Wind turbine generator mechanical part situation awareness method and device based on multi-source data

Abstract

The invention provides a wind turbine generator mechanical part situation awareness method and device based on multi-source data. The method comprises the steps that situation awareness environment parameters are determined and acquired based on the damage condition and influence factors of a wind turbine generator mechanical part; the situation awareness environment parameters are preprocessed; performing feature extraction and fusion on the preprocessed situation awareness environment parameters to obtain a multi-source information fusion matrix; and taking the multi-source information fusion matrix as the input of a situation awareness model, and utilizing the output of the situation awareness model to identify the situation of the mechanical part of the wind turbine generator. According to the method, the vibration signals, the output power, the temperature and the humidity of the wind turbine generator are collected, feature extraction and fusion are carried out on the collected signals, and the reliability of situation awareness of mechanical parts of the wind turbine generator is improved. In addition, linear function mapping is carried out on the input data through a mapper, so that the input data are located in the same shared feature space, multi-sensor information multi-mode fusion is achieved, and the sensing accuracy is further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of situation awareness of wind turbine components, and in particular to a method and device for situation awareness of mechanical components of a wind turbine based on multi-source data. Background Art

[0002] With the growth of energy demand and the improvement of environmental protection awareness, wind power generation as a clean and renewable energy has gradually become one of the main power generation methods in the world. However, wind farms face many challenges, such as abnormal operation of wind turbines, fault detection and location, etc. In order to improve the performance and reliability of wind turbines, situational awareness and prediction of wind turbines have become the focus of current research.

[0003] Traditionally, the condition monitoring of wind turbines mainly relies on limited sensor data, such as vibration and temperature. However, these single-source data often cannot fully and accurately reflect the true condition of wind turbines. At the same time, due to the wide distribution of wind farms and the large number of turbines, the installation and maintenance costs of sensors are relatively high. Therefore, how to use multi-source data to perform situational awareness of mechanical components of wind turbines has become an urgent problem that needs to be solved. Multi-source data includes but is not limited to information such as wind speed, wind direction, temperature, humidity, power output, etc., which can be obtained through various channels such as meteorological stations and sensor networks. Using these data, more comprehensive and accurate wind turbine status information can be obtained, thereby achieving accurate monitoring and fault prediction of the unit.

[0004] In current research, although some grid operation status assessment methods based on multi-source data have been proposed, there are still few methods for the mechanical components of wind turbines. Summary of the invention

[0005] In view of this, the present invention provides a method and device for situation awareness of mechanical components of a wind turbine generator set based on multi-source data to solve at least one of the problems mentioned above.

[0006] In order to achieve the above object, the present invention adopts the following scheme:

[0007] According to a first aspect of the present invention, there is provided a method for situation awareness of mechanical components of a wind turbine generator set based on multi-source data, the method comprising: determining and acquiring situation awareness environmental parameters based on damage conditions and influencing factors of the mechanical components of the wind turbine generator set; preprocessing the situation awareness environmental parameters; performing feature extraction and fusion on the preprocessed situation awareness environmental parameters to obtain a multi-source information fusion matrix; using the multi-source information fusion matrix as an input of a situation awareness model, and using the output of the situation awareness model to perform situation identification of mechanical components of the wind turbine generator set.

[0008] As an embodiment of the present invention, the situation awareness environmental parameters in the above method include a vibration signal, a power output signal, a temperature signal and a humidity signal of the wind turbine.

[0009] As an embodiment of the present invention, the preprocessing of the situation awareness environment parameters in the above method includes: converting the vibration signal using short-time Fourier transform; normalizing the converted vibration signal, the power output signal, the temperature signal and the humidity signal; forming a first information matrix with the normalized vibration signal and power output signal; and forming a second information matrix with the normalized temperature signal and humidity signal.

[0010] As an embodiment of the present invention, the above method extracts and fuses the preprocessed situation awareness environment parameters to obtain a multi-source information fusion matrix, including: extracting local features of the first information matrix; introducing a global information fusion strategy based on the local features to obtain a first feature matrix to be fused; extracting feature information of the second information matrix based on the GRU network to obtain a second feature matrix; and fusing the first feature matrix and the second feature matrix through a fully connected layer to obtain a multi-source information fusion matrix.

[0011] As an embodiment of the present invention, in the above method, the situation awareness model maps data from different acquisition sources to a shared feature space by setting a mapping function during training.

[0012] As an embodiment of the present invention, the loss function selected by the situation awareness model in the above method during the training process is as follows:

[0013]

[0014] In the above formula, z i (y) is the advantage score of the wind turbine mechanical component state y obtained by the mapper, is the vector u in the multi-source data fusion matrix i Predicted probability of belonging to state y of a mechanical component of the wind turbine.

[0015] According to a second aspect of the present invention, there is provided a situation awareness device for mechanical components of a wind turbine generator set based on multi-source data, the device comprising: an environmental parameter acquisition unit, used to determine and acquire situation awareness environmental parameters based on damage conditions and influencing factors of mechanical components of the wind turbine generator set; a preprocessing unit, used to preprocess the situation awareness environmental parameters; a feature fusion unit, used to extract features and fuse the preprocessed situation awareness environmental parameters to obtain a multi-source information fusion matrix; a situation awareness unit, used to use the multi-source information fusion matrix as an input of a situation awareness model, and use the output of the situation awareness model to perform situation identification of mechanical components of the wind turbine generator set.

[0016] As an embodiment of the present invention, the situation awareness environmental parameters in the above-mentioned device include a vibration signal, a power output signal, a temperature signal and a humidity signal of the wind turbine.

[0017] As an embodiment of the present invention, the preprocessing unit in the above-mentioned device includes: a conversion module, which is used to convert the vibration signal using short-time Fourier transform; a normalization module, which is used to normalize the converted vibration signal, the power output signal, the temperature signal and the humidity signal; a first matrix composition module, which is used to form a first information matrix with the normalized vibration signal and the power output signal; and a second matrix composition module, which is used to form a second information matrix with the normalized temperature signal and the humidity signal.

[0018] As an embodiment of the present invention, the feature fusion unit in the above-mentioned device includes: a local feature extraction module, used to extract local features of the first information matrix; a first feature matrix acquisition module, used to introduce a global information fusion strategy based on the local features to obtain a first feature matrix to be fused; a second feature matrix acquisition module, used to extract feature information of the second information matrix based on a GRU network to obtain a second feature matrix; a feature fusion module, used to fuse the first feature matrix and the second feature matrix through a fully connected layer to obtain a multi-source information fusion matrix.

[0019] As an embodiment of the present invention, the situation awareness model in the above-mentioned device maps data from different acquisition sources to a shared feature space by setting a mapping function during training.

[0020] As an embodiment of the present invention, the loss function selected by the situation awareness model in the above device during the training process is as follows:

[0021]

[0022] In the above formula, z i (y) is the advantage score of the wind turbine mechanical component state y obtained by the mapper, is the vector u in the multi-source data fusion matrix i Predicted probability of belonging to state y of a mechanical component of the wind turbine.

[0023] According to a third aspect of the present invention, there is provided an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0024] According to a fourth aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0025] The method and device for situation awareness of mechanical components of wind turbines based on multi-source data proposed in the present invention collects vibration signals, output power, temperature, and humidity of the main shaft system by arranging multiple sensors, extracts and fuses the collected signals, and improves the reliability of situation awareness of mechanical components of wind turbines. In addition, the present invention combines situation awareness theory to process the multi-source information matrix, first performs a function mapping on the input data through a mapper, so that the input data is in the same shared feature space, realizes multi-mode fusion of multi-sensor information, and further improves the accuracy of perception. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0027] Figure 1 It is a flow chart of a method for situation awareness of mechanical components of a wind turbine generator system based on multi-source data provided in an embodiment of the present application;

[0028] Figure 2 It is a schematic diagram of a flow chart of preprocessing situation awareness environment parameters provided by an embodiment of the present invention;

[0029] Figure 3 It is a schematic diagram of a process for extracting and fusing features of situation awareness environment parameters provided by an embodiment of the present invention;

[0030] Figure 4 It is a structural schematic diagram of a situation awareness device for mechanical components of a wind turbine generator set based on multi-source data provided by an embodiment of the present invention;

[0031] Figure 5 is a schematic diagram of the structure of a preprocessing unit provided in an embodiment of the present invention;

[0032] Figure 6 is a structural schematic diagram of a feature fusion unit provided in an embodiment of the present invention;

[0033] Figure 7 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0034] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0035] Since there are still few methods for situational awareness of mechanical components of wind turbines, this patent aims to provide a method and device for situational awareness of mechanical components of wind turbines based on multi-source data. By effectively utilizing multi-source data, the accuracy and reliability of state monitoring and fault diagnosis of wind turbines can be improved, providing more stable and efficient operation management for the wind power industry. In addition, this method also has the potential for promotion and application, and has certain reference value in state monitoring and fault diagnosis in other industrial fields.

[0036] like Figure 1 The figure is a flow chart of a method for situation awareness of mechanical components of a wind turbine generator system based on multi-source data provided by an embodiment of the present application. The method comprises the following steps:

[0037] Step S101: Determine and obtain situation awareness environment parameters based on the damage conditions and influencing factors of mechanical components of the wind turbine generator system.

[0038] The common damage conditions during the operation of wind turbines are as follows:

[0039] 1. Blade damage: Affected by severe weather conditions, excessive rotation speed, material fatigue and other factors, the blades of wind turbines may be damaged by cracks, deformation or falling off, resulting in a decrease in wind energy conversion efficiency.

[0040] 2. Spindle and bearing failure: Under different working conditions, each set of vibration, temperature, sound, and current signals generated by the spindle system are inconsistent; the bearing may be damaged, the spindle may be bent or broken, etc. These situations can be judged by collecting the radial and axial vibration signals of the bearing by the vibration sensor.

[0041] 3. Gearbox failure: The gearbox is an important component of a wind turbine. Due to long-term high-speed operation and load changes, the gearbox may have problems such as gear damage, bearing failure, and lubricating oil leakage, affecting transmission efficiency and reliability.

[0042] 4. Generator failure: Failure may cause unstable or interrupted power output. Common failures include winding short circuit, insulation aging, bearing damage, etc., which can be judged by current or output power.

[0043] Through the analysis of the above-mentioned damage conditions and influencing factors, the present application can select the vibration signal, power output signal, temperature signal and humidity signal of the wind turbine as situation awareness environmental parameters. Accordingly, in order to obtain these situation awareness environmental parameters, the present application needs to arrange corresponding vibration sensors, output power sensors, temperature sensors and humidity sensors at key locations.

[0044] Step S102: pre-processing the situation awareness environment parameters.

[0045] In this step, the collected original environmental parameters are processed in order to prepare for subsequent feature extraction and fusion.

[0046] Preferably, Figure 2 As shown, this step may further include the following sub-steps:

[0047] Step S1021: transform the vibration signal using short-time Fourier transform.

[0048] The vibration signal in the wind turbine is an unstable signal relative to the power output and temperature and humidity signals. Therefore, this application converts the vibration signal based on short-time Fourier transform. The definition of the vibration signal time-frequency diagram process obtained by this method is as follows:

[0049]

[0050] Among them: f(t) is the time domain signal, g(t-τ) is the window function in Fourier transform, τ is the center of the window function, STFT(t,ω) is the spectrum of the signal f(t) at time t, and ω is the angular frequency of the vibration signal f(t). The transformed signal contains time domain features.

[0051] Step S1022: normalizing the converted vibration signal, the power output signal, the temperature signal and the humidity signal.

[0052] Here, different types of data are normalized to eliminate dimensional differences and ensure that they are compared and calculated at the same scale. For example, Min-Max normalization or Z-score normalization can be used for normalization.

[0053] Step S1023: The normalized vibration signal and the power output signal are combined into a first information matrix.

[0054] In terms of feature correlation, vibration signals are usually important information for monitoring the operating status of mechanical components in wind turbines. When mechanical components fail or operate abnormally, the vibration signal and the power output signal are often highly correlated, so combining them together helps to extract more comprehensive and useful features. For example, an increase in the load of a mechanical component may lead to an increase in power output, while the vibration signal will also change, and a specific vibration mode may correspond to a specific power output state, thereby helping to detect abnormal conditions.

[0055] In addition, in terms of signal characteristics, vibration signals and power output signals are more similar in frequency and time domain characteristics, and they can both capture timely changes in the working status or performance of the equipment. Since these signals usually have high time resolution, combining them for analysis can better capture synchronous changes and transient characteristics.

[0056] Therefore, the present application can more conveniently perform feature extraction and data fusion by combining the vibration signal and the power output signal into a first information matrix. For the vibration signal and the power output signal, their joint features can be extracted and the potential relationship between them can be explored, which is helpful for fault diagnosis and anomaly detection.

[0057] Step S1024: The normalized temperature signal and humidity signal are combined into a second information matrix.

[0058] Temperature and humidity signals usually reflect the physical state of the environment, and there is often a certain correlation between the two signals. In particular, in some situations, changes in ambient temperature may affect changes in humidity, and vice versa. In addition, temperature and humidity signals change relatively slowly and usually have lower time resolution and longer correlation periods. Therefore, combining these two signals to form a second information matrix is ​​helpful for analyzing the overall environmental conditions rather than immediate state changes.

[0059] Assuming that a total of n sensor information is selected to construct the multi-source data coupling input of the mechanical components of the wind turbine, the sensor records m information from t1 to t2, and all types of data are normalized, the first information matrix composed of vibration signals and power output signals is expressed as C∈R m×n The second information matrix composed of temperature and humidity signals is expressed as D∈R m×n .

[0060] Step S103: extracting and fusing the preprocessed situation awareness environment parameters to obtain a multi-source information fusion matrix.

[0061] Preferably, Figure 3 As shown, this step may further include the following sub-steps:

[0062] Step S1031: Extract the local features of the first information matrix.

[0063] In this embodiment, the mapping relationship between multi-source data can be learned through the sliding window technique or a convolutional neural network to perform information fusion at the decision level. Taking the convolutional neural network as an example, for the above first information matrix C ∈ R m×n , use a convolutional kernel H with a length of 1 and a width of h to extract the local features of the multi-source information matrix C. This process can be expressed as:

[0064]

[0065] In the above formula, i = 1, 2,..., n, a is the offset, s is the convolution operation step size (s < h), and j represents the number of convolution operations

[0066] After the first information matrix undergoes the convolution operation, its feature matrix Y ∈ R n([(x-d) / s]+1) .

[0067] Step S1032: Introduce a global information fusion strategy based on the local features to obtain the first feature matrix to be fused.

[0068] Simply extracting features through the pooling layer of a convolutional neural network is difficult to effectively obtain complete information from an image. Therefore, this application further introduces a global information fusion strategy (GIF) to increase the global receptive field of the network and obtain more effective features.

[0069] Based on the GIF strategy, two global feature descriptors are provided for the final result, namely the global information descriptor and the global feature descriptor. The global information descriptor is obtained by inputting the global feature descriptor into the max pooling layer. In this strategy, the features are first divided into Ph and Pw blocks horizontally and vertically. The newly generated block elements are composed of a large number of elements from the original positions and a small number of elements from other blocks to synthesize a new complete global element.

[0070] After this processing, the first feature matrix A = [a (1) , a (2) , …, a (n) T .

[0071] Step S1033: Extract the feature information of the second information matrix based on the GRU network to obtain the second feature matrix.

[0072] Use the GRU (Gated Recurrent Unit) network to extract the temporal features from the second information matrix (the combination of temperature and humidity signals) to form the second feature matrix, which is to fully capture the time dependence and sequence pattern of the signals. ​

[0073] GRU is a variant of recurrent neural network (RNN), suitable for processing and extracting features of time series data. It introduces gate mechanisms (update gate and reset gate) to control the flow of information, solving the gradient vanishing and gradient exploding problems in traditional RNN. Using GRU, short-term and long-term dependencies in data can be better learned.

[0074] First, the second information matrix can be prepared to ensure that its shape meets the GRU input requirements. Usually, the input data needs to be a three-dimensional tensor with a shape of (number of samples, time steps, number of features). Then use the deep learning framework to define the GRU layer. Here, you can set the parameters of the GRU layer, such as the number of units and the return sequence option. Finally, the second information matrix is ​​input into the GRU network to extract the sequence features and obtain the second feature matrix.

[0075] Step S1034: fusing the first feature matrix and the second feature matrix through a fully connected layer to obtain a multi-source information fusion matrix.

[0076] This step is to fuse the features from different sources (the first information matrix and the second information matrix) to form a unified multi-source information fusion matrix. This step combines and integrates these features through the fully connected layer, so that the subsequent steps can use more information for analysis and decision-making.

[0077] First, you need to ensure that the first feature matrix and the second feature matrix have been extracted, and the shapes are (number of samples, number of features 1) and (number of samples, number of features 2) respectively. Then use the tf.concat method to concatenate the two feature matrices on the feature dimension (i.e. axis = 1) to form a new concatenated feature matrix. Assume that the concatenated shape is (number of samples, number of features 1 + number of features 2).

[0078] Then define an input layer whose shape is the shape of the concatenated feature matrix (feature number 1 + feature number 2).

[0079] Then add a fully connected layer, which is used to fuse the concatenated features. The number of units in the fully connected layer is 256 (this value can be adjusted according to actual needs), and the activation function uses ReLU to ensure the nonlinear characteristics of the network.

[0080] The final output is the multi-source information fusion matrix, which integrates the features from different information sources. Through these steps, the first feature matrix and the second feature matrix are effectively fused together to form a multi-source information fusion matrix containing more comprehensive information. Such a fusion step can provide richer and more accurate data support for subsequent analysis, prediction and decision-making.

[0081] Step S104: using the multi-source information fusion matrix as an input of a situation awareness model, and using the output of the situation awareness model to perform situation identification of mechanical components of a wind turbine generator set.

[0082] In this embodiment, a long short-term memory network (LSTM) in a deep learning model is used to process a multi-source information fusion matrix to perform situation awareness and fault identification, and the situation awareness model is trained so that it can identify the health status of mechanical components of a wind turbine from an input matrix.

[0083] When training the LSTM situational awareness model, due to different acquisition platforms, the multi-source data fusion matrix may have different distribution conditions. Therefore, preferably, the present application maps all data from different acquisition sources into a shared feature space by setting a mapping function. The mapper consists of four parts: a fully connected layer, a batch normalization layer, a pooling layer, and a classification layer. Given a vector u in the multi-source data fusion matrix i As input, the classifier module outputs a column vector z i , the process can be expressed by the following formula: i =F(u i ), F represents the fully connected layer.

[0084] Through the above steps, all data from different acquisition sources are mapped into a shared feature space and classified by the mapper model to identify the status of the mechanical components of the wind turbine. This process ensures accurate quantitative analysis of the health status of the wind turbine.

[0085] In addition, the traditional LSTM model usually uses common loss functions, which may have various problems, such as: excessive attention to outliers during training, while ignoring the changes in normal data; for unbalanced sequences with large changes, traditional loss functions may not be able to well reflect the learning of the model for different parts, resulting in deviations in the model during training. In response to the above problems, this application uses the following loss function to optimize the network model during the model training process, which is defined as follows:

[0086]

[0087] In the above formula, z i (y) is the advantage score of the wind turbine mechanical component state y obtained by the mapper, is the vector u in the multi-source data fusion matrix i Predicted probability of belonging to state y of a mechanical component of the wind turbine.

[0088] This loss function can better take into account outliers and length differences in data sequences, thereby improving the performance of the model when processing data.

[0089] When training the situation awareness model, the data set is first divided into a training set and a test set. During the training process, the situation awareness model library is established. In the test phase, the situation awareness model can self-learn new states and add the new states to the situation awareness model library, thereby realizing self-learning of situation awareness of wind turbine mechanical components.

[0090] As can be seen from the above, the wind turbine mechanical component situation awareness method based on multi-source data proposed in the present invention collects the vibration signal, output power, temperature, and humidity of the main shaft system by arranging multiple sensors, extracts and fuses the collected signals, and improves the reliability of the situation awareness of the wind turbine mechanical components. In addition, the present invention combines the situation awareness theory to process the multi-source information matrix, first performs a function mapping on the input data through a mapper, so that the input data is in the same shared feature space, realizes the multi-mode fusion of multi-sensor information, and further improves the accuracy of perception.

[0091] like Figure 4 The figure shows a schematic diagram of the structure of a wind turbine mechanical component situation awareness device based on multi-source data provided by an embodiment of the present application. The device includes: an environmental parameter acquisition unit 410, a preprocessing unit 420, a feature fusion unit 430 and a situation awareness unit 440, which are connected in sequence, wherein:

[0092] The environmental parameter acquisition unit 410 is used to determine and acquire situation awareness environmental parameters based on the damage conditions and influencing factors of the mechanical components of the wind turbine generator set.

[0093] The preprocessing unit 420 is used to preprocess the situation awareness environment parameters.

[0094] The feature fusion unit 430 is used to extract and fuse the preprocessed situation awareness environment parameters to obtain a multi-source information fusion matrix.

[0095] The situation awareness unit 440 is used to use the multi-source information fusion matrix as an input of a situation awareness model, and use the output of the situation awareness model to perform situation identification of mechanical components of a wind turbine generator set.

[0096] Preferably, the situation awareness environmental parameters in the above-mentioned device include vibration signals, power output signals, temperature signals and humidity signals of the wind turbine set.

[0097] Preferably, Figure 5 As shown, the pre-processing unit 420 in the above device includes:

[0098] A conversion module 421, used for converting the vibration signal using short-time Fourier transform;

[0099] A normalization module 422, used for normalizing the converted vibration signal, the power output signal, the temperature signal and the humidity signal;

[0100] A first matrix forming module 423, used for forming a first information matrix from the normalized vibration signal and the power output signal;

[0101] The second matrix forming module 424 forms a second information matrix with the normalized temperature signal and humidity signal.

[0102] Preferably, Figure 6 As shown, the feature fusion unit 430 in the above device includes:

[0103] The local feature extraction module 431 is used to extract local features of the first information matrix.

[0104] The first feature matrix acquisition module 432 is used to introduce a global information fusion strategy based on the local features to obtain a first feature matrix to be fused.

[0105] The second feature matrix acquisition module 433 is used to extract feature information of the second information matrix based on the GRU network to obtain a second feature matrix.

[0106] The feature fusion module 434 is used to fuse the first feature matrix and the second feature matrix through a fully connected layer to obtain a multi-source information fusion matrix.

[0107] Preferably, during training, the situation awareness model in the above device maps data from different acquisition sources to a shared feature space by setting a mapping function.

[0108] Preferably, the loss function selected during the training process of the situation awareness model in the above device is as follows:

[0109]

[0110] In the above formula, z i (y) is the advantage score of the wind turbine mechanical component state y obtained by the mapper, is the vector u in the multi-source data fusion matrix i Predicted probability of belonging to state y of a mechanical component of the wind turbine.

[0111] As can be seen from the above, the wind turbine mechanical component situation awareness device based on multi-source data proposed by the present invention collects the vibration signal, output power, temperature, and humidity of the main shaft system by arranging multiple sensors, extracts and fuses the collected signals, and improves the reliability of the situation awareness of the wind turbine mechanical components. In addition, the present invention combines the situation awareness theory to process the multi-source information matrix, first performs a function mapping on the input data through a mapper, so that the input data is in the same shared feature space, realizes the multi-mode fusion of multi-sensor information, and further improves the accuracy of perception.

[0112] Figure 7 is a schematic diagram of an electronic device provided by an embodiment of the present invention. Figure 7 The electronic device shown is a general data processing device, which includes a general computer hardware structure, which at least includes a processor 801 and a memory 802. The processor 801 and the memory 802 are connected via a bus 803. The memory 802 is suitable for storing one or more instructions or programs executable by the processor 801. The one or more instructions or programs are executed by the processor 801 to implement the steps in the above-mentioned method for situation awareness of mechanical components of a wind turbine based on multi-source data.

[0113] The processor 801 may be an independent microprocessor or a collection of one or more microprocessors. Thus, the processor 801 executes the commands stored in the memory 802 to execute the method flow of the embodiment of the present invention as described above to realize the processing of data and the control of other devices. The bus 803 connects the above-mentioned multiple components together, and at the same time connects the above-mentioned components to the display controller 804 and the display device and the input / output (I / O) device 805. The input / output (I / O) device 805 may be a mouse, a keyboard, a modem, a network interface, a touch input device, a somatosensory input device, a printer, and other devices known in the art. Typically, the input / output (I / O) device 805 is connected to the system via an input / output (I / O) controller 806.

[0114] The memory 802 may store software components such as an operating system, a communication module, an interaction module, and an application program. Each of the modules and applications described above corresponds to a set of executable program instructions that implement one or more functions and methods described in the embodiments of the invention.

[0115] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method for situation awareness of mechanical components of a wind turbine generator system based on multi-source data are implemented.

[0116] As can be seen from the above, the method and device for situation awareness of mechanical components of wind turbines based on multi-source data proposed in the present invention collects vibration signals, output power, temperature, and humidity of the main shaft system by arranging multiple sensors, extracts and fuses the collected signals, and improves the reliability of situation awareness of mechanical components of wind turbines. In addition, the present invention combines situation awareness theory to process the multi-source information matrix, first performs a function mapping on the input data through a mapper, so that the input data is in the same shared feature space, realizes multi-sensor information multi-mode fusion, and further improves the accuracy of perception.

[0117] The preferred embodiments of the present invention are described above with reference to the accompanying drawings. Many features and advantages of these embodiments are clear from this detailed description, and therefore the claims are intended to cover all such features and advantages of these embodiments that fall within their true spirit and scope. In addition, since many modifications and changes are easily conceivable to those skilled in the art, it is not intended to limit the embodiments of the present invention to the precise structure and operation illustrated and described, but all suitable modifications and equivalents falling within its scope may be covered.

[0118] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0119] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0120] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0121] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0122] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for situation awareness of mechanical components of a wind turbine generator system based on multi-source data, characterized in that: The method comprises: Determine and obtain situational awareness environmental parameters based on the damage status and influencing factors of wind turbine mechanical components; Preprocessing the situation awareness environment parameters; The pre-processed situation awareness environment parameters are subjected to feature extraction and fusion to obtain a multi-source information fusion matrix; The multi-source information fusion matrix is ​​used as the input of the situation awareness model, and the situation recognition of the mechanical components of the wind turbine generator set is performed using the output of the situation awareness model.

2. The method for wind turbine mechanical component situation awareness based on multi-source data according to claim 1, characterized in that: The situation awareness environmental parameters include vibration signals, power output signals, temperature signals and humidity signals of the wind turbine generator set.

3. The method for wind turbine mechanical component situation awareness based on multi-source data according to claim 2, characterized in that: The preprocessing of the situation awareness environment parameters includes: Converting the vibration signal using short-time Fourier transform; Normalizing the converted vibration signal, the power output signal, the temperature signal and the humidity signal; The normalized vibration signal and the power output signal are combined into a first information matrix; The normalized temperature signal and humidity signal are combined into a second information matrix.

4. The method for wind turbine mechanical component situation awareness based on multi-source data according to claim 3, characterized in that: The method of extracting and fusing the preprocessed situation awareness environment parameters to obtain a multi-source information fusion matrix includes: Extracting local features of the first information matrix; Based on the local features, a global information fusion strategy is introduced to obtain a first feature matrix to be fused; Extracting feature information of the second information matrix based on the GRU network to obtain a second feature matrix; The first feature matrix and the second feature matrix are fused through a fully connected layer to obtain a multi-source information fusion matrix.

5. The method for wind turbine mechanical component situation awareness based on multi-source data according to claim 1, characterized in that: During training, the situation awareness model maps data from different acquisition sources to a shared feature space by setting a mapping function.

6. The method for wind turbine mechanical component situation awareness based on multi-source data according to claim 1, characterized in that: The loss function selected by the situation awareness model during training is as follows: In the above formula, z i (y) is the advantage score of the wind turbine mechanical component state y obtained by the mapper, is the vector u in the multi-source data fusion matrix i Predicted probability of belonging to state y of a mechanical component of the wind turbine.

7. A wind turbine mechanical component situation awareness device based on multi-source data, characterized in that: The device comprises: An environmental parameter acquisition unit, used to determine and acquire situation awareness environmental parameters based on the damage conditions and influencing factors of mechanical components of the wind turbine; A preprocessing unit, used for preprocessing the situation awareness environment parameters; A feature fusion unit is used to extract and fuse the preprocessed situation awareness environment parameters to obtain a multi-source information fusion matrix; The situation awareness unit is used to use the multi-source information fusion matrix as the input of the situation awareness model, and use the output of the situation awareness model to perform situation identification of mechanical components of the wind turbine.

8. The wind turbine mechanical component situation awareness device based on multi-source data according to claim 7, characterized in that: The situation awareness environmental parameters include vibration signals, power output signals, temperature signals and humidity signals of the wind turbine generator set.

9. The wind turbine mechanical component situation awareness device based on multi-source data according to claim 8, characterized in that: The pre-processing unit comprises: A conversion module, used for converting the vibration signal by using short-time Fourier transform; A normalization module, used for normalizing the converted vibration signal, the power output signal, the temperature signal and the humidity signal; A first matrix forming module, used for forming a first information matrix from the normalized vibration signal and the power output signal; The second matrix composition module composes the normalized temperature signal and humidity signal into a second information matrix.

10. The wind turbine mechanical component situation awareness device based on multi-source data according to claim 9, characterized in that: The feature fusion unit comprises: A local feature extraction module, used to extract local features of the first information matrix; A first feature matrix acquisition module, used for introducing a global information fusion strategy based on the local features to obtain a first feature matrix to be fused; A second feature matrix acquisition module, used for extracting feature information of the second information matrix based on a GRU network to obtain a second feature matrix; The feature fusion module is used to fuse the first feature matrix and the second feature matrix through a fully connected layer to obtain a multi-source information fusion matrix.

11. The wind turbine mechanical component situation awareness device based on multi-source data according to claim 7, characterized in that: During training, the situation awareness model maps data from different acquisition sources to a shared feature space by setting a mapping function.

12. The wind turbine mechanical component situation awareness device based on multi-source data according to claim 7, characterized in that: The loss function selected by the situation awareness model during training is as follows: In the above formula, z i (y) is the advantage score of the wind turbine mechanical component state y obtained by the mapper, is the vector u in the multi-source data fusion matrix i Predicted probability of belonging to state y of a mechanical component of the wind turbine.

13. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images