Furniture door panel production method and system based on intelligent manufacturing

Abstract

The present application belongs to the technical field of automatic control of production process, and discloses a furniture door plate production method and system based on intelligent manufacturing, which comprises the following steps: obtaining a to-be-produced furniture door plate and a raw plate, and constructing a edge line manufacturing object; obtaining an edge line stable state by analyzing the mapping relationship between the characteristics of the cutting process, the characteristics of the edge sealing process and the quality detection results; constructing a candidate manufacturing path set, and obtaining a simulated edge line stable state by simulating and distributing the candidate manufacturing path; obtaining an optimal manufacturing path by analyzing the relevance of the simulated edge line stable state on the historical deviation characteristics and the equipment occupation state; predicting the optimal manufacturing path based on the edge line manufacturing object, detecting the edge line stability of the door plate after the edge sealing is completed, and rearranging the optimal manufacturing path by comparing the prediction result with the detection result. The present application realizes the edge line expression, dynamic deduction and manufacturing path closed-loop optimization of the hidden differences of the base material.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology for production processes, and more specifically, to a method and system for producing furniture door panels based on intelligent manufacturing. Background Technology

[0002] Furniture door panel production typically includes processes such as raw panel cutting, edge banding, conveying and transfer, quality inspection, and subsequent sorting. As customized furniture orders develop towards multi-specification, small-batch, and mixed-flow production, the production process is gradually incorporating coordinated control among cutting equipment, edge banding equipment, conveying and switching equipment, and quality inspection equipment to improve door panel processing efficiency, production continuity, and quality consistency.

[0003] Existing production methods for door panel processing typically arrange the processing flow based on board specifications, door panel dimensions, edge banding type, or equipment status. However, they do not adequately consider the continuous impact of the door panel edge line's origin area in the original board, the cutting direction, and the characteristics of the edge banding process on edge line stability. Some latent differences in the substrate do not directly manifest as obvious defects during the cutting stage but gradually reflect as edge bulging, edge line retraction, localized loose bonding, or uneven edge lines during edge banding pressing, cooling release, or subsequent inspection. Existing methods struggle to transform these latent differences into edge line-level states that can sustainably participate in manufacturing control. They also find it difficult to simulate and make decisions based on the edge line stability state for different cutting parameters, edge banding parameters, and manufacturing paths, leading to the inability to optimize subsequent manufacturing parameters and processes in a timely manner. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a furniture door panel production method based on intelligent manufacturing, comprising: S1. Obtain the furniture door panel and the original panel to be produced, identify the source area and edge banding category of each edge banding line to be sealed in the original panel, and construct the edge manufacturing object; S2. Extract the characteristics of the cutting process, the edge sealing process, and the quality inspection results of the edge manufacturing object. By analyzing the mapping relationship between the characteristics of the cutting process, the edge sealing process, and the quality inspection results, obtain the stable state of the edge. S3. By defining and connecting the nodes of the cutting process characteristics and the edge sealing process characteristics, a set of candidate manufacturing paths is constructed. By simulating the allocation of candidate manufacturing paths, the stable state of the simulated edge line is obtained. S4. By analyzing the correlation between the historical deviation characteristics of the simulated edge stability state and the equipment occupancy state, a decision is made on the candidate manufacturing path to obtain the optimal manufacturing path. S5. Predict the optimal manufacturing path based on the edge manufacturing object, and after the edge sealing is completed, test the stability of the door panel edge. By comparing the prediction results and the test results, rearrange the optimal manufacturing path.

[0005] Preferably, the method in S2 includes: S21. Obtain the feature parameters of the historical cutting stage and construct a training dataset of historical edge samples. S22. Obtain the quality detection result dataset corresponding to the historical edge sample dataset, construct the quality state matrix, use the principal component analysis algorithm to reduce the dimensionality of the quality state matrix, extract the principal state factors, and generate supervised learning label vectors. S23. Use the historical edge sample training dataset and supervised learning label vectors to train and obtain the edge stable state initialization model; S24. Obtain the feature parameters of the cutting stage of the edge manufacturing object, input the feature parameters of the cutting stage into the line stability initialization model, obtain the prediction result corresponding to the edge manufacturing object, and generate the initial edge stability state. S25. During the edge sealing process, collect the characteristic parameters of the edge sealing stage and perform state deduction on the initial edge line stability state to obtain the edge line stability state.

[0006] Preferably, the method in S25 includes: S251. Read the source region of the edge manufacturing object, and calculate the normalized occurrence frequency of abnormalities in the source edge of each region after edge sealing in the historical production data to obtain the regional risk coefficient. S252. Obtain the angle between the edge texture direction and the original board texture direction, and correct the area risk coefficient based on the angle to obtain the stress release tendency value. S253. Collect the pressing load sequence when the edge line passes through the pressing mechanism and the feed speed sequence of the edge line in the edge sealing equipment, and calculate the bonding sensitivity tendency value of the pressing load sequence and the feed speed sequence. S254. Obtain historical production samples of the same edge manufacturing object, execute S251 to S253 for each historical production sample and perform range statistics respectively to obtain the initial value range of the corresponding stress release tendency value and the fit sensitivity tendency value respectively. S255. Within the initial value range, randomly sample and generate multiple particle samples according to a preset probability distribution, and construct the state vector of each particle sample. S256. Obtain the characteristic parameters of the edge sealing stage, construct the edge sealing process input vector, and construct the state space transition model based on the spatial mapping relationship between the state vector and the edge sealing process input vector. S257. Calculate the probability distribution of each particle's evolution from the previous state to the current state using the state-space transition model. S258. Obtain the observations for each particle, calculate the particle weights based on the degree of matching and probability distribution between the observations and the predicted state, and calculate the edge stable state based on the particle weights.

[0007] Preferably, the method for calculating the edge stability state based on particle weights in S258 includes: The weights of all particles are normalized, and high-weight particles are retained while low-weight particles are discarded through a resampling mechanism. Based on the resampled particle set, and according to the particle weights and state vectors, the weighted average of the stress release tendency value and the fitting sensitivity tendency value is calculated respectively, to obtain the edge stability state after state deduction.

[0008] Preferably, the method in S3 includes: S31. Define the characteristic parameters of the cutting stage and the characteristic parameters of the edge banding stage as cutting nodes and edge banding nodes, respectively. S32. Establish causal edges between the cutting node and the edge sealing node and the result node respectively, and connect the cutting node and the edge sealing node in sequence to obtain the candidate manufacturing path; S33. By selecting different cutting nodes and edge sealing nodes, enumerate different candidate manufacturing paths to obtain a set of candidate manufacturing paths; S34. Randomize the path simulation order in the candidate manufacturing path set, and assign the edge manufacturing objects to the corresponding cutting nodes and edge sealing nodes of the candidate manufacturing paths in sequence according to the path simulation order for simulation. Obtain the simulated stress release tendency value and the simulated bonding sensitivity tendency value, and construct the simulated edge stable state.

[0009] Preferably, the method in S4 includes: S41. Obtain door panel edge samples that have been tested in historical production records, calculate historical stress release tendency value and historical bonding sensitivity tendency value, and obtain historical edge stability state. S42. By performing state difference analysis on candidate manufacturing paths, a selection list of optimal manufacturing paths is generated; the selection list includes a first selection list, a second selection list, and a path simulation list.

[0010] Preferably, the method in S42 includes: S421. For each simulated edge stability state, calculate the absolute difference between the simulated stress release tendency value and the historical stress release tendency value, and the absolute difference between the simulated fit sensitivity tendency value and the historical fit sensitivity tendency value, to obtain the stress state difference value and the fit state difference value respectively. S422, Accumulate stress state difference value and bonding state difference value, and record the accumulated result as the state difference index of the corresponding candidate manufacturing path; S423. For edge manufacturing objects, determine whether the simulated stress release tendency value is greater than the historical stress release tendency value; if the simulated stress release tendency value is greater than the historical stress release tendency value, then execute steps S424 and S425; if the simulated stress release tendency value is not greater than the historical stress release tendency value, then execute step S426. S424. Sort the candidate manufacturing paths corresponding to the stable states of different simulated edge lines in descending order of state difference index to obtain the first selection list of the optimal manufacturing path. S425. Select the candidate manufacturing paths as target paths in the first selection list in order, and then execute S429. S426. Determine whether the simulated fit sensitivity tendency value is greater than the historical fit sensitivity tendency value. If the simulated fit sensitivity tendency value is greater than the historical fit sensitivity tendency value, then execute S427; if the simulated fit sensitivity tendency value is not greater than the historical fit sensitivity tendency value, then execute S428. S427. Sort the values ​​of stress state difference and bonding state difference in descending order to obtain the second selection list, and then execute S429. S428. Generate a path simulation list according to the path simulation order, mark the candidate manufacturing paths as target paths in the order of the path simulation list, and execute S429.

[0011] Preferably, the method in S429 includes: Based on the target path, read the cutter and edge banding machine identifiers corresponding to the application cutting node and edge banding node respectively, and read the equipment availability status of the corresponding cutter and edge banding machine according to the cutter and edge banding machine identifiers and make status judgments. If both the cutting machine and the edge banding machine are available, the target path is determined to be available and is taken as the optimal manufacturing path. If either the cutting machine or the edge banding machine is in an unavailable state, the target path is determined to be unavailable. The process continues to traverse the next target path and check the equipment status until the target path becomes available.

[0012] Preferably, the method in S5 includes: S51. When the test result of the edge-sealed object is better than the predicted stable result, reduce the stress state difference value and bonding state difference value of the corresponding cutting node and edge-sealing node in the simulation respectively. S52. When the test result of the edge-sealed object is worse than the predicted stable result, increase the stress state difference value and bonding state difference value of the corresponding cutting node and edge-sealing node in the simulation respectively. S53. When the same type of edge manufacturing object continuously exhibits anomalies at the same cutting node and edge sealing node, the corresponding optimal manufacturing path will be marked as rejected. S54. Based on the adjustment results of the stress state difference value and the bonding state difference value of the cutting node and the sealing node respectively, S42 is re-executed for each edge manufacturing object to obtain the optimal manufacturing path after rearrangement.

[0013] This invention also provides a furniture door panel production system based on intelligent manufacturing, applied to the aforementioned furniture door panel production method based on intelligent manufacturing, comprising: The manufacturing object construction module obtains the furniture door panel and the original panel to be produced, identifies the source area and edge banding category of each edge line to be sealed in the door panel in the original panel, and constructs the edge line manufacturing object. The steady state extraction module extracts the cutting process characteristics, edge sealing process characteristics, and quality inspection results of the edge manufacturing object. By analyzing the mapping relationship between the cutting process characteristics, edge sealing process characteristics, and quality inspection results, the steady state of the edge is obtained. The manufacturing path simulation module constructs a set of candidate manufacturing paths by defining and connecting nodes of the characteristics of the cutting process and the edge sealing process. By simulating the allocation of candidate manufacturing paths, the stable state of the simulated edge line is obtained. The manufacturing path decision module determines the optimal manufacturing path by analyzing the correlation between the historical deviation characteristics of the simulated edge stability state and the equipment occupancy state. The path prediction feedback module predicts the optimal manufacturing path based on the edge manufacturing object, and detects the stability of the door panel edge after the edge sealing is completed. By comparing the prediction results with the detection results, the optimal manufacturing path is rearranged.

[0014] The technical effects and advantages of this invention regarding the production method and system for furniture door panels based on intelligent manufacturing are as follows: (1) The invention extracts the characteristics of the cutting process, the edge sealing process and the quality inspection results of the edge manufacturing object, and obtains the edge stability state through the mapping relationship, establishes the edge stability state initialization model, and combines the edge sealing stage characteristics such as the pressing load sequence and the feed speed sequence during the edge sealing process. Through particle samples, state space transfer model, particle weight and resampling mechanism, the stress release tendency value and bonding sensitivity tendency value are dynamically deduced. It solves the shortcomings of the existing methods in lacking parameter traceability and dynamic analysis between cutting characteristics, edge sealing characteristics and edge quality performance, so that the implicit differences of the substrate can be transformed into a predictable, comparable and dynamic state that can be used for subsequent path selection.

[0015] (2) This invention constructs a set of candidate manufacturing paths by defining and connecting the characteristics of the cutting process and the edge sealing process, and obtains the simulated edge line stability state through allocation simulation; further, it combines the simulated edge line stability state, historical deviation characteristics and equipment occupancy status to make a decision on the candidate manufacturing paths and obtain the optimal manufacturing path; it solves the shortcomings of the existing method that even if the latent state of the substrate is discovered, it is difficult to enter the production path decision and subsequent batch adaptive adjustment, and reduces the probability of delayed quality problems such as edge sealing bulging, shrinkage, loose bonding or uneven edge lines recurring in the same type of edge line. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall process of the furniture door panel production method based on intelligent manufacturing of the present invention.

[0017] Figure 2 This is a schematic diagram of the process flow for step S2 in the furniture door panel production method based on intelligent manufacturing of the present invention.

[0018] Figure 3 This is a schematic diagram of the modular structure of the furniture door panel production system based on intelligent manufacturing according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] This application provides a furniture door panel production method based on intelligent manufacturing. This invention achieves edge-based expression, dynamic deduction, and closed-loop optimization of the manufacturing path for implicit differences in the substrate.

[0021] Example 1, please refer to Figure 1 and Figure 2 In Embodiment 1 of the present invention, the furniture door panel production method based on intelligent manufacturing is implemented in detail through the following steps: S1. Obtain the furniture door panel to be produced and construct the edge manufacturing object for the edge banding to be completed; The methods for constructing edge creation objects to be sealed in S1 include: S11. Match the door panel cutting outline with the original panel layout drawing, and identify the source area of ​​each edge line to be sealed in the original panel. In one embodiment of the present invention, based on the original board layout data, the door panel cutting outline data, and the distance data of the source area relative to the original board boundary, the source area is divided into the original board edge area, the original board middle cutting area, and the original board transition area; wherein, the source area that is less than a first preset threshold from the original board boundary is divided into the original board edge area, the source area that is greater than a second preset threshold from the original board boundary is divided into the original board middle cutting area, and the source area that is between the first preset threshold and the second preset threshold from the original board boundary is divided into one of the original board transition areas; The specific methods for obtaining the first and second preset thresholds include: firstly, acquiring information on the source areas corresponding to the door panel edges and the edge banding stability test results from multiple historical production batches, and calculating the distance value from each source area to the nearest boundary of the original panel; statistically analyzing the distance values ​​in segments according to preset distance intervals, and calculating the edge banding stability index of the door panel edges within each distance interval, including at least one or more of the following: edge bulging rate, edge shrinkage rate, local bonding loosening rate, and edge unevenness rate; constructing a curve showing the relationship between the distance value and the edge stability index based on the edge banding stability index corresponding to each distance interval, and... Trend analysis is performed on the change relationship curve to identify the fluctuation characteristics of the edge stability index as a function of distance. When the change amplitude of the edge stability index relative to the adjacent distance interval exceeds the preset fluctuation ratio during the increase of distance value, the corresponding distance value is determined as the position where the edge stability index begins to fluctuate significantly, and this distance value is determined as the first preset threshold. The change relationship curve is analyzed along the direction of increasing distance. When the change amplitude of the edge stability index in multiple consecutive distance intervals is less than the preset stable range, the corresponding distance value is determined as the position where the edge stability index tends to stabilize, and this distance value is determined as the second preset threshold.

[0022] S12. Extract the edge banding type corresponding to the edge line to obtain the edge banding category identifier; S13. Combine the source area and edge banding category identifier into the edge manufacturing object corresponding to the edge line.

[0023] S2. Extract the characteristics of the cutting process, the edge sealing process, and the quality inspection results of the edge manufacturing object. By analyzing the mapping relationship between the characteristics of the cutting process, the edge sealing process, and the quality inspection results, obtain the stable state of the edge. The methods in S2 include: S21. Obtain the feature parameters of the historical cutting stage and construct a training dataset of historical edge samples. In one embodiment of the present invention, the feature parameters of the cutting stage are obtained by collecting the source area of ​​the original board, board density, moisture content, texture direction, cutting direction, edge length, ambient temperature and ambient humidity after the cutting, edge sealing and quality inspection are completed; and a dataset containing no less than 1,000 historical edge samples is obtained in three consecutive production batches, of which 80% of the samples are used as training set and 20% of the samples are used as independent test set. Historical data on the origin of original plates includes: S22. Obtain the quality detection result dataset corresponding to the historical edge sample dataset, construct the quality state matrix, use the principal component analysis algorithm to reduce the dimensionality of the quality state matrix, extract the principal state factors, and generate supervised learning label vectors. In one embodiment of the present invention, the process of constructing the quality state matrix includes: Quality inspection data were collected from no fewer than 1000 edge line samples across three consecutive production batches. After the door panels were sealed, quality inspection data for each edge line was collected using an edge line visual inspection device and a laser measurement device. This included five quality inspection indicators: edge line bulge height, edge line retraction, bonding gap, edge line flatness deviation, and edge warping. After standardization, a 5×5 covariance matrix was obtained. Eigenvalue decomposition of this covariance matrix yielded eigenvalues ​​λ1=2.68, λ2=1.12, and λ... λ3=0.66, λ4=0.38, and λ5=0.16, with corresponding variance contribution rates of 53.6%, 22.4%, 13.2%, 7.6%, and 3.2%, respectively; among them, the first eigenvector is [0.58, 0.49, 0.52, 0.31, 0.24], the second eigenvector is [-0.21, 0.63, -0.18, 0.69, 0.22], and the third eigenvector is [0.34, -0.27, 0.18, 0.41, 0.78]. Multiple candidate state factors are constructed in descending order of eigenvalues. Each candidate state factor is formed by a linear combination of the original features represented by the corresponding eigenvector. The eigenvalues ​​are used to characterize the variance of the data explained by the candidate components. For example, the first candidate state factor: PC1 = 0.58X1 + 0.49X2 + 0.52X3 + 0.31X4 + 0.24X5; Second candidate state factor: PC2 = -0.21X1 + 0.63X2 - 0.18X3 + 0.69X4 + 0.22X5; Third candidate state factor: PC3 = 0.34X1 - 0.27X2 + 0.18X3 + 0.41X4 + 0.78X5; X1 to X5 correspond to the original features such as edge bulge height, edge retraction amount, bonding gap, edge flatness deviation and edge warping amount, respectively. Calculate the cumulative variance contribution rate of several candidate state factors, and mark the candidate components whose cumulative variance contribution rate exceeds the preset contribution rate threshold as main state factors; for example, the cumulative variance contribution rate of the first to third candidate state factors is 89.2%, which exceeds the preset contribution rate threshold of 85%, therefore, the first to third candidate state factors are marked as main state factors. The principal components are sorted in descending order of variance contribution rate to obtain a list of principal state factors. For example, in this embodiment, the list of principal state factors is (first candidate state factor, second candidate state factor, third candidate state factor). The data types corresponding to the two largest feature vector dimension values ​​in each principal state factor are selected as the factor constituent data types, and the normalized values ​​of the feature vector dimension values ​​are used as weight values ​​for weighted fusion to generate stable state dimension features. For example, for the first candidate state factor, its two largest feature vector dimension values ​​(positive values) are 0.58 and 0.49, corresponding to the edge bulge height and edge shrinkage amount, respectively. These two data types are used as factor constituent data types, and 0.58 and 0.49 are normalized to generate weight values ​​and then weighted fusion to generate stable state dimension features, which are proposed as fitting instability features. The supervised learning label vector is obtained by sorting the variance contribution rates and concatenating the steady-state dimension features of each master state factor.

[0024] S23. Use the historical edge sample training dataset and supervised learning label vectors to train and obtain the edge stable state initialization model; Among them, the parameters of each sample in the historical edge sample training dataset are normalized by Z-score standardization to form a feature input matrix. The supervised learning label vector is used as the target output. The gradient boosting decision tree algorithm is used to establish the initialization model of the feature parameters of the cutting stage to the edge stable state. In one specific embodiment, the initial number of decision trees is set to 100, the maximum tree depth is set to 4, the learning rate is set to 0.1, the subsampling ratio is set to 0.8, and the squared error loss function is used as the optimization objective. Based on this, a parameter search space is constructed, where the number of decision trees ranges from 100 to 500, the maximum tree depth ranges from 4 to 10, the learning rate ranges from 0.01 to 0.2, and the minimum number of leaf node samples ranges from 20 to 100. Parameter optimization is performed using a combination of five-fold cross-validation and grid search. For each set of candidate parameters, training and validation are performed in turn on five subsets of the training set. Each training round uses an early stopping mechanism; training is terminated early when the loss on the validation set decreases by less than 0.001 for 20 consecutive iterations. Finally, the optimal parameter combination is obtained. After training, for each candidate model obtained by grid search, the coefficient of determination of the candidate model on the independent test set is calculated and denoted as the test coefficient of determination. The coefficient of determination is obtained by comparing the sum of squared residuals of the model prediction with the sum of squared total deviations of the target variable. It is used to characterize the model's ability to explain the changes in the target variable. The closer the value is to 1, the better the model fit. Historical edge samples from the independent test set are input into the candidate model to obtain prediction results. The Pearson correlation coefficient between the prediction results and the quality inspection of the supervised learning label vector is calculated and denoted as the test correlation coefficient. The test determination coefficient and the test correlation coefficient are weighted and fused, and the weighted fusion result is denoted as the independent test evaluation index; the weights are all set to 0.5. In the training set, the mean squared error, mean absolute error, and coefficient of determination of the candidate models are calculated and used as evaluation indicators. Dimensionless normalization is applied to each evaluation indicator to obtain the corresponding normalized correlation coefficient. For the mean squared error and mean absolute error of the validation set, inverse normalization is used to make the smaller the error, the higher the score. For the coefficient of determination, forward normalization is used to make the higher the goodness of fit, the higher the score. Then, the absolute values ​​of the Pearson correlation coefficients between different evaluation indicators and independent test evaluation indicators on all candidate models are calculated as the corresponding correlation strengths. The corresponding correlation strengths of different candidate models under the same evaluation indicator are summed to obtain the total correlation strength. The total correlation strength of each evaluation indicator is then normalized to obtain the strength contribution ratio of each evaluation indicator to the generalization performance of the model. The strength contribution ratio is then determined as the fusion weight of the evaluation indicators. The evaluation metrics of each candidate model are weighted and fused based on the fusion weights, and the weighted fusion result is used as the comprehensive evaluation score of the candidate model; the candidate model with the highest comprehensive evaluation score is selected as the edge-stable state initialization model.

[0025] S24. Obtain the feature parameters of the cutting stage of the edge manufacturing object, input the feature parameters of the cutting stage into the line stability state initialization model, obtain the prediction result corresponding to the edge manufacturing object, and generate the initial edge stability state; wherein, the initial edge stability state is represented in the form of a vector.

[0026] S25. During the edge sealing process, collect the characteristic parameters of the edge sealing stage and perform state deduction on the initial edge line stability state to obtain the edge line stability state. The method in S25 includes: S251. Read the source region of the edge manufacturing object, and calculate the normalized occurrence frequency of abnormalities in the source edge of each region after edge sealing in the historical production data to obtain the regional risk coefficient. In this embodiment, the area is divided into edge area, middle area and transition area; abnormality refers to the situation of edge bulging, edge line retraction, local loosening or uneven edge line, which can be identified by human identification; S252. Obtain the angle between the edge texture direction and the original board texture direction, and correct the area risk coefficient based on the angle to obtain the stress release tendency value. In one embodiment of the present invention, the specific method of correction is as follows: when the included angle is within the range of 0° to 15°, the regional risk coefficient remains unchanged; when the included angle is within the range of 15° to 45°, a first correction coefficient is added to the regional risk level; when the included angle is within the range of 45° to 90°, a second correction coefficient is added to the regional risk level; wherein, the correction is carried out by accumulation, the second correction coefficient is greater than the first correction coefficient, and all correction coefficients are less than 1; then the corrected regional risk level is mapped to a stress release tendency value between 0 and 1 to obtain the stress release tendency; for example, the first correction coefficient is 0.15 and the second correction coefficient is 0.30; S253. Collect the pressing load sequence when the edge line passes through the pressing mechanism and the feed speed sequence of the edge line in the edge sealing equipment, and calculate the bonding sensitivity tendency value of the pressing load sequence and the feed speed sequence. In one embodiment of the present invention, the pressing load sequence and the feed rate sequence are divided into sliding windows, and the coefficient of variation of the pressing load and the coefficient of variation of the rate are calculated in each sliding window, denoted as CF and CV, respectively. The coefficient of variation is obtained by the ratio of the standard deviation to the mean of the sliding window. The Pearson correlation coefficient FV of the pressing load sequence and the feed rate sequence in the sliding window is further calculated to characterize the degree of synchronous fluctuation between the two. The bonding sensitivity index SI is calculated according to the following formula: SI=α×CF+β×CV+γ×|FV|, where α, β and γ are preset weight coefficients, and α+β+γ=1. Then, the bonding sensitivity index SI is mapped to the interval of 0 to 1 using a normalization function to obtain the bonding sensitivity tendency value PT. S254. Obtain historical production samples of the same edge manufacturing object, execute S251 to S253 for each historical production sample and perform range statistics respectively to obtain the initial value range of the corresponding stress release tendency value and the fit sensitivity tendency value respectively. S255. Within the initial value range, multiple particle samples are randomly generated according to a preset probability distribution, and a state vector of each particle sample is constructed. Each particle sample contains two state variables: stress release tendency and fit sensitivity tendency. S256. Obtain the characteristic parameters of the edge sealing stage, construct the edge sealing process input vector, and construct the state space transition model based on the spatial mapping relationship between the state vector and the edge sealing process input vector. In one embodiment of the present invention, the state vector is first defined as follows: Where σ(k) represents the stress release tendency value at time k, and η(k) represents the bonding sensitivity tendency value at time k; based on historical production data, the state change patterns under different source regions, different cutting directions, and different edge sealing stage characteristic parameters are statistically analyzed, and a state transition equation is used for fitting. The expression of the state transition equation is: ; Where A is the state transition matrix, which describes the natural evolution of stress release tendency value and bonding sensitivity tendency value with the processing process; B is the control input matrix; u(k) is the edge sealing process input vector; and w(k) is the process noise. The system collects real-time data on the source area of ​​the original board, the cutting direction, and process parameters generated during the operation of the edge banding equipment, such as pressure roller pressure, feed speed, dwell time, and pressure roller pressure change rate, and constructs a process input vector. Where P(k) represents the pressure roller pressure, V(k) represents the feed rate, T(k) represents the residence time, and ΔP(k) represents the pressure roller pressure change rate. Further establish observation vectors Observation equations between the state vector and the state vector , where H is the observation matrix, used to describe the mapping relationship between the hidden state of the particle and the actual acquisition process parameters, and v(k) is the observation noise; The state transition matrix A, control input matrix B, and observation matrix H are identified and solved using historical sample data. Specifically, N consecutive sets of historical production samples are selected to construct the state matrix. Precursor state matrix and process input matrix ; The state transition equation is rewritten as follows: ,in, The parameter matrix to be identified is used; the least squares estimation method is employed to minimize the objective function. For the objective, the parameter matrix Solve the problem to obtain the parameter estimation results. Further utilize the observed samples to construct an observation matrix identification model. ,in, The observation matrix parameters are obtained through least squares estimation. For example, in a batch of historical production data containing 500 door panels and a total of 2,000 edge banding lines, the pressure of the pressure roller, the feed speed, the dwell time, and the test results after cooling during the edge banding process of each edge banding line are extracted as training samples. The parameters of A, B, and H are calibrated according to the least squares estimation process described above, and the calibration results are used as the initial parameters of the subsequent online state prediction model, thereby establishing a state-space transition model that can reflect the dynamic changes in stress release tendency and bonding sensitivity tendency during the edge banding process.

[0027] S257. Calculate the probability distribution of each particle's evolution from the previous state to the current state using the state-space transition model. S258. Obtain the observations for each particle, calculate the particle weights based on the matching degree and probability distribution between the observations and the predicted state, and calculate the edge stable state based on the particle weights. In one embodiment of the present invention, the absolute errors of the observed values ​​and the predicted values ​​of different particles under different conditions in terms of stress release tendency and fit sensitivity tendency are calculated. The absolute errors of the same particle in terms of stress release tendency and fit sensitivity tendency are accumulated, and a weight normalization method based on the reciprocal of the absolute error coefficient is used to process the results to obtain the initial score value of each particle. The initial score value is multiplied by the probability value, and the multiplication result is normalized to obtain the weight of the particle. The method for calculating the edge stability state based on particle weights in S258 includes: The weights of all particles are normalized, and high-weight particles are retained while low-weight particles are discarded through a resampling mechanism. In one embodiment of the present invention, for example, suppose there are currently 5 particles E1 to E5, with corresponding errors of 0.1, 0.2, 0.5, 1.0 and 2.0 respectively. Then the weights calculated based on the reciprocal of the error are approximately 0.53, 0.27, 0.11, 0.05 and 0.04. When performing 10 resampling operations, E1 and E2 may be retained 5 times and 3 times respectively due to their higher weights, E3 may be retained once, E4 may be retained once, while E5 may be completely eliminated due to its lowest weight. This allows the new particle set to be more concentrated in the state region that matches the actual observation results, thereby realizing the joint estimation and dynamic update of stress release tendency value and fit sensitivity tendency value.

[0028] Based on the resampled particle set, the weighted average of the stress release tendency value and the fitting sensitivity tendency value is calculated according to the particle weight and state vector, respectively, to obtain the edge stability state after state deduction; whereby the edge stability state is represented in the form of a vector.

[0029] In existing production methods, the continuous changes in the substrate state during the cutting, edge sealing, and subsequent stabilization processes make parameter tracing and dynamic analysis difficult. This invention extracts the cutting process characteristics, edge sealing process characteristics, and quality inspection results of the edge manufacturing object in S2, and obtains the edge stability state through mapping relationships. In a further embodiment, the scheme utilizes principal component analysis to extract quality state factors, establishes an edge stability state initialization model using supervised learning, and combines edge sealing stage characteristics such as pressing load sequences and feed speed sequences during the edge sealing process. Through particle sampling, state space transition models, particle weights, and resampling mechanisms, it dynamically extrapolates stress release tendency values ​​and bonding sensitivity tendency values. Through the above steps, the shortcomings of existing methods in lacking parameter traceability and dynamic analysis between cutting characteristics, edge sealing characteristics and edge quality performance are solved, so that the latent differences of the substrate can be transformed into a predictable, comparable and dynamic state that can be used for subsequent path selection.

[0030] S3. Construct a set of candidate manufacturing paths based on the characteristic parameters of the cutting stage and the edge sealing stage. By simulating the allocation of candidate manufacturing paths, obtain the stable state of the simulated edge line. The methods in S3 include: S31. Define the characteristic parameters of the cutting stage and the characteristic parameters of the edge banding stage as cutting nodes and edge banding nodes, respectively. S32. Establish causal edges between the cutting node and the edge sealing node and the result node respectively, and connect the cutting node and the edge sealing node in sequence to obtain the candidate manufacturing path; S33. By selecting different cutting nodes and edge sealing nodes, enumerate different candidate manufacturing paths to obtain a set of candidate manufacturing paths; S34. Randomize the path simulation order in the candidate manufacturing path set, and assign the edge manufacturing objects to the corresponding cutting nodes and edge sealing nodes of the candidate manufacturing paths in sequence according to the path simulation order for simulation. Obtain the simulated stress release tendency value and the simulated bonding sensitivity tendency value, and construct the simulated edge stability state; wherein, the simulated edge stability state is represented in the form of a vector.

[0031] S4. By simulating the correlation between the stable state of the edge line in terms of historical deviation characteristics and the equipment occupancy status, a decision is made on the candidate manufacturing path to obtain the optimal manufacturing path. The methods in S4 include: S41. Obtain door panel edge samples that have been tested in historical production records, calculate historical stress release tendency value and historical bonding sensitivity tendency value, and obtain historical edge stability state. S42. By performing state difference analysis on candidate manufacturing paths, a selection list of optimal manufacturing paths is generated; the selection list includes a first selection list, a second selection list, and a path simulation list. The methods in S42 include: S421. For each simulated edge stability state, calculate the absolute difference between the simulated stress release tendency value and the historical stress release tendency value, and the absolute difference between the simulated fit sensitivity tendency value and the historical fit sensitivity tendency value, to obtain the stress state difference value and the fit state difference value respectively. S422, Accumulate stress state difference value and bonding state difference value, and record the accumulated result as the state difference index of the corresponding candidate manufacturing path; S423. For edge manufacturing objects, determine whether the simulated stress release tendency value is greater than the historical stress release tendency value; if the simulated stress release tendency value is greater than the historical stress release tendency value, then execute steps S424 and S425; if the simulated stress release tendency value is not greater than the historical stress release tendency value, then execute step S426. S424. Sort the candidate manufacturing paths corresponding to the stable states of different simulated edge lines in descending order of state difference index to obtain the first selection list of the optimal manufacturing path. The significance of S424 lies in the fact that edge manufacturing objects with high stress release tendency values ​​are more prone to problems such as edge deformation, edge shrinkage, or decreased bonding stability due to the release of internal residual stress during the cutting and subsequent edge sealing processes. By prioritizing the combination of cutting nodes and edge sealing nodes with state difference indices, the risk of stress accumulation during the processing can be reduced by utilizing proven stable processing paths.

[0032] S425. Select the candidate manufacturing paths as target paths in the first selection list in order, and then execute S429. S426. Determine whether the simulated fit sensitivity tendency value is greater than the historical fit sensitivity tendency value. If the simulated fit sensitivity tendency value is greater than the historical fit sensitivity tendency value, then execute S427; if the simulated fit sensitivity tendency value is not greater than the historical fit sensitivity tendency value, then execute S428. S427. Sort the values ​​of stress state difference and bonding state difference in descending order to obtain the second selection list, and then execute S429. In one embodiment of the present invention, the specific method for priority selection includes: performing inverse normalization weight calculation on the stress state difference value and bonding state difference value of the candidate manufacturing paths respectively; recording the calculation results as stress difference priority coefficient and bonding difference priority coefficient respectively; weighting and accumulating them with weights of 0.6 and 0.4 to obtain a comprehensive priority coefficient; and sorting the candidate manufacturing paths in descending order according to the comprehensive priority coefficient to obtain a second selection list.

[0033] The significance of S427 lies in the fact that edge-line manufacturing objects with high edge-fit sensitivity are more sensitive to edge-line processing quality, edge integrity, and subsequent edge-sealing process conditions, and are prone to problems such as poor local bonding, uneven edges, or insufficient edge-sealing strength. By prioritizing the selection of cutting nodes with lower stress state difference values, the edge-line processing accuracy and edge quality can be improved. Further assigning them to edge-sealing nodes with lower bonding state difference values ​​helps to ensure the stability of adhesive layer formation, pressing process, and edge tape attachment process, thereby improving edge-fit quality and reducing the probability of edge-sealing defects.

[0034] S428. Generate a path simulation list according to the path simulation order, mark the candidate manufacturing paths as target paths in the order of the path simulation list, and execute S429. S429. Based on the target path, read the cutter identifier and edge banding machine identifier corresponding to the application cutting node and edge banding node respectively, and read the equipment availability status of the corresponding cutter and edge banding machine according to the cutter identifier and edge banding machine identifier and perform status judgment. If both the cutting machine and the edge banding machine are available, the target path is determined to be available and is taken as the optimal manufacturing path. If either the cutting machine or the edge banding machine is in an unavailable state, the target path is determined to be unavailable. The process continues to traverse the next target path and check the equipment status until the target path becomes available.

[0035] S5. Predict the optimal manufacturing path based on the edge manufacturing object, and after the edge sealing is completed, test the stability of the door panel edge. By comparing the prediction results and the test results, rearrange the optimal manufacturing path.

[0036] The methods in S5 include: S51. When the test result of the edge-sealed object is better than the predicted stable result, reduce the stress state difference value and bonding state difference value of the corresponding cutting node and edge-sealing node in the simulation respectively. S52. When the test result of the edge-sealed object is worse than the predicted stable result, increase the stress state difference value and bonding state difference value of the corresponding cutting node and edge-sealing node in the simulation respectively. S53. When the same type of edge manufacturing object continuously exhibits anomalies at the same cutting node and edge sealing node, the corresponding optimal manufacturing path will be marked as rejected. In an embodiment of the present invention, for example, if five consecutive door panels processed from the same cutting node and edge sealing node all show edge bulging or local loosening of the fit, the corresponding optimal manufacturing path will be eliminated; in addition, both reduction and improvement will be adjusted by 10% of the original difference value.

[0037] S54. Based on the adjustment results of the stress state difference value and the bonding state difference value of the cutting node and the sealing node respectively, S42 is re-executed for each edge manufacturing object to obtain the optimal manufacturing path after rearrangement.

[0038] Furthermore, even if existing methods can identify quality fluctuations in a certain type of door panel or edge, they often struggle to address the issue of collaborative selection based on cutting equipment, edge banding equipment, and production paths. In S3, this invention constructs a candidate manufacturing path set by defining and connecting nodes based on the characteristics of the cutting and edge banding processes, and obtains a simulated edge stability state through allocation simulation. In S4, it further combines the simulated edge stability state, historical deviation characteristics, and equipment occupancy status to make a decision on the candidate manufacturing paths, obtaining the optimal manufacturing path. In S5, it adjusts or rearranges the optimal manufacturing path by comparing the predicted results after edge banding with the detection results. The above methods address the shortcomings of existing methods where, even if the latent state of the substrate is discovered, it is difficult to incorporate into production path decisions and subsequent batch adaptive adjustments. This enables the latent material state to truly participate in intelligent manufacturing scheduling, thereby reducing the probability of delayed quality problems such as edge bulging, shrinkage, loose bonding, or uneven edges recurring on similar edge lines.

[0039] Example 2, please refer to Figure 3 In Embodiment 2 of the present invention, the furniture door panel production system based on intelligent manufacturing is implemented in detail through the following modules: The manufacturing object construction module obtains the furniture door panel and the original panel to be produced, identifies the source area and edge banding category of each edge line to be sealed in the door panel in the original panel, and constructs the edge line manufacturing object. The steady state extraction module extracts the cutting process characteristics, edge sealing process characteristics, and quality inspection results of the edge manufacturing object. By analyzing the mapping relationship between the cutting process characteristics, edge sealing process characteristics, and quality inspection results, the steady state of the edge is obtained. The manufacturing path simulation module constructs a set of candidate manufacturing paths by defining and connecting nodes of the characteristics of the cutting process and the edge sealing process. By simulating the allocation of candidate manufacturing paths, the stable state of the simulated edge line is obtained. The manufacturing path decision module determines the optimal manufacturing path by simulating the correlation between the historical deviation characteristics of the edge stability state and the equipment occupancy state. The path prediction feedback module predicts the optimal manufacturing path based on the edge manufacturing object, and detects the stability of the door panel edge after the edge sealing is completed. By comparing the prediction results with the detection results, the optimal manufacturing path is rearranged.

[0040] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0041] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0042] All formulas in this manual are dimensionless and calculated numerically. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0043] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A furniture door panel production method based on intelligent manufacturing, characterized in that, include: S1. Obtain the furniture door panel and the original panel to be produced, identify the source area and edge banding category of each edge banding line to be sealed in the original panel, and construct the edge manufacturing object; S2. Extract the characteristics of the cutting process, the edge sealing process, and the quality inspection results of the edge manufacturing object. By analyzing the mapping relationship between the characteristics of the cutting process, the edge sealing process, and the quality inspection results, obtain the stable state of the edge. S3. By defining and connecting the nodes of the cutting process characteristics and the edge sealing process characteristics, a set of candidate manufacturing paths is constructed. By simulating the allocation of candidate manufacturing paths, the stable state of the simulated edge line is obtained. S4. By analyzing the correlation between the historical deviation characteristics of the simulated edge stability state and the equipment occupancy state, a decision is made on the candidate manufacturing path to obtain the optimal manufacturing path. S5. Predict the optimal manufacturing path based on the edge manufacturing object, and after the edge sealing is completed, test the stability of the door panel edge. By comparing the prediction results and the test results, rearrange the optimal manufacturing path.

2. The furniture door panel production method based on intelligent manufacturing according to claim 1, characterized in that, The method in S2 includes: S21. Obtain the feature parameters of the historical cutting stage and construct a training dataset of historical edge samples. S22. Obtain the quality detection result dataset corresponding to the historical edge sample dataset, construct the quality state matrix, use the principal component analysis algorithm to reduce the dimensionality of the quality state matrix, extract the principal state factors, and generate supervised learning label vectors. S23. Use the historical edge sample training dataset and supervised learning label vectors to train and obtain the edge stable state initialization model; S24. Obtain the feature parameters of the cutting stage of the edge manufacturing object, input the feature parameters of the cutting stage into the line stability initialization model, obtain the prediction result corresponding to the edge manufacturing object, and generate the initial edge stability state. S25. During the edge sealing process, collect the characteristic parameters of the edge sealing stage and perform state deduction on the initial edge line stability state to obtain the edge line stability state.

3. The furniture door panel production method based on intelligent manufacturing according to claim 2, characterized in that, The method in S25 includes: S251. Read the source region of the edge manufacturing object, and calculate the normalized occurrence frequency of abnormalities in the source edge of each region after edge sealing in the historical production data to obtain the regional risk coefficient. S252. Obtain the angle between the edge texture direction and the original board texture direction, and correct the area risk coefficient based on the angle to obtain the stress release tendency value. S253. Collect the pressing load sequence when the edge line passes through the pressing mechanism and the feed speed sequence of the edge line in the edge sealing equipment, and calculate the bonding sensitivity tendency value of the pressing load sequence and the feed speed sequence. S254. Obtain historical production samples of the same edge manufacturing object, execute S251 to S253 for each historical production sample and perform range statistics respectively to obtain the initial value range of the corresponding stress release tendency value and the fit sensitivity tendency value respectively. S255. Within the initial value range, randomly sample and generate multiple particle samples according to a preset probability distribution, and construct the state vector of each particle sample. S256. Obtain the characteristic parameters of the edge sealing stage, construct the edge sealing process input vector, and construct the state space transition model based on the spatial mapping relationship between the state vector and the edge sealing process input vector. S257. Calculate the probability distribution of each particle's evolution from the previous state to the current state using the state-space transition model. S258. Obtain the observations for each particle, calculate the particle weights based on the degree of matching and probability distribution between the observations and the predicted state, and calculate the edge stable state based on the particle weights.

4. The furniture door panel production method based on intelligent manufacturing according to claim 3, characterized in that, The method for calculating the edge stability state based on particle weights in S258 includes: The weights of all particles are normalized, and high-weight particles are retained while low-weight particles are discarded through a resampling mechanism. Based on the resampled particle set, and according to the particle weights and state vectors, the weighted average of the stress release tendency value and the fitting sensitivity tendency value is calculated respectively, to obtain the edge stability state after state deduction.

5. The furniture door panel production method based on intelligent manufacturing according to claim 4, characterized in that, The method in S3 includes: S31. Define the characteristic parameters of the cutting stage and the characteristic parameters of the edge banding stage as cutting nodes and edge banding nodes, respectively. S32. Establish causal edges between the cutting node and the edge sealing node and the result node respectively, and connect the cutting node and the edge sealing node in sequence to obtain the candidate manufacturing path; S33. By selecting different cutting nodes and edge sealing nodes, enumerate different candidate manufacturing paths to obtain a set of candidate manufacturing paths; S34. Randomize the path simulation order in the candidate manufacturing path set, and assign the edge manufacturing objects to the corresponding cutting nodes and edge sealing nodes of the candidate manufacturing paths in sequence according to the path simulation order for simulation. Obtain the simulated stress release tendency value and the simulated bonding sensitivity tendency value, and construct the simulated edge stable state.

6. The furniture door panel production method based on intelligent manufacturing according to claim 5, characterized in that, The method in S4 includes: S41. Obtain door panel edge samples that have been tested in historical production records, calculate historical stress release tendency value and historical bonding sensitivity tendency value, and obtain historical edge stability state. S42. By performing state difference analysis on candidate manufacturing paths, a selection list of optimal manufacturing paths is generated; the selection list includes a first selection list, a second selection list, and a path simulation list.

7. The furniture door panel production method based on intelligent manufacturing according to claim 6, characterized in that, The method in S42 includes: S421. For each simulated edge stability state, calculate the absolute difference between the simulated stress release tendency value and the historical stress release tendency value, and the absolute difference between the simulated fit sensitivity tendency value and the historical fit sensitivity tendency value, to obtain the stress state difference value and the fit state difference value respectively. S422, Accumulate stress state difference value and bonding state difference value, and record the accumulated result as the state difference index of the corresponding candidate manufacturing path; S423. For edge manufacturing objects, determine whether the simulated stress release tendency value is greater than the historical stress release tendency value; if the simulated stress release tendency value is greater than the historical stress release tendency value, then execute steps S424 and S425; if the simulated stress release tendency value is not greater than the historical stress release tendency value, then execute step S426. S424. Sort the candidate manufacturing paths corresponding to the stable states of different simulated edge lines in descending order of state difference index to obtain the first selection list of the optimal manufacturing path. S425. Select the candidate manufacturing paths as target paths in the first selection list in order, and then execute S429. S426. Determine whether the simulated fit sensitivity tendency value is greater than the historical fit sensitivity tendency value. If the simulated fit sensitivity tendency value is greater than the historical fit sensitivity tendency value, then execute S427; if the simulated fit sensitivity tendency value is not greater than the historical fit sensitivity tendency value, then execute S428. S427. Sort the values ​​of stress state difference and bonding state difference in descending order to obtain the second selection list, and then execute S429. S428. Generate a path simulation list according to the path simulation order, mark the candidate manufacturing paths as target paths in the order of the path simulation list, and execute S429.

8. The furniture door panel production method based on intelligent manufacturing according to claim 7, characterized in that, The method in S429 includes: Based on the target path, read the cutter and edge banding machine identifiers corresponding to the application cutting node and edge banding node respectively, and read the equipment availability status of the corresponding cutter and edge banding machine according to the cutter and edge banding machine identifiers and make status judgments. If both the cutting machine and the edge banding machine are available, the target path is determined to be available and is taken as the optimal manufacturing path. If either the cutting machine or the edge banding machine is in an unavailable state, the target path is determined to be unavailable. The process continues to traverse the next target path and check the equipment status until the target path becomes available.

9. The furniture door panel production method based on intelligent manufacturing according to claim 8, characterized in that, The method in S5 includes: S51. When the test result of the edge-sealed object is better than the predicted stable result, reduce the stress state difference value and bonding state difference value of the corresponding cutting node and edge-sealing node in the simulation respectively. S52. When the test result of the edge-sealed object is worse than the predicted stable result, increase the stress state difference value and bonding state difference value of the corresponding cutting node and edge-sealing node in the simulation respectively. S53. When the same type of edge manufacturing object continuously exhibits anomalies at the same cutting node and edge sealing node, the corresponding optimal manufacturing path will be marked as rejected. S54. Based on the adjustment results of the stress state difference value and the bonding state difference value of the cutting node and the sealing node respectively, S42 is re-executed for each edge manufacturing object to obtain the optimal manufacturing path after rearrangement.

10. A furniture door panel production system based on intelligent manufacturing, applied to the furniture door panel production method based on intelligent manufacturing as described in any one of claims 1 to 9, characterized in that, include: The manufacturing object construction module obtains the furniture door panel and the original panel to be produced, identifies the source area and edge banding category of each edge line to be sealed in the door panel in the original panel, and constructs the edge line manufacturing object. The steady state extraction module extracts the cutting process characteristics, edge sealing process characteristics, and quality inspection results of the edge manufacturing object. By analyzing the mapping relationship between the cutting process characteristics, edge sealing process characteristics, and quality inspection results, the steady state of the edge is obtained. The manufacturing path simulation module constructs a set of candidate manufacturing paths by defining and connecting nodes of the characteristics of the cutting process and the edge sealing process. By simulating the allocation of candidate manufacturing paths, the stable state of the simulated edge line is obtained. The manufacturing path decision module determines the optimal manufacturing path by analyzing the correlation between the historical deviation characteristics of the simulated edge stability state and the equipment occupancy state. The path prediction feedback module predicts the optimal manufacturing path based on the edge manufacturing object, and detects the stability of the door panel edge after the edge sealing is completed. By comparing the prediction results with the detection results, the optimal manufacturing path is rearranged.

Images