An integrated system for the development of glass aluminum frame doors based on the Internet of Things

Abstract

The invention discloses a glass aluminum frame door research and development integrated system developed based on the Internet of Things, belonging to the technical field of Internet of Things research and development, comprising a demand analysis module, an optimization and adjustment module, a manufacturing monitoring module, an installation guidance module, a quality inspection module, a maintenance management module, a user interaction module, an evolutionary game module and a data analysis module; the invention enhances the scientificity and objectivity of material selection, reduces costs to the greatest extent, improves economic benefits, can efficiently make the best choice, improves the overall quality of the glass aluminum frame door through scientific material selection, thereby improving user satisfaction and market competitiveness, ensuring that resources are optimally configured during the research and development and production processes, can help the system find the best connection method between modules, improve flexibility and response speed, reduce interference from human factors, and improve the accuracy of decision-making.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things research and development, and in particular to a glass aluminum frame door research and development integrated system developed based on the Internet of Things. Background Art

[0002] Glass aluminum frame doors are widely used in construction projects, and the quality of their design and manufacturing directly affects the safety and aesthetics of the building. Traditional design and manufacturing methods often rely on manual experience and a single platform, which makes it difficult to ensure efficiency and consistency. The introduction of Internet of Things technology provides new solutions for the research and development and production of glass aluminum frame doors. With the rapid development of intelligence and informatization today, the traditional manufacturing industry is undergoing a profound change. The development of Internet of Things (IoT) technology provides intelligent, networked, and integrated solutions for various industries. As an important component in the construction field, the design, manufacturing, and use of glass aluminum frame doors are also deeply affected by Internet of Things technology. By deeply integrating Internet of Things technology with the research and development, design, manufacturing, and quality inspection of glass aluminum frame doors, the intelligence level and market competitiveness of the products can be greatly improved.

[0003] The material selection of the existing glass aluminum frame door R&D integration system is affected by human subjective judgment, which increases material costs and reduces economic benefits. It is also impossible to make the best choice efficiently, which reduces the overall quality of the glass aluminum frame door. In addition, the existing glass aluminum frame door R&D integration system cannot optimize the allocation of resources during the R&D and production process, and the system flexibility and response speed are greatly reduced. There are also many human interference factors, which reduce the accuracy of decision-making. Therefore, we propose a glass aluminum frame door R&D integration system based on the Internet of Things. Summary of the invention

[0004] The purpose of the present invention is to solve the defects in the prior art and to propose a glass aluminum frame door research and development integrated system based on the Internet of Things.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A glass aluminum frame door R&D integrated system developed based on the Internet of Things, including a demand analysis module, an optimization and adjustment module, a manufacturing monitoring module, an installation guidance module, a quality inspection module, a maintenance management module, a user interaction module, an evolutionary game module and a data analysis module;

[0007] The demand analysis module is used to mine the user's potential needs and preferences for glass aluminum frame doors.

[0008] As a further solution of the present invention, the specific steps of potential demand and preference mining of the demand analysis module are as follows:

[0009] S1.1: Collect raw data from market research reports, user feedback forms, and social media comments. Preprocess the raw data to remove noise, duplicate data, and incomplete data. Convert unstructured data into structured data. Use the word frequency-inverse document frequency method to extract keywords and topics. Unify the standards for data in different formats.

[0010] S1.2: Use named entity recognition technology to identify users, product features, and key brand entities in the data, and then identify the relationships between entities. Use the identified entities as nodes and relationships as edges to form a graph structure to build a user demand map;

[0011] S1.3: Search for each group of paths in the user demand graph to find the connection between user needs and product features. Identify the most important entities by calculating the centrality of the nodes, and use the LDA topic model to analyze user comments to extract the main needs and preferences. Then, use the Apriori algorithm to mine the associations and potential patterns between user needs, and generate corresponding reference documents based on the mined potential user needs and preferences.

[0012] The optimization and adjustment module includes a design optimization unit, a material selection unit and a manufacturing process unit;

[0013] Among them, the design optimization unit is used to use 3D modeling and simulation technology to design the door body and optimize the design effect; the material selection unit is used to comprehensively evaluate the performance, cost and sustainability of different materials and select the optimal material combination; the manufacturing process module is used to optimize the manufacturing process based on the design plan and material selection results.

[0014] As a further solution of the present invention, the material selection unit selects the optimal material combination in the following specific steps:

[0015] S2.1: Based on the overall goal of material selection, evaluation criteria, specific sub-criteria and candidate materials, a hierarchical model structure diagram is established. For each level of the hierarchical model, a pairwise comparison judgment matrix A = [a ij ] and use a scale of 1 to 9 to indicate the importance of each pair of criteria, where a ij Indicates the importance of criterion i relative to criterion j;

[0016] S2.2: Perform eigenvalue decomposition on the judgment matrix, calculate the eigenvector corresponding to the maximum eigenvalue, and normalize it to obtain the weight vector of each standard. No material is selected as the initial state, and the initial state is used as the root node to initialize the search tree, where each node represents a state and each edge represents a material selection;

[0017] S2.3: Starting from the root node, select child nodes according to the upper confidence interval strategy, and select downward layer by layer until reaching the node that is not fully expanded at present, that is, the leaf node. At the leaf node reached, if the node is not the terminal node, expand the new node, where the new node represents the selection of a new material in the current state;

[0018] S2.4: Starting from the expanded new node, a complete simulation is performed, i.e., materials are randomly selected from the current state until the terminal state is reached, and results are randomly generated according to the predetermined performance, cost and sustainability evaluation criteria, and the cumulative reward of the simulation results is calculated;

[0019] S2.5: Send the cumulative reward of the simulation results back to all nodes on the path, update the number of visits and average rewards of each node, repeat the selection, expansion, simulation and backtracking steps, iterate multiple times, until the preset time or number of simulations is reached;

[0020] S2.6: Combine the weights of each standard calculated by the hierarchical analysis method with the simulation results of the search tree, calculate the comprehensive score of each material combination, and select the material combination with the highest comprehensive score as the optimal solution.

[0021] As a further solution of the present invention, the specific calculation formula of the upper confidence interval strategy described in S2.3 is as follows:

[0022]

[0023] In the formula, Q a represents the score of node a; n a Represents the number of times node a is visited; R j represents the return of j visits; UCB a represents the upper confidence value of node a; C represents the exploration parameter; N represents the number of visits to the parent node.

[0024] The manufacturing monitoring module is used to monitor various parameters in the manufacturing process in real time through IoT sensors and make dynamic adjustments;

[0025] The installation guidance module provides intuitive installation guidance to help installers complete the installation work;

[0026] The quality inspection module performs a comprehensive inspection on the glass aluminum frame door and generates a inspection report.

[0027] As a further solution of the present invention, the specific steps of the quality inspection module glass aluminum frame door inspection are as follows:

[0028] S4.1: Use a high-resolution camera to shoot the finished glass aluminum frame door to obtain images from multiple angles, then convert the color image into a grayscale image, construct a Gaussian kernel based on the filter size and standard deviation selected by the staff, and perform a convolution operation on the Gaussian kernel and the image to obtain a smoothed image;

[0029] S4.2: Calculate the horizontal and vertical gradients of the smoothed image by convolution with the Sobel operator, set high and low thresholds, check each pixel in the image along the gradient direction, mark points with gradient magnitude greater than the high threshold as strong edges, mark points with gradient magnitude between the high and low thresholds as weak edges, and suppress points with gradient magnitude less than the low threshold to zero, so as to obtain the final edge image;

[0030] S4.3: From each set of processed image data, extract the geometric features of the door body's length, width and diagonal, as well as the texture features of the surface smoothness and detail texture. Then compare the extracted geometric features with the standard size to determine whether the size is qualified. Use texture analysis to detect whether there are scratches, bubbles, and cracks on the surface. Then detect the flatness of the edges to determine whether there are any damages or burrs. Feedback the test results to the production link and adjust the production process and parameters.

[0031] The maintenance management module is used to monitor the use status of the door body in real time, predict failures and provide maintenance suggestions;

[0032] The user interaction module is used to build an interactive platform between users and the R&D team for users to feedback problems and suggestions;

[0033] The evolutionary game module is used to simulate the market competition environment and predict the effects of different designs and marketing strategies.

[0034] As a further solution of the present invention, the specific steps of predicting the effects of different designs and marketing strategies of the evolutionary game module are as follows:

[0035] S5.1: Collect information about different manufacturers or products in the market, the design strategies and marketing strategies that each manufacturer or product can choose, and the revenue information under each strategy combination, simulate the market competition process based on the collected market information, and obtain the initial strategy distribution based on the probability of each competitor choosing different strategies at the initial moment;

[0036] S5.2: Based on the strategy selection of each market competitor and the strategy combination of other entities, calculate the average return of each strategy, use the current proportion of each strategy as the weight, and calculate the weighted sum of all strategy returns. Then use the replicator dynamic equation to calculate the proportion change rate of each strategy, and obtain the change in the popularity of the strategy in the market based on the proportion change rate of each strategy;

[0037] S5.3: Update the current strategy proportion according to the change rate of the proportion of each strategy to obtain the strategy distribution of the next time step, repeatedly simulate the evolution of the strategy proportion over time, observe the changes of each strategy proportion, and record the distribution of the strategy proportion at each time step;

[0038] S5.4: Based on the simulation results, identify the strategy information that exists stably in the market, analyze the market performance and characteristics of each strategy, and define the strategy that occupies a large proportion in the market and remains stable as the dominant strategy, define the strategy that is gradually eliminated as the disadvantageous strategy, and record the characteristics and market performance of each dominant strategy;

[0039] S5.5: Based on the identification results of advantageous and disadvantageous strategies, combined with the characteristics of advantageous strategies, optimize the design and marketing plans, put the optimized strategies into the market for testing, and collect feedback data.

[0040] The data analysis module is used to provide decision support for product design, manufacturing and use processes, and to model and analyze the relationships between various R&D processes to identify key links and potential risks.

[0041] As a further solution of the present invention, the specific steps of the key links and potential risk identification of the data analysis module are as follows:

[0042] S6.1: Identify the modules in the system, analyze each module, determine its input and output, list the other modules that each group of modules depends on, and their impact on other modules, with nodes representing modules and edges representing dependencies or impacts;

[0043] S6.2: Draw directed edges between nodes, with the direction of the edges pointing from the influencer to the influenced, to represent the influence or dependency of one module on another. Then, according to the nature of the modules, divide each group of nodes into decision nodes, opportunity nodes, and value nodes. Combine all the identified modules and relationships to draw a complete influence diagram.

[0044] S6.3: Initialize pheromones and heuristic information on the nodes and edges of the influence graph, randomly generate multiple groups of individuals, and initialize the position information of each group of individuals. Each group of individuals calculates the selection probability of each optional node based on the pheromone and heuristic information, and selects the node with the largest probability to move;

[0045] S6.4: At the beginning of the next iteration, volatilize the pheromones on all edges to reduce their concentration to avoid excessive concentration of pheromones on certain paths. According to the quality of the path found by the individual, add pheromones to the path. The higher the quality of the path, the more pheromones are added. Repeat the path construction and pheromone update steps, iterate for multiple rounds of optimization, and evaluate the quality of each path after each iteration.

[0046] S6.5: Record the best paths and their corresponding pheromone concentrations, analyze the paths with the highest pheromone concentrations in the final impact graph, determine the key modules and dependencies on these paths, identify paths with lower pheromone concentrations but that may cause system problems, evaluate the risk factors on these paths, and develop corresponding risk management strategies.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. The present invention establishes a hierarchical model structure diagram based on the overall goal, evaluation criteria, specific sub-criteria and candidate materials of material selection, constructs a judgment matrix for pairwise comparison for the criteria of each layer in the hierarchical model, and uses a scale of 1 to 9 to represent the importance between the two criteria, performs eigenvalue decomposition on the judgment matrix, calculates the eigenvector corresponding to the maximum eigenvalue, and normalizes it to obtain the weight vector of each criterion, takes no material selected as the initial state, takes the initial state as the root node, initializes the search tree, starts from the root node, selects child nodes according to the upper confidence interval strategy, selects layer by layer downward until reaching the leaf node that is not fully expanded at present, if the node is not the terminal node, expands a new node, wherein the new node represents the selection of a new material under the current state, and starts from the expanded new node to perform a complete simulation, that is, from the current state Starting from the previous state, materials are randomly selected until the termination state is reached. Results are randomly generated according to the predetermined performance, cost and sustainability evaluation criteria, and the cumulative rewards of the simulation results are calculated. The cumulative rewards of the simulation results are transmitted back to all nodes on the path, and the number of visits and average returns of each node are updated. The selection, expansion, simulation and backtracking steps are repeated, and it is iterated multiple times until the preset time or number of simulations is reached. The weights of each standard calculated by the hierarchical analysis method are combined with the simulation results of the search tree, and the comprehensive score of each material combination is calculated. The material combination with the highest comprehensive score is selected as the optimal solution, which enhances the scientificity and objectivity of material selection, minimizes costs, improves economic benefits, and can efficiently make the best choice. Through scientific material selection, the overall quality of glass aluminum frame doors is improved, thereby improving user satisfaction and market competitiveness.

[0049] 2. The present invention identifies each module in the system, analyzes each module, determines its input and output, lists other modules that each group of modules depends on, and their influence on other modules, integrates all identified modules and relationships, draws a complete influence diagram, initializes pheromones and heuristic information on the nodes and edges of the influence diagram, randomly generates multiple groups of individuals, and initializes the position information of each group of individuals. Each group of individuals calculates the selection probability of each optional node based on the pheromone and heuristic information, and selects the node with the largest probability to move. At the beginning of the next round of iteration, the pheromones on all edges are volatilized to reduce their concentration to avoid excessive concentration of pheromones on certain paths, and according to the quality of the path found by the individual, the node is moved on the path. Increase pheromones. The higher the path quality, the more pheromones are added. Repeat the path construction and pheromone update steps, and iterate for multiple rounds of optimization. After each round of iteration, evaluate the quality of each path, record the best path and its corresponding pheromone concentration, analyze the paths with the highest pheromone concentration in the final impact graph, determine the key modules and dependencies on these paths, and identify paths with low pheromone concentrations that may cause system problems. Evaluate the risk factors on these paths, formulate corresponding risk management strategies, and ensure that resources are optimally allocated during the R&D and production processes. This can help the system find the best connection method between modules, improve flexibility and response speed, reduce interference from human factors, and improve the accuracy of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0051] Figure 1 This is a system block diagram of a glass aluminum frame door R&D integrated system developed based on the Internet of Things proposed by the present invention. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0053] Example 1

[0054] Reference Figure 1 , a glass aluminum frame door R&D integrated system developed based on the Internet of Things, including demand analysis module, optimization and adjustment module, manufacturing monitoring module, installation guidance module, quality inspection module, maintenance management module, user interaction module, evolutionary game module and data analysis module.

[0055] The demand analysis module is used to explore users' potential needs and preferences for glass aluminum frame doors.

[0056] Specifically, raw data is collected from market research reports, user feedback forms, and social media comments. Noise data, duplicate data, and incomplete data are removed through preprocessing of the raw data, and the unstructured data is converted into structured data. The word frequency-inverse document frequency method is used to extract keywords and topics. At the same time, the standards of data in different formats are unified. Named entity recognition technology is used to identify users, product features, and key brand entities in the data, and then the relationships between entities are identified. The identified entities are used as nodes and the relationships as edges to form a graph structure to construct a user demand map. Each group of paths is searched in the user demand map to find the connection between user needs and product features. The most important entities are identified by calculating the centrality of the nodes, and the LDA topic model is used to analyze user comments to extract major needs and preferences. The Apriori algorithm is then used to mine the associations and potential patterns between user needs, and corresponding reference documents are generated based on the mined user potential needs and preferences.

[0057] The optimization and adjustment module includes a design optimization unit, a material selection unit and a manufacturing process unit.

[0058] Among them, the design optimization unit is used to use 3D modeling and simulation technology to design the door body and optimize the design effect; the material selection unit is used to comprehensively evaluate the performance, cost and sustainability of different materials and select the optimal material combination; the manufacturing process module is used to optimize the manufacturing process based on the design plan and material selection results.

[0059] Specifically, according to the overall goal of material selection, evaluation criteria, specific sub-criteria and candidate materials, a hierarchical model structure diagram is established, and for each level of the hierarchical model, a pairwise comparison judgment matrix A = [a ij ] and use a scale of 1 to 9 to indicate the importance of each pair of criteria, where a ijIndicates the importance of standard i relative to standard j, performs eigenvalue decomposition on the judgment matrix, calculates the eigenvector corresponding to the maximum eigenvalue, and normalizes it to obtain the weight vector of each standard. The initial state is represented by not selecting any material. The initial state is used as the root node to initialize the search tree, where each node represents a state and each edge represents a material selection. Starting from the root node, child nodes are selected according to the upper confidence interval strategy, and selection is made layer by layer until a node that is not fully expanded is reached, that is, a leaf node. At the leaf node reached, if the node is not a terminal node, a new node is expanded, where the new node represents the selection of a new material in the current state, from the expansion Starting from a new node, a complete simulation is carried out, that is, randomly selecting materials from the current state until the terminal state is reached, randomly generating results according to the predetermined performance, cost and sustainability evaluation criteria, and calculating the cumulative reward of the simulation results, and transmitting the cumulative reward of the simulation results back to all nodes on the path, updating the number of visits and average return of each node, repeating the selection, expansion, simulation and backtracking steps, iterating multiple times until the preset time or number of simulations is reached, combining the weights of each standard calculated by the hierarchical analysis method with the simulation results of the search tree, and calculating the comprehensive score of each material combination, and selecting the material combination with the highest comprehensive score as the optimal solution.

[0060] In this embodiment, the specific calculation formula of the upper confidence interval strategy is as follows:

[0061]

[0062] In the formula, Q a represents the score of node a; n a Represents the number of times node a is visited; R j represents the return of j visits; UCB a represents the upper confidence value of node a; C represents the exploration parameter; N represents the number of visits to the parent node.

[0063] The manufacturing monitoring module is used to monitor various parameters in the manufacturing process in real time through the Internet of Things sensors and make dynamic adjustments; the installation guidance module provides intuitive installation guidance to help installers complete the installation work.

[0064] Example 2

[0065] Reference Figure 1 , a glass aluminum frame door R&D integrated system developed based on the Internet of Things, including demand analysis module, optimization and adjustment module, manufacturing monitoring module, installation guidance module, quality inspection module, maintenance management module, user interaction module, evolutionary game module and data analysis module.

[0066] The quality inspection module conducts a comprehensive inspection of the glass aluminum frame door and generates a test report.

[0067] Specifically, a high-resolution camera is used to shoot the finished glass aluminum frame door to obtain images from multiple angles, and then the color image is converted into a grayscale image. A Gaussian kernel is constructed according to the filter size and standard deviation selected by the staff, and the Gaussian kernel is convolved with the image to obtain a smoothed image. The horizontal gradient and vertical gradient of the smoothed image are calculated by convolution with the Sobel operator, and a high threshold and a low threshold are set. Each pixel in the image is checked along the gradient direction. If the point with a gradient amplitude greater than the high threshold is marked as a strong edge, the point with a gradient amplitude between the high threshold and the low threshold is marked as a weak edge, and the point with a gradient amplitude less than the low threshold is suppressed to zero to obtain the final edge image. From each set of processed image data, the length, width and diagonal geometric features of the door body, as well as the surface smoothness and detail texture various texture features are extracted, and then the extracted geometric features are compared with the standard size to determine whether the size is qualified. The surface is detected for scratches, bubbles and cracks through texture analysis, and then the flatness of the edge is detected to determine whether there is damage and burr problems. The test results are fed back to the production link, and the production process and parameters are adjusted.

[0068] The maintenance management module is used to monitor the usage status of the door in real time, predict faults and provide maintenance suggestions; the user interaction module is used to build an interactive platform between users and the R&D team for users to feedback problems and suggestions; the evolutionary game module is used to simulate the market competition environment and predict the effects of different designs and marketing strategies.

[0069] Specifically, different manufacturers or products in the market, the design strategies and marketing strategies that each manufacturer or product can choose, and the profit information under each strategy combination are collected. The market competition process is simulated based on the collected market information, and the initial strategy distribution is obtained based on the probability of each competitor choosing different strategies at the initial moment. The average profit of each strategy is calculated based on the strategy selection of each market competitor and the strategy combination of other entities. The current proportion of each strategy is used as the weight, and the weighted sum of all strategy profits is calculated. The proportion change rate of each strategy is calculated using the replicator dynamic equation, and the popularity change of the strategy in the market is obtained based on the proportion change rate of each strategy. Update the current strategy proportion at the same rate to obtain the strategy distribution of the next time step, repeat the simulation of the evolution of the strategy proportion over time, observe the changes in the proportion of each strategy, and record the distribution of the strategy proportion at each time step. According to the simulation results, identify the strategy information that is stably present in the market, analyze the market performance and characteristics of each strategy, and regard the strategies that occupy a large proportion in the market and remain stable as dominant strategies, and the strategies that are gradually eliminated as disadvantageous strategies, record the characteristics and market performance of each dominant strategy, optimize the design and marketing plan based on the identification results of dominant and disadvantageous strategies and the characteristics of dominant strategies, put the optimized strategies into the market for testing, and collect feedback data.

[0070] The data analysis module is used to provide decision support for product design, manufacturing and use processes, and to model and analyze the relationships between various R&D processes to identify key links and potential risks.

[0071] Specifically, identify each module in the system, analyze each module, determine its input and output, list the other modules that each group of modules depends on, and their influence on other modules, use nodes to represent modules, edges to represent dependencies or influences, draw directed edges between nodes, and the direction of the edges is from the influencer to the affected, to represent the influence or dependence of one module on another module, and then divide each group of nodes into decision nodes, opportunity nodes, and value nodes according to the nature of the modules, combine all the identified modules and relationships, draw a complete influence diagram, initialize pheromones and heuristic information on the nodes and edges of the influence diagram, randomly generate multiple groups of individuals, and initialize the position information of each group of individuals, and each group of individuals calculates the position of each optional node according to the pheromone and heuristic information. Select the probability and choose the node with the largest probability to move. At the beginning of the next round of iteration, volatilize the pheromones on all edges to reduce their concentration to avoid excessive concentration of pheromones on certain paths. According to the quality of the path found by the individual, add pheromones to the path. The higher the quality of the path, the more pheromones are added. Repeat the path construction and pheromone update steps, and iterate for multiple rounds of optimization. After each round of iteration, evaluate the quality of each path, record the best path and its corresponding pheromone concentration, analyze the paths with the highest pheromone concentration in the final impact graph, determine the key modules and dependencies on these paths, and identify paths with low pheromone concentrations that may cause system problems. Evaluate the risk factors on these paths and formulate corresponding risk management strategies.

Claims

1. A glass aluminum frame door R&D integrated system based on the Internet of Things, characterized in that: It includes demand analysis module, optimization and adjustment module, manufacturing monitoring module, installation guidance module, quality inspection module, maintenance management module, user interaction module, evolutionary game module and data analysis module; The demand analysis module is used to dig out the user's potential demand and preference for glass aluminum frame doors; The optimization and adjustment module includes a design optimization unit, a material selection unit and a manufacturing process unit; The design optimization unit is used to design the door body using 3D modeling and simulation technology, and optimize the design effect; the material selection unit is used to comprehensively evaluate the performance, cost and sustainability of different materials, and select the optimal material combination; the manufacturing process unit is used to optimize the manufacturing process according to the design scheme and material selection results; The specific steps of selecting the optimal material combination by the material selection unit are as follows: S2.1: Based on the overall goal of material selection, evaluation criteria, specific sub-criteria and candidate materials, a hierarchical model structure diagram is established. For each level of the hierarchical model, a pairwise comparison judgment matrix A = [a ij ] and use a scale of 1 to 9 to indicate the importance of each pair of criteria, where a ij Indicates the importance of criterion i relative to criterion j; S2.2: Perform eigenvalue decomposition on the judgment matrix, calculate the eigenvector corresponding to the maximum eigenvalue, and normalize it to obtain the weight vector of each standard. No material is selected as the initial state, and the initial state is used as the root node to initialize the search tree, where each node represents a state and each edge represents a material selection; S2.3: Starting from the root node, select child nodes according to the upper confidence interval strategy, and select downward layer by layer until reaching the node that is not fully expanded at present, that is, the leaf node. At the leaf node reached, if the node is not the terminal node, expand the new node, where the new node represents the selection of a new material in the current state; S2.4: Starting from the expanded new node, a complete simulation is performed, i.e., materials are randomly selected from the current state until the terminal state is reached, and results are randomly generated according to the predetermined performance, cost and sustainability evaluation criteria, and the cumulative reward of the simulation results is calculated; S2.5: Send the cumulative reward of the simulation results back to all nodes on the path, update the number of visits and average rewards of each node, repeat the selection, expansion, simulation and backtracking steps, iterate multiple times, until the preset time or number of simulations is reached; S2.6: Combine the weights of each standard calculated by the AHP method with the simulation results of the search tree, calculate the comprehensive score of each material combination, and select the material combination with the highest comprehensive score as the optimal solution; The manufacturing monitoring module is used to monitor various parameters in the manufacturing process in real time through IoT sensors and make dynamic adjustments; The installation guidance module provides intuitive installation guidance to help installers complete the installation work; The quality inspection module performs a comprehensive inspection on the glass aluminum frame door and generates an inspection report; The maintenance management module is used to monitor the use status of the door body in real time, predict faults and provide maintenance suggestions; The user interaction module is used to build an interactive platform between users and the R&D team for users to feedback problems and suggestions; The evolutionary game module is used to simulate the market competition environment and predict the effects of different designs and marketing strategies; The data analysis module is used to provide decision support for product design, manufacturing and use processes, and to model and analyze the relationship between various R&D processes to identify key links and potential risks; The specific steps of key links and potential risk identification of the data analysis module are as follows: S6.1: Identify the modules in the system, analyze each module, determine its input and output, list the other modules that each group of modules depends on, and their impact on other modules, with nodes representing modules and edges representing dependencies or impacts; S6.2: Draw directed edges between nodes, with the direction of the edges pointing from the influencer to the influenced, to represent the influence or dependency of one module on another. Then, according to the nature of the modules, divide each group of nodes into decision nodes, opportunity nodes, and value nodes. Combine all the identified modules and relationships to draw a complete influence diagram. S6.3: Initialize pheromones and heuristic information on the nodes and edges of the influence graph, randomly generate multiple groups of individuals, and initialize the position information of each group of individuals. Each group of individuals calculates the selection probability of each optional node based on the pheromone and heuristic information, and selects the node with the largest probability to move; S6.4: At the beginning of the next iteration, volatilize the pheromones on all edges to reduce their concentration to avoid excessive concentration of pheromones on certain paths. According to the quality of the path found by the individual, add pheromones to the path. The higher the quality of the path, the more pheromones are added. Repeat the path construction and pheromone update steps, iterate for multiple rounds of optimization, and evaluate the quality of each path after each iteration. S6.5: Record the best paths and their corresponding pheromone concentrations, analyze the paths with the highest pheromone concentrations in the final impact graph, determine the key modules and dependencies on these paths, identify paths with lower pheromone concentrations but that may cause system problems, evaluate the risk factors on these paths, and develop corresponding risk management strategies.

2. According to the glass aluminum frame door R&D integrated system developed based on the Internet of Things according to claim 1, it is characterized in that: The specific steps of potential demand and preference mining in the demand analysis module are as follows: S1.1: Collect raw data from market research reports, user feedback forms, and social media comments. Preprocess the raw data to remove noise, duplicate data, and incomplete data. Convert unstructured data into structured data. Use the word frequency-inverse document frequency method to extract keywords and topics. Unify the standards for data in different formats. S1.2: Use named entity recognition technology to identify users, product features, and key brand entities in the data, and then identify the relationships between entities. Use the identified entities as nodes and relationships as edges to form a graph structure to build a user demand map; S1.3: Search for each group of paths in the user demand graph to find the connection between user needs and product features. Identify the most important entities by calculating the centrality of the nodes, and use the LDA topic model to analyze user comments to extract the main needs and preferences. Then, use the Apriori algorithm to mine the associations and potential patterns between user needs, and generate corresponding reference documents based on the mined potential user needs and preferences.

3. According to the glass aluminum frame door R&D integrated system developed based on the Internet of Things as described in claim 2, it is characterized in that: The specific calculation formula for the upper confidence interval strategy described in S2.3 is as follows: In the formula, Q a represents the score of node a; n a Represents the number of times node a is visited; R j represents the return of j visits; UCB a represents the upper confidence value of node a; C represents the exploration parameter; N represents the number of visits to the parent node.

4. According to the glass aluminum frame door R&D integrated system developed based on the Internet of Things as described in claim 3, it is characterized in that: The specific steps of the quality inspection module glass aluminum frame door inspection are as follows: S4.1: Use a high-resolution camera to shoot the finished glass aluminum frame door to obtain images from multiple angles, then convert the color image into a grayscale image, construct a Gaussian kernel based on the filter size and standard deviation selected by the staff, and perform a convolution operation on the Gaussian kernel and the image to obtain a smoothed image; S4.2: Calculate the horizontal and vertical gradients of the smoothed image by convolution with the Sobel operator, set high and low thresholds, check each pixel in the image along the gradient direction, mark points with gradient magnitude greater than the high threshold as strong edges, mark points with gradient magnitude between the high and low thresholds as weak edges, and suppress points with gradient magnitude less than the low threshold to zero, so as to obtain the final edge image; S4.3: From each set of processed image data, extract the geometric features of the door body's length, width and diagonal, as well as the texture features of the surface smoothness and detail texture. Then compare the extracted geometric features with the standard size to determine whether the size is qualified. Use texture analysis to detect whether there are scratches, bubbles, and cracks on the surface. Then detect the flatness of the edges to determine whether there are any damages or burrs. Feedback the test results to the production link and adjust the production process and parameters.

5. According to the glass aluminum frame door R&D integrated system developed based on the Internet of Things as described in claim 1, it is characterized in that: The specific steps for predicting the effects of different designs and marketing strategies of the evolutionary game module are as follows: S5.1: Collect information about different manufacturers or products in the market, the design strategies and marketing strategies that each manufacturer or product can choose, and the revenue information under each strategy combination, simulate the market competition process based on the collected market information, and obtain the initial strategy distribution based on the probability of each competitor choosing different strategies at the initial moment; S5.2: Based on the strategy selection of each market competitor and the strategy combination of other entities, calculate the average return of each strategy, use the current proportion of each strategy as the weight, and calculate the weighted sum of all strategy returns. Then use the replicator dynamic equation to calculate the proportion change rate of each strategy, and obtain the change in the popularity of the strategy in the market based on the proportion change rate of each strategy; S5.3: Update the current strategy proportion according to the change rate of the proportion of each strategy to obtain the strategy distribution of the next time step, repeatedly simulate the evolution of the strategy proportion over time, observe the changes of each strategy proportion, and record the distribution of the strategy proportion at each time step; S5.4: Based on the simulation results, identify the strategy information that exists stably in the market, analyze the market performance and characteristics of each strategy, and define the strategy that occupies a large proportion in the market and remains stable as the dominant strategy, define the strategy that is gradually eliminated as the disadvantageous strategy, and record the characteristics and market performance of each dominant strategy; S5.5: Based on the identification results of advantageous and disadvantageous strategies, combined with the characteristics of advantageous strategies, optimize the design and marketing plans, put the optimized strategies into the market for testing, and collect feedback data.

Images