Random load spectrum accelerated editing method based on multi-domain information fusion

Abstract

The invention relates to the technical field of mechanical structure durability tests, and discloses a random load spectrum accelerated editing method based on multi-domain information fusion, comprising the following steps: S1, collecting an original random load spectrum of a component; s2, extracting related characteristics of energy distribution, damage distribution and peak value distribution; s3, fusing the energy distribution, the damage distribution and the peak value distribution by using a DS evidence theory to obtain a trust function reserved for the load point after multi-domain information fusion; s4, determining a threshold value of a trust function reserved by the load point after multi-domain information fusion; s5, according to the determined threshold value, identifying and extracting a signal segment with large damage contribution in the load spectrum, comparing whether the statistical parameter error and the damage retention amount before and after editing meet set constraint conditions, and if yes, executing S6; if not, returning to S4; and S6, completing the accelerated durability editing work of the load spectrum. According to the scheme, the problem of accelerated editing of the existing load spectrum is solved, and efficient accelerated editing and reduction of the service load are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical structure durability tests, and particularly relates to a random load spectrum acceleration editing method based on multi-domain information fusion. Background Art

[0002] In the process of mechanical structure durability development, the full-life cycle load spectrum is of great significance. It can accurately display the change of the load borne by the structure during service over time from multiple dimensions such as time history, frequency characteristics, amplitude fluctuation, energy and damage distribution, and statistical parameters, providing key basic data for the mechanical structure durability development and serving as an important basis for the smooth progress of subsequent work.

[0003] However, in the mechanical structure durability bench verification stage, there are many difficulties in completely reproducing the actual service load spectrum. During actual service, the load spectrum borne by the mechanical structure is complex and lengthy, containing a large number of load cycles, and the contribution degrees of these cycles to the structure damage are different. Some cycles have a greater impact on the structure damage, while some have a relatively smaller impact. If all load cycles are completely reproduced in the durability test, it will not only consume a large amount of R & D time and resources, but also greatly extend the test cycle, which is unacceptable for industries pursuing high-efficiency R & D. Therefore, how to accelerate the editing and reduction processing of the service load to effectively shorten the test cycle has become a key problem that the entire industry urgently needs to solve.

[0004] The core of load spectrum acceleration editing lies in accurately identifying the cycles that contribute greatly to the structure damage, but the traditional methods have significant drawbacks. Traditional methods often analyze load cycles only from a single damage domain or energy domain. If only starting from the damage domain and removing the cycles with small contribution to the structure damage according to the damage degree, the important information in the energy domain and statistical domain will be ignored, resulting in the edited load spectrum being unable to comprehensively and accurately reflect the actual service condition of the mechanical structure. On the contrary, if only considering from the energy domain, although the energy distribution of load cycles can be concerned, the key content in the damage domain is easily ignored, ultimately resulting in deviation in structure damage prediction. Summary of the Invention

[0005] The present invention aims to provide a random load spectrum acceleration editing method based on multi-domain information fusion to solve the problems that in the existing mechanical structure durability development, the traditional load spectrum acceleration editing method cannot comprehensively and accurately reflect the actual service condition and there is deviation in damage prediction, and to achieve efficient acceleration editing and reduction of the service load, effectively shortening the test cycle.

[0006] To achieve the above object, the present invention adopts the following technical solutions: A random load spectrum acceleration editing method based on multi-domain information fusion, comprising the following steps:

[0007] S1: Collect the original random load spectrum of the component;

[0008] S2: Extract the relevant characteristics of energy distribution, damage distribution, and peak distribution from the original random load spectrum;

[0009] S3: Use the DS evidence theory to perform multi-domain information fusion on the energy distribution, damage distribution, and peak distribution of the original random load spectrum, and obtain the trust function retained for the load points after multi-domain information fusion;

[0010] S4: Determine the threshold of the trust function retained for the load points after multi-domain information fusion;

[0011] S5: According to the determined threshold, identify and extract the signal segments with large damage contribution in the load spectrum, and compare whether the statistical parameter errors and damage retention amounts before and after editing meet the set constraint conditions. If the requirements are met, execute S6; if the requirements are not met, return to S4 to re-determine the threshold;

[0012] S6: Complete the accelerated durability editing work of the load spectrum.

[0013] Preferably, in S2, the extraction method of energy distribution is as follows: Use wavelet transform in time-frequency analysis technology to process the original random load spectrum, obtain its frequency information and the distribution of each frequency on the time axis, and use this as the energy distribution of the original random load spectrum.

[0014] Preferably, in S2, the extraction of energy distribution includes the following steps:

[0015] Step 1: Determine the mother wavelet function;

[0016] Step 2: Calculate the wavelet coefficients;

[0017] Step 3: Reconstruct the wavelet components;

[0018] Step 4: Reconstruct the original random load spectrum signal;

[0019] Step 5: Calculate the cumulative power spectral density.

[0020] Preferably, in S2, the extraction of damage distribution includes the following steps:

[0021] The first step: Determine the damage index;

[0022] The second step, perform coordinate transformation to calculate the stress value;

[0023] The third step, preprocess and perform rainflow cycle analysis;

[0024] The fourth step, calculate the cyclic damage contribution;

[0025] The fifth step, determine the damage distribution of the original random load spectrum.

[0026] Preferably, in S2, the extraction of the peak distribution includes the following steps:

[0027] Step 1: Mean removal processing;

[0028] When calculating the peaks of the original random load spectrum, first perform a mean removal operation on the original random load spectrum. The formula is

[0029] L′(t) = L(t) - mean{L(t)}

[0030] In the formula, L ′ (t) represents the load spectrum after mean removal;

[0031] mean{L(t)} represents the mean of the original random load spectrum;

[0032] Step 2: Take the absolute value to convert the valleys into peaks.

[0033] Preferably, in S3, the acquisition of the belief function includes the following steps:

[0034] S3.1: Introduce the DS theory and apply it to load spectrum editing;

[0035] S3.2: Define the belief function, likelihood function, and belief interval;

[0036] S3.3: DS combination rule;

[0037] S3.4: Expression of the basic probability assignment of each distribution evidence;

[0038] S3.5 Evidence combination and acquisition of belief function values.

[0039] Preferably, in S3.4, the basic probability assignment of load points retained based on energy distribution evidence expression:

[0040]

[0041] Among them, APSD max and APSD min respectively represent the maximum and minimum values in the energy distribution, and k represents the serial number of the load point.

[0042] Preferably, in S3.4, the basic probability assignment of load points retained based on damage distribution evidence expression

[0043]

[0044] Among them, D max and D min respectively represent the maximum and minimum values in the damage distribution, and k represents the serial number of the load point.

[0045] Preferably, in S3.4, the basic probability assignment of load point retention based on peak distribution evidence Expression

[0046]

[0047] where L max and L min represent the maximum and minimum values in the peak distribution respectively, and k represents the serial number of the load point.

[0048] Preferably, in S4, a genetic algorithm is used to optimize and solve the threshold. Taking the load spectrum compression ratio c(e) of the load spectrum before and after editing as the objective function, and the belief function threshold e as the design variable, and taking the errors of statistical parameters (root mean square and kurtosis coefficient) and the retained amount of damage between the load spectra before and after compression as the constraints, the specific threshold optimization model is as follows:

[0049]

[0050] where L(t) and L e (t) are the time series of the load spectra before and after editing respectively; K and K e are the kurtosis coefficients of the load spectra before and after editing respectively; RMS and RMS e are the root mean square values of the load spectra before and after editing respectively; D and D e are the damage values of the load spectra before and after editing respectively.

[0051] Compared with the prior art, the beneficial effects of this solution are as follows:

[0052] (1) Comprehensively and accurately reflect the actual service condition: Traditional load spectrum acceleration editing methods are often limited to analyzing load cycles in a single damage domain or energy domain, which results in a large amount of key information being missing and it is difficult to accurately reflect the full picture of the loads borne by mechanical structures during actual service. This solution breaks through this limitation and innovatively combines multi-domain information such as the damage domain, energy domain, and peak distribution to deeply analyze the load characteristics of mechanical structures during actual service. By extracting relevant characteristics of energy distribution, damage distribution, and peak distribution, it ensures that the edited load spectrum is highly consistent with the original load spectrum in terms of damage prediction and energy distribution, successfully avoiding the problem of distorted reflection of the actual service condition caused by one-sided information.

[0053] (2) Reduce the bias in damage prediction: The traditional load spectrum accelerated editing method only analyzes the load cycle from a single domain, which easily ignores other key information, leading to bias in structural damage prediction. This scheme uses the evidence theory (DS theory), which has significant advantages in dealing with uncertainty and incomplete information. It can integrate multi-domain information and construct a trust function for load point retention. By reasonably determining the threshold, accurately identifying and extracting signal segments with large damage contributions, and comprehensively considering the comprehensive impact of various factors on structural damage, the bias in damage prediction can be effectively reduced, providing a solid and reliable basis for the durability design and life assessment of mechanical structures.

[0054] (3) Efficient and accelerated editing and reduction, shortening the test cycle: The load spectrum in actual service is extremely complex and lengthy. If it is fully reproduced in the durability test, it will consume a lot of time and resources. This solution can accurately identify the cycles that contribute most to structural damage and decisively eliminate the cycles that contribute less, thereby achieving efficient and accelerated editing and reduction of service loads. While ensuring the effectiveness of the edited load spectrum, it greatly reduces the number of load cycles that need to be simulated in the durability test, thereby significantly shortening the test cycle, effectively improving R&D efficiency, and reducing R&D costs. It has immeasurable value for fields such as automobiles, aerospace, and engineering machinery that have extremely high requirements for R&D efficiency.

[0055] (4) Promote academic research and technological development: This solution deeply involves cross-disciplinary research in multiple fields such as multi-dimensional information fusion, damage prediction, and energy distribution. In-depth exploration of these fields can not only open up new development paths for load spectrum editing technology, but also provide innovative theories and methods for the durability design of mechanical structures. It has far-reaching academic research value and lays a solid foundation for the sustainable development of related fields. With the strong support of information fusion algorithms, this solution realizes the precise editing of load spectra, closely combines theoretical research with engineering practice, and effectively promotes new breakthroughs in the durability development of mechanical structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of load spectrum energy distribution calculation of the present invention;

[0057] Figure 2 It is a schematic diagram of the calculation of load spectrum damage distribution of the present invention;

[0058] Figure 3 It is a schematic diagram of load spectrum peak value calculation of the present invention;

[0059] Figure 4 The present invention is a flowchart of a random load spectrum accelerated editing method based on multi-domain information fusion. DETAILED DESCRIPTION

[0060] The following is further described in detail through specific implementation methods:

[0061] Embodiment

[0062] See Figures 1 to 4 , a random load spectrum acceleration editing method based on multi-domain information fusion, comprising the following steps:

[0063] S1: Collect the original random load spectrum of the component, which is derived from the measured load signal of the component;

[0064] S2: Extract the relevant characteristics of energy distribution, damage distribution and peak distribution from the original random load spectrum;

[0065] The extraction method of energy distribution is as follows: Use wavelet transform in time-frequency analysis technology to process the original random load spectrum, obtain its frequency information and the distribution of each frequency on the time axis, and use this as the energy distribution of the original random load spectrum. The specific extraction steps are as follows:

[0066] Step 1: Determine the mother wavelet function:

[0067]

[0068] In the formula, represents the wavelet coefficient after transformation, where s and k are the independent variables of the function;

[0069] L(t) is the original random load spectrum L(t);

[0070] a0 (a0>0) represents the scale factor;

[0071] s represents the number of layers into which the original random load spectrum is decomposed by wavelet transform;

[0072] represents the mother wavelet function, which is the basic function of the entire wavelet analysis;

[0073] is the operation of stretching and translating the mother wavelet function, is to stretch the time t according to the scale factor a0 and the decomposition layer number s; k in kτ0 is a discrete variable, and τ0 (τ0≠0) is a parameter reflecting the translation step length, which is used to control the translation of the mother wavelet function on the time axis;

[0074] In practical engineering applications, binary wavelet transform is usually adopted, and a0 = 2 is selected as the scale factor, and the time translation step length τ0 = 1;

[0075] Step 2: Calculate the wavelet coefficient;

[0076] Perform an inner product operation on the original random load spectrum L(t) and the mother wavelet function to obtain the high-frequency wavelet coefficient and the low-frequency wavelet coefficient w s,k ;

[0077] w s,k = <L(t), φ s,k (t)> = W φ (s,k)

[0078] Wherein, refers to the mother wavelet function at the scale parameter s and the translation parameter k;

[0079] Step Three: Reconstruct the wavelet components;

[0080] According to the obtained wavelet coefficients, reconstruct the wavelet component D s (t) of the original random load spectrum;

[0081]

[0082] Step Four: Reconstruct the original random load spectrum signal;

[0083] Sum up all the wavelet components to reconstruct the original random load spectrum signal L(t);

[0084]

[0085] Step Five: Calculate the cumulative power spectral density;

[0086] In the wavelet transform process, the information carried by signals at different frequencies is different. The low-frequency wavelet components mainly contain the mean information of the signal, and the high-frequency wavelet components are used to characterize the amplitude change of the load spectrum. Square and sum the high-frequency wavelet signals at all scales to obtain the cumulative power spectral density APSD(t):

[0087]

[0088] The extraction of the damage distribution is mainly through the following steps:

[0089] Step One: Determine the damage index;

[0090] In the fatigue damage study, select the combination of the maximum principal stress and the shear stress on the critical plane as the damage parameter, and this combination essentially represents the equivalent stress on the critical plane and use it as the damage index, and the expression is as follows:

[0091]

[0092] Wherein, B and K are constants;

[0093] represents the shear stress on the critical plane at the fatigue hot spot;

[0094] represents the normal stress on the critical plane at the fatigue hot spot;

[0095] Step 2: Calculate the stress value through coordinate transformation;

[0096] According to different damage criteria, the relevant parameters take values between 0, 1, and constants. By performing coordinate transformation on the spatial stress state σi j (t) at the fatigue hot spot, the normal stress on the critical plane at the fatigue hot spot is obtained and the shear stress

[0097]

[0098] where i, j = x, y, z respectively correspond to the three axes of x, y, and z, and the coordinate transformation matrix are respectively the cosines of the angles between the normal vector and the X, Y, Z axes; the coordinate transformation matrix are respectively the cosines of the angles between the shear vector and the X, Y, Z axes;

[0099] Step 3: Preprocessing and rainflow cycle analysis;

[0100] After preprocessing the equivalent stress of the critical plane, peak-valley value extraction and rainflow cycle analysis are carried out. After obtaining the rainflow cycle analysis results of the equivalent stress time history, with the help of the multiaxial damage model and Miner's damage criterion, the damage contribution value of each cycle is calculated;

[0101]

[0102] where the left side of the above equation represents the damage parameter, and the right side represents the multiaxial fatigue damage model;

[0103] Since different damage parameters will lead to different expressions, so:

[0104] When the damage parameter is the stress amplitude,

[0105] f(N f ) = σ′ f (2N f ) b

[0106] In the formula, represents the number of cycles corresponding to the equivalent stress ;

[0107] σ′ f represents the fatigue strength coefficient;

[0108] b represents the fatigue strength index;

[0109] When the damage parameter is the strain amplitude,

[0110]

[0111] where, represents the number of cycles corresponding to the equivalent stress ;

[0112] σ′ f represents the fatigue strength coefficient;

[0113] b represents the fatigue strength exponent;

[0114] E is the Young's modulus;

[0115] ε′ f represents the shear strain;

[0116] c represents the shear strain fatigue strength exponent;

[0117] When the damage parameter is the combination of the maximum principal strain and the shear strain energy,

[0118]

[0119] where, represents the number of cycles corresponding to the equivalent stress ;

[0120] σ′ f represents the fatigue strength coefficient;

[0121] b represents the fatigue strength exponent;

[0122] E is the Young's modulus;

[0123] ε′ f represents the shear strain;

[0124] c represents the shear strain fatigue strength exponent;

[0125] Step 4, calculate the cyclic damage contribution;

[0126] According to the above multiaxial fatigue damage model, calculate the cyclic damage contribution of each cycle:

[0127]

[0128] Step 5, determine the damage distribution of the original random load spectrum;

[0129] Evenly distribute the calculated cyclic damage contribution to each load point that makes up the cycle, that is, the cyclic damage contribution of each load point is D i / 2. After performing rainflow counting on the entire load spectrum, the damage distribution of the entire load spectrum can be calculated.

[0130] The peak distribution of the load spectrum can reflect the distribution of each peak (maximum value) in the load spectrum. The specific extraction method is as follows:

[0131] Step 1: Mean removal processing;

[0132] When calculating the peaks of the original random load spectrum, first perform a mean removal operation on the original random load spectrum. By removing the mean of the original random load spectrum, the average trend component in the signal is eliminated, facilitating more accurate analysis of the peak characteristics in the subsequent steps.

[0133] L′(t) = L(t) - mean{L(t)}

[0134] where L ′ (t) represents the load spectrum after mean removal;

[0135] mean{L(t)} represents the mean of the original random load spectrum;

[0136] Step 2: Absolute value conversion;

[0137] After completing the mean removal processing, take the absolute value |L′(t)| of the load spectrum. This operation can convert the valleys in the load spectrum into peaks, enabling the analysis to comprehensively cover all extreme value situations and providing conditions for accurately obtaining the peak distribution of the load spectrum;

[0138] S3: Use the DS evidence theory to perform multi-domain information fusion on the energy distribution, damage distribution, and peak distribution of the original random load spectrum to obtain the belief function retained for the load points after multi-domain information fusion;

[0139] S3.1: Introduce the DS theory into load spectrum editing;

[0140] Apply the DS theory to load spectrum editing. The energy distribution, damage distribution, and peak distribution of the original random load spectrum are used as the evidence sources for load spectrum editing. The retention or deletion of each load point (denoted as event C1 and event C2 respectively) constitutes the frame of discernment in the evidence theory;

[0141] Denote it as Θ = {C1, C2}. Each event in the frame of discernment is mutually independent and is called an element of Θ. 2 Θ is the power set of Θ. For any set A, it satisfies

[0142]

[0143] where represents the empty set, and the mapping m is the basic probability assignment function on 2 Θ and is also called the mass function. m(A) represents the basic probability assignment of set A;

[0144] S3.2: Define the belief function, plausibility function, and belief interval;

[0145] Define the belief function Bel, plausibility function Pl, and belief interval as follows

[0146]

[0147] S3.3: DS combination rule;

[0148] When there are n mass functions under the identification framework (i.e., there are multiple pieces of evidence), the DS combination rule is shown in the formula:

[0149]

[0150] where, K is the normalization constant, represents the symbol for the combination of each mass function;

[0151] S3.4: Basic probability assignment expression for evidence of each distribution

[0152] S3.4.1: Basic probability assignment for load point retention based on energy distribution evidence Expression:

[0153]

[0154] where, APSD max and APSD min represent the maximum and minimum values in the energy distribution respectively, and k represents the serial number of the load point;

[0155] S3.4.2: Basic probability assignment for load point retention based on damage distribution evidence Expression

[0156]

[0157] where, D max and D min represent the maximum and minimum values in the damage distribution respectively, and k represents the serial number of the load point;

[0158] S3.4.3: Basic probability assignment for load point retention based on peak distribution evidence Expression

[0159]

[0160] where, L max and L min represent the maximum and minimum values in the peak distribution respectively, and k represents the serial number of the load point.

[0161] S3.5 Evidence synthesis and obtaining trust function values;

[0162] Use the basic formula of evidence theory to synthesize the evidence sources and obtain the trust function value Bel of the retained load points after fusion k (C1), make a decision on retaining or deleting the load points by using the trust function value;

[0163]

[0164] S4: Determine the threshold of the trust function for the retained load points after multi-domain information fusion;

[0165] Use the genetic algorithm to optimize and solve the threshold. Take the load spectrum compression ratio c(e) of the load spectrum before and after editing as the objective function, take the trust function threshold e as the design variable, and take the errors of the statistical parameters (root mean square and kurtosis coefficient) and the retained amount of damage between the load spectra before and after compression as the constraint conditions. The specific threshold optimization model is as follows:

[0166]

[0167] Among them, L(t) and L e (t) are the time series of the load spectra before and after editing respectively; K and K e are the kurtosis coefficients of the load spectra before and after editing respectively; RMS and RMS e are the root mean square values of the load spectra before and after editing respectively; D and D e are the damage values of the load spectra before and after editing respectively;

[0168] S5: Identify and extract the signal segments with large damage contributions in the load spectrum;

[0169] According to the determined threshold, identify and extract the signal segments with large damage contributions in the load spectrum, and compare whether the errors of the statistical parameters (root mean square and kurtosis coefficient) and the retained amount of damage before and after editing meet the set constraint conditions. If the requirements are met, execute S6; if the requirements are not met, return to S4 to re-determine the threshold.

[0170] S6: Complete the accelerated durability editing work of the load spectrum.

[0171] The above are only the embodiments of the present invention. Specific technical solutions and / or common knowledge such as characteristics well known in the solution are not described in detail here. It should be noted that for those skilled in the art, without departing from the technical solution of the present invention, several deformations and improvements can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicability of the patent. The protection scope required by this application should be subject to the content of its claims, and the specific implementation manners and the like recorded in the specification can be used to interpret the content of the claims.

Claims

1. A random load spectrum acceleration editing method based on multi-domain information fusion, characterized in that: It includes the following steps: S1: Collect the original random load spectrum of the component; S2: Extract the relevant characteristics of energy distribution, damage distribution, and peak distribution from the original random load spectrum; S3: Use the DS evidence theory to perform multi-domain information fusion on the energy distribution, damage distribution, and peak distribution of the original random load spectrum, and obtain the belief function retained for the load points after multi-domain information fusion; S4: Determine the threshold of the belief function retained for the load points after multi-domain information fusion; S5: According to the determined threshold, identify and extract the signal segments with large damage contribution in the load spectrum, and compare whether the statistical parameter errors and damage retention amounts before and after editing meet the set constraint conditions. If the requirements are met, execute S6; if the requirements are not met, return to S4 to re-determine the threshold; S6: Complete the accelerated durability editing work of the load spectrum.

2. The accelerated editing method for random load spectra based on multi-domain information fusion according to claim 1, wherein: In S2, the extraction method of energy distribution is as follows: Use wavelet transform in time-frequency analysis technology to process the original random load spectrum, obtain its frequency information and the distribution status of each frequency on the time axis, and use this as the energy distribution of the original random load spectrum.

3. A random load spectrum acceleration editing method based on multi-domain information fusion according to claim 2, characterized in that: In S2, the extraction of energy distribution includes the following steps: Step 1: Determine the mother wavelet function; Step 2: Calculate the wavelet coefficients; Step 3: Reconstruct the wavelet components; Step 4: Reconstruct the original random load spectrum signal; Step 5: Calculate the cumulative power spectral density.

4. A random load spectrum acceleration editing method based on multi-domain information fusion according to claim 3, characterized in that: In S2, the extraction of damage distribution includes the following steps: The first step: Determine the damage index; The second step, perform coordinate transformation to calculate the stress value; The third step, preprocess and perform rainflow cycle analysis; The fourth step, calculate the cyclic damage contribution amount; The fifth step, determine the damage distribution of the original random load spectrum.

5. A random load spectrum acceleration editing method based on multi-domain information fusion according to claim 4, characterized in that: In S2, the extraction of peak distribution includes the following steps: Step 1: Perform mean removal processing; When calculating the peaks of the original random load spectrum, first perform mean removal operation on the original random load spectrum, and the formula is L′(t) = L(t) - mean{L(t)} In the formula, L′(t) represents the load spectrum after mean removal; mean{L(t)} represents the mean of the original random load spectrum; Step 2: Take the absolute value to convert the valleys into peaks.

6. The accelerated editing method for random load spectra based on multi-domain information fusion according to claim 5, wherein: In S3, the acquisition of the belief function includes the following steps: S3.1: Introduce the DS theory and apply it to load spectrum editing; S3.2: Define the belief function, likelihood function, and belief interval; S3.3: DS combination rule; S3.4: Express the basic probability assignment of each distribution evidence; S3.5 Evidence combination and acquisition of belief function values.

7. A random load spectrum acceleration editing method based on multi-domain information fusion according to claim 6, characterized in that: In S3.4, the basic probability assignment of the load point based on the energy distribution evidence is retained is expressed as: Among them, APSD max and APSD min respectively represent the maximum and minimum values in the energy distribution, and k represents the serial number of the load point.

8. A method for accelerating the editing of a random load spectrum based on multi-domain information fusion according to claim 7, characterized in that: In S3.4, the basic probability assignment of load point retention based on damage distribution evidence Expression Among them, D max and D min represent the maximum and minimum values in the damage distribution respectively, and k represents the serial number of the load point.

9. A random load spectrum acceleration editing method based on multi-domain information fusion according to claim 8, characterized in that: In S3.4, the basic probability assignment of load point retention based on peak distribution evidence Expression Among them, L max and L min represent the maximum and minimum values in the peak distribution respectively, and k represents the serial number of the load point.

10. A random load spectrum acceleration editing method based on multi-domain information fusion according to claim 9, characterized in that: In S4, use the genetic algorithm to optimize and solve the threshold. Take the load spectrum compression ratio c(e) of the load spectrum before and after editing as the objective function, take the belief function threshold e as the design variable, and take the statistical parameter (root mean square and kurtosis coefficient) errors and damage retention amounts between the load spectra before and after compression as the constraint conditions. The specific threshold optimization model is as follows: Among them, L(t) and L e (t) are the time series of the load spectra before and after editing, respectively; K and K e are the kurtosis coefficients of the load spectra before and after editing, respectively; RMS and RMS e are the root mean square values of the load spectra before and after editing, respectively; D and D e are the damage values of the load spectra before and after editing, respectively.

Images