Spacecraft porthole structure sound source positioning method based on gradient descent solution

Abstract

The present application relates to a kind of spacecraft porthole structure sound source positioning method based on gradient descent solution, comprising 1) determining signal range;2) calculating time difference of arrival;3) determine starting position;4) calculate shortest path;5) establish objective function;6) positioning solution.The present application can carry out real-time detection to the collision positioning position of through structure, provide guarantee for the operation safety of on-orbit vehicle, when space debris and system collide, collision position will produce elastic lamb wave signal, utilize piezoelectric sensor at different positions to monitor lamb wave signal, and the collision position of fragment and system can be calculated by specific detection algorithm.The present application solves the collision positioning problem of spacecraft porthole structure propagation path not being straight line, and the required piezoelectric sensor array is simple, and strong in applicability;Meanwhile, gradient descent method based on correction step compensation and deviation coefficient speeds up the iteration speed without additional calculation, saves time cost.

Description

Technical Field

[0001] This invention belongs to the field of spacecraft on-orbit structural health monitoring technology, specifically relating to a method for locating sound sources in spacecraft window structures based on gradient descent calculation. Background Technology

[0002] The risk of space debris is one of the biggest problems affecting the long-term sustainability of outer space activities. Unavoidable accidents in space, such as explosions or collisions caused by meteoroids and abandoned space facilities, generate a large amount of space debris, which poses safety hazards to the normal operation of spacecraft.

[0003] Currently, the trajectories of space debris at the centimeter level and above can be predicted, allowing spacecraft to avoid them through active orbital maneuvers. Millimeter-sized debris poses the greatest threat, as it is neither preventable nor avoidable. If it collides with a spacecraft, the spacecraft in operation is at risk of damage, potentially leading to damage to critical components and leaks in sealed containers, severely impacting the operational safety of the spacecraft and the lives of the astronauts.

[0004] Ultrasonic Lamb waves have shown great potential in collision detection due to their advantages such as long propagation distance, small attenuation, and high sensitivity to small defects. In isotropic structures, the Time Difference of Arrival (TDOA) method, which determines the collision location based on the time difference of the probe signal arriving at different sensors, has been widely used. However, this method is based on the isotropic wave velocity assumption and the linear propagation path, which greatly limits its application under practical conditions.

[0005] The geometric features of spacecraft windows significantly affect the propagation path of Lamb waves, making it difficult to effectively locate collision sites using both the direct time-of-flight method and the time difference of flight method. The following problems and limitations exist in acoustic collision localization based on perforated discontinuous structures:

[0006] (1) Lamb wave propagation is hindered by voids, and the propagation path is no longer a straight line. The positioning method needs to select the shortest ray path from multiple ray paths.

[0007] (2) Common ultrasonic testing and acoustic emission testing methods have poor ability to assess and locate discontinuous window structures in special parts of spacecraft and cannot effectively monitor the real-time location of collision points.

[0008] Therefore, finding a method for locating acoustic emission sources that can be applied to the window structure of real spacecraft remains challenging. Summary of the Invention

[0009] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for locating the sound source of a spacecraft window structure based on gradient descent calculation. This method can detect the collision location of the penetrating structure in real time, thus ensuring the operational safety of the spacecraft in orbit. When space debris collides with the system, an elastic Lamb wave signal is generated at the collision location. By using ultrasonic sensors at different locations to monitor the Lamb wave signal, the collision location between the debris and the system can be calculated through a specific detection algorithm.

[0010] The technical problem solved by this invention is achieved through the following technical solution:

[0011] A method for locating sound sources in a spacecraft window structure based on gradient descent calculation is characterized by the following: the positioning system used in the method includes piezoelectric sensors, signal amplifiers, a signal acquisition card, and a computer. The piezoelectric sensors are arranged at all four corners of the spacecraft window structure, and the distances of the piezoelectric sensors at the four corners from the center of the spacecraft window structure are the same. Each piezoelectric sensor is connected to the signal amplifier, which amplifies the elastic wave signal obtained by the piezoelectric sensors and transmits it to the signal acquisition card. The signal acquisition card transmits the acquired information to the computer. The piezoelectric sensors are distributed in four quadrants with the center of the spacecraft window structure as the origin (x0, y0). The coordinate position of the i-th piezoelectric sensor is (x0, y0). i y i );

[0012] The steps of the positioning method are as follows:

[0013] 1) Determine the signal range

[0014] A sliding energy window is used to process the 125kHz to 250kHz signal band received by each piezoelectric sensor. First, a signal window of a certain width is extracted from the starting time point of the Lamb wave signal, and the total energy of the signal in the window is taken as the energy value at the starting time point. Then, moving forward one step from that point, the signal of the next window width is extracted, and the total energy of the signal in the next window is taken as the energy value at the next point. The signal window moves along the time axis according to the step size, and the energy value of the signal in the window is taken as the energy value of the signal at the starting point of the window.

[0015]

[0016] Where: T(t) i () represents the absolute value of energy at each point;

[0017] T w Indicates the length of the energy window;

[0018] I(t) represents the total energy of the energy window corresponding to the current point;

[0019] Plot all energy value points to obtain the time-energy diagram of the Lamb wave signal, and identify the peak point with a normalized energy peak value greater than 0.4 as a collision occurrence. The signals of the 5000 points before and the 10000 points after the energy peak point that meet the condition are taken as the time range of the collision occurrence event.

[0020] 2) Calculate the time difference of arrival

[0021] The signal envelope within a selected range is determined using Hilbert transform to ascertain the signal's trend. The upper and lower envelopes are then processed using the AIC criterion and averaged to accurately obtain the start time of the S0 mode. To reduce errors caused by spurious signals and facilitate calculation, the absolute value of the obtained function value is calculated and multiplied by a scaling factor. The calculation formula is as follows:

[0022] AIC(t w )=K|t w ·ln(var(T(1,t w )))+(T w -t w -1)·ln(var(T(1+t w ,T w )))|

[0023]

[0024] T represents the selected range, which includes signal lengths from 1 to T. w ;

[0025] t w This indicates that only signals within the selected range T are included in the calculation;

[0026] var represents the variance function;

[0027] var(T(1,t w )) indicates that only the sequence (1, t) within the selected range T is calculated. w The variance of the signal;

[0028] max(T(1,T w )) represents the maximum value of the signal within the selected range;

[0029] max(pks(T)) represents the maximum peak value of the AIC curve;

[0030] K represents the scaling factor;

[0031] The arrival times of the S0 mode Lamb wave signals from the four piezoelectric sensors were recorded. The red line represents the Lamb wave signal within the selected range, the black dashed line represents the AIC curve, and the green line represents the start time of the S0 mode.

[0032] For the i-th sensor, the arrival time of the Lamb wave signal received in the debris collision event is t. i This corresponds to the time point of the first peak of the i-th AIC curve. Calculate the actual arrival difference between the i-th and j-th sensors and express it as Δt. ij ,

[0033] t i =pks(AIC i )

[0034] Δt ij =t i -t j

[0035] 3) Determine the starting position

[0036] By comparing the start times of signal reception by piezoelectric sensors in different quadrants, the piezoelectric sensor that receives the signal first is closest to the acoustic emission source. Therefore, we can approximate the initial coordinates of the acoustic emission source point as follows:

[0037] (x0,y0)=(C,C), if the first signal received by the sensor is located in the first quadrant;

[0038] (x0,y0)=(-C,C), if the first signal received by the sensor is located in the second quadrant;

[0039] (x0,y0)=(-C,-C), if the first signal received by the sensor is located in the third quadrant;

[0040] (x0,y0)=(C-,C), if the first signal received by the sensor is located in the fourth quadrant;

[0041] Where C represents a constant;

[0042] 4) Calculate the shortest path

[0043] To determine the shortest signal path, the aperture structure of each piezoelectric sensor is divided into different regions. For the second quadrant sensor, the tangent points between the piezoelectric sensor and the center of the spacecraft window structure are a and b. This plane is divided into six regions by the sensor tangent. The directional path from the piezoelectric sensor S to the acoustic emission source O is defined as a vector. vector and Starting from O and ending at a and b, Xob is calculated and The value obtained from the vector outer product:

[0044]

[0045] Xoa is also calculated and recorded in the same way. If both Xoa and Xob are non-positive or non-negative, it means that the acoustic emission source O is located in region A or D.

[0046] Otherwise, the acoustic emission source O is located in another region;

[0047] Then, through calculation and Yob is obtained by taking the value of the vector dot product;

[0048]

[0049] Xoa is also calculated and recorded in the same way. If both Xoa and Xob are negative, it means that the acoustic emission source O is located in region F.

[0050] Otherwise, calculate the magnitude of the vector for further judgment:

[0051]

[0052] If Zob is negative, it indicates that the acoustic emission source O is located in region E;

[0053] Otherwise, if |x|≥|y|, then the acoustic emission source O is located in region B; if |x|<|y|, then it is located in region C.

[0054] Clearly, when the acoustic emission source O is located in regions A, D, E, and F, the shortest path d2 for the sensor to receive the signal is a straight line;

[0055]

[0056] When the acoustic emission source O is located in region B or C, the shortest path d2 is the sum of multiple path segments. The points of tangency between the acoustic emission source O and the center of the spacecraft window structure are c and d. When the acoustic emission source O is located in region B or c, the shortest path for the sensor to receive the signal is dd2. ac or dd2 bd :

[0057]

[0058]

[0059] Where Lac and Lbd represent the arc lengths of two points on the circular hole;

[0060] By determining the region where the acoustic emission source O is located, the shortest path distance can be calculated:

[0061] If the acoustic emission source O is located in regions A, D, E, and F;

[0062] d2 = dd2 acIf the acoustic emission source O is located in region B;

[0063] d2 = dd2 bd If the acoustic emission source O is located in region C;

[0064] The shortest path for piezoelectric sensors located in other quadrants is calculated using the steps described above.

[0065] 5) Establish the objective function

[0066] Calculate the shortest path difference between the acoustic emission source O and the piezoelectric sensors at different locations:

[0067] Δd ij =d i -d j

[0068] The difference between the theoretical distance difference and the actual distance difference can be used to establish the objective function of gradient descent:

[0069]

[0070] Where: d ij Indicates the distance between different sensors;

[0071] v represents the propagation speed of the Lamb wave in the medium;

[0072] 6) Location solution

[0073] When performing localization calculations, the initial point (x0, y0) is set as the starting point of the iterative calculation. The region where the position is located is determined, the shortest path distance in the corresponding region is calculated, the position is updated using gradient descent, and it is determined whether the next position meets the termination condition. If the condition is not met, the region where the position is located and the corresponding objective function are re-evaluated and updated to continue the iterative calculation.

[0074] The calculation equation is as follows:

[0075]

[0076]

[0077] (x k+1 ,y k+1 (x) represents the coordinates of the next suspected collision point after the iteration. k ,y k H represents the coordinates of the currently identified collision point. k -1 And it represents the step size of the gradient descent method;

[0078] The following formula represents the degree of deviation between the current position and the ideal position, and is called the deviation coefficient:

[0079]

[0080] If k≤λ

[0081] Where: β represents the calculation factor, used to prevent overcorrection of the step size;

[0082] max||g|| is the maximum absolute value of all elements in the gradient vector;

[0083] This method is used to process the first few iterations, and then the step size is updated using the compensated step size method (k>λ).

[0084] S k =[x k ,y k ] T -[x k-1 ,y k-1 ] T

[0085]

[0086]

[0087] Repeat the iteration until the stopping criteria ||g|| < ε and ||E are met. n+1 -E n ||<ε, and the positioning result is obtained.

[0088] The advantages and beneficial effects of this invention are as follows:

[0089] 1. This invention enables real-time detection of collision locations in penetrating structures, ensuring the operational safety of on-orbit spacecraft. When space debris collides with a system, an elastic Lamb wave signal is generated at the collision location. Ultrasonic sensors at different locations are used to monitor the Lamb wave signal, and a specific detection algorithm can be used to calculate the collision location between the debris and the system. First, the propagation law of Lamb waves at the porthole structure was studied through finite element simulation. The results show that the Lamb wave will pass through the hole region along the shortest tangential path. Second, the signal region of the collision event is determined using a sliding energy window and an adaptive energy threshold method. The arrival time is captured using the AIC criterion, and the fastest propagation path model of the Lamb wave is established based on the geometric characteristics of the holed discontinuous structure. Finally, the objective equation of the gradient descent method is established based on the propagation path model, and the location calculation is performed.

[0090] 2. This invention solves the collision positioning problem when the propagation path of a spacecraft window structure is not a straight line.

[0091] 3. The piezoelectric sensor array required by this invention is simple and has strong applicability.

[0092] 4. The gradient descent method based on modified step size compensation and deviation coefficient of this invention accelerates the iteration speed and saves time costs without generating additional calculations. Attached Figure Description

[0093] Figure 1 Here is a schematic diagram of the positioning system structure of the present invention:

[0094] Figure 2 This is a schematic diagram of the Lamb wave dispersion curve of the present invention:

[0095] Figure 3 This is a schematic diagram of the Lamb wave signal of the present invention;

[0096] Figure 4 This is a schematic diagram showing the arrival times of the S0 modes of the various piezoelectric sensors of the present invention;

[0097] Figure 5 This is a schematic diagram of the region division and propagation path of the present invention;

[0098] Figure 6 This is a flowchart of the method of the present invention. Detailed Implementation

[0099] The present invention will be further described in detail below through specific embodiments. The following embodiments are merely descriptive and not limiting, and should not be used to limit the scope of protection of the present invention.

[0100] like Figure 1 As shown, the positioning system used in this invention includes piezoelectric sensors, signal amplifiers, signal acquisition cards, and a computer. The piezoelectric sensors are arranged at all four corners of the spacecraft's porthole structure, with each corner sensor equidistant from the center of the porthole structure. Each piezoelectric sensor is connected to the signal amplifier, which amplifies the elastic wave signal obtained by the piezoelectric sensors and transmits it to the signal acquisition card. The signal acquisition card then transmits the acquired information to the computer. The piezoelectric sensors are distributed in four quadrants with the center of the spacecraft's porthole structure as the origin (x0, y0). The coordinate position of the i-th piezoelectric sensor is (x0, y0). i y i );

[0101] like Figure 6 As shown, a method for locating sound sources in a spacecraft window structure based on gradient descent calculation is innovative in that the steps of the location method are as follows:

[0102] 1) Determine the signal range

[0103] A sliding energy window is used to process the 125kHz to 250kHz signal band received by each piezoelectric sensor. First, a signal window of a certain width is extracted from the starting time point of the Lamb wave signal, and the total energy of the signal in the window is taken as the energy value at the starting time point. Then, moving forward one step from that point, the signal of the next window width is extracted, and the total energy of the signal in the next window is taken as the energy value at the next point. The signal window moves along the time axis according to the step size, and the energy value of the signal in the window is taken as the energy value of the signal at the starting point of the window.

[0104]

[0105] Where: T(t) i () represents the absolute value of energy at each point;

[0106] T w Indicates the length of the energy window;

[0107] I(t) represents the total energy of the energy window corresponding to the current point;

[0108] Plot all energy points to obtain the time-energy diagram of the Lamb wave signal. Define a peak point with a normalized energy peak greater than 0.4 as a collision occurrence. The signals of the 5000 points before and 10000 points after the qualifying energy peak point are considered as the time range of the collision event. Figure 3 As shown;

[0109] 2) Calculate the time difference of arrival

[0110] The signal envelope within a selected range is determined using Hilbert transform to ascertain the signal's trend. The upper and lower envelopes are then processed using the AIC criterion and averaged to accurately obtain the start time of the S0 mode. To reduce errors caused by spurious signals and facilitate calculation, the absolute value of the obtained function value is calculated and multiplied by a scaling factor. The calculation formula is as follows:

[0111] AIC(t w )=K|t w ·ln(var(T(1,t w )))+(T w -t w -1)·ln(var(T(1+t w ,T w )))|

[0112]

[0113] T represents the selected range, which includes signal lengths from 1 to T. w ;

[0114] t wThis indicates that only signals within the selected range T are included in the calculation;

[0115] var represents the variance function;

[0116] var(T(1,t w )) indicates that only the sequence (1, t) within the selected range T is calculated. w The variance of the signal;

[0117] max(T(1,T w )) represents the maximum value of the signal within the selected range;

[0118] max(pks(T)) represents the maximum peak value of the AIC curve;

[0119] K represents the scaling factor;

[0120] The arrival times of the S0 mode Lamb wave signals from the four piezoelectric sensors were recorded, such as... Figure 4 As shown, the red line represents the Lamb wave signal within the selected range, the black dashed line represents the AIC curve, and the green line represents the start time of the S0 mode.

[0121] For the i-th sensor, the arrival time of the Lamb wave signal received in the debris collision event is t. i This corresponds to the time point of the first peak of the i-th AIC curve. Calculate the actual arrival difference between the i-th and j-th sensors and express it as Δt. ij ,

[0122] t i =pks(AIC i )

[0123] Δt ij =t i -t j

[0124] 3) Determine the starting position

[0125] By comparing the start times of signal reception by piezoelectric sensors in different quadrants, the piezoelectric sensor that receives the signal first is closest to the acoustic emission source. Therefore, we can approximate the initial coordinates of the acoustic emission source point as follows:

[0126] (x0,y0)=(C,C), if the first signal received by the sensor is located in the first quadrant;

[0127] (x0,y0)=(-C,C), if the first signal received by the sensor is located in the second quadrant;

[0128] (x0,y0)=(-C,-C), if the first signal received by the sensor is located in the third quadrant;

[0129] (x0,y0)=(C-,C), if the first signal received by the sensor is located in the fourth quadrant;

[0130] Where C represents a constant;

[0131] 4) Calculate the shortest path

[0132] To determine the shortest signal path, the aperture structure of each piezoelectric sensor is divided into different regions, such as... Figure 5 For sensors in the second quadrant, such as Figure 5 As shown in a), the tangent points between the piezoelectric sensor and the center of the spacecraft window structure are a and b. This plane is divided into six regions by the sensor tangent. The directional path from the piezoelectric sensor S to the acoustic emission source O is defined as a vector. vector and Starting from O and ending at a and b, Xob is calculated and The value obtained from the vector outer product:

[0133]

[0134] Xoa is calculated and recorded in the same way. If both Xoa and Xob are non-positive or non-negative, it indicates that the acoustic emission source O is located in region A or D. Figure 5 As shown in b);

[0135] Otherwise, the acoustic emission source O is located in another region;

[0136] Then, through calculation and Yob is obtained by taking the value of the vector dot product;

[0137]

[0138] Xoa is also calculated and recorded in the same way. If both Xoa and Xob are negative, it means that the acoustic emission source O is located in region F.

[0139] Otherwise, calculate the magnitude of the vector for further judgment:

[0140]

[0141] If Zob is negative, it indicates that the acoustic emission source O is located in region E;

[0142] Otherwise, if |x|≥|y|, then the acoustic emission source O is located in region B; if |x|<|y|, then it is located in region C.

[0143] Clearly, when the acoustic emission source O is located in regions A, D, E, and F, the shortest path d2 for the sensor to receive the signal is a straight line;

[0144]

[0145] When the acoustic emission source O is located in region B or C, the shortest path d2 is the sum of multiple path segments. The points of tangency between the acoustic emission source O and the center of the spacecraft window structure are c and d, as shown below. Figure 5 As shown in c), when the acoustic emission source O is located in region B or c, the shortest path for the sensor to receive the signal is dd2. ac or dd2 bd :

[0146]

[0147]

[0148] Where Lac and Lbd represent the arc lengths of two points on the circular hole;

[0149] By determining the region where the acoustic emission source O is located, the shortest path distance can be calculated:

[0150] If the acoustic emission source O is located in regions A, D, E, and F;

[0151] d2 = dd2 ac If the acoustic emission source O is located in region B;

[0152] d2 = dd2 bd If the acoustic emission source O is located in region C;

[0153] The shortest path for piezoelectric sensors located in other quadrants is calculated using the steps described above.

[0154] 5) Establish the objective function

[0155] Calculate the shortest path difference between the acoustic emission source O and the piezoelectric sensors at different locations:

[0156] Δd ij =d i -d j

[0157] The difference between the theoretical distance difference and the actual distance difference can be used to establish the objective function of gradient descent:

[0158]

[0159] Where: d ij Indicates the distance between different sensors;

[0160] v represents the propagation speed of the Lamb wave in the medium, such as... Figure 2 The group velocity curve shown is determined;

[0161] 6) Location solution

[0162] When performing localization calculations, the initial point (x0, y0) is set as the starting point of the iterative calculation. The region where the position is located is determined, the shortest path distance in the corresponding region is calculated, the position is updated using gradient descent, and it is determined whether the next position meets the termination condition. If the condition is not met, the region where the position is located and the corresponding objective function are re-evaluated and updated to continue the iterative calculation.

[0163] The calculation equation is as follows:

[0164]

[0165]

[0166] (x k+1 ,y k+1 (x) represents the coordinates of the next suspected collision point after the iteration. k ,y k H represents the coordinates of the currently identified collision point. k -1 And it represents the step size of the gradient descent method;

[0167] The following formula represents the degree of deviation between the current position and the ideal position, and is called the deviation coefficient:

[0168]

[0169] If k≤λ

[0170] Where: β represents the calculation factor, used to prevent overcorrection of the step size;

[0171] max||g|| is the maximum absolute value of all elements in the gradient vector;

[0172] This method is used to process the first few iterations, and then the step size is updated using the compensated step size method (k>λ).

[0173] S k =[x k ,y k ] T -[x k-1 ,y k-1 ] T

[0174]

[0175] If k > λ

[0176] Repeat the iteration until the stopping criteria ||g|| < ε and ||E are met. n+1 -E n ||<ε, and the positioning result is obtained.

[0177] Although embodiments and drawings of the present invention have been disclosed for illustrative purposes, those skilled in the art will understand that various substitutions, variations and modifications are possible without departing from the spirit and scope of the present invention and the appended claims. Therefore, the scope of the present invention is not limited to the contents disclosed in the embodiments and drawings.

Claims

1. A method for locating sound sources in a spacecraft window structure based on gradient descent calculation, characterized in that: The positioning system employed in this method includes piezoelectric sensors, signal amplifiers, a signal acquisition card, and a computer. The piezoelectric sensors are positioned at each of the four corners of the spacecraft's porthole structure, with each sensor at the corner equidistant from the center of the porthole structure. Each piezoelectric sensor is connected to the signal amplifier, which amplifies the elastic wave signals obtained by the piezoelectric sensors and transmits them to the signal acquisition card. The signal acquisition card then transmits the acquired information to the computer. The piezoelectric sensors are distributed across the four quadrants with the center of the spacecraft's porthole structure as the origin (x0, y0). The coordinate position of the i-th piezoelectric sensor is (x0, y0). i y i ); The steps of the positioning method are as follows: 1) Determine the signal range A sliding energy window is used to process the 125kHz to 250kHz signal band received by each piezoelectric sensor. First, a signal window of a certain width is extracted from the starting time point of the Lamb wave signal, and the total energy of the signal in the window is taken as the energy value at the starting time point. Then, moving forward one step from that point, the signal of the next window width is extracted, and the total energy of the signal in the next window is taken as the energy value at the next point. The signal window moves along the time axis according to the step size, and the energy value of the signal in the window is taken as the energy value of the signal at the starting point of the window. Where: T(t) i () represents the absolute value of energy at each point; T w Indicates the length of the energy window; I(t) represents the total energy of the energy window corresponding to the current point; Plot all energy value points to obtain the time-energy diagram of the Lamb wave signal, and identify the peak point with a normalized energy peak value greater than 0.4 as a collision occurrence. The signals of the 5000 points before and the 10000 points after the energy peak point that meet the condition are taken as the time range of the collision occurrence event. 2) Calculate the time difference of arrival The signal envelope within a selected range is determined using Hilbert transform to ascertain the signal's trend. The upper and lower envelopes are then processed using the AIC criterion and averaged to accurately obtain the start time of the S0 mode. To reduce errors caused by spurious signals and facilitate calculation, the absolute value of the obtained function value is calculated and multiplied by a scaling factor. The calculation formula is as follows: AIC(t w )=K|t w ·ln(var(T(1,t w )))+(T w -t w -1)·ln(var(T(1+t w ,T w )))| T represents the selected range, which includes signal lengths from 1 to T. w ; t w This indicates that only signals within the selected range T are included in the calculation; var represents the variance function; var(T(1,t w )) indicates that only the sequence (1, t) within the selected range T is calculated. w The variance of the signal; max(T(1,T w )) represents the maximum value of the signal within the selected range; max(pks(T)) represents the maximum peak value of the AIC curve; K represents the scaling factor; The arrival times of the S0 mode Lamb wave signals from the four piezoelectric sensors were recorded. The red line represents the Lamb wave signal within the selected range, the black dashed line represents the AIC curve, and the green line represents... S0 The start time of the pattern; For the i-th sensor, the arrival time of the Lamb wave signal received in the debris collision event is t. i This corresponds to the time point of the first peak of the i-th AIC curve. The actual arrival difference between the i-th and j-th sensors is calculated and expressed as Δt. ij , t i =pks(AIC i ) Δt ij =t i -t j 3) Determine the starting position By comparing the start times of signal reception by piezoelectric sensors in different quadrants, the piezoelectric sensor that receives the signal first is closest to the acoustic emission source. Therefore, we can approximate the initial coordinates of the acoustic emission source point as follows: (x0,y0)=(C,C), if the first signal received by the sensor is located in the first quadrant; (x0,y0)=(-C,C), if the first signal received by the sensor is located in the second quadrant; (x0,y0)=(-C,-C), if the first signal received by the sensor is located in the third quadrant; (x0,y0)=(C-,C), if the first signal received by the sensor is located in the fourth quadrant; Where C represents a constant; 4) Calculate the shortest path To determine the shortest signal path, the aperture structure of each piezoelectric sensor is divided into different regions. For the second quadrant sensor, the tangent points between the piezoelectric sensor and the center of the spacecraft window structure are a and b. This plane is divided into six regions by the sensor tangent. The directional path from the piezoelectric sensor S to the acoustic emission source O is defined as a vector. vector and Starting from O and ending at a and b, Xob is calculated... and The value obtained from the vector outer product: Xoa is also calculated and recorded in the same way. If both Xoa and Xob are non-positive or non-negative, it means that the acoustic emission source O is located in region A or D. Otherwise, the acoustic emission source O is located in another region; Then, through calculation and Yob is obtained by taking the value of the vector dot product; Xoa is also calculated and recorded in the same way. If both Xoa and Xob are negative, it means that the acoustic emission source O is located in region F. Otherwise, calculate the magnitude of the vector for further judgment: If Zob is negative, it indicates that the acoustic emission source O is located in region E; Otherwise, if |x|≥|y|, then the acoustic emission source O is located in region B; if |x|<|y|, then it is located in region C. Clearly, when the acoustic emission source O is located in regions A, D, E, and F, the shortest path d2 for the sensor to receive the signal is a straight line; When the acoustic emission source O is located in region B or C, the shortest path d2 is the sum of multiple path segments. The points of tangency between the acoustic emission source O and the center of the spacecraft window structure are c and d. When the acoustic emission source O is located in region B or c, the shortest path for the sensor to receive the signal is dd2. ac or dd2 bd : Where Lac and Lbd represent the arc lengths of two points on the circular hole; By determining the region where the acoustic emission source O is located, the shortest path distance can be calculated: If the acoustic emission source O is located in regions A, D, E, and F; d2 = dd2 ac If the acoustic emission source O is located in region B; d2 = dd2 bd If the acoustic emission source O is located in region C; The shortest path for piezoelectric sensors located in other quadrants is calculated using the steps described above. 5) Establish the objective function Calculate the shortest path difference between the acoustic emission source O and the piezoelectric sensors at different locations: Δd ij =d i -d j The difference between the theoretical distance difference and the actual distance difference can be used to establish the objective function of gradient descent: Where: d ij Indicates the distance between different sensors; v represents the propagation speed of the Lamb wave in the medium; 6) Location solution When performing localization calculations, the initial point (x0, y0) is set as the starting point of the iterative calculation. The region where the position is located is determined, the shortest path distance in the corresponding region is calculated, the position is updated using gradient descent, and it is determined whether the next position meets the termination condition. If the condition is not met, the region where the position is located and the corresponding objective function are re-evaluated and updated to continue the iterative calculation. The calculation equation is as follows: [x k+1 ,and k+1 ]=[x k ,and k ]-H k -1 ▽f(x k ,and k ) (x k+1 ,y k+1 (x) represents the coordinates of the next suspected collision point after the iteration. k ,y k H represents the coordinates of the currently identified collision point. k -1 And it represents the step size of the gradient descent method; The following formula represents the degree of deviation between the current position and the ideal position, and is called the deviation coefficient: If k≤λ Where: β represents the calculation factor, used to prevent overcorrection of the step size; max||g|| is the maximum absolute value of all elements in the gradient vector; This method is used to process the first few iterations, and then the step size is updated using the compensated step size method (k>λ). S k =[x k ,and k ] T -[x k-1 ,and k-1 ] T z k =[▽f(x k ,and k )-▽f(x k-1 ,and k-1 )] T If k > λ Repeat the iteration until the stopping criteria ||g|| < ε and ||E are met. n+1 -E n ||<ε, and the positioning result is obtained.

Images