A wind turbine gearbox fault diagnosis system, method, computer equipment, and storage medium

Abstract

The present invention relates to the technical field of wind power generation, and discloses a method for diagnosing faults in a wind turbine gearbox, comprising the following steps: obtaining the front temperature of the gearbox, the rear temperature of the gearbox, the oil temperature of the gearbox, the hub speed, the nacelle wind speed, the pitch angle and the generator power; performing data preprocessing on the nacelle wind speed and the generator power to obtain a wind speed-power curve; performing fault feature extraction on the front and rear temperature of the gearbox, the oil temperature of the gearbox, the hub speed, the pitch angle and the wind speed-power curve to obtain a fault feature extraction value, and establishing a fault diagnosis network model, and the fault diagnosis network model training is successful; after processing the operating data to be diagnosed, the data is input into the fault diagnosis network model to obtain a fault diagnosis result. The method does not rely on the vibration data of the gearbox, but extracts 7 parameter characteristic values, and uses a neural network algorithm to realize real-time prediction of gearbox faults.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular to a wind turbine gearbox fault diagnosis system, method, computer equipment and storage medium. Background Art

[0002] Common types of wind turbine failures include electrical system failures, sensor and blade / pitch device failures, gearbox failures, etc. According to statistics, the gearbox damage rate in my country's wind farms is as high as 40% to 50%, which is the component with the highest failure rate in the unit and the main cause of wind turbine shutdown. Therefore, gearbox status detection in the early stage of gearbox failure, and fault diagnosis and analysis based on this, can effectively diagnose the fault in the early stage, which is conducive to reducing maintenance time and reducing economic losses caused by gearbox failure, which is of great significance to improving the economic benefits and safety of wind farms.

[0003] As one of the most effective signal processing methods in rotating machinery condition monitoring, vibration signal analysis is widely used in fault diagnosis of wind turbine gearboxes. However, the gearbox fault signal of wind turbine gearboxes is often submerged in background noise. Traditional signal processing methods such as wavelet decomposition and empirical mode decomposition separate different components in the signal. These signal processing methods are easily affected by signal timing distortion and discontinuity, which causes the corresponding characteristic points to mutate, resulting in large timing distortion of the decomposed signal. In particular, empirical mode decomposition will be affected by discontinuity points and mutation points, causing the envelope and its corresponding extreme point to be distorted. The intrinsic mode function components obtained by decomposition will also be disturbed by the distortion points, causing the decomposed modal function to be aliased, which ultimately leads to inaccurate diagnosis results. Summary of the invention

[0004] The object of the present invention is to provide a wind turbine gearbox fault diagnosis system, system and method, which solves the problem of inaccurate diagnosis results of the current gearbox fault diagnosis method based on vibration signals.

[0005] The present invention is achieved through the following technical solutions:

[0006] A method for diagnosing a wind turbine gearbox fault comprises the following steps:

[0007] S1. Acquire a gearbox status signal, which includes a gearbox front end temperature, a gearbox rear end temperature, a gearbox oil temperature, a hub speed, a nacelle wind speed, a pitch angle, and a generator power;

[0008] S2. Preprocess the data of the nacelle wind speed and the generator power to obtain the corresponding power midpoint at each wind speed and establish a wind speed-power curve;

[0009] S3, extracting fault features from the gearbox front end temperature, gearbox rear end temperature, gearbox oil temperature, hub speed, pitch angle and wind speed-power curve to obtain fault feature extraction values;

[0010] S4, dividing the fault feature extraction values ​​into a training sample set and a test sample set, establishing a fault diagnosis network model, and using the training sample set to train the fault diagnosis network model; inputting the test sample set into the preliminarily trained fault diagnosis network model for verification until a preset accuracy rate is reached, and the fault diagnosis network model training is successful;

[0011] S5. After processing the operating data to be diagnosed using S2 to S3, the data is input into the trained fault diagnosis network model to obtain the fault diagnosis result.

[0012] Further, S2 is specifically:

[0013] 1.1. Delete invalid data and average the retained data, replacing the instantaneous value with the short-time average value;

[0014] 1.2 Select the data band between the cut-in wind speed and the cut-out wind speed, and divide the data band into several groups according to different wind speeds;

[0015] 1.3 Based on the overall density function, filter out the abnormal power data;

[0016] 1.4 Through averaging, the corresponding power median point at each wind speed is calculated to obtain the wind speed power curve.

[0017] Furthermore, in 1.3, the kernel density-mean method is used to construct the overall density function, specifically:

[0018] Let K() be the kernel function, h be the window width, X 1， X 2 , X 3 , …, X n is a univariate continuous sample, then the kernel density estimate of the overall density function f(x) at any point x is:

[0019]

[0020] Where K(x)≥0,

[0021] Further, in S3, the fault feature extraction value is normalized. The normalization process is as follows: Taking the minimum and maximum values ​​of the parameter data corresponding to each fault feature extraction value as the boundaries and the average value of the parameter data as the benchmark, each specific value is normalized to [0, 1] in proportion. The corresponding formula is:

[0022]

[0023] Where: is the average value of the initial sample data; U is a specific value in the initial sample; U* is the transformed sample data; Umax is the maximum value of the initial sample data; Umin is the minimum value of the initial sample data.

[0024] Furthermore, the fault diagnosis network model includes an input layer, an output layer and a hidden layer.

[0025] Furthermore, the input layer is composed of 6 input indicators, including the gearbox oil temperature characteristic indicator, the gearbox front end temperature characteristic indicator, the gearbox front end temperature characteristic indicator, the hub speed characteristic indicator, the wind speed-power characteristic indicator and the pitch angle characteristic indicator;

[0026] The diagnostic results output by the output layer are normal state, wear state or broken tooth state.

[0027] Further, let the data set be:

[0028]

[0029] Then the gearbox fault state is defined as:

[0030]

[0031] Wherein: n = 1, 2, 3; when the output is 000, it means the unit is in normal state; when the output is 010, it means the gearbox is in a worn state; when the output is 100, it means the gearbox is in a broken tooth state.

[0032] The present invention also discloses a wind turbine gearbox fault diagnosis system, comprising:

[0033] Data acquisition module, used to obtain gearbox status signal;

[0034] The data preprocessing module is used to preprocess the nacelle wind speed and generator power, filter out power data with abnormal distribution, obtain the corresponding power median point at each wind speed, and obtain the wind speed-power curve;

[0035] A fault feature extraction module is used to extract fault features of the gearbox front temperature, gearbox rear temperature, gearbox oil temperature, hub speed, pitch angle and wind speed-power curve to obtain a fault feature extraction value;

[0036] The fault diagnosis module is used to input the fault feature extraction value into the trained fault diagnosis network model to obtain the fault diagnosis result.

[0037] The present invention also discloses a computer 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 wind turbine gearbox fault diagnosis method when executing the computer program.

[0038] The present invention also discloses a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and is characterized in that when the computer program is executed by a processor, the steps of the wind turbine gearbox fault diagnosis method are implemented.

[0039] Compared with the prior art, the present invention has the following beneficial technical effects:

[0040] The present invention discloses a wind turbine gearbox fault diagnosis method, which does not rely on gearbox vibration data, but extracts parameter characteristic values ​​composed of gearbox front end temperature, gearbox rear end temperature, gearbox oil temperature, hub speed, cabin wind speed, pitch angle and generator power. The seven parameters are core parameters of the gearbox, and their parameter values ​​can represent the gearbox operation characteristics. A neural network algorithm is used to realize real-time prediction of gearbox faults. Practice has proved that the method is simple to operate, and the prediction is relatively accurate by comparing simulation results with actual operation data, which has certain guiding significance for gearbox fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 Modeling and implementing a flow chart for fault diagnosis of the present invention;

[0042] Figure 2 It is the neural network diagnosis model of the present invention. DETAILED DESCRIPTION

[0043] The present invention is further described in detail below in conjunction with specific embodiments, which are intended to explain the present invention rather than to limit it.

[0044] Common faults of wind turbine gearboxes mainly include gear damage, bearing damage, and shaft breakage according to the location of occurrence. Gear damage mainly includes: tooth surface wear, tooth surface bonding and abrasion, tooth surface contact fatigue, bending fatigue and broken teeth. Bearing damage mainly includes wear failure, fatigue failure, corrosion failure, fracture failure, indentation failure, and bonding failure. Shaft failures mainly include shaft bending, axial movement, and shaft misalignment.

[0045] According to the operation records of multiple wind farms, yaw gearbox failures are mainly divided into two categories: wear and broken teeth.

[0046] According to different wear mechanisms, gear wear can be divided into four basic types: abrasive wear, adhesive wear, fatigue wear and corrosive wear. Abrasive wear is mainly caused by grooves and micro-cutting, while adhesive wear is closely related to surface molecular forces and frictional heat. Fatigue wear is the result of the initiation and expansion of surface fatigue cracks under cyclic stress, while corrosive wear is caused by the chemical action of the environmental medium.

[0047] During the operation of the gear, the gear may break due to severe impact, eccentric load and uneven material. According to the crack extension and the cause of the broken tooth, the broken tooth includes overload fracture (including impact fracture), fatigue fracture and random fracture, etc. The broken tooth is often caused by the gradual expansion of fine cracks.

[0048] The present invention proposes a wind turbine gearbox fault diagnosis method based on operation data feature extraction, comprising the following steps:

[0049] S1. Acquire a gearbox status signal, which includes a gearbox front end temperature, a gearbox rear end temperature, a gearbox oil temperature, a hub speed, a nacelle wind speed, a pitch angle, and a generator power;

[0050] S2. Preprocess the data of the nacelle wind speed and the generator power to obtain the corresponding power midpoint at each wind speed and obtain the wind speed-power curve;

[0051] S3, extracting fault features from the gearbox front end temperature, gearbox rear end temperature, gearbox oil temperature, hub speed, pitch angle and wind speed-power curve to obtain fault feature extraction values, and dividing the fault feature extraction values ​​into a training sample set and a test sample set;

[0052] S4, establish a fault diagnosis network model, and use the training sample set to train the fault diagnosis network model; input the test sample set into the preliminarily trained fault diagnosis network model for verification until the preset accuracy rate is reached, and the fault diagnosis network model training is successful;

[0053] S5. After processing the operating data to be diagnosed using S1 to S3, the data is input into the trained fault diagnosis network model to obtain the fault diagnosis result.

[0054] First, the existing parameters in the running system are screened, then data preprocessing (i.e. data cleaning) is performed, and finally fault features are extracted from the selected feature parameters to obtain a training sample set for the neural network diagnosis model to find the relationship between parameters and a test sample set for verifying the accuracy of the model.

[0055] like Figure 1As shown in Figure 1, the model consists of two major steps: (1) Modeling process, training a model that can reflect the relationship between parameters, and then using a sample set to test the model accuracy. (2) Implementation process, importing real-time data into the trained model to obtain prediction results.

[0056] The data preprocessing process is divided into the following steps:

[0057] (1) Delete invalid data and average the data, replacing the instantaneous value with the short-time average value;

[0058] (2) selecting a data band between the cut-in wind speed and the cut-out wind speed, and dividing the data band into several groups according to different wind speeds;

[0059] (3) According to the overall density function f(x), the abnormal power data are screened out;

[0060] (4) The corresponding midpoint power at each wind speed is calculated through averaging.

[0061] Furthermore, the kernel density-mean method is used to construct the overall density function, specifically:

[0062] Let K() be the kernel function, h be the window width, X 1， X 2 , X 3 , …, X n is a univariate continuous sample, then the kernel density estimate of the overall density function f(x)f(x) at any point x is:

[0063]

[0064] Where K(x)≥0,

[0065] Furthermore, the data is normalized, and the data format is converted or unified into a form suitable for modeling. The minimum and maximum values ​​of each parameter data are used as boundaries, and the average value of the parameter data is used as a benchmark. Each specific value is normalized to [0, 1] in proportion. The corresponding formula is as follows (2):

[0066]

[0067] Where: is the average value of the initial sample data; U is a specific value in the initial sample; U* is the transformed sample data; Umax is the maximum value of the initial sample data; Umin is the minimum value of the initial sample data.

[0068] The gearbox state signal is extracted using the ReliefF algorithm, which selects parameters by analyzing and comparing the weight values ​​between various parameters, where the weight value represents the degree of correlation between parameters. The basic idea of ​​the algorithm is to find a set of similar samples, group them together, and classify different sample sets, and then update the feature weights through the weight calculation formula.

[0069] The present invention selects 7 parameters from the listed operating parameters as gearbox status signal parameters, namely, gearbox front end temperature, gearbox rear end temperature, gearbox oil temperature, hub speed, cabin wind speed, pitch angle, and generator power.

[0070] In S3, the fault extraction feature model is used to extract the feature values ​​corresponding to the six parameters.

[0071] Fault Diagnosis Neural Network Design:

[0072] The basic structure of BP neural network includes three layers of input, output and hidden layer. The present invention uses Matlab software platform to establish a three-layer BP neural network model to predict gearbox failure. The input layer is 6 input indicators: gearbox oil temperature characteristic indicator, gearbox front temperature characteristic indicator, gearbox front temperature characteristic indicator, hub speed characteristic indicator, wind speed-power characteristic indicator and pitch angle characteristic indicator. Suppose the set after processing the selected operating parameter data is:

[0073]

[0074] Then the yaw gearbox fault state is defined as:

[0075]

[0076] Where: n = 1, 2, 3; when the output is 000, it means the unit is in normal state, when the output is 010, it means the gearbox is in a worn state, and when the output is 100, it means the gearbox has a broken tooth fault and should be handled urgently. The output results of the neural network are shown in Table 1, and the neural network diagram is shown in the attached Figure 2 , X represents input and z represents output.

[0077] Table 1 Neural network output diagnosis results

[0078] model Output Results Normal operation 000 Gearbox normal Wear status 010 Tooth surface wear Broken tooth status 100 Tooth surface fracture, downtime

[0079] The present invention also discloses a wind turbine gearbox fault diagnosis system, comprising:

[0080] Data acquisition module, used to obtain gearbox status signal;

[0081] The data preprocessing module is used to preprocess the nacelle wind speed and generator power, filter out power data with abnormal distribution, obtain the corresponding power median point at each wind speed, and obtain the wind speed-power curve;

[0082] A fault feature extraction module is used to extract fault features of the gearbox front temperature, gearbox rear temperature, gearbox oil temperature, hub speed, pitch angle and wind speed-power curve to obtain a fault feature extraction value;

[0083] The fault diagnosis module is used to input the fault feature extraction value into the trained fault diagnosis network model to obtain the fault diagnosis result.

[0084] The wind turbine gearbox fault diagnosis method of the present invention may be implemented in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may be implemented in 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 codes.

[0085] If a wind turbine gearbox fault diagnosis method of the present invention is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. Computer-readable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals. Among them, the computer storage medium can be any available medium or data storage device that can be accessed by the computer, including but not limited to magnetic storage (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical storage (such as CD, DVD, BD, HVD, etc.), and semiconductor storage (such as ROM, EPROM, EEPROM, non-volatile memory (NANDFLASH), solid-state drive (SSD)), etc.

[0086] In an exemplary embodiment, a computer device is also provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the wind turbine gearbox fault diagnosis method when executing the computer program. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc.

[0087] The present invention is based on data collection and operation data, and establishes a neural network diagnostic model for the unit gearbox. The ReliefF algorithm and the kernel density-mean method are used to extract 7 operating parameters that can reflect the operating conditions of the gearbox, and 6 fault characteristic indicators are extracted as inputs of the neural network diagnostic model to diagnose the normal state, wear fault and broken tooth fault of the yaw gearbox, a total of 3 operating states. The method is simple to operate, accurate in prediction, and has certain guiding significance for gearbox fault diagnosis.

Claims

1. A wind turbine gearbox fault diagnosis method, characterized in that it comprises the following steps: S1. Acquire a gearbox status signal, which includes a gearbox front end temperature, a gearbox rear end temperature, a gearbox oil temperature, a hub speed, a nacelle wind speed, a pitch angle, and a generator power; S2. Preprocess the data of the nacelle wind speed and the generator power to obtain the corresponding power midpoint at each wind speed and establish a wind speed-power curve; S3, extracting fault features from the gearbox front end temperature, gearbox rear end temperature, gearbox oil temperature, hub speed, pitch angle and wind speed-power curve to obtain fault feature extraction values; S4, dividing the fault feature extraction values ​​into a training sample set and a test sample set, establishing a fault diagnosis network model, and using the training sample set to train the fault diagnosis network model; inputting the test sample set into the preliminarily trained fault diagnosis network model for verification until a preset accuracy rate is reached, and the fault diagnosis network model training is successful; S5, after processing the operating data to be diagnosed by S2-S3, input it into the trained fault diagnosis network model to obtain the fault diagnosis result; S2 is specifically: 1.

1. Delete invalid data and average the retained data, replacing the instantaneous value with the short-time average value; 1.

2. Select the data band between the cut-in wind speed and the cut-out wind speed, and divide the data band into several groups according to different wind speeds; 1.

3. According to the overall density function, filter out the abnormal power data; 1.

4. Calculate the corresponding power median point at each wind speed through averaging processing to obtain the wind speed power curve; In 1.3, the kernel density-mean method is used to construct the overall density function, specifically; Let K( ) be the kernel function and h be the window width; is a univariate continuous sample; then the kernel density estimate of the overall density function f(x) at any point x is: Where K(x)≥0, dx=1.

2. A wind turbine gearbox fault diagnosis method according to claim 1, characterized in that: In S3, the fault feature extraction values ​​are normalized. The normalization process is as follows: the minimum and maximum values ​​of the parameter data corresponding to each fault feature extraction value are used as the boundaries, and the average value of the parameter data is used as the benchmark, and each specific value is normalized to [0, 1] in proportion. The corresponding formula is: Where: is the average value of the initial sample data; U is a specific value in the initial sample; U* is the transformed sample data; Umax is the maximum value of the initial sample data; Umin is the minimum value of the initial sample data.

3. A wind turbine gearbox fault diagnosis method according to claim 1, characterized in that: The fault diagnosis network model includes input layer, output layer and hidden layer.

4. A wind turbine gearbox fault diagnosis method according to claim 3, characterized in that: The input layer consists of 6 input indicators, including the gearbox oil temperature characteristic indicator, the gearbox front end temperature characteristic indicator, the gearbox front end temperature characteristic indicator, the hub speed characteristic indicator, the wind speed-power characteristic indicator and the pitch angle characteristic indicator; The diagnostic results output by the output layer are normal state, wear state or broken tooth state.

5. A wind turbine gearbox fault diagnosis method according to claim 1, characterized in that: The dataset is: ； Then the gearbox fault state is defined as: ； Wherein: n=1, 2, 3; when the output is 000, it means the unit is in normal state; when the output is 010, it means the gearbox is in a worn state; when the output is 100, it means the gearbox is in a broken tooth state.

6. A wind turbine gearbox fault diagnosis system implementing the wind turbine gearbox fault diagnosis method according to any one of claims 1 to 5, characterized in that: include: Data acquisition module, used to obtain gearbox status signal; The data preprocessing module is used to preprocess the nacelle wind speed and generator power, filter out power data with abnormal distribution, obtain the corresponding power median point at each wind speed, and obtain the wind speed-power curve; A fault feature extraction module is used to extract fault features of the gearbox front temperature, gearbox rear temperature, gearbox oil temperature, hub speed, pitch angle and wind speed-power curve to obtain a fault feature extraction value; The fault diagnosis module is used to input the fault feature extraction value into the trained fault diagnosis network model to obtain the fault diagnosis result.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the wind turbine gearbox fault diagnosis method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the wind turbine gearbox fault diagnosis method according to any one of claims 1 to 5 are implemented.

Images