Fan fault diagnosis method and system

Abstract

The invention provides a fan fault diagnosis method and system, and the method comprises the steps: installing a sensor at a key part, collecting an acceleration signal, and carrying out the filtering processing; performing time domain and frequency domain analysis, and extracting a root-mean-square value, an amplitude entropy value and an instantaneous frequency value; establishing a model based on the instantaneous frequency to calculate a vibration complexity index; a fault diagnosis model is established in combination with the indexes to analyze the component fault probability; setting a risk threshold, comparing the analysis result and generating a diagnosis conclusion; according to the invention, the accuracy and the real-time monitoring capability of fan fault diagnosis can be improved, the maintenance cost is reduced, the reliability and the intelligent level of equipment are improved, and the user experience is enhanced.

Description

Technical Field

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

[0002] As an important mechanical equipment, fans are widely used in industrial production, power generation, building ventilation and other fields. With the acceleration of industrialization and technological advancement, the application scope of fans is becoming wider and wider, and the requirements for their reliability and efficiency are becoming higher and higher. However, during the long-term operation of fans, due to various factors such as wear, vibration, overload, etc., they are prone to failure, resulting in equipment performance degradation or even shutdown, which not only increases maintenance costs, but may also affect the normal operation of the entire production line.

[0003] In recent years, with the development of Internet of Things technology, big data analysis and artificial intelligence technology, the condition monitoring and fault diagnosis technology of mechanical equipment has made significant progress; by installing sensors to collect data on the operating status of equipment and using data analysis technology to process these data, the health status of equipment can be effectively identified and potential faults can be discovered in advance, so that preventive measures can be taken to reduce unplanned downtime and improve the availability and reliability of equipment;

[0004] However, traditional fault diagnosis methods often rely on manual experience and judgment, lack effective data processing methods, and find it difficult to accurately extract useful information from large amounts of data. At the same time, fans are affected by a variety of factors in actual working environments, such as environmental noise, mechanical vibration, etc. These factors can lead to poor quality of the collected signals and affect the accuracy of fault diagnosis. In addition, the fan has a complex internal structure and strong interactions between different components. When a part fails, it may trigger a chain reaction, resulting in diverse and difficult to distinguish fault modes. Although some fault diagnosis systems can identify faults, they often take a long time to complete data processing and analysis, making it difficult to meet the needs of real-time monitoring. Most current fault diagnosis technologies use fixed rules or simple algorithms, lack self-learning and adaptability, and cannot adjust diagnostic strategies as the operating conditions of the equipment change. Summary of the invention

[0005] The present invention provides a method and system for diagnosing fan faults, which can effectively improve the accuracy and efficiency of data processing by filtering, analyzing in the time domain, analyzing in the frequency domain, and analyzing in the complexity of the fan acceleration signal. In particular, by calculating a series of indicators such as the root mean square value, the amplitude entropy value, and the vibration complexity index, and establishing a fault diagnosis model in combination with these indicators, the operating status and fault probability of the fan can be more accurately evaluated.

[0006] In one aspect, the present invention provides a method for diagnosing a fan fault, comprising:

[0007] Step 1: Install acceleration sensors and data collection devices at key locations of the wind turbine;

[0008] Step 2: filtering the collected fan acceleration signal;

[0009] Step 3, performing time domain analysis on the processed acceleration signal to obtain the root mean square value and amplitude entropy value corresponding to the acceleration signal;

[0010] Step 4: Perform frequency domain analysis on the processed acceleration signal to obtain the instantaneous frequency value corresponding to the acceleration signal;

[0011] Step 5: establishing a complex evaluation model based on the instantaneous frequency value and calculating the vibration complexity index;

[0012] Step six, establishing a fault diagnosis model based on the root mean square value, amplitude entropy value and vibration complexity index corresponding to the acceleration signal, and analyzing the probability of failure of each key component;

[0013] Step seven, pre-set a risk probability threshold, compare the probability threshold with the probability result obtained by analysis, and generate a diagnosis result based on the comparison result.

[0014] According to a fan fault diagnosis method provided by the present invention, in the filtering process of the collected fan acceleration signal, the filtering process is implemented by the following formula:

[0015]

[0016] Among them, a(t) is the original acceleration signal collected, a f (t) is the filtered acceleration signal; M is the number of frequency points to be removed, h i (t) is the adaptive filter coefficient at the i-th frequency point, ω i is the i-th frequency point to be removed, δ(ω-ω i ) is the frequency ω i The impulse function at ;

[0017] For each frequency point ω that needs to be removed i , using the impulse function δ(ω-ω i ) selects the frequency point in the frequency domain and then performs a convolution operation with the original signal a(t) to obtain the response at the frequency point. Subsequently, the adaptive filter coefficient h is used i (t) Weight the response and subtract the weighted response from the original signal to remove the effect of the frequency point. Finally, sum all the frequency points that need to be removed to get the final filtering result a f (t).

[0018] According to a fan fault diagnosis method provided by the present invention, the process of performing time domain analysis on the processed acceleration signal to obtain the root mean square value and amplitude entropy value corresponding to the acceleration signal includes:

[0019] Calculate the filtered acceleration signal a based on the RMS value calculation formula f The root mean square value of (t); the root mean square value calculation formula is:

[0020]

[0021] Among them, RMS is the filtered acceleration signal a f The root mean square value of (t), T is the sampling time length;

[0022] The root mean square value RMS is used to reflect the acceleration signal a f (t) The energy level, the larger the RMS value, the more severe the vibration;

[0023] Calculate the filtered acceleration signal a based on the amplitude entropy value calculation formula f (t) The amplitude entropy value; the amplitude entropy value calculation formula is:

[0024]

[0025] Among them, AE is the filtered acceleration signal a f (t), n is the number of preset amplitude intervals, P j is the acceleration signal a f The probability that the amplitude at each moment in (t) falls within the jth amplitude interval;

[0026] The amplitude entropy value AE is used to reflect the acceleration signal a f The complexity of (t), the larger the amplitude entropy value AE, the more complex the signal is and the more unpredictable it is.

[0027] According to a fan fault diagnosis method provided by the present invention, the process of performing frequency domain analysis on the processed acceleration signal to obtain the instantaneous frequency value corresponding to the acceleration signal includes:

[0028] For the acceleration signal a f (t) Perform Hilbert exchange and obtain the exchange result

[0029] Construct the analytical signal z{t} based on the exchange result,

[0030] The phase is calculated based on the analytical signal z{t},

[0031] in, is the phase, denote the real and imaginary parts respectively;

[0032] Preset instantaneous frequency calculation formula:

[0033]

[0034] Where f{t} is the instantaneous frequency, Δt is the time interval between adjacent frequency points;

[0035] Phase Substitute into the instantaneous frequency calculation formula to calculate and obtain the acceleration signal a f (t) The instantaneous frequency f{t} at each moment.

[0036] According to a fan fault diagnosis method provided by the present invention, a complex evaluation model is established based on the instantaneous frequency value, and in calculating the vibration complexity index, the complex evaluation model is:

[0037]

[0038] Wherein, VCI is the vibration complexity index, N is the total number of frequency points of instantaneous frequency f{t}, is the instantaneous frequency of the kth frequency point, and is the average instantaneous frequency of all frequency points.

[0039] The vibration complexity index VCI reflects the complexity of the instantaneous frequency changing with time; the smaller the vibration complexity index VCI is, the smaller the change of the instantaneous frequency is, and the signal is relatively simple; the larger the vibration complexity index VCI is, the greater the change of the instantaneous frequency is, and the signal is relatively complex, which means that the signal contains more frequency components or the frequency changes more dramatically with time.

[0040] According to a fan fault diagnosis method provided by the present invention, in the fault diagnosis model established based on the root mean square value, amplitude entropy value and vibration complexity index corresponding to the acceleration signal, the fault diagnosis model is:

[0041]

[0042] Among them, P(F) is the probability of failure occurrence, β0, β1, β2, and β3 are model parameters, which are determined based on the training of training data, and each model parameter is adaptively adjusted based on the vibration characteristics under different working conditions.

[0043] On the other hand, the present invention provides a fan fault diagnosis system, based on the above-mentioned fan fault diagnosis method, comprising:

[0044] The signal acquisition module is installed in the key parts of the fan and is used to obtain the acceleration signals of each key part;

[0045] A preprocessing module, connected to the signal acquisition module, for filtering the original acceleration signal of the fan;

[0046] A time domain analysis module, which is connected to the preprocessing module and is used to perform time domain analysis on the acceleration signal after filtering to obtain a root mean square value and an amplitude entropy value corresponding to the acceleration signal;

[0047] A frequency domain analysis module, which is connected to the preprocessing module and is used to perform frequency domain analysis on the acceleration signal after filtering to obtain an instantaneous frequency value corresponding to the acceleration signal;

[0048] A complexity analysis module, which is connected to the frequency domain analysis module, establishes a complexity evaluation model based on the instantaneous frequency value, and calculates the vibration complexity index;

[0049] A fault diagnosis module, which is connected to the time domain analysis module and the complexity analysis module, establishes a fault diagnosis model based on the root mean square value, amplitude entropy value and vibration complexity index corresponding to the acceleration signal, and analyzes the probability of failure of each key component;

[0050] A result generating unit is connected to the fault diagnosis module, and pre-sets a risk probability threshold value, and is used to compare the probability threshold value with the probability result obtained by analysis, and generate a diagnosis result based on the comparison result.

[0051] According to a fan fault diagnosis system provided by the present invention, the signal acquisition module includes an acceleration sensor and a data collection device; the acceleration sensor is used to obtain the acceleration signal of each key equipment, and the data collection device is connected to the acceleration sensor to organize the acceleration signal and transmit it to the preprocessing module for subsequent processing and analysis.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] 1. Improved the accuracy of fault diagnosis:

[0054] By filtering the acceleration signal of the fan, the interference components in the signal are effectively removed, which improves the accuracy of subsequent analysis. By calculating indicators such as the root mean square value, amplitude entropy value and vibration complexity index, the operating status of the fan can be more comprehensively evaluated, thereby improving the accuracy of fault diagnosis.

[0055] 2. Enhanced real-time monitoring capabilities:

[0056] The use of fast data processing algorithms can achieve real-time or quasi-real-time data analysis, timely detect potential faults, and reduce unplanned downtime; by setting risk probability thresholds, decisions can be made quickly to improve the efficiency of equipment management;

[0057] 3. Reduced maintenance costs:

[0058] Early fault warning reduces unplanned maintenance caused by faults and reduces maintenance costs. Accurate fault location avoids unnecessary component replacement and saves maintenance resources.

[0059] 4. Improved equipment reliability:

[0060] By continuously monitoring the operating status of the fan, potential faults can be discovered in time, improving the overall reliability of the equipment; by optimizing the maintenance strategy, the service life of the equipment is extended, and the reliability and stability of the equipment are improved;

[0061] 5. Enhanced intelligence level:

[0062] By establishing a fault diagnosis model, the intelligent level of fault diagnosis is improved and the reliance on manual experience is reduced; through the automatic adjustment mechanism of model parameters, the adaptability of the diagnosis model is improved and the intelligent learning ability of the system is enhanced;

[0063] 6. Improved user experience:

[0064] By improving the accuracy and real-time performance of fault diagnosis, user confidence is enhanced and user satisfaction is improved; by reducing unplanned downtime, production continuity is ensured and the competitiveness of the enterprise is improved;

[0065] In summary, the present invention can significantly improve the accuracy and real-time performance of fan fault diagnosis, reduce maintenance costs, and improve the reliability and intelligence level of equipment, thereby bringing higher economic and social benefits to enterprises.

[0066] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0067] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0069] Figure 1 It is a flow chart of a fan fault diagnosis method provided by an embodiment of the present invention;

[0070] Figure 2 It is a structural schematic diagram of a fan fault diagnosis system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0071] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0072] Example 1

[0073] A method for diagnosing fan faults, see Figure 1 , the steps include:

[0074] Step 1: Install acceleration sensors and data collection devices at key locations of the wind turbine;

[0075] Step 2: filtering the collected fan acceleration signal;

[0076] Step 3, performing time domain analysis on the processed acceleration signal to obtain the root mean square value and amplitude entropy value corresponding to the acceleration signal;

[0077] Step 4: Perform frequency domain analysis on the processed acceleration signal to obtain the instantaneous frequency value corresponding to the acceleration signal;

[0078] Step 5: Establish a complex evaluation model based on the instantaneous frequency value and calculate the vibration complexity index;

[0079] Step 6: Establish a fault diagnosis model based on the RMS value, amplitude entropy value and vibration complexity index corresponding to the acceleration signal to analyze the probability of failure of each key component;

[0080] Step seven, pre-set a risk probability threshold, compare the probability threshold with the probability result obtained by analysis, and generate a diagnosis result based on the comparison result.

[0081] The principle of the above embodiment is: installing acceleration sensors and data collection devices at key parts of the fan to ensure that important signals reflecting the operating status of the fan can be captured; filtering the original acceleration signal to remove noise interference, retaining effective information, and providing a higher quality data basis for subsequent analysis; analyzing the characteristics of the processed acceleration signal in the time domain, and calculating the root mean square value (RMS) and amplitude entropy value (Amplitude Entropy), which reflect the energy distribution and complexity of the signal; converting the acceleration signal to the frequency domain, calculating the instantaneous frequency value to identify the contribution of different frequency components, and further revealing the characteristics of the signal; constructing a complexity evaluation model based on the instantaneous frequency value, and calculating the vibration complexity index, which helps to quantify the complexity of the signal; combining the root mean square value, amplitude entropy value and vibration complexity index to establish a fault diagnosis model, and by analyzing the changing trends of these indicators to determine whether a key component has a fault and the probability; setting a risk probability threshold, comparing the fault probability obtained by the diagnosis model with the threshold, and generating a final diagnosis report based on the comparison results to determine whether there is a fault risk;

[0082] The beneficial effects of the above embodiments are: by combining the analysis of the time domain and the frequency domain, the status of the fan can be evaluated from multiple angles, thereby improving the accuracy of fault detection;

[0083] The rapid response mechanism can detect abnormal situations in time, warn of potential failures in advance, and reduce the risk of unplanned downtime;

[0084] By accurately predicting failures, unnecessary repairs can be avoided, while also extending equipment life and reducing overall maintenance costs;

[0085] Regular monitoring and analysis of fan status can help to promptly identify and resolve potential problems and improve the stability and reliability of the entire system;

[0086] The automated analysis process reduces the need for manual intervention and improves the speed and efficiency of fault diagnosis;

[0087] The establishment and improvement of fault diagnosis models make maintenance work more intelligent and reduce dependence on expert experience and intuition.

[0088] In order to further optimize the above embodiment, the collected fan acceleration signal is filtered, and the filtering process is implemented by the following formula:

[0089]

[0090] Among them, a(t) is the original acceleration signal collected, a f (t) is the filtered acceleration signal; M is the number of frequency points to be removed, h i(t) is the adaptive filter coefficient at the i-th frequency point, ω i is the i-th frequency point to be removed, δ(ω-ω i ) is the frequency ω i The impulse function at ;

[0091] For each frequency point ω that needs to be removed i , using the impulse function δ(ω-ω i ) selects the frequency point in the frequency domain and then performs a convolution operation with the original signal a(t) to obtain the response at the frequency point. Subsequently, the adaptive filter coefficient h is used i (t) Weight the response and subtract the weighted response from the original signal to remove the effect of the frequency point. Finally, sum all the frequency points that need to be removed to get the final filtering result a f (t).

[0092] It should be noted that in order to explain the formula of filtering processing in detail, the following example is provided;

[0093] Assume that the acceleration signal a(t) of a key component of a wind turbine is collected. The signal contains five simple sampling points as shown below:

[0094] a(0)=0.01; a(1)=0.02; a(2)=0.03; a(3)=0.04;

[0095] a(4)=0.05;

[0096] Suppose we want to remove the effects of two specific frequency points: ω1 = 31.83 Hz and ω2 = 63.66 Hz. At the same time, assume that we have determined the corresponding adaptive filter coefficients h1(t) and h2(t). To simplify the example, we will assume that these coefficients are constants, h1(t) = 0.2 and h2(t) = 0.1.

[0097] First, we need to calculate the impulse function δ(ω-ω i ). In the discrete case, we can approximate the impulse function with a unit impulse function. Suppose we use a unit impulse function δ(n) that takes the value 1 when n = 0 and takes the value 0 elsewhere.

[0098] For ω1=31.83Hz and ω2=63.66Hz, we need to find the corresponding sampling points. Assuming the sampling frequency is fs=1000Hz, the corresponding sampling point index is:

[0099] The sampling point index corresponding to ω1 is n1 = ω1 / fs*N = 31.83 / 1000*5 = 0.15915. Since we only have five sampling points, we assume n1 = 1 (the nearest integer).

[0100] The sampling point index corresponding to ω2 is n2=ω2 / fs*N=63.66 / 1000*5=0.3183. Similarly, we assume that n2=2 (the nearest integer).

[0101] Next, the sampling point index and other parameters are substituted into the model, and the filtered acceleration signal a can be calculated. f (t).

[0102] In order to further optimize the above embodiment, the process of performing time domain analysis on the processed acceleration signal to obtain the root mean square value and amplitude entropy value corresponding to the acceleration signal includes:

[0103] Calculate the filtered acceleration signal a based on the RMS value calculation formula f The root mean square value of (t); the root mean square value calculation formula is:

[0104]

[0105] Among them, RMS is the filtered acceleration signal a f The root mean square value of (t), T is the sampling time length;

[0106] The root mean square value RMS is used to reflect the acceleration signal a f (t) The energy level, the larger the RMS value, the more severe the vibration;

[0107] Calculate the filtered acceleration signal a based on the amplitude entropy value calculation formula f (t) The amplitude entropy value; the amplitude entropy value calculation formula is:

[0108]

[0109] Among them, AE is the filtered acceleration signal a f (t), n is the number of preset amplitude intervals, P j is the acceleration signal a f The probability that the amplitude at each moment in (t) falls within the jth amplitude interval;

[0110] The amplitude entropy value AE is used to reflect the acceleration signal a f The complexity of (t), the larger the amplitude entropy value AE, the more complex the signal is and the more unpredictable it is.

[0111] It should be noted that the root mean square (RMS) is a statistic used to measure the strength or energy of a signal. RMS can be understood as the "effective value" of a signal, which reflects the overall energy level of the signal. In vibration analysis, a larger RMS value usually means that the equipment is vibrating more violently and there may be potential problems. Sudden changes in the RMS value may be an early indication of equipment failure. For example, if the RMS value increases significantly over a period of time, it may indicate loose parts, imbalance or other faults inside the machine.

[0112] Amplitude entropy is a nonlinear feature used to evaluate the complexity and irregularity of a signal. Amplitude entropy evaluates the complexity of a signal by calculating the probability of the signal amplitude distribution. Specifically, amplitude entropy is calculated by dividing the signal into multiple amplitude intervals and then calculating the probability of amplitude occurrence in each interval; the higher the AE value, the more complex and unpredictable the signal. In vibration analysis, AE can be used to identify whether there are random or nonlinear components in the signal. If a mechanical system begins to fail, the vibration signal it produces may become more complex, causing the AE value to increase. Therefore, AE can be a useful indicator for monitoring the health of mechanical systems.

[0113] Both RMS and AE are important vibration analysis indicators that can help us understand the operating status of the equipment. RMS focuses more on evaluating the overall energy level of the signal, while AE focuses on evaluating the complexity and irregularity of the signal. The combination of the two can provide more comprehensive fault diagnosis information.

[0114] In order to further optimize the above embodiment, the process of performing frequency domain analysis on the processed acceleration signal to obtain the instantaneous frequency value corresponding to the acceleration signal includes:

[0115] For the acceleration signal a f (t) Perform Hilbert exchange and obtain the exchange result

[0116] Construct the analytical signal z{t} based on the exchange result,

[0117] The phase is calculated based on the analytical signal z{t},

[0118] in, is the phase, denote the real and imaginary parts respectively;

[0119] Preset instantaneous frequency calculation formula:

[0120]

[0121] Where f{t} is the instantaneous frequency, Δt is the time interval between adjacent frequency points;

[0122] Phase Substitute into the instantaneous frequency calculation formula to calculate and obtain the acceleration signal a f (t) The instantaneous frequency f{t} at each moment.

[0123] It should be noted that since there is no explicit formula for the Hilbert transform in the discrete case, tools such as MATLAB or Python are usually used for calculation. The result of the Hilbert-Huang HHT transform is the instantaneous frequency (IF) and instantaneous amplitude (IA) of the signal. Specifically, the Hilbert transform can obtain the analytical form of the signal, and then calculate the instantaneous frequency and instantaneous amplitude. The instantaneous frequency describes the change of the signal frequency over time. It can capture the information of the frequency change over time in the signal. The Hilbert transform result represents the Hilbert transform of the original signal, which is used to construct the analytical signal. The phase is calculated from the analytical signal, which represents the phase angle of the signal at each moment.

[0124] In order to further optimize the above embodiment, a complex evaluation model is established based on the instantaneous frequency value. In calculating the vibration complexity index, the complex evaluation model is:

[0125]

[0126] Wherein, VCI is the vibration complexity index, N is the total number of frequency points of instantaneous frequency f{t}, is the instantaneous frequency of the kth frequency point, and is the average instantaneous frequency of all frequency points.

[0127] The vibration complexity index VCI reflects the complexity of the instantaneous frequency changing with time; the smaller the vibration complexity index VCI, the smaller the change in the instantaneous frequency and the relatively simple signal; the larger the vibration complexity index VCI, the greater the change in the instantaneous frequency and the relatively complex signal, which means that the signal contains more frequency components or the frequency changes more dramatically with time.

[0128] It should be noted that the vibration complexity index VCI reflects the complexity of the instantaneous frequency changing over time. Specifically:

[0129] A smaller VCI value indicates that the instantaneous frequency changes less and the signal is relatively simple.

[0130] A larger VCI value indicates that the instantaneous frequency changes more significantly, the signal is more complex, and may contain more frequency components or the frequency changes more dramatically over time.

[0131] In mechanical equipment fault diagnosis, a larger VCI value may indicate that the equipment is abnormal, because the vibration during normal operation is usually relatively stable. Therefore, VCI can be used as an effective fault diagnosis indicator to help detect whether the equipment is in a healthy state.

[0132] In order to further optimize the above embodiment, a fault diagnosis model is established based on the RMS value, amplitude entropy value and vibration complexity index corresponding to the acceleration signal. The fault diagnosis model is:

[0133]

[0134] Among them, P(F) is the probability of failure occurrence, β0, β1, β2, and β3 are model parameters, which are determined based on the training of training data, and each model parameter is adaptively adjusted based on the vibration characteristics under different working conditions.

[0135] It should be noted that in practical applications, the input data needs to be standardized or normalized to ensure model performance. A large amount of labeled data needs to be collected to train the model. This data should contain samples in normal operation and fault conditions. Use cross-validation and other techniques to verify the generalization ability of the model. In practical applications, the threshold of the fault diagnosis model can be adjusted according to actual needs, such as setting a higher threshold to reduce false positives.

[0136] Example 2

[0137] A fan fault diagnosis system, based on the above fan fault diagnosis method, see Figure 2 ,include:

[0138] The signal acquisition module is installed in the key parts of the fan and is used to obtain the acceleration signals of each key part;

[0139] A preprocessing module, which is connected to the signal acquisition module and is used to filter the original acceleration signal of the fan;

[0140] A time domain analysis module, which is connected to the preprocessing module and is used to perform time domain analysis on the acceleration signal after filtering to obtain a root mean square value and an amplitude entropy value corresponding to the acceleration signal;

[0141] A frequency domain analysis module, which is connected to the preprocessing module and is used to perform frequency domain analysis on the acceleration signal after filtering to obtain an instantaneous frequency value corresponding to the acceleration signal;

[0142] A complexity analysis module, which is connected to the frequency domain analysis module, establishes a complexity evaluation model based on the instantaneous frequency value, and calculates the vibration complexity index;

[0143] The fault diagnosis module is connected with the time domain analysis module and the complexity analysis module, and establishes a fault diagnosis model based on the root mean square value, amplitude entropy value and vibration complexity index corresponding to the acceleration signal to analyze the probability of failure of each key component;

[0144] The result generating unit is connected to the fault diagnosis module, and pre-sets a risk probability threshold value, and is used to compare the probability threshold value with the probability result obtained by analysis, and generate a diagnosis result based on the comparison result.

[0145] It should be noted that the signal acquisition module includes an acceleration sensor and a data collection device; the acceleration sensor is used to obtain the acceleration signal of each key device, and the data collection device is connected to the acceleration sensor to organize the acceleration signal and transmit it to the preprocessing module for subsequent processing and analysis.

[0146] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A fan fault diagnosis method, characterized in that: include: Step 1: Install acceleration sensors and data collection devices at key locations of the wind turbine; Step 2: filtering the collected fan acceleration signal; Step 3, performing time domain analysis on the processed acceleration signal to obtain the root mean square value and amplitude entropy value corresponding to the acceleration signal; Step 4: Perform frequency domain analysis on the processed acceleration signal to obtain the instantaneous frequency value corresponding to the acceleration signal; Step 5: establishing a complex evaluation model based on the instantaneous frequency value and calculating the vibration complexity index; Step six, establishing a fault diagnosis model based on the root mean square value, amplitude entropy value and vibration complexity index corresponding to the acceleration signal, and analyzing the probability of failure of each key component; Step seven, pre-set a risk probability threshold, compare the probability threshold with the probability result obtained by analysis, and generate a diagnosis result based on the comparison result.

2. A fan fault diagnosis method according to claim 1, characterized in that: In the filtering process of the collected fan acceleration signal, the filtering process is implemented by the following formula: Among them, a(t) is the original acceleration signal collected, a f (t) is the filtered acceleration signal; M is the number of frequency points that need to be removed, h i (t) is the adaptive filter coefficient at the i-th frequency point, ω i is the i-th frequency point to be removed, δ(ω-ω i ) is the frequency ω i The impulse function at ; For each frequency point ω that needs to be removed i , using the impulse function δ(ω-ω i ) selects the frequency point in the frequency domain, and then performs a convolution operation with the original signal a(t) to obtain the response at the frequency point; then, the adaptive filter coefficient h is used i (t) Weight the response and subtract the weighted response from the original signal to remove the influence of the frequency point; finally, sum all the frequency points that need to be removed to obtain the final filtering result a f (t).

3. A fan fault diagnosis method according to claim 2, characterized in that: The process of performing time domain analysis on the processed acceleration signal to obtain the root mean square value and amplitude entropy value corresponding to the acceleration signal includes: Calculate the filtered acceleration signal a based on the RMS value calculation formula f The root mean square value of (t); the root mean square value calculation formula is: Among them, RMS is the filtered acceleration signal a f The root mean square value of (t), T is the sampling time length; The root mean square value RMS is used to reflect the acceleration signal a f (t) The energy level, the larger the RMS value, the more severe the vibration; Calculate the filtered acceleration signal a based on the amplitude entropy value calculation formula f (t) The amplitude entropy value; the amplitude entropy value calculation formula is: Among them, AE is the filtered acceleration signal a f (t), n is the number of preset amplitude intervals, P j is the acceleration signal a f The probability that the amplitude at each moment in (t) falls within the jth amplitude interval; The amplitude entropy value AE is used to reflect the acceleration signal a f The complexity of (t), the larger the amplitude entropy value AE, the more complex the signal is and the more unpredictable it is.

4. A fan fault diagnosis method according to claim 3, characterized in that: The process of performing frequency domain analysis on the processed acceleration signal to obtain the instantaneous frequency value corresponding to the acceleration signal includes: For the acceleration signal a f (t) Perform Hilbert exchange and obtain the exchange result Construct the analytical signal z{t} based on the exchange result, The phase is calculated based on the analytical signal z{t}, in, is the phase, denote the real and imaginary parts respectively; Preset instantaneous frequency calculation formula: Where f{t} is the instantaneous frequency, Δt is the time interval between adjacent frequency points; Phase Substitute into the instantaneous frequency calculation formula to calculate and obtain the acceleration signal a f (t) The instantaneous frequency f{t} at each moment.

5. A fan fault diagnosis method according to claim 4, characterized in that: In the process of establishing a complex evaluation model based on the instantaneous frequency value and calculating the vibration complexity index, the complex evaluation model is: Wherein, VCI is the vibration complexity index, N is the total number of frequency points of instantaneous frequency f{t}, is the instantaneous frequency of the kth frequency point, and is the average instantaneous frequency of all frequency points; The vibration complexity index VCI reflects the complexity of the instantaneous frequency changing with time; the smaller the vibration complexity index VCI is, the smaller the change of the instantaneous frequency is, and the signal is relatively simple; the larger the vibration complexity index VCI is, the greater the change of the instantaneous frequency is, and the signal is relatively complex, which means that the signal contains more frequency components or the frequency changes more dramatically with time.

6. A fan fault diagnosis method according to claim 5, characterized in that: In the fault diagnosis model established based on the root mean square value, amplitude entropy value and vibration complexity index corresponding to the acceleration signal, the fault diagnosis model is: Among them, P(F) is the probability of failure occurrence, β0, β1, β2, and β3 are model parameters, which are determined based on the training of training data, and each model parameter is adaptively adjusted based on the vibration characteristics under different working conditions.

7. A fan fault diagnosis system, based on a fan fault diagnosis method according to claims 1 to 6, characterized in that: include: The signal acquisition module is installed in the key parts of the fan and is used to obtain the acceleration signals of each key part; A preprocessing module, connected to the signal acquisition module, for filtering the original acceleration signal of the fan; A time domain analysis module, which is connected to the preprocessing module and is used to perform time domain analysis on the acceleration signal after filtering to obtain a root mean square value and an amplitude entropy value corresponding to the acceleration signal; A frequency domain analysis module, which is connected to the preprocessing module and is used to perform frequency domain analysis on the acceleration signal after filtering to obtain an instantaneous frequency value corresponding to the acceleration signal; A complexity analysis module, which is connected to the frequency domain analysis module, establishes a complexity evaluation model based on the instantaneous frequency value, and calculates the vibration complexity index; A fault diagnosis module, which is connected to the time domain analysis module and the complexity analysis module, establishes a fault diagnosis model based on the root mean square value, amplitude entropy value and vibration complexity index corresponding to the acceleration signal, and analyzes the probability of failure of each key component; A result generating unit is connected to the fault diagnosis module, and pre-sets a risk probability threshold value, and is used to compare the probability threshold value with the probability result obtained by analysis, and generate a diagnosis result based on the comparison result.

8. A fan fault diagnosis system according to claim 7, characterized in that: The signal acquisition module includes an acceleration sensor and a data collection device; the acceleration sensor is used to obtain the acceleration signal of each key device, and the data collection device is connected to the acceleration sensor to organize the acceleration signal and transmit it to the preprocessing module for subsequent processing and analysis.

Images