Wind turbine health status detection method, model training method and device

Abstract

The present disclosure provides a method for detecting the health status of a wind turbine, a model training method, and an apparatus, and relates to the field of deep learning, and in particular to the field of wind farm technology. The specific implementation scheme is: obtaining the status data of the wind turbine of the wind turbine; inputting the status data of the wind turbine into the power prediction model to obtain the estimated power; the power prediction model can estimate the normal power corresponding to the status data of the wind turbine; determining the gap information between the estimated power and the actual power corresponding to the status data of the wind turbine; and determining the health value of the wind turbine based on the gap information. In the embodiment of the present disclosure, since the gap between the actual power and the estimated power reflects to a certain extent the degree to which the wind turbine deviates from the normal state, the health value of the wind turbine determined based on the gap information between the estimated power and the actual power can accurately describe the health status of the wind turbine.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to the field of artificial intelligence technology. Background Art

[0002] The global wind power industry is developing rapidly. As the number of wind turbines continues to increase, wind turbines are gradually being deployed in more remote land and ocean areas. In this context, it is necessary to monitor the status of wind turbines in industrial scenarios to facilitate better maintenance. Summary of the Invention

[0003] The present disclosure provides a health status detection method, a model training method, and a device for a wind turbine.

[0004] According to one aspect of the present disclosure, a method for detecting the health status of a wind turbine is provided, comprising:

[0005] Obtain wind turbine status data of wind turbines;

[0006] Input the wind turbine status data into the power prediction model to obtain the estimated power; the power prediction model can estimate the normal power corresponding to the wind turbine status data;

[0007] Determine the difference between the estimated power and the actual power corresponding to the wind turbine status data;

[0008] Based on the gap information, the health value of the wind turbine is determined.

[0009] According to another aspect of the present disclosure, a model training method is provided, comprising:

[0010] Obtain training samples, which include wind turbine status samples, actual power, and category labels;

[0011] Input the wind turbine status samples into the power prediction model to obtain the estimated power of the training samples; the power prediction model can estimate the normal power corresponding to the wind turbine status samples;

[0012] Determine the difference between the estimated power of the training sample and the actual power corresponding to the training sample;

[0013] Input the gap information into the health status estimation model to be trained to obtain the estimated health value;

[0014] Determining estimated categories of training samples based on the estimated health values, where the estimated categories include positive samples and negative samples;

[0015] Determine the loss value based on the estimated category and the category label of the training sample;

[0016] The health status prediction model to be trained is adjusted based on the loss value, and the training is terminated when the training convergence condition is met to obtain the health status prediction model.

[0017] According to another aspect of the present disclosure, a device for detecting the health status of a wind turbine is provided, comprising:

[0018] A first acquisition module is used to acquire wind turbine status data of a wind turbine;

[0019] The first estimation module is used to input the wind turbine status data into the power prediction model to obtain the estimated power; the power prediction model can estimate the normal power corresponding to the wind turbine status data;

[0020] A first gap determination module is used to determine gap information between the estimated power and the actual power corresponding to the wind turbine status data;

[0021] The health status determination module is used to determine the health value of the wind turbine based on the gap information.

[0022] According to another aspect of the present disclosure, there is provided a model training device, comprising:

[0023] The second acquisition module is used to obtain training samples, which include wind turbine status samples, actual power and category labels;

[0024] The second estimation module is used to input the wind turbine status sample into the power prediction model to obtain the estimated power of the training sample; the power prediction model can estimate the normal power corresponding to the wind turbine status sample;

[0025] A second gap determination module is used to determine the gap information between the estimated power of the training sample and the actual power corresponding to the training sample;

[0026] The third estimation module is used to input the gap information into the health status estimation model to be trained to obtain an estimated health value;

[0027] A category determination module, configured to determine an estimated category of a training sample based on the estimated health value, wherein the estimated category includes a positive sample and a negative sample;

[0028] A loss determination module is used to determine the loss value based on the estimated category and the category label of the training sample;

[0029] The training module is used to adjust the health status prediction model to be trained based on the loss value, and end the training when the training convergence condition is met to obtain the health status prediction model.

[0030] According to another aspect of the present disclosure, there is provided an electronic device, comprising:

[0031] at least one processor; and

[0032] a memory communicatively connected to the at least one processor; wherein,

[0033] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to perform the method of any embodiment of the present disclosure.

[0034] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method according to any embodiment of the present disclosure.

[0035] According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program, which implements the method according to any embodiment of the present disclosure when executed by a processor.

[0036] In the embodiment of the present disclosure, since the gap between the actual power and the estimated power reflects to a certain extent the degree to which the wind turbine deviates from the normal state, in the embodiment of the present disclosure, the wind turbine health value determined based on the gap information between the estimated power and the actual power can accurately describe the health status of the wind turbine.

[0037] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0039] Figure 1 is a flow chart of a method for detecting the health status of a wind turbine according to an embodiment of the present disclosure;

[0040] Figure 2 is a schematic diagram of a scenario for detecting the health status of a wind turbine according to another embodiment of the present disclosure;

[0041] Figure 3 is a flowchart of a model training method according to an embodiment of the present disclosure;

[0042] Figure 4 is a structural diagram of a wind turbine health status detection device according to an embodiment of the present disclosure;

[0043] Figure 5 is a structural diagram of a model training device according to an embodiment of the present disclosure;

[0044] Figure 64 is a block diagram of an electronic device used to implement the health status detection method / model training method of a wind turbine according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0045] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be appreciated by those skilled in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0046] For wind turbines (hereinafter referred to as wind turbines), the traditional maintenance method is periodic maintenance. This method not only increases maintenance costs and maintenance time, but also may cause economic losses. Therefore, in order to achieve real-time monitoring of the operating status of wind turbines and reasonable maintenance of wind turbines, a method for detecting the health status of wind turbines is proposed. The process can be implemented as follows: Figure 1 As shown:

[0047] S101, obtaining status data of a wind turbine of a wind turbine.

[0048] Most wind turbines are equipped with a Supervisory Control and Data Acquisition (SCADA) system. The present disclosure can collect wind turbine status data based on SCADA, and of course, the specific collection method is not specifically limited.

[0049] The wind turbine status data may include at least one of the following: wind speed, wind direction, temperature, rotation speed, voltage and current, output power, yaw angle, pitch angle and other information.

[0050] In some embodiments, the collected data related to the wind turbine is used to complete temperature prediction, wind turbine performance analysis and wind turbine reliability analysis, thereby obtaining temperature prediction results, wind turbine performance analysis results and wind turbine reliability analysis results. Among them, the temperature prediction results are used to predict temperature trends at future times. The wind turbine performance analysis results can mainly analyze the wind turbine's power generation, utilization time, equipment availability, loss circuit, energy utilization rate, etc., so as to locate the cause of the wind farm's power generation loss and discover problems with equipment performance. The wind turbine reliability analysis results can be understood as the ability of the wind turbine to complete the specified power generation within its service life under specified environmental and working conditions (wind zone, region, temperature, humidity, etc.).

[0051] In the embodiment of the present disclosure, the wind turbine status data may further include at least one of a temperature prediction result, a wind turbine performance analysis result, and a wind turbine reliability analysis result.

[0052] In the embodiment of the present disclosure, one type of information in the above-mentioned wind turbine status data may be used for subsequent processing, or multiple types of information may be used in combination for subsequent processing, which is not limited in the embodiment of the present disclosure.

[0053] S102 , inputting the wind turbine status data into a power prediction model to obtain estimated power; the power prediction model can estimate the normal power corresponding to the wind turbine status data.

[0054] It can be understood that the embodiment of the present disclosure uses data sampled under normal conditions of the wind turbine to fit a power prediction model. The power prediction model can then provide the normal power corresponding to normal conditions based on the input wind turbine status data.

[0055] S103: Determine the difference between the estimated power and the actual power corresponding to the wind turbine status data.

[0056] S104: Determine the health value of the wind turbine based on the gap information.

[0057] In the disclosed embodiments, the power prediction model can be used to estimate the normal power corresponding to the wind turbine status data, i.e., the estimated power. The difference between the actual power and the estimated power, to a certain extent, reflects the degree to which the wind turbine has deviated from its normal state. Therefore, in the disclosed embodiments, the wind turbine health value determined based on this difference between the estimated and actual power can accurately describe the health status of the wind turbine.

[0058] In some embodiments, wind speed information may be more representative than other wind turbine status information, and the impact of other information on power can be attributed to the impact of wind speed. Therefore, in the embodiments of the present disclosure, wind turbine status data includes wind speed information.

[0059] Since the amount of wind speed information data is relatively large, and considering that the difference between wind speed and output power in a short period of time is not large, in order to process the wind speed information more quickly, the wind speed information can be preprocessed, wherein the preprocessing can include time series processing, which can be implemented as follows: based on the time series, multiple time windows are obtained, and the corresponding wind speed information in each time window is obtained. Specifically, in a time window with a width of t, t consecutive wind speed information sampling data v1, v2, v3, ..., v t , the output powers corresponding to t sampling data are p1, p2, p3, ..., p t , the output power in the time window is averaged to obtain the power mean value, as shown in expression (1):

[0060]

[0061] in, represents the mean power, t is the total output power of t, p i Represents the i-th output power, where i is a positive integer.

[0062] Because initial wind turbine status data may contain missing data, and considering the instantaneous variability of wind speed and output power, wind turbine status data within the same time window can be smoothed. During implementation, a smoothing function can be called to perform smoothing. This disclosure does not limit the method for smoothing wind turbine status data; any method that can achieve data smoothing can be used in the embodiments of this disclosure.

[0063] In summary, in the embodiments of the present disclosure, the wind turbine status data is divided in a time window manner, which can effectively save computing resources while obtaining the health value of the wind turbine.

[0064] In some embodiments, when the wind speed at each sampling point is selected as the wind turbine status data, the actual power corresponding to the wind speed is the output power corresponding to the wind speed at each sampling point.

[0065] In some embodiments, the statistical errors between the estimated power and the actual power can be determined based on multiple data analysis methods to obtain multiple statistical errors; multiple sub-parameters can be determined based on the multiple statistical errors to obtain gap information including multiple sub-parameters.

[0066] In the embodiment of the present disclosure, by processing data using a variety of data analysis methods, the gap between actual power and estimated power can be described from multiple perspectives. By combining the results obtained by a variety of data analysis methods, reasonable and effective gap information can be obtained, thereby improving the accuracy of the wind turbine health value.

[0067] The data analysis method for determining the statistical error between the estimated power and the actual power may be the absolute error between the estimated power and the actual power, or the mean square error between the estimated power and the actual power.

[0068] The absolute error expression is shown in formula (2):

[0069]

[0070] Where ε1 represents the absolute error, Indicates the estimated power, Indicates the mean power.

[0071] The expression of mean square error is shown in formula (3):

[0072]

[0073] Where ε2 represents the mean square error, The meanings are the same as above and will not be repeated here.

[0074] Any method that can determine the difference between the estimated power and the actual power corresponding to the wind turbine status data is applicable to the embodiments of the present disclosure, and the present disclosure is not limited thereto. For example, the difference information can also be determined based on the ratio error of the estimated power and the actual power, and its expression is shown in formula (4):

[0075]

[0076] Where ε3 represents the ratio error, The meanings are the same as above and will not be repeated here.

[0077] In the embodiment of the present disclosure, the statistical error between the estimated power and the actual power is determined using the absolute error and the mean square error. The implementation of the two methods can save computing resources and can better describe the gap between the actual power and the estimated power.

[0078] In some embodiments, weights of sub-parameters included in the gap information may be obtained; based on the obtained weights, a weighted sum is performed on each sub-parameter in the gap information to obtain a wind turbine health value.

[0079] Taking the absolute error between the estimated power and the actual power and the mean square error between the estimated power and the actual power as an example, the health value of the wind turbine is shown in expression (5):

[0080] HI=α1ε1+α2ε2 (5)

[0081] Where HI represents the health value of the wind turbine, and α1 and α2 represent weights.

[0082] When the actual power in the same time window adopts the power average of the output powers of multiple wind speed information, ε1 represents the absolute error between the estimated power in the same time window and the power average, and ε2 represents the mean square error between the estimated power and the power average.

[0083] In some embodiments, in addition to using the power mean corresponding to the time window, when the wind speed information at each sampling point is selected as the wind turbine status data, the output power corresponding to each wind speed information can also be used as the actual power. In this case, each wind speed information within the same time window has its own absolute error and mean square error. In formula (5), ε1 is the vector composed of the absolute errors of multiple wind speed information within the same time window, and similarly, ε2 is the vector composed of the mean square errors of multiple wind speed information within the same time window.

[0084] The weight may be allocated based on experimental data, or may be set to 0.5, which is not limited in this disclosure.

[0085] From expression (5), we can see that the relationship between the health status value and the absolute error and the mean square error is linearly correlated. When the gap between the actual power and the estimated power is larger, the absolute error and the mean square error are larger, which proves that the health status value is larger. When the health status value reaches a certain threshold, it can be seen that the wind turbine is in an abnormal state.

[0086] In the embodiment of the present disclosure, the wind turbine health value is determined based on multiple sub-parameters using a weighted summation method. The weights of the sub-parameters can be adjusted based on the proportions of the sub-parameters, and a more accurate wind turbine health value can be obtained using fewer computing resources.

[0087] In other embodiments, the health status of the wind turbine can also be evaluated based on a health status prediction model constructed based on a neural network. This method can be implemented as follows: the sub-parameters contained in the gap information are input into the health status prediction model constructed based on the neural network to obtain the health value of the wind turbine output by the health status prediction model. As shown in Figure (2), when the estimated power is obtained based on the above method, taking the sub-parameters as absolute error and mean square error as an example, the absolute error and mean square error are obtained based on the wind turbine status data at this moment and the estimated power, and the absolute error and mean square error are input into the health status prediction model to obtain the health value of the wind turbine.

[0088] In some embodiments, a threshold may be set. When the health value of the wind turbine is greater than the threshold, the health state of the wind turbine is determined to be abnormal. When the health value of the wind turbine is lower than the threshold, the health state of the wind turbine is determined to be normal.

[0089] In the embodiment of the present disclosure, the health status prediction model constructed based on multiple sub-parameters has strong robustness and fault tolerance, and has strong information integration capabilities. It can learn relatively accurate wind turbine health values and thus realize the detection of wind turbine models.

[0090] As described above, in the embodiments of the present disclosure, it is necessary to use a power prediction model to obtain the estimated power. In one possible implementation, a fitting function can be established based on the relationship between the wind turbine status sample and the actual power, and then the power prediction model can be fitted. The fitting function of the power prediction model can be in the form of a polynomial, which can be shown as expression (6):

[0091] P=a0*v 0 +a1*v 1 +a2*v 2 +a3*v 3 +…+a n *v n (6)

[0092] Where, when the input is the corresponding wind speed mean in the time window, P is the power mean corresponding to the time window, v is the average wind speed corresponding to the time window, a0, a1, a2, ..., a n The coefficients of the fitting function can be adjusted based on the mean power and the average wind speed. For ease of calculation, only the first four terms of the fitting function can be taken.

[0093] Where, when the input is the wind speed information corresponding to each sampling point, P is the output power corresponding to the wind speed information, v is the wind speed information corresponding to the sampling point, a0, a1, a2, ..., a n The meaning is the same as above, that is, the fitting function coefficient, which will not be described here one by one.

[0094] When constructing a fitting function based on collected data, the collected data can be the average data of a sliding window, or the data collected for each sampling point can be used to obtain the fitting function coefficients, and the relationship between the two can be fitted using a polynomial function. This method is simple and easy to operate.

[0095] In another embodiment, the estimated power can be calculated based on a power prediction model constructed using a neural network. The training process for the power prediction model can be implemented as follows: obtaining fan status samples and the actual power corresponding to the fan status samples; establishing an initial power prediction model based on the fan status samples and the actual power corresponding to the fan status samples to obtain the estimated power; determining a loss value based on the actual power and the estimated power; and adjusting the model parameters of the initial power prediction model based on the loss value. When the power prediction model meets the convergence conditions, the training is terminated to obtain the power prediction model.

[0096] In some embodiments, the loss value between the actual power and the estimated power can be determined based on the mean square error, which is expressed as shown in formula (7):

[0097]

[0098] Among them, loss represents the loss value between actual power and estimated power, N represents N time windows, represents the estimated power of the jth time window, Represents the actual power of the j-th time window, where j is a positive integer.

[0099] The convergence condition may be that the loss value between the actual power and the estimated power approaches stability, or the number of iterations reaches a preset number.

[0100] Based on the same technical concept, the present disclosure also provides a method for training a health status prediction model, which can be implemented as follows: Figure 3 As shown:

[0101] S301: Obtain training samples, where the training samples include wind turbine status samples, actual power, and category labels.

[0102] S302 , inputting the wind turbine status sample into the power prediction model to obtain the estimated power of the training sample; the power prediction model can estimate the normal power corresponding to the wind turbine status sample.

[0103] S303: Determine the difference between the estimated power of the training sample and the actual power corresponding to the training sample.

[0104] The difference information here is similar to the determination method of the aforementioned difference information, and will not be described in detail here.

[0105] S304: Input the gap information into the health status estimation model to be trained to obtain an estimated health value.

[0106] S305 : Determine an estimated category of the training sample based on the estimated health value, where the estimated category includes positive samples and negative samples.

[0107] S306 : Determine a loss value based on the estimated category and the category label of the training sample.

[0108] S307 , adjusting the health status prediction model to be trained based on the loss value, and ending the training when the training convergence condition is met to obtain the health status prediction model.

[0109] In this disclosed embodiment, the difference between actual and estimated power reflects, to a certain extent, the degree to which the wind turbine deviates from normal conditions. Therefore, in this disclosed embodiment, an estimated health value is derived based on this difference between estimated and actual power, accurately describing the health status of the wind turbine. Classification labels are introduced based on the estimated health value, thereby facilitating the training of a health status prediction model.

[0110] The training samples may include normal fan data obtained under normal fan conditions and abnormal fan data obtained under abnormal fan conditions. Since the amount of abnormal fan data may be small, a simulated fan model can be used to simulate abnormal fan conditions and obtain the abnormal data as the abnormal fan data.

[0111] In some embodiments, when the training samples are sample data obtained by sampling when the fan is in normal state, the category label corresponding to the training samples is a positive sample; when the training samples are sample data obtained by sampling when the fan is in abnormal state, the category label corresponding to the training samples is a negative sample.

[0112] The method for determining category labels in the disclosed embodiment does not require manual labeling and can accurately and automatically determine category labels, thereby accelerating the training speed of the model and improving the training efficiency of the model.

[0113] When the health value of the fan is greater than a certain threshold, the health value of the fan is confirmed to be a negative sample, confirming that the fan is in an abnormal state at the moment. When the health value of the fan is not greater than a certain threshold, the health value of the fan is confirmed to be a positive sample, confirming that the fan is in a normal state at the moment.

[0114] Based on the same technical concept, the present disclosure also proposes a device for detecting the health status of a fan, such as Figure 4 The device shown comprises:

[0115] The first acquisition module 401 is used to acquire the status data of the wind turbine of the wind turbine;

[0116] The first estimation module 402 is used to input the wind turbine status data into the power prediction model to obtain the estimated power; the power prediction model can estimate the normal power corresponding to the wind turbine status data;

[0117] A first gap determination module 403 is used to determine gap information between the estimated power and the actual power corresponding to the wind turbine status data;

[0118] The health status determination module 404 is configured to determine a health value of the wind turbine based on the gap information.

[0119] In some embodiments, the first gap determination module is configured to:

[0120] Based on multiple data analysis methods, the statistical errors between the estimated power and the actual power are determined respectively to obtain multiple statistical errors;

[0121] A plurality of sub-parameters are determined based on a plurality of statistical errors to obtain gap information including the plurality of sub-parameters.

[0122] In some embodiments, the determining module is configured to:

[0123] Get the weights of the sub-parameters included in the gap information;

[0124] Based on the obtained weights, each sub-parameter in the gap information is weighted and summed to obtain the wind turbine health value.

[0125] In some embodiments, the first gap module is configured to:

[0126] The sub-parameters contained in the gap information are input into a health status prediction model built based on a neural network to obtain the health value of the wind turbine output by the health status prediction model.

[0127] In some embodiments, the statistical error between the estimated power and the actual power in the first gap module includes at least one of the following:

[0128] The absolute error between the estimated power and the actual power, and the mean square error between the estimated power and the actual power.

[0129] In some embodiments, the wind turbine status data includes wind speed information within a plurality of time windows;

[0130] The actual power is the average power within the time window.

[0131] Based on the same technical concept, the present disclosure also proposes a model training device, such as Figure 5 The device shown comprises:

[0132] The second acquisition module 501 is used to acquire training samples, where the training samples include wind turbine status samples, actual power and category labels;

[0133] The second estimation module 502 is used to input the wind turbine status sample into the power prediction model to obtain the estimated power of the training sample; the power prediction model can estimate the normal power corresponding to the wind turbine status sample;

[0134] A second gap determination module 503 is used to determine the gap information between the estimated power of the training sample and the actual power corresponding to the training sample;

[0135] The third estimation module 504 is used to input the gap information into the health status estimation model to be trained to obtain an estimated health value;

[0136] A category determination module 505 is configured to determine an estimated category of a training sample based on the estimated health value, wherein the estimated category includes positive samples and negative samples;

[0137] A loss determination module 506 is configured to determine a loss value based on the estimated category and the category label of the training sample;

[0138] The training module 507 is used to adjust the health status prediction model to be trained based on the loss value, and terminate the training when the training convergence condition is met to obtain the health status prediction model.

[0139] In some embodiments, when the training samples are sample data obtained by sampling when the fan is in normal state, the category label corresponding to the training samples is a positive sample; when the training samples are sample data obtained by sampling when the fan is in abnormal state, the category label corresponding to the training samples is a negative sample.

[0140] For the description of specific functions and examples of each module and submodule of the device in the embodiment of the present disclosure, please refer to the relevant description of the corresponding steps in the above method embodiment, which will not be repeated here.

[0141] In the technical solutions disclosed herein, the acquisition, storage, and application of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0142] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0143] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0144] like Figure 6 As shown, the device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. Various programs and data required for the operation of the device 600 can also be stored in the RAM 603. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0145] Various components in device 600 are connected to I / O interface 605, including an input unit 606, such as a keyboard, mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a magnetic disk, optical disk, etc.; and a communication unit 609, such as a network card, modem, wireless communication transceiver, etc. The communication unit 609 allows device 600 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0146] The computing unit 601 can be various general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as the health status detection method / model training method for a wind turbine. For example, in some embodiments, the health status detection method / model training method for a wind turbine can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as the storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the health status detection method / model training method for a wind turbine described above can be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to execute the health status detection method / model training method of the wind turbine in any other appropriate manner (for example, by means of firmware).

[0147] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0148] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0149] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0150] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0151] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0152] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0153] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not limited herein.

[0154] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for detecting the health status of a wind turbine, comprising: Obtain wind turbine status data of wind turbines; Inputting the wind turbine status data into a power prediction model to obtain an estimated power; the power prediction model can estimate the normal power corresponding to the wind turbine status data; Determining a difference between the estimated power and the actual power corresponding to the wind turbine status data; The gap information includes an absolute error and a mean square error between the estimated power and the actual power; Determining a health value of the wind turbine based on the gap information includes: performing weighted summation of the absolute error and the mean square error based on respective weights of the absolute error and the mean square error to obtain the health value of the wind turbine.

2. The method according to claim 1, wherein Determining the difference between the estimated power and the actual power corresponding to the wind turbine status data includes: determining statistical errors between the estimated power and the actual power based on multiple data analysis methods to obtain multiple statistical errors; A plurality of sub-parameters are determined based on the plurality of statistical errors to obtain gap information including the plurality of sub-parameters.

3. The method according to claim 2, wherein: The determining of the wind turbine health value based on the gap information further includes: The sub-parameters included in the gap information are input into a health status prediction model constructed based on a neural network to obtain the wind turbine health value output by the health status prediction model.

4. The method according to any one of claims 1 to 3, wherein The wind turbine status data includes multiple wind speed information within the same time window; The actual power is the power average of the output powers corresponding to each wind speed information within the time window.

5. A model training method for training the health status prediction model according to claim 3, comprising: Acquire training samples, where the training samples include wind turbine status samples, actual power, and category labels; Inputting the wind turbine status sample into a power prediction model to obtain the estimated power of the training sample; the power prediction model can estimate the normal power corresponding to the wind turbine status sample; Determining the difference between the estimated power of the training sample and the actual power corresponding to the training sample; Inputting the gap information into a health status prediction model to be trained to obtain an estimated health value; Determining an estimated category of the training sample based on the estimated health value, wherein the estimated category includes a positive sample and a negative sample; Determining a loss value based on the estimated category and the category label of the training sample; The health status prediction model to be trained is adjusted based on the loss value, and the training is terminated when a training convergence condition is met to obtain a health status prediction model.

6. The method according to claim 5, wherein when the training sample is sample data obtained by sampling under normal conditions of the wind turbine, the category label corresponding to the training sample is a positive sample; In the case that the training samples are sample data obtained by sampling when the wind turbine is in an abnormal state, the category label corresponding to the training samples is a negative sample.

7. A device for detecting the health status of a fan, comprising: A first acquisition module is used to acquire wind turbine status data of a wind turbine; A first estimation module is configured to input the wind turbine status data into a power prediction model to obtain an estimated power; the power prediction model is capable of estimating a normal power corresponding to the wind turbine status data; A first gap determination module is configured to determine gap information between the estimated power and the actual power corresponding to the wind turbine status data; the gap information includes an absolute error and a mean square error between the estimated power and the actual power; The health status determination module is used to determine the health value of the wind turbine based on the gap information, including: performing weighted summation of the absolute error and the mean square error based on their respective weights to obtain the health value of the wind turbine.

8. The device according to claim 7, wherein The first gap determination module is configured to: determining statistical errors between the estimated power and the actual power based on multiple data analysis methods to obtain multiple statistical errors; A plurality of sub-parameters are determined based on the plurality of statistical errors to obtain gap information including the plurality of sub-parameters.

9. The device according to claim 8, wherein The health status determination module is further configured to: The sub-parameters included in the gap information are input into a health status prediction model constructed based on a neural network to obtain the wind turbine health value output by the health status prediction model.

10. The device according to any one of claims 7 to 9, wherein: The wind turbine status data includes multiple wind speed information within the same time window; The actual power is the power average of the output powers corresponding to each wind speed information within the time window.

11. A model training device for training the health status prediction model according to claim 9, comprising: A second acquisition module is used to acquire training samples, where the training samples include wind turbine status samples, actual power and category labels; A second estimation module is configured to input the wind turbine status sample into a power prediction model to obtain an estimated power of the training sample; the power prediction model is capable of estimating the normal power corresponding to the wind turbine status sample; A second gap determination module is used to determine the gap information between the estimated power of the training sample and the actual power corresponding to the training sample; A third estimation module is used to input the gap information into the health status estimation model to be trained to obtain an estimated health value; a category determination module, configured to determine an estimated category of the training sample based on the estimated health value, wherein the estimated category includes a positive sample and a negative sample; a loss determination module, configured to determine a loss value based on the estimated category and the category label of the training sample; A training module is used to adjust the health status prediction model to be trained based on the loss value, and to end the training when the training convergence condition is met to obtain the health status prediction model.

12. The device according to claim 11, wherein when the training sample is sample data obtained by sampling when the wind turbine is in a normal state, the category label corresponding to the training sample is a positive sample; In the case that the training samples are sample data obtained by sampling when the wind turbine is in an abnormal state, the category label corresponding to the training samples is a negative sample.

13. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.

14. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 6.

15. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 6.

Images