Wind turbine component temperature fault detection method, system and wind turbine generator set

Abstract

The present invention provides a method, system and wind turbine generator set for detecting temperature faults of fan components, belonging to the technical field of fan faults. The method comprises: obtaining a first predicted temperature of the fan component based on data of a first parameter and first historical data within a target time period; obtaining a second predicted temperature of the fan component based on data of a second parameter and second historical data within the target time period; obtaining a third predicted temperature based on the first predicted temperature and the second predicted temperature; subtracting the third predicted temperature from the actual temperature of the fan component within the target time period to obtain a residual; when the residual is greater than a preset fault threshold, determining that a temperature fault exists in the fan component. The present invention predicts the first predicted temperature and the second predicted temperature of the target time period in two different ways. Compared with a method of obtaining predicted temperature values ​​in only one way, the accuracy of the predicted temperature values ​​is improved, thereby improving the robustness and accuracy of detecting temperature faults of fan components.

Description

Technical Field

[0001] The present invention relates to the technical field of fan faults, and in particular to a method and system for detecting temperature faults of fan components and a wind turbine generator set. Background Art

[0002] Wind turbines are renewable energy generation equipment designed to efficiently convert wind energy into electricity. Their reliability and performance are crucial to the sustainable development of the energy industry. However, during operation, large components such as gearboxes and main shafts are prone to temperature-related faults. These faults can lead to performance degradation or even damage. Establishing effective temperature-related fault detection methods is essential for the safe operation of wind turbines.

[0003] Currently, temperature-related fault detection methods in the industry all use a single method to predict temperature, and then perform fault detection based on the predicted temperature. Using a single temperature prediction method can result in inaccurate temperature prediction results, which in turn leads to inaccurate temperature-related fault detection results. Summary of the Invention

[0004] The present invention provides a method, system and wind turbine generator set for detecting temperature faults of fan components, so as to solve the defects of the temperature fault detection method of fan components in the prior art, such as unstable fault detection and low prediction accuracy, and realize high-precision prediction of temperature faults.

[0005] The present invention provides a method for detecting temperature faults of fan components, comprising:

[0006] Obtaining a first predicted temperature of the fan component based on data of a first parameter and first historical data within a target period; the first parameter being a parameter in a heat dissipation path of the fan component that is nonlinearly related to temperature; and the first historical data including historical data of the first parameter and historical temperature data of the fan component;

[0007] Obtaining a second predicted temperature of the fan component based on data of a second parameter within a target period and second historical data; the second parameter is a parameter of the fan component that is nonlinearly related to temperature; and the second historical data includes historical data of the second parameter and historical temperature data of the fan component;

[0008] Obtaining a third predicted temperature based on the first predicted temperature and the second predicted temperature;

[0009] Subtracting the third predicted temperature from the actual temperature of the fan component during the target period to obtain a residual;

[0010] When the residual is greater than a preset fault threshold, it is determined that a temperature fault exists in the fan component.

[0011] According to a method for detecting temperature faults of fan components provided by the present invention, obtaining a first predicted temperature of the fan component based on first characteristic data and first historical data during a target period specifically includes:

[0012] The data of the first parameter within the target time period is input into an equivalent heat circuit model with determined hyperparameters to obtain the first predicted temperature; the equivalent heat circuit model with determined hyperparameters is obtained using the following method:

[0013] Acquiring the first historical data;

[0014] According to the heat dissipation path, an equivalent thermal circuit model based on physical mechanism is constructed;

[0015] A hyperparameter search is performed on the equivalent thermal circuit model according to the first historical data to obtain an equivalent thermal circuit model with determined hyperparameters.

[0016] According to a method for detecting temperature faults of fan components provided by the present invention, a hyperparameter search is performed on the equivalent thermal circuit model based on the first historical data to obtain an equivalent thermal circuit model with determined hyperparameters, specifically comprising:

[0017] According to the first historical data, a particle swarm algorithm is used to perform a hyperparameter search on the equivalent thermal circuit model to obtain an equivalent thermal circuit model with determined hyperparameters.

[0018] According to a method for detecting temperature faults of fan components provided by the present invention, after obtaining the first historical data and before performing a hyperparameter search on the equivalent thermal circuit model based on the first historical data, the method for obtaining the equivalent thermal circuit model having determined the hyperparameters further includes:

[0019] The first historical data is preprocessed; the data preprocessing includes missing value filling and data cleaning.

[0020] According to a method for detecting temperature faults of fan components provided by the present invention, a second predicted temperature of the fan component is obtained based on data of a second parameter within a target period and second historical data, specifically comprising:

[0021] The second characteristic data of the target time period is input into the trained time series neural network model to obtain the second predicted temperature; the trained time series neural network model is obtained using the following method:

[0022] obtaining the second historical data;

[0023] The historical data of the second parameter in the second historical data is used as input, and the temperature historical data is used as a label to train the time series neural network model to obtain the trained time series neural network model.

[0024] According to a method for detecting temperature faults of fan components provided by the present invention, after obtaining the second historical data and before training the time series neural network model, the method for obtaining the trained time series neural network model further includes:

[0025] Automatically screening the second historical data for features using a chi-square test-based method;

[0026] Data preprocessing is performed on the second historical data after automatic feature screening; the data preprocessing includes missing value filling and data cleaning.

[0027] According to a method for detecting temperature faults of fan components provided by the present invention, obtaining a third predicted temperature based on the first predicted temperature and the second predicted temperature specifically includes:

[0028] The first predicted temperature and the second predicted temperature are subjected to weighted average probability voting to obtain a weighted temperature prediction result; the weight of the weighted average probability voting is set using an adaptive weight method.

[0029] According to a method for detecting temperature faults of fan components provided by the present invention, the preset fault threshold is determined based on the first historical data and historical fault temperature values ​​using the Laida criterion.

[0030] The present invention also provides a fan component temperature fault detection system, comprising:

[0031] a first predicted temperature acquisition module, configured to obtain a first predicted temperature of the fan component based on data of a first parameter and first historical data within a target period; the first parameter being a parameter in a heat dissipation path of the fan component that is nonlinearly related to temperature; and the first historical data including historical data of the first parameter and historical temperature data of the fan component;

[0032] a second predicted temperature acquisition module, configured to obtain a second predicted temperature of the fan component based on data of a second parameter and second historical data within a target period; the second parameter being a parameter of the fan component that is nonlinearly related to temperature; and the second historical data including historical data of the second parameter and historical temperature data of the fan component;

[0033] a third predicted temperature acquisition module, configured to obtain a third predicted temperature according to the first predicted temperature and the second predicted temperature;

[0034] a residual obtaining module, configured to obtain a residual by subtracting the third predicted temperature from the actual temperature of the fan component within the target period;

[0035] The fault determination module is used to determine that a temperature fault exists in the fan component when the residual is greater than a preset fault threshold.

[0036] The present invention also provides a wind turbine generator set, comprising the above-mentioned wind turbine component temperature fault detection system.

[0037] The present invention provides a method, system and wind turbine generator set for detecting temperature faults of fan components. First, based on the data of the first parameter and the first historical data in the target time period, a first predicted temperature of the fan component is obtained; and based on the data of the second parameter and the second historical data in the target time period, a second predicted temperature of the fan component is obtained; then, based on the first predicted temperature and the second predicted temperature, a third predicted temperature is obtained; the third predicted temperature is subtracted from the actual temperature of the fan component in the target time period to obtain a residual; when the residual is greater than a preset fault threshold, it is determined that a temperature fault exists in the fan component. The present invention predicts the first predicted temperature and the second predicted temperature of the target time period respectively in two different ways, calculates the final third predicted temperature value based on the predicted two temperature values, and then uses the third predicted temperature value to perform temperature fault detection. Compared with the method of obtaining the predicted temperature value in only one way, the present application scheme improves the accuracy of the predicted temperature value, thereby improving the robustness and accuracy of the detection of temperature faults of fan components. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 It is a flow chart of a method for detecting temperature faults of fan components provided by the present invention;

[0040] Figure 2 It is a structural schematic diagram of the equivalent thermal circuit model provided by the present invention;

[0041] Figure 3 It is a structural diagram of the fan component temperature fault detection system provided by the present invention;

[0042] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0043] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0044] The following combination Figure 1-Figure 4 The present invention describes a method and system for detecting temperature faults of wind turbine components and a wind turbine generator set.

[0045] Figure 1 The flowchart of the method for detecting temperature faults of fan components provided by the present invention is illustrated;

[0046] It should be noted that the fan components mentioned in the fan component temperature fault detection method provided by the present invention refer to large fan components, such as the gearbox high-speed shaft, main shaft or generator, that is, the provided fan component temperature fault detection method can be applied to any large fan component.

[0047] like Figure 1 As shown, the present invention provides a method for detecting temperature faults of fan components, comprising:

[0048] 100. Obtain a first predicted temperature of a fan component based on data of a first parameter within a target period and first historical data.

[0049] The first parameter is a parameter in the heat dissipation path of the fan component that is nonlinearly related to temperature; the first historical data includes the historical data of the first parameter and the historical temperature data of the fan component. For example, when predicting a temperature-related fault of a gearbox high-speed shaft, the parameters linearly related to the gearbox high-speed shaft temperature are the gearbox oil temperature and the gearbox low-speed shaft temperature. Therefore, the parameters in the heat dissipation path of the gearbox high-speed shaft other than the aforementioned linearly related parameters are parameters nonlinearly related to temperature. Therefore, the first parameter is the parameter in the heat dissipation path of the gearbox high-speed shaft other than the gearbox oil temperature and the gearbox low-speed shaft temperature.

[0050] The historical data mentioned here refers to the data of the period before the target period. The first predicted temperature is the first predicted temperature within the target period.

[0051] The present invention can obtain the relationship between the temperature of the fan component and the parameters in the heat dissipation path of the fan component that are nonlinearly related to temperature based on the first historical data, and then calculate the first predicted temperature of the fan component within the target time period based on the relationship and the data of the first parameter within the target time period.

[0052] 200. Obtain a second predicted temperature of the fan component based on the data of the second parameter within the target period and the second historical data.

[0053] The second parameter is a parameter related to temperature nonlinearity in the wind turbine component; the second historical data includes historical data of the second parameter and historical temperature data of the wind turbine component. The second historical data may include the first historical data and, in addition, parameters related to temperature nonlinearity other than the heat dissipation path of the wind turbine component. For example, when detecting a temperature-related fault on a gearbox high-speed shaft, the second historical data may include time, wind speed, grid power, active power, cabin temperature, grid voltage, and generator speed.

[0054] The first historical data and the second historical data of the present invention can both use second-level data in a SCADA (Supervisory Control And Data Acquisition) system.

[0055] The present invention can obtain the relationship between the temperature of the fan component and the parameter nonlinearly related to temperature through the second historical data, and then obtain the second predicted temperature of the fan component in the target period based on the relationship and the data of the second parameter in the target period.

[0056] 300. Obtain a third predicted temperature according to the first predicted temperature and the second predicted temperature.

[0057] Specifically, the third predicted temperature may be calculated based on the first predicted temperature and the second predicted temperature by adopting a weighted average probability voting method, or the third predicted temperature may be calculated by adopting a method of calculating the average of the first predicted temperature and the second predicted temperature.

[0058] 400. Subtract the third predicted temperature from the actual temperature of the fan component during the target period to obtain a residual.

[0059] 500. When the residual is greater than a preset fault threshold, it is determined that a temperature fault exists in the fan component.

[0060] The method for detecting temperature faults of fan components provided by the present invention predicts the first predicted temperature and the second predicted temperature of the target time period respectively in two different ways, calculates the final third predicted temperature value based on the two predicted temperature values, and then uses the third predicted temperature value to perform temperature fault detection. Compared with the method of obtaining the predicted temperature value by only one method, the present application scheme improves the accuracy of the predicted temperature value, thereby improving the robustness and accuracy of the detection of temperature faults of fan components.

[0061] As an optional implementation manner, obtaining a first predicted temperature of a fan component according to the first characteristic data and the first historical data in the target period specifically includes:

[0062] The data of the first parameter in the target period is input into the equivalent heat circuit model with the hyperparameters determined to obtain the first predicted temperature. The equivalent heat circuit model with the hyperparameters determined is obtained using the following method:

[0063] Acquire first historical data; construct an equivalent thermal circuit model based on a physical mechanism according to the heat dissipation path; perform a hyperparameter search on the equivalent thermal circuit model according to the first historical data to obtain an equivalent thermal circuit model with determined hyperparameters.

[0064] For example, in the specific implementation process, when the fan component is the fan main shaft, the equivalent thermal circuit model constructed is as follows: Figure 2 As shown, according to Figure 2 , and combined with the circuit node voltage method, the equations of the equivalent thermal circuit model are listed:

[0065]

[0066] Convert the above equations into a system of state equations:

[0067]

[0068] Here, T1 represents the grease temperature, T2 represents the spindle surface temperature, T3 represents the cabin temperature, R1, R2, and R3 represent the thermal resistance between components, and C1, C2, and C3 represent the heat capacity. These parameters are related to the unit characteristics and layout and require independent modeling for each unit. Pg represents thermal power loss (friction loss). Figure 2 The meanings of the parameters in are the same as those in the above formula.

[0069] The above state equations are the equivalent thermal circuit model.

[0070] In order to determine the hyperparameters in the state equation group (equivalent thermal circuit model) and adjust the differences between different units, the hyperparameters R, C, etc. in the equivalent thermal circuit model are searched through the first historical data, specifically through the mapping relationship between the fan component temperature and parameters related to temperature nonlinearity in the first historical data, to obtain an equivalent thermal circuit model with determined hyperparameters.

[0071] The hyperparameter search of the equivalent thermal circuit model can use the mapping relationship in 80% of the first historical data for search, and 20% of the data is used for testing and subsequent threshold or weight setting.

[0072] After obtaining the equivalent thermal circuit model with determined hyperparameters, the data of the first parameter in the target time period is input into the model to obtain the first predicted temperature.

[0073] The present invention uses two methods for predicting fan component heat dissipation. The first method constructs an equivalent thermal circuit model based on physical mechanisms, based on the heat dissipation paths of the fan components. Specifically, the first method employs an equivalent mechanism model based on thermodynamic principles. This model ensures the equivalence of fundamental fan thermodynamic principles and the rationality of the modeling. Simultaneously, based on the first historical data, the hyperparameters of the equivalent thermal circuit model are appropriately adjusted to accommodate differences in unit characteristics and improve prediction accuracy.

[0074] In some embodiments, a method for performing hyperparameter search on an equivalent thermal circuit model specifically includes:

[0075] According to the first historical data, a particle swarm algorithm is used to search for hyperparameters of the equivalent thermal circuit model, and an equivalent thermal circuit model with determined hyperparameters is obtained.

[0076] The present invention adopts a particle swarm algorithm to search for hyperparameters of an equivalent thermal circuit model, thereby improving the speed of the hyperparameter search.

[0077] In some embodiments, after obtaining the first historical data, before performing a hyperparameter search on the equivalent thermal circuit model based on the first historical data, the method for obtaining the equivalent thermal circuit model with determined hyperparameters further includes: performing data preprocessing on the first historical data.

[0078] Among them, data preprocessing includes missing value filling and data cleaning.

[0079] Specifically, missing columns can be filled with the default value 0, and missing rows of data can be filled according to the field type: if the field type is a continuous value, linear filling is performed; if the field type is a status bit, front filling is performed.

[0080] Data cleaning may include: performing data cleaning according to temperature range and data feature rules, removing constant values ​​and out-of-limit abnormal values.

[0081] After data cleaning, if the field type is continuous value, normalization is performed; if the field type is discrete data (state bit), one-hot encoding is performed.

[0082] The present invention improves the quality of the data by preprocessing the first historical data, thereby helping to improve the accuracy of the equivalent thermal circuit model whose hyperparameters are subsequently determined.

[0083] As an optional implementation, the second predicted temperature of the fan component is obtained based on the data of the second parameter in the target time period and the second historical data, specifically including: inputting the second characteristic data of the target time period into the trained time series neural network model to obtain the second predicted temperature.

[0084] The trained time series neural network model is obtained using the following method:

[0085] Obtain second historical data; use the historical data of the second parameter in the second historical data as input and the temperature historical data as a label to train the time series neural network model to obtain a trained time series neural network model.

[0086] The time series neural network model is the second method used in this invention to predict the temperature during the target period. The backbone network of the time series neural network model is a residual convolutional neural network. The main structure includes several layers of residual convolution modules, pooling layers, normalization layers, fully connected layers, and ReLU activation function layers.

[0087] The present invention adopts the mapping relationship of the parameters in the second historical data for modeling. For example, 80% of the second historical data in the past 30 days may be extracted for training, and 20% may be extracted for testing, correction and threshold setting.

[0088] The input for training the time series neural network model is a multivariate time series parameter (i.e., the second parameter) that is nonlinearly related to the temperature of the fan components. The output is historical temperature data, extracted using a sliding window approach, where S is the window size. The time series neural network model uses backpropagation for parameter updates, using the Adam optimizer, and the training loss function is the root mean square error (RMSE).

[0089] RMSE=sqrt(Σ(y i _true-y i _predict) 2 / N)

[0090] Where yi_true represents the true temperature value of the i-th point in the current time series, yi_predict represents the predicted temperature value of the i-th point in the current time series, and N represents the number of sequence points.

[0091] After the training of the time series neural network model is completed, the second characteristic data of the target period is input into the model to obtain the second predicted temperature of the fan component in the target period.

[0092] Of the two methods for predicting fan component failures in this paper, the first method uses historical data from the fan component's heat dissipation path, excluding data outside the heat dissipation path. Therefore, the data is incomplete. To address this, the second method uses a time series neural network model to learn high-dimensional implicit information from the time series to supplement the data missing in the first method and further improve prediction accuracy.

[0093] The present invention adopts a time series neural network model for time series modeling, which enhances the model's ability to extract high-order data information and improves temperature prediction accuracy.

[0094] As an optional implementation, after obtaining the second historical data and before training the time series neural network model, the method for obtaining the trained time series neural network model further includes:

[0095] Automatic feature screening is performed on the second historical data using a chi-square test-based method; data preprocessing is performed on the second historical data after the automatic feature screening; wherein the data preprocessing includes missing value filling and data cleaning.

[0096] The present invention adds automatic feature screening of timing fault features, thereby improving the efficiency of fault feature screening.

[0097] It should be noted that the method for preprocessing the second historical data is the same as the method for preprocessing the first historical data. The specific content can be referred to the data preprocessing method for the first historical data, which will not be repeated here.

[0098] In some embodiments, obtaining a third predicted temperature based on the first predicted temperature and the second predicted temperature specifically includes performing a weighted average probability vote on the first predicted temperature and the second predicted temperature to obtain a weighted temperature prediction result, wherein the weight of the weighted average probability vote is set using an adaptive weighting method.

[0099] In the specific implementation process, the root mean square error (RMSE) of the actual temperature and the predicted temperature can be calculated on the test data of the two different methods mentioned above. The larger the RMSE, the smaller the weight, and the sum of the weights of the two models is 1.

[0100] The present invention obtains the predicted temperature of the fan component in two different ways and combines the weighted average probability voting method to obtain the final predicted temperature, thereby improving the accuracy of the fan component temperature prediction result within the target period.

[0101] In some embodiments, in step 500 , the preset fault threshold is determined based on the first historical data and historical fault temperature values ​​using the Laida criterion.

[0102] In the present invention, a preset fault threshold is determined based on the first historical data and the historical fault temperature value and using the Laida criterion, thereby improving the accuracy of the fault threshold and further improving the accuracy of the temperature fault detection result.

[0103] The fan component temperature fault detection system provided by the present invention is described below. The fan component temperature fault detection system described below and the fan component temperature fault detection method described above can be referenced to each other.

[0104] Figure 3The schematic diagram of the structure of the fan component temperature fault detection system provided by the present invention is illustrated.

[0105] like Figure 3 As shown, the fan component temperature fault detection system provided by the present invention includes:

[0106] A first predicted temperature acquisition module 201 is configured to obtain a first predicted temperature of the fan component based on data of a first parameter and first historical data within a target period; the first parameter being a parameter in a heat dissipation path of the fan component that is nonlinearly related to temperature; and the first historical data including historical data of the first parameter and historical temperature data of the fan component.

[0107] A second predicted temperature acquisition module 202 is configured to obtain a second predicted temperature of the fan component based on data of a second parameter and second historical data within a target period; the second parameter is a parameter of the fan component that is nonlinearly related to temperature; and the second historical data includes historical data of the second parameter and historical temperature data of the fan component.

[0108] A third predicted temperature acquisition module 203 is configured to obtain a third predicted temperature based on the first predicted temperature and the second predicted temperature;

[0109] A residual obtaining module 204 is configured to obtain a residual by subtracting the third predicted temperature from the actual temperature of the fan component within the target period;

[0110] The fault determination module 205 is configured to determine that a temperature fault exists in the fan component when the residual error is greater than a preset fault threshold.

[0111] As an optional implementation manner, the first predicted temperature acquisition module 201 is specifically configured to input the data of the first parameter within the target time period into the equivalent thermal circuit model with determined hyperparameters to obtain the first predicted temperature.

[0112] In the first predicted temperature acquisition module 201 , an equivalent thermal circuit model with hyperparameters is determined, which is obtained by performing a hyperparameter search on the equivalent thermal circuit model. The equivalent thermal circuit model is constructed based on the heat dissipation path.

[0113] The above-mentioned fan component temperature fault detection system also includes:

[0114] The first historical data acquisition module is used to acquire first historical data.

[0115] The equivalent thermal circuit model construction module is used to construct an equivalent thermal circuit model based on physical mechanisms according to the heat dissipation path.

[0116] The hyperparameter determination module is used to perform a hyperparameter search on the equivalent thermal circuit model according to the first historical data to obtain an equivalent thermal circuit model with determined hyperparameters.

[0117] In some embodiments, the hyperparameter determination module is specifically configured to perform a hyperparameter search on the equivalent thermal circuit model using a particle swarm algorithm based on the first historical data to obtain an equivalent thermal circuit model with determined hyperparameters.

[0118] In some embodiments, the fan component temperature fault detection system further includes:

[0119] The first data preprocessing module is configured to perform data preprocessing on the first historical data after acquiring the first historical data and before performing a hyperparameter search on the equivalent thermal circuit model based on the first historical data, wherein the data preprocessing includes missing value filling and data cleaning.

[0120] As an optional implementation, the second predicted temperature acquisition module 202 is specifically configured to input the second characteristic data of the target period into the trained time series neural network model to obtain the second predicted temperature.

[0121] The fan component temperature fault detection system also includes:

[0122] The second historical data acquisition module is used to acquire second historical data.

[0123] The model training module is used to train the time series neural network model using the historical data of the second parameter in the second historical data as input and the temperature historical data as a label to obtain a trained time series neural network model.

[0124] In some embodiments, the fan component temperature fault detection system further includes: a feature screening module for automatically screening features of the second historical data using a chi-square test-based method after acquiring the second historical data and before training the time series neural network model.

[0125] The second data preprocessing module is used to perform data preprocessing on the second historical data after automatic feature screening, wherein the data preprocessing includes missing value filling and data cleaning.

[0126] As an optional embodiment, the third predicted temperature acquisition module 203 is specifically configured to perform weighted average probability voting on the first predicted temperature and the second predicted temperature to obtain a weighted temperature prediction result, wherein the weight of the weighted average probability voting is set using an adaptive weighting method.

[0127] The fault threshold preset in the fault determination module 205 is set based on the first historical data and the historical fault temperature values ​​using the Laida criterion.

[0128] As an optional implementation, the fan component temperature fault detection system further includes a fault threshold setting module, which is used to set the fault threshold using the Laida criterion according to the first historical data and the historical fault temperature value.

[0129] The present invention further provides a wind turbine generator set, including the wind turbine component temperature fault detection system described in the above solutions, the system comprising:

[0130] A first predicted temperature acquisition module 201 is configured to obtain a first predicted temperature of the fan component based on data of a first parameter and first historical data within a target period; the first parameter being a parameter in a heat dissipation path of the fan component that is nonlinearly related to temperature; and the first historical data including historical data of the first parameter and historical temperature data of the fan component.

[0131] A second predicted temperature acquisition module 202 is configured to obtain a second predicted temperature of the fan component based on data of a second parameter and second historical data within a target period; the second parameter is a parameter of the fan component that is nonlinearly related to temperature; and the second historical data includes historical data of the second parameter and historical temperature data of the fan component.

[0132] A third predicted temperature acquisition module 203 is configured to obtain a third predicted temperature based on the first predicted temperature and the second predicted temperature;

[0133] A residual obtaining module 204 is configured to obtain a residual by subtracting the third predicted temperature from the actual temperature of the fan component within the target period;

[0134] The fault determination module 205 is configured to determine that a temperature fault exists in the fan component when the residual error is greater than a preset fault threshold.

[0135] Figure 4 An example of a physical structure diagram of an electronic device is shown below. Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other via the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the fan component temperature fault detection method, which includes:

[0136] A first predicted temperature of the fan component is obtained according to the data of the first parameter in the target time period and the first historical data.

[0137] A second predicted temperature of the fan component is obtained according to the data of the second parameter in the target period and the second historical data.

[0138] A third predicted temperature is obtained according to the first predicted temperature and the second predicted temperature.

[0139] The third predicted temperature is subtracted from the actual temperature of the fan component during the target period to obtain a residual.

[0140] When the residual error is greater than a preset fault threshold, it is determined that a temperature fault exists in the fan component.

[0141] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0142] On the other hand, the present invention further provides a computer program product, comprising a computer program, which may be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the fan component temperature fault detection method provided by the above methods, which includes:

[0143] A first predicted temperature of the fan component is obtained according to the data of the first parameter in the target time period and the first historical data.

[0144] A second predicted temperature of the fan component is obtained according to the data of the second parameter in the target period and the second historical data.

[0145] A third predicted temperature is obtained according to the first predicted temperature and the second predicted temperature.

[0146] The third predicted temperature is subtracted from the actual temperature of the fan component during the target period to obtain a residual.

[0147] When the residual error is greater than a preset fault threshold, it is determined that a temperature fault exists in the fan component.

[0148] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for detecting temperature faults of fan components provided by the above methods is implemented. The method includes:

[0149] A first predicted temperature of the fan component is obtained according to the data of the first parameter in the target time period and the first historical data.

[0150] A second predicted temperature of the fan component is obtained according to the data of the second parameter in the target period and the second historical data.

[0151] A third predicted temperature is obtained according to the first predicted temperature and the second predicted temperature.

[0152] The third predicted temperature is subtracted from the actual temperature of the fan component during the target period to obtain a residual.

[0153] When the residual error is greater than a preset fault threshold, it is determined that a temperature fault exists in the fan component.

[0154] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0155] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0156] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for detecting temperature faults of fan components, characterized in that: include: Obtaining a first predicted temperature of the fan component based on data of the first parameter within a target period and first historical data; The first parameter is a parameter in the heat dissipation path of the fan component that is nonlinearly related to temperature; the first historical data includes historical data of the first parameter and historical temperature data of the fan component; Obtaining a second predicted temperature of the fan component based on data of a second parameter within a target period and second historical data; the second parameter is a parameter of the fan component that is nonlinearly related to temperature; and the second historical data includes historical data of the second parameter and historical temperature data of the fan component; Obtaining a third predicted temperature based on the first predicted temperature and the second predicted temperature by using a weighted average probability voting method or an average method of the first predicted temperature and the second predicted temperature; Subtracting the third predicted temperature from the actual temperature of the fan component during the target period to obtain a residual; When the residual is greater than a preset fault threshold, it is determined that a temperature fault exists in the fan component.

2. The method for detecting temperature faults of fan components according to claim 1, characterized in that: Obtaining a first predicted temperature of the fan component according to the first characteristic data and the first historical data during the target period specifically includes: The data of the first parameter within the target time period is input into an equivalent heat circuit model with determined hyperparameters to obtain the first predicted temperature; the equivalent heat circuit model with determined hyperparameters is obtained using the following method: Acquiring the first historical data; According to the heat dissipation path, an equivalent thermal circuit model based on physical mechanism is constructed; A hyperparameter search is performed on the equivalent thermal circuit model according to the first historical data to obtain an equivalent thermal circuit model with determined hyperparameters.

3. The method for detecting temperature faults of fan components according to claim 2, characterized in that: Performing a hyperparameter search on the equivalent thermal circuit model based on the first historical data to obtain an equivalent thermal circuit model with determined hyperparameters specifically includes: According to the first historical data, a particle swarm algorithm is used to perform a hyperparameter search on the equivalent thermal circuit model to obtain an equivalent thermal circuit model with determined hyperparameters.

4. The method for detecting temperature faults of fan components according to claim 2, characterized in that: After acquiring the first historical data, and before performing a hyperparameter search on the equivalent thermal circuit model based on the first historical data, the method for acquiring the equivalent thermal circuit model having determined the hyperparameters further includes: The first historical data is preprocessed; the data preprocessing includes missing value filling and data cleaning.

5. The method for detecting temperature faults of fan components according to claim 1, characterized in that: Obtaining a second predicted temperature of the fan component based on the data of the second parameter within the target period and the second historical data, specifically including: The second characteristic data of the target time period is input into the trained time series neural network model to obtain the second predicted temperature; the trained time series neural network model is obtained using the following method: obtaining the second historical data; The historical data of the second parameter in the second historical data is used as input, and the temperature historical data is used as a label to train the time series neural network model to obtain the trained time series neural network model.

6. The method for detecting temperature faults of fan components according to claim 5, characterized in that: After acquiring the second historical data and before training the time series neural network model, the method for acquiring the trained time series neural network model further includes: Automatically screening the second historical data for features using a chi-square test-based method; Data preprocessing is performed on the second historical data after automatic feature screening; the data preprocessing includes missing value filling and data cleaning.

7. The method for detecting temperature-related faults in fan components according to claim 1, wherein obtaining a third predicted temperature based on the first predicted temperature and the second predicted temperature comprises: Performing a weighted average probability vote on the first predicted temperature and the second predicted temperature to obtain a weighted temperature prediction result; The weights of the weighted average probability voting are set using an adaptive weighting method.

8. The method for detecting temperature faults of fan components according to claim 1, wherein the preset fault threshold is determined by using the Laida criterion based on the first historical data and historical fault temperature values.

9. A fan component temperature fault detection system, characterized in that: include: a first predicted temperature acquisition module, configured to obtain a first predicted temperature of the fan component based on data of the first parameter within a target period and first historical data; The first parameter is a parameter in the heat dissipation path of the fan component that is nonlinearly related to temperature; the first historical data includes historical data of the first parameter and historical temperature data of the fan component; A second predicted temperature acquisition module is configured to obtain a second predicted temperature of the fan component based on data of a second parameter within a target period and second historical data; the second parameter is a parameter of the fan component that is nonlinearly related to temperature; The second historical data includes historical data of the second parameter and historical temperature data of the fan component; a third predicted temperature acquisition module, configured to obtain a third predicted temperature based on the first predicted temperature and the second predicted temperature by adopting a weighted average probability voting method or an average method of the first predicted temperature and the second predicted temperature; a residual obtaining module, configured to obtain a residual by subtracting the third predicted temperature from the actual temperature of the fan component within the target period; The fault determination module is used to determine that a temperature fault exists in the fan component when the residual is greater than a preset fault threshold.

10. A wind turbine generator system, characterized in that: It includes the fan component temperature fault detection system as described in claim 9.

Images