A fan bearing state evaluation method and system based on a sound energy flow deviation index

Abstract

This invention proposes a method and system for assessing the condition of wind turbine bearings based on the sound energy flow deviation index. The method involves: acquiring multiple sound signals under normal stable operation of the wind turbine bearing as reference samples; acquiring the sound signal under stable operation of the current wind turbine bearing as an evaluation sample; calculating the wavelet packet energy flow of each reference sample and the wavelet packet energy flow of the evaluation sample; calculating the wavelet packet energy flow deviation reference value for each pair of wavelet packet energy flows of the reference samples, and selecting the maximum reference value for wavelet packet energy flow deviation; calculating the wavelet packet energy flow deviation between the evaluation sample and any reference sample; if the wavelet packet energy flow deviation is greater than the maximum reference value for wavelet packet energy flow deviation, the wind turbine bearing is abnormal; otherwise, the wind turbine bearing is normal. This method features high assessment speed and high accuracy.

Description

Technical Field

[0001] This invention relates to the field of civil engineering, and specifically to a method and system for assessing the condition of wind turbine bearings based on the sound energy flow deviation index. Background Technology

[0002] Fans are common equipment in industrial production, and their malfunctions can seriously affect production efficiency and safety. Common types of fan malfunctions include bearing wear, blade damage, motor failure, and fan imbalance. These malfunctions can cause vibration, increased noise, and increased energy consumption during operation, and in severe cases, may lead to equipment shutdown, damage to surrounding equipment, or personal injury. Therefore, timely detection and prevention of fan malfunctions are crucial to ensuring the normal operation of production equipment.

[0003] Currently, wind turbine safety management mainly relies on regular maintenance, judging the turbine's condition by inspecting and repairing various parts. This can easily lead to blind maintenance and wasted resources. The wind turbine bearing is the heart of the turbine; its normal operation is crucial for overall production safety and efficiency. According to statistics, bearing-related failures account for up to 40% of wind turbine malfunctions. Therefore, assessing the condition of wind turbine bearings is beneficial for predicting turbine failures. A commonly used method for wind turbine bearing condition assessment is vibration signal fault diagnosis. However, traditional vibration signal fault diagnosis methods have limitations due to the measurement methods used, sometimes failing to extract effective information, thus hindering accurate assessment of the wind turbine bearing condition. Summary of the Invention

[0004] In order to overcome the defects in the prior art, the purpose of this invention is to provide a method and system for evaluating the condition of wind turbine bearings based on the sound energy flow deviation index.

[0005] To achieve the above-mentioned objectives of this invention, this invention provides a method for evaluating the condition of wind turbine bearings based on the sound energy flow deviation index, characterized by comprising the following steps:

[0006] Multiple sound signals under normal stable operation of the fan bearing were acquired as reference samples;

[0007] Acquire the sound signal of the current wind turbine bearing under stable operation as an evaluation sample;

[0008] Calculate the wavelet packet energy flow for each reference sample and the wavelet packet energy flow for the evaluation sample;

[0009] For each reference sample, calculate the wavelet packet energy flow deviation reference value pairwise, and select the reference value with the largest wavelet packet energy flow deviation.

[0010] Calculate the wavelet packet energy flow deviation between the evaluation sample and any reference sample;

[0011] If the wavelet packet energy flow deviation is greater than the maximum reference value for wavelet packet energy flow deviation, then the fan bearing is abnormal; otherwise, the fan bearing is normal.

[0012] This method extracts the wavelet packet energy flow characteristics of the sound signal of the wind turbine bearing and analyzes its energy flow deviation to achieve the condition assessment of the wind turbine bearing. It has the characteristics of high speed assessment and high accuracy.

[0013] In one option of the wind turbine bearing condition assessment method based on the sound energy flow deviation index, the original array signal of the sound under stable operation of the wind turbine bearing is obtained, and the original array signal is beamformed to obtain the sound signal.

[0014] Since the acquisition of the wind turbine bearing sound signal is carried out in a closed laboratory, the original array signal contains severe reverberation and echo. This alternative solution can effectively remove reverberation and echo, and obtain a cleaner and more accurate wind turbine bearing sound signal, providing more accurate data support for the judgment of wind turbine bearing abnormalities.

[0015] In one alternative scheme of this wind turbine bearing condition assessment method based on the sound energy flow deviation index, the steps for calculating the wavelet packet energy flow of the sound signal are as follows:

[0016] Wavelet packet decomposition is performed on the sound signal to obtain the wavelet packet decomposition coefficients;

[0017] Based on the wavelet packet decomposition coefficients, a wavelet packet energy feature vector with the same length as the sound signal is reconstructed;

[0018] The wavelet packet energy feature vector is transformed into a new wavelet packet phase space to obtain the wavelet packet energy flow of the sound signal. During the transformation, the time delay is equal to the embedding dimension.

[0019] In one option of this wind turbine bearing condition assessment method based on sound energy flow deviation index, the wavelet packet energy feature vector is: in The nth reconstruction coefficient of the p-th wavelet packet node in the i-th layer. n refers to the ordinal number of the reconstruction coefficient of the p-th wavelet packet node in the i-th layer, and N is the length of the audio signal;

[0020] The wavelet packet energy flow of the sound signal is

[0021] In one option of the wind turbine bearing condition assessment method based on the sound energy flow deviation index, the calculation formula for the wavelet packet energy flow deviation is as follows:

[0022]

[0023] Among them, WPI A For the wavelet packet energy flow of a sound signal, WPI B For the wavelet packet energy flow of another sound signal.

[0024] The present invention also proposes a wind turbine bearing condition assessment system, including a processing module, a storage module, and an array sensing module for acquiring sound signals of wind turbine bearings;

[0025] The array sensing module is communicatively connected to the processing module and sends the acquired wind turbine bearing sound signal to the processing module. The processing module is communicatively connected to the storage module, which stores at least one executable instruction. The executable instruction causes the processing module to perform the operation corresponding to the wind turbine bearing status assessment method based on the sound energy flow deviation index described above, based on the wind turbine bearing sound signal, to assess the wind turbine bearing status.

[0026] Optionally, the array sensing module includes a microphone array placed near the wind turbine to receive the wind turbine bearing sound signal. The obtained wind turbine bearing sound is a multi-channel signal. The microphone array also performs buffering, amplification, low-pass filtering, and A / D conversion on the received wind turbine bearing sound signal.

[0027] It also includes a signal synchronization set device composed of an array driver board and a development board for synchronously receiving multi-channel signals; the signal synchronization set device is communicatively connected to the processing module and sends multi-channel wind turbine bearing sound signals to the processing module.

[0028] Optionally, the processing module performs beamforming on the multi-channel wind turbine bearing sound signals to obtain a single-channel wind turbine bearing sound signal, and the processing module evaluates the condition of the wind turbine bearing based on the single-channel wind turbine bearing sound signal.

[0029] The beneficial effects of this invention are:

[0030] This invention uses arrayed sound signals, which have a certain effect of noise reduction and reverberation removal. Furthermore, the signal obtained after wavelet packet reconstruction can also achieve filtering and noise reduction, thereby increasing the signal-to-noise ratio of the sound signal and providing a reliable data source for subsequent construction of wavelet packet energy flow. By analyzing the deviation of wavelet packet energy flow, the wind turbine status can be identified.

[0031] This invention overcomes the shortcomings of traditional vibration signal fault diagnosis methods and combines non-contact detection of acoustic signals, which has the advantages of easy signal collection and easy detection of early faults.

[0032] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0033] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0034] Figure 1 This is a flowchart of Example 1;

[0035] Figure 2 This is a schematic diagram of a binary decomposition tree;

[0036] Figure 3 This is a schematic diagram of the principle and structure of the array sensing module. Detailed Implementation

[0037] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0038] In the description of this invention, unless otherwise specified and limited, it should be noted that the terms "installation", "connection" and "linking" should be interpreted broadly. For example, they can refer to mechanical or electrical connections, or internal connections between two components. They can be direct connections or indirect connections through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms according to the specific circumstances.

[0039] Example 1

[0040] like Figure 1 As shown, this invention provides a method for evaluating the condition of wind turbine bearings based on the sound energy flow deviation index, comprising the following steps:

[0041] Multiple sound signals under normal, stable operation of the wind turbine bearing are acquired as reference samples. These reference samples can be obtained from a database or directly acquired. When directly acquired, the same acquisition method as for the evaluation samples can be used. Assume there are λ reference samples, named Data-1 to Data-λ, each containing m data points, for example: Data-λ = (x1, x2, ..., xm), where xm refers to the m-th data point.

[0042] The sound signal under stable operation of the current wind turbine bearing is acquired as an evaluation sample, denoted as Data-r, containing m data points. In this embodiment, the evaluation sample is acquired by setting up an array sensing module. Specifically, as shown... Figure 3 As shown, the array sensing module includes a microphone array, an array driver board, and a development board. The microphone array includes a microphone module and an array mounting board. The microphone array is placed near the fan, close to the motor side, to receive sound signals from the fan bearings. Simultaneously, it performs preprocessing on the received signals, including buffering, amplification, low-pass filtering, and A / D conversion. The array driver board and development board together constitute a signal synchronization device, used to synchronously receive multi-channel signals, ensuring the accuracy of the phase of each channel. The data is then transmitted to a host computer or processing module via a network cable, providing raw array data for data processing and fault diagnosis.

[0043] Since the sound signal acquisition of the wind turbine bearing was conducted in a closed laboratory, the original array signal contained severe reverberation and echo. Therefore, in this embodiment, beamforming was performed on the original array data to obtain the output sound signal y(t). The output sound signal y(t) is a weighted sum of multiple channels after processing; the number of channels corresponds to the number of microphones on the microphone array. In this embodiment, the output sound signal y(t) is a weighted sum of 8 channels after processing.

[0044] Calculate the wavelet packet energy flow WPI-1 to WPI-λ for each reference sample and the wavelet packet energy flow WPI-r for the evaluation sample.

[0045] The wavelet packet energy flow of the reference sample and the evaluation sample are calculated using the same method.

[0046] First, iterative wavelet packet decomposition is performed on the sound signal y(t) in the reference sample or evaluation sample using a low-pass filter h(k) and a high-pass filter g(k) to obtain the wavelet packet decomposition coefficients:

[0047]

[0048] in, Let be the wavelet decomposition coefficients of the p-th node at the i-th layer. and The coefficients represent the wavelet decomposition coefficients at the 2p-th and 2p+1-th nodes in the (i+1)-th layer, corresponding to the approximation coefficients and detail coefficients, respectively. The approximation coefficients reflect the overall characteristics of the signal, while the detail coefficients reflect the local details. In wavelet transform, the approximation coefficients contain the low-frequency components of the signal, while the detail coefficients contain the high-frequency components. In a binary decomposition tree, 2^i nodes (frequency bands) can be obtained in the i-th layer, and the nodes can be numbered (i, p) (p = 0, 1, 2, ..., 2^i-1), as shown below. Figure 2 As shown.

[0049] Then, a wavelet packet energy feature vector with the same length as the sound signal is reconstructed based on the wavelet packet decomposition coefficients. Conversely to the recursive splitting operation in equation (1), the reconstruction process can be expressed based on the wavelet decomposition coefficients as follows:

[0050]

[0051] in This represents the wavelet packet coefficients of the p-th node in the i-th layer of the reconstruction. This indicates that a zero is inserted next to each point d (corresponding to the downsampling operation in the decomposition process). That is, during wavelet packet reconstruction, zeros are padded between every two wavelet decomposition coefficients, and a 2x upsampling operation is performed. H(k-2τ) and G(k-2τ) refer to the low-pass filter and high-pass filter, respectively. Compared with h(k) and g(k) above, they differ only in time reversal, where k refers to the k-th wavelet packet coefficient. Except for node (i,p), all wavelet packet node coefficients in the i-th layer are set to zero. That is, when reconstructing the p-th wavelet packet node coefficient in the i-th layer, the coefficients of the remaining wavelet packet nodes are set to 0. Therefore, 2^i signals with the same length as the original signal y(t) can be reconstructed, retaining the frequency information of the reconstructed nodes. Since the frequency information of the wind turbine varies with its state, the distribution characteristics of the reconstructed coefficients also differ, providing a theoretical basis for state recognition. The frequency information of the reconstructed nodes retained here provides data support for state recognition.

[0052] It is foreseeable that the reconstructed coefficients obtained from wavelet packet nodes contain rich class information, and when a fault occurs, the coefficients of some nodes will change significantly. As an intuitive feature, energy is first calculated from the wavelet packet reconstructed coefficients. This paper reconstructs the energy features distributed on the wavelet packet nodes and explores new energy images among different wavelet packet nodes.

[0053] To distinguish that what is needed in this application are reconstruction coefficients, according to equations (1) and (2), the reconstruction coefficients of the p-th wavelet packet node (i,p) of the i-th layer are... Represented as n refers to the ordinal number of the reconstruction coefficients of the p-th wavelet packet node in the i-th layer, i.e. The nth reconstruction coefficient of the p-th wavelet packet node in the i-th layer is used to calculate the corresponding wavelet packet energy feature:

[0054]

[0055] Where N is the length of the sound signal.

[0056] Finally, the wavelet packet energy eigenvectors are transformed into a new wavelet packet phase space to obtain the wavelet packet energy flow of the sound signal. The number of these energy eigenvectors is 2. i Transforming the energy eigenvectors into a new wavelet packet phase space allows for the reconstruction of the dynamic structure of the frequency distribution within the wavelet packet energy flow. The j-th phase vector in the m-dimensional phase space is set as... In this patent, the time delay τ is set to be equal to the embedding dimension m. Therefore, the wavelet packet energy flow can be redefined and defined mathematically as follows:

[0057]

[0058] In the formula, i represents the decomposition level, and q represents the dimension of the wavelet packet energy flow, which is the dimension of the wavelet packet energy flow obtained after determining the embedding dimension. In this application, q = m, and the size is 2. i / 2 ×2 i / 2 In this multidimensional data construction process, the dynamic structure of WPI can reflect the discrimination information of different fault categories, which is beneficial to the performance evaluation of wind turbine bearings.

[0059] The wavelet packet energy flow WPI-1 to WPI-n of each reference sample and the wavelet packet energy flow WPI-r of the evaluation sample are obtained by following the above steps.

[0060] Then, for each reference sample, the wavelet packet energy flow deviation reference value SWPI is calculated pairwise for each wavelet packet energy flow WPI-1 to WPI-n, and the calculation formula is as follows:

[0061]

[0062] Among them, WPI A For the wavelet packet energy flow of an audio signal involved in the calculation, WPI B The wavelet packet energy flow of another sound signal involved in the calculation.

[0063] Select the reference value with the largest wavelet packet energy flow deviation (SWPI) from all wavelet packet energy flow deviation reference values. max :

[0064] SWPI| max =max(SWPI1…SWPI (n-1)！ (6)

[0065] And calculate the wavelet packet energy flow deviation (SWPI) between the evaluation sample's wavelet packet energy flow (WPI-r) and any reference sample's wavelet packet energy flow. r The calculation formula is referenced in formula (5).

[0066] Comparison of wavelet packet energy flow deviation (SWPI) r Maximum reference value for wavelet packet energy flow deviation (SWPI)| max If the wavelet packet energy flow deviates from the SWPI r Not greater than the maximum reference value of wavelet packet energy flow deviation (SWPI)| max If the fan bearing is normal, then the wavelet packet energy flow deviates from the SWPI. rThe maximum reference value for wavelet packet energy flow deviation is greater than SWPI| max If so, the fan bearing is abnormal.

[0067] Example 2

[0068] This embodiment also provides a wind turbine bearing condition assessment system, which includes a processing module, a storage module, and an array sensing module for acquiring sound signals from wind turbine bearings.

[0069] The array sensing module is communicatively connected to the processing module, sending the acquired wind turbine bearing sound signals to the processing module. Specifically, the array sensing module includes a microphone array placed near the wind turbine to receive the wind turbine bearing sound signals. The received wind turbine bearing sound signals are multi-channel signals. The microphone array also buffers, amplifies, low-pass filters, and performs A / D conversion on the received wind turbine bearing sound signals. It also includes a signal synchronization set device composed of an array driver board and a development board for synchronously receiving multi-channel signals. The signal synchronization set device is communicatively connected to the processing module, sending the multi-channel wind turbine bearing sound signals to the processing module. The processing module beamforms the multi-channel wind turbine bearing sound signals to obtain a single-channel wind turbine bearing sound signal.

[0070] The processing module and the storage module are interconnected. The storage module is used to store at least one executable instruction. The executable instruction enables the processing module to perform the operation corresponding to the wind turbine bearing condition assessment method based on the sound energy flow deviation index in Example 1 according to the single-channel wind turbine bearing sound signal, and to assess the condition of the wind turbine bearing.

[0071] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0072] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for assessing the condition of wind turbine bearings based on the sound energy flow deviation index, characterized in that, Includes the following steps: Multiple sound signals under normal stable operation of the fan bearing were acquired as reference samples; Acquire the sound signal of the current wind turbine bearing under stable operation as an evaluation sample; The wavelet packet energy flow of each reference sample and the wavelet packet energy flow of the evaluation sample are calculated. The steps for calculating the wavelet packet energy flow of the sound signal are as follows: Wavelet packet decomposition is performed on the sound signal to obtain the wavelet packet decomposition coefficients; Based on the wavelet packet decomposition coefficients, a wavelet packet energy feature vector with the same length as the sound signal is reconstructed; The wavelet packet energy feature vector is in The nth reconstruction coefficient of the p-th wavelet packet node in the i-th layer. n refers to the ordinal number of the reconstruction coefficient of the p-th wavelet packet node in the i-th layer, and N is the length of the audio signal; The wavelet packet energy flow of the sound signal is The wavelet packet energy feature vector is transformed into a new wavelet packet phase space to obtain the wavelet packet energy flow of the sound signal. During the transformation, the time delay is equal to the embedding dimension. For each reference sample, calculate the wavelet packet energy flow deviation reference value pairwise, and select the reference value with the largest wavelet packet energy flow deviation. The wavelet packet energy flow deviation between the evaluation sample and any reference sample is calculated using the following formula: Among them, WPI A For the wavelet packet energy flow of a sound signal, WPI B For the wavelet packet energy flow of another sound signal; If the wavelet packet energy flow deviation is greater than the maximum reference value for wavelet packet energy flow deviation, then the fan bearing is abnormal; otherwise, the fan bearing is normal.

2. The method for evaluating the condition of wind turbine bearings based on the sound energy flow deviation index according to claim 1, characterized in that, The original array signal of the sound under stable operation of the wind turbine bearing is acquired, and the original array signal is beamformed to obtain the sound signal.

3. A wind turbine bearing condition assessment system, characterized in that, It includes a processing module, a storage module, and an array sensing module for acquiring sound signals from the wind turbine bearings; The array sensing module is communicatively connected to the processing module and sends the acquired wind turbine bearing sound signal to the processing module. The processing module is communicatively connected to the storage module. The storage module is used to store at least one executable instruction. The executable instruction causes the processing module to perform the operation corresponding to the wind turbine bearing status assessment method based on the sound energy flow deviation index as described in claim 1 or 2 according to the wind turbine bearing sound signal, and to assess the wind turbine bearing status.

4. The wind turbine bearing condition assessment system according to claim 3, characterized in that, The array sensing module includes a microphone array placed near the wind turbine to receive the wind turbine bearing sound signal. The wind turbine bearing sound is a multi-channel signal. The microphone array also performs buffering, amplification, low-pass filtering and A / D conversion on the received wind turbine bearing sound signal. It also includes a signal synchronization set device composed of an array driver board and a development board for synchronously receiving multi-channel signals; the signal synchronization set device is communicatively connected to the processing module and sends multi-channel wind turbine bearing sound signals to the processing module.

5. The wind turbine bearing condition assessment system according to claim 4, characterized in that, The processing module beamforms the multi-channel wind turbine bearing sound signals to obtain a single-channel wind turbine bearing sound signal, and then evaluates the condition of the wind turbine bearing based on this single-channel sound signal.

Images