determine the fault status of the generator set. However, since the operating environment of the generator set changes with the external weather or operating time, a single threshold cannot accurately determine the fault status of the generator set.

[0004] Chinese patent publication number CN106523276A discloses a monitoring device for a wind turbine. The wind turbine monitoring device in this invention patent monitors the actual wind speed, wind direction, and humidity at the wind turbine site. Three comparators compare the wind speed, wind direction, and humidity with threshold values, ultimately outputting a binary signal consisting of a pass signal and a fail signal. Based on this signal, a remote terminal can monitor the operating status of the wind turbine on site in a timely manner over a GSM network.

[0005] However, this wind turbine monitoring device relies too much on the judgment of wind speed parameter thresholds and cannot combine equipment operating parameters with electrical parameters for collaborative analysis. Its ability to process complex signals needs to be further improved. It also relies on fixed threshold alarms and lacks predictive capabilities, making it impossible to achieve fault model identification and equipment life prediction effects.

[0006] Therefore, we propose a monitoring method that can accurately judge the working performance of the fan. Summary of the Invention

[0007] The object of the present invention is to provide a method and device for monitoring the working performance of a fan, which are used to solve the problem of inaccurate judgment of the working performance of the fan in the traditional way.

[0008] The present invention is achieved through the following technical solutions:

[0009] A method for monitoring fan operating performance, comprising:

[0010] Collect environmental data, equipment working status data, and equipment electrical data, and perform pre-processing;

[0011] Perform multi-source data integration on the pre-processed equipment working status data and equipment electrical data to obtain a data set;

[0012] Extract input feature vectors from the dataset;

[0013] Build a fault identification model and obtain the fault status data of the equipment based on the input feature vector;

[0014] Build a health assessment model to evaluate the operating status of the equipment based on fault status data;

[0015] If the equipment is in good operating condition, the current power generation of the equipment is determined based on the collected data. If the current power generation is lower than the average value of the historical power generation, the pitch angle is dynamically adjusted until the wind energy utilization rate of the equipment reaches the threshold.

[0016] If there is a problem with the equipment's operating status, an early warning will be issued through the sound and light alarm unit.

[0017] Furthermore, the environmental data includes wind speed data, ambient temperature data, ambient humidity data, snow thickness data on the top of the wind turbine, and peak value data of lightning current;

[0018] The equipment working status data includes the vibration acceleration data of the gearbox, the torque load data of the transmission chain, and the oil temperature data and oil viscosity change data of the oil tank;

[0019] The equipment electrical data includes power generation current data and power generation voltage data of the generator.

[0020] Furthermore, the input feature vector includes the spectrum peak of vibration acceleration, the fluctuation variance of torque, the harmonic distortion rate of current, the oil temperature stability index and the viscosity deviation.

[0021] Furthermore, the peak value of the spectrum of the vibration acceleration , and its calculation formula is:

[0022]

[0023] is the vibration acceleration data of the gearbox;

[0024] The torque fluctuation variance , and its calculation formula is:

[0025]

[0026] in, is the timestamp, is the total time of fluctuation, is the torque load data of the transmission chain, For the Torque load data of the second drive chain, is the mean value of the torque load data within the time window;

[0027] Harmonic distortion rate of the current , and its calculation formula is:

[0028]

[0029] in, is the generator’s current data, is the harmonic order, is the highest harmonic order included in the calculation;

[0030] The oil temperature stability index , and its calculation formula is:

[0031]

[0032] in, For the The oil temperature value of the sampling time, is the mean oil temperature in the time window, is the number of sampling points in the time window;

[0033] The oil viscosity change data , and its calculation formula is:

[0034]

[0035] in, is the current oil viscosity measurement value, It is the reference viscosity under standard working conditions.

[0036] Furthermore, a support vector machine is used to construct the fault identification model, and the fault status data includes a fault type and a confidence score.

[0037] Furthermore, the fault type is determined by the decision function of the fault identification model The calculation is as follows:

[0038] ;

[0039] in, is the total number of input feature vectors, is the Lagrange multiplier, is the radial basis kernel function, The first support vectors, is the input feature vector to be classified, is the bias term, Fault type label.

[0040] Furthermore, the confidence score is calculated as follows:

[0041] Calculate the distance from the input feature vector to the hyperplane :

[0042]

[0043] in, is the normal vector of the hyperplane;

[0044] The distance Normalize to the interval [0, 1] and define the normalized distance:

[0045]

[0046] in, is the maximum distance of the training set;

[0047] Mapping normalized distance to confidence score , the calculation formula is:

[0048]

[0049] is the slope parameter.

[0050] Furthermore, the health assessment model is constructed as follows:

[0051] Defining health indicators for:

[0052]

[0053] in, For the The weights of the fault types, and , MTBF is the mean time between failures.

[0054] A fan performance monitoring device includes a sensor monitoring module, an analysis and control module, a circuit protection module and a remote interaction module;

[0055] The sensor monitoring module includes an environmental monitoring unit, a state monitoring unit, and an electrical parameter monitoring unit. The environmental monitoring unit includes an ultrasonic anemometer fixedly mounted on the outside of the top of the wind turbine generator set, which is used to detect environmental wind data; a temperature and humidity sensor fixedly mounted on the outside of the wind turbine generator set, which is used to detect environmental temperature and humidity data; a piezoelectric snow depth sensor fixedly mounted on the top of the wind turbine generator set, which is used to detect the thickness of snow accumulation; and a lightning current sensor fixed in the lightning protection down conductor of the power generation blade, which is used to detect the peak value data of the lightning current.

[0056] The condition monitoring unit includes a vibration sensor fixed to the outside of the gearbox and the generator set bearing, which is used to detect abnormal vibration data; a strain gauge torque sensor attached to the outside of the generator set drive shaft, which is used to detect the torque load data of the transmission chain; a temperature sensor and a viscosity sensor fixed to the outside of the generator set oil tank, which are used to detect the oil temperature data and oil viscosity change data of the oil tank;

[0057] The electrical parameter monitoring unit includes a current transformer and a voltage divider connected to the internal circuit of the generator set, and is used to collect current and voltage signals in the control circuit;

[0058] The analysis and control module includes a data fusion platform, a diagnostic engine layer equipped with a cloud server cluster, an adaptive pitch control unit, and an oscillation suppression unit. The data fusion platform is used to integrate various types of data collected by the sensor monitoring module and extract input feature vectors from the data set; the diagnostic engine layer is configured with a fault identification model and an equipment health assessment model; the adaptive pitch control unit is used to dynamically adjust the pitch angle and optimize the wind energy utilization coefficient when the equipment is in good operating condition and the power generation is low; the oscillation suppression unit is used to suppress tower oscillations through an active damping controller when the equipment is in good operating condition and the tower amplitude exceeds a threshold.

[0059] The circuit protection module is composed of a surge protector with a three-level protection module. The first level of the surge protector is set at the main power distribution cabinet of the generator set, the second level is set at the power interface position, and the third level is set at the circuit interface end;

[0060] The remote interaction module includes a visualization interface and an audible and visual alarm unit, and both the visualization interface and the audible and visual alarm unit are electrically connected to the analysis and control module.

[0061] Furthermore, the temperature and humidity sensor is externally designed with a radiation shield.

[0062] The technical solution of the present invention has at least the following advantages and beneficial effects:

[0063] The present invention discloses a method and device for monitoring the working performance of a wind turbine, which can provide comprehensive input for subsequent comprehensive diagnosis of a wind turbine by collecting and analyzing multi-source data;

[0064] In addition, by adopting a hierarchical model architecture of fault identification model and health assessment model, it is possible to avoid direct reliance on raw data, improve the rigor of the assessment logic, and thus ensure the accuracy of the obtained wind turbine operating status results;

[0065] In addition, by extracting input feature vectors from the data set, the pertinence of the input data can be ensured, thereby facilitating the subsequent fault identification module to identify faults of the wind turbine. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 A schematic flow chart of a method of the present invention is shown;

[0067] Figure 2 A schematic diagram of the structure of a device of the present invention;

[0068] Figure 3 It is a structural diagram of the analysis control module of the present invention. DETAILED DESCRIPTION

[0069] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0070] Example 1

[0071] like Figure 1 A method for monitoring fan operating performance is shown, comprising:

[0072] Collect environmental data, equipment working status data, and equipment electrical data, and perform pre-processing;

[0073] The environmental data includes wind speed data, ambient temperature data, ambient humidity data, snow thickness data on the top of the wind turbine, and peak value data of lightning current;

[0074] It should be noted that the snow thickness data on the top of the wind turbine is measured by a piezoelectric snow depth sensor. The piezoelectric snow depth sensor uses an ultrasonic probe to emit ultrasonic waves and uses the ultrasonic reflection time difference to measure the snow thickness. When the snow thickness is >30cm during continuous snowfall, the blade heating system is activated to melt snow and de-ice.

[0075] When the measured humidity is >80% and the temperature is close to 0°C, the low temperature warning is triggered to start the blade heating system. When the ambient temperature is >40°C, the high temperature protection is triggered to limit the power operation to protect the equipment.

[0076] and when a thunderstorm is predicted to be approaching based on peak lightning current data, adjusting the blade angle to reduce exposure risk;

[0077] The equipment working status data includes the vibration acceleration data of the gearbox, the torque load data of the transmission chain, and the oil temperature data and oil viscosity change data of the oil tank;

[0078] The equipment electrical data includes the power generation current data and power generation voltage data of the generator;

[0079] The preprocessing methods include filtering, aligning the timestamps of multi-source data using interpolation formulas, and unifying the formats of various collected data into a standardized format. This can eliminate the noise of the collected data and achieve temporal and spatial alignment between multi-source data, which is beneficial for subsequent processing.

[0080] Perform multi-source data integration on the pre-processed equipment working status data and equipment electrical data to obtain a data set;

[0081] Extract input feature vectors from the dataset;

[0082] In addition, the input feature vector includes the spectrum peak of vibration acceleration, the fluctuation variance of torque, the harmonic distortion rate of current, the oil temperature stability index and the viscosity deviation; wherein the input feature vector is expressed as follows: ;

[0083] The peak value of the spectrum of the vibration acceleration is calculated as follows:

[0084]

[0085] is the vibration acceleration data of the gearbox;

[0086] The spectrum peak of the vibration acceleration is analyzed in the FFT frequency domain to capture the abnormal high-frequency components in the gearbox vibration, effectively identifying mechanical faults such as bearing wear and gear tooth breakage. In addition, frequency domain analysis can filter out time domain noise, highlight the characteristics of periodic faults, and has high computational efficiency.

[0087] The torque fluctuation variance , and its calculation formula is:

[0088]

[0089] in, is the timestamp, is the total time of fluctuation, is the torque load data of the transmission chain, For the Torque load data of the second drive chain, is the mean value of the torque load data within the time window;

[0090] Torque fluctuation variance is used to quantify the stability of the transmission chain load. Increased variance may indicate coupling loosening, blade imbalance, or transmission shaft fatigue. Variance calculation is simple and sensitive to fluctuations, making it suitable for real-time monitoring of dynamic load changes.

[0091] The harmonic distortion rate of the current is calculated as follows:

[0092]

[0093] in, is the generator’s current data, is the harmonic order, is the highest harmonic order included in the calculation;

[0094] The harmonic distortion rate of current reflects the electrical health of the generator or converter. An increase in the harmonic distortion rate may indicate a winding short circuit, power device aging, or grid interference. Percentage normalization facilitates setting a unified threshold and can distinguish between fundamental and harmonic energy, making it more targeted.

[0095] The oil temperature stability index , and its calculation formula is:

[0096]

[0097] in, For the The oil temperature value of the sampling time, is the mean oil temperature in the time window, is the number of sampling points in the time window;

[0098] The oil temperature stability indicator is used to monitor the thermal stability of the lubrication system. An increase in the standard deviation may be caused by cooling failure, oil line blockage, or friction anomalies. Quantifying temperature fluctuations through the standard deviation avoids misjudgment of a single threshold and adapts to dynamic operating conditions under different ambient temperatures.

[0099] The oil viscosity change data , and its calculation formula is:

[0100]

[0101] in, is the current oil viscosity measurement value, It is the reference viscosity under standard working conditions;

[0102] Oil viscosity change data is used to detect oil aging or contamination. Viscosity deviation may reduce lubrication efficiency and accelerate mechanical wear. The percentage form directly reflects the degree of deviation. The reference value It can be dynamically adjusted according to working conditions and has high flexibility.

[0103] Build a fault identification model and obtain the fault status data of the equipment based on the input feature vector;

[0104] Build a health assessment model to evaluate the operating status of equipment based on fault status data. By adopting a layered model architecture of fault identification model and health assessment model, it can avoid direct reliance on raw data, improve the rigor of the assessment logic, and thus ensure the accuracy of the obtained wind turbine operating status results.

[0105] If the equipment is in good operating condition, the current power generation of the equipment is determined based on the collected data. If the current power generation is lower than the average value of the historical power generation, the pitch angle is dynamically adjusted until the wind energy utilization rate of the equipment reaches the threshold.

[0106] The power generation of the equipment is calculated from the equipment electrical data, and the average value of historical power generation is the average power generation at the current moment in the past 10 years;

[0107] The process of dynamically adjusting the pitch angle is as follows: when the equipment is in good operating condition and the power generation is low, it means that the blades of the wind turbine are not facing the wind direction, which leads to insufficient power generation. , optimize wind energy utilization coefficient , the formula is:

[0108]

[0109] in, The tip speed ratio after correction is a key parameter to measure the operating status of the wind turbine and directly affects the wind energy capture efficiency. is the base of natural logarithms;

[0110]

[0111]

[0112] in, is the tip speed ratio, is the initial pitch angle, is the angular velocity of the impeller, is the blade radius, is the wind speed;

[0113] According to the Betz limit, wind turbines can capture up to 59.3% of wind energy, that is, The maximum theoretical value is 0.593. In practical applications, it should be as close to this value as possible, but it is usually affected by factors such as mechanical loss and aerodynamic efficiency. The actual optimal value is between 0.4 and 0.5, so when When the value is between 0.4 and 0.5, the pitch angle can be stopped. changes.

[0114] If there is a problem with the equipment's operating status, an early warning will be issued through the sound and light alarm unit.

[0115] Example 2

[0116] As an embodiment, a support vector machine (SVM) is used to construct the fault identification model, and the fault status data includes a fault type and a confidence score.

[0117] In addition, the fault type is determined by the decision function of the fault identification model The calculation is as follows:

[0118]

[0119] in, is the total number of input feature vectors, is the Lagrange multiplier, is the radial basis kernel function, which is used to map data into a high-dimensional space. The first support vectors, is the input feature vector to be classified (such as vibration spectrum peak, harmonic distortion rate, etc.), is a bias term used to adjust the position of the classification hyperplane. is the fault type label;

[0120] In addition, the radial basis kernel function :

[0121]

[0122] in, is the kernel parameter.

[0123] It should be noted that the fault type is a pre-built type set, which includes mechanical faults: gear / bearing damage (determined by abnormal high-frequency vibration characteristics or abnormal spectrum peaks in specific frequency bands), drive chain abnormalities (determined by a sudden increase in torque variance), and blade imbalance (determined by periodic torque fluctuations); electrical faults: winding short circuit (determined by a significant increase in harmonic distortion), converter aging (determined by current waveform distortion), and grid interference (driven by abnormal voltage fluctuations); lubrication system faults: oil temperature abnormalities (determined by a continuous increase in oil temperature standard deviation) and oil contamination (determined by viscosity deviation exceeding a threshold); environmental-related faults: blade icing (triggered by snow thickness > 30 cm), lightning strike risk (determined when the lightning current peak exceeds the limit), and high-temperature derating (triggered when the ambient temperature > 40°C).

[0124] As needed, the confidence score calculation process is as follows:

[0125] Calculate the distance from the input feature vector to the hyperplane :

[0126]

[0127] in, is the normal vector of the hyperplane;

[0128] The distance Normalize to the interval [0, 1] and define the normalized distance:

[0129]

[0130] in, is the maximum distance of the training set;

[0131] The normalized distance is mapped to a confidence score through the Sigmoid function (S-shaped function) , the calculation formula is:

[0132]

[0133] is the slope parameter, which is used to control the steepness of the confidence curve. The typical value is , is the base of natural logarithms;

[0134] By normalizing the distance and the Sigmoid function, the hard decision results of SVM are converted into continuous confidence scores, which improves the interpretability of the model output and the slope parameter Dynamic adjustment is performed through cross-validation to adapt to different data distributions and avoid the increase in false alarm rate caused by fixed thresholds.

[0135] Example 3

[0136] As an embodiment, the health assessment model is constructed as follows:

[0137] Define health indicators as:

[0138]

[0139] in, For the The weights of the fault types, and , MTBF is the mean time between failures;

[0140] Among them, the health index is used to judge whether the health of wind power equipment is good or has problems. It is necessary to set the corresponding threshold according to the actual situation. There is a problem with the wind power equipment. The wind power equipment is in good operating condition;

[0141] In addition, by dynamically calculating the health level based on the type and confidence of the fault status data, it is possible to avoid direct reliance on the original data, thereby improving the hierarchy and accuracy of the evaluation logic.

[0142] Example 4

[0143] like Figure 2-Figure 3 A wind turbine performance monitoring device shown includes a sensor monitoring module, an analysis and control module, a circuit protection module and a remote interaction module;

[0144] The sensor monitoring module includes an environmental monitoring unit, a state monitoring unit, and an electrical parameter monitoring unit. The environmental monitoring unit includes an ultrasonic anemometer fixedly mounted on the outside of the top of the wind turbine generator set, which is used to detect environmental wind data; a temperature and humidity sensor fixedly mounted on the outside of the wind turbine generator set, which is used to detect environmental temperature and humidity data. In particular, the temperature and humidity sensor is designed with a radiation shield on the outside; a piezoelectric snow depth sensor fixedly mounted on the top of the wind turbine generator set, which is used to detect the thickness of snow accumulation; and a lightning current sensor fixed in the lightning protection down conductor of the power generation blade, which is used to detect the peak value data of the lightning current.

[0145] The condition monitoring unit includes a vibration sensor fixed to the outside of the gearbox and the generator set bearings for detecting abnormal vibration data; a strain gauge torque sensor attached to the outside of the generator set drive shaft for detecting the torque load data of the drive chain; a temperature sensor and a viscosity sensor fixed to the outside of the generator set oil tank for detecting the oil temperature data and oil viscosity change data of the oil tank. The sensor housings are all made of 316L stainless steel coated with a nano-hydrophobic coating.

[0146] 316L stainless steel itself has excellent corrosion resistance, especially in chloride environments. The addition of a nano-hydrophobic coating can further prevent moisture and corrosive substances from directly contacting the stainless steel surface, providing an extra layer of protection and extending the life of the equipment. The nano-hydrophobic coating can reduce the adhesion of water droplets, dust, and other contaminants to the surface. When water droplets roll off, they carry away dust and impurities on the surface, achieving a self-cleaning effect, which is particularly important for sensors that need to be kept clean to maintain efficient operation.

[0147] It should be noted that although the temperature sensor and viscosity sensor are fixed on the outside of the generator set tank, their respective probes responsible for sensing and detection are deep inside the generator set tank, so that the probes can detect the oil temperature data in the tank and the changes in oil viscosity.

[0148] The electrical parameter monitoring unit includes a current transformer and a voltage divider connected to the internal circuit of the generator set, and is used to collect current and voltage signals in the control circuit;

[0149] The analysis and control module includes a data fusion platform, a diagnostic engine layer equipped with a cloud server cluster, an adaptive pitch control unit, and an oscillation suppression unit. The data fusion platform is used to integrate various types of data collected by the sensor monitoring module and extract input feature vectors from the data set; the diagnostic engine layer is configured with a fault identification model and an equipment health assessment model; the adaptive pitch control unit is used to dynamically adjust the pitch angle and optimize the wind energy utilization coefficient when the equipment is in good operating condition and the power generation is low; the oscillation suppression unit is used to suppress tower oscillations through an active damping controller when the equipment is in good operating condition and the tower amplitude exceeds a threshold.

[0150] It is important to note that when wind turbines are operating, their towers may vibrate in strong winds or due to internal faults, which can affect the stability of the blade rotation process and, in turn, the overall power generation. Therefore, a vibration sensor is installed on the tower, and an active damper is installed at the top of the tower near the connection section of the nacelle. When the sensor determines that the tower amplitude exceeds a threshold, the active damping controller is activated to suppress the tower oscillation using the active damper. The formula is:

[0151]

[0152] in, represents the displacement of the tower in the vibration direction, represents the instantaneous speed of tower vibration, is the stiffness coefficient, is the damping coefficient;

[0153] The circuit protection module is composed of a surge protector with a three-level protection module. The first level of the surge protector is set at the main power distribution cabinet of the generator set, the second level is set at the power interface position, and the third level is set at the circuit interface end;

[0154] In addition, the circuit protection module also includes an electronic fuse that provides short-circuit and overcurrent protection, and a lightning protection grounding component. The lightning protection grounding component includes a lightning rod, a grounding wire, and an underground ground grid structure. The surge protector can protect against transient shocks such as lightning and grid operation overvoltage, while the electronic fuse can provide short-circuit and overcurrent protection for the power module, and grounding protection is performed through the lightning protection grounding component.

[0155] In addition, the remote interaction module includes a visualization interface and an audible and visual alarm unit, and both the visualization interface and the audible and visual alarm unit are electrically connected to the analysis and control module.

[0156] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for monitoring fan performance, characterized in that: include: Collect environmental data, equipment working status data, and equipment electrical data, and perform pre-processing; The environmental data includes wind speed data, ambient temperature data, ambient humidity data, snow thickness data on the top of the wind turbine, and peak value data of lightning current; The equipment working status data includes the vibration acceleration data of the gearbox, the torque load data of the transmission chain, and the oil temperature data and oil viscosity change data of the oil tank; The equipment electrical data includes the power generation current data and power generation voltage data of the generator; Perform multi-source data integration on the pre-processed equipment working status data and equipment electrical data to obtain a data set; Extract input feature vectors from the dataset; The input feature vectors include the spectrum peak of vibration acceleration, the fluctuation variance of torque, the harmonic distortion rate of current, the oil temperature stability index and the viscosity deviation; Build a fault identification model and obtain the fault status data of the equipment based on the input feature vector; The fault identification model is constructed using a support vector machine, and the fault status data includes a fault type and a confidence score; The fault type is determined by the decision function of the fault identification model The calculation is as follows: in, is the total number of input feature vectors, is the Lagrange multiplier, is the radial basis kernel function, For the input feature vectors, is the bias term, is the fault type label; The confidence score is calculated as follows: Define the radial basis kernel function : in, is the kernel parameter; Calculate the distance from the input feature vector to the hyperplane : in, is the normal vector of the hyperplane; The distance Normalize to the interval [0, 1] and define the normalized distance: in, is the maximum distance of the training set; Mapping normalized distance to confidence score , the calculation formula is: is the slope parameter; Build a health assessment model to evaluate the operating status of the equipment based on fault status data; The health assessment model is constructed as follows: Defining health indicators for: in, For the The weights of the fault types, and , MTBF is the mean time between failures; If the equipment is in good operating condition, the current power generation of the equipment is determined based on the collected data. If the current power generation is lower than the average value of the historical power generation, the pitch angle is dynamically adjusted until the wind energy utilization rate of the equipment reaches the threshold. If there is a problem with the equipment's operating status, an early warning will be issued through the sound and light alarm unit.

2. The method for monitoring fan operating performance according to claim 1, wherein: The spectrum peak of the vibration acceleration , and its calculation formula is: is the vibration acceleration data of the gearbox; The torque fluctuation variance , and its calculation formula is: in, is the timestamp, is the total time of fluctuation, For the Torque load data of the second drive chain, is the mean value of the torque load data within the time window; Harmonic distortion rate of the current , and its calculation formula is: in, is the generator’s current data, is the harmonic order, is the highest harmonic order included in the calculation; The oil temperature stability index , and its calculation formula is: in, For the The oil temperature value of the sampling time, is the mean oil temperature in the time window, is the number of sampling points in the time window; The oil viscosity change data , and its calculation formula is: in, is the current oil viscosity measurement value, It is the reference viscosity under standard working conditions.

3. A fan performance monitoring device, characterized in that: It includes sensor monitoring module, analysis and control module, circuit protection module and remote interaction module; The sensor monitoring module includes an environmental monitoring unit, a state monitoring unit and an electrical parameter monitoring unit, wherein the environmental monitoring unit includes an ultrasonic anemometer fixedly installed on the top outside of the wind turbine generator set, which is used to detect environmental wind data; a temperature and humidity sensor fixedly installed on the outside of the wind turbine generator set, which is used to detect environmental temperature and humidity data; A piezoelectric snow depth sensor fixedly installed on the top of the wind turbine is used to detect the thickness of snow accumulation; a lightning current sensor fixed in the lightning protection down conductor of the generator blade is used to detect the peak value data of the lightning current; The condition monitoring unit includes a vibration sensor fixed to the outside of the gearbox and the generator set bearing, which is used to detect abnormal vibration data; a strain gauge torque sensor attached to the outside of the generator set drive shaft, which is used to detect the torque load data of the transmission chain; a temperature sensor and a viscosity sensor fixed to the outside of the generator set oil tank, which are used to detect the oil temperature data and oil viscosity change data of the oil tank; The electrical parameter monitoring unit includes a current transformer and a voltage divider connected to the internal circuit of the generator set, and is used to collect current and voltage signals in the control circuit; The analysis and control module includes a data fusion platform, a diagnostic engine layer equipped with a cloud server cluster, an adaptive pitch control unit, and an oscillation suppression unit. The data fusion platform is used to integrate various types of data collected by the sensor monitoring module and extract input feature vectors from the data set; the diagnostic engine layer is configured with a fault identification model and an equipment health assessment model; the adaptive pitch control unit is used to dynamically adjust the pitch angle and optimize the wind energy utilization coefficient when the equipment is in good operating condition and the power generation is low; the oscillation suppression unit is used to suppress tower oscillations through an active damping controller when the equipment is in good operating condition and the tower amplitude exceeds a threshold. The circuit protection module is composed of a surge protector with a three-level protection module. The first level of the surge protector is set at the main power distribution cabinet of the generator set, the second level is set at the power interface position, and the third level is set at the circuit interface end; The remote interaction module includes a visualization interface and an audible and visual alarm unit, and both the visualization interface and the audible and visual alarm unit are electrically connected to the analysis and control module.

4. The fan operating performance monitoring device according to claim 3, characterized in that: The temperature and humidity sensor is externally designed with a radiation shield.

Images