Remote intelligent diagnosis method for fan faults

Abstract

The present invention provides a remote intelligent diagnosis method for fan faults, which relates to the field of intelligent diagnosis technology and includes: starting the target fan equipment to obtain air volume detection results; when the preset air volume requirement is not met, performing a fault detection to obtain a fault detection result, including both appearance faults and non-appearance faults; when it is a non-appearance fault, performing operation status collection to obtain an operation status collection result; transmitting to the secondary fault detection model of the remote fault processing center to obtain the secondary fault detection result; obtaining life information, including expected service life, service life, historical fault interval, and fault repair time; performing life loss analysis to obtain a life loss coefficient, compensating the secondary fault detection result, and obtaining a fault diagnosis result. The present invention solves the technical problem that traditional fault diagnosis methods cannot obtain the operating status information of fan bearings in real time and comprehensively, resulting in poor accuracy and reliability of fault diagnosis results.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent diagnosis, and in particular to a remote intelligent diagnosis method for fan faults. Background Art

[0002] Wind turbine equipment usually requires regular inspection and maintenance to ensure its normal operation and timely handling of possible faults. However, traditional inspection and maintenance methods have many problems. First, traditional inspection methods rely on manual observation and appearance inspection, which is not only time-consuming and labor-intensive, but also prone to overlooking some internal or hidden faults. Second, traditional methods cannot monitor the operating status information of wind turbine equipment in real time and can only obtain some limited data through periodic inspections. This may cause faults to be ignored or delay maintenance, leading to more serious damage. Third, traditional methods mainly rely on manual experience for fault diagnosis, which has limited accuracy. At the same time, the diagnosis of appearance faults is easy, but the diagnosis of non-appearance faults, such as vibration and temperature, is more difficult. Summary of the Invention

[0003] This application provides a remote intelligent diagnosis method for fan faults, aiming to solve the technical problem that traditional fault diagnosis methods mainly rely on visual inspection and manual maintenance, and are unable to obtain real-time and comprehensive operating status information of fan bearings, resulting in poor accuracy and reliability of fault diagnosis results.

[0004] In view of the above problems, the present application provides a remote intelligent diagnosis method for fan faults.

[0005] The first aspect disclosed in the present application provides a remote intelligent diagnosis method for fan faults, the method comprising: starting a target fan device, performing air volume detection during use, and obtaining an air volume detection result; when the air volume detection result is not satisfactory to a preset air volume requirement, connecting the image acquisition device to perform a primary fault detection on the target fan device and obtaining a primary fault detection result, wherein the primary fault detection result includes an appearance fault and a non-appearance fault; when the primary fault detection result is a non-appearance fault, connecting the sensor network to perform operating status acquisition on a target bearing and obtaining an operating status acquisition result, wherein the operating status acquisition result includes operating temperature information and operating vibration information; transmitting the operating temperature information and the operating vibration information to a secondary fault detection model of the remote fault processing center to obtain a secondary fault detection result, wherein the secondary fault detection result includes fault level information; obtaining life information of the target bearing, wherein the life information includes expected service life, service life, historical fault interval, and fault repair time; performing life loss analysis based on the life information to obtain a life loss coefficient, and compensating the secondary fault detection result based on the life loss coefficient to obtain a fault diagnosis result.

[0006] Another aspect disclosed in the present application provides a remote intelligent diagnosis system for fan faults, which is used for the above method and includes: an air volume detection module for starting a target fan device, performing air volume detection during use, and obtaining an air volume detection result; a primary fault detection module for connecting to the image acquisition device to perform a primary fault detection on the target fan device when the air volume detection result is not satisfactory to a preset air volume requirement, and obtaining a primary fault detection result, wherein the primary fault detection result includes an appearance fault and a non-appearance fault; and an operating status acquisition module for connecting to the sensor network to perform operating status acquisition on the target bearing when the primary fault detection result is a non-appearance fault, and obtaining an operating status acquisition result. Among them, the operating status collection results include working temperature information and working vibration information; the secondary fault detection module is used to transmit the working temperature information and the working vibration information to the secondary fault detection model of the remote fault processing center to obtain the secondary fault detection results, and the secondary fault detection results include fault level information; the life information acquisition module is used to obtain the life information of the target bearing, and the life information includes the expected service life, service life, historical fault interval, and fault repair time; the diagnosis result acquisition module is used to perform life loss analysis based on the life information to obtain a life loss coefficient, compensate the secondary fault detection results based on the life loss coefficient, and obtain a fault diagnosis result.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] Through multiple steps, including air volume detection, primary fault detection, and secondary fault detection, this method covers both visual and non-visual fault detection, improving the fault detection rate and accuracy. A sensor network collects information about the bearing's operating status, including operating temperature and vibration, enabling comprehensive monitoring and assessment of the fan bearing's operating condition. Life loss analysis is performed based on life information, and the life loss coefficient is calculated. This coefficient is then applied to compensate for secondary fault detection results, improving the accuracy and reliability of fault diagnosis. In summary, this method can improve the diagnostic effectiveness and reliability of fan faults, enable remote intelligent diagnosis, reduce manual inspection and maintenance costs, and improve the reliability and operational efficiency of fan equipment.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1A flow chart of a remote intelligent diagnosis method for fan faults is provided for the embodiment of the present application;

[0011] Figure 2 A schematic diagram of the structure of a remote intelligent diagnosis system for fan faults is provided for the embodiment of the present application.

[0012] Explanation of the reference numerals: air volume detection module 10 , primary fault detection module 20 , operation status acquisition module 30 , secondary fault detection module 40 , life information acquisition module 50 , diagnosis result acquisition module 60 . DETAILED DESCRIPTION

[0013] The embodiment of the present application solves the technical problem that traditional fault diagnosis methods mainly rely on appearance inspection and manual maintenance, and are unable to obtain real-time and comprehensive operating status information of fan bearings, resulting in poor accuracy and reliability of fault diagnosis results, by providing a remote intelligent diagnosis method for fan faults.

[0014] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically introduced in conjunction with the drawings in the specification.

[0015] Example 1

[0016] like Figure 1 As shown, an embodiment of the present application provides a remote intelligent diagnosis method for fan faults, which is applied to a remote intelligent diagnosis system for fan faults. The system is communicatively connected with a remote fault processing center, a sensor network, and an image acquisition device. The sensor network includes a temperature sensor and a vibration sensor. The method includes:

[0017] Start the target fan equipment, perform air volume detection during use, and obtain the air volume detection results;

[0018] The remote intelligent diagnosis method for fan faults provided in the embodiment of the present application is applied to a remote intelligent diagnosis system for fan faults, wherein the system is communicatively connected with a remote fault processing center, a sensor network, and an image acquisition device, wherein the sensor network includes a temperature sensor and a vibration sensor.

[0019] Start the target fan equipment and make it start normal operation. Use the air volume sensor to measure the air volume generated by the fan equipment in real time. Different locations can be selected for measurement to ensure that comprehensive air volume data is obtained. Based on the real-time measured data, the air volume detection result is calculated and obtained. This is a specific value that represents the current air volume level of the fan equipment. Through air volume detection, the current air volume situation of the fan equipment can be understood, thereby providing necessary data support for subsequent fault diagnosis.

[0020] When the air volume detection result is not satisfied with the preset air volume requirement, the target fan device is connected to the image acquisition device to perform a fault detection to obtain a fault detection result, wherein the fault detection result includes an appearance fault and a non-appearance fault;

[0021] Connect the image acquisition device to the target wind turbine equipment to ensure that the image information of the wind turbine equipment can be obtained. A camera or other suitable image acquisition device can be used. Use the image acquisition device to perform multi-angle image acquisition of the target wind turbine equipment, that is, capture images of the wind turbine equipment from different angles and directions to more comprehensively analyze its status. Based on the obtained image data, image processing technology is used to perform defect detection on the surface of the wind turbine equipment, including identifying and analyzing possible appearance faults, such as cracks, wear, or other visible surface problems that cause wind turbine failures. Based on the results of image processing, a fault detection result is obtained. These results include identified appearance faults and possible non-appearance faults. Appearance faults refer to problems that can be directly observed through images, while non-appearance faults may require further analysis and testing to determine.

[0022] When the primary fault detection result is a non-appearance fault, connecting the sensor network to collect the operating status of the target bearing to obtain the operating status collection result, wherein the operating status collection result includes operating temperature information and operating vibration information;

[0023] If a fault detection result is non-visual, a sensor network is connected to the target bearing to collect the operating status of the target bearing and obtain the operating status results. Specifically, the sensor network is connected to the target bearing to ensure that the bearing's operating status information can be obtained. The sensor network includes temperature sensors and vibration sensors. The sensor network collects the operating status of the target bearing in real time. The temperature sensor measures the bearing's operating temperature, and the vibration sensor monitors the bearing's operating vibration. Based on the sensor detection data, the operating status results are obtained. These results include the bearing's operating temperature and operating vibration information, reflecting the bearing's current operating condition.

[0024] transmitting the operating temperature information and the operating vibration information to a secondary fault detection model of the remote fault processing center to obtain a secondary fault detection result, wherein the secondary fault detection result includes fault level information;

[0025] Transmit operating temperature and vibration information from the target bearing's location to a remote fault processing center. This can be achieved through a network connection, data transmission protocol, or other appropriate means to ensure the reliability and security of data transmission to guarantee data integrity and privacy. Deploy and run a secondary fault detection model on the remote fault processing center. This model can be built based on machine learning algorithms and artificial intelligence technologies to analyze and process the transmitted data. Applying the secondary fault detection model to analyze the operating temperature and vibration information yields secondary fault detection results, including fault level information, which indicates the severity or priority of the bearing fault.

[0026] Acquiring life information of the target bearing, wherein the life information includes expected service life, service life, historical failure interval, and failure repair time;

[0027] Estimated service life refers to the expected service life of a bearing, estimated based on its design and specifications. This is obtained based on the manufacturer's technical parameters and relevant empirical data. Service life refers to the time a target bearing has actually been in operation. By recording the time a bearing has been in service and monitoring its operating status, the bearing's service life can be estimated. Historical failure intervals refer to the time intervals between bearing failures. By recording the time points of bearing failures, the average or statistical characteristics of these historical failure intervals can be calculated to understand the frequency and patterns of bearing failures. Fault repair time refers to the time required to repair and repair a bearing after a failure. Recording the repair time after each failure can be used to assess the reliability and repair efficiency of the bearing. By obtaining life information for a target bearing, it is possible to understand the bearing's usage, failure frequency, and repair efficiency.

[0028] A life loss analysis is performed based on the life information to obtain a life loss coefficient, and the secondary fault detection result is compensated based on the life loss coefficient to obtain a fault diagnosis result.

[0029] A life loss analysis is performed based on life information such as the expected service life, service life, and historical failure intervals of the target bearing. By comparing the actual service life with the expected service life and analyzing the historical failure intervals, the current life loss of the bearing is evaluated. Based on the results of the life loss analysis, the life loss coefficient is calculated. The life loss coefficient represents the degree of loss of the current bearing life relative to the expected service life. This coefficient can be used to quantify the remaining life and health status of the bearing.

[0030] The resulting life loss coefficient is used to compensate for the secondary fault detection results. Based on the magnitude of the life loss coefficient, the fault detection results are adjusted and corrected to reflect the actual bearing fault condition. By compensating the secondary fault detection results, a final fault diagnosis result is obtained. This result takes into account the bearing's life loss and can more accurately determine the bearing's current condition and possible faults.

[0031] Furthermore, connecting the image acquisition device to perform a fault detection on the target wind turbine device and obtaining a fault detection result includes:

[0032] Capturing multi-angle images of the target wind turbine device by the image acquisition device to obtain a set of wind turbine device images;

[0033] Surface defect detection is performed based on the wind turbine equipment image set, and the primary fault detection result is obtained based on the surface defect detection result.

[0034] Image acquisition devices, such as high-definition cameras, are used to capture images of the target wind turbine equipment from multiple angles. By capturing images at different positions, angles, and distances, comprehensive perspectives and information are obtained. Computer vision technology and image processing algorithms are used to inspect the image collection of wind turbine equipment for surface defects, including visible surface issues such as cracks, wear, and scratches. Feature extraction, image segmentation, and pattern recognition techniques are employed to compare and analyze the images with known defect patterns. The results of surface defect detection provide a primary fault detection result, indicating potential fault conditions on the surface of the wind turbine equipment. This allows for the effective detection and identification of surface defects in wind turbine equipment, providing important data support for fault diagnosis.

[0035] Furthermore, performing surface defect detection based on the wind turbine device image set includes:

[0036] Acquire a sample defect image set, perform feature extraction on the sample defect image set, and acquire a sample defect feature set;

[0037] Performing grayscale conversion on the first wind turbine device image to obtain a grayscale image of the wind turbine device;

[0038] Setting a grayscale interval set according to the sample defect feature set;

[0039] The grayscale image of the wind turbine device is segmented based on the grayscale interval set, and areas meeting the grayscale interval set are marked as defective areas.

[0040] A set of wind turbine equipment images containing different types of defects are collected as a sample defect image set. These defects may include surface problems such as cracks, wear, and scratches. Computer vision technology and image processing algorithms are used to extract features from the sample defect image set, including color features, texture features, and shape features. Through feature extraction, the feature vector of each sample defect image is obtained. These feature vectors are combined into a sample defect feature set for subsequent defect classification and identification.

[0041] A grayscale conversion algorithm is applied to the image of the first wind turbine device. This process converts the color image into a grayscale image so that the brightness value of each pixel contains only one-dimensional information. After the grayscale conversion, a grayscale image of the first wind turbine device is obtained. The image is represented by grayscale levels, which can be more conveniently processed and analyzed.

[0042] The extracted sample defect feature set is used as a reference, which contains feature vectors of different types of defects. According to the distribution of the sample defect feature set, a set of appropriate grayscale intervals is determined. These grayscale intervals cover the grayscale range of common defects so that subsequent image segmentation can accurately mark the defect area.

[0043] Based on the acquired grayscale image of the wind turbine equipment as the input for segmentation, the grayscale image of the wind turbine equipment is segmented using the set grayscale interval set, and pixels with grayscale values within the grayscale interval are marked as defective areas. By marking the pixels that meet the grayscale interval set conditions, the defective areas in the wind turbine equipment image are obtained. These defective areas can be represented by different marking methods, such as color or binarization, so that the areas that meet the expected defect type can be accurately extracted.

[0044] Further, including:

[0045] Determining whether the primary fault detection result is an appearance fault;

[0046] If so, the fault diagnosis is terminated and a fault detection result is output as the fault diagnosis result.

[0047] For the primary fault detection result, the characteristics and location of the defect area are analyzed to determine whether it is a cosmetic fault. Appearance faults typically appear on the surface of the wind turbine equipment and can be identified through image analysis. If the primary fault detection result is determined to be a cosmetic fault, the fault diagnosis is terminated and the primary fault detection result is output as the final fault diagnosis result. This indicates that the wind turbine equipment has a surface defect or externally visible problem and does not require further in-depth analysis. If the primary fault detection result is not determined to be a cosmetic fault, meaning it is not a cosmetic fault, then further in-depth fault diagnosis and analysis will be required.

[0048] Furthermore, obtaining the secondary fault detection result includes:

[0049] Retrieving and acquiring fault detection data of the wind turbine bearing through the remote fault processing center, and extracting a sample vibration information set, a sample temperature information set, and a sample fault detection result based on the fault detection data, wherein the sample fault detection result includes a sample fault level;

[0050] Constructing a vibration fault detection channel based on a neural network, and using the sample vibration information set and the sample fault detection results for training until convergence;

[0051] Constructing a temperature fault detection channel based on a neural network, and using the sample temperature information set and the sample fault detection results for training until convergence;

[0052] The vibration fault detection channel and the temperature fault detection channel are connected to obtain the secondary fault detection model.

[0053] Through the remote fault processing center, fault detection data for the fan bearings is obtained. This data includes vibration data, temperature data, and corresponding fault detection results. Vibration information of the fan bearings is extracted from the fault detection data, including the time domain characteristics, frequency domain characteristics, or statistical characteristics of the vibration signal. This information is then used to build and train the fault detection channel. Similarly, temperature information of the fan bearings is extracted from the fault detection data. This information can be bearing surface temperature data or ambient temperature data. Fault detection results of samples are obtained from the fault detection data, including the fault type and level. These results are used to train and evaluate the fault detection channel.

[0054] Select an appropriate neural network structure, such as a recurrent neural network (RNN), which will serve as the foundational model for the vibration fault detection channel. The sample vibration information set serves as the neural network's input data, and the vibration signal undergoes preprocessing, normalization, and feature extraction to better adapt it to the neural network model. The sample fault detection results serve as the neural network's output data, and the fault level is represented using methods such as one-hot encoding. The neural network model is trained using the sample vibration information set and sample fault detection results. Through backpropagation and optimization algorithms, the network parameters are continuously adjusted until the network's training error converges to an acceptable range.

[0055] By building a vibration fault detection channel based on a neural network and using sample vibration information sets and sample fault detection results for training, a model that can automatically detect fan bearing vibration faults is established. This model can be used in real-time monitoring to provide accurate vibration fault diagnosis results.

[0056] The temperature fault detection channel is constructed and trained in exactly the same way. For the sake of brevity, it will not be repeated here.

[0057] By connecting the vibration fault detection channel and the temperature fault detection channel, the output results of the two channels can be combined together through series connection, parallel connection or other methods. By connecting the outputs of the two fault detection channels, a comprehensive secondary fault detection model is obtained. This model can comprehensively consider vibration and temperature information to improve the accuracy and reliability of fault detection.

[0058] Furthermore, obtaining the secondary fault detection result also includes:

[0059] Establishing an operation status collection window, and extracting a first collection node and a second collection node according to the operation status collection window, wherein the first collection node and the second collection node are adjacent operation status collection nodes;

[0060] respectively acquiring first operating temperature information and first operating vibration information of a first acquisition node, and second operating temperature information and second operating vibration information of a second acquisition node;

[0061] An abnormal support function is introduced to calculate the temperature abnormal support of the first operating temperature information. The abnormal support function is expressed as follows:

[0062] Sup(x1,x2)=k[1+(x1-x2) e ] -1 ;

[0063] Wherein, sup(x1,x2) is the first abnormal support of the second operating temperature information x2 to the first operating temperature information x1, k is the amplitude of the abnormal support function, k is [0,1], e is the support attenuation factor, e≥0;

[0064] The vibration anomaly support degree of the first working vibration information is calculated according to the anomaly support function. When the temperature anomaly support degree and / or the vibration anomaly support degree reaches a preset temperature support degree threshold and a preset vibration support degree threshold, the first acquisition node is marked as an abnormal node and fault detection is performed.

[0065] Determine the time range of the collection window based on actual conditions and specific needs. For example, setting a 10-second collection window means collecting data continuously for 10 seconds, starting from the starting time point. Within the running status collection window, the continuous time is divided into multiple nodes, each containing corresponding data. The first and second collection nodes are selected for stacking. These two nodes are adjacent nodes, and their positions can be determined based on the time sequence.

[0066] According to the above steps, the sensor network is connected to collect the operating status of the target bearing, and the first operating status collection result of the first collection node is obtained, from which the first operating temperature information and the first operating vibration information are extracted; similarly, the second operating status collection result of the second collection node is obtained, from which the second operating temperature information and the second operating vibration information are extracted.

[0067] This anomaly support function is used to calculate the anomaly support of the second operating temperature information for the first operating temperature information. Specifically, by calculating the difference between x1 and x2, the deviation between them is obtained. This difference represents the degree of offset of the second operating temperature information relative to the first operating temperature information. The deviation is adjusted by multiplying the difference by the amplitude parameter k. The amplitude parameter k ranges from 0 to 1 and is used to control the magnitude of the anomaly support. By adjusting the amplitude parameter k and the support attenuation factor e, the magnitude and attenuation rate of the anomaly support can be controlled. The higher the anomaly support, the greater the difference between the second operating temperature information and the first operating temperature information, and the possibility of an anomaly.

[0068] By substituting the first operating temperature information and the second operating temperature information into the above-mentioned abnormal support function, the temperature abnormality support of the first operating temperature information is calculated. This value reflects the abnormality of the second operating temperature information relative to the first operating temperature information, that is, the difference between the first node and the second node. The higher the temperature abnormality support, the greater the difference between the second operating temperature information and the first operating temperature information, and the more likely an abnormal situation is.

[0069] Similar to the step of calculating the temperature anomaly support, the second working vibration information and the first working vibration information are substituted into the anomaly support function for calculation to obtain the vibration anomaly support.

[0070] Based on actual conditions and specific needs, preset temperature support thresholds and vibration support thresholds are defined. These thresholds indicate the acceptable degree of deviation. If the temperature anomaly support exceeds the preset temperature support threshold, or the vibration anomaly support exceeds the preset vibration support threshold, as long as either or both conditions are met, the first acquisition node can be marked as an abnormal node, indicating that the node is not an accidental anomaly and anomaly detection is performed on the node.

[0071] Furthermore, the life loss coefficient is obtained, including:

[0072] Calculating the ratio of the service life to the expected service life to obtain a life attenuation coefficient;

[0073] Calculate the ratio of the historical fault interval to the number of faults to obtain the average fault interval, and calculate the ratio of the fault repair time to the number of faults to obtain the average repair time;

[0074] Calculating the ratio of the mean repair time to the mean failure interval to obtain a failure loss coefficient;

[0075] The life loss coefficient is obtained by weighted summing of the life attenuation coefficient and the failure loss coefficient.

[0076] Divide the service life by the expected service life to obtain the life attenuation coefficient. This coefficient reflects the degree of attenuation of the actual service life of the equipment relative to the expected life. A higher life attenuation coefficient indicates that the actual service life of the equipment is lower than expected, and maintenance, replacement or improvement plans may need to be considered.

[0077] Based on the existing fan failure records, the interval time of each failure is calculated, and the total number of failures is counted. The sum of the historical failure intervals is divided by the total number of failures to obtain the value of the average failure interval, which can reflect the average occurrence interval time of fan failures; based on the existing fan failure records, the repair time of each failure is calculated, and the total number of failures is counted. The sum of the fault repair time is divided by the total number of failures to obtain the value of the average repair time, which can reflect the average repair time of fan failures. These indicators are used to indicate the frequency of failures and maintenance efficiency.

[0078] Divide the mean repair time by the mean failure interval to obtain the failure loss coefficient. This coefficient reflects the time required to repair a failure relative to the frequency of failures. A higher failure loss coefficient indicates a relatively longer repair time, which may cause greater losses to production and operations.

[0079] According to actual needs, different weights are given to the life attenuation coefficient and the failure loss coefficient, and then the weighted sum of the two is obtained to obtain the life loss coefficient. This coefficient comprehensively considers the impact of equipment life attenuation and failure on production and operation, and is used to evaluate the degree of equipment life loss.

[0080] Furthermore, it also includes:

[0081] When the secondary fault detection result is that a fault exists, obtaining a sample fault detection result, a sample vibration threshold set, and a sample temperature threshold set based on the fault detection data of the fan bearing;

[0082] Constructing a mapping relationship between sample fault detection results and a sample vibration threshold set, and a mapping relationship between sample fault detection results and a sample temperature threshold set;

[0083] According to the secondary fault detection results, the working vibration threshold and the working temperature threshold are mapped and obtained;

[0084] Based on the working vibration threshold and the working temperature threshold, fault locations are traversed within the operating status acquisition result to obtain a vibration fault location and a temperature fault location.

[0085] When the secondary fault detection result indicates that a fault exists, sample fault detection results are extracted from the fault detection data, which include the type and level of the fault; based on the fault detection data, vibration data of the sample is obtained, and a corresponding vibration threshold set is calculated or set according to different fault types and levels; based on the fault detection data, temperature data of the sample is obtained, and a corresponding temperature threshold set is calculated or set according to different fault types and levels.

[0086] Based on the sample fault detection results, a corresponding set of vibration thresholds is set. This mapping relationship can be established using methods such as rules, statistical analysis, or expert experience. Based on the sample fault detection results, a corresponding set of temperature thresholds is set. This mapping relationship is also established using methods such as rules, statistical analysis, or expert experience. By establishing this mapping relationship, the fault detection results can be matched with the corresponding thresholds to determine whether the fan bearing is abnormal or faulty.

[0087] Based on the secondary fault detection results, a pre-established mapping relationship is used to find the corresponding operating vibration threshold. This threshold is used to determine whether the fan vibration exceeds the normal range, thereby determining whether a vibration fault exists. Based on the secondary fault detection results, a pre-established mapping relationship is used to find the corresponding operating temperature threshold. This threshold is used to determine whether the fan temperature exceeds the normal range, thereby determining whether a temperature fault exists.

[0088] Using the operating vibration and temperature thresholds as a benchmark, the system traverses the operating status collection results, checking whether the vibration and temperature of each data point exceed the thresholds. If so, the corresponding fault location is recorded. Based on the fault locations recorded during the traversal process, the vibration and temperature fault locations are determined. These locations can indicate the specific locations of abnormal vibration or temperature in the fan bearing. This helps accurately diagnose fan bearing faults and implement appropriate repair measures, thereby improving equipment reliability and continuous operation.

[0089] In summary, the wind turbine fault remote intelligent diagnosis method provided by the embodiment of the present application has the following technical effects:

[0090] 1. Through multiple links such as air volume detection, primary fault detection and secondary fault detection, it covers the detection of appearance faults and non-appearance faults, improving the fault detection rate and accuracy;

[0091] 2. The sensor network collects information about the bearing's operating status, including operating temperature and vibration, to enable comprehensive monitoring and evaluation of the wind turbine's bearing operating conditions.

[0092] 3. Perform life loss analysis based on life information, calculate the life loss coefficient, and apply it to compensate for secondary fault detection results, thereby improving the accuracy and reliability of fault diagnosis results.

[0093] In summary, this method can improve the diagnostic effect and reliability of fan faults, realize remote intelligent diagnosis, reduce manual inspection and maintenance costs, and improve the reliability and operation efficiency of fan equipment.

[0094] Example 2

[0095] Based on the same inventive concept as the remote intelligent diagnosis method for fan faults in the above embodiment, Figure 2 As shown, the present application provides a remote intelligent diagnosis system for wind turbine faults, the system is communicatively connected with a remote fault processing center, a sensor network, and an image acquisition device, the sensor network including a temperature sensor and a vibration sensor, and the system includes:

[0096] The air volume detection module 10 is used to start the target fan equipment, perform air volume detection during use, and obtain air volume detection results;

[0097] A primary fault detection module 20 is configured to connect to the image acquisition device to perform a primary fault detection on the target fan device when the air volume detection result does not meet the preset air volume requirement, and obtain a primary fault detection result, wherein the primary fault detection result includes an appearance fault and a non-appearance fault;

[0098] An operating status acquisition module 30 is configured to connect to the sensor network to acquire operating status of the target bearing and obtain operating status acquisition results when the primary fault detection result is a non-appearance fault, wherein the operating status acquisition results include operating temperature information and operating vibration information;

[0099] a secondary fault detection module 40, configured to transmit the operating temperature information and the operating vibration information to a secondary fault detection model of the remote fault processing center to obtain a secondary fault detection result, wherein the secondary fault detection result includes fault level information;

[0100] A life information acquisition module 50 is used to acquire the life information of the target bearing, wherein the life information includes the expected service life, service life, historical failure interval, and failure repair time;

[0101] The diagnosis result acquisition module 60 is configured to perform a life loss analysis based on the life information to obtain a life loss coefficient, and compensate the secondary fault detection result based on the life loss coefficient to obtain a fault diagnosis result.

[0102] Furthermore, the system further includes a fault detection result acquisition module to perform the following operation steps:

[0103] Capturing multi-angle images of the target wind turbine device by the image acquisition device to obtain a set of wind turbine device images;

[0104] Surface defect detection is performed based on the wind turbine equipment image set, and the primary fault detection result is obtained based on the surface defect detection result.

[0105] Furthermore, the system further includes a defect area acquisition module to perform the following operation steps:

[0106] Acquire a sample defect image set, perform feature extraction on the sample defect image set, and acquire a sample defect feature set;

[0107] Performing grayscale conversion on the first wind turbine device image to obtain a grayscale image of the wind turbine device;

[0108] Setting a grayscale interval set according to the sample defect feature set;

[0109] The grayscale image of the wind turbine device is segmented based on the grayscale interval set, and areas meeting the grayscale interval set are marked as defective areas.

[0110] Furthermore, the system further includes a primary fault detection result judgment module to perform the following operation steps:

[0111] Determining whether the primary fault detection result is an appearance fault;

[0112] If so, the fault diagnosis is terminated and a fault detection result is output as the fault diagnosis result.

[0113] Furthermore, the system further includes a secondary fault detection model acquisition module to perform the following operation steps:

[0114] Retrieving and acquiring fault detection data of the wind turbine bearing through the remote fault processing center, and extracting a sample vibration information set, a sample temperature information set, and a sample fault detection result based on the fault detection data, wherein the sample fault detection result includes a sample fault level;

[0115] Constructing a vibration fault detection channel based on a neural network, and using the sample vibration information set and the sample fault detection results for training until convergence;

[0116] Constructing a temperature fault detection channel based on a neural network, and using the sample temperature information set and the sample fault detection results for training until convergence;

[0117] The vibration fault detection channel and the temperature fault detection channel are connected to obtain the secondary fault detection model.

[0118] Furthermore, the system further includes a fault detection and analysis module to perform the following steps:

[0119] Establishing an operation status collection window, and extracting a first collection node and a second collection node according to the operation status collection window, wherein the first collection node and the second collection node are adjacent operation status collection nodes;

[0120] respectively acquiring first operating temperature information and first operating vibration information of a first acquisition node, and second operating temperature information and second operating vibration information of a second acquisition node;

[0121] An abnormal support function is introduced to calculate the temperature abnormal support of the first operating temperature information. The abnormal support function is expressed as follows:

[0122] Sup(x1,x2)=k[1+(x1-x2) e ] -1 ;

[0123] Wherein, sup(x1,x2) is the first abnormal support of the second operating temperature information x2 to the first operating temperature information x1, k is the amplitude of the abnormal support function, k is [0,1], e is the support attenuation factor, e≥0;

[0124] The vibration anomaly support degree of the first working vibration information is calculated according to the anomaly support function. When the temperature anomaly support degree and / or the vibration anomaly support degree reaches a preset temperature support degree threshold and a preset vibration support degree threshold, the first acquisition node is marked as an abnormal node and fault detection is performed.

[0125] Furthermore, the system further includes a life loss coefficient acquisition module to perform the following operation steps:

[0126] Calculating the ratio of the service life to the expected service life to obtain a life attenuation coefficient;

[0127] Calculate the ratio of the historical fault interval to the number of faults to obtain the average fault interval, and calculate the ratio of the fault repair time to the number of faults to obtain the average repair time;

[0128] Calculating the ratio of the mean repair time to the mean failure interval to obtain a failure loss coefficient;

[0129] The life loss coefficient is obtained by weighted summing of the life attenuation coefficient and the failure loss coefficient.

[0130] Furthermore, the system further includes a fault location acquisition module to perform the following operation steps:

[0131] When the secondary fault detection result is that a fault exists, obtaining a sample fault detection result, a sample vibration threshold set, and a sample temperature threshold set based on the fault detection data of the fan bearing;

[0132] Constructing a mapping relationship between sample fault detection results and a sample vibration threshold set, and a mapping relationship between sample fault detection results and a sample temperature threshold set;

[0133] According to the secondary fault detection results, the working vibration threshold and the working temperature threshold are mapped and obtained;

[0134] Based on the working vibration threshold and the working temperature threshold, fault locations are traversed within the operating status acquisition result to obtain a vibration fault location and a temperature fault location.

[0135] Through the above detailed description of the remote intelligent diagnosis method for fan faults in this specification, those skilled in the art can clearly understand the remote intelligent diagnosis method for fan faults in this embodiment. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0136] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A remote intelligent diagnosis method for fan faults, characterized in that: Applied to a remote intelligent diagnosis system for fan faults, the system is communicatively connected to a remote fault processing center, a sensor network, and an image acquisition device. The sensor network includes a temperature sensor and a vibration sensor. The method includes: Start the target fan equipment, perform air volume detection during use, and obtain the air volume detection results; When the air volume detection result is not satisfied with the preset air volume requirement, the target fan device is connected to the image acquisition device to perform a fault detection to obtain a fault detection result, wherein the fault detection result includes an appearance fault and a non-appearance fault; When the primary fault detection result is a non-appearance fault, connecting the sensor network to collect the operating status of the target bearing to obtain the operating status collection result, wherein the operating status collection result includes operating temperature information and operating vibration information; transmitting the operating temperature information and the operating vibration information to a secondary fault detection model of the remote fault processing center to obtain a secondary fault detection result, wherein the secondary fault detection result includes fault level information; Acquiring life information of the target bearing, wherein the life information includes expected service life, service life, historical failure interval, and failure repair time; performing a life loss analysis based on the life information to obtain a life loss coefficient, and compensating the secondary fault detection result based on the life loss coefficient to obtain a fault diagnosis result; The method of connecting the image acquisition device to perform a fault detection on the target wind turbine device and obtaining a fault detection result includes: Capturing multi-angle images of the target wind turbine device by the image acquisition device to obtain a set of wind turbine device images; Performing surface defect detection based on the wind turbine device image set, and obtaining the primary fault detection result based on the surface defect detection result; Obtaining secondary fault detection results includes: Retrieving and acquiring fault detection data of the wind turbine bearing through the remote fault processing center, and extracting a sample vibration information set, a sample temperature information set, and a sample fault detection result based on the fault detection data, wherein the sample fault detection result includes a sample fault level; Constructing a vibration fault detection channel based on a neural network, and using the sample vibration information set and the sample fault detection results for training until convergence; Constructing a temperature fault detection channel based on a neural network, and using the sample temperature information set and the sample fault detection results for training until convergence; The vibration fault detection channel and the temperature fault detection channel are connected to obtain the secondary fault detection model.

2. The method according to claim 1, wherein Performing surface defect detection based on the wind turbine equipment image set includes: Acquire a sample defect image set, perform feature extraction on the sample defect image set, and acquire a sample defect feature set; Performing grayscale conversion on the first wind turbine device image to obtain a grayscale image of the wind turbine device; Setting a grayscale interval set according to the sample defect feature set; The grayscale image of the wind turbine device is segmented based on the grayscale interval set, and areas meeting the grayscale interval set are marked as defective areas.

3. The method according to claim 1, wherein include: Determining whether the primary fault detection result is an appearance fault; If so, the fault diagnosis is terminated and a fault detection result is output as the fault diagnosis result.

4. The method according to claim 1, wherein Obtaining secondary fault detection results also includes: Establishing an operation status collection window, and extracting a first collection node and a second collection node according to the operation status collection window, wherein the first collection node and the second collection node are adjacent operation status collection nodes; respectively acquiring first operating temperature information and first operating vibration information of a first acquisition node, and second operating temperature information and second operating vibration information of a second acquisition node; An abnormal support function is introduced to calculate the temperature abnormal support of the first operating temperature information. The abnormal support function is expressed as follows: ; in, Second operating temperature information First operating temperature information The first abnormal support of is the amplitude of the anomaly support function, Pick , To support the attenuation factor, ; The vibration anomaly support degree of the first working vibration information is calculated according to the anomaly support function. When the temperature anomaly support degree and / or the vibration anomaly support degree reaches a preset temperature support degree threshold and a preset vibration support degree threshold, the first acquisition node is marked as an abnormal node and fault detection is performed.

5. The method according to claim 1, wherein Obtain life loss factors, including: Calculating the ratio of the service life to the expected service life to obtain a life attenuation coefficient; Calculate the ratio of the historical fault interval to the number of faults to obtain the average fault interval, and calculate the ratio of the fault repair time to the number of faults to obtain the average repair time; Calculating the ratio of the mean repair time to the mean failure interval to obtain a failure loss coefficient; The life loss coefficient is obtained by weighted summing of the life attenuation coefficient and the failure loss coefficient.

6. The method according to claim 1, wherein Also includes: When the secondary fault detection result is that a fault exists, obtaining a sample fault detection result, a sample vibration threshold set, and a sample temperature threshold set based on the fault detection data of the fan bearing; Constructing a mapping relationship between sample fault detection results and a sample vibration threshold set, and a mapping relationship between sample fault detection results and a sample temperature threshold set; According to the secondary fault detection results, the working vibration threshold and the working temperature threshold are mapped and obtained; Based on the working vibration threshold and the working temperature threshold, fault locations are traversed within the operating status acquisition result to obtain a vibration fault location and a temperature fault location.

7. The remote intelligent diagnosis system for fan faults is characterized by: The system is communicatively connected to a remote fault processing center, a sensor network, and an image acquisition device. The sensor network includes a temperature sensor and a vibration sensor, and is used to implement the remote intelligent diagnosis method for wind turbine faults according to any one of claims 1 to 6, including: The air volume detection module is used to start the target fan equipment, perform air volume detection during use, and obtain air volume detection results; A primary fault detection module, configured to connect to the image acquisition device to perform a primary fault detection on the target fan equipment when the air volume detection result is not satisfactory to the preset air volume requirement, and obtain a primary fault detection result, wherein the primary fault detection result includes an appearance fault and a non-appearance fault; an operating status acquisition module, configured to connect to the sensor network to acquire operating status of the target bearing and obtain operating status acquisition results when the primary fault detection result is a non-appearance fault, wherein the operating status acquisition results include operating temperature information and operating vibration information; a secondary fault detection module, configured to transmit the operating temperature information and the operating vibration information to a secondary fault detection model of the remote fault processing center to obtain a secondary fault detection result, wherein the secondary fault detection result includes fault level information; A life information acquisition module is used to obtain the life information of the target bearing, wherein the life information includes the expected service life, service life, historical failure interval, and failure repair time; a diagnosis result acquisition module, configured to perform a life loss analysis based on the life information to obtain a life loss coefficient, and compensate the secondary fault detection result based on the life loss coefficient to obtain a fault diagnosis result; The method of connecting the image acquisition device to perform a fault detection on the target wind turbine device and obtaining a fault detection result includes: Capturing multi-angle images of the target wind turbine device by the image acquisition device to obtain a set of wind turbine device images; Performing surface defect detection based on the wind turbine device image set, and obtaining the primary fault detection result based on the surface defect detection result; Obtaining secondary fault detection results includes: Retrieving and acquiring fault detection data of the wind turbine bearing through the remote fault processing center, and extracting a sample vibration information set, a sample temperature information set, and a sample fault detection result based on the fault detection data, wherein the sample fault detection result includes a sample fault level; Constructing a vibration fault detection channel based on a neural network, and using the sample vibration information set and the sample fault detection results for training until convergence; Building a temperature fault detection channel based on a neural network, and using the sample temperature information set and the sample fault detection results for training until convergence; The vibration fault detection channel and the temperature fault detection channel are connected to obtain the secondary fault detection model.

Images