An automatic detection method for strain sensor failure in full-machine fatigue testing

Abstract

The present invention proposes an automatic detection method for failure of strain sensors in full-aircraft fatigue tests. By identifying the response characteristics of failed sensors, constructing mathematical models, establishing failure criteria, and determining failure thresholds, the method achieves efficient automatic detection of failed strain sensors during full-aircraft fatigue tests, improves detection efficiency, and solves the problems of low efficiency and one-sided criteria in manual identification of failed strain sensors in previous full-aircraft fatigue tests. It avoids the risk of time gaps in damage monitoring of key structural parts due to untimely detection of sensor failure, and provides a reliable technical means for timely detection, repair, or replacement of failed sensors.

Description

Technical Field

[0001] The invention belongs to the technical field of aircraft fatigue testing, and in particular relates to an automatic detection method for failure of a strain sensor in a full aircraft fatigue test. Background Art

[0002] During full-aircraft fatigue testing, a large number of strain sensors are arranged on fatigue-critical structures. Since the testing lasts for several years, some sensors may fail during the test. If not discovered in time, strain data will be lost, and there will be time gaps in structural damage monitoring, which will bring test safety risks. After the sensor fails, the distorted strain response will also have an adverse effect on damage identification based on strain monitoring.

[0003] At present, the main method of judging whether a sensor has failed is to observe with the naked eye whether the strain response curve of each sensor is normal. The main basis for judgment is to check whether the strain response exceeds the range or has no response. This method is inefficient and has one-sided judgment criteria. Summary of the Invention

[0004] Purpose of the present invention: The present invention proposes an automatic detection method for strain sensor failure in a full-machine fatigue test. By identifying the response characteristics of the failed sensor, constructing a mathematical model, and establishing failure criteria, efficient automatic detection of failed strain sensors during the test is achieved, providing a reliable technical means for timely discovery, repair or replacement of failed sensors.

[0005] The technical solution of the present invention:

[0006] An automatic detection method for strain sensor failure in a full-machine fatigue test comprises the following steps:

[0007] Step 1: Extract the strain sensor failure data characteristics based on the failure data of the same type of strain sensor in previous full-aircraft fatigue tests;

[0008] Step 2: Mathematically model the failure data characteristics of the strain sensor and determine the typical characteristic parameters and mathematical expressions of each failure data characteristic;

[0009] Step 3: Establish strain sensor failure criteria based on typical characteristic parameters of failure data characteristics;

[0010] Step 4: During the new full-aircraft fatigue test commissioning phase, determine the sensor failure criterion threshold parameter range based on the commissioning data;

[0011] Step 5: In the new full-aircraft fatigue test, the response data of all strain sensors are collected through the data acquisition system, and the typical characteristic parameters of the response data of each strain sensor are calculated;

[0012] Step 6: Based on the strain sensor failure criterion and the typical characteristic parameters obtained in step 5, determine whether each strain sensor has failed.

[0013] Step 7: Summarize the statistics of failed strain sensors so that test personnel can check, repair and replace the corresponding sensors.

[0014] Furthermore, in step one, the strain sensor failure data characteristics include: characteristic one: the strain sensor response data exceeds the strain sensor range, characteristic two: the strain sensor response data is a constant value and does not change with the external load, and characteristic three: the strain sensor response data shows an increasing / decreasing phenomenon that is independent of the external load size.

[0015] Furthermore, the typical characteristic parameters involved in feature 1 include: mean standard deviation s, least squares regression slope k between response data and time series;

[0016] Typical characteristic parameters involved in feature 2 include: mean standard deviation s, least squares regression slope k between response data and time series;

[0017] Typical characteristic parameters involved in feature three include: the least squares regression slope k between the response data and the time series.

[0018] Features 1 to 3 s and k are calculated using formulas (1) to (3) respectively:

[0019]

[0020]

[0021]

[0022] Where x is the time series of sensor data, y is the sensor response data sequence, and n is the number of data points in y.

[0023] Furthermore, in step 3, the strain sensor failure criterion includes:

[0024] If the absolute value of the least squares regression slope between the response data of the strain sensor and the time series is not greater than 10 -6 , then it is judged that the strain sensor fails, and the failure data feature is feature 2;

[0025] If the absolute value of the least squares regression slope between the strain sensor response data and the time series is greater than 10 -6 And the absolute value of the mean is not greater than ε th , the standard deviation is greater than s th1, then the strain sensor is judged to be failed, and the failure data feature is feature three;

[0026] If the absolute value of the least squares regression slope between the strain sensor response data and the time series is greater than 10 -6 And the absolute value of the mean is greater than ε th , the standard deviation is greater than s th2 , then the strain sensor is judged to be failed, and the failure data feature is feature 1;

[0027] Otherwise, the sensor is judged to be normal;

[0028] Among them, ε th 、s th1 、s th2 is the failure criterion threshold.

[0029] Furthermore, in step 4, during the new full-machine fatigue test debugging phase, debugging data of strain sensors known to have not failed are collected, and the failure criterion threshold parameter value is determined based on the debugging data of strain sensors known to have not failed.

[0030] Furthermore, in step 4, based on the actual data of different full-machine fatigue tests, the failure criterion threshold parameter range is as follows:

[0031] ε th Between [500,2000], s th1 Between [500,1500], s th2 In [1000,2500].

[0032] Furthermore, the method further includes step seven: compiling statistics of failed strain sensors for testing, repair and replacement by test personnel.

[0033] Furthermore, in step 4, before conducting full-aircraft fatigue tests on different aircraft, it is necessary to go through the full-aircraft fatigue test debugging phase to re-determine the sensor failure criterion threshold parameter values.

[0034] Beneficial effects of the present invention:

[0035] To solve the problems of low efficiency and one-sided judgment in manual identification of failed strain sensors in full-machine fatigue tests, and time gaps in damage monitoring of key structural parts caused by untimely detection, a method for automatically detecting whether strain sensors have failed is established to ensure the safe and smooth progress of the test.

[0036] The method provides a relatively accurate and rapid method for determining whether a strain sensor has failed for full-aircraft fatigue tests with a test cycle of several years and hundreds or even thousands of strain sensors. This method is conducive to early detection of failed sensors and timely repair or replacement, avoiding safety hazards in the test caused by time gaps in the monitoring of key structural parts due to undetected sensor failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 The following is a technical flow diagram of an automatic detection method for strain sensor failure in full-machine fatigue testing;

[0038] Figure 2 (a) is a schematic diagram of the data feature ① of the failed sensor;

[0039] Figure 2 (b) is a schematic diagram of the data feature ② of the failed sensor;

[0040] Figure 2 (c) is a schematic diagram of data characteristics ③ of the failed sensor;

[0041] Figure 3 (a) is a schematic diagram showing the probability distribution of the mean value of the response data of the strain sensors that have not failed during the full fatigue test commissioning phase of a certain type of aircraft in a specific embodiment;

[0042] Figure 3 (b) is a schematic diagram showing the probability distribution of the sample standard deviation of the strain sensors whose response data of the strain sensors that have not failed have an absolute value greater than 500 during the full fatigue test commissioning phase of a certain type of aircraft in a specific embodiment;

[0043] Figure 3 (c) is a schematic diagram showing the probability distribution of the sample standard deviation of the strain sensors whose absolute value of the mean of the response data of the strain sensors that have not failed is less than 500 during the full fatigue test commissioning phase of a certain type of aircraft in a specific embodiment;

[0044] Figure 4 (a) is a schematic diagram of the data history of a failed sensor that meets failure characteristic ① and is identified using the established failure criterion in a specific embodiment;

[0045] Figure 4 (b) is a schematic diagram of the data history of a failed sensor that meets failure characteristic ③ and is identified using the established failure criterion in a specific embodiment;

[0046] Figure 4 (c) is a schematic diagram of the data history of a failed sensor (having both characteristics ① and ② when viewed in sections) that meets failure characteristic ③ and is identified using the established failure criterion in a specific embodiment. DETAILED DESCRIPTION

[0047] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0048] A method for automatically detecting failure of a strain sensor in a full-machine fatigue test, one of the possible specific implementations of which is as follows, is characterized by comprising the following steps:

[0049] The first step is to extract the data characteristics after the strain sensor fails based on the application data of the same type of strain sensor in the past full-machine fatigue test. By analyzing the response change pattern of each group of strain sensors over time, three typical characteristics of the response data after the strain sensor fails are summarized: ① Exceeding the range of the strain sensor, for example Figure 2 As shown in (a), the strain sensor range is generally ±20000. After the sensor fails, the response data reaches -1000000; ② It does not change with the external load, and the strain response is always constant. For example, Figure 2 As shown in (b), after the sensor fails, its response data is always 0; ③ The strain value shows an increasing / decreasing phenomenon that is independent of the external load, for example Figure 2 As shown in (c), after the sensor fails, its response data shows a continuous monotonically increasing or monotonically decreasing phenomenon, which is inconsistent with the fluctuation of the external load. Among them, feature ③ has been omitted in previous tests and is prone to false alarms in structural damage monitoring;

[0050] The second step is to conduct mathematical modeling for the three features of the first step. Using statistical analysis methods, the characteristic parameters of features ① and ③ are determined as: mean Sample standard deviation s, least squares regression slope k between strain data and time series; the characteristic parameter of feature ② is determined to be the least squares regression slope k between strain data and time series;

[0051] The mean The calculation formulas for the sample standard deviation s and slope k are formula (1), formula (2), and formula (3), respectively. Where x is the time series of the extracted data, y is the sensor data series, and n is the number of data points in y. Slope k is the slope of the line obtained by performing a linear regression on (x, y) using the least squares method.

[0052]

[0053]

[0054]

[0055] The third step is to establish the criteria for determining whether the strain sensor is failed as follows:

[0056] If |k|≤10 -6 ,

[0057] Then the response of the strain sensor is always constant (corresponding to feature ②), and the sensor has failed;

[0058] If |k|>10 -6 and s>s th1 ,

[0059] The response of the strain sensor shows an increasing / decreasing phenomenon that is independent of the external load (corresponding to feature ③), and the sensor has failed.

[0060] If |k|>10 -6 ,and s>s th2 ,

[0061] The response of the strain sensor exceeds the sensor's range (corresponding to feature ①), and the sensor has failed; otherwise, the strain sensor is normal.

[0062] Among them, the ε th 、s th1 、s th2 The threshold value for failure criterion is different for full-aircraft fatigue tests of different aircraft.

[0063] The fourth step is to use the debugging data to perform statistical analysis on the data of all the strain sensors that have not failed during the full-machine fatigue test debugging phase, and use formulas (1) to (3) to calculate the mean value of the response data of each strain sensor that has not failed. Sample standard deviation s, slope k, respectively statistical and the probability distribution of s. Figure 3 Statistics of fatigue test and debugging data of a certain type of aircraft, such as Figure 3 (a) shows the mean Most of the values of are distributed between [-500,500] and obey the normal distribution; Figure 3 (b) shows the mean The standard deviation of the strain sensor follows a log-normal distribution, and most of them are distributed between [300,1250]. Figure 3 (c) shows the mean The standard deviation of the strain sensor also obeys the log-normal distribution, and most of them are distributed between [0,500].

[0064] According to the above statistics, an iterative method is used to determine ε between [500, 2000] thThe value of is 500, and s is determined between [500,1500]. th1 The value of is 1000, and s is determined between [1000,2500]. th2 The value is 1500.

[0065] In the fifth step, during the full-machine fatigue test, the data of all strain sensors newly collected by the data acquisition system are used to calculate the mean value of the response data of each strain sensor in the second step using formulas (1) to (3) in sequence. Sample standard deviation s, slope k;

[0066] Step 6: For each strain sensor, the mean value calculated in step 5 is The sample standard deviation s and slope k are substituted into the failure criterion determined in the third and fourth steps to automatically determine whether the strain sensor has failed. Obviously, it is easy to identify the constant feature ②, and no case is given here. Figure 4 (a)~ Figure 4 The case shown in (c) is to identify the data features of the failed sensor based on the criteria determined in the third and fourth steps, which correspond to features ①, ③, and ③ respectively. Their typical feature parameter combinations (k, s) are (27.55, -317852.8, 465707.6), (-0.55, 158.3, 4857.1), and (0.011, 109.7, 10826.9) respectively.

[0067] in, Figure 4 Case (b) is often overlooked in previous tests. After a sensor fails, its response value does not exceed the range or remains constant, which is difficult to notice. However, the data shows an increase or decrease that is unrelated to the external load and is invalid. Figure 4 For the sensor in (c), only one point has a value of 120,000, and the other values are all 0. From the value of the characteristic parameter, it satisfies feature ③, but from the segmentation point of view, it is a combination of features ① and ②.

[0068] In the seventh step, the number of the detected failed strain sensor and its corresponding characteristics are automatically written into a file in a specified format, and the feedback is fed back to the test personnel to complete the investigation and confirmation as soon as possible, and to carry out repairs or replacements in a timely manner.

[0069] The above description is merely a detailed description of specific embodiments of the present invention. Any unspecified portions are conventional techniques. However, the scope of the present invention is not limited thereto. Any changes or substitutions that can be readily conceived by a person skilled in the art within the technical scope disclosed in the present invention are intended to be encompassed within the scope of the present invention. The scope of the present invention shall be determined by the scope of the claims.

Claims

1. A method for automatically detecting failure of strain sensors in full-machine fatigue testing, characterized by: The method comprises the following steps: Step 1: Based on the failure data of the same type of strain sensor in previous full-aircraft fatigue tests, extract the strain sensor failure data characteristics, including: Feature 1: the strain sensor response data exceeds the strain sensor range; Feature 2: the strain sensor response data is constant and does not change with the external load; Feature 3: the strain sensor response data shows an increasing / decreasing phenomenon that is independent of the external load; Step 2: Mathematically model the failure data characteristics of the strain sensor and determine the typical characteristic parameters and mathematical expressions of each failure data characteristic; Step 3: Establish the strain sensor failure criterion based on the typical characteristic parameters of the failure data characteristics, including: if the absolute value of the least square regression slope between the response data of the strain sensor and the time series is not greater than 10 -6 , then it is judged that the strain sensor fails, and the failure data feature is feature 2; If the absolute value of the least squares regression slope between the strain sensor response data and the time series is greater than 10 -6 And the absolute value of the mean is not greater than ε th , the standard deviation is greater than s th1 , then the strain sensor is judged to be failed, and the failure data feature is feature three; If the absolute value of the least squares regression slope between the strain sensor response data and the time series is greater than 10 -6 And the absolute value of the mean is greater than ε th , the standard deviation is greater than s th2 , then the strain sensor is judged to be failed, and the failure data feature is feature 1; Otherwise, the sensor is judged to be normal; among them, ε th 、 s th1 、 s th2 is the failure criterion threshold; Step 4: During the new full-aircraft fatigue test commissioning phase, determine the sensor failure criterion threshold parameter range based on the commissioning data; Step 5: In the new full-aircraft fatigue test, the response data of all strain sensors are collected through the data acquisition system, and the typical characteristic parameters of the response data of each strain sensor are calculated; Step 6: Based on the strain sensor failure criterion and the typical characteristic parameters obtained in step 5, determine whether each strain sensor has failed; Step 7: Summarize the statistics of failed strain sensors so that test personnel can check, repair and replace the corresponding sensors.

2. The method according to claim 1, wherein: Typical characteristic parameters involved in feature 1 include: mean , standard deviation s , the least squares regression slope between the response data and the time series k , calculated using formulas (1) to (3) respectively; (1) (2) (3) in, is the time series of sensor data, is the sensor response data sequence, n for The number of data points in the Typical characteristic parameters involved in feature 2 include: mean , standard deviation s , the least squares regression slope between the response data and the time series k ; Typical characteristic parameters involved in feature three include: the least squares regression slope between the response data and the time series k .

3. The method according to claim 2, wherein: In the fourth step, during the new full-machine fatigue test debugging phase, the debugging data of the strain sensors known to have not failed are collected, and the debugging data of all the strain sensors known to have not failed are calculated. ε th 、 s th1 、 s th2 , and separately count ε th 、 s th1 、 s th2 The distribution of , based on which the failure criterion threshold parameter range is determined.

4. The method according to claim 3, wherein: In step 4, the failure criterion threshold parameter range is as follows. The specific value needs to be determined through iterative calculation based on the data from the full-machine fatigue test commissioning phase: ε th Between [500,2000], s th1 Between [500,1500], s th2 Between [1000,2500].

5. The method according to claim 4, characterized in that: The method further includes step seven: compiling statistics of failed strain sensors for testing, repair, and replacement by test personnel.

6. The method according to claim 5, characterized in that: In step 4, before automatically detecting strain sensor failure during full-aircraft fatigue tests on different aircraft, the sensor failure criterion threshold parameter range must be re-determined based on data from the full-aircraft fatigue test debugging phase.

Images