Intelligent bending and edging system and method for solid wood custom furniture

Abstract

The application discloses an intelligent bending and edge covering system and method for solid wood customized furniture, and belongs to the technical field of furniture manufacturing. The system comprises a data acquisition module, a material analysis module, a demand analysis module, a design optimization module, an identification and positioning module, a path planning module, an edge covering strategy module, a production control module, a quality detection module, a production simulation module and a data storage module. The application reduces the number of parameter combinations that need to be evaluated, thereby reducing the calculation and time costs, can provide a more comprehensive bending path planning scheme, ensures that all design requirements are met, enables the system to quickly adapt to different production requirements and process requirements, improves the versatility and adaptability of the system, enables the system to flexibly cope with different production conditions and requirements, helps to avoid errors caused by experience and subjective judgment in traditional methods, reduces material waste, improves the use efficiency of materials, thereby reduces production costs and improves economic benefits.

Description

Technical Field

[0001] This invention relates to the field of furniture manufacturing technology, and in particular to an intelligent bending and edge-wrapping system and method for custom solid wood furniture. Background Technology

[0002] With the continuous development of the furniture industry, consumers' demand for personalized and high-quality furniture is growing. Solid wood custom furniture, due to its unique natural texture and elegant appearance, has always been regarded as a representative of high-end furniture. As consumers' demand for personalized furniture increases, the custom furniture market is growing rapidly. However, the production process of solid wood furniture faces many challenges, especially in bending and edging techniques. Traditional production processes present several challenges. Bending and edging rely on manual operation, resulting in low efficiency and inconsistent quality, making it difficult to meet modern production needs. These processes are characterized by high labor costs, low production efficiency, and high complexity, especially in bending and edging. Bending and edging are key aspects affecting the quality and appearance of solid wood furniture, and traditional methods struggle to guarantee consistency and precision.

[0003] Existing intelligent bending and edge-wrapping systems and methods for solid wood custom furniture require evaluation of a large number of parameter combinations, resulting in high computational and time costs. Furthermore, they cannot provide comprehensive bending path planning solutions, leading to low system versatility and adaptability. In addition, existing intelligent bending and edge-wrapping systems and methods for solid wood custom furniture suffer from errors caused by experience and subjective judgment, common in traditional methods, increasing material waste and reducing material utilization efficiency. Therefore, we propose an intelligent bending and edge-wrapping system and method for solid wood custom furniture. Summary of the Invention

[0004] The purpose of this invention is to address the deficiencies in the existing technology by proposing an intelligent bending and edge-wrapping system and method for custom solid wood furniture.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] An intelligent bending and edge-binding system for custom solid wood furniture includes a data acquisition module, a material analysis module, a demand analysis module, a design optimization module, an identification and positioning module, a path planning module, an edge-binding strategy module, a production control module, a quality inspection module, a production simulation module, and a data storage module.

[0007] The data acquisition module is used to collect various physical property data of solid wood materials.

[0008] As a further aspect of the present invention, the physical property data of the solid wood material collected by the data acquisition module specifically include density, hardness, humidity, texture characteristics, elastic modulus, bending strength, compressive strength, toughness, shrinkage rate, color difference, oil content, thermal conductivity, sound absorption coefficient, specific gravity, wood type, and chemical composition.

[0009] The material analysis module is used to analyze the collected material data and construct material property maps.

[0010] As a further aspect of the present invention, the specific steps for constructing the material property spectrum by the material analysis module are as follows:

[0011] Q1.1: Clean, normalize and standardize the various physical attribute data collected by the data acquisition module to remove noise and outliers. Treat each piece of solid wood as a node, and the feature vector of the node is composed of the pre-processed physical attribute data. Define the edges between nodes according to the similarity between materials, and the weight of the edge represents the similarity.

[0012] Q1.2: Input the feature vector of each node and the weight matrix of the edge as input data into the pre-trained graph neural network model, collect the features of each group of neighboring nodes j∈Π(i) of node i, and perform feature aggregation, where Π(i) represents the set of neighboring nodes of node i. Then, fuse the node's own features with the aggregated neighbor features, and update the node's feature vector through a set of non-linear activation functions.

[0013] Q1.3: Repeated feature aggregation and feature update are performed, gradually fusing information from more distant nodes through multi-layer graph convolution. The output features of each layer become the input features of the next layer. After K layers of graph convolution, the output node i is the final node embedding representation that integrates the information of the node itself and its multi-layer neighbor nodes.

[0014] Q1.4: Embedding Nodes The data is input into a set of regression models. The predicted value of node i is obtained through the regression model. The loss value between the predicted value and the actual value is calculated using the RMSE loss function. The gradient of the RMSE loss function with respect to each layer of the regression model is obtained based on the calculated loss value. Then, the gradient is passed back to each layer through the backpropagation algorithm, and the parameters of each layer are updated. Through multiple iterations of training, the model parameters are continuously updated until the range of change of the loss function reaches the preset convergence threshold.

[0015] Q1.5: The graph neural network model collects real-time physical property data of various solid wood materials and processes them through graph convolutional layers to obtain real-time node embeddings. Then, through the trained regression model, the predicted value of node i is obtained and used as the output of the predicted material lifespan.

[0016] As a further aspect of the present invention, the specific calculation formula for feature aggregation in Q1.2 is as follows:

[0017]

[0018] In the formula, Represents the feature aggregation result of node i's neighbors at layer k; Π(i) represents the set of neighboring nodes of node i; c ij The normalization coefficient represents the relationship between node i and its neighbor node j; The feature vector representing node j at the k-th layer;

[0019] The specific calculation formula for updating the feature vector of the node mentioned in Q1.2 is as follows:

[0020]

[0021] In the formula, W represents the feature vector of node i in the (k+1)th layer; σ represents the non-linear activation function, usually ReLU; (k) This represents the weight matrix of the k-th layer; The feature vector representing node i at the k-th layer.

[0022] The demand analysis module is used to integrate customer needs, market trends and historical order data to analyze customer customization needs in order to generate the optimal design solution.

[0023] The design optimization module is used to balance material costs, production efficiency, and aesthetics to optimize the generated design scheme;

[0024] The identification and positioning module is used to accurately identify and locate solid wood materials.

[0025] As a further aspect of the present invention, the specific steps for the identification and positioning module to accurately identify and locate solid wood materials are as follows:

[0026] Q2.1: Collect image data containing solid wood materials, label the solid wood materials in the images, generate a training dataset, use the image data in the training dataset as input images, input them into a pre-trained convolutional neural network, extract high-level feature representations, and obtain the corresponding feature maps;

[0027] Q2.2: The extracted feature map is transformed into a query matrix, a key matrix, and a value matrix through three different linear transformations. The dot product of the query matrix and the key matrix is ​​calculated, and the calculated dot product is scaled. The attention weights are obtained through the softmax function.

[0028] Q2.3: Based on the preset number of attention heads, the query matrix, key matrix, and value matrix are divided into equal parts, and then the query matrix, key matrix, and value matrix are reshaped respectively. For each attention head, the attention is calculated using the corresponding query matrix, key matrix, and value matrix.

[0029] Q2.4: Concatenate the outputs of all attention heads together, then generate the final output through a linear transformation. Concatenate the features processed by the multi-head attention mechanism with the original features, and use a matching algorithm to find the location of the solid wood material in the concatenated feature map. At the same time, output the position coordinates of the solid wood material in the image and its bounding box.

[0030] The path planning module is used to plan the bending path based on the material shape and location data.

[0031] As a further aspect of the present invention, the specific steps for the path planning module to plan the bending path are as follows:

[0032] Q3.1: Randomly generate multiple sets of bending paths as the initial population, where each individual in the population represents a bending path. Calculate the bending path P for each set based on path length, processing time, and energy consumption. l fitness value f(P) l Then, individuals with fitness higher than a preset threshold are selected for the next generation using a roulette wheel selection method.

[0033] Q3.2: Randomly select two parent individuals P from the population. a With P b Multiple crossover points are randomly selected in the gene sequence of an individual, and gene segments of the parents are exchanged at each crossover point to generate new offspring individuals. Then, one or more groups of individuals are randomly selected from the current population, and one or more mutation points are randomly selected in the gene sequence of the individuals, and the gene value is perturbed according to the Gaussian distribution.

[0034] Q3.3: Randomly generate the initial position and initial velocity of each group of individuals in the initial population, and calculate the fitness value of the current position of each group of individuals. Compare the fitness value of the current position of each group of individuals with the fitness value of the historical best position of that individual. If the current fitness value is better than the historical best fitness value, then update the individual's best position to the current position.

[0035] Q3.4: Use the Gaussian process as a surrogate model to fit the existing evaluation data, and randomly select a portion of individuals for real evaluation to obtain the corresponding fitness values ​​as training data. Then use the surrogate model to calculate the expected improvement value of all candidate individuals, and select the individual with the largest expected improvement value for actual experiment or simulation to obtain its true fitness value.

[0036] Q3.5: Add the newly evaluated individuals and their fitness values ​​to the training dataset, update the surrogate model, update the individual information based on the real evaluated fitness values, adjust the population, and then use the new real fitness values ​​to update the individual's velocity and position.

[0037] Q3.6: Repeat the iteration and update the population until the preset maximum number of generations is reached or the fitness change value is within the preset convergence interval. Select the bending path with the highest fitness value from the final population as the optimal solution and apply it to the actual bending edge operation. At the same time, dynamically adjust the bending path according to the actual bending situation.

[0038] The edge-binding strategy module is used to generate the optimal edge-binding strategy based on real-time bending data;

[0039] The production control module is used to monitor the production process in real time and adjust production parameters.

[0040] The quality inspection module is used to inspect the quality of finished products and identify potential quality defects;

[0041] The production simulation module is used to generate a three-dimensional model of the furniture and to simulate the entire production process.

[0042] The data storage module is used to store all production data, testing data, and optimization parameters.

[0043] A smart bending and edge-binding method for custom solid wood furniture, the specific steps of which are as follows:

[0044] Ⅰ: Collect solid wood material data, construct material property maps, and identify materials;

[0045] II: Acquire images of solid wood materials and extract image feature data;

[0046] III: Plan the bending path based on the material properties and identification results;

[0047] IV: Design an edging scheme and simulate the actual edging process to verify its feasibility and effectiveness;

[0048] V: Optimize the edging scheme and implement it using an intelligent bending edging system;

[0049] VI: Conduct a quality assessment and make adjustments to the completed bent and edged furniture.

[0050] As a further aspect of the present invention, the specific steps of the optimized edge-binding scheme in step V are as follows:

[0051] Q4.1: Based on the material properties and bending path, generate an initial edge-binding scheme, take the current edge-binding scheme as the root node, take the results of different edge-binding strategies as child nodes, and build a corresponding search tree based on the root node and each group of child nodes;

[0052] Q4.2: Calculate the upper limit of the confidence interval for each child node. Starting from the root node, select the child node with the highest upper limit of the confidence interval. Check if the child node has any child nodes that have not been fully explored. If so, expand the node. If not, the current child node is considered fully expanded and the process returns to the parent node. Continue traversing and repeat the selection and expansion until the preset depth or leaf node is reached.

[0053] Q4.3: Starting from the selected child node, perform the simulation, randomly select an executable action, execute the selected action to transition to the next state, update the current node state, calculate the current node's reward through the potential function, continue the simulation until the preset depth is reached or the leaf node is terminated, and record the current path and the corresponding reward.

[0054] Q4.4: Starting from the leaf node, backtrack the nodes and corresponding rewards in the simulation path back to the root node step by step, and update the number of visits for each node and action in the path, while also updating the average reward for each node and action in the path.

[0055] Q4.5: Calculate the potential function of each node using the updated node revenue, determine the revenue increment based on the calculated potential function value, and update the edge parameters in the edge-binding scheme. If the update of any edge parameter causes the potential function to increase, retain the update; otherwise, restore the original state and try updating other variables.

[0056] Q4.6: Repeat the search and update of the edge-binding parameters in the edge-binding scheme until the change value of the potential function is within the preset convergence interval. Then, stop the search and update, and starting from the root node, select the optimal path with the highest current search tree profit value as the optimal edge-binding scheme. Record each edge-binding parameter in the optimal scheme and its corresponding value, and output the edge-binding scheme.

[0057] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0058] 1. This invention randomly generates multiple sets of bending paths as the initial population. The fitness value of each bending path is calculated based on path length, processing time, and energy consumption. Then, an individual with a fitness higher than a preset threshold is selected for the next generation using a roulette wheel selection method. Two sets of parent individuals are randomly selected from the population, and crossover and mutation operations are performed. The initial position and initial velocity of each individual in the initial population are randomly generated, and the fitness value of each individual's current position is calculated. The fitness value of each individual's current position is compared with the fitness value of its historical best position. If the current fitness value is better than the historical best fitness value, the individual's best position is updated to the current position. A Gaussian process is used as a surrogate model to fit the existing evaluation data, and a portion of individuals are randomly selected for real evaluation to obtain the corresponding fitness values ​​as training data. The surrogate model is then used to calculate the expected improvement value of all candidate individuals, and the selected individual is... The system selects the individual with the highest expected improvement value for actual experiments or simulations to obtain its true fitness value. The newly evaluated individual and its fitness value are then added to the training dataset to update the surrogate model. Individual information is updated based on the true fitness value, and the population is adjusted. The velocity and position of the individual are then updated using the new true fitness value. This iterative process is repeated until a preset maximum number of generations is reached or the fitness change value is within a preset convergence interval. The bending path with the highest fitness value is selected from the final population as the optimal solution and applied to the actual bending and edge-wrapping operation. Simultaneously, the bending path is dynamically adjusted based on the actual bending situation, reducing the number of parameter combinations that need to be evaluated, thereby reducing computational and time costs. This provides a more comprehensive bending path planning scheme, ensuring that all design requirements are met, enabling the system to quickly adapt to different production needs and process requirements, and improving the system's versatility and adaptability.

[0059] 2. This invention uses the current edge-binding scheme as the root node and the results of different edge-binding strategies as child nodes. A corresponding search tree is constructed based on the root node and each group of child nodes. The upper limit of the confidence interval for each child node is calculated. Starting from the root node, the child node with the highest upper limit of the confidence interval is selected. It is checked whether this child node has any unexplored child nodes. If so, the node is expanded. This selection and expansion process is repeated until a preset depth or leaf node is reached. Simulation is performed starting from the selected child node, and the nodes and corresponding rewards in the simulation path are gradually backtracked to the root node. The number of visits to each node and action is updated. The updated node rewards are used to calculate the potential function of each node. The reward increment is determined based on the calculated potential function value, and the edge binding is updated simultaneously. If updating any of the edge-binding parameters in the scheme leads to an increase in the potential function, the update is retained; otherwise, the original state is restored and updates to other variables are attempted. This process of searching and updating the edge-binding parameters in the scheme is repeated until the change in the potential function is within a preset convergence interval. At this point, the search and update process terminates, and starting from the root node, the optimal path with the highest current search tree return value is selected as the optimal edge-binding scheme. Each edge-binding parameter in the optimal scheme and its corresponding value are recorded, and the scheme is output. This allows the system to flexibly respond to different production conditions and requirements, helping to avoid errors caused by experience and subjective judgment common in traditional methods, reducing material waste, improving material utilization efficiency, thereby reducing production costs and increasing economic benefits. Attached Figure Description

[0060] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0061] Figure 1 This is a system block diagram of an intelligent bending and edge-wrapping system for custom solid wood furniture proposed in this invention;

[0062] Figure 2 This is a flowchart of an intelligent bending and edge-wrapping method for custom solid wood furniture proposed in this invention. Detailed Implementation

[0063] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0064] Example 1

[0065] Reference Figure 1An intelligent bending and edge-binding system for custom solid wood furniture includes a data acquisition module, a material analysis module, a demand analysis module, a design optimization module, an identification and positioning module, a path planning module, an edge-binding strategy module, a production control module, a quality inspection module, a production simulation module, and a data storage module.

[0066] The data acquisition module is used to collect various physical property data of solid wood materials.

[0067] It should be further explained that the physical property data of solid wood materials collected by the data acquisition module specifically include density, hardness, moisture content, texture characteristics, modulus of elasticity, bending strength, compressive strength, toughness, shrinkage rate, color difference, oil content, thermal conductivity, sound absorption coefficient, specific gravity, wood species, and chemical composition.

[0068] The materials analysis module is used to analyze the collected materials data and construct materials property maps.

[0069] Specifically, the various physical attribute data collected by the data acquisition module are cleaned, normalized, and standardized to remove noise and outliers. Each piece of solid wood is considered a node, and the feature vector of this node is composed of the pre-processed physical attribute data. Edges between nodes are defined based on the similarity between materials, and the weight of the edge represents the similarity. The feature vector of each node and the weight matrix of the edge are used as input data into the pre-trained graph neural network model. Features of each group of neighboring nodes j∈Π(i) of node i are collected and feature aggregation is performed, where Π(i) represents the set of neighboring nodes of node i. Then, the node's own features are fused with the aggregated neighbor features, and the node's feature vector is updated through a set of non-linear activation functions. Feature aggregation and feature updating are repeated. Information from more distant nodes is gradually fused through multi-layer graph convolution, and the output features of each layer become the input features of the next layer. After K layers of graph convolution, the final node embedding representation of node i is output, which integrates the information of the node itself and its multi-layer neighboring nodes. Embedding nodes The data is input into a set of regression models. Through the regression models, the predicted value of node i is obtained. The loss value between the predicted value and the actual value is calculated using the RMSE loss function. Based on the calculated loss value, the gradient of the RMSE loss function with respect to each layer of the regression model is obtained. Then, the gradient is passed back to each layer through the backpropagation algorithm, and the parameters of each layer are updated. Through multiple iterations of training, the model parameters are continuously updated until the range of the loss function reaches the preset convergence threshold. The graph neural network model collects real-time physical property data of various solid wood materials and processes them through graph convolutional layers to obtain real-time node embeddings. Then, through the trained regression model, the predicted value of node i is obtained and output as the predicted material lifespan.

[0070] In this embodiment, the specific calculation formula for feature aggregation is as follows:

[0071]

[0072] In the formula, Represents the feature aggregation result of node i's neighbors at layer k; Π(i) represents the set of neighboring nodes of node i; c ij The normalization coefficient represents the relationship between node i and its neighbor node j; The feature vector representing node j at the k-th layer;

[0073] The specific calculation formula for updating the feature vector of a node is as follows:

[0074]

[0075] In the formula, W represents the feature vector of node i in the (k+1)th layer; σ represents the non-linear activation function, usually ReLU; (k) This represents the weight matrix of the k-th layer; The feature vector representing node i at the k-th layer.

[0076] The demand analysis module integrates customer needs, market trends, and historical order data to analyze customer customization requirements and generate the optimal design solution; the design optimization module balances material costs, production efficiency, and aesthetics to optimize the generated design solution; and the identification and positioning module accurately identifies and positions solid wood materials.

[0077] Specifically, image data containing solid wood materials are collected, and the solid wood materials in the images are labeled to generate a training dataset. The image data in the training dataset is used as input images and fed into a pre-trained convolutional neural network to extract high-level feature representations and obtain corresponding feature maps. The extracted feature maps are then subjected to three different linear transformations to generate query matrices, key matrices, and value matrices. The dot product of the query matrix and key matrix is ​​calculated and scaled. The attention weights are obtained through the softmax function. Based on the preset number of attention heads, the query matrix, key matrix, and value matrix are equally divided and then reshaped. For each attention head, the attention is calculated using the corresponding query matrix, key matrix, and value matrix. The outputs of all attention heads are concatenated and then the final output is generated through a linear transformation. The features processed by the multi-head attention mechanism are then concatenated with the original features, and a matching algorithm is used to find the location of the solid wood material in the concatenated feature map. The position coordinates of the solid wood material in the image and its bounding box are output.

[0078] The path planning module is used to plan the bending path based on the material shape and location data.

[0079] Specifically, multiple sets of bending paths are randomly generated as the initial population, where each individual in the population represents a bending path. The bending path P for each set is calculated based on the path length, processing time, and energy consumption. l fitness value f(P) l Then, using a roulette wheel selection method, individuals with fitness higher than a preset threshold are selected to enter the next generation. Two sets of parent individuals P are randomly selected from the population. a With P b Multiple crossover points are randomly selected in the gene sequences of individuals. At each crossover point, gene segments from the parents are exchanged to generate new offspring. Then, one or more groups of individuals are randomly selected from the current population, and one or more mutation points are randomly selected in their gene sequences. The gene values ​​are perturbed according to a Gaussian distribution. The initial position and initial velocity of each group of individuals in the initial population are randomly generated, and the fitness value of each group of individuals at its current position is calculated. The fitness value of each group of individuals at its current position is compared with the fitness value of the individual's historical best position. If the current fitness value is better than the historical best fitness value, the individual's best position is updated to the current position. The Gaussian process is used as a surrogate model to fit the existing evaluation data, and a portion of individuals are randomly selected for true evaluation. The system performs real-world evaluations to obtain corresponding fitness values ​​as training data. Then, it uses a surrogate model to calculate the expected improvement value of all candidate individuals and selects the individual with the largest expected improvement value for actual experiments or simulations to obtain its true fitness value. The newly evaluated individuals and their fitness values ​​are added to the training dataset, the surrogate model is updated, and the individual information is updated based on the true fitness value. The population is then adjusted, and the velocity and position of the individuals are updated using the new true fitness value. This process is repeated iteratively, updating the population, until a preset maximum number of generations is reached or the fitness change value is within a preset convergence interval. The bending path with the highest fitness value is selected from the final population as the optimal solution and applied to the actual bending edge operation. At the same time, the bending path is dynamically adjusted according to the actual bending situation.

[0080] The edge binding strategy module is used to produce the optimal edge binding strategy based on real-time bending data; the production control module is used to monitor the production process in real time and adjust production parameters; the quality inspection module is used to inspect the finished products and identify potential quality defects; the production simulation module is used to generate a 3D model of the furniture and simulate the entire production process; and the data storage module is used to store all production data, inspection data, and optimization parameters.

[0081] Example 2

[0082] Reference Figure 2 A smart bending and edge-binding method for custom solid wood furniture, the specific steps of which are as follows:

[0083] Ⅰ: Collect solid wood material data, construct material property maps, and identify materials.

[0084] II: Image acquisition and extraction of image feature data for solid wood materials.

[0085] III: Plan the bending path based on the material properties and identification results.

[0086] IV: Design an edge binding scheme and simulate the actual edge binding process to verify its feasibility and effectiveness.

[0087] V: Optimize the edge binding scheme and implement it in practice using an intelligent bending edge binding system.

[0088] Specifically, based on material properties and bending paths, an initial edge-binding scheme is generated. The current edge-binding scheme is used as the root node, and the results of different edge-binding strategies are used as child nodes. A corresponding search tree is constructed based on the root node and each group of child nodes. The upper limit of the confidence interval for each child node is calculated. Starting from the root node, the child node with the highest upper limit of the confidence interval is selected. It is checked whether this child node has any unexplored child nodes. If so, the node is expanded. If not, the current child node is considered fully expanded, and the process returns to the parent node. The selection and expansion are repeated until a preset depth or leaf node is reached. Starting from the selected child node, simulation is performed. An executable action is randomly selected, and the selected action is executed to transition to the next state. The current node state is updated, and the benefit of the current node is calculated using a potential function. The simulation continues until a preset depth or a terminating leaf node is reached, and the results are recorded. Starting from the leaf node, the current path and its corresponding reward are simulated by backtracking back to the root node. For each node and action in the path, the number of visits is updated, and the average reward of each node and action is also updated. The potential function of each node is calculated using the updated node reward, and the reward increment is determined based on the calculated potential function value. At the same time, the edge-binding parameters in the edge-binding scheme are updated. If the update of any edge-binding parameter leads to an increase in the potential function, the update is retained; otherwise, the original state is restored, and updates of other variables are tried. The search and edge-binding parameter updates in the edge-binding scheme are repeated until the change in the potential function value is within the preset convergence interval. Then, the search and update are terminated, and starting from the root node, the optimal path with the highest reward value in the current search tree is selected as the optimal edge-binding scheme. Each edge-binding parameter in the optimal scheme and its corresponding value are recorded, and the edge-binding scheme is output.

[0089] VI: Conduct a quality assessment and make adjustments to the completed bent and edged furniture.

Claims

1. A smart bending and edge-binding method for custom solid wood furniture, characterized in that, The intelligent bending and edge-binding system for custom solid wood furniture includes a data acquisition module, a material analysis module, a demand analysis module, a design optimization module, an identification and positioning module, a path planning module, an edge-binding strategy module, a production control module, a quality inspection module, a production simulation module, and a data storage module. The data acquisition module is used to collect various physical property data of solid wood materials; The material analysis module is used to analyze the collected material data and construct material property maps; The demand analysis module is used to integrate customer needs, market trends and historical order data to analyze customer customization needs in order to generate the optimal design solution. The design optimization module is used to balance material costs, production efficiency, and aesthetics to optimize the generated design scheme; The identification and positioning module is used for accurate identification and positioning of solid wood materials; The path planning module is used to plan the bending path based on the material shape and location data; The edge-binding strategy module is used to generate the optimal edge-binding solution based on real-time bending data; The production control module is used to monitor the production process in real time and adjust production parameters. The quality inspection module is used to inspect the quality of finished products and identify potential quality defects; The production simulation module is used to generate a three-dimensional model of the furniture and to simulate the entire production process. The data storage module is used to store all production data, testing data, and optimization parameters; The intelligent bending and edge-binding method comprises the following steps: Ⅰ: Collect solid wood material data, construct material property maps, and identify materials; II: Acquire images of solid wood materials and extract image feature data; III: Plan the bending path based on the material properties and identification results; The specific steps for planning the bending path are as follows: Q3.1: Randomly generate multiple sets of bending paths as the initial population, where each individual in the population represents a bending path. Calculate the bending path for each set based on path length, processing time, and energy consumption. fitness value Then, individuals with fitness higher than a preset threshold are selected for the next generation using a roulette wheel selection method; Q3.2: Randomly select two groups of parent individuals from the population. and Multiple crossover points are randomly selected in the gene sequence of an individual. Gene segments of the parents are exchanged at each crossover point to generate new offspring. Then, one or more groups of individuals are randomly selected from the current population, and one or more mutation points are randomly selected in the gene sequence of the individuals. The gene value of the mutation point is perturbed according to the Gaussian distribution. Q3.3: Randomly generate the initial position and initial velocity of each group of individuals in the initial population, and calculate the fitness value of the current position of each group of individuals. Compare the fitness value of the current position of each group of individuals with the fitness value of the historical best position of that group of individuals. If the current fitness value is better than the historical best fitness value, then update the individual's best position to the current position. Q3.4: Use the Gaussian process as a surrogate model to fit the existing evaluation data, and randomly select a portion of individuals for real evaluation to obtain the corresponding fitness values ​​as training data. Then use the surrogate model to calculate the expected improvement value of all candidate individuals, and select the individual with the largest expected improvement value for actual experiment or simulation to obtain its true fitness value. Q3.5: Add the newly evaluated individuals and their fitness values ​​to the training dataset, update the surrogate model, update the individual information based on the real evaluated fitness values, adjust the population, and then use the new real fitness values ​​to update the individual's velocity and position. Q3.6: Repeat the iteration and update the population until the preset maximum number of generations is reached or the fitness change value is within the preset convergence interval. Select the bending path with the highest fitness value from the final population as the optimal solution and apply it to the actual bending edge operation. At the same time, dynamically adjust the bending path according to the actual bending situation. IV: Design an edging scheme and simulate the actual edging process to verify its feasibility and effectiveness; V: Optimize the edging scheme and implement it using an intelligent bending edging system; The specific steps of the optimized edge-binding scheme are as follows: Q4.1: Based on the material properties and bending path, generate an initial edge-binding scheme, take the current edge-binding scheme as the root node, take the results of different edge-binding strategies as child nodes, and build a corresponding search tree based on the root node and each group of child nodes; Q4.2: Calculate the upper limit of the confidence interval for each child node. Starting from the root node, select the child node with the highest upper limit of the confidence interval. Check if the child node has any child nodes that have not been fully explored. If so, expand the child node. If not, the current child node is considered fully expanded and returns to the parent node. Continue traversing and repeat the selection and expansion until the preset depth or leaf node is reached. Q4.3: Starting from the selected child node, perform the simulation, randomly select an executable action, execute the selected action to transition to the next state, update the current node state, calculate the current node's reward through the potential function, continue the simulation until the preset depth is reached or the leaf node is terminated, and record the current path and the corresponding reward. Q4.4: Starting from the leaf node, backtrack the nodes and corresponding rewards in the simulation path back to the root node step by step, and update the number of visits for each node and action in the path, while also updating the average reward for each node and action in the path. Q4.5: Calculate the potential function of each node using the updated node revenue, determine the revenue increment based on the calculated potential function value, and update the edge parameters in the edge-binding scheme. If the update of any edge parameter causes the potential function to increase, retain the update; otherwise, restore the original state and try updating other variables. Q4.6: Repeat the search and update of the edge-binding parameters in the edge-binding scheme until the change value of the potential function is within the preset convergence interval. Then, stop the search and update, and starting from the root node, select the optimal path with the highest current search tree profit value as the optimal edge-binding scheme. Record each edge-binding parameter in the optimal scheme and its corresponding value, and output the edge-binding scheme. VI: Conduct a quality assessment and make adjustments to the completed bent and edged furniture.

2. The intelligent bending and edge-binding method for custom solid wood furniture according to claim 1, characterized in that, The specific steps for constructing the material property map are as follows: Q1.1: Clean, normalize and standardize the various physical attribute data collected by the data acquisition module to remove noise and outliers. Treat each piece of solid wood as a node, and the feature vector of the node is composed of the pre-processed physical attribute data. Define the edges between nodes according to the similarity between materials, and the weight of the edge represents the similarity. Q1.2: Input the feature vector of each node and the weight matrix of the edges as input data into the pre-trained graph neural network model, and collect data for each group of nodes. Neighbor nodes The features are then aggregated, where Representative node The neighbor node set is then used to fuse the node's own features with the aggregated neighbor features, and the node's feature vector is updated using a set of non-linear activation functions. Q1.3: Repeatedly perform feature aggregation and feature updates, gradually fusing information from more distant nodes through multi-layer graph convolution, with the output features of each layer becoming the input features of the next layer. After K layers of graph convolution, the output nodes... The final node embedding representation that integrates information about the node itself and its multi-level neighbor nodes. ; Q1.4: Embedding Nodes The data is input into a set of regression models, and the nodes are obtained through the regression models. The predicted value is calculated using the RMSE loss function to determine the loss between the predicted value and the actual value. The gradient of the RMSE loss function for each layer of the regression model is obtained based on the calculated loss value. Then, the gradient is passed back to each layer through the backpropagation algorithm, and the parameters of each layer are updated. Through multiple iterations of training, the model parameters are continuously updated until the range of the loss function reaches the preset convergence threshold. Q1.5: The graph neural network model collects real-time physical property data of various solid wood materials and processes it through graph convolutional layers to obtain real-time node embeddings. Then, through a trained regression model, the node embeddings are obtained. The predicted value is then output as the predicted material lifespan.

3. The intelligent bending and edge-binding method for custom solid wood furniture according to claim 1, characterized in that, This also includes the precise identification and positioning of solid wood materials, with the following specific steps: Q2.1: Collect image data containing solid wood materials, label the solid wood materials in the images, generate a training dataset, use the image data in the training dataset as input images, input them into a pre-trained convolutional neural network, extract high-level feature representations, and obtain the corresponding feature maps; Q2.2: The extracted feature map is transformed into a query matrix, a key matrix, and a value matrix through three different linear transformations. The dot product of the query matrix and the key matrix is ​​calculated, and the calculated dot product is scaled. The attention weights are obtained through the softmax function. Q2.3: Based on the preset number of attention heads, the query matrix, key matrix, and value matrix are divided into equal parts, and then the query matrix, key matrix, and value matrix are reshaped respectively. For each attention head, the attention is calculated using the corresponding query matrix, key matrix, and value matrix. Q2.4: Concatenate the outputs of all attention heads together, then generate the final output through a linear transformation. Concatenate the features processed by the multi-head attention mechanism with the original features, and use a matching algorithm to find the location of the solid wood material in the concatenated feature map. At the same time, output the position coordinates of the solid wood material in the image and its bounding box.

4. The intelligent bending and edge-binding method for custom solid wood furniture according to claim 1, characterized in that, The data acquisition module collects physical property data of solid wood materials, including density, hardness, humidity, texture characteristics, modulus of elasticity, bending strength, compressive strength, toughness, shrinkage rate, color difference, oil content, thermal conductivity, sound absorption coefficient, specific gravity, wood type, and chemical composition.

Images