Product life cycle management method and device based on diffusion model

Abstract

The invention relates to a product life cycle management method and device based on a diffusion model, and the method comprises the steps: obtaining multi-source original data, and carrying out the feature extraction of the original data based on a set weight and a feature extraction function, and obtaining an extraction feature; taking the extracted features as the input of a Transform encoder, and calling a deep learning model to carry out deep analysis on the extracted features to generate a global design primitive. A diffusion model is called to convert the global design primitive into a product operation parameter and a product design parameter, and the product operation parameter and the product design parameter serve as model configuration parameters to initialize the diffusion model. And taking the global design guidance data as the input of the initialized diffusion model, and converting the global design guidance data into corresponding product design execution data. And under a given cost function, obtaining a product design parameter when a cost function value corresponding to the product design execution data is minimum as a product design optimization parameter, and carrying out optimization adjustment on the product design execution data.

Description

Technical Field

[0001] The present invention relates to the technical field of product lifecycle management, and in particular to a product lifecycle management method and device based on a diffusion model. Background Art

[0002] Product lifecycle management technology (PLM technology) mainly involves how to manage the entire life cycle of a product from concept design to market retirement. PLM technology helps companies manage complex product information, optimize product development processes and improve collaboration efficiency by integrating information, processes, business systems and personnel. With the rapid changes in the market and the diversification of consumer demand, PLM systems play an increasingly important role in helping companies respond quickly to the market, reduce costs, improve product quality and innovation capabilities.

[0003] At present, PLM systems are increasingly integrating advanced data processing technologies, such as big data analysis, artificial intelligence (AI), and cloud computing, to adapt to more complex and dynamic product development needs. For example, AI technology is used to automate complex design decision-making processes and optimize product design, while cloud technology provides cross-regional and cross-platform data sharing and collaboration capabilities. The integration of these technologies not only improves the efficiency of the PLM system, but also greatly enhances its flexibility and adaptability. In addition, with the rise of sustainable development and green manufacturing, PLM systems have also begun to integrate environmental impact assessment tools to help designers consider the environmental impact of products during the design phase to achieve eco-friendly product development. This trend reflects the market's growing demand for sustainable products and companies' emphasis on environmental responsibility.

[0004] Although PLM systems have made significant progress, there are still some problems and challenges in practical applications. First, data integration and processing are still a big problem, especially how to effectively process and utilize large amounts of data from different sources and in different formats. Current PLM systems often need to be integrated with multiple other enterprise systems (such as ERP, SCM), and data compatibility and consistency issues often lead to inefficiency. Secondly, although the integration of emerging technologies has brought new functions and improved efficiency to PLM systems, it has also brought high technology updates and maintenance costs. For many small and medium-sized enterprises, these costs may be a considerable burden. In addition, the rapid changes in the market require PLM systems to be more flexible and quickly adapt to new market demands and technological changes. However, existing PLM systems are still insufficient in terms of rapid iteration and innovation, and it is difficult to fully meet the industry's needs for speed and flexibility. Typical existing technology solutions use modular software architectures, with each module responsible for different functions, such as CAD (computer-aided design) data management, document management, supply chain management, etc. These systems usually rely on a central database to store all relevant information to ensure data consistency and accuracy.

[0005] Therefore, existing product lifecycle management technologies are highly dependent on central data and have poor system flexibility and scalability. Summary of the invention

[0006] Based on this, it is necessary to provide a product lifecycle management method and device based on a diffusion model that can reduce the dependence on central data and has good system flexibility and scalability to address the above technical problems.

[0007] The present invention provides a product lifecycle management method based on a diffusion model, the method comprising:

[0008] Acquire raw data from multiple sources, and perform feature extraction on the raw data based on set weights and feature extraction functions to obtain extracted features for characterizing useful information or key information;

[0009] The extracted features are used as input of a Transformer encoder to call a deep learning model to perform deep analysis on the extracted features, thereby generating a global design primitive that reflects the overall design intent of the original data;

[0010] Calling a diffusion model to convert the global design primitives into product operation parameters and product design parameters, and initializing the diffusion model with the product operation parameters and product design parameters as model configuration parameters;

[0011] Using the global design guidance data as input of the initialized diffusion model, and calling the initialized diffusion model to convert the global design guidance data into corresponding product design execution data;

[0012] Under a given cost function, obtaining a product design parameter when the cost function value corresponding to the product design execution data is minimum as a product design optimization parameter, and optimizing and adjusting the product design execution data based on the product design optimization parameter;

[0013] Among them, the multi-source original data includes structured data and unstructured data, the extracted features are global feature vectors of the original data, the global design primitives are feature vectors output by the Transformer encoder after parsing the global feature vectors, and the product design execution data are executable product design detail process data based on the model configuration parameters.

[0014] In one embodiment, the method of acquiring raw data from multiple sources and performing feature extraction on the raw data based on a set weight and feature extraction function to obtain extracted features for characterizing useful information or key information includes:

[0015] Selecting multiple data segments from the original data, each data segment is provided with a corresponding weight, and extracting useful information or key information from each data segment according to the weight corresponding to the data segment, and performing standardization processing on the extracted useful information or key information; and

[0016] The feature extraction function is called to extract original features from multiple data segments of the original data, and feature weighting processing is performed on the original data based on the feature parameter set extracted corresponding to each data segment to obtain the weighted global feature vector.

[0017] In one embodiment, the extracting features are used as input of a Transformer encoder to call a deep learning model to perform deep analysis on the extracting features to generate a global design primitive that reflects the overall design intent of the original data, including:

[0018] Determine a first weight matrix and a second weight matrix based on the weighted global feature vector, wherein the first weight matrix is ​​used to adjust the contribution of each feature in the weighted summation, and the second weight matrix is ​​used to perform a linear transformation on the input features;

[0019] The offset of the linear transformation is adjusted by the bias term, and the hyperbolic tangent activation function is called to process the nonlinear relationship in the input features to obtain the weight vector of the original data;

[0020] A normalization function is called to convert the weight vector into a probability distribution to obtain a feature vector with the highest probability value, wherein the feature vector with the highest probability value is a global design primitive for reflecting the overall design intent of the original data.

[0021] In one embodiment, calling the diffusion model to convert the global design primitives into product operation parameters and product design parameters, and initializing the diffusion model with the product operation parameters and product design parameters as model configuration parameters, includes:

[0022] Acquire an initial condition set based on the original data of the multiple sources, and call an initialization function to generate initialization parameters according to the initial condition set and global design primitives, wherein the initialization parameters include the product operation parameters and product design parameters;

[0023] Multiple aspects of product design are defined according to the initialization parameters to obtain the model configuration parameters, and the diffusion model is initialized based on the model configuration parameters.

[0024] In one embodiment, the using the global design guidance data as the input of the initialized diffusion model, and calling the initialized diffusion model to convert the global design guidance data into corresponding product design execution data, includes:

[0025] Acquire historical product design states based on the original data from multiple sources, and call a local diffusion function to process the historical product design states and global design primitives according to a set time step to obtain a new product design state;

[0026] Calling an integral function to update a product design state according to the new product design state output by the local diffusion function within the time step to generate the product design execution data;

[0027] The set time step is the time step between the global design primitive and the historical product design state, and is used to control the time granularity of the product design iteration process.

[0028] In one embodiment, obtaining the product design parameters when the cost function value corresponding to the product design execution data is minimum under a given cost function as product design optimization parameters, and optimizing and adjusting the product design execution data based on the product design optimization parameters, includes:

[0029] Calling the cost function to calculate a plurality of cost function values ​​corresponding to the product design execution data, wherein the product design execution data has a corresponding product design parameter under each cost function value;

[0030] The product design parameters corresponding to the product design execution data when the cost function value is minimum are obtained through an optimization algorithm to obtain the product design optimization parameters.

[0031] In one embodiment, the method further comprises:

[0032] Acquire market and user feedback data in real time based on the original data from multiple sources, and call a feedback difference function according to the difference data between the market and user feedback data and the product design optimization parameters;

[0033] The difference data is written into the product design execution data through an update function to generate new product design execution data, and the new product design execution data is used to update the current product design execution data in real time.

[0034] The present invention also provides a product lifecycle management device based on a diffusion model, the device comprising:

[0035] A feature extraction module is used to obtain raw data from multiple sources, and perform feature extraction on the raw data based on the set weights and feature extraction functions to obtain extracted features for characterizing useful information or key information;

[0036] A deep analysis module, used to use the extracted features as input of a Transformer encoder to call a deep learning model to perform deep analysis on the extracted features, and generate a global design primitive that reflects the overall design intent of the original data;

[0037] A model initialization module, used for calling a diffusion model to convert the global design primitives into product operation parameters and product design parameters, and initializing the diffusion model using the product operation parameters and product design parameters as model configuration parameters;

[0038] A model processing module, used to use the global design guidance data as the input of the initialized diffusion model, and call the initialized diffusion model to convert the global design guidance data into corresponding product design execution data;

[0039] A parameter optimization module, used to obtain the product design parameters when the cost function value corresponding to the product design execution data is minimum under a given cost function as product design optimization parameters, and optimize and adjust the product design execution data based on the product design optimization parameters;

[0040] Among them, the multi-source original data includes structured data and unstructured data, the extracted features are global feature vectors of the original data, the global design primitives are feature vectors output by the Transformer encoder after parsing the global feature vectors, and the product design execution data are executable product design detail process data based on the model configuration parameters.

[0041] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the product lifecycle management method based on the diffusion model as described in any one of the above is implemented.

[0042] The present invention also provides a computer storage medium storing a computer program, wherein when the computer program is executed by a processor, the product lifecycle management method based on the diffusion model as described above is implemented.

[0043] The above-mentioned product lifecycle management method and device based on the diffusion model obtains original data from multiple sources, and extracts features from the original data based on the set weights and feature extraction functions to obtain extracted features for characterizing useful information or key information. Subsequently, the extracted features are used as the input of the Transformer encoder to call the deep learning model to perform deep analysis on the extracted features, and generate global design primitives for reflecting the overall design intent of the original data. The diffusion model is then called to convert the global design primitives into product operating parameters and product design parameters, and the diffusion model is initialized with the product operating parameters and product design parameters as model configuration parameters. Then the global design guidance data is used as the input of the initialized diffusion model, and the initialized diffusion model is called to convert the global design guidance data into corresponding product design execution data. Finally, the product design parameters corresponding to the product design execution data when the cost function value is the minimum under a given cost function are obtained as product design optimization parameters, and the product design execution data is optimized and adjusted based on the product design optimization parameters. This method optimizes the process from global design to local implementation by introducing a design method based on a diffusion model, thereby enhancing the system's response speed to market changes and the depth of data processing, while reducing dependence on the central database, improving the system's flexibility and scalability, and being able to more effectively support product innovation and personalized design to meet the needs of the modern market. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 One of the flow charts of the product life cycle management method based on the diffusion model provided by the present invention;

[0046] Figure 2 A schematic diagram of the overall process of product life cycle management of a product life cycle management method based on a diffusion model in a specific implementation provided by the present invention;

[0047] Figure 3 The second flowchart of the product life cycle management method based on the diffusion model provided by the present invention;

[0048] Figure 4 The third flowchart of the product life cycle management method based on the diffusion model provided by the present invention;

[0049] Figure 5 The fourth flowchart of the product life cycle management method based on the diffusion model provided by the present invention;

[0050] Figure 6 The fifth flowchart of the product life cycle management method based on the diffusion model provided by the present invention;

[0051] Figure 7 The sixth flowchart of the product life cycle management method based on the diffusion model provided by the present invention;

[0052] Figure 8 The seventh flowchart of the product life cycle management method based on the diffusion model provided by the present invention;

[0053] Fig. 9 A schematic diagram of the structure of a product lifecycle management device based on a diffusion model provided by the present invention;

[0054] Fig.10 This is a diagram of the internal structure of the computer device provided by the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] Combine the following Figure 1-Figure 10 The product lifecycle management method and device based on the diffusion model of the present invention are described.

[0057] like Figure 1 As shown, in one embodiment, a product lifecycle management method based on a diffusion model includes the following steps:

[0058] Step S110, acquiring original data from multiple sources, and performing feature extraction on the original data based on the set weights and feature extraction functions to obtain extracted features for characterizing useful information or key information.

[0059] Specifically, the server obtains raw data from multiple sources, and performs feature extraction on the raw data based on set weights and feature extraction functions to obtain extracted features for characterizing useful information or key information.

[0060] The original data from multiple sources include structured data and unstructured data, and the extracted features are global feature vectors of the original data.

[0061] Combination Figure 2 As shown, in a specific embodiment, the product lifecycle management method based on the diffusion model provided by the present invention first collects and extracts key information from multiple data sources, including market data, user feedback, technical parameters, etc. The purpose of feature extraction is to obtain valuable information from these complex raw data to provide a basis for subsequent product design decisions. Data processing technology is used in the feature extraction process to ensure that the extracted information is accurate and reliable, providing a solid foundation for product design.

[0062] In this embodiment, in product lifecycle management (PLM), feature extraction is the key to identifying and extracting useful information from raw data. In the system, feature extraction not only focuses on extracting features from audio (as in the previous example), but is also widely used to extract useful information from various business and engineering data. A general feature extraction method will be used here, which is suitable for processing various data types, such as product performance data, customer feedback, market dynamics, etc.

[0063] The feature extraction algorithm formula is:

[0064] F=Sum(w_i*Features(Extract(data_i),data_params_i)).

[0065] Among them, F is the weighted global feature vector, which integrates the features of all input data and provides a comprehensive perspective for subsequent analysis and decision-making processes. w_i is the weight of the i-th data segment. These weights are assigned according to the importance or reliability of the data segment to ensure that more critical information contributes more to the final feature vector. Features is a feature extraction function that is responsible for adjusting and standardizing the information extracted from the raw data to ensure the validity and consistency of the data in subsequent processing. Extract is a general data extraction function that is responsible for extracting raw features from various types of input data, such as extracting temperature changes from sensor data and emotional tendencies from user feedback. data_i is the i-th input data segment, which can be any form of business or engineering data. data_params_i is a set of parameters for feature extraction of the i-th data segment. These parameters may include the accuracy of feature extraction, the settings of specific algorithms, etc., to ensure the adaptability and flexibility of the data extraction process.

[0066] In a complex product development process, data comes from a wide range of sources, including but not limited to mechanical performance test data, market feedback, customer service records, and economic data related to the product. For example, for a newly developed smartphone, key data may include processor performance test results, consumer preferences obtained from market research, initial user feedback, and data on the durability of the phone. Each type of data has its own unique characteristics and extraction requirements. For example, data extracted from processor performance tests may focus on peak operating speed and thermal efficiency, while data extracted from market research may focus more on consumer acceptance of new features or price sensitivity. Through the above formula, each type of data is first extracted through the Extract function to obtain preliminary features, and then further processed and standardized through the Features function to adapt to the subsequent analysis and decision-making process. By setting different weights "w_i" for different data, the system can emphasize the importance of certain key information. For example, in the early stages of product design, technical performance data may be more important than market data, so it can be given a higher weight. This method not only enhances the relevance and representativeness of the feature vector "F", but also makes the entire data processing process more flexible and adaptable. This comprehensive and weighted feature extraction method enables the PLM system to more accurately grasp key information throughout the product development cycle, optimize the decision-making process, and improve product success rate and market competitiveness.

[0067] Step S120, using the extracted features as input to the Transformer encoder to call a deep learning model to perform deep analysis on the extracted features, and generate global design primitives that reflect the overall design intent of the original data.

[0068] Specifically, the server uses the extracted features as the input of the Transformer encoder to call the deep learning model to perform deep analysis on the extracted features and generate global design primitives that reflect the overall design intent of the original data.

[0069] Among them, the global design primitive is the feature vector output by the Transformer encoder after parsing the global feature vector.

[0070] Combination Figure 2 As shown, in a specific embodiment, the product lifecycle management method based on the diffusion model provided by the present invention uses the Transformer encoder to process the extracted features to identify the intrinsic relationships and patterns in the data. The focus of this stage is to use deep learning algorithms to parse and transform data to generate global design primitives. These primitives summarize the key directions and requirements of product design, such as performance standards, cost targets, etc., laying the foundation for more detailed design work.

[0071] In this embodiment, the extracted features are then input into the Transformer encoder, which is processed to further parse the extracted features, find patterns and associations using deep learning techniques, and generate global design primitives that reflect the overall design intent. Therefore, the Transformer encoder processing relies on the data quality and exhaustiveness of the original data and extracted features, and its output is a high-level design guide. In a product lifecycle management (PLM) system, the purpose of using the Transformer encoder to process the global feature vector is to convert large amounts of potentially unstructured data information into useful, structured design guidance. The Transformer encoder plays a key role here, as it can capture the complex relationships between input features through a self-attention mechanism, thereby providing deep insights into product design and development.

[0072] The Transformer encoder parsing algorithm formula is:

[0073] E=Softmax(sum_j(V_j*tanh(W_F*F+b_F))).

[0074] Among them, E is the feature vector output by the encoder, which represents the global design primitives. These primitives are the basis for subsequent product design decisions and detailed development. V_j and W_F are both weight matrices. V_j is used to adjust the contribution of each feature in the final weighted sum, while W_F is used to perform a preliminary linear transformation on the input feature F. F is the global feature vector obtained by data feature extraction during the feature extraction process. b_F is a bias term used to adjust the offset of the linear transformation, increase the flexibility of the model and its ability to adapt to different data. Tanh is a hyperbolic tangent activation function, which is used to increase the nonlinear capability in the processing process and helps to handle complex and nonlinear relationships. Softmax is a normalization function used to convert the output weight vector into a probability distribution, which helps to more clearly select the most important features in subsequent processing.

[0075] Step S130 , calling the diffusion model to convert the global design primitives into product operation parameters and product design parameters, and initializing the diffusion model using the product operation parameters and product design parameters as model configuration parameters.

[0076] Specifically, the server calls the diffusion model to convert the global design primitives into product operation parameters and product design parameters, and initializes the diffusion model using the product operation parameters and product design parameters as model configuration parameters.

[0077] Combination Figure 2 As shown, in a specific embodiment, the product lifecycle management method based on the diffusion model provided by the present invention, the output of the encoder contains the core concepts and guidance of the global design, and these outputs serve as the input for the initialization of the diffusion model. The task of the diffusion model is to convert these abstract global guidance into specific operations and design parameters, which need to be precisely adjusted according to the specific needs and market positioning of the product. The initialization of the diffusion model directly depends on the results of the encoder processing, so the global design direction determines the setting of the local model and the parameter adjustment direction.

[0078] In this embodiment, in the context of the product lifecycle management (PLM) system, the initialization of the diffusion model is the key to the entire system, which ensures the accurate transformation from the global design primitives to the specific design and production details. The process pays special attention to the conversion of the abstract information (E) obtained from the global diffusion stage into the specific parameters (LD_params) of the actual operation and design decision. The algorithm formula is:

[0079] LD_params=Initialize(E, initial_conditions).

[0080] Among them, LD_params are the initialization parameters of the diffusion model, which define specific aspects of product design, such as material selection, component size, production process, etc. Initialize is the initialization function, which is a key algorithm or process used to combine the global output E with the initial conditions to generate the specific operating parameters required for the diffusion model. E is the output of the global diffusion stage, which contains guidance information about the overall product design, such as the main functions of the product design, performance goals, budget constraints, etc. initial_conditions is the initial condition set, which includes product type, design specifications, market demand, regulatory requirements, etc., and is the basis for customizing and personalizing the diffusion model.

[0081] In practical applications of product development, such as designing a new high-efficiency automobile engine, the global diffusion stage may have determined high-level goals such as energy saving and emission reduction and related global characteristics. The task of diffusion model initialization is to convert these global design goals into specific engineering parameters and production guidelines. For example, if the global output E indicates that the engine design needs to optimize combustion efficiency and reduce emissions, the initialization process will involve determining the specific cylinder size, piston stroke, combustion chamber design, etc. Initial conditions may include specific materials used (such as aluminum alloy or titanium alloy), technical standards (such as European emission standards), and cost constraints. The Initialize function plays a key role here, and it may use machine learning, optimization algorithms, or rule-based systems to process this information. The process includes:

[0082] (1) Parameter determination: Select the most appropriate technical parameters and materials based on E and initial conditions.

[0083] (2) Performance prediction: Use simulation and calculation methods to predict product performance under selected parameters.

[0084] (3) Iterative optimization: Based on performance prediction results and cost-benefit analysis, parameters are adjusted to meet design goals and cost constraints.

[0085] In this embodiment, "LD_params" may include detailed production process parameters, such as machine tool settings, assembly line speeds, inspection and quality control standards, etc., which are refined based on the global guidance and initial conditions in "E". The advantage of this approach is that it can ensure the consistency and continuity between the global goals of the design and the local implementation, thereby effectively controlling the complexity and uncertainty in the product development process. Through precise initialization, the PLM system can better manage and predict every step of the product from design to production, ensuring that the final product can meet market demand and regulatory requirements while optimizing cost and time. This integrated and systematic approach is particularly critical to cope with the rapid changes and high competition in today's market.

[0086] Step S140 , taking the global design guidance data as input of the initialized diffusion model, and calling the initialized diffusion model to transform the global design guidance data into corresponding product design execution data.

[0087] Specifically, the server uses the pre-input global design guidance data as the input of the initialized diffusion model, and calls the initialized diffusion model to convert the global design guidance data into corresponding product design execution data.

[0088] The product design execution data is executable product design detail process data based on model configuration parameters.

[0089] Combination Figure 2 As shown, in a specific embodiment, the product lifecycle management method based on the diffusion model provided by the present invention can be used to generate detailed design once the diffusion model is initialized and configured. The generated detailed design includes all technical details, specific material selection, size, etc., which are all based on the parameters set in the model initialization phase. This process is to convert the theoretical parameters and design directions defined in the initialization phase into practical and implementable, detailed product designs. The design details generated by the local model directly affect the manufacturing and functional realization of the product. In the product lifecycle management (PLM) system, generating detailed design is a key process to convert global design guidance into specific and executable product design details. This process utilizes a diffusion model (LD), combined with the previous design state (D_prev) and global guidance (E), and iteratively updates and refines the design through an integral process. The algorithm formula is:

[0090] D=Integrate(LD(E,D_prev),dt).

[0091] Among them, D is the generated detailed design, which is the final output and contains all the necessary technical and production details, such as dimensions, materials, tolerances, etc. LD is the local diffusion function, which is a key algorithm or process used to calculate the new design state based on the current and previous design states, as well as the global guidance information (E). E is the output of the global diffusion stage, which provides direction and constraints for local design, such as performance standards, cost targets, etc. D_prev is the design state at the previous moment, which provides historical data support for the diffusion model to help the model understand the evolution of the design. Integrate is an integral function used to update the design state according to the output of the local diffusion function within a set time step (dt). dt is the time step, which is the time granularity that controls the design iteration process, and its size affects the update speed and accuracy of the design details.

[0092] In actual product development, such as developing a new smartphone, the process involves the transformation from global design goals (such as thin and light design, high durability, optimized user interface) to specific engineering implementation (such as screen size, battery life, material selection, etc.). The global design goal (E) may guide the design team to optimize the energy consumption and physical size of the device, while the local diffusion function (LD) converts these abstract goals into specific technical parameters, such as the choice of battery technology, the decision of screen technology (OLED or LCD), etc. The process needs to be iterated continuously, and each iteration is adjusted and optimized through the design state (D_prev) at the previous moment to ensure that the design not only meets the functional requirements, but also meets the cost and market competitiveness. For example, if the battery life caused by the preliminary design is insufficient, the LD function will adjust the relevant parameters (such as battery capacity, power management strategy) and update the design state in the next time step (dt) through the integral function. This approach allows the design team to continuously optimize the product to achieve the best market competitiveness while meeting the functional and performance requirements. This iterative integration process is very important because it allows the design team to flexibly adjust and respond to sudden technological and market changes during the development cycle. In this way, the PLM system can ensure the continuity and adaptability of the product design process, thereby improving the market success rate and customer satisfaction of the final product. This detailed design generation method is suitable for fast-changing high-tech product fields such as smart devices, consumer electronics and advanced industrial equipment.

[0093] Step S150, obtaining product design parameters when the cost function value corresponding to the product design execution data is minimum under a given cost function as product design optimization parameters, and optimizing and adjusting the product design execution data based on the product design optimization parameters.

[0094] Specifically, the server obtains the product design parameters when the cost function value corresponding to the product design execution data is minimized under a given cost function as product design optimization parameters, and optimizes and adjusts the product design execution data based on the product design optimization parameters.

[0095] Combination Figure 2 As shown, in a specific embodiment, the product lifecycle management method based on the diffusion model provided by the present invention, after the detailed design is completed, the next step is performance optimization. The performance optimization process involves comprehensive evaluation and optimization of the economy, efficiency and market competitiveness of the design. Performance optimization needs to evaluate the performance and cost-effectiveness of the product based on the data provided by the detailed design, and make necessary adjustments to the design to ensure the best resource utilization and market performance, which directly affects the market success of the product. It is a direct evaluation and improvement process based on the output of the detailed design.

[0096] In this embodiment, performance optimization is a key process in the product development process, which can ensure that the final design not only meets the technical and functional requirements, but also achieves the best in terms of cost and efficiency. This process involves using a cost function to evaluate and optimize the design so that the production and operating costs of the entire product are minimized while maintaining or improving product performance. The algorithm formula is as follows:

[0097] P=Argmin_cost(D, cost_function).

[0098] Where P is the optimized product design parameters, which ensure cost efficiency and resource optimization while meeting all technical and market requirements. Argmin_cost is an optimization algorithm used to find the design parameters that minimize the value of a given cost function. D is the previously generated detailed design, which contains all the design details and technical specifications of the product. cost_function is the cost function, a mathematical model or expression used to evaluate the economic and performance indicators of the design, such as production cost, material cost, energy consumption, environmental impact, etc.

[0099] In modern product development, it is extremely important to optimize the design to reduce costs while improving performance and efficiency. For example, imagine developing a new home air purifier. The design team needs to consider not only its air purification efficiency, but also energy consumption, service life, manufacturing cost, and user maintenance cost. In this case, the "cost_function" will consider multiple factors:

[0100] Manufacturing costs include material costs and labor costs in the production process.

[0101] Energy efficiency is concerned with the amount of electricity a product consumes during use.

[0102] Maintenance costs, including the cost of replacing filters and other maintenance costs that users may face during use.

[0103] In this embodiment, the performance optimization process uses the Argmin_cost algorithm to explore different design variants to find the best balance between cost and benefit. The algorithm iteratively tests various design parameters (D) through simulation or computational methods (such as finite element analysis, computational fluid dynamics, etc.), and uses cost_function to evaluate the cost efficiency of these designs after each iteration. For example, the design team may experiment with filters of different materials, each of which has different costs and purification efficiencies. Through the Argmin_cost algorithm, the total cost of each design solution (including manufacturing costs and expected user maintenance costs) can be calculated, and those design parameters (P) that provide the best cost-effectiveness ratio can be selected. In addition, environmental sustainability can also be incorporated into the cost function, such as evaluating the carbon footprint of the product throughout its life cycle. Such a comprehensive evaluation ensures that the product is not only competitive in the market, but also meets the expectations of modern consumers for environmental protection and economic benefits. In this way, the PLM system can ensure the optimal use of resources during product design and development, and ultimately achieve an economical and efficient product solution.

[0104] In this embodiment, performance optimization ensures the theoretically optimal design of the product, and real-time feedback adjustment is also required to ensure that these designs can adapt to the dynamic needs of the actual market and users. This process is used to continuously adjust and optimize product design based on market feedback, user experience and other real-time data. Real-time feedback adjustment enables product design to respond quickly to market changes, enhance product adaptability and user satisfaction. This process is a natural extension of performance optimization, shifting product design from theoretical optimization to practical application optimization. Real-time feedback adjustment can ensure that product design can quickly adapt to market changes and user needs during the product life cycle management process. By integrating user feedback, the system can iteratively improve product design, optimize performance and enhance user satisfaction.

[0105] The algorithm formula for real-time feedback adjustment is:

[0106] D_updated=Update(D,Delta_feedback(feedback,P)).

[0107] Among them, D_updated is the updated product design, which includes adjustments to the original design (D) based on feedback data. Update is the update function, which is a processing algorithm used to integrate feedback differences into the current design and generate a new design version. Delta_feedback is the feedback difference function, which is responsible for calculating the difference between the real-time feedback received and the current design parameters (P), and making necessary adjustment suggestions. feedback is real-time market and user feedback data, which can come from multiple channels, such as online reviews, customer service records, market research, etc. P is the optimized parameters, including the design parameters determined in the performance optimization phase, which define the core performance and functional characteristics of the product.

[0108] In the modern product development process, the real-time feedback and quick response to this feedback are the keys to success. For example, suppose a company has just launched a smartwatch and early users begin to report issues with the touch screen being slow to respond. This type of feedback is collected through online product review platforms and customer service departments. The real-time feedback adjustment process is as follows:

[0109] Collect feedback: The system automatically collects user feedback about the new smartwatch through various channels.

[0110] Analyze feedback: Use data analysis tools to parse feedback content and determine the prevalence and severity of the problem. For example, a touch screen feedback problem may be caused by hardware limitations or insufficient software optimization.

[0111] Calculate feedback difference (Delta_feedback): This function analyzes the difference between the current design parameters (P) and the feedback data to determine the specific parameters that need to be adjusted, such as improving the sensitivity of the touch screen or updating the touch algorithm.

[0112] Update design: Update design parameters and product specifications based on Delta_feedback suggestions, such as replacing better screen components or adjusting software processing logic.

[0113] Through this process, D_updated is not only a fine-tuning of the original design, but also a new version optimized according to market and user needs. This iterative improvement can greatly increase the market competitiveness and user satisfaction of the product.

[0114] The above-mentioned product lifecycle management method based on the diffusion model obtains the original data from multiple sources, and extracts the features of the original data based on the set weights and feature extraction functions to obtain the extracted features for characterizing useful information or key information. Subsequently, the extracted features are used as the input of the Transformer encoder to call the deep learning model to deeply analyze the extracted features and generate global design primitives for reflecting the overall design intent of the original data. The diffusion model is then called to convert the global design primitives into product operating parameters and product design parameters, and the diffusion model is initialized with the product operating parameters and product design parameters as model configuration parameters. Then the global design guidance data is used as the input of the initialized diffusion model, and the initialized diffusion model is called to convert the global design guidance data into the corresponding product design execution data. Finally, the product design parameters corresponding to the product design execution data when the cost function value is the minimum under a given cost function are obtained as product design optimization parameters, and the product design execution data is optimized and adjusted based on the product design optimization parameters. This method optimizes the process from global design to local implementation by introducing a design method based on a diffusion model, thereby enhancing the system's response speed to market changes and the depth of data processing, while reducing dependence on the central database, improving the system's flexibility and scalability, and being able to more effectively support product innovation and personalized design to meet the needs of the modern market.

[0115] like Figure 3 As shown, in one embodiment, the product lifecycle management method based on the diffusion model provided by the present invention obtains raw data from multiple sources, and extracts features from the raw data based on the set weights and feature extraction functions to obtain extracted features for characterizing useful information or key information, specifically including the following steps:

[0116] Step S112, selecting multiple data segments from the original data, each data segment is set with a corresponding weight, and extracting useful information or key information from each data segment according to the weight corresponding to the data segment, and standardizing the extracted useful information or key information.

[0117] Step S114, calling a feature extraction function to extract original features from multiple data segments of the original data, and performing feature weighting processing on the original data based on the extracted feature parameter set corresponding to each data segment to obtain a weighted global feature vector.

[0118] like Figure 4 As shown, in one embodiment, the product lifecycle management method based on the diffusion model provided by the present invention uses the extracted features as the input of the Transformer encoder to call the deep learning model to perform deep analysis on the extracted features, and generates a global design primitive for reflecting the overall design intent of the original data, specifically including the following steps:

[0119] Step S122, determining a first weight matrix and a second weight matrix based on the weighted global feature vector, the first weight matrix is ​​used to adjust the contribution of each feature in the weighted summation, and the second weight matrix is ​​used to perform a linear transformation on the input features.

[0120] Step S124, adjusting the offset of the linear transformation through the bias term, and calling the hyperbolic tangent activation function to process the nonlinear relationship in the input features to obtain the weight vector of the original data.

[0121] Step S126, calling a normalization function to convert the weight vector into a probability distribution to obtain a feature vector with the highest probability value, where the feature vector with the highest probability value is a global design primitive used to reflect the overall design intent of the original data.

[0122] like Figure 5 As shown, in one embodiment, the product lifecycle management method based on the diffusion model provided by the present invention calls the diffusion model to convert the global design primitives into product operation parameters and product design parameters, and initializes the diffusion model with the product operation parameters and product design parameters as model configuration parameters, specifically including the following steps:

[0123] Step S132, acquiring an initial condition set based on original data from multiple sources, and calling an initialization function to generate initialization parameters according to the initial condition set and global design primitives, wherein the initialization parameters include product operation parameters and product design parameters.

[0124] Step S134, defining multiple aspects of product design according to the initialization parameters, obtaining model configuration parameters, and initializing the diffusion model based on the model configuration parameters.

[0125] like Figure 6 As shown, in one embodiment, the product lifecycle management method based on the diffusion model provided by the present invention uses the global design guidance data as the input of the initialized diffusion model, and calls the initialized diffusion model to convert the global design guidance data into corresponding product design execution data, which specifically includes the following steps:

[0126] Step S142, acquiring historical product design states based on original data from multiple sources, and invoking a local diffusion function to process the historical product design states and global design primitives according to a set time step to obtain new product design states.

[0127] Step S144, calling the integral function to update the product design state according to the new product design state output by the local diffusion function within the time step to generate product design execution data.

[0128] The set time step is the time step between the global design primitive and the historical product design state, which is used to control the time granularity of the product design iteration process.

[0129] like Figure 7 As shown, in one embodiment, the product lifecycle management method based on the diffusion model provided by the present invention obtains the product design parameters when the cost function value corresponding to the product design execution data is the minimum under a given cost function as the product design optimization parameters, and optimizes and adjusts the product design execution data based on the product design optimization parameters, specifically including the following steps:

[0130] Step S152, calling the cost function to calculate multiple cost function values ​​corresponding to the product design execution data, the product design execution data having corresponding product design parameters under each cost function value.

[0131] Step S154, obtaining the product design parameters corresponding to the product design execution data when the cost function value is minimum through the optimization algorithm, so as to obtain the product design optimization parameters.

[0132] like Figure 8 As shown, in one embodiment, the product lifecycle management method based on the diffusion model provided by the present invention further includes the following steps:

[0133] Step S810, acquiring market and user feedback data in real time based on original data from multiple sources, and calling a feedback difference function according to the difference data between the market and user feedback data and product design optimization parameters.

[0134] Step S820, writing the difference data into the product design execution data through the update function to generate new product design execution data, and the new product design execution data is used to update the current product design execution data in real time.

[0135] The product lifecycle management device based on the diffusion model provided by the present invention is described below. The product lifecycle management device based on the diffusion model described below and the product lifecycle management method based on the diffusion model described above can be referenced to each other.

[0136] like Fig. 9 As shown, in one embodiment, a product lifecycle management device based on a diffusion model includes a feature extraction module 910 , a depth analysis module 920 , a model initialization module 930 , a model processing module 940 and a parameter optimization module 950 .

[0137] The feature extraction module 910 is used to obtain raw data from multiple sources, and perform feature extraction on the raw data based on the set weights and feature extraction functions to obtain extracted features for characterizing useful information or key information.

[0138] The deep analysis module 920 is used to use the extracted features as the input of the Transformer encoder to call the deep learning model to perform deep analysis on the extracted features and generate global design primitives that reflect the overall design intent of the original data.

[0139] The model initialization module 930 is used to call the diffusion model to convert the global design primitives into product operation parameters and product design parameters, and initialize the diffusion model using the product operation parameters and product design parameters as model configuration parameters.

[0140] The model processing module 940 is used to use the global design guidance data as the input of the initialized diffusion model, and call the initialized diffusion model to convert the global design guidance data into corresponding product design execution data.

[0141] The parameter optimization module 950 is used to obtain the product design parameters when the cost function value corresponding to the product design execution data is minimized under a given cost function as product design optimization parameters, and optimize and adjust the product design execution data based on the product design optimization parameters.

[0142] Among them, the multi-source original data includes structured data and unstructured data, the extracted features are the global feature vectors of the original data, the global design primitives are the feature vectors output by the Transformer encoder after parsing the global feature vectors, and the product design execution data are executable product design detail process data based on the model configuration parameters.

[0143] In this embodiment, in the product lifecycle management device based on the diffusion model provided by the present invention, the feature extraction module is specifically used for:

[0144] Select multiple data segments from the original data, each data segment is set with a corresponding weight, and useful information or key information is extracted from each data segment according to the weight corresponding to the data segment, and the extracted useful information or key information is standardized.

[0145] The feature extraction function is called to extract original features from multiple data fragments of the original data, and feature weighting processing is performed on the original data based on the feature parameter set extracted corresponding to each data fragment to obtain a weighted global feature vector.

[0146] In this embodiment, the product lifecycle management device based on the diffusion model provided by the present invention, the deep analysis module is specifically used for:

[0147] A first weight matrix and a second weight matrix are determined based on the weighted global eigenvector, the first weight matrix is ​​used to adjust the contribution of each feature in the weighted summation, and the second weight matrix is ​​used to perform a linear transformation on the input features.

[0148] The offset of the linear transformation is adjusted through the bias term, and the hyperbolic tangent activation function is called to process the nonlinear relationship in the input features to obtain the weight vector of the original data.

[0149] The normalization function is called to convert the weight vector into a probability distribution to obtain the eigenvector with the highest probability value. The eigenvector with the highest probability value is the global design primitive used to reflect the overall design intent of the original data.

[0150] In this embodiment, in the product lifecycle management device based on the diffusion model provided by the present invention, the model initialization module is specifically used for:

[0151] An initial condition set is acquired based on original data from multiple sources, and an initialization function is called to generate initialization parameters according to the initial condition set and global design primitives. The initialization parameters include product operation parameters and product design parameters.

[0152] Multiple aspects of product design are defined according to initialization parameters to obtain model configuration parameters, and the diffusion model is initialized based on the model configuration parameters.

[0153] In this embodiment, the product lifecycle management device based on the diffusion model provided by the present invention, the model processing module is specifically used for:

[0154] The historical product design status is obtained based on the original data from multiple sources, and the local diffusion function is called to process the historical product design status and the global design primitives according to the set time step to obtain the new product design status.

[0155] The integration function is called to update the product design state according to the new product design state output by the local diffusion function within the time step to generate product design execution data.

[0156] The set time step is the time step between the global design primitive and the historical product design state, which is used to control the time granularity of the product design iteration process.

[0157] In this embodiment, the product lifecycle management device based on the diffusion model provided by the present invention, the parameter optimization module is specifically used for:

[0158] The cost function is called to calculate multiple cost function values ​​corresponding to the product design execution data, and the product design execution data has corresponding product design parameters under each cost function value.

[0159] The product design parameters corresponding to the product design execution data when the cost function value is minimum are obtained through the optimization algorithm to obtain the product design optimization parameters.

[0160] In this embodiment, the product lifecycle management device based on the diffusion model provided by the present invention further includes a real-time feedback adjustment module, which is used to:

[0161] Based on the original data from multiple sources, market and user feedback data are obtained in real time, and the feedback difference function is called to obtain the difference data between market and user feedback data and product design optimization parameters.

[0162] The difference data is written into the product design execution data through the update function to generate new product design execution data, and the new product design execution data is used to update the current product design execution data in real time.

[0163] Fig.10 The physical structure diagram of an electronic device is shown as an example. The electronic device may be a smart terminal, and its internal structure diagram may be as follows: Fig.10 As shown. The electronic device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the electronic device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above-mentioned product life cycle management method based on the diffusion model is implemented.

[0164] Those skilled in the art will understand that Fig.10 The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present invention, and does not constitute a limitation on the electronic device to which the scheme of the present invention is applied. The specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0165] On the other hand, the present invention further provides a computer storage medium storing a computer program, which implements the above-mentioned product lifecycle management method based on the diffusion model when executed by a processor.

[0166] In another aspect, a computer program product or a computer program is provided, the computer program product or the computer program comprising computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium, and when the processor executes the computer instructions, the product lifecycle management method based on the diffusion model is implemented.

[0167] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory.

[0168] By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0169] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0170] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A product lifecycle management method based on diffusion model, It is characterized in that The method comprises: Acquire raw data from multiple sources, and perform feature extraction on the raw data based on set weights and feature extraction functions to obtain extracted features for characterizing useful information or key information; The extracted features are used as input of a Transformer encoder to call a deep learning model to perform deep analysis on the extracted features, thereby generating a global design primitive that reflects the overall design intent of the original data; Calling a diffusion model to convert the global design primitives into product operation parameters and product design parameters, and initializing the diffusion model with the product operation parameters and product design parameters as model configuration parameters; Using the global design guidance data as input of the initialized diffusion model, and calling the initialized diffusion model to convert the global design guidance data into corresponding product design execution data; Under a given cost function, obtaining a product design parameter when the cost function value corresponding to the product design execution data is minimum as a product design optimization parameter, and optimizing and adjusting the product design execution data based on the product design optimization parameter; Among them, the multi-source original data includes structured data and unstructured data, the extracted features are global feature vectors of the original data, the global design primitives are feature vectors output by the Transformer encoder after parsing the global feature vectors, and the product design execution data are executable product design detail process data based on the model configuration parameters.

2. The product lifecycle management method based on the diffusion model according to claim 1, It is characterized in that The method of acquiring raw data from multiple sources and extracting features from the raw data based on the set weights and feature extraction functions to obtain extracted features for characterizing useful information or key information includes: Selecting multiple data segments from the original data, each data segment is provided with a corresponding weight, and extracting useful information or key information from each data segment according to the weight corresponding to the data segment, and performing standardization processing on the extracted useful information or key information; and The feature extraction function is called to extract original features from multiple data segments of the original data, and feature weighting processing is performed on the original data based on the feature parameter set extracted corresponding to each data segment to obtain the weighted global feature vector.

3. The product lifecycle management method based on the diffusion model according to claim 2, It is characterized in that The extracting features are used as input of a Transformer encoder to call a deep learning model to perform deep analysis on the extracting features, and generate a global design primitive that reflects the overall design intent of the original data, including: Determine a first weight matrix and a second weight matrix based on the weighted global feature vector, wherein the first weight matrix is ​​used to adjust the contribution of each feature in the weighted summation, and the second weight matrix is ​​used to perform a linear transformation on the input features; The offset of the linear transformation is adjusted by the bias term, and the hyperbolic tangent activation function is called to process the nonlinear relationship in the input features to obtain the weight vector of the original data; A normalization function is called to convert the weight vector into a probability distribution to obtain a feature vector with the highest probability value, wherein the feature vector with the highest probability value is a global design primitive for reflecting the overall design intent of the original data.

4. The product lifecycle management method based on the diffusion model according to claim 3, It is characterized in that The calling of the diffusion model to convert the global design primitives into product operation parameters and product design parameters, and initializing the diffusion model with the product operation parameters and product design parameters as model configuration parameters, includes: Acquire an initial condition set based on the original data of the multiple sources, and call an initialization function to generate initialization parameters according to the initial condition set and global design primitives, wherein the initialization parameters include the product operation parameters and product design parameters; Multiple aspects of product design are defined according to the initialization parameters to obtain the model configuration parameters, and the diffusion model is initialized based on the model configuration parameters.

5. The product lifecycle management method based on the diffusion model according to claim 3, It is characterized in that The using the global design guidance data as the input of the initialized diffusion model, and calling the initialized diffusion model to convert the global design guidance data into corresponding product design execution data, includes: Acquire historical product design states based on the original data from multiple sources, and call a local diffusion function to process the historical product design states and global design primitives according to a set time step to obtain a new product design state; Calling an integral function to update a product design state according to the new product design state output by the local diffusion function within the time step to generate the product design execution data; The set time step is the time step between the global design primitive and the historical product design state, and is used to control the time granularity of the product design iteration process.

6. The product lifecycle management method based on the diffusion model according to claim 1, It is characterized in that The obtaining, under a given cost function, product design parameters when the cost function value corresponding to the product design execution data is minimum as product design optimization parameters, and optimizing and adjusting the product design execution data based on the product design optimization parameters, comprises: Calling the cost function to calculate a plurality of cost function values ​​corresponding to the product design execution data, wherein the product design execution data has a corresponding product design parameter under each cost function value; The product design parameters corresponding to the product design execution data when the cost function value is minimum are obtained through an optimization algorithm to obtain the product design optimization parameters.

7. The product lifecycle management method based on the diffusion model according to any one of claims 1 to 6, It is characterized in that The method further comprises: Acquire market and user feedback data in real time based on the original data from multiple sources, and call a feedback difference function according to the difference data between the market and user feedback data and the product design optimization parameters; The difference data is written into the product design execution data through an update function to generate new product design execution data, and the new product design execution data is used to update the current product design execution data in real time.

8. A product life cycle management device based on a diffusion model, It is characterized in that The device comprises: A feature extraction module is used to obtain raw data from multiple sources, and perform feature extraction on the raw data based on the set weights and feature extraction functions to obtain extracted features for characterizing useful information or key information; A deep analysis module, used to use the extracted features as input of a Transformer encoder to call a deep learning model to perform deep analysis on the extracted features, and generate a global design primitive that reflects the overall design intent of the original data; A model initialization module, used for calling a diffusion model to convert the global design primitives into product operation parameters and product design parameters, and initializing the diffusion model using the product operation parameters and product design parameters as model configuration parameters; A model processing module, used to use the global design guidance data as the input of the initialized diffusion model, and call the initialized diffusion model to convert the global design guidance data into corresponding product design execution data; A parameter optimization module, used to obtain the product design parameters when the cost function value corresponding to the product design execution data is minimum under a given cost function as product design optimization parameters, and optimize and adjust the product design execution data based on the product design optimization parameters; Among them, the multi-source original data includes structured data and unstructured data, the extracted features are global feature vectors of the original data, the global design primitives are feature vectors output by the Transformer encoder after parsing the global feature vectors, and the product design execution data are executable product design detail process data based on the model configuration parameters.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program. It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer storage medium storing a computer program, It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images