A method and system for predicting the thermal fatigue life of cylinder heads
By constructing a physical information sequence learning network and combining multi-source sensor data of the cylinder head with a long short-term memory neural network, the problems of long time consumption of pure physical simulation and data-driven sample dependence in cylinder head thermal fatigue life prediction are solved, and efficient and accurate thermal fatigue life prediction is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING INST OF TECH
- Filing Date
- 2026-02-11
- Publication Date
- 2026-05-26
AI Technical Summary
Existing methods for predicting cylinder head thermal fatigue life suffer from several limitations: purely physical simulations are time-consuming and modeling accuracy depends on empirical corrections; purely data-driven methods require a large number of labeled samples and the models lack physical interpretability; and the combination of physical and statistical regression methods has not achieved deep integration, resulting in technical shortcomings.
By acquiring multi-source sensor time-series data of the cylinder head, extracting feature parameter sequences, constructing physical constitutive equations, and combining them with long short-term memory neural networks, a physical information sequence learning network is constructed to achieve deep integration of physics and data-driven approaches. The network is then trained to improve prediction accuracy and generalization ability.
It can improve prediction accuracy and model generalization with small sample sizes without requiring a large number of labeled samples, achieve deep integration of physical laws and data-driven approaches, improve prediction efficiency and interpretability, and solve the shortcomings of existing technologies.
Smart Images
Figure CN122087576A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of life prediction technology, and in particular to a method and system for predicting the thermal fatigue life of a cylinder head. Background Technology
[0002] The cylinder head is a critical engine component. During operation, it is subjected to complex thermo-mechanical coupled loads, which can easily lead to thermal fatigue cracks, severely affecting engine reliability and service life. Therefore, there is an urgent need for a cylinder head thermal fatigue life prediction method that integrates physical priors and data-driven capabilities, and possesses both high accuracy and strong generalization. Currently, there are two main methods for predicting the thermal fatigue life of cylinder heads: one is prediction using CFD finite element analysis combined with fatigue life models, and the other is pure data-driven prediction based on support vector machines, deep neural networks, etc. Some studies have also attempted to combine physical models with statistical regression, but these are all two-stage fusion methods.
[0003] However, existing technologies have some limitations: pure physical simulation methods are time-consuming for a single analysis, and modeling accuracy relies on empirical corrections. Meanwhile, purely data-driven methods require a large number of labeled samples, resulting in high experimental costs and models lacking physical interpretability and exhibiting weak generalization ability. Furthermore, the combination of physical and statistical regression methods, implemented in two stages, fails to achieve deep integration of the two approaches and still suffers from technical shortcomings. Summary of the Invention
[0004] In view of the shortcomings of the prior art, the purpose of this invention is to provide a method for predicting the thermal fatigue life of cylinder heads, which can solve the problems of long analysis time and reliance on empirical correction for modeling accuracy in pure physical simulation methods. Meanwhile, pure data-driven methods require a large number of labeled samples, resulting in high experimental costs and models lacking physical interpretability and generalization ability. Furthermore, the combination of physical and statistical regression methods, implemented in two stages, does not achieve deep integration of the two, and still suffers from technical shortcomings.
[0005] A first aspect of this invention provides a method for predicting the thermal fatigue life of a cylinder head, comprising: S1: Acquire multi-source sensor timing data of the cylinder head.
[0006] S2: Extract the sequence of characteristic parameters related to thermal fatigue from multi-source sensing time-series data.
[0007] S3: Construct the physical constitutive equations.
[0008] S4: Combine long short-term memory neural networks and physical constitutive equations to construct a physical information sequence learning network.
[0009] S5: Input the feature parameter sequence into the physical information sequence learning network to train the physical information sequence learning network.
[0010] S6: Obtain load data for the operating condition to be predicted.
[0011] S7: Input the load data of the working condition to be predicted into the trained physical information sequence learning network to obtain the prediction result of the thermal fatigue life of the cylinder head.
[0012] A second aspect of the present invention provides a cylinder head thermal fatigue life prediction system, comprising: a processor and a memory.
[0013] The memory stores programs or instructions that can run on a processor, and when the program or instructions are executed by the processor, they implement the steps of the cylinder head thermal fatigue life prediction method of the first aspect.
[0014] A third aspect of the present invention provides a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the cylinder head thermal fatigue life prediction method of the first aspect.
[0015] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this embodiment of the invention, a physical information sequence learning network is used to constrain network learning with prior physical knowledge. This eliminates the need for a large number of labeled samples, improving prediction accuracy and model generalization under small sample conditions, and solving the problems of sample dependence, poor interpretability, and weak generalization ability inherent in purely data-driven systems. Furthermore, the physical information sequence learning network breaks the two-stage fusion model of physical models and statistical regression, embedding physical laws into the entire network training process, achieving integrated fusion rather than a simple splicing application, thus addressing the technical shortcomings of insufficient fusion. Attached Figure Description
[0016] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. Obviously, the drawings described below are merely some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without any creative effort.
[0017] Figure 1 This is a schematic flowchart of a cylinder head thermal fatigue life prediction method provided in an embodiment of the present invention.
[0018] Figure 2 This is a schematic diagram of a cylinder head thermal fatigue life prediction system provided in an embodiment of the present invention. Detailed Implementation
[0019] To enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. It should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0020] The cylinder head thermal fatigue life prediction method provided by the present invention will be described in detail below with reference to the accompanying drawings, through specific embodiments and application scenarios.
[0021] Reference manual attached Figure 1 The diagram shows a flowchart of a cylinder head thermal fatigue life prediction method provided by an embodiment of the present invention.
[0022] This invention provides a method for predicting the thermal fatigue life of a cylinder head, which may include the following steps: S1: Acquire multi-source sensor timing data of the cylinder head.
[0023] S2: Extract the sequence of characteristic parameters related to thermal fatigue from multi-source sensing time-series data.
[0024] Specifically, the characteristic parameter sequence includes: temperature field sequence, combustion chamber pressure sequence, and strain field sequence.
[0025] In this embodiment of the invention, the targeted extraction of feature parameter sequences related to thermal fatigue from multi-source sensing time-series data can effectively eliminate redundant and irrelevant data and simplify the network input dimension. This reduces the computational load of subsequent model training, improves training efficiency, and allows focus on core thermal fatigue influencing factors, making feature input more targeted and effective. This lays a high-quality data foundation for the accurate training of the subsequent physical information sequence learning network, while avoiding the problem of decreased model generalization ability caused by irrelevant data interference.
[0026] S3: Construct the physical constitutive equations.
[0027] Specifically, the physical constitutive equations include: stress-strain constitutive equations containing the thermoplastic behavior of materials, damage evolution equations, and thermal fatigue life criteria.
[0028] In one possible implementation, S3 specifically includes sub-steps S301 to S303: S301: The stress-strain constitutive equations are obtained using the Chaboche nonlinear kinematic hardening model.
[0029] Among them, the Chaboche nonlinear kinematic hardening model is a classic material plastic constitutive model. It is used to accurately describe the nonlinear kinematic hardening characteristics and cyclic plastic deformation behavior of metallic materials under cyclic loading and thermo-mechanical coupled loading. It is also an important model for characterizing the translation of the material yield surface during the loading process and is widely used in stress-strain analysis of mechanical components.
[0030] Specifically, the formula for the stress-strain constitutive equation is as follows: in, Indicates stress, E Indicates the elastic modulus. Indicates total strain. Indicates elastic strain. Indicates plastic strain, Indicates thermal strain, Indicates the increment of plastic strain. s Represents the stress deviatoric tensor. α Represents the back stress tensor. f Represents the yield function. K Indicates the material hardening parameters. n Indicates the material hardening parameters. dt The symbol represents the time increment, and < and > represent Macaulay brackets.
[0031] It should be noted that the stress-strain constitutive equations constructed based on the Chaboche nonlinear kinematic hardening model can accurately describe the nonlinear hardening and cyclic plastic deformation characteristics of cylinder head materials. This closely matches the actual stress and deformation laws of cylinder heads under complex thermo-mechanical coupled loads. Compared with conventional linear models, this model is more in line with engineering practice and provides stress-strain relationship support that conforms to the essence of material mechanics for subsequent embedding of physical constitutive equations into the network, ensuring the scientific nature and accuracy of physical constraints.
[0032] S302: The damage evolution equation is obtained through the Lemaitre continuous damage mechanics model.
[0033] Among them, the Lemaitre continuous damage mechanics model refers to the classical theoretical model of material fatigue damage evolution, which is specifically used to describe the entire process of internal micro-defect initiation, expansion, accumulation and macroscopic fracture of metallic materials under cyclic loading, high temperature and thermo-mechanical coupling.
[0034] Specifically, the formula for the damage evolution equation is as follows: in, Indicates the rate of damage evolution. D Represents damage variables, N Indicates the number of loops. MIndicates material damage parameters, m Indicates material damage parameters.
[0035] It should be noted that the damage evolution equation constructed using the Lemaitre continuous damage mechanics model can quantitatively describe the entire process of micro-defect initiation, expansion and accumulation in materials from the perspective of continuous media. It has clear physical meaning and rigorous theory, and can accurately reflect the thermal fatigue damage development law of cylinder head materials under high temperature and cyclic loading. It provides strong physical constraints for the network and greatly improves the reliability and interpretability of life prediction.
[0036] S303: The thermal fatigue life criterion is obtained through the Manson-Coffin equation.
[0037] Among them, the Manson-Coffin equation is the most classic and widely used empirical-semi-theoretical formula for predicting low-cycle fatigue and thermal fatigue life of metallic materials. It is mainly used to describe the relationship between plastic strain amplitude and the number of fatigue failure cycles of materials, and is particularly suitable for low-cycle thermal fatigue scenarios such as cylinder heads under high-temperature cyclic loading with plastic strain control.
[0038] Specifically, the formula for the thermal fatigue life criterion is as follows: in, Indicates the plastic strain amplitude. Indicates the fatigue ductility coefficient. Indicates the fatigue strength coefficient. Indicates the number of failure cycles. b Indicates fatigue strength index, c It represents the fatigue ductility index.
[0039] It should be noted that the thermal fatigue life criterion established by using the Manson-Coffin equation can directly correlate the plastic strain amplitude with the number of fatigue failure cycles, which is consistent with the failure mechanism of low-cycle thermal fatigue of metallic materials. The formula is simple and has strong engineering applicability, providing a mature and reliable basis for network life determination and effectively improving the engineering practicality and consistency of the prediction results.
[0040] In this embodiment of the invention, by sequentially constructing the Chaboche stress-strain constitutive equation, the Lemaitre damage evolution equation, and the Manson-Coffin thermal fatigue life criterion, a complete and self-consistent thermal fatigue physical constraint system is formed. This system accurately fits the material mechanical behavior and damage failure law under the thermo-mechanical coupled load of the cylinder head, providing rigorous and reliable physical prior knowledge for the subsequent physical information sequence learning network, and effectively improving the model's prediction accuracy, physical interpretability, and generalization ability.
[0041] S4: Combine long short-term memory neural networks and physical constitutive equations to construct a physical information sequence learning network.
[0042] In one possible implementation, S4 specifically involves: using a Long Short-Term Memory (LSTM) neural network as the backbone network of the physical information sequence learning network, and embedding the physical constitutive equation into the backbone network to construct the physical information sequence learning network.
[0043] Specifically, the total loss function is as follows: in, This represents the total loss parameter. This represents the data fitting loss. Represents the residuals of the physical equations. Represents the residuals of the damage evolution equation. This represents the first weighting coefficient. This represents the second weighting coefficient. Represents the residuals of the physical equations. Indicates the weight of the first subtask. Indicates the weight of the second subtask. This represents the square of the norm.
[0044] In this embodiment of the invention, a physical information sequence learning network is constructed by deeply integrating a long short-term memory neural network with physical constitutive equations. This retains the powerful feature extraction and long-term dependency modeling capabilities of LSTM for time-series load data, while introducing strict physical constraints. This allows the model to combine the advantages of data-driven fitting with the rationality of physical laws, significantly improving prediction accuracy, interpretability, and generalization ability. At the same time, it reduces the dependence on a large number of labeled samples, enabling reliable training with small sample sizes.
[0045] S5: Input the feature parameter sequence into the physical information sequence learning network to train the physical information sequence learning network.
[0046] In one possible implementation, S5 specifically includes: The feature parameter sequence is input into the physical information sequence learning network. Under the constraint of monitoring physical consistency, the physical information sequence learning network is trained until the value of the total loss function is less than the preset loss function value.
[0047] It should be noted that those skilled in the art can set the value of the preset loss function according to actual needs, and this invention does not limit this.
[0048] Specifically, the physical consistency constraints for monitoring include: loss monotonicity, stress range constraints, and damage boundaries.
[0049] In this embodiment of the invention, high-quality feature parameter sequences are input into a network that incorporates physical constraints for training. Under the guidance of physical constitutive equations, the solution space is reduced and the convergence speed is accelerated. This reduces the dependence on a large number of samples and ensures that the learning process conforms to the laws of material thermal fatigue mechanics. As a result, the model has high accuracy, strong generalization ability and good physical interpretability, avoiding unreasonable predictions caused by purely data-driven approaches.
[0050] S6: Obtain load data for the operating condition to be predicted.
[0051] S7: Input the load data of the working condition to be predicted into the trained physical information sequence learning network to obtain the prediction result of the thermal fatigue life of the cylinder head.
[0052] In one possible implementation, S7 specifically includes sub-steps S701 to S704: S701: Input the load data of the working condition to be predicted into the trained physical information sequence learning network to obtain the damage prediction value.
[0053] It should be noted that by directly inputting the load data of the working condition to be predicted into the trained physical information sequence learning network, the damage prediction value can be output quickly. While ensuring compliance with the physical laws of thermal fatigue, the prediction efficiency is greatly improved, and complex finite element iterative calculations are avoided, thus achieving efficient, accurate and interpretable damage assessment.
[0054] S702: Compare the damage prediction value and the damage boundary to obtain the failure determination result.
[0055] It should be noted that by directly comparing the damage prediction value with the preset damage boundary, a clear and quantitative determination of cylinder head thermal fatigue failure is achieved. The determination criteria are rigorous and the physical meaning is clear. It can quickly distinguish between safe and failure states, providing a reliable basis for subsequent life prediction and improving the logic and engineering practicality of the entire prediction process.
[0056] S703: Using the Monte Carlo method, the trained physical information sequence learning network is propagated forward multiple times, and the lifetime prediction mean and confidence interval are output.
[0057] The Monte Carlo method is a computational method that uses random sampling and a large number of repeated trials to solve numerical problems, statistical laws, and uncertainties.
[0058] It should be noted that using the Monte Carlo method to perform multiple forward propagation calculations on the network can fully consider the uncertainties of the load and model, output the mean lifetime and confidence interval, making the prediction results more statistically reliable and of engineering reference value. At the same time, it quantifies the fluctuation range of the evaluation results, improving the rigor and credibility of the lifetime assessment.
[0059] S704: Combining the failure determination results, the mean life prediction value, and the confidence interval, the thermal fatigue life prediction results are obtained.
[0060] It should be noted that the final thermal fatigue life prediction result is formed by combining the failure judgment results, the mean of the life prediction, and the confidence interval. This achieves an organic combination of qualitative judgment and quantitative evaluation, ensuring that the results meet the physical failure criteria and have statistical reliability and uncertainty quantification capabilities, thus greatly improving the rigor, engineering applicability, and decision-making reference value of the prediction results.
[0061] In this embodiment of the invention, relying on a well-trained physical information sequence learning network, the complete life result can be output simply by inputting the load data of the working condition to be predicted. The entire process does not require complex finite element simulation and multiple experiments. While ensuring physical rationality and statistical reliability, it achieves efficient, automated and high-precision cylinder head thermal fatigue life assessment, which greatly improves the efficiency of engineering applications and decision support capabilities.
[0062] In this embodiment of the invention, the transmission mode of data packets can be dynamically adjusted according to actual transmission conditions, maximizing the utilization of the available capacity of the transmission window while ensuring that data packet segmentation meets the minimum transmission unit limit, avoiding low transmission efficiency due to excessively small segments. Through priority sorting, high-priority data can be allocated resources first, improving the transmission success rate of critical data and meeting real-time requirements, thereby achieving efficient utilization of transmission resources and accurate guarantee of task priority.
[0063] The cylinder head thermal fatigue life prediction method provided in this application can be executed by a cylinder head thermal fatigue life prediction device. This application uses the cylinder head thermal fatigue life prediction device executing the method as an example to illustrate the cylinder head thermal fatigue life prediction device provided in this application.
[0064] Reference manual attached Figure 2 The diagram shows a structural schematic of a cylinder head thermal fatigue life prediction system provided in an embodiment of the present invention.
[0065] This invention provides a cylinder head thermal fatigue life prediction system 20, including: a processor 201 and a memory 202; The memory 202 stores programs or instructions that can run on the processor 201. When the program or instructions are executed by the processor 201, they implement the steps of the cylinder head thermal fatigue life prediction method described above and achieve the same technical effect. To avoid repetition, the present invention will not elaborate further.
[0066] It should be understood that the processor 201 in this embodiment of the invention may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.
[0067] It should also be understood that the memory 202 in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DR RAM).
[0068] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.
[0069] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
[0070] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0071] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0072] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.
[0073] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0074] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0075] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0076] This invention provides a readable storage medium comprising: storing a program or instructions on the readable storage medium, wherein when the program or instructions are executed by a processor, the program or instructions implement the steps of the above-described cylinder head thermal fatigue life prediction method and achieve the same technical effect. To avoid repetition, this invention will not elaborate further.
[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the protection scope of the present invention.
Claims
1. A method for predicting the thermal fatigue life of a cylinder head, characterized in that, include: S1: Acquire multi-source sensor timing data of the cylinder head; S2: Extract the sequence of feature parameters related to thermal fatigue from the multi-source sensing time-series data; S3: Constructing the physical constitutive equations; S4: Combine the long short-term memory neural network and the physical constitutive equation to construct a physical information sequence learning network; S5: Input the feature parameter sequence into the physical information sequence learning network to train the physical information sequence learning network; S6: Obtain load data for the operating condition to be predicted; S7: Input the load data of the working condition to be predicted into the trained physical information sequence learning network to obtain the prediction result of the thermal fatigue life of the cylinder head.
2. The method for predicting the thermal fatigue life of a cylinder head according to claim 1, characterized in that, The characteristic parameter sequence specifically includes: temperature field sequence, combustion chamber pressure sequence, and strain field sequence.
3. The method for predicting the thermal fatigue life of a cylinder head according to claim 1, characterized in that, The physical constitutive equations specifically include: stress-strain constitutive equations incorporating the thermoplastic behavior of materials, damage evolution equations, and thermal fatigue life criteria.
4. The method for predicting the thermal fatigue life of a cylinder head according to claim 3, characterized in that, S3 specifically includes: S301: The stress-strain constitutive equation is obtained using the Chaboche nonlinear kinematic hardening model; S302: The damage evolution equation is obtained through the Lemaitre continuous damage mechanics model; S303: The thermal fatigue life criterion is obtained through the Manson-Coffin equation.
5. The method for predicting the thermal fatigue life of a cylinder head according to claim 1, characterized in that, Specifically, S4 involves using a Long Short-Term Memory (LSTM) neural network as the backbone of the physical information sequence learning network, and embedding the physical constitutive equation into the backbone network to construct the physical information sequence learning network.
6. The method for predicting the thermal fatigue life of a cylinder head according to claim 1, characterized in that, Specifically, S5 is: The feature parameter sequence is input into the physical information sequence learning network. Under the constraint of monitoring physical consistency, the physical information sequence learning network is trained until the value of the total loss function is less than the preset loss function value.
7. The method for predicting the thermal fatigue life of a cylinder head according to claim 6, characterized in that, The specific physical consistency constraints for monitoring include: loss monotonicity, stress range constraints, and damage boundaries.
8. The method for predicting the thermal fatigue life of a cylinder head according to claim 1, characterized in that, Specifically, S7 includes: S701: Input the load data of the working condition to be predicted into the trained physical information sequence learning network to obtain the damage prediction value; S702: Compare the predicted damage value with the damage boundary to obtain the failure determination result; S703: The trained physical information sequence learning network is propagated multiple times using the Monte Carlo method, and the mean lifetime prediction and confidence interval are output. S704: Combining the failure determination result, the mean of the life prediction, and the confidence interval, the thermal fatigue life prediction result is obtained.
9. A cylinder head thermal fatigue life prediction system, characterized in that, include: Processor and memory; The memory stores programs or instructions that can run on the processor, which, when executed by the processor, implement the steps of the cylinder head thermal fatigue life prediction method as described in any one of claims 1 to 8.
10. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the cylinder head thermal fatigue life prediction method as described in any one of claims 1 to 8.