Fault comprehensive diagnosis system and method suitable for aviation aerostatic press

Abstract

The invention relates to the technical field of aviation aerostatics, in particular to a fault comprehensive diagnosis system and method suitable for an aviation aerostatics, and the system comprises a data collection module, a data preprocessing module, a feature extraction module, a fault diagnosis module, and a result display and alarm module. Through cooperation of the data acquisition module, the data preprocessing module, the feature extraction module, the fault diagnosis module and the result display and alarm module, the data acquisition module provides basic data support, the data preprocessing module improves the data quality and accuracy, and the feature extraction module extracts useful feature information. The fault diagnosis module accurately judges the type of a fault, the result display and alarm module ensures that related personnel can timely obtain fault information and take corresponding measures, the whole system forms a closed-loop monitoring and diagnosis process, the data processing rate of the aviation aerostatic press can be effectively increased, and the data processing accuracy is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of aviation air compressors, and in particular relates to a comprehensive fault diagnosis system and method suitable for aviation air compressors. Background Art

[0002] As a key equipment on the aircraft, the stable operation of the air compressor is crucial to flight safety. If the air compressor fails, it may affect the normal air supply, braking system, tire inflation and other key functions of the aircraft, thereby threatening the safe flight of the aircraft. Therefore, in order to ensure the safe flight of the aircraft, the aircraft air compressor needs to be diagnosed and repaired in time.

[0003] At present, the fault diagnosis method for aviation air compressors is usually to use professional testing instruments, such as vibration analyzers, infrared thermal imagers, etc., to conduct more in-depth testing of the air compressor to determine the specific location and cause of the fault;

[0004] However, when confirming the specific location and cause of the fault, it is necessary to collect the operating data of the air compressor, such as exhaust temperature, pressure change, fuel consumption, etc., and analyze the operating data; however, since there is a lot of operating data, a large amount of data may lead to slow data processing rate and is easily interfered by external signals, affecting the accuracy of data processing. Therefore, we need to propose a comprehensive fault diagnosis system and method suitable for aviation air compressors to solve the above-mentioned problems, so that it can improve the data processing rate while reducing the impact of external signals on data processing and improving the accuracy of data processing. Summary of the invention

[0005] In view of the above problems, the present invention provides a comprehensive fault diagnosis system applicable to an aviation air compressor, comprising: a data acquisition module, the data acquisition module is used to collect multi-dimensional operation data of exhaust temperature, pressure change and fuel consumption during the operation of the aviation air compressor in real time, so as to ensure that various operation parameters of the aviation air compressor can be obtained in real time and accurately;

[0006] A data preprocessing module, which is used to clean, normalize and filter the collected data to improve the quality and accuracy of the data;

[0007] A feature extraction module, which is used to extract time domain features and frequency domain features from the preprocessed data to extract feature information useful for fault diagnosis from the original data;

[0008] A fault diagnosis module, wherein the fault diagnosis module is used to establish a fault diagnosis model and use the fault diagnosis model to accurately determine the fault type of the aviation air compressor according to the extracted time domain features and frequency domain features;

[0009] A result display and alarm module, which intuitively displays the information output by the fault diagnosis module on a monitoring display screen or a terminal device, and sets different levels of alarm thresholds according to the severity of the fault. When the diagnosis result reaches the corresponding alarm condition, the corresponding alarm mechanism is activated to ensure that relevant personnel can timely understand the operating status and fault condition of the aviation air compressor, and take corresponding countermeasures according to the severity of the fault to ensure the safe flight of the aircraft;

[0010] The data acquisition module is electrically connected to the data preprocessing module, the data preprocessing module is electrically connected to the feature extraction module, the feature extraction module is electrically connected to the fault diagnosis module, and the fault diagnosis module is electrically connected to the result display and alarm module.

[0011] Furthermore, the data acquisition module includes a temperature acquisition unit for collecting the exhaust temperature during the operation of the aviation compressor, a pressure acquisition unit for collecting the pressure change during the operation of the aviation compressor, and a fuel consumption acquisition unit for collecting the fuel consumption during the operation of the aviation compressor. The temperature acquisition unit, the pressure acquisition unit and the fuel consumption acquisition unit are all electrically connected to the data preprocessing module.

[0012] Furthermore, the process of the fuel consumption collection unit for collecting fuel consumption is as follows:

[0013] A31. Install the turbine flow sensor in the fuel supply pipeline of the aviation air compressor to ensure that the fuel can pass through the turbine flow sensor smoothly without leakage;

[0014] A32. Set the sampling frequency of the turbine flow sensor. The turbine flow sensor collects the rotation frequency of the turbine in real time, and then calculates the flow rate based on the rotation frequency. The calculation formula is as follows:

[0015] Among them, f is the rotation frequency of the turbine, K is the instrument factor of the turbine flow sensor, and Q is the fuel flow rate.

[0016] Furthermore, the data preprocessing module performs cleaning processing on the collected data as follows:

[0017] B11. According to the technical standards for the normal operation of the air compressor and past operating experience, determine the normal range of exhaust temperature, pressure and fuel consumption parameters. The normal range of exhaust temperature is 50℃-150℃, the normal range of intake pressure is 3-6 atmospheres, and the normal range of fuel consumption is 0.1-0.5 liters per minute;

[0018] B12. Classify and archive the exhaust temperature, pressure and fuel consumption, and then traverse the data sequence of each parameter collected according to the normal value range, and judge each data point to confirm whether there is data beyond the normal value range. If there is data beyond the normal value range, it is judged as abnormal data and directly eliminated. If it is within the normal value range, it is retained as normal data;

[0019] B13. Sort the normal data, and then compare the adjacent data points in turn. If the values ​​of two adjacent data points in all parameters are exactly the same, the following data points are considered duplicate data, the first data point is retained, and the duplicate data points are deleted. If the values ​​of two adjacent data points in all parameters are not the same, it is determined that there is no duplicate data and it is retained.

[0020] Furthermore, the process of the data preprocessing module normalizing the cleaned data is as follows:

[0021] B21. Calculate the average value of the data set using the mean calculation formula based on the cleaned data. The mean calculation formula is as follows:

[0022] Among them, μ is the mean value of the data set, x i is the cleaned data set, n is the number of data in the data set;

[0023] B22. Calculate the standard deviation of the data based on the calculated mean value. The standard deviation calculation formula is as follows:

[0024] Among them, μ is the mean value of the data set, x i is the cleaned data set, σ is the standard deviation of the cleaned data;

[0025] B23. Normalize the original data in each data set. The normalization formula is as follows:

[0026] Among them, x is the original data in the data set, and z is the normalized data value.

[0027] Furthermore, the process of filtering the normalized data by the data preprocessing module is as follows:

[0028] B31. According to the characteristics of the normalized data and the filtering target, select the Haar wavelet or Daubechies wavelet filter and determine the number of wavelet decomposition layers;

[0029] B32. Perform wavelet decomposition on the normalized data sequence to obtain the approximate coefficients and detail coefficients of each layer. The calculation formula is as follows:

[0030] A j (k)=∑n h(nk)*x(n),

[0031] D j (k)=∑ n g(nk)*x(n),

[0032] Among them, A j (k) is the approximate coefficient, x(n) is the normalized data sequence, n is the data point number, k is the coefficient number, h(n) is the low-pass filter coefficient, g(n) is the ordinary filter coefficient, D j (k) is the detail coefficient;

[0033] B33, the threshold M is determined for each layer detail coefficient by the minimum maximum threshold method, and the detail coefficient D of each layer is j (k) Perform threshold processing and calculate the detail coefficient after threshold processing Detail coefficient after threshold processing The calculation formula is as follows:

[0034]

[0035] B34. The approximate coefficients and details after threshold processing are reconstructed by inverse discrete wavelet transform to obtain the filtered normalized data sequence y(n). The calculation formula of wavelet reconstruction is as follows:

[0036] Among them, h(n) is the low-pass filter coefficient, g(n) is the normal filter coefficient;

[0037] B35. Set the smoothing window size L to smooth the filtered data. The smoothing formula is as follows:

[0038] Among them, y(i) is the average value of the data points within the range of window size L with the current data point i as the center as the smoothed value;

[0039] B36. Using the time axis of the data sequence collected by the temperature sensor as a reference, the pressure data and fuel consumption data are matched with the reference time axis through the timestamp information, so that each data can correspond one to one in time.

[0040] Furthermore, the process of the feature extraction module for extracting time domain features is as follows:

[0041] C11. Calculate the average value of the data sequence based on the preprocessed data sequence The calculation formula is as follows:

[0042] Where N is the total number of preprocessed data, and y(i) is the preprocessed data;

[0043] C12, based on the average Calculate its variance δ 2 , to confirm the fluctuation of time domain characteristics, the variance calculation formula is as follows:

[0044] Where N is the total number of data after preprocessing;

[0045] The steps of frequency domain feature extraction by the feature extraction module are as follows:

[0046] C21, performing a reverse order operation on the index of the preprocessed data sequence;

[0047] C22, use the fixed rotation factor to perform m-level operation iterations to obtain the transformed value x(a), (a=0,1,…,H-1), the iterative operation formula is as follows:

[0048]

[0049] H=2 m , where x(a) is the value of the discrete result of the discrete sequence y(i) at frequency point i, is the rotation factor, is the frequency interval between adjacent frequency discrete points, a and i are frequency index and time index respectively, H is the length of discrete time series y(i), y(i) is the discrete time series data after preprocessing, j is the imaginary unit, and j satisfies 2 = -1;

[0050] C23. Calculate the spectrum amplitude |x(a)| according to the transformed value x(a). The calculation formula is as follows:

[0051] Where Re[x(a)] is the real part of x(a), Im[x(a)] is the imaginary part of x(a);

[0052] C24. After obtaining the spectrum amplitude of each frequency component, find the a value corresponding to the frequency with the largest spectrum amplitude by comparing the spectrum amplitudes, and then calculate the frequency f corresponding to the a value. a , the calculation formula is as follows:

[0053] Among them, f s is the sampling frequency of the preprocessed data, a is the frequency index, and H is the length of the discrete time series y(i).

[0054] Furthermore, the steps of establishing a fault diagnosis model in the fault diagnosis module are as follows:

[0055] D11. Collect a large amount of historical operation data of aviation compressors with marked fault types, so that these historical operation data contain time domain features and frequency domain features. The fault type is marked with y i Indicates that y i = +1 or -1;

[0056] D12. According to the data distribution characteristics and the complexity of fault classification, a prediction function is established. The prediction function is:

[0057] Among them, K(x' i ,x j ') is the predicted value, x' i and x j 'are two input vectors of historical operation data with marked fault types, is the bandwidth parameter of the prediction function;

[0058] D13. Use the support vector machine training algorithm to solve the optimization formula to determine the support vector machine multiplier β i , the optimization formula is as follows:

[0059]

[0060] The constraints are:

[0061] Among them, β i and β j are the multipliers of the two support vector machines, y i and j are the labels corresponding to the fault types after training, τ is the penalty parameter, and n is the number of training samples;

[0062] D14, will satisfy 0≤β i The sample points corresponding to the training samples with ≤τ are used as support vectors (x s ,y s ), use the support vector to calculate the bias term b, the calculation formula is as follows:

[0063] b=y s -Σ i∈SV β i y i K(x' i ,x s ), where SV is the set of all support vectors, x s is the eigenvector corresponding to the support vector, y s is the category label of the feature vector corresponding to the support vector;

[0064] The process of the fault diagnosis module for determining the fault type of the aviation air compressor is as follows:

[0065] D21, combining the extracted time domain features and frequency domain features into a feature vector x;

[0066] D22. Calculate the feature vector x using the decision function to confirm the positive and negative values ​​of the calculation results. The calculation formula is as follows:

[0067] Among them, K(x' i ,x) is the calculation result of the prediction function, f(x) is the calculation value of the decision function, b is the bias term of the prediction function, β i is the multiplier of the support vector machine, y i is the fault category label corresponding to the training sample;

[0068] D23. When f(x)>0, it is judged to correspond to y i = +1 indicates the fault type. When f(x) is less than 0, it is judged to correspond to y i = -1 indicates the fault type.

[0069] Furthermore, the operation flow of the result display and alarm module for displaying results and alarming is as follows:

[0070] E1. According to the impact of the failure of the air compressor on flight safety and the urgency of maintenance, different levels of alarm thresholds are set in advance for different types of failures;

[0071] E2. Generate data results based on the diagnosis results of the fault diagnosis module. The data results include key information such as fault type and fault probability. The data results are then transmitted to a display device for visual display. The display device includes a maintenance personnel terminal device and a cockpit monitoring display screen:

[0072] E3. Perform logical judgment on the fault type and the corresponding fault probability through the display device to confirm whether the displayed fault type and the corresponding fault probability reach the alarm threshold condition. There are two situations:

[0073] E31. When the displayed fault type and the corresponding fault probability reach the alarm threshold condition, the corresponding level of alarm action is performed;

[0074] E32. When the displayed fault type and the corresponding fault probability do not reach the alarm threshold condition, continue with step E2.

[0075] Based on the above-described comprehensive fault diagnosis system applicable to an aviation air compressor, the present invention further provides a comprehensive fault diagnosis method applicable to an aviation air compressor, comprising the following steps:

[0076] S1. Collect multi-dimensional operation data of exhaust temperature, pressure change and fuel consumption of the aviation air compressor in real time through the data acquisition module;

[0077] S2, cleaning, normalizing and filtering the collected data through a data preprocessing module;

[0078] S3, extracting time domain features and frequency domain features from the preprocessed data through a feature extraction module;

[0079] S4. Construct a fault diagnosis model and perform fault diagnosis based on the extracted time domain features and frequency domain features. If a fault exists, different levels of warnings are issued according to the severity of the fault, and the fault information is visualized. If no fault exists, continue to perform fault diagnosis.

[0080] The beneficial effects of the present invention are:

[0081] 1. The present invention cooperates with a data acquisition module, a data preprocessing module, a feature extraction module, a fault diagnosis module and a result display and alarm module. The data acquisition module provides basic data support, the data preprocessing module improves data quality and accuracy, the feature extraction module extracts useful feature information, the fault diagnosis module accurately determines the fault type, and the result display and alarm module ensures that relevant personnel can obtain fault information in time and take corresponding countermeasures. The entire system forms a closed-loop monitoring and diagnosis process, which can effectively improve the rate of data processing on aviation air compressors, reduce the impact of external signals on data processing, improve the accuracy of data processing, and ensure the safe flight of aircraft.

[0082] 2. The present invention cleans, normalizes and filters the collected data through the cooperation of the data preprocessing module and the feature extraction module to improve the quality and accuracy of the data. Finally, the time domain features and frequency domain features of the processed data are extracted to accurately extract feature information useful for fault diagnosis from the original data and reduce the interference of external signals on the original data.

[0083] 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 pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0085] Figure 1 A block diagram of a comprehensive fault diagnosis system according to an embodiment of the present invention is shown;

[0086] Figure 2 A flowchart showing the cleaning process of the collected data by the data preprocessing module according to an embodiment of the present invention is shown;

[0087] Figure 3 A flowchart showing a data preprocessing module filtering normalized data according to an embodiment of the present invention;

[0088] Figure 4 It shows a flowchart of the operation of the result display and alarm module according to an embodiment of the present invention;

[0089] Figure 5 A flowchart of a comprehensive fault diagnosis method according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0090] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0091] The embodiment of the present invention provides a comprehensive fault diagnosis system applicable to an aviation air compressor, such as Figure 1-4 As shown, it includes a data acquisition module, a data preprocessing module, a feature extraction module, a fault diagnosis module and a result display and alarm module; the data acquisition module is electrically connected to the data preprocessing module, the data preprocessing module is electrically connected to the feature extraction module, the feature extraction module is electrically connected to the fault diagnosis module, and the fault diagnosis module is electrically connected to the result display and alarm module;

[0092] The data acquisition module is used to collect multi-dimensional operating data of exhaust temperature, pressure change and fuel consumption during the operation of the aviation air compressor in real time, ensuring that various operating parameters of the aviation air compressor can be obtained in real time and accurately, providing basic data support for subsequent data analysis and fault diagnosis;

[0093] The data acquisition module includes a temperature acquisition unit for acquiring the exhaust temperature during the operation of the aviation air compressor, a pressure acquisition unit for acquiring the pressure change during the operation of the aviation air compressor, and a fuel consumption acquisition unit for acquiring the fuel consumption during the operation of the aviation air compressor. The temperature acquisition unit, the pressure acquisition unit, and the fuel consumption acquisition unit are all electrically connected to the data preprocessing module;

[0094] The process of exhaust temperature collection by the temperature collection unit is as follows:

[0095] A11. Install the thermocouple temperature sensor near the exhaust port of the aviation air compressor, ensuring that the thermocouple temperature sensor probe is in full contact with the high-temperature gas and is securely installed;

[0096] A12. Set the sampling frequency of the thermocouple temperature sensor, and collect the resistance value corresponding to the exhaust temperature of the aviation air compressor in real time through the thermocouple temperature sensor;

[0097] A13. Calculate the actual exhaust temperature based on the resistance value collected by the thermocouple temperature sensor. The calculation formula is as follows:

[0098] Where, T is the actual measured resistance value, R t is the resistance of the thermal resistor at temperature t, R0 is the resistance of the thermocouple at reference temperature 0°C, and α is the temperature coefficient of the thermocouple;

[0099] The process of the pressure acquisition unit for collecting pressure changes is as follows:

[0100] A21. Use strain gauge pressure sensors to be installed at the intake and exhaust pipes of the aircraft compressor to ensure that the sensing surface of the strain gauge pressure sensor can accurately sense pressure changes and that the installation direction is correct to avoid pressure measurement deviations caused by improper installation of the strain gauge pressure sensor.

[0101] A22. Set the sampling frequency of the strain gauge pressure sensor, and collect the resistance change rate in real time through the strain gauge pressure sensor, and calculate the actual pressure value through the resistance change rate. The calculation formula is as follows;

[0102] Where, E is the elastic modulus, ΔR is the change in strain gauge resistance, R is the initial resistance of the strain gauge, and K is the sensitivity coefficient of the strain gauge;

[0103] The process of the fuel consumption collection unit for collecting fuel consumption is as follows:

[0104] A31. Install the turbine flow sensor in the fuel supply pipeline of the aviation air compressor to ensure that the fuel can pass through the turbine flow sensor smoothly without leakage, and ensure that the turbine flow sensor can accurately measure the fuel flow;

[0105] A32. Set the sampling frequency of the turbine flow sensor. The turbine flow sensor collects the rotation frequency of the turbine in real time, and then calculates the flow rate based on the rotation frequency. The calculation formula is as follows:

[0106] Where, f is the rotation frequency of the turbine, K is the instrument factor of the turbine flow sensor, and Q is the fuel flow rate;

[0107] Through the cooperation of the temperature collection unit, the pressure collection unit and the fuel consumption collection unit, the multi-dimensional operation data of the exhaust temperature, pressure change and fuel consumption during the operation of the aviation compressor can be accurately collected, thereby improving the real-time and accuracy of data collection.

[0108] The data preprocessing module is used to clean, normalize and filter the collected data to improve the quality and accuracy of the data and provide a reliable data basis for the subsequent feature extraction module and fault diagnosis module;

[0109] like Figure 2 As shown, the process of the data preprocessing module cleaning the collected data is as follows:

[0110] B11. According to the technical standards for the normal operation of the air compressor and past operating experience, determine the normal range of exhaust temperature, pressure and fuel consumption parameters. The normal range of exhaust temperature is 50℃-150℃, the normal range of intake pressure is 3-6 atmospheres, and the normal range of fuel consumption is 0.1-0.5 liters per minute;

[0111] B12. Classify and archive the exhaust temperature, pressure and fuel consumption, and then traverse the data sequence of each parameter collected according to the normal value range, and judge each data point to confirm whether there is data beyond the normal value range. If there is data beyond the normal value range, it is judged as abnormal data and directly eliminated. If it is within the normal value range, it is retained as normal data;

[0112] B13. Sort the normal data, and then compare the adjacent data points in turn. If the values ​​of two adjacent data points in all parameters are exactly the same, the following data points are considered duplicate data, the first data point is retained, and the duplicate data points are deleted. If the values ​​of two adjacent data points in all parameters are different, it is determined that there is no duplicate data and it is retained.

[0113] The process of the data preprocessing module normalizing the cleaned data is as follows:

[0114] B21. Calculate the average value of the data set using the mean calculation formula based on the cleaned data. The mean calculation formula is as follows:

[0115] Among them, μ is the mean value of the data set, x i is the cleaned data set, n is the number of data in the data set;

[0116] B22. Calculate the standard deviation of the data based on the calculated mean value. The standard deviation calculation formula is as follows:

[0117] Among them, μ is the mean value of the data set, x i is the cleaned data set, σ is the standard deviation of the cleaned data;

[0118] B23. Normalize the original data in each data set. The normalization formula is as follows:

[0119] Among them, x is the original data in the data set, and z is the normalized data value;

[0120] The above normalization steps are applicable to the processing of temperature data, pressure data and fuel consumption data. When processing each type of data, different types of data sets are used. The temperature data, pressure data and fuel consumption data need to be normalized separately in order to convert the data into a unified dimension range for subsequent processing.

[0121] like Figure 3 As shown, the process of filtering the normalized data by the data preprocessing module is as follows:

[0122] B31. According to the characteristics of the normalized data and the filtering target, select the Haar wavelet or Daubechies wavelet filter, and determine the number of wavelet decomposition layers. The number of decomposition layers determines the degree of subdivision of the data in different frequency bands, which is generally determined through experiments or based on the frequency range and noise distribution of the data.

[0123] B32. Perform wavelet decomposition on the normalized data sequence to obtain the approximate coefficients and detail coefficients of each layer. The calculation formula is as follows:

[0124] A j (k)=Σ n h(nk)*x(n),

[0125] D j (k)=∑ n g(nk)*x(n),

[0126] Among them, A j(k) is the approximate coefficient, x(n) is the normalized data sequence, n is the data point number, k is the coefficient number, h(n) is the low-pass filter coefficient, g(n) is the ordinary filter coefficient, D j (k) is the detail coefficient;

[0127] B33, the threshold M is determined for each layer detail coefficient by the minimum maximum threshold method, and the detail coefficient D of each layer is j (k) Perform threshold processing and calculate the detail coefficient after threshold processing To remove the wavelet coefficients corresponding to the noise, retain the useful signal correlation coefficients, and the detail coefficients after threshold processing The calculation formula is as follows:

[0128]

[0129] B34. The approximate coefficients and details after threshold processing are reconstructed by inverse discrete wavelet transform to obtain the filtered normalized data sequence y(n). The calculation formula of wavelet reconstruction is as follows:

[0130] Among them, h(n) is the low-pass filter coefficient, g(n) is the normal filter coefficient;

[0131] B35. Set the smoothing window size L to smooth the filtered data. The smoothing formula is as follows:

[0132] Among them, y(i) is the average value of the data points within the range of window size L with the current data point i as the center as the smoothed value;

[0133] B36. Using the time axis of the data sequence collected by the temperature sensor as a reference, the pressure data and fuel consumption data are matched with the reference time axis through the timestamp information, so that each data can correspond to each other in time, ensuring that the correlation and comparison between the data are accurate when the features of multiple groups of data are extracted simultaneously in the future;

[0134] The feature extraction module is used to extract time domain features and frequency domain features from the preprocessed data, so as to extract feature information useful for fault diagnosis from the original data and provide input for subsequent fault diagnosis;

[0135] The process of the feature extraction module for extracting time domain features is as follows:

[0136] C11. Calculate the average value of the data sequence based on the preprocessed data sequence The calculation formula is as follows:

[0137] Where N is the total number of preprocessed data, and y(i) is the preprocessed data;

[0138] C12, based on the average Calculate its variance δ 2 , to confirm the fluctuation of time domain characteristics, the variance calculation formula is as follows:

[0139] Where N is the total number of data after preprocessing;

[0140] The larger the variance value, the greater the signal fluctuation, which means that the data may contain more noise or unstable factors. The smaller the variance value, the smaller the signal fluctuation, which means that the data is more stable or the noise level is lower. When extracting time domain features, the discrete degree characteristics of data distribution are obtained by calculating the variance.

[0141] The steps of frequency domain feature extraction by the feature extraction module are as follows:

[0142] C21, performing a bit-reversal operation on the index of the preprocessed data sequence, that is, reversing each bit of the original index to obtain a new index, and then rearranging the data sequence according to the new index;

[0143] C22, use the fixed rotation factor to perform m-level operation iterations to obtain the transformed value x(a), (a=0,1,…,H-1), the iterative operation formula is as follows:

[0144]

[0145] H=2 m , where x(a) is the value of the discrete result of the discrete sequence y(i) at frequency point i, is the rotation factor, is the frequency interval between adjacent frequency discrete points, a and i are frequency index and time index respectively, H is the length of discrete time series y(i), y(i) is the discrete time series data after preprocessing, j is the imaginary unit, and j satisfies 2 = -1;

[0146] C23. Calculate the spectrum amplitude |x(a)| according to the transformed value x(a). The calculation formula is as follows:

[0147] Where Re[x(a)] is the real part of x(a), Im[x(a)] is the imaginary part of x(a);

[0148] C24. After obtaining the spectrum amplitude of each frequency component, find the a value corresponding to the frequency with the largest spectrum amplitude by comparing the spectrum amplitudes, and then calculate the frequency f corresponding to the a value. a , the calculation formula is as follows:

[0149] Among them, f s is the sampling frequency of the preprocessed data, a is the frequency index, and H is the length of the discrete time series y(i);

[0150] The fault diagnosis module is used to establish a fault diagnosis model, and use the fault diagnosis model to accurately determine the fault type of the aircraft compressor according to the extracted time domain features and frequency domain features, and provide timely fault information to maintenance personnel so that corresponding maintenance measures can be taken.

[0151] The steps of establishing the fault diagnosis model by the fault diagnosis module are as follows:

[0152] D11. Collect a large amount of historical operation data of aviation compressors with marked fault types, so that these historical operation data contain time domain features and frequency domain features. The fault type is marked with y i Indicates that y i = +1 or -1, +1 and -1 represent different categories respectively;

[0153] D12. According to the data distribution characteristics and the complexity of fault classification, a prediction function is established. The prediction function is:

[0154] Among them, K(x' i ,x j ') is the predicted value, x' i and x j 'are two input vectors of historical operation data with marked fault types, To predict the bandwidth parameters of the function, the bandwidth parameters need to be adjusted and optimized according to the data;

[0155] D13. Use the support vector machine training algorithm to solve the optimization formula to determine the support vector machine multiplier β i , the optimization formula is as follows:

[0156]

[0157] The constraints are:

[0158] Among them, β i and β j are the multipliers of the two support vector machines, y i and j are the labels corresponding to the fault types after training, τ is the penalty parameter, and n is the number of training samples;

[0159] D14, will satisfy 0≤β i The sample points corresponding to the training samples with ≤τ are used as support vectors (x s ,ys ), use the support vector to calculate the bias term b, the calculation formula is as follows:

[0160] b=y s -Σ i∈SV β i y i K(x' i ,x s ), where SV is the set of all support vectors, x s is the eigenvector corresponding to the support vector, y s is the category label of the feature vector corresponding to the support vector;

[0161] like Figure 4 As shown, the process of the fault diagnosis module for determining the fault type of the aviation air compressor is as follows:

[0162] D21, combining the extracted time domain features and frequency domain features into a feature vector x;

[0163] D22. Calculate the feature vector x using the decision function to confirm the positive and negative values ​​of the calculation results. The calculation formula is as follows:

[0164] Among them, K(x' i ,x) is the calculation result of the prediction function, f(x) is the calculation value of the decision function, b is the bias term of the prediction function, β i is the multiplier of the support vector machine, y i is the fault category label corresponding to the training sample;

[0165] D23. When f(x)>0, it is judged to correspond to y i = +1 indicates the fault type. When f(x) is less than 0, it is judged to correspond to y i = Fault type indicated by -1;

[0166] The result display and alarm module intuitively displays the fault type, fault probability and other information output by the fault diagnosis module on the monitoring display screen or terminal device. At the same time, different levels of alarm thresholds are set according to the severity of the fault. When the diagnosis result reaches the corresponding alarm condition, the corresponding alarm mechanism is activated to ensure that relevant personnel can timely understand the operating status and fault condition of the aviation compressor, and take corresponding response measures according to the severity of the fault to ensure the safe flight of the aircraft.

[0167] The operation flow of the result display and alarm module for result display and alarm is as follows:

[0168] E1. According to the impact of the failure of the air compressor on flight safety and the urgency of maintenance, different levels of alarm thresholds are set in advance for different types of failures;

[0169] E2. Generate data results based on the diagnosis results of the fault diagnosis module. The data results include key information such as fault type and fault probability. The data results are then transmitted to a display device for visual display. The display device includes a maintenance personnel terminal device and a cockpit monitoring display screen:

[0170] E3. Perform logical judgment on the fault type and the corresponding fault probability through the display device to confirm whether the displayed fault type and the corresponding fault probability reach the alarm threshold condition. There are two situations:

[0171] E31. When the displayed fault type and the corresponding fault probability reach the alarm threshold condition, the corresponding level of alarm action is performed;

[0172] E32. When the displayed fault type and the corresponding fault probability do not reach the alarm threshold condition, continue with step E2;

[0173] Through the cooperation of data acquisition module, data preprocessing module, feature extraction module, fault diagnosis module and result display and alarm module, the data acquisition module provides basic data support, the data preprocessing module improves data quality and accuracy, the feature extraction module extracts useful feature information, the fault diagnosis module accurately determines the fault type, and the result display and alarm module ensures that relevant personnel can obtain fault information in time and take corresponding countermeasures. The whole system forms a closed-loop monitoring and diagnosis process, which can effectively improve the data processing rate of aviation compressors, reduce the impact of external signals on data processing, improve the accuracy of data processing, and ensure the safe flight of aircraft.

[0174] Based on the above-described comprehensive fault diagnosis system applicable to an aviation air compressor, the present invention also provides a comprehensive fault diagnosis method applicable to an aviation air compressor, such as Figure 5 As shown, the following steps are included:

[0175] S1. Collect multi-dimensional operation data of exhaust temperature, pressure change and fuel consumption of the aviation air compressor in real time through the data acquisition module;

[0176] S2, cleaning, normalizing and filtering the collected data through a data preprocessing module;

[0177] S3, extracting time domain features and frequency domain features from the preprocessed data through a feature extraction module;

[0178] S4. Construct a fault diagnosis model and perform fault diagnosis based on the extracted time domain features and frequency domain features. If a fault exists, different levels of warnings are issued according to the severity of the fault, and the fault information is visualized. If no fault exists, continue to perform fault diagnosis.

[0179] 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 substitutions for some of the technical features therein; and these modifications or substitutions 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 invention.

Claims

1. A comprehensive fault diagnosis system suitable for aviation air compressors, characterized in that: include: A data acquisition module, which is used to collect multi-dimensional operating data such as exhaust temperature, pressure change and fuel consumption during the operation of the aircraft compressor in real time, so as to ensure that various operating parameters of the aircraft compressor can be obtained in real time and accurately; A data preprocessing module, which is used to clean, normalize and filter the collected data to improve the quality and accuracy of the data; A feature extraction module, which is used to extract time domain features and frequency domain features from the preprocessed data to extract feature information useful for fault diagnosis from the original data; A fault diagnosis module, wherein the fault diagnosis module is used to establish a fault diagnosis model and use the fault diagnosis model to accurately determine the fault type of the aviation air compressor according to the extracted time domain features and frequency domain features; A result display and alarm module, which intuitively displays the information output by the fault diagnosis module on a monitoring display screen or a terminal device, and sets different levels of alarm thresholds according to the severity of the fault. When the diagnosis result reaches the corresponding alarm condition, the corresponding alarm mechanism is activated to ensure that relevant personnel can timely understand the operating status and fault condition of the aviation air compressor, and take corresponding countermeasures according to the severity of the fault to ensure the safe flight of the aircraft; The data acquisition module is electrically connected to the data preprocessing module, the data preprocessing module is electrically connected to the feature extraction module, the feature extraction module is electrically connected to the fault diagnosis module, and the fault diagnosis module is electrically connected to the result display and alarm module.

2. A comprehensive fault diagnosis system for an aviation air compressor according to claim 1, characterized in that: The data acquisition module includes a temperature acquisition unit for collecting the exhaust temperature during the operation of the aviation air compressor, a pressure acquisition unit for collecting the pressure change during the operation of the aviation air compressor, and a fuel consumption acquisition unit for collecting the fuel consumption during the operation of the aviation air compressor. The temperature acquisition unit, the pressure acquisition unit and the fuel consumption acquisition unit are all electrically connected to the data preprocessing module.

3. A comprehensive fault diagnosis system for an aviation air compressor according to claim 2, characterized in that: The process of the fuel consumption collection unit for collecting fuel consumption is as follows: A31. Install the turbine flow sensor in the fuel supply pipeline of the aviation air compressor to ensure that the fuel can pass through the turbine flow sensor smoothly without leakage; A32. Set the sampling frequency of the turbine flow sensor. The turbine flow sensor collects the rotation frequency of the turbine in real time, and then calculates the flow rate based on the rotation frequency. The calculation formula is as follows: Among them, f is the rotation frequency of the turbine, K is the instrument factor of the turbine flow sensor, and Q is the fuel flow rate.

4. A comprehensive fault diagnosis system for an aviation air compressor according to claim 3, characterized in that: The process of the data preprocessing module cleaning the collected data is as follows: B11. According to the technical standards for the normal operation of the air compressor and past operating experience, determine the normal range of exhaust temperature, pressure and fuel consumption parameters. The normal range of exhaust temperature is 50℃-150℃, the normal range of intake pressure is 3-6 atmospheres, and the normal range of fuel consumption is 0.1-0.5 liters per minute; B12. Classify and archive the exhaust temperature, pressure and fuel consumption, and then traverse the data sequence of each parameter collected according to the normal value range, and judge each data point to confirm whether there is data beyond the normal value range. If there is data beyond the normal value range, it is judged as abnormal data and directly eliminated. If it is within the normal value range, it is retained as normal data; B13. Sort the normal data, and then compare the adjacent data points in turn. If the values ​​of two adjacent data points in all parameters are exactly the same, the following data points are considered duplicate data, the first data point is retained, and the duplicate data points are deleted. If the values ​​of two adjacent data points in all parameters are not the same, it is determined that there is no duplicate data and it is retained.

5. A comprehensive fault diagnosis system for an aviation air compressor according to claim 4, characterized in that: The process of the data preprocessing module normalizing the cleaned data is as follows: B21. Calculate the average value of the data set using the mean calculation formula based on the cleaned data. The mean calculation formula is as follows: Among them, μ is the mean value of the data set, x i is the cleaned data set, n is the number of data in the data set; B22. Calculate the standard deviation of the data based on the calculated mean value. The standard deviation calculation formula is as follows: Among them, μ is the mean value of the data set, x i is the cleaned data set, σ is the standard deviation of the cleaned data; B23. Normalize the original data in each data set. The normalization formula is as follows: Among them, x is the original data in the data set, and z is the normalized data value.

6. A comprehensive fault diagnosis system for aviation air compressors according to claim 5, characterized in that: The process of filtering the normalized data by the data preprocessing module is as follows: B31. According to the characteristics of the normalized data and the filtering target, select the Haar wavelet or Daubechies wavelet filter and determine the number of wavelet decomposition layers; B32. Perform wavelet decomposition on the normalized data sequence to obtain the approximate coefficients and detail coefficients of each layer. The calculation formula is as follows: A j (k)=∑ n h(n-k)*x(n), D j (k)=∑ n g(n-k)*x(n), Among them, A j (k) is the approximate coefficient, x(n) is the normalized data sequence, n is the data point number, k is the coefficient number, h(n) is the low-pass filter coefficient, g(n) is the ordinary filter coefficient, D j (k) is the detail coefficient; B33, the threshold M is determined for each layer detail coefficient by the minimum maximum threshold method, and the detail coefficient D of each layer is j (k) Perform threshold processing and calculate the detail coefficient after threshold processing Detail coefficient after threshold processing The calculation formula is as follows: B34. The approximate coefficients and details after threshold processing are reconstructed by inverse discrete wavelet transform to obtain the filtered normalized data sequence y(n). The calculation formula of wavelet reconstruction is as follows: Among them, h(n) is the low-pass filter coefficient, g(n) is the normal filter coefficient; B35. Set the smoothing window size L to smooth the filtered data. The smoothing formula is as follows: Among them, y(i) is the average value of the data points within the range of window size L with the current data point i as the center as the smoothed value; B36. Using the time axis of the data sequence collected by the temperature sensor as a reference, the pressure data and fuel consumption data are matched with the reference time axis through the timestamp information, so that each data can correspond one to one in time.

7. A comprehensive fault diagnosis system for an aviation air compressor according to claim 6, characterized in that: The process of the feature extraction module for extracting time domain features is as follows: C11. Calculate the average value of the data sequence based on the preprocessed data sequence The calculation formula is as follows: Where N is the total number of preprocessed data, and y(i) is the preprocessed data; C12, based on the average Calculate its variance δ 2 , to confirm the fluctuation of time domain characteristics, the variance calculation formula is as follows: Where N is the total number of data after preprocessing; The steps of frequency domain feature extraction by the feature extraction module are as follows: C21, performing a reverse order operation on the index of the preprocessed data sequence; C22, use the fixed rotation factor to perform m-level operation iterations to obtain the transformed value x(a), (a=0,1,…,H-1), the iterative operation formula is as follows: H=2 m , where x(a) is the value of the discrete result of the discrete sequence y(i) at frequency point i, is the rotation factor, is the frequency interval between adjacent frequency discrete points, a and i are frequency index and time index respectively, H is the length of discrete time series y(i), y(i) is the discrete time series data after preprocessing, j is the imaginary unit, and j satisfies 2 = -1; C23. Calculate the spectrum amplitude |x(a)| according to the transformed value x(a). The calculation formula is as follows: Where Re[x(a)] is the real part of x(a), Im[x(a)] is the imaginary part of x(a); C24. After obtaining the spectrum amplitude of each frequency component, find the a value corresponding to the frequency with the largest spectrum amplitude by comparing the spectrum amplitudes, and then calculate the frequency f corresponding to the a value. a , the calculation formula is as follows: Among them, f s is the sampling frequency of the preprocessed data, a is the frequency index, and H is the length of the discrete time series y(i).

8. A comprehensive fault diagnosis system for aviation air compressors according to claim 7, characterized in that: The steps of establishing the fault diagnosis model by the fault diagnosis module are as follows: D11. Collect a large amount of historical operation data of aviation compressors with marked fault types, so that these historical operation data contain time domain features and frequency domain features. The fault type is marked with y i Indicates that y i = +1 or -1; D12. According to the data distribution characteristics and the complexity of fault classification, a prediction function is established. The prediction function is: Among them, K(x' i ,x' j ) is the predicted value, x' i and x' j are the input vectors of two historical operating data with marked fault types, φ is the bandwidth parameter of the prediction function; D13. Use the support vector machine training algorithm to solve the optimization formula to determine the support vector machine multiplier β i , the optimization formula is as follows: The constraints are: Among them, β i and β j are the multipliers of the two support vector machines, y i and j are the labels corresponding to the fault types after training, τ is the penalty parameter, and n is the number of training samples; D14, will satisfy 0≤β i The sample points corresponding to the training samples with ≤τ are used as support vectors (x s ,y s ), use the support vector to calculate the bias term b, the calculation formula is as follows: b=y s -∑ i∈SV β i y i K(x' i ,x s ), where SV is the set of all support vectors, x s is the eigenvector corresponding to the support vector, y s is the category label of the feature vector corresponding to the support vector; The process of the fault diagnosis module for determining the fault type of the aviation air compressor is as follows: D21, combining the extracted time domain features and frequency domain features into a feature vector x; D22. Calculate the feature vector x using the decision function to confirm the positive and negative values ​​of the calculation results. The calculation formula is as follows: Among them, K(x' i ,x) is the calculation result of the prediction function, f(x) is the calculation value of the decision function, b is the bias term of the prediction function, β i is the multiplier of the support vector machine, y i is the fault category label corresponding to the training sample; D23. When f(x)>0, it is judged to correspond to y i = +1 indicates the fault type. When f(x) is less than 0, it is judged to correspond to y i = -1 indicates the fault type.

9. A comprehensive fault diagnosis system for aviation air compressors according to claim 8, characterized in that: The operation flow of the result display and alarm module for result display and alarm is as follows: E1. According to the impact of the failure of the air compressor on flight safety and the urgency of maintenance, different levels of alarm thresholds are set in advance for different types of failures; E2. Generate data results based on the diagnosis results of the fault diagnosis module. The data results include key information such as fault type and fault probability. The data results are then transmitted to a display device for visual display. The display device includes a maintenance personnel terminal device and a cockpit monitoring display screen: E3. Perform logical judgment on the fault type and the corresponding fault probability through the display device to confirm whether the displayed fault type and the corresponding fault probability reach the alarm threshold condition. There are two situations: E31. When the displayed fault type and the corresponding fault probability reach the alarm threshold condition, the corresponding level of alarm action is performed; E32. When the displayed fault type and the corresponding fault probability do not reach the alarm threshold condition, continue with step E2.

10. A method for comprehensive fault diagnosis of an aviation air compressor, based on a comprehensive fault diagnosis system for an aviation air compressor according to any one of claims 1 to 9, characterized in that: The steps include: S1. Collect multi-dimensional operation data of exhaust temperature, pressure change and fuel consumption of the aviation air compressor in real time through the data acquisition module; S2, cleaning, normalizing and filtering the collected data through a data preprocessing module; S3, extracting time domain features and frequency domain features from the preprocessed data through a feature extraction module; S4. Construct a fault diagnosis model and perform fault diagnosis based on the extracted time domain features and frequency domain features. If a fault exists, different levels of warnings are issued according to the severity of the fault, and the fault information is visualized. If no fault exists, continue to perform fault diagnosis.

Images