Test spectrum compilation method based on particle swarm optimization

Abstract

The invention discloses a test spectrum compilation method based on a particle swarm algorithm, and relates to the technical field of aero-engine test spectrum compilation, and the method comprises the steps: carrying out the coordinate transformation of a random multi-axis load, and converting a stress-strain spectrum of the random multi-axis load into a stress history and a strain history on a new plane; carrying out peak-valley value extraction on the random multi-axis load; performing cycle number statistics on the random multi-axis load, and converting a stress history and a strain history on the new plane into stress-strain cycles; carrying out damage calculation on the random multi-axis load, taking a plane where the maximum damage of the random multi-axis load is located as a critical plane, and extracting a stress cycle and a strain cycle on the critical plane; and according to the cycle number statistical result and the damage calculation result, reversely deducing test spectrum parameters based on a particle swarm algorithm. Damage equivalence before and after test spectrum compilation can be ensured, and test spectrum parameter search efficiency is improved.

Description

Technical Field

[0001] The invention relates to the technical field of aero-engine test spectrum compilation, and in particular to a test spectrum compilation method based on a particle swarm algorithm. Background Art

[0002] In recent decades, the research on the compilation of single-parameter fatigue test load spectra has made great progress and has been widely used in aerospace, vehicles, engineering machinery, ships and other fields. However, in actual engineering applications, most components are subjected to random multi-parameter loads for a long time, which can easily cause multi-axial fatigue damage and then fail. Therefore, single-parameter fatigue test spectra are no longer suitable for fatigue assessment of these components, and it is necessary to develop multi-parameter fatigue load test spectrum compilation technology.

[0003] The current multi-parameter fatigue load test spectrum compilation technology often cannot ensure the consistency of damage before and after the test spectrum compilation. The matching principle used in the compilation process lacks sufficient theoretical basis, which may affect the accuracy and reliability of the test spectrum compilation. In addition, the existing technology usually cannot fully consider the phase difference between different loads, which leads to deviations between the test conditions and the actual working conditions. The current multi-parameter fatigue load test spectrum compilation method is relatively simple, lacks diversified technical means and strategies, and is difficult to meet the test needs of different components.

[0004] For single-point and multi-axial stress failure modes under multi-parameter loads, the existing method uses the multi-axis rain flow counting method to cyclically count and group the main damage parameters on the critical surface, and then performs a multi-parameter cyclic load search on each group. The cyclic loads searched in each group are connected in a certain order to obtain a multi-parameter fatigue test spectrum. It uses a traversal method to search for multi-parameter cyclic loads, which is relatively inefficient and cannot clarify the specific grouping method.

[0005] Therefore, how to ensure the damage equivalence before and after the test spectrum is compiled and improve the search efficiency of the test spectrum parameters has become an urgent problem to be solved in the field of aero-engine test spectrum compilation technology. Summary of the invention

[0006] In order to solve the above problems, the present invention proposes a test spectrum compilation method based on particle swarm algorithm, which can ensure the damage equivalence before and after the test spectrum compilation and improve the efficiency of test spectrum parameter search.

[0007] In order to achieve the above object, the present invention is implemented by the following technical solutions:

[0008] The present invention is a test spectrum compilation method based on particle swarm algorithm, the method comprising:

[0009] Performing coordinate transformation on the random multiaxial load, and transforming the stress-strain spectrum of the random multiaxial load into a stress history and a strain history on a new plane;

[0010] Extracting peak and valley values ​​of the random multiaxial load;

[0011] Performing cycle statistics on the random multiaxial load, and converting the stress history and strain history on the new plane into stress-strain cycles;

[0012] Performing damage calculation on the random multiaxial load, taking the plane where the maximum damage of the random multiaxial load is located as the critical plane, and extracting the stress cycle and strain cycle on the critical plane;

[0013] According to the cycle number statistical results and damage calculation results, the test spectrum parameters are inferred based on the particle swarm algorithm.

[0014] A further improvement of the present invention is that the coordinate transformation of the random multi-axial load is performed to transform the stress-strain spectrum of the random multi-axial load into the stress history and strain history on a new plane, specifically including:

[0015] Based on the critical plane method, the angles θ and ψ of the plane where the maximum damage of the random multiaxial load is located are determined, the angle θ increases from 0° to 360° at intervals of 1°, and the angle ψ increases from 0° to 180° at intervals of 1°, and any plane in the three-dimensional space under the Cartesian coordinate system is generated by rotating the angles θ and ψ, and the stress-strain spectrum of the random multiaxial load is transformed into a stress history and a strain history on the new plane;

[0016] The formula for the coordinate transformation is:

[0017]

[0018]

[0019]

[0020] In the formula, σ represents the stress tensor of a point in the original random stress spectrum; σ x , τ xy , τ xz , τ xy , σ y , τ yz , τ xz , τ yz , σ z represents the stress component corresponding to the point, and σ x , σ y , σ z represents normal stress. The subscript of normal stress represents the normal direction of the plane of action of normal stress.xy , τ xz , τ yz represents shear stress. The first subscript of shear stress represents the normal direction of the plane of action of shear stress, and the second subscript represents the coordinate direction parallel to the shear stress. In addition, on the positive side, that is, the normal direction on the action plane is along the positive direction of the coordinate axis, the stress along the positive direction of the coordinate axis is positive; on the negative side, that is, the normal direction on the action plane is along the negative direction of the coordinate axis, the stress along the negative direction of the coordinate axis is positive; ε represents the strain tensor of a point in the original random strain spectrum; ε x , 1 / 2γ xy , 1 / 2γ xz , 1 / 2γ xy , ε y , 1 / 2γ yz , 1 / 2γ xz , 1 / 2γ yz , ε z represents the strain component corresponding to the point, and ε x , ε y , ε z represents the normal strain. The subscript of the normal strain represents the normal direction of the plane of action of the normal strain. 1 / 2γ xy , 1 / 2γ xz , 1 / 2γ yz represents shear strain. The first subscript of shear strain represents the normal direction of the plane of action of shear strain. The second subscript represents the coordinate direction parallel to the shear strain. In addition, on the positive side, that is, the normal direction on the action plane is along the positive direction of the coordinate axis, the strain along the positive direction of the coordinate axis is positive; on the negative side, that is, the normal direction on the action plane is along the negative direction of the coordinate axis, the strain along the negative direction of the coordinate axis is positive; M represents the transformation matrix; M T represents the transpose of the transformation matrix; σ' represents the stress history on the new plane; ε' represents the strain history on the new plane; θ and ψ represent the angles of the plane where the maximum damage of random multiaxial loading is located.

[0021] A further improvement of the present invention is that the peak-valley value extraction of the random multi-axial load specifically includes:

[0022] Obtain the main channel and auxiliary channel load histories of the stress component history and the strain component history. If the corresponding points in the load histories of the main channel and the auxiliary channel have peak and valley values ​​in the two load histories, retain the corresponding points; if the corresponding points in the load histories of the main channel and the auxiliary channel do not have peak and valley values ​​in the two load histories, remove the corresponding points. That is, the removal of the corresponding points needs to meet the following conditions:

[0023] (S i -S i-1 )(S i -S i+1 )≤0

[0024] (A i -A i-1 )(A i -A i+1 )<0

[0025] In the formula, S i-1 , S i , S i+1 represent points in the main channel load history; A i-1 , A i , A i+1 represent corresponding points in the auxiliary channel load history.

[0026] A further improvement of the present invention lies in: according to the peak-valley value extraction result, performing rainflow counting on the main channel load history based on the four-point method, continuously reading four points in the main channel load history. If the load range of the middle two points is not affected by the load range of the outer two points, then extract the middle two points as a cycle; otherwise, the middle two points do not form a cycle. The formula for determining whether the middle two points form a cycle is:

[0027] |S i+2 -S i+1 |<|S i+1 -S i | and |S i+2 -S i+1 |<|S i+3 -S i+2 |

[0028] In the formula, S i , S i+1 , S i+2 , S i+3 represent points in the main channel load history;

[0029] According to the rainflow counting result of the main channel load history, determine the load subscript information of each cycle in the main channel load history. Based on the load subscript information, obtain the corresponding load values and load ranges in the auxiliary channel load history through indexing;

[0030] According to each cycle in the main channel load history, sequentially identify the corresponding cycles in the auxiliary channel load history, and obtain the cycle number statistical result of the random multi-axial load, that is, convert the stress history and strain history on the new plane into stress-strain cycles.

[0031] A further improvement of the present invention lies in: performing damage calculation on the random multi-axial load, taking the plane where the maximum damage of the random multi-axial load is located as the critical plane, and extracting the stress cycle and strain cycle on the critical plane, specifically including:

[0032] Construct a multiaxial fatigue life model to calculate the total damage at each level, including:

[0033] Based on the modified SWT life model, the damage of a single stress cycle is calculated:

[0034]

[0035] In the formula, represents the amplitude of the ith normal strain cycle on any plane; σ i,max Indicates the maximum normal stress corresponding to the normal strain cycle history on the plane; σ' f Represents fatigue strength coefficient; ε' f represents fatigue ductility coefficient; b represents fatigue strength index; c represents fatigue ductility index; E represents elastic modulus; N fi represents the fatigue life under the i-th cycle load;

[0036] Based on Miner's cumulative damage principle, the total damage of the stress cycle is calculated:

[0037]

[0038] Where n represents the number of load cycles; N fi represents the fatigue life under the i-th cyclic load; D represents the total damage of the stress cycle;

[0039] According to the total damage of the stress cycle, the plane where the maximum damage is located is determined as the critical plane, and the stress cycle and strain cycle on the critical plane are extracted.

[0040] A further improvement of the present invention is that: the test spectrum parameters are inferred based on the cycle number statistical results and damage calculation results based on the particle swarm algorithm, specifically including:

[0041] The expression for a single cycle of strain control with sinusoidal loading is:

[0042]

[0043]

[0044] In the formula, ε xx represents the positive strain component; γ xy represents the shear strain component; t represents time; ε a represents the normal strain amplitude; γ a represents the shear strain amplitude; λ represents the amplitude ratio; represents the phase difference between tension and torsion strain; ω represents the loading frequency; sin() represents the sine function;

[0045] Based on the statistical results of the number of cycles, the number of cycles is graded, and the total damage of each grade is calculated based on the modified SWT life model;

[0046] When inferring the test spectrum parameters based on the particle swarm algorithm, the dimension of the particles is the number of solutions. The test spectrum needs to solve three parameters, namely ε a ,λ, Assume that particles represented by N candidate solutions form a population, and the position information of the i-th particle in the population is:

[0047] X i =(x i1 ,x i2 ,x i3 )

[0048] Where, X i represents the position information of the i-th particle; x i1 、x i2 、x i3 They represent the three parameters that need to be solved in the test spectrum, namely ε a ,λ,

[0049] The velocity of the ith particle is:

[0050] V i =(v i1 ,v i2 ,v i3 )

[0051] Where V i represents the velocity of the ith particle; v i1 、v i2 、v i3 Respectively represent ε a ,λ, The corresponding speeds;

[0052] Randomly initialize the parameters of the population, parameter ε a The value range of is 1 / 3 of the minimum value to 3 times of the maximum value of the load variation range of the critical plane, the value range of parameter λ is between 0 and 3, and the value range of parameter The value range of is between 0 and 90°;

[0053] Construct a fitness function, the expression of which is:

[0054] |D R -D C |≈0

[0055] Where D R Indicates the damage caused by the original spectrum; D C Indicates the damage caused by the test spectrum;

[0056] The calculation formula for the velocity of the i-th particle in each iteration is:

[0057] V i,t+1 =ωV i,t +c1rand·(p best,t -X i,t )+c2rand·(g best,t -X i,t )

[0058] X i,t+1 =X i,t +V i,t+1

[0059] Where V i,t+1 Indicates the velocity of the ith particle when the current iteration number is t+1; V i,t represents the velocity of the i-th particle when the current iteration number is t; X i,t Indicates the position information of the i-th particle when the current iteration number is t; X i,t+1 The position information of the i-th particle when the current iteration number is t+1; ω represents the inertia weight; rand represents a randomly generated value between 0 and 1; c1 and c2 represent learning factors; p best,t represents the current individual optimal solution; g best,t represents the current optimal solution of the group; t represents the current number of iterations;

[0060] Compare the current fitness of each particle with the current individual optimal solution. If the current fitness is smaller than the current individual optimal solution, use the current individual optimal solution as the replaced current fitness.

[0061] Compare the current individual optimal solution of each particle with the current group optimal solution. If the current individual optimal solution is smaller than the current group optimal solution, use the current group optimal solution as the replaced current individual optimal solution.

[0062] When the number of iterations reaches the maximum number of iterations, the position information and fitness of the particle are output. The position information represents the parameter ε a ,λ, and compile an equal-amplitude test spectrum.

[0063] The beneficial effects of the present invention are as follows: the present invention proposes a test spectrum compilation method based on a particle swarm algorithm, which converts the stress-strain spectrum of the random multiaxial load into a stress history and a strain history on a new plane by performing coordinate transformation on the random multiaxial load; extracts peak and valley values ​​of the random multiaxial load; performs cycle number statistics on the random multiaxial load, and converts the stress history and strain history on the new plane into a stress-strain cycle; performs damage calculation on the random multiaxial load, takes the plane where the maximum damage of the random multiaxial load is located as the critical plane, and extracts the stress cycle and strain cycle on the critical plane; and reversely infers the test spectrum parameters based on the particle swarm algorithm according to the cycle number statistics results and the damage calculation results, thereby realizing the conversion of the random multiaxial load spectrum to the equal-amplitude test spectrum under the premise of ensuring damage equivalence, and based on the particle swarm algorithm, the search efficiency of the test spectrum parameters can be improved, which makes up for the disadvantage of low traversal search efficiency in the traditional method, ensures the damage equivalence before and after compilation, and has a wide range of engineering application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A flow chart of a test spectrum compilation method based on a particle swarm algorithm provided in an embodiment of the present invention;

[0065] Figure 2 A schematic diagram of the stress component history of the original stress-strain spectrum provided by an embodiment of the present invention;

[0066] Figure 3 A schematic diagram of the strain component history of the original stress-strain spectrum provided by an embodiment of the present invention;

[0067] Figure 4 A flowchart of multi-axis peak-valley value extraction provided by an embodiment of the present invention;

[0068] Figure 5 A schematic diagram of a cycle formed by two points in a four-point counting rule provided in an embodiment of the present invention;

[0069] Figure 6 A schematic diagram of two points not forming a cycle in the four-point counting rule provided in an embodiment of the present invention;

[0070] Figure 7 A flowchart of a four-point method for rain flow counting provided by an embodiment of the present invention;

[0071] Figure 8 A schematic diagram of a multi-axis load cycle counting method provided by an embodiment of the present invention;

[0072] Fig. 9 A flow chart of a particle swarm algorithm provided by an embodiment of the present invention;

[0073] Fig.10 A schematic diagram of a multi-axial load cycle of equal amplitude provided by an embodiment of the present invention;

[0074] Fig.11 Schematic diagram of single cycles at various levels of an equal amplitude test spectrum provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0075] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0076] See also Figure 1 , a test spectrum compilation method based on a particle swarm algorithm in this embodiment, the method comprises:

[0077] Performing coordinate transformation on the random multiaxial load, and transforming the stress-strain spectrum of the random multiaxial load into a stress history and a strain history on a new plane;

[0078] Extracting peak and valley values ​​of the random multiaxial load;

[0079] Performing cycle statistics on the random multiaxial load, and converting the stress history and strain history on the new plane into stress-strain cycles;

[0080] Performing damage calculation on the random multiaxial load, taking the plane where the maximum damage of the random multiaxial load is located as the critical plane, and extracting the stress cycle and strain cycle on the critical plane;

[0081] According to the cycle number statistical results and damage calculation results, the test spectrum parameters are inferred based on the particle swarm algorithm.

[0082] See also Figure 2-3 , the coordinate transformation of the random multi-axial load to convert the stress-strain spectrum of the random multi-axial load into the stress history and strain history on the new plane specifically includes:

[0083] Based on the critical plane method, the angles θ and ψ of the plane where the maximum damage of the random multiaxial load is located are determined, the angle θ increases from 0° to 360° at intervals of 1°, and the angle ψ increases from 0° to 180° at intervals of 1°, and any plane in the three-dimensional space under the Cartesian coordinate system is generated by rotating the angles θ and ψ, and the stress-strain spectrum of the random multiaxial load is transformed into a stress history and a strain history on the new plane;

[0084] The formula for the coordinate transformation is:

[0085]

[0086]

[0087]

[0088] ε'=MεM T

[0089] In the formula, σ represents the stress tensor of a point in the original random stress spectrum; σ x , τ xy , τ xz , τ xy , σ y , τ yz , τ xz , τ yz , σ z represents the stress component corresponding to the point, and σ x , σ y , σ z represents normal stress. The subscript of normal stress represents the normal direction of the plane of action of normal stress. xy , τ xz , τ yz represents shear stress. The first subscript of shear stress represents the normal direction of the plane of action of shear stress, and the second subscript represents the coordinate direction parallel to the shear stress. In addition, on the positive side, that is, the normal direction on the action plane is along the positive direction of the coordinate axis, the stress along the positive direction of the coordinate axis is positive; on the negative side, that is, the normal direction on the action plane is along the negative direction of the coordinate axis, the stress along the negative direction of the coordinate axis is positive; ε represents the strain tensor of a point in the original random strain spectrum; ε x , 1 / 2γ xy , 1 / 2γ xz , 1 / 2γ xy , ε y , 1 / 2γ yz , 1 / 2γ xz , 1 / 2γ yz , ε z represents the strain component corresponding to the point, and ε x , ε y , ε z represents the normal strain. The subscript of the normal strain represents the normal direction of the plane of action of the normal strain. 1 / 2γ xy , 1 / 2γ xz , 1 / 2γ yzrepresents shear strain. The first subscript of shear strain represents the normal direction of the plane of action of shear strain. The second subscript represents the coordinate direction parallel to the shear strain. In addition, on the positive side, that is, the normal direction on the action plane is along the positive direction of the coordinate axis, the strain along the positive direction of the coordinate axis is positive; on the negative side, that is, the normal direction on the action plane is along the negative direction of the coordinate axis, the strain along the negative direction of the coordinate axis is positive; M represents the transformation matrix; M T represents the transpose of the transformation matrix; σ' represents the stress history on the new plane; ε' represents the strain history on the new plane; θ and ψ represent the angles of the plane where the maximum damage of random multiaxial loading is located.

[0090] See also Figure 4 , the peak-to-valley value extraction of the random multi-axial load specifically includes:

[0091] Obtain the main channel and auxiliary channel load histories of the stress component history and the strain component history. If the corresponding points in the load histories of the main channel and the auxiliary channel have peak and valley values ​​in the two load histories, retain the corresponding points; if the corresponding points in the load histories of the main channel and the auxiliary channel do not have peak and valley values ​​in the two load histories, remove the corresponding points. That is, the removal of the corresponding points needs to meet the following conditions:

[0092] (S i -S i-1 )(S i -S i+1 )≤0

[0093] (A i -A i-1 )(A i -A i+1 )<0

[0094] In the formula, S i-1 , S i , S i+1 Represents a point in the main channel load history; A i-1 , A i , A i+1 Indicates the point in the corresponding auxiliary channel load history.

[0095] According to the peak-valley value extraction results, the rain flow counting of the main channel load history is performed based on the four-point method, see Figure 5-6 , continuously read the four points in the load history of the main channel. If the load range of the middle two points is not affected by the load range of the outer two points, the middle two points are extracted as a cycle. Otherwise, the middle two points do not constitute a cycle. See Figure 7 , the formula to determine whether the two middle points form a cycle is:

[0096] |S i+2 -S i+1|<|S i+1 -S i |AND|S i+2 -S i+1 |<|S i+3 -S i+2 |

[0097] wherein, S i , S i+1 , S i+2 , S i+3 represent points in the main channel load history;

[0098] Determine the load subscript information for each cycle in the main channel load history according to the rainflow counting result of the main channel load history. Based on the load subscript information, obtain the corresponding load value and load range in the secondary channel load history through indexing;

[0099] See Figure 8 , according to each cycle in the main channel load history, sequentially identify the corresponding cycles in the secondary channel load history, and obtain the cycle number statistical result of the random multiaxial load, that is, convert the stress history and strain history on the new plane into stress-strain cycles.

[0100] Performing damage calculation on the random multiaxial load, taking the plane where the maximum damage of the random multiaxial load is located as the critical plane, and extracting the stress cycle and strain cycle on the critical plane, specifically including:

[0101] Construct a multiaxial fatigue life model to calculate the total damage of each level, including:

[0102] Based on the modified SWT life model, calculate the damage of a single stress cycle:

[0103]

[0104] wherein, represents the amplitude of the i-th positive strain cycle on any plane; σ i,max represents the maximum positive stress corresponding to the positive strain cycle history on this plane; σ' f represents the fatigue strength coefficient; ε' f represents the fatigue ductility coefficient; b represents the fatigue strength index; c represents the fatigue ductility index; E represents the elastic modulus; N fi represents the fatigue life under the action of the i-th cyclic load;

[0105] Based on the Miner cumulative damage principle, calculate the total damage of the stress cycle:

[0106]

[0107] Where n represents the number of load cycles; N fi represents the fatigue life under the i-th cyclic load; D represents the total damage of the stress cycle, and the result is 1.29×10 -6 ;

[0108] According to the total damage of the stress cycle, the plane where the maximum damage is located is determined as the critical plane, and the stress cycle and strain cycle on the critical plane are extracted.

[0109] See also Fig. 9 , the test spectrum parameters are inferred based on the particle swarm algorithm according to the cycle number statistical results and damage calculation results, specifically including:

[0110] See also Fig.10 , the expression for a single cycle of strain control with sinusoidal loading is:

[0111]

[0112]

[0113] In the formula, ε xx represents the positive strain component; γ xy represents the shear strain component; t represents time; ε a represents the normal strain amplitude; γ a represents the shear strain amplitude; λ represents the amplitude ratio; represents the phase difference between tension and torsion strain; ω represents the loading frequency; sin() represents the sine function;

[0114] Based on the statistical results of the cycle number, the cycle number is divided into 4 levels. Based on the modified SWT life model, the total damage of each level is calculated. The total damage of each level is 6.120×10 -12 4.782×10 -9 , 1.107×10 -7 , 1.176×10 -6 ;

[0115] When inferring the test spectrum parameters based on the particle swarm algorithm, the dimension of the particles is the number of solutions. The test spectrum needs to solve three parameters, namely ε a ,λ, Assume that particles represented by N candidate solutions form a population, and the position information of the i-th particle in the population is:

[0116] X i =(x i1 ,x i2 ,x i3 )

[0117] Where, X irepresents the position information of the i-th particle; x i1 、x i2 、x i3 They represent the three parameters that need to be solved in the test spectrum, namely ε a ,λ,

[0118] The velocity of the ith particle is:

[0119] V i =(v i1 ,v i2 ,v i3 )

[0120] Where V i represents the velocity of the ith particle; v i1 、v i2 、v i3 Respectively represent ε a ,λ, The corresponding speeds;

[0121] Randomly initialize the parameters of the population, parameter ε a The value range of is 1 / 3 of the minimum value to 3 times of the maximum value of the load variation range of the critical plane, the value range of parameter λ is between 0 and 3, and the value range of parameter The value range of is between 0 and 90°;

[0122] Construct a fitness function, the expression of which is:

[0123] |D R -D C |≈0

[0124] Where D R Indicates the damage caused by the original spectrum; D C Indicates the damage caused by the test spectrum;

[0125] The calculation formula for the velocity of the i-th particle in each iteration is:

[0126] V i,t+1 =ωV i,t +c1rand·(p best,t -X i,t )+c2rand·(g best,t -X i,t )

[0127] X i,t+1 =X i,t +V i,t+1

[0128] Where V i,t+1Indicates the velocity of the ith particle when the current iteration number is t+1; V i,t represents the velocity of the i-th particle when the current iteration number is t; X i,t Indicates the position information of the i-th particle when the current iteration number is t; X i,t+1 The position information of the i-th particle when the current iteration number is t+1; ω represents the inertia weight; rand represents a randomly generated value between 0 and 1; c1 and c2 represent learning factors; p best,t represents the current individual optimal solution; g best,t represents the current optimal solution of the group; t represents the current number of iterations;

[0129] Compare the current fitness of each particle with the current individual optimal solution. If the current fitness is smaller than the current individual optimal solution, use the current individual optimal solution as the replaced current fitness.

[0130] Compare the current individual optimal solution of each particle with the current group optimal solution. If the current individual optimal solution is smaller than the current group optimal solution, use the current group optimal solution as the replaced current individual optimal solution.

[0131] When the number of iterations reaches the maximum number of iterations, the optimization is exited and the position information and fitness of the particle are output. The position information represents the parameter ε a ,λ, and compile the equal-amplitude test spectrum, see Fig.11 , and obtain a single cycle of equal amplitude test spectrum per level.

[0132] Through the introduction of the above embodiments, the present invention adopts a test spectrum compilation method based on a particle swarm algorithm, and through coordinate transformation of the random multi-axial load, the stress-strain spectrum of the random multi-axial load is converted into the stress history and strain history on the new plane; the peak and valley values ​​of the random multi-axial load are extracted; the number of cycles of the random multi-axial load is counted, and the stress history and strain history on the new plane are converted into stress-strain cycles; the damage calculation of the random multi-axial load is performed, and the plane where the maximum damage of the random multi-axial load is located is used as the critical plane, and the stress cycle and strain cycle on the critical plane are extracted; according to the cycle number statistics results and the damage calculation results, the test spectrum parameters are inferred based on the particle swarm algorithm, so that under the premise of ensuring damage equivalence, the conversion of the random multi-axial load spectrum to the equal-amplitude test spectrum can be realized, and based on the particle swarm algorithm, the search efficiency of the test spectrum parameters can be improved, which makes up for the shortcomings of low traversal search efficiency in traditional methods, ensures the damage equivalence before and after compilation, and has a wide range of engineering application value.

[0133] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have the same meaning as in the context of the prior art, and will not be interpreted with an idealized or overly formal meaning unless defined similarly here.

[0134] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A test spectrum compilation method based on particle swarm algorithm, characterized in that: The method comprises: Performing coordinate transformation on the random multiaxial load, and transforming the stress-strain spectrum of the random multiaxial load into a stress history and a strain history on a new plane; Extracting peak and valley values ​​of the random multiaxial load; Performing cycle statistics on the random multiaxial load, and converting the stress history and strain history on the new plane into stress-strain cycles; Performing damage calculation on the random multiaxial load, taking the plane where the maximum damage of the random multiaxial load is located as the critical plane, and extracting the stress cycle and strain cycle on the critical plane; According to the cycle number statistical results and damage calculation results, the test spectrum parameters are inferred based on the particle swarm algorithm.

2. The test spectrum compilation method based on particle swarm algorithm according to claim 1 is characterized in that: The coordinate transformation of the random multi-axial load to transform the stress-strain spectrum of the random multi-axial load into the stress history and strain history on the new plane specifically includes: Based on the critical plane method, the angles θ and ψ of the plane where the maximum damage of the random multiaxial load is located are determined, the angle θ increases from 0° to 360° at intervals of 1°, and the angle ψ increases from 0° to 180° at intervals of 1°, and any plane in the three-dimensional space under the Cartesian coordinate system is generated by rotating the angles θ and ψ, and the stress-strain spectrum of the random multiaxial load is transformed into a stress history and a strain history on the new plane; The formula for the coordinate transformation is: σ'=MσM T e'=MεM T In the formula, σ represents the stress tensor of a point in the original random stress spectrum; σ x , τ xy , τ xz , τ xy , σ y , τ yz , τ xz , τ yz , σ z represents the stress component corresponding to the point, and σ x , σ y , σ z represents normal stress. The subscript of normal stress represents the normal direction of the plane of action of normal stress. xy , τ xz , τ yz represents shear stress. The first subscript of shear stress represents the normal direction of the plane of action of shear stress, and the second subscript represents the coordinate direction parallel to the shear stress. In addition, on the positive side, that is, the normal direction on the action plane is along the positive direction of the coordinate axis, the stress along the positive direction of the coordinate axis is positive; on the negative side, that is, the normal direction on the action plane is along the negative direction of the coordinate axis, the stress along the negative direction of the coordinate axis is positive; ε represents the strain tensor of a point in the original random strain spectrum; ε x , 1 / 2γ xy , 1 / 2γ xz , 1 / 2λ xy , ε y , 1 / 2γ yz , 1 / 2γ xz , 1 / 2γ yz , ε z represents the strain component corresponding to the point, and ε x , ε y , ε z represents the normal strain. The subscript of the normal strain represents the normal direction of the plane of action of the normal strain. 1 / 2γ xy , 1 / 2γ xz , 1 / 2γ yz represents shear strain. The first subscript of shear strain represents the normal direction of the plane of action of shear strain. The second subscript represents the coordinate direction parallel to the shear strain. In addition, on the positive side, that is, the normal direction on the action plane is along the positive direction of the coordinate axis, the strain along the positive direction of the coordinate axis is positive; on the negative side, that is, the normal direction on the action plane is along the negative direction of the coordinate axis, the strain along the negative direction of the coordinate axis is positive; M represents the transformation matrix; M T represents the transpose of the transformation matrix; σ' represents the stress history on the new plane; ε' represents the strain history on the new plane; θ and ψ represent the angles of the plane where the maximum damage of random multiaxial loading is located.

3. The test spectrum compilation method based on particle swarm algorithm according to claim 1 is characterized in that: The extracting of peak and valley values ​​of the random multi-axial load specifically includes: Obtain the main channel and auxiliary channel load histories of the stress component history and the strain component history. If the corresponding points in the load histories of the main channel and the auxiliary channel have peak and valley values ​​in the two load histories, retain the corresponding points; if the corresponding points in the load histories of the main channel and the auxiliary channel do not have peak and valley values ​​in the two load histories, remove the corresponding points. That is, the removal of the corresponding points needs to meet the following conditions: (S i -S i-1 )(S i -S i+1 )≤0 (A i -A i-1 )(A i -A i+1 )<0 In the formula, S i-1 , S i , S i+1 Represents a point in the main channel load history; A i-1 , A i , A i+1 Indicates the point in the corresponding auxiliary channel load history.

4. The test spectrum compilation method based on particle swarm algorithm according to claim 3 is characterized in that: According to the peak-valley value extraction result, the rain flow count is performed on the main channel load history based on the four-point method, and four points in the main channel load history are continuously read. If the load range of the middle two points is not affected by the load range of the outer two points, the middle two points are extracted as a cycle, otherwise the middle two points do not constitute a cycle. The formula for judging whether the middle two points constitute a cycle is: |S i+2 -S i+1 |<|S i+1 -S i |and|S i+2 -S i+1 |<|S i+3 -S i+2 | In the formula, S i , S i+1 , S i+2 , S i+3 Represents a point in the main channel load history; Determine the load subscript information of each cycle in the load history of the main channel according to the rain flow counting result of the load history of the main channel, and obtain the corresponding load value and load range in the load history of the auxiliary channel by indexing based on the load subscript information; According to each cycle in the load history of the main channel, the corresponding cycles in the load history of the auxiliary channel are identified in turn, and the statistical results of the number of cycles of the random multi-axial load are obtained, that is, the stress history and strain history on the new plane are converted into stress-strain cycles.

5. The test spectrum compilation method based on particle swarm algorithm according to claim 1 is characterized in that: The damage calculation for the random multiaxial load is performed, the plane where the maximum damage of the random multiaxial load is located is used as the critical plane, and the stress cycle and strain cycle on the critical plane are extracted, specifically including: constructing a multiaxial fatigue life model to calculate the total damage at each level, including: Based on the modified SWT life model, the damage of a single stress cycle is calculated: In the formula, represents the amplitude of the ith normal strain cycle on any plane; σ i,max Indicates the maximum normal stress corresponding to the normal strain cycle history on the plane; σ' f Represents fatigue strength coefficient; ε' f represents fatigue ductility coefficient; b represents fatigue strength index; c represents fatigue ductility index; E represents elastic modulus; N fi represents the fatigue life under the i-th cycle load; Based on Miner's cumulative damage principle, the total damage of the stress cycle is calculated: Where n represents the number of load cycles; N fi represents the fatigue life under the i-th cyclic load; D represents the total damage of the stress cycle; According to the total damage of the stress cycle, the plane where the maximum damage is located is determined as the critical plane, and the stress cycle and strain cycle on the critical plane are extracted.

6. The test spectrum compilation method based on particle swarm algorithm according to claim 1 is characterized in that: The method of inferring the test spectrum parameters based on the particle swarm algorithm according to the cycle number statistical results and the damage calculation results specifically includes: The expression for a single cycle of strain control with sinusoidal loading is: In the formula, ε xx represents the positive strain component; γ xy represents the shear strain component; t represents time; ε a represents the normal strain amplitude; γ a represents the shear strain amplitude; λ represents the amplitude ratio; represents the phase difference between tension and torsion strain; ω represents the loading frequency; sin() represents the sine function; Based on the statistical results of the number of cycles, the number of cycles is graded, and the total damage of each grade is calculated based on the modified SWT life model; When inferring the test spectrum parameters based on the particle swarm algorithm, the dimension of the particles is the number of solutions. The test spectrum needs to solve three parameters, namely ε a ,λ, Assume that particles represented by N candidate solutions form a population, and the position information of the i-th particle in the population is: X i =(x i1 ,x i2 ,x i3 ) Where, X i represents the position information of the i-th particle; x i1 、x i2 、x i3 They represent the three parameters that need to be solved in the test spectrum, namely ε a ,λ, The velocity of the ith particle is: V i =(v i1 ,v i2 ,v i3 ) Where V i represents the velocity of the ith particle; v i1 、v i2 、v i3 Respectively represent ε a ,λ, The corresponding speeds; Randomly initialize the parameters of the population, parameter ε a The value range of is 1 / 3 of the minimum value to 3 times of the maximum value of the load variation range of the critical plane, the value range of parameter λ is between 0 and 3, and the value range of parameter The value range of is between 0 and 90°; Construct a fitness function, the expression of which is: |D R -D C |≈0 Where D R Indicates the damage caused by the original spectrum; D C Indicates the damage caused by the test spectrum; The calculation formula for the velocity of the i-th particle in each iteration is: V i,t+1 =ωV i,t +c1rand·(p best,t -X i,t )+c2rand·(g best,t -X i,t ) X i,t+1 =X i,t +V i,t+1 In the formula, v i,t+1 represents the velocity of the i-th particle when the current iteration number is t+1; v i,t represents the velocity of the i-th particle when the current iteration number is t; X i,t Indicates the position information of the i-th particle when the current iteration number is t; X i,t+1 The position information of the i-th particle when the current iteration number is t+1; ω represents the inertia weight; rand represents a randomly generated value between 0 and 1; c1 and c2 represent learning factors; p best,t represents the current individual optimal solution; g best,t represents the current optimal solution of the group; t represents the current number of iterations; Compare the current fitness of each particle with the current individual optimal solution. If the current fitness is smaller than the current individual optimal solution, use the current individual optimal solution as the replaced current fitness. Compare the current individual optimal solution of each particle with the current group optimal solution. If the current individual optimal solution is smaller than the current group optimal solution, use the current group optimal solution as the replaced current individual optimal solution. When the number of iterations reaches the maximum number of iterations, the position information and fitness of the particle are output. The position information represents the parameter ε a ,λ, and compile an equal-amplitude test spectrum.

Images