An online fault detection method for wind turbines

Abstract

The present invention discloses an online fault monitoring method for a wind turbine, belonging to the technical field of online fault detection of wind power generation units, which specifically integrates fractional-order extended dispersion entropy and cumulative sum control chart, and includes the following steps: collecting historical vibration signals of core components of the wind turbine; using fractional-order extended dispersion entropy to capture deep dynamic information changes in the component vibration signals; detecting possible abnormal conditions in the process of dynamic information changes with the help of the cumulative sum control chart and sending out alarm information; comparing the detection results with the actual operation conditions of the wind turbine to determine the detection performance of the method. The present invention provides a method that can accurately and quickly detect abnormal conditions of the wind turbine and send out alarms, which can be simply and economically realized, provides a strong basis for discovering faults of the wind turbine and carrying out maintenance, and ensures the reliable and stable operation of the wind turbine.

Description

Technical Field

[0001] The present invention belongs to the technical field of on-line fault detection of wind turbine generators, and more specifically, relates to a fault detection method for wind turbine generators based on fractional-order extended dispersion entropy and cumulative sum control chart. Background Technique

[0002] With the gradual depletion of fossil energy, wind energy, as a renewable energy source with huge resource potential, has attracted the attention of countries around the world. However, most wind turbine generators are located in remote areas and usually operate in harsh environments, so they often fail, and in severe cases, they even shut down. Accidental failures and shutdowns will lead to a significant increase in the operation and maintenance costs of wind turbine generators. To solve this problem, the condition monitoring of wind turbine generators has attracted extensive attention.

[0003] In recent years, the research on extracting fault information from vibration signals to reflect the performance degradation of the core components of wind turbine generators and detecting early faults has become the focus. Time-domain features and frequency-domain features are commonly used features because they are simple and have clear physical meanings. However, these commonly used features also have their limitations. For example, the root mean square has good stability for the development of damage, but it cannot detect early faults and give accurate warnings, while the kurtosis index has high sensitivity to early faults, but its stability for monitoring the bearing degradation process is poor. Frequency-domain features are easily affected by noise frequencies. Therefore, how to select the optimal features to reflect the state of wind turbine generators is still a challenging task. Summary of the Invention

[0004] Aiming at the fact that most of the currently commonly used features cannot reveal the fault features hidden behind the vibration signals and have poor noise elimination ability, a fault detection method for wind turbine generators based on fractional-order extended dispersion entropy and cumulative sum control chart is proposed to achieve high-efficiency and high-precision state detection, reduce unnecessary downtime, and provide strong guidance for ensuring the reliable, safe, and economic operation of wind farms.

[0005] To achieve the above object, the present invention provides an on-line fault detection method for wind turbine generators, including:

[0006] (1) Collect the vibration signals of the components of the wind turbine generator;

[0007] (2) Use fractional-order extended dispersion entropy to capture the deep dynamic information changes in the vibration signals of the components;

[0008] (3) Detect possible abnormal conditions in the process of dynamic information changes with the help of a cumulative sum control chart and send out alarm information to complete the fault detection of the wind turbine generator;

[0009] (4) Compare the detection results with the actual operating conditions to determine the detection performance for online fault detection of vibration data in real-time operation.

[0010] In some alternative embodiments, the vibration signals include historical data recorded from the gearbox, high-speed shaft bearings of an operating wind turbine, and a wind farm.

[0011] In some alternative embodiments, step (2) includes:

[0012] Map the vibration signals using the normal cumulative distribution function and convert the mapped signals into a symbol sequence represented by integers.

[0013] Given the time embedding dimension m and time delay d, perform phase space reconstruction on the symbol sequence and the vibration signals to obtain the corresponding phase space sequences of the symbol sequence and the vibration signals respectively.

[0014] Calculate the symbol factor of the vibration signals, construct the dispersion pattern of the symbol sequence, count the number of occurrences of the same value in the dispersion pattern, and find the probability of the occurrence of the same value.

[0015] Calculate the fractional-order extended dispersion entropy based on the probability of the occurrence of the same value.

[0016] In some alternative embodiments, step (3) includes:

[0017] Where E is the fractional-order extended dispersion entropy sequence, g + and g - are the positive and negative cumulative values respectively, h is the drift correction parameter. Given a threshold T, if g + > T or g - > T, the cumulative sum control chart can detect abnormal conditions and send an alarm to the operator in a timely manner.

[0018] In some alternative embodiments, compare the detection results with the actual situation to verify the ability to effectively detect changes in the state of the wind turbine and the early warning ability, and count the false alarm rate.

[0019] In some alternative embodiments, for online fault detection of vibration data in real-time operation, it includes:

[0020] Collect the real-time vibration signals of the core components of the wind turbine.

[0021] Utilize the fractional-order extended dispersion entropy to capture the deep dynamic information changes in the real-time vibration signals of the components.

[0022] Detect possible abnormal conditions in the process of dynamic information changes with the help of the cumulative sum control chart and send an alarm message to complete the fault detection of the wind turbine.

[0023] Generally speaking, compared with the prior art, the above technical solution conceived by the present invention can achieve the following beneficial effects:

[0024] The present invention specifically integrates fractional-order extended dispersion entropy and cumulative sum control chart, which can be used for fault detection of wind turbines. In order to capture weak changes in signals and improve the robustness to noise, fractional-order extended dispersion entropy is innovatively proposed to track the changes in component states, and it is organically integrated with the cumulative sum control chart to propose a fault detection method for wind turbines. In this method, fractional-order extended dispersion entropy is sensitive to weak changes in signals and has good noise robustness. The cumulative sum control chart is simple to implement as a monitoring tool, has a fast processing speed, and high dynamic detection accuracy. If an abnormal situation occurs in the process, it can send an alarm message as soon as possible. Therefore, the fault detection method for wind turbines that integrates fractional-order extended dispersion entropy and cumulative sum control chart can improve the efficiency of equipment condition monitoring and achieve fast and accurate online fault detection. Description of the Drawings

[0025] Figure 1 is a flowchart for implementing an online fault detection method for wind turbines provided by an embodiment of the present invention;

[0026] Figure 2 is the structure of a fractional-order extended dispersion entropy provided by an embodiment of the present invention;

[0027] Figure 3 is the result of an online fault detection method for wind turbines provided by an embodiment of the present invention. Detailed Embodiments

[0028] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0029] The objective of the present invention is to provide a new detection method for online fault detection of wind turbines, with higher detection efficiency and accuracy, and to solve problems such as the traditional method being insensitive to weak changes in vibration signals.

[0030] As Figure 1 shown, an online fault detection method for wind turbines according to an embodiment of the present invention includes the following steps:

[0031] Step 1: First, collect the vibration signals of the operating states of the core components of the wind turbine;

[0032] Among them, the core components of the wind turbine include: gearbox, high-speed shaft, generator bearing, etc. The vibration signals are the running wind turbine and historical data.

[0033] Step 2: Analyze the vibration signals of a certain component of the wind turbine collected by using the fractional-order extended dispersion entropy, and track its dynamic evolution trend;

[0034] The parameter settings of the fractional-order extended dispersion entropy are shown in Table 1.

[0035] Table 1 Parameter settings of fractional-order extended dispersion entropy

[0036]

[0037] Among them, as Figure 2 shown, the method of Step 2 is specifically as follows:

[0038] Given a vibration signal time series x = {x(i), i = 1, 2,..., N}, map x to y = {y(i)} by using the normal cumulative distribution function, and then there is:

[0039]

[0040] Among them, μ is the mean of x; σ is the standard deviation of x. The value range of y is between 0 and 1. Subsequently, y is transformed into a symbol sequence represented by integers:

[0041] z(i) = round(y(i) × c + 0.5)

[0042] Among them, c is the number of symbols; round is the rounding operation; z is the symbol sequence, and its value is an integer from 1 to c.

[0043] Given a time embedding dimension m and a time delay d, perform phase space reconstruction on z and x to obtain the corresponding sequence in the phase space, and its formula is:

[0044] z m,d (j) = [z(j), z(j + d),..., z(j + (m - 1)d)]

[0045] x m,d (j) = {x(j), x(j + d),..., x(j + (m - 1)d)}

[0046] Calculate the symbol factor of the original sequence x:

[0047]

[0048] d x = {|x(i + 1) - x(i)|}, i = 1, 2,..., N - 1

[0049] where d is the difference between adjacent elements in x m,d ; λ is a regulation parameter.

[0050] Construct the dispersion pattern of the sequence:

[0051] q m,d (j) = (z(j), z(j + d),..., z(j + (m - 1)d), Δ(j))

[0052] Count the number of occurrences of the same q m,d and find its probability:

[0053]

[0054] Calculate the fractional - order extended dispersion entropy:

[0055]

[0056] where α is the fractional - order, taking values from - 1 to 1; Γ(·) and ψ(·) are the gamma function and the digamma function respectively, and the expressions are:

[0057]

[0058]

[0059] where x(i) is the i - th value in the time series x, N is the length of the time series, Δ(j) is the j - th symbol factor, type(q m,d ) is the type of the dispersion pattern, represents the (1 - α) - th power of the probability of the i - th dispersion pattern occurring, p i represents the probability of the i - th dispersion pattern occurring.

[0060] Step 3: Detect possible abnormal conditions in the dynamic information change process of the fractional - order extended dispersion entropy with the help of the cumulative sum control chart, and send an alarm message to complete the fault detection of the wind turbine.

[0061] The parameter settings of the cumulative sum control chart are shown in Table 2.

[0062] Table 2 Parameter settings of the cumulative sum control chart

[0063]

[0064] where the method of Step 3 is specifically:

[0065]

[0066] where E is the fractional - order extended dispersion entropy sequence; g + and g -They are the positive and negative cumulative values respectively, with the initial value being 0; h is the drift correction parameter; given a threshold T, if g + >T or g - >T, the cumulative sum control chart can detect abnormal conditions and send an alarm to the operator in a timely manner.

[0067] Step 4: Compare the detection results with the actual situation to verify whether the state change and early warning ability of the wind turbine can be effectively detected;

[0068] According to Figure 3 the results, the fractional-order extended dispersion entropy can effectively track the dynamic changes of the health state of the wind turbine. Based on this, the cumulative sum control chart can give an early warning before the fault fails, and the false alarm rate is low.

[0069] Step 5: Apply the method to the vibration data of real-time operation for online fault detection. When applying online, implement Steps 1 to 3 of the specification.

[0070] It should be noted that according to the needs of implementation, each step / component described in this application can be split into more steps / components, or two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0071] Those skilled in the art can easily understand that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An on-line fault detection method for a wind turbine, characterized in that, Including: (1) Collect the vibration signals of the components of the wind turbine; (2) Use the fractional-order extended dispersion entropy to capture the deep dynamic information changes in the vibration signals of the components; (3) Detect the possible abnormal conditions in the dynamic information change process with the help of the cumulative sum control chart and send out alarm messages to complete the fault detection of the wind turbine; (4) Compare the detection results with the actual operating conditions to determine the detection performance for online fault detection of the vibration data in real-time operation; Step (2) includes: Map the vibration signals using the normal cumulative distribution function and convert the mapped signals into symbol sequences represented by integers; Given the time embedding dimension m and time delay d, perform phase space reconstruction on the symbol sequence and the vibration signals to obtain the corresponding phase space sequences of the symbol sequence and the vibration signals respectively; Calculate the symbol factors of the vibration signals, construct the dispersion patterns of the symbol sequence, count the number of times the same values appear in the dispersion patterns, and calculate the probabilities of the same values appearing; Calculate the fractional-order extended dispersion entropy according to the probabilities of the same values appearing; Step (3) includes: Among them, E is the fractional-order extended dispersion entropy sequence, and g + and g - are the positive and negative cumulative values respectively, h is the drift correction parameter. Given a threshold T, if g + > T or g - > T, the cumulative sum control chart can detect anomalies and send alerts to the operator in a timely manner.

2. The method according to claim 1, wherein Compare the detection results with the actual situation to verify the ability to effectively detect the state changes of the wind turbine and the early warning ability, and count the false alarm rate.

3. The method according to claim 1, characterized in that, For online fault detection of the vibration data in real-time operation, it includes: Collect the real-time vibration signals of the core components of the wind turbine; Use the fractional-order extended dispersion entropy to capture the deep dynamic information changes in the real-time vibration signals of the components; Detect the possible abnormal conditions in the dynamic information change process with the help of the cumulative sum control chart and send out alarm messages to complete the fault detection of the wind turbine.

Images