A Diesel Engine Structural Design Method Based on Digital Twin

Abstract

This invention discloses a diesel engine structural design method based on digital twins, comprising the following steps: establishing a digital twin model and initializing an improved pollen algorithm; collecting real-time operating condition data to generate boundary conditions for the twin model; performing population evolution of the algorithm, selecting elite individuals for mutation and crossover, and dynamically adjusting the search step size; driving twin model simulation based on boundary conditions and design parameters, and calculating fitness; dynamically adjusting algorithm parameters based on fitness feedback to generate new design parameters; iteratively verifying until convergence, and outputting the optimal parameters; updating the twin model to generate the optimized structure. This invention achieves a dual improvement in the accuracy and efficiency of diesel engine structural design, and can adapt to changes in operating conditions.

Description

Technical Field

[0001] This invention relates to the field of engineering design optimization technology, and in particular to a diesel engine structural design method based on digital twins. Background Technology

[0002] With the development of modern industrial technology, the design optimization of mechanical equipment has become an important means to improve equipment performance and working efficiency. Especially in the field of engine design, designers can simulate and optimize various engine components, such as cylinders, combustion chambers, and exhaust systems, using simulation technologies such as computer-aided design, computational fluid dynamics, and finite element analysis. These traditional simulation technologies can provide certain predictions and assessments during the design phase, providing a theoretical basis for manufacturing and subsequent debugging. However, existing simulation technologies still have some limitations, particularly in handling complex and variable operating conditions, where the accuracy and adaptability of the design are relatively poor, and they cannot fully consider the many variable factors that may be encountered in actual operation.

[0003] Although there are design methods based on optimization algorithms that can search and optimize the design space to some extent, they usually rely on static assumptions and cannot be dynamically adjusted according to actual working conditions.

[0004] Digital twin technology, as an emerging technology, enables dynamic monitoring and optimization of physical systems through real-time data feedback and synchronous updates of virtual models. However, existing digital twin technologies are mostly used for monitoring and maintenance during equipment operation, lacking applications that combine them with design optimization algorithms. In current technologies, the integration of digital twins and optimization algorithms remains relatively preliminary.

[0005] Therefore, how to provide a diesel engine structural design method based on digital twins is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] One objective of this invention is to propose a diesel engine structural design method based on digital twins. This invention fully utilizes digital twin technology and advanced optimization algorithms, and optimizes diesel engine design parameters by combining real-time operating data with simulation models. It has the advantages of efficient dynamic adjustment capability, high optimization accuracy, and adaptability to complex operating condition changes.

[0007] A diesel engine structure design method based on digital twin according to an embodiment of the present invention includes the following steps: Establish digital twin models of key components of diesel engines, set model boundary conditions and material property parameters, and initialize the improved flower pollination algorithm population; Collect operating condition data, transmit the operating condition data to the digital twin model, and generate real-time operating condition boundary conditions; The population evolution operation of the improved flower pollination algorithm is implemented. Elite individuals are selected from the current population based on fitness. Mutation and crossover operations are performed on the elite individuals to generate offspring population. The global search step size and local search step size are dynamically adjusted according to the population distribution status. Based on the real-time working condition boundary conditions and the structural design parameters corresponding to the current population individuals, the digital twin model is driven to perform simulation calculations to obtain performance response data. Based on the performance response data and the preset design target, the fitness value of the population individuals is calculated. The fitness value is transmitted as feedback information to the improved flower pollination algorithm. The conversion probability, mutation rate and crossover rate are dynamically adjusted according to the fitness value. Based on the adjusted parameters, the population is controlled to perform the next global search or local search, and new structural design parameters are generated. The new structural design parameters are input into the digital twin model to perform iterative verification until the convergence condition is met, then the iteration is stopped and the optimal structural design parameters are output. The optimal structural design parameters are mapped to the digital twin model, the diesel engine model structure is updated, and an optimized digital twin model of the diesel engine structure is generated.

[0008] Optionally, the establishment of the digital twin model and the initialization of the flower pollination algorithm population specifically include: Obtain three-dimensional geometric models of key components of the diesel engine, model the key components using computer-aided design software, and generate a preliminary framework for a digital twin model. Based on the physical characteristics of each component of the diesel engine, the material property parameters of each component are set, and the selection is made according to the actual application requirements; Set the boundary conditions for the digital twin model, and input the boundary conditions based on the engine's working environment and operating conditions; Based on the established material properties and boundary conditions, the digital twin model is simulated using finite element analysis and computational fluid dynamics tools to calculate the performance of components under different working conditions. Initialize the improved flower pollination algorithm population, set preliminary optimization parameters according to the design goal, define the search space of design parameters, and select the structural design parameters of the initial population individuals.

[0009] Optionally, the generation of real-time operating condition boundary conditions specifically includes: Multi-source operating condition data are collected by a sensor array deployed on a physical prototype of the diesel engine; The collected multi-source operating condition data is subjected to time synchronization and filtering and noise reduction processing to remove outliers and generate a synchronized operating condition data sequence. Transmit the synchronous operating condition data sequence to the data interface of the digital twin model; parse the synchronous operating condition data sequence in the digital twin model and extract key feature parameters; input the key feature parameters as load conditions for the simulation calculation of the digital twin model to the corresponding load interface of the solver; The digital twin model performs simulation calculations based on the input load conditions and outputs the distribution field data of diesel engine components under the current operating conditions.

[0010] Optionally, the improved flower pollination algorithm specifically includes: For each individual in the current population, a random number is generated. When the random number is less than the preset conversion probability, a global pollination operation is performed, a global pollination step size is generated, and the individual position is updated based on the current global best individual. When the random number is greater than or equal to the preset conversion probability, a local pollination operation is performed, two other individuals are randomly selected from the current population, and a local pollination step size is generated to update the individual position. All the updated individuals constitute a temporary population. Calculate the fitness value of each individual in the temporary population; Merge the current population with the temporary population, and select the top few individuals by fitness value as elite individuals; Two individuals are randomly selected from the elite individuals to generate offspring individuals; random perturbations are superimposed on the parameters of each dimension to generate a set of offspring individuals; the current population, the temporary population and the set of offspring individuals are merged, and individuals with the highest fitness values ​​and the same number of individuals as the population size are selected to form the next generation population. Calculate the fitness variance of the next generation population and adjust the global search step size based on the fitness variance, while decreasing the local search step size; The individual with the best fitness value from the next generation of the population is selected as the current global best individual.

[0011] Optionally, the calculation of the fitness value of an individual in the population specifically includes: Extract the structural design parameters corresponding to each individual in the current population in sequence, and input the structural design parameters and real-time working condition boundary conditions into the digital twin model. The digital twin model updates the geometric features and material properties of components based on structural design parameters, applies load boundaries at corresponding locations based on real-time working condition boundary conditions, performs multiphysics coupled simulation calculations, and outputs performance response data; Read the preset design target file, which contains the maximum allowable stress threshold and the highest operating temperature threshold for each component; The performance response data is compared with the preset design target item by item, and the ratio between the actual simulation value of each indicator and the target threshold is calculated as the degree of compliance. The fitness values ​​of each achievement level are weighted and summed, and the weighted sum is used as the fitness value of the current individual under the current working conditions.

[0012] Optionally, the dynamic adjustment of the fitness value to the transition probability, mutation rate, and crossover rate specifically includes: The fitness values ​​of individuals in the current population are statistically analyzed, and the conversion probability, mutation rate, and crossover rate are dynamically adjusted based on the fitness improvement rate, the ratio of the population fitness variance to the current best fitness value, and the proportion of elites. The fitness improvement rate is the relative increase in the optimal fitness value between adjacent iterations, and the elite ratio is the proportion of individuals in the current population whose fitness value is higher than the average fitness value of several historical iterations to the population size. The direction and magnitude of the conversion probability are set based on the fitness improvement rate; the direction and magnitude of the mutation rate are set based on the ratio of the population fitness variance to the current best fitness value; and the direction and magnitude of the crossover rate are set based on the elite ratio. The adjustment ranges for the transformation probability, mutation rate, and crossover rate are set separately. The parameters after each adjustment are limited, and the adjusted parameters are used for the pollination operation in the next iteration.

[0013] Optionally, the step of inputting the new structural design parameters into the digital twin model to perform iterative verification specifically includes: After generating new structural design parameters in each iteration, a preset number of individuals with fitness values ​​ranked before the current population are selected as candidate individuals, and the structural design parameters of the candidate individuals are extracted as parameters to be verified. Input the parameters to be verified into the digital twin model to perform multiphysics simulation calculations and output the corresponding simulation performance data; Read the measured performance data of the physical prototype with the same boundary conditions as the current real-time operating conditions from the historical database, compare the simulated performance data with the measured performance data item by item, and calculate the absolute value of the relative error of each performance index; When the absolute value of the relative error of any performance index exceeds the preset accuracy threshold, the back propagation algorithm is used to correct the corresponding material property parameters or boundary condition correction coefficients in the digital twin model based on the relative error value, generating an updated digital twin model, and the updated digital twin model is used for subsequent iterations of simulation calculations. Repeat the above correction process until the absolute values ​​of the relative errors of all performance indicators converge to within the preset accuracy threshold. Determine whether the difference between the optimal fitness value in the current population and the preset design target threshold is less than the convergence tolerance. When both conditions are met, stop the iteration and output the structural design parameters corresponding to the current optimal fitness value as the final optimization result.

[0014] Optionally, the digital twin model of the diesel engine structure specifically includes: Geometric models of key diesel engine components are constructed based on optimal structural design parameters. The optimal structural design parameters are the combination of structural parameters that meet the preset design objectives and are output after iterative optimization by an improved flower pollination algorithm. The material property field distribution associated with the geometric model is based on the optimal structural design parameters to assign differentiated values ​​to the materials of the components, forming a non-uniform material distribution characteristic. The sensor virtual deployment nodes are embedded in the geometric model, and the arrangement positions of the sensor virtual deployment nodes correspond one-to-one with the sensor array of the physical prototype. The load mapping interface is dynamically associated with real-time operating condition boundary conditions. The load mapping interface receives real-time operating condition data collected and processed by the physical prototype and converts the real-time operating condition data into dynamic load boundaries for simulation calculation of the digital twin model. The parameter feedback interface is linked with the improved flower pollination algorithm. The parameter feedback interface receives the structural design parameters generated by the algorithm iteration and updates the geometric model. It also sends the performance response data of each simulation calculation back to the improved flower pollination algorithm.

[0015] The beneficial effects of this invention are: (1) By using digital twin technology, various data during the operation of the diesel engine are collected and fed back in real time, and the design parameters are dynamically adjusted, so that the design can be more accurately adapted to the actual working conditions and avoid errors and deviations under static assumptions in traditional design methods.

[0016] (2) By combining the improved pollen propagation algorithm with global and local search, the efficiency of the optimization process is improved. It can quickly converge to the global optimum in a complex design space, reducing the number of design iterations and shortening the R&D cycle.

[0017] (3) Through the closed-loop feedback mechanism of real-time feedback and optimization algorithm, the coupling effect between multiple physical fields can be fully considered in the design stage, which can improve the performance of the engine in terms of fuel efficiency, power output, emission control and other aspects under various working conditions, and increase the stability and reliability of the engine. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a diesel engine structural design method based on digital twin proposed in this invention; Figure 2 This is a schematic diagram illustrating the fitness calculation of a diesel engine structural design method based on digital twins proposed in this invention. Figure 3This is a flowchart of an improved flower pollination algorithm for a diesel engine structural design method based on digital twins proposed in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figures 1-3 A diesel engine structural design method based on digital twins includes the following steps: Establish digital twin models of key components of diesel engines, set model boundary conditions and material property parameters, and initialize the improved flower pollination algorithm population; Collect operating condition data, transmit the operating condition data to the digital twin model, and generate real-time operating condition boundary conditions; The population evolution operation of the improved flower pollination algorithm is implemented. Elite individuals are selected from the current population based on fitness. Mutation and crossover operations are performed on the elite individuals to generate offspring population. The global search step size and local search step size are dynamically adjusted according to the population distribution status. Based on the real-time working condition boundary conditions and the structural design parameters corresponding to the current population individuals, the digital twin model is driven to perform simulation calculations to obtain performance response data. Based on the performance response data and the preset design target, the fitness value of the population individuals is calculated. The fitness value is transmitted as feedback information to the improved flower pollination algorithm. The conversion probability, mutation rate and crossover rate are dynamically adjusted according to the fitness value. Based on the adjusted parameters, the population is controlled to perform the next global search or local search, and new structural design parameters are generated. The new structural design parameters are input into the digital twin model to perform iterative verification until the convergence condition is met, then the iteration is stopped and the optimal structural design parameters are output. The optimal structural design parameters are mapped to the digital twin model, the diesel engine model structure is updated, and an optimized digital twin model of the diesel engine structure is generated.

[0021] In this embodiment, establishing the digital twin model and initializing the flower pollination algorithm population specifically includes: Obtain three-dimensional geometric models of key components of the diesel engine, including cylinders, pistons, crankshafts, combustion chambers, turbochargers, etc., and model the key components using computer-aided design software to generate a preliminary framework of digital twin models. Based on the physical characteristics of each component of the diesel engine, the material property parameters of each component are set, including the material's density, elastic modulus, thermal conductivity, strength, high temperature resistance, etc., and selected according to the actual application requirements, such as using aluminum alloy for the cylinder and 4032 aluminum alloy for the piston. Set the boundary conditions for the digital twin model. Based on the engine's working environment and operating conditions, including temperature, pressure, fluid velocity, etc., input the boundary conditions, such as the temperature range of the cylinder wall and the exhaust temperature of the turbocharger. Based on the established material properties and boundary conditions, the digital twin model is simulated using finite element analysis and computational fluid dynamics tools to calculate the stress, heat conduction, airflow distribution and other properties of the components under different working conditions. The boundary conditions are the temperature and pressure ranges that each component may experience, which are estimated by historical data or long-term operating data under actual working conditions. Initialize the improved flower pollination algorithm population, set the population size, maximum number of iterations and initial random distribution of the population, set preliminary optimization parameters according to the design goal, define the search space of design parameters, and select the structural design parameters of the initial population individuals, including cylinder size, fuel injection quantity and injection timing.

[0022] In this embodiment, generating real-time operating condition boundary conditions specifically includes: Multi-source operating condition data are collected by a sensor array deployed on a physical prototype of the diesel engine; The collected multi-source operating condition data is subjected to time synchronization and filtering and noise reduction processing to remove outliers and generate a synchronized operating condition data sequence. The synchronous operating condition data sequence is transmitted to the data interface of the digital twin model; the synchronous operating condition data sequence is parsed in the digital twin model to extract key feature parameters including temperature, pressure, and vibration amplitude; the key feature parameters are used as load conditions for simulation calculation of the digital twin model and input to the corresponding load interface of the solver. The digital twin model performs simulation calculations based on the input load conditions and outputs the distribution field data of diesel engine components under the current operating conditions, including temperature field, stress field, and strain field.

[0023] In this embodiment, the improved flower pollination algorithm specifically includes: For each individual in the current population, a random number is generated. When the random number is less than the preset transition probability, a global pollination operation is performed, generating a global pollination step size and updating the individual's position based on the current global best individual. When the random number is greater than or equal to the preset transition probability, a local pollination operation is performed, randomly selecting two other individuals from the current population and generating a local pollination step size to update the individual's position. All updated individuals constitute a temporary population. The population size is set to 100; the maximum number of iterations is set to 300; and the preset transition probability p is set to 0.7. Calculate the fitness value of each individual in the temporary population; The current population is merged with the temporary population, and the top-ranked individuals by fitness value are selected as elite individuals; in this implementation, the top 20% are selected. Two individuals are randomly selected from the elite individuals to generate offspring individuals; random perturbations are superimposed on the parameters of each dimension to generate a set of offspring individuals; the current population, temporary population, and offspring individual set are merged, and individuals with the highest fitness values ​​and the same number of individuals as the population size are selected to form the next generation population; the crossover operation sets the crossover probability to 0.8, that is, the crossover rate is 0.8, and two individuals are randomly selected from the elite individuals each time to perform discrete crossover with this probability; random perturbation is performed by superimposing random perturbations that follow a Gaussian distribution on the parameters of each dimension. The mean of the Gaussian distribution is 0, and the standard deviation is set to 0.1 times the standard deviation of the parameter of that dimension in the current population, that is, the mutation rate is 0.1; The fitness variance of the next generation population is calculated, and the global search step size is adjusted according to the fitness variance, while the local search step size is decreased. In this embodiment, the fitness variance of the next generation population is calculated. When the fitness variance is greater than a first preset threshold, the global search step size is multiplied by a first scaling factor. When the fitness variance is less than a second preset threshold, the global search step size is multiplied by a second scaling factor. The local search step size is linearly decreased according to the ratio of the current iteration number to the maximum iteration number. The first preset threshold is set to 0.001, and the second threshold is set to 0.0001. The first scaling factor is set to 1.1, and the second scaling factor is set to 0.9. The initial value of the local search step size is set to 0.1, and the termination value is set to 0.01. It is updated according to a linear decreasing law, that is, the current iteration step size is equal to the initial step size multiplied by (1 minus the ratio of the current iteration number to the maximum iteration number). The individual with the best fitness value from the next generation of the population is selected as the current global best individual.

[0024] In this embodiment, calculating the fitness value of an individual in the population specifically includes: Extract the structural design parameters corresponding to each individual in the current population in sequence, and input the structural design parameters and real-time working condition boundary conditions into the digital twin model. The digital twin model updates the geometric features and material properties of components based on structural design parameters, applies load boundaries at corresponding locations based on real-time working condition boundary conditions, performs multiphysics coupled simulation calculations, and outputs performance response data; Read the preset design target file, which contains the maximum allowable stress threshold and the highest operating temperature threshold for each component; The performance response data is compared with the preset design target item by item, and the ratio between the actual simulation value of each indicator and the target threshold is calculated as the degree of compliance. The fitness values ​​of each achievement level are weighted and summed, and the weighted sum is used as the fitness value of the current individual under the current working conditions.

[0025] In this embodiment, the dynamic adjustment of the fitness value for the transition probability, mutation rate, and crossover rate specifically includes: The fitness values ​​of individuals in the current population are statistically analyzed, and the conversion probability, mutation rate, and crossover rate are dynamically adjusted based on the fitness improvement rate, the ratio of the population fitness variance to the current best fitness value, and the proportion of elites. The fitness improvement rate is the relative increase in the optimal fitness value between adjacent iterations, and the elite ratio is the proportion of individuals in the current population whose fitness value is higher than the average fitness value of several historical iterations to the population size. The direction and magnitude of the conversion probability are set based on the fitness improvement rate; the direction and magnitude of the mutation rate are set based on the ratio of the population fitness variance to the current best fitness value; and the direction and magnitude of the crossover rate are set based on the elite ratio. The adjustment ranges for the transformation probability, mutation rate, and crossover rate are set separately. The parameters after each adjustment are limited, and the adjusted parameters are used for the pollination operation in the next iteration.

[0026] In this embodiment, the specific implementation method for dynamically adjusting the transition probability, mutation rate, and crossover rate based on the fitness value is as follows: When the fitness improvement rate is greater than the preset threshold of 0.05, it indicates that the algorithm is converging rapidly. The conversion probability should be appropriately reduced to enhance local exploitation capability, and the coefficient should be 0.95. When the fitness improvement rate is less than the preset threshold of 0.01 and remains less than the threshold for 5 consecutive rounds, the conversion probability should be increased, and the coefficient should be 1.05. The adjustment range of the conversion probability is limited to between 0.4 and 0.9. When the ratio of the population fitness variance to the current best fitness is less than 0.01, it indicates that the population is becoming homogeneous, so the mutation rate should be increased, and a coefficient of 1.1 should be used. When the ratio of the population fitness variance to the current best fitness is greater than the threshold of 0.1, it indicates that the population is too dispersed, so the mutation rate should be reduced, and a coefficient of 0.9 should be used. The mutation rate should be limited to a range of 0.01 to 0.3. When the proportion of elites is less than 0.2, it indicates that there are few outstanding individuals, and it is necessary to strengthen information exchange and increase the crossover rate, using a coefficient of 1.1; when the proportion of elites is greater than 0.6, it indicates that there are many outstanding individuals, and it is necessary to appropriately reduce the crossover rate to protect outstanding genes, using a coefficient of 0.95; the crossover rate is limited to a range of 0.3 to 0.9.

[0027] In this embodiment, inputting the new structural design parameters into the digital twin model to perform iterative verification specifically includes: After generating new structural design parameters in each iteration, a predetermined number of individuals with fitness values ​​ranked first in the current population are selected as candidate individuals, and the structural design parameters of the candidate individuals are extracted as parameters to be verified; the number of individuals is set to be the top 10% of the individuals with fitness values ​​ranked first. Input the parameters to be verified into the digital twin model to perform multiphysics simulation calculations and output the corresponding simulation performance data; Read the measured performance data of the physical prototype with the same boundary conditions as the current real-time operating conditions from the historical database, compare the simulated performance data with the measured performance data item by item, and calculate the absolute value of the relative error of each performance index; When the absolute value of the relative error of any performance index exceeds the preset accuracy threshold, the corresponding material property parameters or boundary condition correction coefficients in the digital twin model are corrected using the backpropagation algorithm based on the relative error value, generating an updated digital twin model, which is then used for subsequent iterations of simulation calculations; the preset accuracy threshold is set to 5%. Repeat the above correction process until the absolute values ​​of the relative errors of all performance indicators converge to within the preset accuracy threshold. Then, determine if the difference between the optimal fitness value and the preset design target threshold in the current population is less than the convergence tolerance. When both conditions are met, stop the iteration and output the structural design parameters corresponding to the current optimal fitness value as the final optimization result. The convergence tolerance for the difference between the optimal fitness value and the preset design target threshold is set to 0.001. Repeat the above correction process until the absolute values ​​of the relative errors of all performance indicators are less than 5% for three consecutive generations, and the difference between the optimal fitness value and the preset design target threshold is less than 0.001. Then, stop the iteration and output the result.

[0028] In this embodiment, the digital twin model of the diesel engine structure specifically includes: Geometric models of key diesel engine components are constructed based on optimal structural design parameters. The optimal structural design parameters are the combination of structural parameters that meet the preset design objectives and are output after iterative optimization by an improved flower pollination algorithm. The material property field distribution associated with the geometric model is based on the optimal structural design parameters to assign differentiated values ​​to the materials of the components, forming a non-uniform material distribution characteristic. The sensor virtual deployment nodes are embedded in the geometric model, and the arrangement positions of the sensor virtual deployment nodes correspond one-to-one with the sensor array of the physical prototype. The load mapping interface is dynamically associated with real-time operating condition boundary conditions. The load mapping interface receives real-time operating condition data collected and processed by the physical prototype and converts the real-time operating condition data into dynamic load boundaries for simulation calculation of the digital twin model. The parameter feedback interface is linked with the improved flower pollination algorithm. The parameter feedback interface receives the structural design parameters generated by the algorithm iteration and updates the geometric model. It also sends the performance response data of each simulation calculation back to the improved flower pollination algorithm.

[0029] Example 1: To verify the feasibility of this invention in practice, it was applied to the piston structure optimization design of a certain type of six-cylinder turbocharged diesel engine. This type of diesel engine is widely used in heavy transportation equipment and operates under complex and variable working conditions for extended periods, resulting in problems such as excessive piston heat load and insufficient fatigue life due to localized stress concentration. Traditional design methods rely on empirical formulas and static simulations, leading to long design cycles and difficulty in achieving optimal performance across the entire operating range. Often, optimization at one operating point results in poor performance under other operating conditions. This invention effectively solves the above problems by constructing a digital twin model to map the piston's operating state in real time and employing an improved flower pollination algorithm for global optimization.

[0030] In this application scenario, a digital twin model of the piston is first established. Based on the piston's three-dimensional geometry, material properties and boundary conditions are set, with the boundary conditions determined according to typical diesel engine operating conditions. An improved flower pollination algorithm is initialized, setting initial parameters such as population size, maximum number of iterations, conversion probability, mutation rate, and crossover rate. The algorithm uses five structural parameters—piston compression height, combustion chamber narrowing rate, fire shore height, pin hole eccentricity, and top wall thickness—as design variables, and piston maximum temperature, maximum equivalent stress, and fatigue life safety factor as optimization objectives. The weighted sum of these objectives is used as the fitness value.

[0031] Real-time operating data is acquired through a sensor array deployed on the physical prototype, including temperature sensors, strain sensors, acceleration sensors, and pressure sensors. After time synchronization and filtering, the acquired data generates real-time operating boundary conditions, which are input into the digital twin model to update load boundaries and drive simulation calculations. In each iteration, each individual in the algorithm population drives the digital twin model to perform thermo-mechanical coupling simulation, outputting the piston's temperature and stress fields, and then calculating the fitness value. Based on fitness feedback, the algorithm dynamically adjusts the transition probability, mutation rate, crossover rate, and search step size to generate a new generation of design parameters until convergence conditions are met.

[0032] After iterative optimization, the algorithm converges and outputs the optimal design parameters. Mapping these optimal parameters back to the digital twin model yields the optimized piston structure. Comparing the initial design and the optimized design, simulations were performed under the same typical operating conditions. The main performance indicators are shown in Table 1. The data shows that the optimized piston exhibits significantly reduced maximum temperature and maximum stress, improved fatigue life safety factor, and a slight improvement in fuel consumption. The optimization process was significantly faster, improving design efficiency.

[0033] Table 1: Performance Comparison Before and After Piston Structure Optimization

[0034] As shown in Table 1, the method proposed in this invention can optimize the piston structure, enabling key performance indicators to basically reach or exceed the preset target thresholds. The maximum piston temperature decreased from 362.7℃ to 350.4℃, below the target of 350℃, reducing the risk of thermal load; the maximum equivalent stress decreased from 215.3MPa to 197.2MPa, below 200MPa, reducing stress concentration; the fatigue life safety factor increased from 1.82 to 2.06, exceeding the threshold of 2.0, meaning that the expected life of the piston under rated operating conditions is extended. The fuel consumption rate decreased by 1.6g / kW·h, which also makes a positive contribution to the overall economic efficiency of the machine. During the optimization process, the algorithm adaptively adjusts the search step size and control parameters, avoiding getting trapped in local optima, and achieves stable convergence within a limited number of iterations, taking only 28 hours, demonstrating high efficiency. Compared with traditional methods that rely on empirical trial and error and single-point simulation, this invention not only improves design quality but also shortens the development cycle, proving its feasibility and superiority in practical engineering.

[0035] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A diesel engine structural design method based on digital twins, characterized in that, Includes the following steps: Establish digital twin models of key components of diesel engines, set model boundary conditions and material property parameters, and initialize the improved flower pollination algorithm population; Collect operating condition data, transmit the operating condition data to the digital twin model, and generate real-time operating condition boundary conditions; The population evolution operation of the improved flower pollination algorithm is implemented. Elite individuals are selected from the current population based on fitness. Mutation and crossover operations are performed on the elite individuals to generate offspring population. The global search step size and local search step size are dynamically adjusted according to the population distribution status. The improved flower pollination algorithm specifically includes: For each individual in the current population, a random number is generated. When the random number is less than the preset conversion probability, a global pollination operation is performed, a global pollination step size is generated, and the individual position is updated based on the current global best individual. When the random number is greater than or equal to the preset conversion probability, a local pollination operation is performed, two other individuals are randomly selected from the current population, and a local pollination step size is generated to update the individual position. All the updated individuals constitute a temporary population. Calculate the fitness value of each individual in the temporary population; Merge the current population with the temporary population, and select the top few individuals by fitness value as elite individuals; Two individuals are randomly selected from the elite individuals to generate offspring individuals; random perturbations are superimposed on the parameters of each dimension to generate a set of offspring individuals; the current population, the temporary population and the set of offspring individuals are merged, and individuals with the highest fitness values ​​and the same number of individuals as the population size are selected to form the next generation population. Calculate the fitness variance of the next generation population and adjust the global search step size based on the fitness variance, while decreasing the local search step size; The individual with the best fitness value from the next generation of the population is selected as the current global best individual. Based on the real-time working condition boundary conditions and the structural design parameters corresponding to the current population individuals, the digital twin model is driven to perform simulation calculations to obtain performance response data. Based on the performance response data and the preset design target, the fitness value of the population individuals is calculated. The fitness value is transmitted as feedback information to the improved flower pollination algorithm. The conversion probability, mutation rate and crossover rate are dynamically adjusted according to the fitness value. Based on the adjusted parameters, the population is controlled to perform the next global search or local search, and new structural design parameters are generated. The new structural design parameters are input into the digital twin model to perform iterative verification until the convergence condition is met, then the iteration is stopped and the optimal structural design parameters are output. The optimal structural design parameters are mapped to the digital twin model, the diesel engine model structure is updated, and an optimized digital twin model of the diesel engine structure is generated.

2. The diesel engine structure design method based on digital twin according to claim 1, characterized in that, The establishment of the digital twin model and the initialization of the flower pollination algorithm population specifically include: Obtain three-dimensional geometric models of key components of the diesel engine, model the key components using computer-aided design software, and generate a preliminary framework for a digital twin model. Based on the physical characteristics of each component of the diesel engine, the material property parameters of each component are set, and the selection is made according to the actual application requirements; Set the boundary conditions for the digital twin model, and input the boundary conditions based on the engine's working environment and operating conditions; Based on the established material properties and boundary conditions, the digital twin model is simulated using finite element analysis and computational fluid dynamics tools to calculate the performance of components under different working conditions. Initialize the improved flower pollination algorithm population, set preliminary optimization parameters according to the design goal, define the search space of design parameters, and select the structural design parameters of the initial population individuals.

3. The diesel engine structure design method based on digital twin according to claim 2, characterized in that, The generation of real-time operating condition boundary conditions specifically includes: Multi-source operating condition data are collected by a sensor array deployed on a physical prototype of the diesel engine; Time synchronization and filtering / denoising processing are performed on the collected multi-source operating condition data to remove outliers and generate a synchronized operating condition data sequence. Transmit the synchronous operating condition data sequence to the data interface of the digital twin model; parse the synchronous operating condition data sequence in the digital twin model and extract key feature parameters; input the key feature parameters as load conditions for the simulation calculation of the digital twin model to the corresponding load interface of the solver; The digital twin model performs simulation calculations based on the input load conditions and outputs the distribution field data of diesel engine components under the current operating conditions.

4. The diesel engine structure design method based on digital twin according to claim 3, characterized in that, The calculation of the fitness value of an individual in the population specifically includes: Extract the structural design parameters corresponding to each individual in the current population in sequence, and input the structural design parameters and real-time working condition boundary conditions into the digital twin model. The digital twin model updates the geometric features and material properties of components based on structural design parameters, applies load boundaries at corresponding locations based on real-time working condition boundary conditions, performs multiphysics coupling simulation calculations, and outputs performance response data. Read the preset design target file, which contains the maximum allowable stress threshold and the highest operating temperature threshold for each component; The performance response data is compared with the preset design target item by item, and the ratio between the actual simulation value of each indicator and the target threshold is calculated as the degree of compliance. The fitness values ​​of each achievement level are weighted and summed, and the weighted sum is used as the fitness value of the current individual under the current working conditions.

5. The diesel engine structure design method based on digital twin according to claim 4, characterized in that, The dynamic adjustment of the fitness value to the transition probability, mutation rate, and crossover rate specifically includes: The fitness values ​​of individuals in the current population are statistically analyzed, and the conversion probability, mutation rate, and crossover rate are dynamically adjusted based on the fitness improvement rate, the ratio of the population fitness variance to the current best fitness value, and the proportion of elites. The fitness improvement rate is the relative increase in the optimal fitness value between adjacent iterations, and the elite ratio is the proportion of individuals in the current population whose fitness value is higher than the average fitness value of several historical iterations to the population size. The direction and magnitude of the conversion probability are set based on the fitness improvement rate; the direction and magnitude of the mutation rate are set based on the ratio of the population fitness variance to the current best fitness value; and the direction and magnitude of the crossover rate are set based on the elite ratio. The adjustment ranges for the transformation probability, mutation rate, and crossover rate are set separately. The parameters after each adjustment are limited, and the adjusted parameters are used for the pollination operation in the next iteration.

6. The diesel engine structure design method based on digital twin according to claim 5, characterized in that, The step of inputting the new structural design parameters into the digital twin model to perform iterative verification specifically includes: After generating new structural design parameters in each iteration, a preset number of individuals with fitness values ​​ranked before the current population are selected as candidate individuals, and the structural design parameters of the candidate individuals are extracted as parameters to be verified. Input the parameters to be verified into the digital twin model to perform multiphysics simulation calculations and output the corresponding simulation performance data; Read the measured performance data of the physical prototype with the same boundary conditions as the current real-time operating conditions from the historical database, compare the simulated performance data with the measured performance data item by item, and calculate the absolute value of the relative error of each performance index; When the absolute value of the relative error of any performance index exceeds the preset accuracy threshold, the back propagation algorithm is used to correct the corresponding material property parameters or boundary condition correction coefficients in the digital twin model based on the relative error value, generating an updated digital twin model, and the updated digital twin model is used for subsequent iterations of simulation calculations. Repeat the above correction process until the absolute values ​​of the relative errors of all performance indicators converge to within the preset accuracy threshold. Determine whether the difference between the optimal fitness value in the current population and the preset design target threshold is less than the convergence tolerance. When both conditions are met, stop the iteration and output the structural design parameters corresponding to the current optimal fitness value as the final optimization result.

7. The diesel engine structure design method based on digital twin according to claim 6, characterized in that, The digital twin model of the diesel engine structure specifically includes: Geometric models of key diesel engine components are constructed based on optimal structural design parameters. The optimal structural design parameters are the combination of structural parameters that meet the preset design objectives and are output after iterative optimization by an improved flower pollination algorithm. The material property field distribution associated with the geometric model is based on the optimal structural design parameters to assign differentiated values ​​to the materials of the components, forming a non-uniform material distribution characteristic. The sensor virtual deployment nodes are embedded in the geometric model, and the arrangement positions of the sensor virtual deployment nodes correspond one-to-one with the sensor array of the physical prototype. The load mapping interface is dynamically associated with real-time operating condition boundary conditions. The load mapping interface receives real-time operating condition data collected and processed by the physical prototype and converts the real-time operating condition data into dynamic load boundaries for simulation calculation of the digital twin model. The parameter feedback interface is linked with the improved flower pollination algorithm. The parameter feedback interface receives the structural design parameters generated by the algorithm iteration and updates the geometric model, and sends the performance response data of each simulation calculation back to the improved flower pollination algorithm.

Images