Intelligent monitoring system for fan tower bolt based on sensor technology

Abstract

The application relates to the technical field of data processing, in particular to a fan tower drum bolt intelligent monitoring system based on sensor technology, which comprises a memory, a processor and a computer program stored in the memory and running on the processor, and the processor realizes the following steps when executing the computer program: EMD decomposition is performed on the original vibration signal of any bolt to obtain a to-be-authenticated component of the original vibration signal; the IMF judgment condition is optimized according to the energy and amplitude of the to-be-authenticated component and the similarity between the original vibration signal and the to-be-authenticated component, and an IMF component is obtained; the IMF component is removed from the original vibration signal to obtain a residual signal, and the residual signal is continuously subjected to EMD decomposition until a stop iteration decomposition condition is met, all IMF components are obtained, and fault monitoring is performed on any bolt according to all the IMF components, thereby improving the accuracy of monitoring the fan tower drum bolt.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a fan tower drum bolt intelligent monitoring system based on sensor technology. BACKGROUND

[0002] The wind turbine is usually located in a relatively harsh environment, and is affected by the overall vibration of the tower drum caused by external wind and the self-weight operation of the unit, so that the tower drum connecting bolt may be loose, or even fatigue fracture, which may cause the wind turbine to collapse and other major safety hazards, so it is extremely important to monitor the wind turbine tower drum bolt.

[0003] Since the vibration signal is very sensitive to slight mechanical changes, it is an effective signal source that can detect problems early, so the vibration signal of the fan tower drum bolt is collected by the sensor, and the vibration signal is decomposed by EMD, and the key information is extracted, and then the extracted key information is used to monitor the vibration signal in real time and identify potential problems.

[0004] However, in the process of decomposing the bolt vibration signal by EMD algorithm, the intermediate component needs to be determined by IMF, and the two determination conditions of IMF in the traditional way are to ensure that the intermediate component can reflect the smooth and symmetrical vibration mode of the original signal, but the actual fan tower drum bolt vibration signal is often non-stationary and sudden, with irregular frequency components and complex signal characteristics, resulting in that part of the intermediate component containing key information cannot fully meet the determination condition of IMF and is excluded when using EMD algorithm to decompose it, and then when the bolt vibration signal is decomposed, part of the key information is lost, which affects the monitoring result of the fan tower drum bolt.

[0005] Therefore, how to optimize the determination condition of IMF in the EMD algorithm to improve the accuracy of monitoring the fan tower drum bolt becomes a problem to be solved. SUMMARY

[0006] Therefore, the embodiments of the present application provide a fan tower drum bolt intelligent monitoring system based on sensor technology to solve the problem of how to optimize the determination condition of IMF in the EMD algorithm to improve the accuracy of monitoring the fan tower drum bolt.

[0007] The embodiments of the present application provide a fan tower drum bolt intelligent monitoring system based on sensor technology, which includes a memory, a processor and a computer program stored in the memory and running on the processor, characterized in that the processor implements the following steps when executing the computer program:

[0008] The original vibration signal of any bolt at the wind turbine tower drum within a preset time period is acquired, during EMD decomposition of the original vibration signal, an envelope of the original vibration signal is acquired, and a component to be identified of the original vibration signal is obtained according to the envelope;

[0009] According to the energy value of the component to be identified and the extreme value in the component to be identified, a response degree of the original vibration signal in the component to be identified is obtained, according to the similarity of the envelope between the original vibration signal and the component to be identified, a change trend similarity degree of the original vibration signal and the component to be identified is obtained, and according to the amplitude value in the component to be identified, a non-stationary characteristic value of the component to be identified is obtained.

[0010] According to the response degree, the change trend similarity degree, and the non-stationary characteristic value, the IMF judgment condition of the component to be identified is optimized to obtain a corresponding IMF component.

[0011] The IMF component is removed from the original vibration signal to obtain a residual signal, and the residual signal is continuously subjected to EMD decomposition until a stop iteration decomposition condition is met to obtain all IMF components of the original vibration signal, and any bolt at the wind turbine tower drum is subjected to fault monitoring according to all IMF components.

[0012] Preferably, according to the energy value of the component to be identified and the extreme value in the component to be identified, the response degree of the original vibration signal in the component to be identified comprises:

[0013] The sum of squares of all amplitude values in the component to be identified is calculated to obtain the energy value of the component to be identified, and the energy value is taken as an independent variable of a hyperbolic tangent function to obtain a first function value;

[0014] The maximum value and the minimum value in the component to be identified are acquired, the absolute value of the difference between the cumulative value of all maximum values and the cumulative value of all minimum values is calculated, and the absolute value of the difference is taken as an independent variable of a hyperbolic tangent function to obtain a second function value;

[0015] The average value between the first function value and the second function value is calculated to obtain the response degree of the original vibration signal in the component to be identified.

[0016] Preferably, according to the similarity of the envelope between the original vibration signal and the component to be identified, the change trend similarity degree of the original vibration signal and the component to be identified comprises:

[0017] an envelope of the original vibration signal is denoted as an original envelope, an envelope of the component to be identified is denoted as a component envelope, all extreme points in the original envelope are obtained, a slope between every two adjacent extreme points in the original envelope is calculated, a slope sequence of the original envelope is obtained, a slope sequence of the component envelope is obtained, an envelope slope feature difference value between the original envelope and the component envelope is obtained according to a difference between the slope sequences of the original envelope and the component envelope, the envelope slope feature difference value is inversely proportional normalized by using an exponential function with a natural constant as a base, and a first similarity index of the original vibration signal and the component to be identified is obtained;

[0018] a DTW distance between the original vibration signal and the component to be identified is obtained, the DTW distance is inversely proportional normalized by using an exponential function with a natural constant as a base, and a second similarity index of the original vibration signal and the component to be identified is obtained;

[0019] an average value between the first similarity index and the second similarity index is calculated, and a change trend similarity degree of the original vibration signal and the component to be identified is obtained.

[0020] Preferably, the envelope slope feature difference value between the original envelope and the component envelope is obtained according to a difference between the slope sequences of the original envelope and the component envelope, and the envelope slope feature difference value between the original envelope and the component envelope includes:

[0021] a difference value between a maximum value and a minimum value in the slope sequence of the original envelope is calculated, a mutation index of the original envelope is obtained, the mutation index of the original envelope and a mutation index of the component envelope are obtained according to the slope sequence of the component envelope, an absolute value of a difference between the mutation index of the original envelope and the mutation index of the component envelope is calculated, and a first slope difference degree between the original envelope and the component envelope is obtained.

[0022] an average value of all data in the slope sequence of the original envelope is calculated, a trend feature value of the original envelope is obtained, the trend feature value of the original envelope and a trend feature value of the component envelope are obtained according to the slope sequence of the component envelope, an absolute value of a difference between the trend feature value of the original envelope and the trend feature value of the component envelope is calculated, and a second slope difference degree between the original envelope and the component envelope is obtained.

[0023] the first slope difference degree and the second slope difference degree are added, and the envelope slope feature difference value between the original envelope and the component envelope is obtained.

[0024] Preferably, the non-stationary feature value of the component to be identified is obtained according to an amplitude in the component to be identified.

[0025] obtaining a maximum amplitude in the component to be identified and an average amplitude of all amplitudes in the component to be identified, normalizing an absolute value of a difference between the maximum amplitude and the average amplitude by using a hyperbolic tangent function to obtain a first characteristic value;

[0026] obtaining all extreme points in the component to be identified, calculating a slope between every two adjacent extreme points in the component to be identified, normalizing a variance of all slopes by using a hyperbolic tangent function to obtain a second characteristic value;

[0027] calculating an average value between the first characteristic value and the second characteristic value to obtain a non-stationary characteristic value of the component to be identified.

[0028] Preferably, the IMF judgment condition of the component to be identified is optimized according to the response degree, the change trend similarity degree and the non-stationary characteristic value to obtain a corresponding IMF component, which comprises:

[0029] calculating an average value of the response degree, the change trend similarity degree and the non-stationary characteristic value to obtain a structural characteristic value of the component to be identified;

[0030] obtaining an initial condition judgment value of the component to be identified according to an IMF judgment condition in the EMD decomposition, performing weighted summation on the initial condition judgment value and the structural characteristic value to obtain a final condition judgment value of the component to be identified, and determining an IMF component of the original vibration signal according to the final condition judgment value.

[0031] Preferably, the initial condition judgment value of the component to be identified is obtained according to the IMF judgment condition in the EMD decomposition, which comprises:

[0032] in the component to be identified, obtaining a number of all extreme points and a number of all zero points, inversely normalizing an absolute value of a difference between the number of all extreme points and the number of all zero points by using an exponential function with a natural constant as a base to obtain a first initial condition value;

[0033] obtaining an intrinsic mode function of the component to be identified, performing integration on the intrinsic mode function, inversely normalizing a result of the integration by using an exponential function with a natural constant as a base to obtain a second initial condition value;

[0034] calculating an average value between the first initial condition value and the second initial condition value to obtain the initial condition judgment value of the component to be identified.

[0035] Preferably, the IMF component of the original vibration signal is determined according to the final condition judgment value, which comprises:

[0036] If the final condition judgment value is within a preset IMF condition judgment value range, the component to be identified is determined as an IMF component of the original vibration signal.

[0037] Compared with the prior art, the embodiment of the present application has the following beneficial effects:

[0038] The original vibration signal of any bolt at the fan tower drum within a preset time period is obtained, in the process of EMD decomposition of the original vibration signal, an envelope line of the original vibration signal is obtained, and a component to be identified of the original vibration signal is obtained according to the envelope line; the response degree of the original vibration signal in the component to be identified is obtained according to the energy value of the component to be identified and the extreme value in the component to be identified, the change trend similarity degree of the original vibration signal and the component to be identified is obtained according to the similarity of the envelope lines between the original vibration signal and the component to be identified, and the non-stationary characteristic value of the component to be identified is obtained according to the amplitude in the component to be identified; the IMF judgment condition of the component to be identified is optimized according to the response degree, the change trend similarity degree and the non-stationary characteristic value, and a corresponding IMF component is obtained; the IMF component is removed from the original vibration signal to obtain a residual signal, and the residual signal is continuously subjected to EMD decomposition until a stop iteration decomposition condition is met, all IMF components of the original vibration signal are obtained, and any bolt at the fan tower drum is subjected to fault monitoring according to all IMF components. Considering that part of the components to be identified containing bolt fault information cannot meet the IMF judgment condition and is removed when the original vibration signal is decomposed by using the EMD algorithm, part of the key information is lost, and the monitoring result of the fan tower drum bolt is affected, the IMF judgment condition in the EMD algorithm is optimized by analyzing the change trend and structure of the component to be identified, a new IMF judgment condition is obtained, the IMF component obtained under the new IMF judgment condition can retain more key information about bolt faults in the original vibration signal, and the accuracy of monitoring the fan tower drum bolt is improved. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0040] Figure 1 It is a flow chart of a fan tower drum bolt intelligent monitoring method based on sensor technology provided by the first embodiment of the present application. DETAILED DESCRIPTION

[0041] Embodiments of the present disclosure are described in detail below with reference to examples shown in the accompanying drawings. The embodiments described below by reference to the drawings are exemplary and are intended to explain the present disclosure, and cannot be understood as a limitation of the present disclosure.

[0042] It should be noted that the terms "first", "second", and the like in the specification of the present disclosure and the above drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Rather, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0043] In order to illustrate the technical solutions of the present application, specific embodiments are described below.

[0044] The specific scenario to which the present application is directed is that: in the process of decomposing the bolt vibration signal using the EMD algorithm, it is necessary to determine the intermediate component IMF, and the two determination conditions of IMF in the traditional way are to ensure that the intermediate component can reflect the smooth and symmetric vibration mode of the original signal, but the actual fan tower bolt vibration signal has irregular frequency components and complex signal characteristics, resulting in that part of the intermediate components containing key information cannot fully meet the determination conditions of IMF when using the EMD algorithm to decompose it, and then when the bolt vibration signal is decomposed, part of the key information is lost, which affects the monitoring result of the fan tower bolt, therefore, the present application optimizes the determination condition of IMF in the EMD algorithm, improves the accuracy of monitoring the fan tower bolt.

[0045] The embodiment of the present application provides a fan tower bolt intelligent monitoring system based on sensor technology, which comprises a processor and a memory, the processor executes the computer program stored in the memory to realize a fan tower bolt intelligent monitoring method based on sensor technology, as shown in the figure, the method comprises the following steps: Figure 1 As shown in the figure, the method comprises the following steps:

[0046] Step S101, obtaining the original vibration signal of any bolt at the fan tower in a preset time period, in the process of EMD decomposition of the original vibration signal, obtaining the envelope line of the original vibration signal, and obtaining the to-be-identified component of the original vibration signal according to the envelope line.

[0047] The sampling frequency is set to 1 time per second, and the vibration signal of any bolt at the fan tower drum within 10 minutes is obtained through the sensor to obtain the original vibration signal Y. The sampling frequency and sampling time can be set according to the specific scene.

[0048] The original vibration signal is decomposed by the EMD algorithm to obtain the to-be-authenticated component ht in the original vibration signal. The general process of the EMD algorithm is as follows: (1) determine all extreme points of the original vibration signal, and use the cubic spline interpolation method to connect all maximum points and all minimum points to obtain the upper envelope line and the lower envelope line of the original vibration signal; (2) calculate the mean value between the upper envelope line and the lower envelope line to obtain the mean envelope line; (3) subtract the mean envelope line from the original vibration signal to obtain the to-be-authenticated component of the original vibration signal; (4) check whether the to-be-authenticated component meets the IMF judgment condition. If yes, the to-be-authenticated signal is taken as an IMF component. If not, a new to-be-authenticated signal is obtained according to the upper envelope line and the lower envelope line of the to-be-authenticated signal, and step (4) is repeated until the IMF component is obtained; (5) remove the IMF component from the original vibration signal to obtain a residual signal, and continue to decompose the residual signal by EMD until the intrinsic mode function of the residual signal is a monotonic function or a constant. The original vibration signal and the to-be-authenticated component are expressed by a time-domain graph, in which the x-axis (abscissa) represents time and the y-axis (ordinate) represents amplitude. The EMD algorithm is prior art, and will not be described here.

[0049] In step S102, the response degree of the original vibration signal in the to-be-authenticated component is obtained according to the energy value of the to-be-authenticated component and the extreme value in the to-be-authenticated component, the change trend similarity degree of the original vibration signal and the to-be-authenticated component is obtained according to the similarity of the envelope line between the original vibration signal and the to-be-authenticated component, and the non-stationary characteristic value of the to-be-authenticated component is obtained according to the amplitude in the to-be-authenticated component.

[0050] The specific conditions for determining the to-be-identified component in the EMD algorithm are: (1) the number of extreme points of the to-be-identified component and the number of zero points must be equal, or at most differ by one; (2) the average of the upper envelope line and the lower envelope line of the to-be-identified component is 0, that is, the upper envelope line and the lower envelope line are symmetrical. The two conditions are to ensure that the waveform of the decomposed signal (to-be-identified component) is symmetrical and neat, so as to reduce unnecessary fluctuations and the interference of asymmetric components. However, the vibration signal of the bolt of the fan tower is often non-stationary, sudden, and has irregular frequency components and complex signal characteristics. This feature causes some to-be-identified components that may contain key information to be excluded because they do not fully meet the IMF determination conditions, resulting in the loss of some key information when the IMF component is finally extracted. Therefore, in the embodiment of the present application, the structural features and change trend of the to-be-identified component are analyzed to reflect the number of key information related to bolt failure in the to-be-identified component.

[0051] Considering that the energy of the IMF component can effectively describe the energy distribution of the signal at different frequency components and can reflect the contribution degree of the IMF component to the original signal, if the energy of the to-be-identified component is higher, it means that the to-be-identified component can better reflect the characteristics in the original signal, that is, the to-be-identified component retains more bolt failure information. At the same time, when the bolt at the fan tower appears abnormal, the original vibration signal shows sudden and violent fluctuations. If the to-be-identified component also shows violent fluctuations, it means that the to-be-identified component retains more key information related to bolt failure in the original vibration signal.

[0052] Therefore, in the embodiment of the present application, the extreme value in the to-be-identified component is used to represent the fluctuation degree of the to-be-identified component, and then the response degree of the original vibration signal in the to-be-identified component is obtained according to the energy of the to-be-identified component and the extreme value in the to-be-identified component, which is used to reflect the number of key information related to bolt failure in the original vibration signal retained in the to-be-identified component. Specifically:

[0053] The sum of squares of all amplitudes in the to-be-identified component is calculated to obtain an energy value of the to-be-identified component, and the energy value is used as an independent variable of a hyperbolic tangent function to obtain a first function value;

[0054] The maximum value and the minimum value in the to-be-identified component are obtained, the absolute value of the difference between the cumulative value of all maximum values and the cumulative value of all minimum values is calculated, and the absolute value of the difference is used as an independent variable of a hyperbolic tangent function to obtain a second function value;

[0055] The average value between the first function value and the second function value is calculated to obtain the response degree of the original vibration signal in the to-be-identified component.

[0056] In an embodiment, the formula for calculating the response degree of the original vibration signal in the component to be identified is:

[0057]

[0058] wherein T(ht) represents the response degree of the original vibration signal in the component to be identified ht, E represents the energy value of the component to be identified, H represents the number of all maxima in the component to be identified, Q represents the number of all minima in the component to be identified, fh represents the hth maxima in the component to be identified, fq represents the qth minima in the component to be identified, and tanh() represents the hyperbolic tangent function. h q wherein tanh() represents the hyperbolic tangent function, and || represents the absolute value symbol.

[0059] It should be noted that the greater E is, the stronger the volatility of the component to be identified is, and the greater T(ht) is, indicating that the component to be identified retains more bolt fault information in the original vibration signal; the greater H is, the more frequent and intense the fluctuation of the signal in the component to be identified is, and the greater T(ht) is, indicating that the component to be identified retains more bolt fault information in the original vibration signal.

[0060] When the bolt at the fan tower fails, due to the complex signal characteristics, if only the component to be identified is analyzed, it is difficult to fully reveal the fault information that the component to be identified may contain. Therefore, based on the retention characteristics of the signal, that is, when the component to be identified and the original vibration signal have similar trends, it is considered that the component to be identified retains the main mode in the original vibration signal, that is, the component to be identified covers more information related to the bolt fault.

[0061] Therefore, in the embodiment of the present application, the DTW distance between the original vibration signal and the component to be identified is used to represent the similarity degree between the original vibration signal and the component to be identified in the overall aspect, wherein the DTW distance is a prior art, which will not be described here; the similarity of the slopes between the envelope lines (including the upper envelope line and the lower envelope line) of the original vibration signal and the envelope lines of the component to be identified is used to represent the similarity degree between the original vibration signal and the component to be identified in the local aspect, and the similarity degree in the overall aspect and the similarity degree in the local aspect are combined to obtain the similarity degree of the change trend between the original vibration signal and the component to be identified, specifically:

[0062] (1) According to the slopes of the envelope lines of the original vibration signal and the envelope lines of the component to be identified, the envelope line slope feature difference value between the envelope lines of the original vibration signal and the envelope lines of the component to be identified is obtained.

[0063] ​Specifically, the envelope of the original vibration signal is denoted as an original envelope, the envelope of the component to be identified is denoted as a component envelope, all extreme points in the original envelope are obtained, the slope between every two adjacent extreme points in the original envelope is calculated to obtain a slope sequence of the original envelope, and a slope sequence of the component envelope is obtained;

[0064] A difference between the maximum value and the minimum value in the slope sequence of the original envelope is calculated to obtain a mutation index of the original envelope, a mutation index of the component envelope is obtained according to the slope sequence of the component envelope, and an absolute value of a difference between the mutation index of the original envelope and the mutation index of the component envelope is calculated to obtain a first slope difference degree between the original envelope and the component envelope;

[0065] An average value of all data in the slope sequence of the original envelope is calculated to obtain a trend feature value of the original envelope, a trend feature value of the component envelope is obtained according to the slope sequence of the component envelope, and an absolute value of a difference between the trend feature value of the original envelope and the trend feature value of the component envelope is calculated to obtain a second slope difference degree between the original envelope and the component envelope;

[0066] The first slope difference degree and the second slope difference degree are added to obtain an envelope slope feature difference value between the original envelope and the component envelope.

[0067] In an embodiment, a calculation formula of the envelope slope feature difference value between the envelope of the original vibration signal and the envelope of the component to be identified is as follows:

[0068]

[0069] wherein ΔK(ht, Y) represents an envelope slope feature difference value between the envelope (original envelope) of the original vibration signal Y and the envelope (component envelope) of the component ht to be identified, k Y,max represents a maximum value in the slope sequence of the original envelope, k Y,min represents a minimum value in the slope sequence of the original envelope, k ht,max represents a maximum value in the slope sequence of the component envelope, k ht,min represents a minimum value in the slope sequence of the component envelope, ∑k Y represents an accumulated value of all data in the slope sequence of the original envelope, ∑k ht represents an accumulated value of all data in the slope sequence of the component envelope, n represents a number of all data in the slope sequence of the original envelope, m represents a number of all data in the slope sequence of the component envelope, and || represents an absolute value symbol.

[0070] It should be noted that |(k Y,max -k Y,min )-(k ht,max -k ht,min )| is the first slope difference. The smaller the first slope difference, the more similar the fluctuation between the original envelope and the component envelope. Thus, the smaller ΔK(ht,Y) is, the more similar the original vibration signal and the component to be identified are in local aspects, and the component to be identified can better reflect the signal changes of the original vibration signal. This is the second slope difference, That is the trend characteristic value of the original envelope, That is, the trend characteristic value of the component envelope. The smaller the second slope difference, the more similar the change trends between the original envelope and the component envelope. Therefore, the smaller ΔK(ht,Y) is, the more similar the original vibration signal and the component to be identified are in local aspects, and the more the component to be identified can reflect the signal changes of the original vibration signal.

[0071] (2) Obtain the DTW distance between the original vibration signal and the component to be identified. Combining the envelope slope feature difference value with the DTW distance, the similarity between the change trends of the original vibration signal and the component to be identified is obtained.

[0072] Specifically, using an exponential function with a natural constant as the base, the envelope slope characteristic difference value is inversely normalized to obtain a first similarity index between the original vibration signal and the component to be identified, and the DTW distance is inversely normalized to obtain a second similarity index between the original vibration signal and the component to be identified;

[0073] An average value between the first similarity index and the second similarity index is calculated to obtain a similarity degree between a change trend of the original vibration signal and the component to be identified.

[0074] In one embodiment, the calculation formula for the similarity between the change trend of the original vibration signal and the component to be identified is:

[0075]

[0076] Among them, R α (ht) represents the similarity between the changing trends of the original vibration signal and the component to be identified ht, DTW(ht,Y) represents the DTW distance between the original vibration signal Y and the component to be identified ht, ΔK(ht,Y) represents the envelope slope characteristic difference between the envelope of the original vibration signal Y (original envelope) and the envelope of the component to be identified ht (component envelope), and exp[-()] represents the exponential function with the natural constant e as the base, which is used for inverse proportional normalization.

[0077] It should be noted that the smaller the DTW(ht, Y) is, the higher the similarity between the original vibration signal and the component to be identified in the overall aspect is, and the larger the exp[-DTW(ht, Y)] is, the larger the R α The larger the ΔK(ht, Y) is, the more the bolt fault information in the original vibration signal retained in the component to be identified is; the smaller the ΔK(ht, Y) is, the more similar the original vibration signal and the component to be identified in the local aspect are, and the larger the exp[-ΔK(ht, Y)] is, the larger the R α The larger the ΔK(ht, Y) is, the more the bolt fault information in the original vibration signal retained in the component to be identified is.

[0078] In addition, considering that when the bolt of the fan tower is in failure, an impact signal or a pulse type fluctuation is often generated, the waveform amplitude in the original vibration signal is high, and the frequency component is irregular, that is, the amplitude distribution in the original vibration signal is relatively complex, and the signal is not stationary and uniform, if the component to be identified of the original vibration signal has the same characteristics, it indicates that the component to be identified can reflect more bolt fault information.

[0079] Therefore, in the embodiment of the application, the non-stationary characteristic value of the component to be identified is obtained through the amplitude in the component to be identified of the original vibration signal, so as to represent the amplitude distribution in the component to be identified, and further reflect the number of key information related to the bolt fault contained in the component to be identified, specifically:

[0080] The maximum amplitude in the component to be identified and the average amplitude of all amplitudes in the component to be identified are obtained, the absolute value of the difference between the maximum amplitude and the average amplitude is normalized by using the hyperbolic tangent function, and the first characteristic value is obtained;

[0081] All extreme points in the component to be identified are obtained, the slope between every two adjacent extreme points in the component to be identified is calculated, and the variance of all slopes is normalized by using the hyperbolic tangent function, and the second characteristic value is obtained;

[0082] The average value between the first characteristic value and the second characteristic value is calculated, and the non-stationary characteristic value of the component to be identified is obtained.

[0083] In an embodiment, the calculation formula of the non-stationary characteristic value of the component to be identified is:

[0084]

[0085] Wherein, R β (ht) represents the non-stationary characteristic value of the component to be identified ht, f maxrepresents the maximum amplitude in the component to be identified, ∑f represents the cumulative value of all amplitudes in the component to be identified, u represents the number of all amplitudes in the component to be identified, S 2 represents the variance between every two adjacent extreme points in the component to be identified, tanh() represents a hyperbolic tangent function, is used for normalization of the result, and || represents an absolute value symbol.

[0086] It should be noted that the average amplitude of all amplitudes in the component to be identified is used to represent the overall signal fluctuation in the component to be identified, and therefore, by taking the average amplitude as a reference benchmark, the deviation degree of the signal in the component to be identified is represented by the absolute value of the difference between the maximum amplitude in the component to be identified and the average amplitude, that is, The greater the R, the greater the deviation degree of the signal in the component to be identified, and the more unstable the signal, and further, the greater the R β (ht), the more bolt fault information contained in the component to be identified. 2 The greater the R, the greater the discrete degree of the signal in the component to be identified, the more complex the amplitude distribution, and the fewer the number of signals with stable change trend in the component to be identified, and further, the greater the R β (ht), the more bolt fault information contained in the component to be identified.

[0087] At this point, through the analysis of the component to be identified of the original vibration signal, the response degree of the original vibration signal in the component to be identified, the similarity degree of the change trend of the original vibration signal and the component to be identified, and the non-stationary characteristic value of the component to be identified are obtained.

[0088] In step S103, the IMF judgment condition of the component to be identified is optimized according to the response degree, the similarity degree of the change trend, and the non-stationary characteristic value, and the corresponding IMF component is obtained.

[0089] In the EMD algorithm, in order to ensure that the waveform of the decomposed signal is symmetrical and neat, and reduce unnecessary fluctuations and the interference of asymmetric components, if a signal is determined as an IMF component, the signal must satisfy two conditions: (1) the number of extreme points and the number of zero points of the signal must be equal, or at most differ by one; (2) the average of the upper envelope line and the lower envelope line of the signal is 0, that is, the upper envelope line and the lower envelope line are symmetrical. Since the actual fan tower bolt vibration signal is often non-stationary, burst, and has irregular frequency components and complex signal characteristics, when the EMD algorithm is used for decomposition, part of the intermediate components that may contain key information cannot fully satisfy the IMF determination condition and is excluded, and then when the bolt vibration signal is decomposed, part of the key information is lost, which affects the monitoring result of the fan tower bolt. Therefore, in the embodiment of the present application, according to the response degree of the original vibration signal in the to-be-identified component obtained in step S102, the similarity degree of the change trend of the original vibration signal and the to-be-identified component, and the non-stationary characteristic value of the to-be-identified component, the IMF determination condition in the EMD algorithm is optimized for IMF determination of the to-be-identified component of the original vibration signal, and the corresponding IMF component is obtained.

[0090] Firstly, according to the IMF determination condition of the to-be-identified component in the EMD algorithm, an initial condition determination value of the to-be-identified component is obtained, specifically:

[0091] In the to-be-identified component, the number of all extreme points and the number of all zero points are obtained, the absolute value of the difference between the number of all extreme points and the number of zero points is inversely proportional normalized by using an exponential function with a natural constant as a base, and a first initial condition value is obtained;

[0092] The intrinsic mode function of the to-be-identified component is obtained, the result of the integral of the intrinsic mode function is inversely proportional normalized by using an exponential function with a natural constant as a base, and a second initial condition value is obtained, where the intrinsic mode function is a prior art and will not be described here;

[0093] The average value between the first initial condition value and the second initial condition value is calculated, and the initial condition determination value of the to-be-identified component is obtained.

[0094] In an embodiment, the calculation formula of the initial condition determination value of the to-be-identified component is:

[0095]

[0096] Where δ represents the initial condition determination value of the to-be-identified component, P represents the number of all extreme points in the to-be-identified component, Z represents the number of all zero points in the to-be-identified component, h(t) represents the intrinsic mode function of the to-be-identified component, Integral of intrinsic mode function representing the component to be identified, x represents the time corresponding to the last waveform in the component to be identified, || represents the absolute value symbol, exp[-()] represents the exponential function with the natural constant e as the base, and is used for inverse proportional normalization.

[0097] It should be noted that exp[-|P-Z|] is the first initial condition value, and the smaller |P-Z| is, the more equal the number of all extreme points in the component to be identified and the number of all zero points are, and the larger the first initial condition value is, the larger δ is, and the more consistent the component to be identified is with the IMF determination condition in the EMD algorithm; the IMF determination condition in the EMD algorithm requires that the mean of the upper envelope line and the lower envelope line of the component to be identified is 0, that is, the component to be identified is symmetric about the x axis in the time domain graph, that is, the integral of the intrinsic mode function of the component to be identified is 0, so the more symmetric the upper envelope line and the lower envelope line of the component to be identified are, and the larger the larger δ is, and the more consistent the component to be identified is with the IMF determination condition in the EMD algorithm.

[0098] Then, the average value between the response degree, the change trend similarity degree, and the non-stationary characteristic value obtained in the calculation step S102 is calculated to obtain the structural characteristic value of the component to be identified, the initial condition determination value and the structural characteristic value are weighted and summed to obtain the final condition determination value of the component to be identified, so as to optimize the IMF determination condition in the EMD algorithm and determine the first IMF component in the original vibration signal. The calculation formula of the final condition determination value of the component to be identified is:

[0099]

[0100] Wherein, τ represents the final condition determination value of the component to be identified, δ represents the initial condition determination value of the component to be identified, T(ht) represents the response degree of the original vibration signal in the component to be identified, R α (ht) represents the change trend similarity degree between the original vibration signal and the component to be identified, R β (ht) represents the non-stationary characteristic value of the component to be identified, μ1 represents the first weight, and μ2 represents the second weight.

[0101] It should be noted that μ1=0.7 and μ2=0.3 are set, which is not limited here, and the implementer can set according to the specific scene. That is, the structural characteristic value of the component to be identified. The larger δ is, the more consistent the component to be identified is with the IMF determination condition in the EMD algorithm, the larger the structural characteristic value is, the more the number of the bolt fault information retained in the component to be identified is, and the larger τ is, which indicates that the component to be identified is not only more consistent with the IMF determination condition in the EMD algorithm, but also contains more information of the bolt fault.

[0102] The preset IMF condition judgment value range is set to (0.7, 1), which is not limited here, and the implementer can set it according to the specific scene. If the final condition judgment value τ of the to-be-judged component is in (0.7, 1), it indicates that the to-be-judged component of the original vibration signal is consistent with the IMF judgment condition in the EMD algorithm, and the to-be-judged component contains more bolt fault information, so the to-be-judged component of the original vibration signal is determined to be the first IMF component of the original vibration signal.

[0103] In step S104, the IMF component is removed from the original vibration signal to obtain a residual signal, and the EMD decomposition is continued on the residual signal until the stop iteration decomposition condition is met to obtain all IMF components of the original vibration signal, and any bolt at the fan tower is monitored for fault according to all IMF components.

[0104] After obtaining the first IMF component in the original vibration signal, the first IMF component is removed from the original vibration signal to obtain a residual signal. The residual signal is decomposed by EMD to obtain a new to-be-judged component of the original vibration signal, and the final condition judgment value of the new to-be-judged component is obtained according to step S103. If the final condition judgment value of the new to-be-judged component is in (0.7, 1), the new to-be-judged component of the original vibration signal is determined to be the second IMF component of the original vibration signal, and the second IMF component is removed from the original vibration signal to obtain a new residual signal. If the final condition judgment value of the new to-be-judged component is not in (0.7, 1), the new to-be-judged component is decomposed by EMD until the second IMF component of the original vibration signal is obtained, and the second IMF component is removed from the original vibration signal to obtain a new residual signal. By analogy, until the new residual signal meets the stop iteration condition in the EMD algorithm, all IMF components in the original vibration signal are obtained.

[0105] All IMF components in the original vibration signal of any bolt at the fan tower are extracted for features such as frequency domain features and time domain features, and the extracted features are input into a decision tree model to obtain a fault mode of any bolt at the fan tower. Here, the implementer can set a machine learning model according to the specific scene, match the obtained fault mode with the existing fault mode in the fault mode library, and determine whether any bolt at the fan tower has loosening, damage or other faults. If it is determined that any bolt at the fan tower has a fault, an alarm is sent to remind the relevant staff. The feature extraction, decision tree model and fault mode library are prior art, and will not be described here.

[0106] It is worth mentioning that the focus of the present application is to optimize the IMF determination condition in the EMD algorithm to obtain more IMF components containing bolt fault information, thereby improving the accuracy of monitoring the fan tower bolt.

[0107] In summary, the present application obtains the original vibration signal of any bolt at the fan tower within a preset time period, obtains the envelope line of the original vibration signal in the process of EMD decomposition of the original vibration signal, and obtains the to-be-authenticated component of the original vibration signal according to the envelope line; obtains the response degree of the original vibration signal in the to-be-authenticated component according to the energy value of the to-be-authenticated component and the extreme value in the to-be-authenticated component, obtains the change trend similarity degree of the original vibration signal and the to-be-authenticated component according to the similarity of the envelope lines between the original vibration signal and the to-be-authenticated component, and obtains the non-stationary characteristic value of the to-be-authenticated component according to the amplitude in the to-be-authenticated component; the IMF determination condition of the to-be-authenticated component is optimized according to the response degree, the change trend similarity degree, and the non-stationary characteristic value, to obtain the corresponding IMF component; the IMF component is removed from the original vibration signal to obtain a residual signal, and the residual signal is continuously decomposed by EMD until the stop iteration decomposition condition is met, to obtain all IMF components of the original vibration signal, and any bolt at the fan tower is monitored for fault according to all IMF components. Among them, considering that part of the to-be-authenticated components containing bolt fault information is excluded when the original vibration signal is decomposed by using the EMD algorithm, resulting in the loss of part of the key information and affecting the monitoring result of the fan tower bolt, the change trend and structure of the to-be-authenticated component are analyzed, the IMF determination condition in the EMD algorithm is optimized, a new IMF determination condition is obtained, the IMF component obtained under the new IMF determination condition can retain more key information about bolt fault in the original vibration signal, and the accuracy of monitoring the fan tower bolt is improved.

[0108] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit it; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. The intelligent monitoring system for wind turbine tower bolts based on sensor technology is characterized by: The system comprises a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the following steps when executing the computer program: Obtaining an original vibration signal of any bolt at a wind turbine tower within a preset time period, obtaining an envelope of the original vibration signal during EMD decomposition of the original vibration signal, and obtaining a component to be identified of the original vibration signal based on the envelope; Obtaining the degree of response of the original vibration signal in the component to be identified based on the energy value of the component to be identified and the extreme value in the component to be identified; obtaining the degree of similarity of the changing trends of the original vibration signal and the component to be identified based on the similarity of the envelopes between the original vibration signal and the component to be identified; and obtaining the non-stationary eigenvalue of the component to be identified based on the amplitude value in the component to be identified; Optimizing the IMF determination condition of the component to be identified according to the response degree, the similarity degree of the change trend, and the non-stationary eigenvalue to obtain the corresponding IMF component; The IMF component is eliminated from the original vibration signal to obtain a residual signal, and the residual signal is continuously subjected to EMD decomposition until the condition for stopping iterative decomposition is met, thereby obtaining all IMF components of the original vibration signal. Fault monitoring is then performed on any bolt at the wind turbine tower based on all IMF components.

2. The intelligent monitoring system for wind turbine tower bolts based on sensor technology according to claim 1 is characterized in that: Obtaining the response degree of the original vibration signal in the component to be identified based on the energy value of the component to be identified and the extreme value in the component to be identified includes: Calculating the sum of squares of all amplitudes of the component to be identified to obtain an energy value of the component to be identified, and using the energy value as an independent variable of a hyperbolic tangent function to obtain a first function value; Obtaining the maximum and minimum values ​​in the component to be identified, calculating the absolute value of the difference between the accumulated value of all the maximum values ​​and the accumulated value of all the minimum values, and using the absolute value of the difference as the independent variable of the hyperbolic tangent function to obtain a second function value; An average value between the first function value and the second function value is calculated to obtain the response degree of the original vibration signal in the component to be identified.

3. The intelligent monitoring system for wind turbine tower bolts based on sensor technology according to claim 1 is characterized in that: The obtaining, based on the similarity of the envelopes between the original vibration signal and the component to be identified, a degree of similarity between the changing trends of the original vibration signal and the component to be identified, includes: Recording the envelope of the original vibration signal as the original envelope, recording the envelope of the component to be identified as the component envelope, obtaining all extreme value points in the original envelope, calculating the slope between every two adjacent extreme value points in the original envelope to obtain a slope sequence of the original envelope, obtaining a slope sequence of the component envelope, obtaining an envelope slope characteristic difference value between the original envelope and the component envelope based on a difference in the slope sequence between the original envelope and the component envelope, and inversely normalizing the envelope slope characteristic difference value using an exponential function with a natural constant as a base to obtain a first similarity index between the original vibration signal and the component to be identified; Obtaining a DTW distance between the original vibration signal and the component to be identified, and inversely normalizing the DTW distance using an exponential function with a natural constant as a base to obtain a second similarity index between the original vibration signal and the component to be identified; An average value between the first similarity index and the second similarity index is calculated to obtain a similarity degree between a change trend of the original vibration signal and the component to be identified.

4. The intelligent monitoring system for wind turbine tower bolts based on sensor technology according to claim 3 is characterized in that: The step of obtaining an envelope slope characteristic difference value between the original envelope and the component envelope according to a difference in slope sequence between the original envelope and the component envelope includes: Calculating the difference between the maximum and minimum values ​​in the slope sequence of the original envelope to obtain a mutation index of the original envelope; obtaining the mutation index of the component envelope according to the slope sequence of the component envelope; calculating the absolute value of the difference between the mutation index of the original envelope and the mutation index of the component envelope to obtain a first slope difference between the original envelope and the component envelope; Calculating the average value of all data in the slope sequence of the original envelope to obtain a trend characteristic value of the original envelope, obtaining the trend characteristic value of the component envelope according to the slope sequence of the component envelope, and calculating the absolute value of the difference between the trend characteristic value of the original envelope and the trend characteristic value of the component envelope to obtain a second slope difference between the original envelope and the component envelope; The first slope difference and the second slope difference are added together to obtain an envelope slope characteristic difference value between the original envelope and the component envelope.

5. The intelligent monitoring system for wind turbine tower bolts based on sensor technology according to claim 1 is characterized in that: Obtaining the non-stationary eigenvalue of the component to be identified according to the amplitude of the component to be identified includes: Obtaining a maximum amplitude in the component to be identified and an average amplitude of all amplitudes in the component to be identified, and normalizing the absolute value of the difference between the maximum amplitude and the average amplitude using a hyperbolic tangent function to obtain a first eigenvalue; Obtaining all extreme points in the component to be identified, calculating the slope between every two adjacent extreme points in the component to be identified, and normalizing the variance of all slopes using a hyperbolic tangent function to obtain a second eigenvalue; An average value between the first eigenvalue and the second eigenvalue is calculated to obtain a non-stationary eigenvalue of the component to be identified.

6. The intelligent monitoring system for wind turbine tower bolts based on sensor technology according to claim 1 is characterized in that: The optimizing the IMF determination condition of the component to be identified according to the response degree, the similarity degree of the change trend, and the non-stationary eigenvalue to obtain the corresponding IMF component includes: Calculating the response degree, the similarity degree of the change trend, and the average value of the non-stationary eigenvalue to obtain the structural eigenvalue of the component to be identified; According to the IMF judgment condition in the EMD decomposition, an initial condition judgment value of the component to be identified is obtained, a weighted sum is performed on the initial condition judgment value and the structural eigenvalue to obtain a final condition judgment value of the component to be identified, and the IMF component of the original vibration signal is determined according to the final condition judgment value.

7. The intelligent monitoring system for wind turbine tower bolts based on sensor technology according to claim 6 is characterized in that: The step of obtaining the initial condition determination value of the component to be identified according to the IMF determination condition in the EMD decomposition includes: In the component to be identified, the number of all extreme points and the number of all zero points are obtained, and the absolute values ​​of the differences between the number of all extreme points and the number of zero points are inversely normalized using an exponential function with a natural constant as the base to obtain a first initial condition value; Obtaining an intrinsic mode function of the component to be identified, integrating the intrinsic mode function, and inversely normalizing a result of the integration using an exponential function with a natural constant as a base to obtain a second initial condition value; An average value between the first initial condition value and the second initial condition value is calculated to obtain an initial condition determination value of the component to be identified.

8. The intelligent monitoring system for wind turbine tower bolts based on sensor technology according to claim 6 is characterized in that: The determining, according to the final condition determination value, the IMF component of the original vibration signal includes: If the final condition determination value is within a preset IMF condition determination value range, the component to be identified is determined to be the IMF component of the original vibration signal.

Images