A multi-factor analysis-based packaging material life prediction method

Abstract

The application discloses a packaging material life prediction method based on multi-factor analysis and relates to the technical field of reliability prediction, which comprises the following steps: based on a phase collision response field, a synchronous amplification path of a failure response variable is tracked along a collision grid area, grid release verification is performed on the collision grid area on the synchronous amplification path, life gate grid is determined, and a life gate grid chain is generated in series; local damage recursion is performed on the life gate grid; when a continuous failure path is first formed in the life gate grid chain, a life measurement value corresponding to a current phase collision time window is counted, closed counterexample playback is performed on the continuous failure path, and packaging material predicted life is obtained. According to the application, the life gate grid chain is determined, so that the life prediction no longer depends on a single maximum stress point, and the accuracy of life bottleneck positioning and life prediction results is improved.

Description

Technical Field

[0001] This invention relates to the field of reliability prediction technology, and in particular to a method for predicting the lifetime of packaging materials based on multi-factor analysis. Background Technology

[0002] With the development of high-density integrated circuits, memory chips, power devices, and multi-chip packaging technologies, the role of packaging materials in device reliability is becoming increasingly prominent. To assess the lifetime of packaging materials in advance during the design phase, existing technologies typically employ finite element analysis (FEM) to establish a packaging structure model. By applying temperature loads, humidity loads, mechanical loads, or electrothermal loads, the stress, strain, temperature field, humidity field, and energy release rate within the packaging structure are calculated. Combined with fatigue damage criteria, interface debonding criteria, or accelerated life models, the reliable lifetime of the packaging material is predicted. FEM has become an important technical tool in packaging reliability analysis and computer-aided engineering design.

[0003] However, existing technologies for predicting the lifetime of encapsulation materials still have two shortcomings. First, current multi-factor finite element lifetime analysis typically treats thermal, humidity, electrothermal, and mechanical loads as parallel boundary conditions and superimposes them as a whole, focusing on load amplitudes and peak values ​​of individual field variables. This makes it difficult to distinguish between effective collision loads that truly trigger failure and pseudo-coupled loads that only overlap in time but contribute weakly to failure propagation. Second, existing lifetime results are often determined based on the maximum stress region, the maximum strain region, or the overall cumulative damage value, easily leading to the direct identification of local high-response regions as lifetime bottlenecks. Therefore, it is difficult to accurately determine which mesh region, which load combination, and which phase time window triggers the termination of the encapsulation material's lifetime. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for predicting the lifetime of packaging materials based on multi-factor analysis to solve the problems of inaccurate identification of effective failure trigger load and inaccurate judgment of lifetime termination time window.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] This invention provides a method for predicting the lifetime of encapsulation materials based on multi-factor analysis. The method includes: constructing a lifetime finite element dataset for the encapsulation material and applying multi-factor loads; generating a disturbance response fingerprint database by recording failure response variables; constructing a load phase-sensitive model using a disturbance fingerprint embedding layer, a load phase mapping layer, and an interface collision discrimination layer; determining the failure response principal axis by combining the disturbance response fingerprint database; mapping the multi-factor loads to the failure response principal axis; identifying effective collision loads; and generating a phase collision load spectrum. The method further includes extracting the phase collision time window and collision grid region corresponding to the effective collision loads in the phase collision load spectrum, and then... The effective collision load is converted into finite element coupled boundary conditions, and local finite element incremental solution is performed on the collision mesh region to generate a phase collision response field. Based on the phase collision response field, the synchronous amplification path of the failure response variable is traced along the collision mesh region, and mesh release verification is performed on the collision mesh region on the synchronous amplification path to determine the lifetime gate mesh, and a lifetime gate mesh chain is generated in series. Local damage recursion is performed on the lifetime gate mesh. When a continuous failure path is first formed in the lifetime gate mesh chain, the lifetime measurement value corresponding to the current phase collision time window is statistically analyzed, and closed proof replay is performed on the continuous failure path to obtain the predicted lifetime of the encapsulation material.

[0008] As a preferred embodiment of the packaging material lifetime prediction method based on multi-factor analysis described in this invention, the specific steps of constructing a lifetime finite element dataset for the packaging material and applying multi-factor loads are as follows:

[0009] Collect structural geometry data, material property data, interface contact data, and test condition data of the packaging material, and bind the mesh region numbers to generate a lifetime finite element dataset;

[0010] Based on the lifetime finite element dataset, multi-factor loads are fully applied to the mesh region corresponding to the encapsulation material, and the failure response variables of the mesh region under the action of multi-factor loads are recorded.

[0011] As a preferred embodiment of the packaging material lifetime prediction method based on multi-factor analysis described in this invention, the specific steps for generating the disturbance response fingerprint database are as follows:

[0012] Based on the failure response variables, one type of load is removed from the multi-factor load in turn and the remaining load is reapplied. The failure response variables before and after the removal are compared by contradiction to obtain the necessary triggering load.

[0013] The necessary triggering load is transferred from the original grid region to an adjacent grid region and reapplied. The failure response variable of the original grid region is observed to decrease with the load transfer, and the failure response variable of the adjacent grid region is observed to increase with the load transfer, so as to obtain the effective propagation relationship.

[0014] Based on the effective propagation relationship, the necessary triggering loads, failure response variables and effective propagation relationships corresponding to the same grid are encapsulated in a closed loop to generate a disturbance response fingerprint library.

[0015] As a preferred embodiment of the packaging material lifetime prediction method based on multi-factor analysis described in this invention, the specific steps for constructing a load phase-sensitive model using a perturbation fingerprint embedding layer, a load phase mapping layer, and an interface collision discrimination layer are as follows:

[0016] The perturbation fingerprint embedding layer merges fingerprints of interface peeling response and shear expansion response triggered by different loads within the same grid area based on failure response variables, and removes isolated response directions triggered by a single load through the main axis pin removal process, generating failure response main axis features.

[0017] The load phase mapping layer establishes a mapping relationship between multi-factor loads and the failure response principal axis based on the failure response principal axis characteristics. Through load switching trial processing, it determines whether multi-factor loads can cause the failure response variable to be amplified along the same failure response principal axis, and generates principal axis latched load characteristics.

[0018] The interface collision discrimination layer, based on the spindle latched load characteristics, performs collision discrimination on the peeling amplification marks and shear amplification marks on the same failure response spindle of multi-factor loads, and generates interface collision discrimination features.

[0019] The failure response principal axis features output by the perturbation fingerprint embedding layer are used as the mapping constraints of the load phase mapping layer. The principal axis latched load features output by the load phase mapping layer are used as the discrimination input of the interface collision discrimination layer. The collision discrimination features formed by the interface collision discrimination layer are used to inversely constrain the principal axis retention and load latching range of the perturbation fingerprint embedding layer and the load phase mapping layer, thus constructing a load phase sensitive model.

[0020] As a preferred embodiment of the packaging material lifetime prediction method based on multi-factor analysis described in this invention, the specific steps for generating the phase collision load spectrum are as follows:

[0021] Load records pointing to the same interface stripping enhancement direction and the same shear propagation enhancement direction within the same grid area are read from the disturbance response fingerprint database. The interface stripping enhancement direction and the shear propagation enhancement direction are determined as the failure response principal axis. The load records are mapped to the principal axis to generate the principal axis triggered load sequence.

[0022] Based on the principal axis triggered load sequence, missing item verification and out-of-order verification are performed on load combinations mapped to the same failure response principal axis within the same time window. If removing any load cannot maintain the synchronous amplification of the failure response variable, and changing the original time order of the loads cannot maintain the synchronous amplification of the failure response variable, then it is determined to be a valid collision load and is bound accordingly to generate a phase collision load spectrum.

[0023] As a preferred embodiment of the packaging material lifetime prediction method based on multi-factor analysis described in this invention, the specific steps for extracting the phase collision time window and collision grid region corresponding to the effective collision load in the phase collision load spectrum are as follows:

[0024] The continuous time period that can cause the simultaneous occurrence of interface stripping enhancement and shearing expansion enhancement is taken as the candidate collision time period, and the time node cancellation verification is performed on the time nodes within the candidate collision time period to obtain the phase collision time window.

[0025] Based on the phase collision time window, the grid region corresponding to the effective collision load is read, and the interface misalignment is checked on the grid region. The grid region that loses synchronous amplification and transmission after misalignment and re-establishes synchronous amplification and transmission after returning to its original position is determined as the collision grid region.

[0026] As a preferred embodiment of the packaging material lifetime prediction method based on multi-factor analysis described in this invention, the specific steps for generating the phase collision response field are as follows:

[0027] Temperature loads are converted into thermal expansion boundaries, humidity loads into hygroscopic expansion boundaries, electrothermal loads into local heating boundaries, and mechanical loads into mechanically equivalent boundaries, which are simultaneously applied to the collision mesh region to generate finite element coupled boundary conditions.

[0028] A dual-track incremental solution is performed on the coupled boundary conditions. The incremental collision response when the effective collision load is fully applied is differentially calculated with the incremental response after removing the single type of effective collision load to obtain the increment of the failure response variable. The unique collision response generated only by the combined action of the effective collision load is screened out through the interface through-closed verification, and the phase collision response field is generated.

[0029] As a preferred embodiment of the multi-factor analysis-based packaging material lifetime prediction method of the present invention, the synchronous amplification path of tracking failure response variables along the collision grid region based on the phase collision response field refers to generating candidate synchronous amplification paths by amplifying the interface peeling driving force increment and shear strain increment along the failure response principal axis of the collision grid region based on the phase collision response field, and performing chain break verification on the candidate synchronous amplification paths to generate synchronous amplification paths.

[0030] As a preferred embodiment of the packaging material lifetime prediction method based on multi-factor analysis described in this invention, the specific steps for generating the lifetime gate mesh chain are as follows:

[0031] Collision response stripping is performed on the candidate collision mesh region on the synchronous amplification path, while preserving the finite element connection relationship of the candidate collision mesh region and only stripping the failure response variable increment generated by the effective collision load.

[0032] Based on the stripped candidate collision grid region, a bypass substitute response is constructed. It is determined whether the synchronous amplification path can bypass the candidate collision grid region, the lifetime gate grid is obtained, and the lifetime gate grids are connected in series to generate a lifetime gate grid chain.

[0033] As a preferred embodiment of the encapsulation material lifetime prediction method based on multi-factor analysis described in this invention, the method involves performing local damage recursion on the lifetime gate grid. When a continuous failure path is first formed in the lifetime gate grid chain, the lifetime measurement value corresponding to the current phase collision time window is statistically analyzed, and the continuous failure path is subjected to closed-loop proof replay to obtain the predicted lifetime of the encapsulation material. The specific steps are as follows:

[0034] Based on the lifetime gate mesh chain and the phase collision response field, according to the arrangement order of the lifetime gate mesh chain, the interface stripping driving force increment, shear strain increment, temperature gradient increment and humidity enrichment increment of the corresponding lifetime gate mesh in each phase collision time window are written into the same chain sequence record to generate the gate response chain sequence table.

[0035] Based on the gate response chain sequence table, the corresponding gate penetration triggering amount is calculated according to the interface stripping driving force increment, shear strain increment, temperature gradient increment and humidity enrichment increment of the same lifetime gate mesh within the phase collision time window.

[0036] Based on the gate breakthrough trigger quantity, local damage recursion and chain sequence relay verification are performed according to the arrangement order of the life gate grid chain. Only life gate grids that have reached the breakthrough judgment threshold and can accept the upstream breakthrough result are marked as relay breakthrough grids, and chain sequence relay breakthrough records are generated.

[0037] In the chain sequence relay connection record, the phase collision time window that makes the discontinuous connection segment first spliced ​​into a continuous failure path is identified as a candidate closing key window, and the lifetime measurement value of the phase collision time window corresponding to the candidate closing key window is calculated.

[0038] The gate penetration trigger value in the candidate closing key window is shifted to the adjacent time window for re-verification. When the continuous failure path cannot be closed after the shift, the original time window is restored and the path closing trigger window is obtained. The lifetime measurement value bound to the path closing trigger window is determined as the predicted lifetime of the packaging material.

[0039] The beneficial effects of this invention are as follows: by constructing a disturbance response fingerprint library and generating a phase collision load spectrum, it is possible to screen out the effective collision loads that truly cause interface peeling enhancement and shear propagation enhancement from multi-factor loads. By determining the lifetime gate mesh chain, lifetime prediction no longer depends on a single maximum stress point, thus improving the accuracy of lifetime bottleneck location and lifetime prediction results in finite element analysis. Attached Figure Description

[0040] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a flowchart of a method for predicting the lifetime of packaging materials based on multi-factor analysis.

[0042] Figure 2 This is a flowchart for generating a disturbance response fingerprint.

[0043] Figure 3 A flowchart for generating the phase collision load spectrum.

[0044] Figure 4 This is a flowchart for identifying lifespan gates.

[0045] Figure 5 A comparison chart of the load removal response.

[0046] Figure 6 This is a comparison chart of predicted lifetime errors. Detailed Implementation

[0047] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0048] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0049] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0050] Reference Figures 1-6 This is one embodiment of the present invention, which provides a method for predicting the lifetime of packaging materials based on multi-factor analysis, including the following steps:

[0051] S1: Construct a lifetime finite element dataset for the packaging material and apply multi-factor loads. By recording failure response variables, generate a disturbance response fingerprint library.

[0052] S1.1: Collect structural geometry data, material property data, interface contact data, and test condition data of the packaging material, and bind the grid region numbers to generate a life finite element dataset.

[0053] Specifically, the positional relationships of the chip, substrate, molding compound, underfill adhesive, adhesive layer, solder joints, and package boundary in the packaging material are collected sequentially, and the length, width, thickness, interlayer contact position, chip edge position, solder joint distribution position, and package outer boundary position are recorded to form the structural geometric data of the packaging material.

[0054] The molding compound, underfill adhesive, adhesive layer, substrate material, and solder joint material in the packaging material are read and recorded to obtain the corresponding elastic modulus, Poisson's ratio, thermal expansion performance, thermal conductivity performance, moisture absorption and diffusion performance, moisture absorption and expansion performance, and interface bonding performance, forming material property data.

[0055] The contact interfaces between the molding compound and the substrate, between the bottom filler and the solder joint, between the adhesive layer and the chip, and between adjacent material layers are collected and recorded. The location, contact area, interface normal direction, interface tangential direction, and material relationship on both sides of the interface are recorded to form interface contact data.

[0056] Collect test condition data of temperature load, humidity load, electrothermal load and mechanical load corresponding to the packaging material, and record the action time, action area, action direction and action intensity of each type of load to form test condition data of the packaging material.

[0057] Finite element mesh regions are divided based on structural geometry data, and a unique mesh region number is assigned to each finite element mesh region. Material property data are bound to the corresponding mesh region number according to the material layer location, interface contact data are bound to the corresponding mesh region number according to the contact relationship between adjacent mesh regions, and test condition data are bound to the corresponding mesh region number according to the load application area and load application direction, thus generating a life finite element dataset.

[0058] S1.2: Based on the lifetime finite element dataset, multi-factor loads are fully applied to the mesh region corresponding to the encapsulation material, and the failure response variables of the mesh region under the action of multi-factor loads are recorded.

[0059] Specifically, based on the life finite element dataset, the temperature load, humidity load, electrothermal load, and mechanical load in the test condition data are identified as multi-factor loads. According to the structural location, material properties, interface relationship, and load application area corresponding to the grid region number, the multi-factor loads are assigned to the grid region corresponding to the packaging material.

[0060] For the grid region inside the material layer, temperature load, electrothermal load and mechanical load are applied to the corresponding material region, and humidity load is applied to the corresponding material region according to the moisture diffusion effect region; for the grid region at the contact position of adjacent material layers, temperature load, humidity load, electrothermal load and mechanical load are applied to the corresponding interface region simultaneously, so that the multi-factor loads are fully applied to the corresponding grid region according to the load application time, load application direction and load application intensity.

[0061] The temperature gradient change, humidity enrichment change, interface delamination driving force change, and shear strain change are recorded according to the grid region number, and these changes are identified as the failure response variables of the grid region under multi-factor loading.

[0062] S1.3: Based on the failure response variables, remove one type of load from the multi-factor loads in sequence and then reapply the remaining loads. Compare the failure response variables before and after removal to obtain the necessary triggering loads.

[0063] Specifically, based on the failure response variables recorded under the complete action of multi-factor loads, temperature load, humidity load, electrothermal load and mechanical load are sequentially taken as loads to be removed. When removing the loads to be removed each time, the action time, action area, action direction and action intensity of the remaining loads are kept unchanged. After reapplying the remaining loads, the temperature gradient change, humidity enrichment change, interface peeling driving force change and shear strain change in the same grid area are recorded.

[0064] The failure response variables formed after reapplying the remaining load are compared with the failure response variables recorded in the same grid area and within the same time period under the complete action of multi-factor loads. If, after removing a certain type of load, the change in interface peeling driving force and shear strain changes from an enhanced state to a non-enhanced state, and the change in temperature gradient and humidity enrichment no longer support the synchronous enhancement of the change in interface peeling driving force and shear strain, then the removed load is determined as the necessary triggering load for the corresponding grid area.

[0065] If the failure response variable remains enhanced after removing a certain type of load, the removed load will not be identified as a necessary triggering load, and the necessary triggering load will be obtained.

[0066] S1.4: Transfer the necessary triggering load from the original grid region to the adjacent grid region and reapply it. Observe whether the failure response variable of the original grid region diminishes with the load transfer and whether the failure response variable of the adjacent grid region increases with the load transfer to obtain the effective propagation relationship.

[0067] Specifically, based on the binding relationship between the necessary trigger load and the grid region number in the finite-life metadata set, the original grid region corresponding to the necessary trigger load is determined, and based on the interface contact data and the material layer position relationship, adjacent grid regions that share an interface with the original grid region or are adjacent to the same material layer are selected from the finite-life metadata set.

[0068] Keeping the load type, duration, direction, and intensity of the necessary triggering load unchanged, the application location of the necessary triggering load is transferred from the original grid region to an adjacent grid region and reapplied. The failure response variables of the original grid region and the adjacent grid region are recorded within the same duration. The failure response variables of the original grid region before and after the load transfer are compared. If the change in the interface peeling driving force or the change in shear strain in the original grid region changes from enhancement to weakening or non-enhancement, it is determined that the failure response variables of the original grid region have disappeared with the load transfer.

[0069] The failure response variables of adjacent grid regions before and after load transfer are compared. If the change in the interface peeling driving force or shear strain of adjacent grid regions changes from non-enhanced to enhanced, and the direction of enhancement is consistent with the direction of the failure response variables formed by the necessary triggering load in the original grid region, then it is determined that the failure response variables of adjacent grid regions are enhanced with load transfer. When the failure response variables of the original grid region decrease with load transfer and the failure response variables of adjacent grid regions are enhanced with load transfer, the propagation correspondence between the original grid region, adjacent grid regions, necessary triggering load, and corresponding failure response variables is determined as an effective propagation relationship.

[0070] S1.5: Based on the effective propagation relationship, the necessary triggering loads, failure response variables and effective propagation relationships corresponding to the same grid are encapsulated in a closed loop to generate a disturbance response fingerprint library.

[0071] Specifically, based on the effective propagation relationship, the necessary triggering loads, failure response variables, original grid region failure response variable fading results, and adjacent grid region failure response variable enhancement results are bound together with the grid region number as the encapsulation index.

[0072] The disturbance response records are generated in the following order: grid region number, necessary triggering load, failure response variable, fading result, enhancement result, and effective propagation relationship. The disturbance response records corresponding to all grid regions are then summarized to generate a disturbance response fingerprint database.

[0073] S2: A load phase-sensitive model is constructed using a perturbation fingerprint embedding layer, a load phase mapping layer, and an interface collision discrimination layer. The failure response principal axis is determined by combining the perturbation response fingerprint library, multi-factor loads are mapped to the failure response principal axis, effective collision loads are identified, and a phase collision load spectrum is generated.

[0074] S2.1: The perturbation fingerprint embedding layer merges the fingerprints of interface peeling response and shear expansion response triggered by different loads within the same grid area based on the failure response variable, and removes isolated response directions triggered by a single load through the main axis pin removal process, generating failure response main axis features.

[0075] Specifically, using the grid region number as the collection object, the necessary triggering loads, failure response variables, and effective propagation relationships within the same grid region are placed into the same grid region record; the interface peeling driving force change in the failure response variables is classified into the interface peeling response according to the interface peeling enhancement direction, and the shear strain change in the failure response variables is classified into the shear propagation response according to the shear propagation enhancement direction.

[0076] Response records with the same grid region number, the same necessary trigger load source, the same interface stripping enhancement direction or the same shear propagation enhancement direction, and which correspond to the original grid region fading result and the adjacent grid region enhancement result in the effective propagation relationship, are merged into the same candidate response direction. If a candidate response direction corresponds to only one type of necessary trigger load and cannot correspond to the adjacent grid region enhancement result in the effective propagation relationship, the candidate response direction is discarded as an isolated response direction. If a candidate response direction corresponds to at least two types of necessary trigger loads and corresponds to the adjacent grid region enhancement result in the effective propagation relationship, the candidate response direction is retained.

[0077] The retained interface stripping enhancement direction and shear expansion enhancement direction are bound according to the grid region number to generate the failure response principal axis feature.

[0078] It should be noted that the principal axis feature of the failure response refers to the dominant failure response direction determined by the interface peeling reinforcement direction and the shear propagation reinforcement direction within the same grid area. It is used to characterize the direction feature that most easily triggers failure propagation under multi-factor loads within the grid area.

[0079] S2.2: The load phase mapping layer establishes a mapping relationship between multi-factor loads and the failure response principal axis based on the failure response principal axis characteristics. Through load rerouting trial processing, it determines whether multi-factor loads can cause the failure response variable to be amplified along the same failure response principal axis, and generates principal axis latched load characteristics.

[0080] Specifically, according to the grid region number, the loads in the multi-factor loads that have the same source as the necessary triggering load, whose action time falls within the same time window, and whose action location corresponds to the same grid region are determined as the initial mapped loads. Based on the effective propagation relationship, the load type, action time, action direction, and action intensity of the initial mapped loads are kept unchanged. The initial mapped loads are transferred from the same grid region to the adjacent grid regions recorded in the effective propagation relationship, and it is re-determined whether the interface peeling driving force change or shear strain change in the adjacent grid regions is still enhanced along the same failure response principal axis.

[0081] If the initial mapped load can cause the failure response variable to amplify along the same failure response principal axis in both the same grid region and adjacent grid regions, then the initial mapped load is determined as the principal axis latching load; if the initial mapped load only causes the failure response variable to amplify in a single grid region, or if the direction of the failure response variable amplification deviates from the same failure response principal axis, then the mapping relationship between the initial mapped load and the failure response principal axis is deleted.

[0082] The spindle latch load, the corresponding grid region number, the corresponding failure response spindle, the action time, and the amplified failure response variable are bound together to generate the spindle latch load feature.

[0083] It should be noted that the spindle latching load characteristic refers to the multi-factor load mapping result that can continuously amplify the failure response variable along the failure response spindle, and is used to characterize the stable correspondence between the load and the failure response spindle.

[0084] S2.3: The interface collision discrimination layer, based on the spindle latched load characteristics, performs collision discrimination on the peeling amplification traces and shear amplification traces on the same failure response spindle of multi-factor loads, and generates interface collision discrimination features.

[0085] Specifically, the interface collision discrimination layer collects the main axis latched loads according to the grid region number, action time, and failure response main axis. The main axis latched loads mapped to the same failure response main axis within the same time window are placed into the same discrimination group. The enhanced interface peeling driving force caused by the main axis latched load is recorded as peeling amplification trace, and the enhanced shear strain caused by the main axis latched load is recorded as shear amplification trace.

[0086] If, within the same discrimination group, at least two types of multi-factor loads both form peeling amplification marks and shear amplification marks along the same principal axis of failure response within the same time window, and the peeling amplification marks and shear amplification marks correspond to the same grid region number or adjacent grid region numbers in the effective propagation relationship, then the corresponding discrimination group is marked as having an interface collision.

[0087] If only peeling amplification marks or only shearing amplification marks exist within the same time window, and the peeling amplification marks and shearing amplification marks do not correspond to the same failure response principal axis, then the corresponding discrimination group is marked as interface collision not established; the multi-factor load, grid region number, action time, failure response principal axis, peeling amplification marks and shearing amplification marks corresponding to interface collision are bound together to generate interface collision discrimination features.

[0088] The failure response principal axis features output by the perturbation fingerprint embedding layer are used as the mapping constraints of the load phase mapping layer. The principal axis latched load features output by the load phase mapping layer are used as the discrimination input of the interface collision discrimination layer. The collision discrimination features formed by the interface collision discrimination layer are used to inversely constrain the principal axis retention and load latching range of the perturbation fingerprint embedding layer and the load phase mapping layer, thus constructing a load phase sensitive model.

[0089] It should be noted that the interface collision discrimination feature refers to the discrimination result used to determine whether multi-factor loads simultaneously form interface peeling amplification and shear amplification on the same failure response principal axis, and is used to distinguish between effective collision loads and pseudo-collision loads.

[0090] The load phase-sensitive model is a finite element load discrimination model based on a perturbation response fingerprint database and calibrated according to rules. It is used to output the failure response principal axis, principal axis latched load characteristics, and interface collision discrimination characteristics.

[0091] S2.4: Read load records pointing to the same interface stripping enhancement direction and the same shear propagation enhancement direction within the same grid area from the disturbance response fingerprint database, and determine the interface stripping enhancement direction and the shear propagation enhancement direction as the failure response principal axis. Perform principal axis mapping on the load records to generate the principal axis triggered load sequence.

[0092] Specifically, based on the perturbation response fingerprint database, load records within the same grid area that can jointly point to the same interface peeling enhancement direction and the same shear propagation enhancement direction are used as training basis. The perturbation fingerprint embedding layer merges the fingerprints of the interface peeling response and shear propagation response, and removes isolated response directions triggered independently by a single load to obtain the failure response principal axis features. The load phase mapping layer maps multi-factor loads to the failure response principal axis, and retains the principal axis latched load features that can continuously amplify along the same failure response principal axis through load rerouting trial. The interface collision discrimination layer performs collision discrimination on peeling amplification traces and shear amplification traces within the same time window and determines the effective collision loads. When the effective collision load determination results of the same training sample are consistent under missing item verification and out-of-order verification, the trained load phase sensitive model is obtained.

[0093] Based on the trained load phase-sensitive model, records containing necessary triggering loads, failure response variables, and effective propagation relationships within the same grid area are filtered using grid area numbers as indices. Records showing an increasing trend in the change of interface peeling driving force in the failure response variables are classified into the interface peeling enhancement direction, and records showing an increasing trend in the change of shear strain in the failure response variables are classified into the shear propagation enhancement direction.

[0094] When the necessary triggering loads within the same grid region simultaneously correspond to the same interface stripping reinforcement direction and the same shear propagation reinforcement direction, and the effective propagation relationship indicates that the same interface stripping reinforcement direction and the same shear propagation reinforcement direction can extend from the same grid region to adjacent grid regions, the corresponding necessary triggering loads, application time, grid region number, interface stripping reinforcement direction, and shear propagation reinforcement direction are determined as load records.

[0095] The interface stripping enhancement direction and shear propagation enhancement direction in the load record are merged into the failure response principal axis, and the load records corresponding to the same failure response principal axis are arranged sequentially according to the action time to generate the principal axis trigger load sequence.

[0096] S2.5: Based on the main axis trigger load sequence, perform missing item verification and out-of-order verification on load combinations mapped to the same failure response main axis within the same time window. If removing any load cannot maintain the synchronous amplification of the failure response variable, and changing the original time order of the load cannot maintain the synchronous amplification of the failure response variable, then it is determined as a valid collision load and is bound accordingly to generate a phase collision load spectrum.

[0097] Specifically, based on the principal axis triggered load sequence, load records are collected according to the same time window, the same grid area number, and the same failure response principal axis to form a load combination; individual loads in the load combination are removed one by one, while keeping the time window, grid area number, and failure response principal axis corresponding to the remaining load unchanged. If the interface peeling enhancement direction and the shear propagation enhancement direction cannot continue to amplify synchronously after any individual load is removed, the load combination is determined to have passed the missing item check.

[0098] When performing out-of-order verification on a load combination, the order of application of various loads in the load combination is adjusted, while keeping the grid region number and the principal axis of the failure response unchanged. If the interface peeling enhancement direction and the shear propagation enhancement direction cannot continue to amplify synchronously after adjusting the order of application, the load combination is determined to have passed the out-of-order verification.

[0099] The load combination that passes both the missing item check and the out-of-order check is determined as the effective collision load. The effective collision load, time window, grid region number, failure response principal axis, interface peeling enhancement direction and shear propagation enhancement direction are bound together to generate the phase collision load spectrum.

[0100] It should be noted that the phase collision load spectrum refers to the load spectrum data formed by binding the effective collision load, phase collision time window, collision grid region, failure response principal axis, interface peeling reinforcement direction and shear propagation reinforcement direction in chronological order. Through the phase collision load spectrum, pseudo-coupled loads that only coincide in time but cannot cause failure propagation can be eliminated, thereby improving the accuracy of phase collision response field calculation and encapsulation material lifetime prediction.

[0101] like Figure 5 This invention describes the process of identifying effective collision loads using a perturbation response fingerprint database and phase collision load spectrum. In the simulation experiment, the same lifetime finite element dataset was used to apply temperature load, humidity load, electrothermal load, and mechanical load completely within each phase collision time window, and the increment of the interface peeling driving force was recorded as a failure response variable. Temperature load removal groups, humidity load removal groups, electrothermal load removal groups, and mechanical load removal groups were constructed separately. Only one type of load was removed each time, while the application time, application area, application direction, and application intensity of the remaining loads remained unchanged. The increment of the interface peeling driving force in the same mesh region was then recalculated. If the increment of the interface peeling driving force decreases relative to the complete multi-factor load after removing a certain type of load, it indicates that the load has a necessary triggering effect on interface peeling enhancement and can be recorded as a necessary triggering load in the perturbation response fingerprint database. If the response remains basically unchanged after removal, it indicates that the load is more likely just a temporally overlapping pseudo-coupling factor. Figure 5 The complete multi-factor load curve is highest near the central phase collision time window, and all removed load curves decrease to varying degrees, especially after the removal of mechanical loads. This indicates that the mechanical load and other loads jointly participate in interface stripping enhancement within this time window. This supports the technical effectiveness of this invention in screening effective collision loads through removal counter-evidence comparison, effective propagation relationship encapsulation, and phase collision load spectrum generation.

[0102] S3: Extract the phase collision time window and collision grid region corresponding to the effective collision load in the phase collision load spectrum, convert the effective collision load within the phase collision time window into finite element coupled boundary conditions, and perform local finite element incremental solution on the collision grid region to generate the phase collision response field.

[0103] S3.1: Select the continuous time period that can cause the interface peeling enhancement and shearing expansion enhancement to occur simultaneously as the candidate collision time period, and perform time node cancellation verification on the time nodes within the candidate collision time period to obtain the phase collision time window.

[0104] Specifically, based on the phase collision load spectrum, the load spectrum records are collected according to the effective collision load, grid region number and failure response principal axis. The continuous time range in which there are effective collision loads in adjacent time nodes and the corresponding failure response principal axis simultaneously maintains interface peeling enhancement and shear propagation enhancement is determined as the candidate collision time period.

[0105] Based on the candidate collision time period, the effective collision load records corresponding to individual time nodes within the candidate collision time period are temporarily cancelled one by one, while the effective collision load records corresponding to the remaining time nodes within the candidate collision time period are retained. Then it is determined whether the interface peeling enhancement and shearing expansion enhancement can still be synchronously maintained on the same failure response axis.

[0106] If the interface stripping enhancement and shear expansion enhancement cannot be maintained synchronously after the effective collision load record corresponding to a certain time node is cancelled, then the corresponding time node is determined as the necessary collision time node; the necessary collision time nodes are connected in chronological order to form a phase collision time window.

[0107] S3.2: Based on the phase collision time window, read the grid region corresponding to the effective collision load, perform interface misalignment verification on the grid region, and determine the grid region that loses synchronous amplification and transmission after misalignment and re-forms synchronous amplification and transmission after returning to its original position as the collision grid region.

[0108] Specifically, based on the phase collision time window and the phase collision load spectrum, within the time range defined by the phase collision time window, candidate mesh regions corresponding to effective collision loads are determined according to the effective collision loads already bound in the phase collision load spectrum, mesh region numbers, failure response principal axes, interface peeling enhancement directions, and shear propagation enhancement directions.

[0109] Based on the interface contact data bound to the grid region number in the finite-lifetime metadata set, the interface position corresponding to the candidate grid region and the interface position adjacent to the candidate grid region are determined. When performing interface misalignment verification on the candidate grid region, the effective collision load, phase collision time window and failure response principal axis remain unchanged. Only the interface position corresponding to the candidate grid region is replaced with the adjacent interface position, and it is re-determined whether the interface peeling enhancement direction and shear propagation enhancement direction can still be synchronously transmitted along the failure response principal axis. If the synchronous amplification transmission disappears after the interface position is replaced, and the synchronous amplification transmission is re-formed after restoring to the interface position before replacement, then the candidate grid region is determined as the collision grid region.

[0110] S3.3: Temperature load is converted into thermal expansion boundary, humidity load is converted into hygroscopic expansion boundary, electrothermal load is converted into local heating boundary, and mechanical load is converted into mechanical equivalent boundary, and is simultaneously applied to the collision mesh region to generate finite element coupled boundary conditions.

[0111] Specifically, based on the phase collision time window, the collision grid region, the effective collision loads already bound in the phase collision load spectrum, the grid region number, the principal axis of the failure response, the interface peeling enhancement direction, and the shear propagation enhancement direction, the effective collision loads acting on the collision grid region within the phase collision time window are determined.

[0112] According to the load type corresponding to the effective collision load, the temperature load is converted into a boundary that generates thermal expansion along the principal axis of the failure response, the humidity load is converted into a boundary that generates hygroscopic expansion along the principal axis of the failure response, the electrothermal load is converted into a boundary that generates local heating along the principal axis of the failure response, and the mechanical load is converted into a boundary that generates bending, tension or compression along the principal axis of the failure response.

[0113] The transformed boundaries are synchronously bound according to the same phase collision time window, the same collision mesh area, and the same failure response principal axis to form principal axis equivalent coupling boundaries. The principal axis equivalent coupling boundaries are applied to the finite element boundary positions corresponding to the collision mesh area, so that the effective collision loads act together along the failure response principal axis within the phase collision time window to generate finite element coupling boundary conditions.

[0114] S3.4: Perform dual-track incremental solution on the coupled boundary conditions, perform differential calculation on the incremental response of the collision when the effective collision load is fully applied and the incremental response of the counter-evidence after removing the single type of effective collision load to obtain the increment of the failure response variable, and filter out the collision-unique response generated only by the combined action of the effective collision load through interface closure verification, and generate the phase collision response field.

[0115] Specifically, the dual-track incremental solution includes collision incremental solution and proof by contradiction incremental solution. Collision incremental solution applies the equivalent coupling boundary of the principal axis corresponding to the effective collision load to the collision mesh region within the phase collision time window to obtain the collision incremental response. Proof by contradiction incremental solution removes the single type of load from the effective collision loads one by one while keeping the phase collision time window, collision mesh region and failure response principal axis unchanged, and applies the equivalent coupling boundary of the principal axis corresponding to the remaining effective collision loads to obtain the proof by contradiction incremental response.

[0116] The temperature gradient change, humidity enrichment change, interface peeling driving force change, and shear strain change in the collision incremental response are respectively differentially calculated with the corresponding changes in the counter-evidence incremental response to obtain the failure response variable increments. When the interface peeling driving force increment and shear strain increment can form a continuous enhancement along the principal axis of the failure response within the collision grid region, and the temperature gradient increment and humidity enrichment increment correspond to the continuous enhancement position, the corresponding failure response variable increment is determined as the collision-unique response.

[0117] The unique collision response is bound according to the phase collision time window, collision grid region and failure response principal axis to generate a phase collision response field.

[0118] S4: Based on the phase collision response field, the synchronous amplification path of the failure response variable is tracked along the collision grid region, and the grid release verification is performed on the collision grid region on the synchronous amplification path to determine the lifetime gate grid and generate a lifetime gate grid chain in series.

[0119] S4.1: Based on the phase collision response field, the grid positions where the interface peeling driving force increment and shear strain increment in the collision grid region are amplified along the failure response principal axis are generated to form candidate synchronous amplification paths. The candidate synchronous amplification paths are then subjected to chain break verification to generate synchronous amplification paths.

[0120] Specifically, based on the phase collision response field, the grid positions in the collision grid region that are simultaneously enhanced along the principal axis of the failure response by the interface peeling driving force increment and shear strain increment are located according to the grid region number, and the grid positions that satisfy the synchronous enhancement are marked as synchronous amplification grid positions.

[0121] Based on the already bound grid region number adjacency relationship and interface contact data in the finite-lifetime metadata set, the synchronous amplification grid positions that are located at the same collision interface, correspond to the same failure response principal axis, and continuously exhibit synchronous enhancement within adjacent phase collision time windows are connected in spatial adjacency order to form candidate synchronous amplification paths.

[0122] The failure response variable transfer relationship between adjacent synchronous amplified grid positions is interrupted segment by segment, while keeping the effective collision load, phase collision time window and failure response principal axis in the phase collision response field unchanged. It is then determined whether the downstream synchronous amplified grid position can still maintain synchronous enhancement of interface peeling driving force increment and shear strain increment after interruption. If the downstream synchronous amplified grid position cannot maintain synchronous enhancement after interruption, the transfer relationship between the corresponding adjacent synchronous amplified grid positions is retained. If the downstream synchronous amplified grid position can still maintain synchronous enhancement after interruption, the transfer relationship between the corresponding adjacent synchronous amplified grid positions is deleted.

[0123] The retained synchronous amplification grid positions and transmission relationships are concatenated in sequence to generate a synchronous amplification path.

[0124] S4.2: Perform collision response stripping on the candidate collision mesh region on the synchronous amplification path, retain the finite element connection relationship of the candidate collision mesh region, and only strip the failure response variable increment generated by the effective collision load.

[0125] Specifically, candidate collision mesh regions are selected according to the order of the synchronous amplification paths, and the finite element connection relationship between the candidate collision mesh regions and the upstream and downstream adjacent mesh regions is kept unchanged based on the adjacent relationship and interface contact data of the mesh regions already bound in the finite lifetime metadata set.

[0126] In the phase collision response field, the increment of failure response variables generated by the effective collision load is located in the candidate collision grid region. The increment of temperature gradient change, humidity enrichment change, interface peeling driving force change and shear strain change are temporarily deducted from the candidate collision grid region. At the same time, the material properties, interface contact relationship and the connection relationship of adjacent grid regions of the candidate collision grid region are retained to form the candidate collision grid region after collision response peeling.

[0127] S4.3: Construct a bypass substitute response based on the stripped candidate collision grid region, determine whether the synchronous amplification path can bypass the candidate collision grid region, obtain the lifetime gate grid, and connect the lifetime gate grids in series to generate a lifetime gate grid chain.

[0128] Specifically, based on the stripped candidate collision grid region, with the upstream and downstream adjacent grid regions located before and after the candidate collision grid region in the synchronous amplification path as the two ends, the interface stripping driving force increment, shear strain increment, temperature gradient increment, and humidity enrichment increment in the phase collision response field of the upstream and downstream adjacent grid regions are kept unchanged. The failure response variable increment of the upstream adjacent grid region is directly transmitted to the downstream adjacent grid region along the failure response principal axis to form a bypass substitute response.

[0129] If the bypass substitute response cannot keep the incremental interface peeling driving force and shear strain of the downstream adjacent grid region synchronously amplified, and the synchronous amplification path is reopened after the failure response variable increment of the candidate collision grid region is restored, then the candidate collision grid region is determined as the lifetime gate grid; according to the order of the lifetime gate grid in the synchronous amplification path, all lifetime gate grids are connected in series to generate a lifetime gate grid chain.

[0130] S5: Perform local damage recursion on the lifetime gate mesh. When a continuous failure path is formed for the first time in the lifetime gate mesh chain, calculate the lifetime measurement value corresponding to the current phase collision time window, and perform closed proof replay on the continuous failure path to obtain the predicted lifetime of the packaging material.

[0131] S5.1: Based on the lifetime gate mesh chain and the phase collision response field, according to the arrangement order of the lifetime gate mesh chain, the interface stripping driving force increment, shear strain increment, temperature gradient increment and humidity enrichment increment of the corresponding lifetime gate mesh within each phase collision time window are written into the same chain sequence record to generate the gate response chain sequence table.

[0132] Specifically, a chain sequence record is established according to the sequential arrangement of each lifetime gate grid in the lifetime gate grid chain, with the lifetime gate grid, phase collision time window, and failure response principal axis as the chain sequence record objects. Within each phase collision time window, the increments of interface peeling driving force, shear strain, temperature gradient, and humidity enrichment belonging to the same lifetime gate grid and the same failure response principal axis in the phase collision response field are included and written into the corresponding chain sequence record according to the arrangement order of the lifetime gate grid chain.

[0133] If the same lifetime gate mesh has the same interface peeling driving force increment, shear strain increment, temperature gradient increment and humidity enrichment increment within the same phase collision time window, it is written into the same chain sequence record. If there is a missing one, the already formed increment is written into the same chain sequence record and the corresponding empty space is reserved. The chain sequence records formed by all lifetime gate meshes in each phase collision time window are summarized according to the arrangement order of the lifetime gate mesh chain to generate the gate response chain sequence table.

[0134] S5.2: Based on the gate response chain sequence table, calculate the corresponding gate penetration trigger amount according to the interface peeling driving force increment, shear strain increment, temperature gradient increment and humidity enrichment increment of the same lifetime gate mesh within the phase collision time window.

[0135] Specifically, based on the gate response chain sequence table, and taking the lifetime gate grid and phase collision time window as the calculation objects, the incremental values ​​of interface peeling driving force, shear strain, temperature gradient, and humidity enrichment corresponding to the same lifetime gate grid are summarized in the gate response chain sequence table.

[0136] Using the maximum increment of interface peeling driving force, maximum increment of shear strain, maximum increment of temperature gradient, and maximum increment of humidity enrichment within all phase collision time windows of the same lifetime gate grid as the same scaling benchmark, the ratio of the current increment of interface peeling driving force to the maximum increment of interface peeling driving force is used as the normalized interface peeling amount; the ratio of the current increment of shear strain to the maximum increment of shear strain is used as the normalized shear strain; the ratio of the current increment of temperature gradient to the maximum increment of temperature gradient is used as the normalized temperature gradient amount; and the ratio of the current increment of humidity enrichment to the maximum increment of humidity enrichment is used as the normalized humidity enrichment amount. When any maximum increment is zero, the corresponding normalization result is recorded as zero.

[0137] When the maximum increment corresponding to the same lifetime gate grid is zero, the corresponding normalized result is recorded as zero. Based on the normalized interface peeling amount, normalized shear strain, normalized temperature gradient, and normalized humidity enrichment amount, the gate penetration triggering amount is calculated, and the expression is:

[0138] ;

[0139] in, This is the gate opening trigger quantity. This represents the normalized interface stripping amount. To normalize the shear strain, This is a normalized temperature gradient quantity. This represents the normalized humidity enrichment level.

[0140] S5.3: Based on the gate penetration trigger quantity, perform local damage recursion and chain sequence relay verification according to the arrangement order of the life gate mesh chain. Only life gate meshes that have reached the penetration judgment threshold and can accept the upstream penetration result are marked as relay penetration meshes, and chain sequence relay penetration records are generated.

[0141] Specifically, the gate penetration trigger amount of the same lifetime gate mesh within the current phase collision time window is accumulated and added to the local damage recursive value formed after the end of the previous phase collision time window for the same lifetime gate mesh, forming the current local damage recursive value; when the current local damage recursive value reaches the penetration judgment threshold (the gate penetration trigger amount is grouped according to the intermittent state, the first penetration state, and the stable state after penetration, and the lowest gate penetration trigger amount that can make the lifetime gate mesh chain close for the first time in the first penetration state is taken as the lower limit, and the highest gate penetration trigger amount that will not be misjudged as non-penetration after misshifting is taken as the upper limit, and the value range is: When the corresponding lifetime gate grid is recorded as the trigger grid, the corresponding lifetime gate grid is recorded as the trigger grid.

[0142] Chain sequence relay verification is performed according to the arrangement order of the life gate grid chain from the head to the tail. When the life gate grid at the head of the chain reaches the continuity judgment threshold, it is directly marked as a relay continuity grid. When the life gate grids after the head of the chain reach the continuity judgment threshold, and the upstream adjacent life gate grid has been marked as a relay continuity grid, and there is a series relationship between the upstream adjacent life gate grid and the current life gate grid in the life gate grid chain, the current life gate grid is marked as a relay continuity grid.

[0143] When the life gate grid after the chain head reaches the breakthrough judgment threshold but cannot accept the relay breakthrough result of the upstream adjacent life gate grid, it is recorded as a non-relay trigger result; the life gate grid number, phase collision time window, current local damage recursion value, breakthrough judgment threshold judgment result, upstream breakthrough result and relay breakthrough mark are written into the same record to generate a chain sequence relay breakthrough record.

[0144] S5.4: In the chain sequence relay connection record, find the phase collision time window that makes the discontinuous connection segment first spliced ​​into a continuous failure path and determine it as a candidate closing key window, and count the lifetime measurement value of the phase collision time window corresponding to the candidate closing key window.

[0145] Specifically, based on the chain sequence relay connection record, the arrangement of relay connection grids in the lifetime gate grid chain is checked according to the order of phase collision time windows. The section formed by the continuous connection of adjacent relay connection grids on the lifetime gate grid chain is recorded as a connection segment, and the state where there are still unconnected grids between the connection segments is recorded as a discontinuous connection segment.

[0146] When a new relay-through mesh is added within the phase collision time window, connecting the previous through segment to the next for the first time, and forming a continuous relay relationship from the beginning to the end of the life gate mesh chain, the phase collision time window is determined as a candidate closed key window. Based on the time window identifier of the candidate closed key window in the chain sequence relay-through record, the life measurement values ​​experienced from the start of the life finite element calculation to the end of the candidate closed key window are statistically analyzed. The life measurement values ​​include the cumulative number of phase collision time windows, the number of temperature cycles, the number of power cycles, and the aging time.

[0147] S5.5: Misalign the gate penetration trigger value in the candidate closing key window to the adjacent time window for re-verification. If the continuous failure path cannot be closed after misalignment and the original time window is restored, the path closing trigger window is obtained, and the lifetime measurement value bound to the path closing trigger window is determined as the predicted lifetime of the packaging material.

[0148] Specifically, the adjacent time windows of the candidate closed key window in the phase collision time window sequence are selected, and the gate connection triggering amount corresponding to each lifetime gate grid in the candidate closed key window is moved as a whole to the adjacent time window to form the chain sequence relay connection record after the shift.

[0149] The chain sequence relay continuity record after the misalignment is reassessed to determine whether the lifetime gate grid chain can still form a continuous failure path from the chain head to the chain tail. When the chain sequence relay continuity record after the misalignment cannot form a continuous failure path, and the continuous failure path can be re-formed after restoring the gate continuity trigger value to the candidate closing key window, it is confirmed that the candidate closing key window has a triggering effect on the closure of the continuous failure path, and the path closure trigger window is obtained. The lifetime measurement value bound to the path closure trigger window is determined as the predicted lifetime of the packaging material.

[0150] like Figure 6This invention demonstrates the error reduction effect of its method compared to traditional lifetime prediction methods. In simulation experiments, a lifetime finite element dataset was established for the same packaging material. Structural geometry data, material property data, interface contact data, and test condition data were kept consistent, and temperature load, humidity load, electrothermal load, and mechanical load were uniformly applied. The maximum stress point method, after solving the complete multi-factor load, directly reads the stress response or interface peeling driving force response of all mesh regions, identifies the single mesh with the largest response as the lifetime bottleneck, and uses the time window when this mesh reaches the failure condition as the predicted lifetime. The overall damage accumulation method statistically analyzes the temperature gradient change, humidity enrichment change, interface peeling driving force change, and shear strain change of all meshes within each phase collision time window, and uses the time window of the region with the largest cumulative damage as the predicted lifetime. This invention generates a disturbance response fingerprint database and a phase collision load spectrum, filters effective collision loads, generates a phase collision response field, tracks the synchronous amplification path, and determines the lifetime gate mesh chain. The predicted lifetime of the packaging material is output after the lifetime gate mesh chain first forms a continuous failure path and is confirmed by closed-loop replay. The curve of the method of the present invention in the figure is lower than that of the two control methods, indicating that it can avoid misjudgment caused by a single maximum stress point or the overall cumulative maximum area, and improve the accuracy of life bottleneck location and life prediction.

[0151] In summary, this invention, by constructing a disturbance response fingerprint database and generating a phase collision load spectrum, can screen out the effective collision loads that truly cause interface peeling enhancement and shear propagation enhancement from multi-factor loads. By determining the lifetime gate mesh chain, lifetime prediction no longer depends on a single maximum stress point, thus improving the accuracy of lifetime bottleneck location and lifetime prediction results in finite element analysis.

[0152] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for predicting the lifetime of packaging materials based on multi-factor analysis, characterized in that, include: A lifetime finite element dataset is constructed for the packaging material and multi-factor loads are applied. By recording failure response variables, a disturbance response fingerprint database is generated. The multi-factor loads include temperature load, humidity load, electrothermal load, and mechanical load; A load phase-sensitive model is constructed by using a perturbation fingerprint embedding layer, a load phase mapping layer, and an interface collision discrimination layer. The failure response principal axis is determined by combining the perturbation response fingerprint database, multi-factor loads are mapped to the failure response principal axis, effective collision loads are identified, and a phase collision load spectrum is generated. Extract the phase collision time window and collision grid region corresponding to the effective collision load in the phase collision load spectrum, convert the effective collision load in the phase collision time window into finite element coupled boundary conditions, and perform local finite element incremental solution on the collision grid region to generate the phase collision response field. Based on the phase collision response field, the synchronous amplification path of the failure response variable is tracked along the collision grid region, and the grid release verification is performed on the collision grid region on the synchronous amplification path to determine the lifetime gate grid and generate a lifetime gate grid chain in series. Local damage recursion is performed on the lifetime gate mesh. When a continuous failure path is first formed in the lifetime gate mesh chain, the lifetime measurement value corresponding to the current phase collision time window is calculated, and the closed proof replay is performed on the continuous failure path to obtain the predicted lifetime of the packaging material.

2. The method for predicting the lifetime of packaging materials based on multi-factor analysis as described in claim 1, characterized in that, The specific steps for constructing a lifetime finite element dataset for the encapsulation material and applying multi-factor loads are as follows: Collect structural geometry data, material property data, interface contact data, and test condition data of the packaging material, and bind the mesh region numbers to generate a lifetime finite element dataset; Based on the lifetime finite element dataset, multi-factor loads are fully applied to the mesh region corresponding to the encapsulation material, and the failure response variables of the mesh region under the action of multi-factor loads are recorded.

3. The method for predicting the lifetime of packaging materials based on multi-factor analysis as described in claim 2, characterized in that, The specific steps for generating the perturbation response fingerprint database are as follows: Based on the failure response variables, one type of load is removed from the multi-factor load in turn and the remaining load is reapplied. The failure response variables before and after the removal are compared by contradiction to obtain the necessary triggering load. The necessary triggering load is transferred from the original grid region to an adjacent grid region and reapplied. The failure response variables of the original grid region are observed to decrease with the load transfer, and the failure response variables of the adjacent grid region are observed to increase with the load transfer, so as to obtain the effective propagation relationship. Based on the effective propagation relationship, the necessary triggering loads, failure response variables and effective propagation relationships corresponding to the same grid are encapsulated in a closed loop to generate a disturbance response fingerprint library.

4. The method for predicting the lifetime of packaging materials based on multi-factor analysis as described in claim 1 or 3, characterized in that, The load phase-sensitive model is constructed by employing a perturbation fingerprint embedding layer, a load phase mapping layer, and an interface collision discrimination layer. The specific steps are as follows: The perturbation fingerprint embedding layer merges fingerprints of interface peeling response and shear expansion response triggered by different loads within the same grid area based on failure response variables, and removes isolated response directions triggered by a single load through the main axis pin removal process, generating failure response main axis features. The load phase mapping layer establishes a mapping relationship between multi-factor loads and the failure response principal axis based on the failure response principal axis characteristics. Through load switching trial processing, it determines whether multi-factor loads can cause the failure response variable to be amplified along the same failure response principal axis, and generates principal axis latched load characteristics. The interface collision discrimination layer, based on the spindle latched load characteristics, performs collision discrimination on the peeling amplification marks and shear amplification marks on the same failure response spindle of multi-factor loads, and generates interface collision discrimination features. The failure response principal axis features output by the perturbation fingerprint embedding layer are used as the mapping constraints of the load phase mapping layer. The principal axis latched load features output by the load phase mapping layer are used as the discrimination input of the interface collision discrimination layer. The collision discrimination features formed by the interface collision discrimination layer are used to inversely constrain the principal axis retention and load latching range of the perturbation fingerprint embedding layer and the load phase mapping layer, thus constructing a load phase sensitive model.

5. The method for predicting the lifetime of packaging materials based on multi-factor analysis as described in claim 4, characterized in that, The specific steps for generating the phase collision load spectrum are as follows: Based on the load phase-sensitive model, load records pointing to the same interface peeling enhancement direction and the same shear propagation enhancement direction within the same grid area are read from the disturbance response fingerprint database. The interface peeling enhancement direction and the shear propagation enhancement direction are determined as the failure response principal axis. The load records are mapped to the principal axis to generate the principal axis triggered load sequence. Based on the principal axis triggered load sequence, missing item verification and out-of-order verification are performed on load combinations mapped to the same failure response principal axis within the same time window. If removing any load cannot maintain the synchronous amplification of the failure response variable, and changing the original time order of the loads cannot maintain the synchronous amplification of the failure response variable, then it is determined to be a valid collision load and is bound accordingly to generate a phase collision load spectrum.

6. The method for predicting the lifetime of packaging materials based on multi-factor analysis as described in claim 5, characterized in that, The specific steps for extracting the phase collision time window and collision grid region corresponding to the effective collision load in the phase collision load spectrum are as follows: The continuous time period that can cause the simultaneous occurrence of interface stripping enhancement and shearing expansion enhancement is taken as the candidate collision time period, and the time node cancellation verification is performed on the time nodes within the candidate collision time period to obtain the phase collision time window. Based on the phase collision time window, the grid region corresponding to the effective collision load is read, and the interface misalignment is checked on the grid region. The grid region that loses synchronous amplification and transmission after misalignment and re-establishes synchronous amplification and transmission after returning to its original position is determined as the collision grid region.

7. The method for predicting the lifetime of packaging materials based on multi-factor analysis as described in claim 1, characterized in that, The specific steps for generating the phase collision response field are as follows: Temperature loads are converted into thermal expansion boundaries, humidity loads into hygroscopic expansion boundaries, electrothermal loads into local heating boundaries, and mechanical loads into mechanically equivalent boundaries, which are simultaneously applied to the collision mesh region to generate finite element coupled boundary conditions. A dual-track incremental solution is performed on the coupled boundary conditions. The incremental collision response when the effective collision load is fully applied is differentially calculated with the incremental response after removing the single type of effective collision load to obtain the increment of the failure response variable. The unique collision response generated only by the combined action of the effective collision load is screened out through the interface through-closed verification, and the phase collision response field is generated.

8. The method for predicting the lifetime of packaging materials based on multi-factor analysis as described in claim 7, characterized in that, The synchronous amplification path for tracing failure response variables along the collision grid region based on the phase collision response field refers to generating candidate synchronous amplification paths by amplifying the interface peeling driving force increment and shear strain increment along the failure response principal axis of the collision grid region based on the phase collision response field, and performing chain break verification on the candidate synchronous amplification paths to generate synchronous amplification paths.

9. The method for predicting the lifetime of packaging materials based on multi-factor analysis as described in claim 8, characterized in that, The specific steps for generating the lifetime gate mesh chain are as follows: Collision response stripping is performed on the candidate collision mesh region on the synchronous amplification path, while preserving the finite element connection relationship of the candidate collision mesh region and only stripping the failure response variable increment generated by the effective collision load. Based on the stripped candidate collision grid region, a bypass substitute response is constructed. It is determined whether the synchronous amplification path can bypass the candidate collision grid region, the lifetime gate grid is obtained, and the lifetime gate grids are connected in series to generate a lifetime gate grid chain.

10. The method for predicting the lifetime of packaging materials based on multi-factor analysis as described in claim 1, characterized in that, The process involves performing local damage recursion on the lifetime gate mesh. When a continuous failure path is first formed in the lifetime gate mesh chain, the lifetime measurement value corresponding to the current phase collision time window is calculated, and a closed-loop reversal playback is performed on the continuous failure path to obtain the predicted lifetime of the encapsulation material. The specific steps are as follows: Based on the lifetime gate mesh chain and the phase collision response field, according to the arrangement order of the lifetime gate mesh chain, the interface stripping driving force increment, shear strain increment, temperature gradient increment and humidity enrichment increment of the corresponding lifetime gate mesh in each phase collision time window are written into the same chain sequence record to generate the gate response chain sequence table. Based on the gate response chain sequence table, the corresponding gate penetration triggering amount is calculated according to the interface stripping driving force increment, shear strain increment, temperature gradient increment and humidity enrichment increment of the same lifetime gate mesh within the phase collision time window. Based on the gate breakthrough trigger quantity, local damage recursion and chain sequence relay verification are performed according to the arrangement order of the life gate grid chain. Only life gate grids that have reached the breakthrough judgment threshold and can accept the upstream breakthrough result are marked as relay breakthrough grids, and chain sequence relay breakthrough records are generated. In the chain sequence relay connection record, the phase collision time window that makes the discontinuous connection segment first spliced ​​into a continuous failure path is identified as a candidate closing key window, and the lifetime measurement value of the phase collision time window corresponding to the candidate closing key window is calculated. The gate penetration trigger value in the candidate closing key window is shifted to the adjacent time window for re-verification. When the continuous failure path cannot be closed after the shift, the original time window is restored and the path closing trigger window is obtained. The lifetime measurement value bound to the path closing trigger window is determined as the predicted lifetime of the packaging material.

Images