A method for automatic trajectory planning of battery protection board blanking
By conducting mechanical testing and database management of battery protection plate materials and optimizing the cutting path in combination with genetic algorithms, the automated trajectory planning of battery protection plate cutting is realized, which solves the problems of low material utilization and excessive scraps, and improves processing efficiency and materials Utilization rate, in line with the concept of green manufacturing.
Patent Information
- Application Number
- CN202410985917.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-23
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-07-23
AI Technical Summary
During the process of unloading the battery protective plate, due to the differences in mechanical, thermal and electrical characteristics of different materials, it is difficult for the existing technology to formulate reasonable cutting strategies, resulting in low material utilization, excessive scraps, and low processing process efficiency and accuracy.
By conducting mechanical properties testing of composite materials and metal materials, establishing a material database, optimizing the cutting path with plastic processing heat treatment models and genetic algorithms, using multi-functional cutting equipment for automated trajectory planning, real-time monitoring and optimization of cutting processes, realizing the classification and classification of scraps and Reuse.
It improves the scientificity and processing efficiency of material selection, reduces material waste and tool wear, extends the service life of cutting tools, significantly improves material utilization, reduces production costs, and conforms to the concept of green manufacturing and circular economy.
Smart Images

Figure CN118543990B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of information technology, and in particular to an automated trajectory planning method for blanking a battery protection board. Background Art
[0002] Battery protection boards have various material properties, including metals and composite materials. The mechanical properties, thermal properties, and electrical properties of different materials vary significantly. When cutting battery protection boards, it is necessary to formulate corresponding cutting strategies for different materials, select cutting processes and cutting parameters. Battery protection boards are usually made up of multiple layers of different materials. Not only are the material properties of each layer different, but the anisotropy of the materials also poses challenges to cutting. Unreasonable cutting paths will lead to low material utilization and a large amount of scraps. It is urgent to conduct a systematic study on the material properties of battery protection boards, establish a material property database, and clarify the mechanical, thermal, and electrical property parameters of different materials, different directions, and different layers.
[0003] The present invention proposes a solution to the above-mentioned disadvantages as follows:
[0004] Analyze the material properties of the battery protection board, obtain basic data and parameters that adapt to the different material properties of metals and composite materials, formulate cutting strategies for different materials, and based on the cutting strategies, study the cutting technology for multi-layer battery protection boards, optimize the cutting path, integrate multifunctional cutting equipment that can handle a variety of materials and complex structures to execute the cutting path, formulate material utilization strategies, classify and reuse the scraps in the cutting process for the production of other parts. Summary of the invention
[0005] In order to solve the above-mentioned technical problems, the present invention provides a method for automatic trajectory planning of battery protection board blanking.
[0006] The technical solution of the present invention is achieved in this way:
[0007] The present invention provides a method for automatic trajectory planning of battery protection board blanking, which mainly includes:
[0008] The mechanical strength and thermal conductivity of the preset composite materials and metal materials are evaluated by mechanical properties testers and thermal conductivity instruments, including evaluating mechanical strength through tensile strength, yield strength, and fracture toughness, and evaluating thermal conductivity through thermal conductivity and thermal diffusivity, to obtain the performance parameters of each material, manage the test results using the material database, and classify each material according to the performance parameters;
[0009] Through the plastic processing heat treatment model, the processing difficulty of each type of material is analyzed according to the material classification results, including cutting force, tool wear rate, and processing temperature. Through the cutting strategy module, a cutting process roadmap is generated for different types of materials with different processing difficulties, and the cutting speed, angle, cutting tool type, including laser cutting machine or plasma cutting machine, as well as cooling and lubrication methods are marked;
[0010] Analyze the cutting process roadmap, combine material performance parameters and structural characteristics, including shape, size and thickness, use genetic algorithm to optimize the cutting path, generate cutting operation plans, and use taboo search module to adjust the cutting order to avoid conflicts;
[0011] Receiving a cutting operation plan through a cutting device, and adjusting cutting parameters of the cutting device according to the cutting operation plan, including a motion path of a cutting knife and a battery protection board, power output, and a feeding speed of the battery protection board;
[0012] The data acquisition module is used to connect the sensors of the cutting equipment to monitor the cutting process in real time, calculate the material utilization rate and cutting quality, and use the statistical process control module to analyze the monitoring data. The principal component analysis module is used to calculate the cutting efficiency and material loss. If the cutting efficiency is less than the preset threshold or the material loss is greater than the preset threshold, the cutting strategy module is dynamically optimized.
[0013] Identify the size and shape of the scraps generated after cutting the battery protection board, combine the identification results with the material type and the brand of the battery corresponding to the battery protection board, and record them in the material management system;
[0014] During the production process of battery parts, the scrap data associated with the corresponding battery brand is queried in the material management system, and the scraps are matched to the battery parts through the shape matching algorithm. The heuristic optimization algorithm is used to calculate the material block layout and re-cutting instructions, which are sent to the battery parts production system to execute the reuse of the matched scraps. Beneficial Effects
[0015] By comprehensively testing and classifying the mechanical strength and thermal conductivity of materials, the scientificity and accuracy of material selection can be improved, which helps to select the most suitable materials for different application scenarios;
[0016] The cutting process roadmap is customized according to the material characteristics, taking into account the difficulty of processing, tool selection and maintenance costs, improving processing efficiency and yield rate, while reducing energy consumption and tool wear during processing; the application of genetic algorithms and taboo search ensures the efficiency and conflict-free nature of the cutting path, reduces material waste, improves cutting accuracy and efficiency, and also extends the service life of the cutting tool; real-time monitoring of the cutting process and the use of statistical process control and principal component analysis can quickly identify and respond to low cutting efficiency or excessive material loss, realize instant adjustment of cutting strategies, and ensure the stability and economy of the production process; through size recognition, shape classification and integration with the material management system, refined management and efficient redistribution of scraps are achieved, which significantly improves material utilization, reduces waste, and reduces production costs; the entire process strengthens the recycling of resources, reduces resource consumption and environmental impact, and is in line with the concepts of green manufacturing and circular economy; through the fast query and matching system, it can efficiently respond to changing production needs, flexibly dispatch scraps for the production of battery parts, shorten the production cycle, and enhance the agility of the supply chain.
[0017] Working principle:
[0018] First, the mechanical strength and thermal conductivity of the preset composite materials and metal materials are evaluated by mechanical property testers and thermal conductivity instruments to obtain the performance parameters of each material and store them in the material database for management. Then, the materials are classified according to the performance parameters, and the processing difficulty of each type of material is analyzed through the plastic processing heat treatment model, including cutting force, tool wear rate and processing temperature. Then, the cutting strategy module generates a cutting process roadmap based on the material classification results, marking the cutting speed, angle, cutting tool type, and cooling and lubrication methods. On this basis, combined with the material performance parameters and structural characteristics, the genetic algorithm is used to optimize the cutting path, generate a cutting operation plan, and adjust the cutting order through the taboo search module to avoid conflicts. The cutting equipment receives the cutting operation plan and adjusts the cutting parameters according to the plan, including the motion path of the cutting knife and the battery protection board, power output, and battery protection board feed speed. At the same time, the data acquisition module connects to the sensors of the cutting equipment to monitor the cutting process in real time, calculate the material utilization rate and Cutting quality, the statistical process control module processes the monitoring data, and the principal component analysis module calculates the cutting efficiency and material loss. If the cutting efficiency is less than the preset threshold or the material loss is greater than the preset threshold, the cutting strategy module is dynamically optimized. In addition, the weight of the scraps dropped after cutting is monitored by the gravity sensor, and the material utilization rate is calculated in combination with the weight of the battery protection board before cutting. The surface image of the cutting surface is monitored by the visual sensor, and the surface features of the cutting surface, including roughness, are extracted to quantify the cutting quality. The scraps generated after cutting the battery protection board are identified in size and classified in shape, and the identification results are combined with the material type and the brand of the battery corresponding to the battery protection board, and recorded in the material management system. Finally, in the production process of battery parts, the scrap data associated with the brand of the corresponding battery in the material management system is queried, and the scraps are matched for the battery parts through the shape matching algorithm. The heuristic optimization algorithm is used to calculate the material block layout and re-cutting instructions, which are sent to the battery parts production system to execute the reuse of the matched scraps. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 It is a flow chart of a method for automatic trajectory planning of battery protection board blanking of the present invention;
[0020] Figure 2 A schematic diagram of an automated trajectory planning method for blanking a battery protection board according to the present invention;
[0021] Figure 3 It is another schematic diagram of the automatic trajectory planning method for blanking a battery protection board according to the present invention. DETAILED DESCRIPTION
[0022] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0023] The preferred implementation methods of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these implementation methods are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.
[0024] It should be noted that when an element is considered to be "connected" to another element, it can be directly connected to the other element, or connected to the other element through an intermediate element. In addition, the "connection" in the following embodiments should be understood as "electrical connection", "communication connection", etc. if there is transmission of electrical signals or data between the connected objects.
[0025] When used herein, the singular forms "a", "an", and "said / the" may also include plural forms, unless the context clearly indicates otherwise. It should also be understood that the terms "include / comprise" or "have" and the like specify the presence of stated features, wholes, steps, operations, components, parts, or combinations thereof, but do not exclude the possibility of the presence or addition of one or more other features, wholes, steps, operations, components, parts, or combinations thereof. At the same time, the term "and / or" used in this specification includes any and all combinations of the relevant listed items.
[0026] See also Figure 1-3 As shown, a method for automatic trajectory planning of battery protection board blanking includes:
[0027] Specifically, step S101 tests 10 composite materials and 10 metal materials using a mechanical properties tester to obtain the tensile strength, yield strength, fracture toughness and mechanical strength performance parameters of each material:
[0028] The tensile strength ranges from 100-1000MPa, the yield strength ranges from 50-800MPa, and the fracture toughness ranges from 10-100MPa·m1 / 2;
[0029] The 20 materials were tested using a thermal conductivity instrument, and the thermal conductivity range was 10-500W / (m·K), and the thermal diffusion coefficient range was 1-10mm2 / s;
[0030] In the material database, establish a corresponding relationship table between material types and their mechanical performance parameters and thermal performance parameters:
[0031] Materials with a tensile strength greater than 800MPa are classified as high-strength materials, those with a tensile strength less than 300MPa are classified as low-strength materials, and the rest are classified as medium-strength materials;
[0032] Materials with a thermal conductivity greater than 200W / (m·K) are classified as high thermal conductivity materials, materials with a thermal conductivity less than 50W / (m·K) are classified as low thermal conductivity materials, and the rest are classified as medium thermal conductivity materials;
[0033] The classification results recorded in the material database include 5 high-strength materials, 10 medium-strength materials, and 5 low-strength materials; 6 high thermal conductivity materials, 9 medium thermal conductivity materials, and 5 low thermal conductivity materials;
[0034] Based on the comprehensive mechanical strength and thermal conductivity, one material was marked as high-quality material, three materials were marked as ordinary materials, and the remaining 16 materials were marked as medium materials;
[0035] The statistical report on material properties shows that among the 10 composite materials, there is 1 high-quality material, 7 medium materials, and 2 ordinary materials; among the 10 metal materials, there are 9 medium materials and 1 ordinary material. According to the statistical report, the mechanical strength and thermal conductivity of composite materials are generally better than those of metal materials.
[0036] Composite materials are used in situations that require high strength and high thermal conductivity, including aerospace structural parts and electronic device heat dissipation substrates; metal materials are used in general situations where strength and thermal conductivity requirements are lower;
[0037] Finally, the performance parameters, classification results, statistical reports, and application guidance information of all materials are associated and stored to build a material performance assessment and application knowledge base, providing intelligent and systematic data support for material selection and product design.
[0038] Specifically, in step S102, when establishing the material processing performance prediction model, mechanical performance parameters and cutting processing experimental data of 100 different materials are collected, 50 sets of data are collected for each material, 80% of which are randomly selected as training sets and 20% as test sets;
[0039] For the continuous numerical indicators of hardness and strength, the support vector regression algorithm is used, the Gaussian kernel function is selected, the kernel function parameters and penalty coefficients are optimized through grid search and cross-validation, and the root mean square error and determination coefficient of the model are evaluated on the test set;
[0040] For discrete indicators of material type, the random forest algorithm is used to optimize the number of decision trees, maximum depth, and minimum number of samples for node division through grid search and cross validation. The accuracy and F1 value of the model are evaluated on the test set. The trained model is used to predict the processing performance of the new material, including predicting that the cutting force of Ti-6Al-4V titanium alloy is 800N, the tool wear rate is 0.02mm / min, and the processing temperature is 500℃. According to the prediction results, the processing difficulty is high. Then, the genetic algorithm is used to optimize the cutting process parameters of the material to cut the surface roughness and cutting The optimization target is time, and the cutting speed, feed rate and cutting depth are the optimization variables. The variable value range is set, including cutting speed 20-100m / min, feed rate 0.05-0.3mm / r, and cutting depth 0.2-1mm. Through the iteration of encoding, selection, crossover and mutation genetic operators, the Pareto optimal solution set is obtained within 100 generations. The optimal parameter combination of cutting speed 60m / min, feed rate 0.15mm / r and cutting depth 0.5mm is selected. According to the melting point of the material 1660℃ and thermal conductivity 7.3W / (m·K), the optimal solution of Ti-6Al-4V Laser cutting was used for titanium alloy. The laser power was set to 800W, the spot diameter was 0.1mm, the pulse frequency was 2000Hz, and the gas pressure was 0.8MPa. Argon protection and oil mist lubrication were used during the cutting process. The finite element software ABAQUS was used to establish a three-dimensional thermal-mechanical coupling model for laser cutting of Ti-6Al-4V titanium alloy. An eight-node heat conduction unit and an eight-node reduced integration unit were used. The mesh size was 0.05mm. The temperature correlation and strain rate correlation of the material were considered. The thermophysical parameters and the Johnson-Cook constitutive equation were defined. The laser heat source used the Gaussian body heat source model. By writing D The FLUX subroutine realizes the movement of thermal load with cutting position, and uses dynamic temperature field distribution to predict cutting molten pool size, transient temperature gradient to predict residual stress distribution, plastic strain to predict cutting deformation, and stress distribution to predict cutting crack initiation. The simulation results show that under the optimized cutting parameters, the maximum size of the cutting molten pool is 0.8mm, the maximum residual stress is 500MPa, the maximum deformation is 0.1mm, and the risk of cutting crack initiation is low. According to the simulation results, a cutting process guidance plan for Ti-6Al-4V titanium alloy is formulated, and the cutting experience rules and optimization parameters of the material are saved in the cutting knowledge base and applied to actual production.
[0041] Through machine learning algorithms and knowledge reasoning, intelligent mapping and optimization between material processing performance, cutting parameters and quality are achieved. Finally, the above functional modules are integrated into the cutting process simulation software, and the parameter setting, processing simulation, optimization analysis and knowledge management of the Ti-6Al-4V titanium alloy cutting process are realized through a graphical interface to form an intelligent and integrated cutting processing CAE system.
[0042] Specifically, the step S103 extracts the geometric shape, size, thickness and structural feature information of the part to be cut according to the cutting process roadmap, and establishes a digital model of the part in combination with the material performance parameters, including the material hardness of HRC60, the tensile strength of 1000MPa, and the thermal conductivity of 50W / (m·K), to provide basic data for subsequent cutting path optimization, and adopts a graph theory algorithm to select a cutting path discretization method according to the geometric features of the cutting contour, including straight lines, arcs, and free curves;
[0043] For straight line segments, equal distance sampling is adopted, and the sampling interval is 2 mm; for arcs with a radius of 50 mm, equal angle sampling is adopted, and the sampling angle is 5°;
[0044] For free curves, key point sampling is used to extract 10 points with large curvature changes as key points;
[0045] The geometric outline of the part to be cut is converted into a graph structure consisting of 50 nodes and 80 edges. The topological structure of the cutting path is initially generated through the shortest path algorithm. Equidistant sampling is used for straight line segments, which is simple to calculate but has a large number of sampling points. Equal angle sampling is used for arcs to ensure cutting accuracy, but is not suitable for arcs with smaller radii. Key point sampling is used for free curves to reduce the amount of calculation, but some curve features will be lost.
[0046] For the initially generated cutting path, a genetic algorithm is used for optimization, with the optimization goal of minimizing the total cutting time and cutting cost. The objective function is the weighted sum of the total cutting time and cutting cost, with weights of 6 and 4 respectively. Considering the constraints of cutting order, tool turning, and cooling time, the chromosome length is 50, the population size is 100, the crossover probability is 8, the mutation probability is 05, and the evolutionary generation is 200. Through convergence analysis, it is found that the algorithm converges after 150 generations. The total cutting time of the optimal cutting path is 500s, and the cutting cost is 1000 yuan. For complex structural parts with a thickness of 50mm, a layered cutting strategy is adopted to divide the cutting path into 5 levels, each with a thickness of 10mm. In each level, a genetic algorithm is applied to optimize the cutting path, setting the cutting height to 2mm and the cutting speed to 500mm / min. By reasonably setting the layered threshold and cutting parameters, the cutting order and transition between layers are optimized, and the optimal solution of the overall cutting path is obtained. The total cutting time is 1500s, which saves 20% of the time compared with non-layered cutting.
[0047] The taboo search module is introduced to perform local search and improvement on the cutting operation scheme obtained by genetic algorithm optimization. The length of the taboo table is set to 20, the taboo object is the cutting path, and the contempt criterion is to reduce the cutting time by 10%. Through domain knowledge, the situation where the cutting turning radius is less than the tool radius is avoided, and a more refined and feasible cutting operation scheme is obtained. The cutting time is reduced by 5%, and the convergence speed is increased by 10%. For the multi-station cutting situation of 5 stations, a multi-objective optimization model is constructed. The objective function is to minimize the total cutting time and maximize the equipment utilization rate, with weights of 7 and 3 respectively. Considering the processing capabilities of different stations, the maximum cutting speed of the first station is 800mm / min, and that of the second station is 600mm / min; considering the processing order and cutting path constraints of 10 workpieces, a scheduling optimization method based on ant colony algorithm is adopted, and the number of ants is set to 50, the number of iterations is 100, the pheromone influence factor is 1, and the heuristic function influence factor is 5. By dynamically balancing the cutting load, the total cutting time is minimized to 2000s, and the equipment utilization rate reaches 85%, while avoiding interference and conflict between cutting paths;
[0048] Based on the performance parameters and thickness of the material, the edge threshold of the cutting line is calculated, and the cutting path is optimized based on the edge threshold so that the distance between the cutting path and the edge of the material exceeds the edge threshold;
[0049] According to the performance parameters of the material, including hardness, strength, and toughness, a machine learning algorithm is used to establish a prediction model between material properties and cutting parameters; by collecting experimental data of different materials under different cutting parameters, characteristic variables are selected, including material thickness, cutting speed, and cutting power, and support vector regression, neural network, and decision tree algorithms are used to evaluate the performance of the prediction model through cross-validation methods, and parameter tuning is performed to obtain the optimal cutting parameters for different materials, including cutting speed and cutting power, as the basis for subsequent calculation of edge thresholds and optimization of cutting paths;Obtain the 3D CAD model of the material to be cut, extract the material thickness information, calculate the edge threshold of the cutting line according to the material thickness and performance parameters, that is, the minimum safe distance between the cutting path and the material edge, and use the formula D=k×t, where D is the edge threshold, t is the material thickness, and k is a coefficient related to material performance. Determine the value range of k according to the specific situation, or use the finite element simulation method to establish a mechanical model of the material cutting process, analyze the cutting stress distribution, and determine the edge threshold to ensure the cutting quality and the strength of the cut piece. Discretize the cutting path and convert the cutting path into a digital number composed of discrete points. The model is constructed, and the B-spline curve and NURBS curve mathematical tools are used to fit and smooth the discrete points to generate a continuous and smooth cutting path curve. The cutting path curve is corrected and optimized according to the edge threshold to keep a sufficient safety distance from the edge of the material. The computational geometry algorithm, including the offset algorithm, is used to offset the optimized cutting path curve to generate an offset curve parallel to the edge of the material. The implementation principle and steps of the offset algorithm are as follows: first calculate the normal vector of each discrete point on the cutting path, then offset a certain distance in the direction of the normal vector to obtain the offset point, and finally connect the offset points to form the offset curve ,For different cutting path types, including straight line, arc, and free curve, different offset strategies and parameter settings are adopted. Then, the distance between the offset curve and the edge of the material is calculated, and the distance is compared with the edge threshold. If the distance is less than the threshold, the offset curve is adjusted until the minimum safety distance requirement is met. For cutting paths with complex shapes, the path decomposition and reorganization strategy is adopted to decompose the cutting path into several simple sub-paths, including straight line segments and arc segments. Each sub-path is subjected to offset processing and distance judgment respectively, and then the processed sub-paths are recombined into a cutting path to ensure that the entire cutting path meets the edge threshold requirement. A cutting process knowledge base is established to summarize the optimal process parameters and edge thresholds under different materials, different thicknesses, and different cutting paths. Big data analysis and machine learning algorithms, including decision trees and random forests, are used to mine the association rules between cutting process parameters and cutting quality, form an optimization decision model for the cutting process, guide the generation and optimization of the cutting path, and improve the intelligence level of the cutting process. In practical applications, cutting process optimization software is developed to integrate the above algorithms and models, and realize the automatic generation, optimization and simulation of the cutting path through a graphical interface to assist process personnel in the design and decision-making of the cutting process. ;
[0050] In one embodiment, when establishing a prediction model of material properties and cutting parameters, experimental data of 100 different materials under different cutting parameters are collected, 5 different thicknesses are selected for each material, the thickness range is 1-20 mm, 10 groups of cutting parameters are set for each thickness, the cutting speed range is 100-1000 mm / min, and the cutting power range is 500-2000 W. A total of 5000 groups of experimental data are obtained, material thickness, cutting speed, and cutting power are selected as characteristic variables, and material removal rate is selected as the target variable. 80% of the data are randomly selected as the training set, and 20% of the data are randomly selected as the test set. The support vector regression algorithm is used. Through grid search and cross validation, the hyperparameters of the model are optimized, including the penalty coefficient C and the kernel function type. The performance of the model is evaluated on the test set, including the mean absolute error MAE and the root mean square error RMSE, and the material removal rate prediction model is obtained. According to the prediction model, for 45# steel with a thickness of 10mm, the optimal cutting speed is predicted to be 600mm / min, the optimal cutting power is 1200W, and the material removal rate is 500mm³ / min. When calculating the edge threshold of the cutting line, for 45# steel with a thickness of 10mm, the manual is consulted to obtain the empirical formula D=0.1×t, and the edge threshold D is calculated to be 1mm. In order to verify In order to verify the rationality of the threshold, a three-dimensional thermal-mechanical coupling model of the cutting process of 45 steel was established using the finite element simulation software ANSYS. The material physical properties were input, including density 7.85g / cm³, thermal conductivity 46.6W / (m·K), specific heat capacity 460J / (kg·K). A 1200W Gaussian heat source was applied to simulate the temperature field and stress field distribution at different cutting speeds, and the stress level near the cutting line was evaluated. The results showed that when the cutting speed was 600mm / min, the equivalent stress within 1mm near the cutting line was less than the material yield strength of 355MPa, which met the strength requirements, verifying the rationality of the edge threshold. When the cutting path is offset, for the arc cutting path composed of 100 discrete points, the normal vector of each discrete point is calculated by the three-point method, and then 1mm is offset in the direction of the normal vector to obtain the offset point. Finally, all the offset points are fitted by the B-spline curve to generate the offset curve. The minimum distance between the offset curve and the edge of the material is calculated to be 0.98mm, which meets the edge threshold requirement. For complex cutting paths, including closed paths composed of straight lines, arcs, and free curves, they are first decomposed into 10 simple sub-paths, and then each sub-path is offset and recombined to obtain the offset curve. The minimum distance between the offset curve and the edge of the material is calculated to be 1.05mm, meeting the edge threshold requirements. Finally, the above cutting experimental data and optimization results are summarized into the cutting process knowledge base. The decision tree algorithm is used to generate a cutting process optimization decision tree with material type, thickness, cutting speed, and cutting power as input features and material removal rate as output target. For new cutting tasks, including 20mm thick Q235 steel, the optimal cutting process parameters predicted by the decision tree are cutting speed 800mm / min, cutting power 1500W, expected material removal rate 800mm³ / min, and edge threshold 2mm, which guides the actual cutting process. .
[0051] Specifically, after the cutting device in step S104 receives the cutting operation plan file containing the processing parameters, the cutting knife path and the battery protection board path are extracted through the G code parser to generate a trajectory composed of straight lines and arc segments, and the key parameters of the cutting power of 1000W and the battery protection board feed speed of 10mm / s are extracted. Then, the parsed cutting parameters are converted into pulses and direction signals recognizable by the motion control card, which are transmitted to the motion control card to control the servo motor and the stepper motor to realize the synchronous movement of the cutting knife and the battery protection board. At the same time, the power parameter of 1000W is converted into an analog voltage signal of 0-10V, which is transmitted to the laser controller to control the output power of the laser.
[0052] During the cutting process, the distance between the cutting blade and the processing surface is detected in real time by a laser rangefinder with a sampling frequency of 1kHz. The sampled data is subjected to Kalman filtering to extract the amplitude and frequency parameters of the vibration displacement of the cutting blade. When the vibration displacement amplitude exceeds 0.1mm or the frequency exceeds 100Hz, it is judged as cutting abnormality, triggering an audible and visual alarm. At the same time, the vibration signal is input into the fuzzy controller. According to the preset fuzzy rules, including "IF the amplitude is large, THEN the power is reduced", the laser power is adjusted to achieve vibration suppression.
[0053] During the cutting process, an infrared thermometer is used to detect the processing surface temperature in real time. When the temperature exceeds the melting point of the material, cutting is stopped to avoid scrapping the workpiece.
[0054] Through online learning algorithms, including recursive least squares method, the best cutting parameters for different materials and thicknesses are identified and optimized, and the optimized cutting parameters are stored in the knowledge base to form the cutting rule of "IF the material is steel AND the thickness is 2mm THEN the power is 800W AND the speed is 8mm / s", realizing intelligent cutting;
[0055] Finally, the real-time status and processing progress data of the cutting equipment are uploaded to the cloud through the MQTT protocol to achieve remote monitoring. It is also integrated with the ERP system and warehouse management system to automatically generate cutting tasks based on product orders and inventory conditions and send them to the cutting equipment to achieve flexible production.
[0056] The fault diagnosis of cutting equipment is based on the Bayesian network model. By online monitoring of machine vibration, temperature, and current parameters, the time domain and frequency domain features are extracted and input into the Bayesian network to calculate the posterior probability of the equipment being in the four states of "normal", "caution", "warning", and "fault". Based on the state prediction results, a maintenance plan is generated, including "the probability of failure is 90%, and the spindle bearing needs to be replaced", to guide equipment maintenance.
[0057] Specifically, in step S105, a K-type thermocouple temperature sensor and a piezoresistive pressure sensor are installed on the cutting equipment. The temperature sensor performs AD conversion through the MAX6675 chip and transmits data every 100ms through the RS485 bus using the Modbus-RTU protocol. The pressure sensor converts the 4-20mA current signal into a 0-5V voltage signal through a signal conditioning circuit and samples the signal at a frequency of 1000 times per second through a 16-bit AD acquisition card.
[0058] The collected temperature data is Kalman filtered, and the temperature mean and temperature slope time domain features are extracted with a 100ms window.
[0059] Perform short-time Fourier transform on the collected vibration data to extract the frequency domain energy characteristics in the 0-500Hz frequency band;
[0060] The laser rangefinder measures the three-dimensional coordinates of the surface of the cut piece 100 times per second, reconstructs the three-dimensional model of the cut piece through the triangulated meshing algorithm, and compares it with the CAD model to calculate the size deviation and material removal of the cut piece.
[0061] Material utilization rate = (cutting part volume) / (raw material volume);
[0062] If the utilization rate is lower than 90%, an alarm is triggered;
[0063] A high-speed camera is used to capture the cutting surface image at a speed of 200 frames per second, and the cutting contour is extracted by the Canny operator. The contour is then measured and it is determined to be a cutting burr defect when the contour exceeds 0.1 mm.
[0064] Feature selection was performed on 30 quality features collected online and offline, and 10 optimal feature subsets were selected using the minimum redundancy maximum relevance (mRMR) algorithm. The cutting quality prediction model was constructed using the SVM classifier, and the average classification accuracy of the model was 95.2% through 5-fold cross validation.
[0065] The I-MR control chart and xs control chart were drawn for the cutting vibration and surface roughness parameters, and the cutoff position was taken as μ±3σ. The process capability index Cpk of cutting vibration and surface roughness was calculated to be 1.33 and 1.45 respectively, both greater than 1.33, indicating that the cutting process was under control.
[0066] The three main components of cutting parameters were extracted by principal component analysis (PCA), and the cumulative contribution rate reached 85%. The regression model of the main components, cutting material removal rate and tool wear rate was obtained by multivariate linear regression analysis, and the regression coefficients R² were 0.89 and 0.92 respectively. When the material removal rate was lower than 100mm³ / s or the tool wear rate was higher than 0.01mm / s, the NSGA-Ⅱ algorithm was used to optimize the cutting parameters. During the optimization, the objective function was to maximize the material removal rate and minimize the tool wear rate. The spindle speed, feed speed and cutting depth were taken as decision variables. The population size was 50, the crossover probability was 0.8, and the mutation probability was 0.1. After 100 iterations, the Pareto optimal solution set was obtained. The cutting parameters with a material removal rate of 120mm³ / s and a tool wear rate of 0.008mm / s were selected as the optimization results in the optimal solution set and applied to actual processing.
[0067] The weight of the scraps dropped after cutting is monitored by a gravity sensor, and the material utilization rate is calculated based on the weight of the battery protection board before cutting.
[0068] A high-precision gravity sensor is installed at the feeding port of the cutting equipment. A resistance strain sensor with a measuring range of 0-10kg and an accuracy of 0.1g is used. Data is transmitted every 100ms through the RS485 bus using the Modbus-RTU protocol. The weight data of the scraps dropped after cutting is collected in real time, and the weight data is smoothed by the Kalman filter algorithm. The state equation of the Kalman filter is: x(k)=Ax(k-1)+Bu(k)+w(k), and the observation equation is: z(k)=Hx(k)+v(k). The historical data is fitted by the least squares method to estimate the parameters of the A, B, and H matrices. According to the fluctuation characteristics of the weight data, the initial values of the Q and R matrices are set. Then, through the prediction and update process of the Kalman filter, the optimal value of the weight state is recursively estimated to eliminate high-frequency noise and burr interference, thereby improving the stability and reliability of weight measurement.
[0069] Before cutting, the weight of the battery protection board to be cut is measured by a high-precision electronic weighing sensor. The weighing sensor adopts a resistance strain type or electromagnetic sensor with a range of 0-5kg and an accuracy of 0.1g. The 24-bit AD acquisition card is used to sample at a frequency of 10 times per second, and the multiple weighing data are fitted by the least squares method to obtain the average weight value of the battery protection board, which is used as the benchmark value for the subsequent material utilization calculation;
[0070] During the cutting process, the weight of the scraps is monitored in real time, and is associated with the weight data of the battery protection board before cutting through the database system. The material utilization rate is calculated by the ratio of the scrap weight to the battery protection board weight, that is, material utilization rate = (battery protection board weight - scrap weight) / battery protection board weight. The utilization rate data is used as a key indicator to evaluate the material utilization efficiency of the cutting process, and is compared with the preset target value to timely discover and warn of material waste problems;
[0071] Conduct statistical process control analysis on material utilization data, use I-MR control chart to monitor the change trend and abnormal situation of material utilization, set upper and lower control limits, including taking the cutoff position as μ±3σ. When 5 consecutive sample points fall outside the control limit, the material utilization rate warning is triggered, and relevant personnel are notified by email or SMS to analyze the cause and formulate improvement measures;
[0072] Data mining technology is used to conduct correlation analysis on material utilization data and cutting process parameters, tool parameters, and material batch multi-source heterogeneous data. The key factors affecting material utilization are discovered through the Apriori association rule algorithm:
[0073] First, according to the minimum support threshold, the frequent itemsets are generated in an iterative manner, that is, the combination of data items that meet the minimum support;
[0074] Secondly, for each frequent item set, according to the minimum confidence threshold, the association rules are generated, that is, the reasoning relationship between the elements in the frequent item set, and the strong association rules are screened out by setting the minimum support and minimum confidence;
[0075] At the same time, a material utilization prediction model is established through decision tree algorithms including ID3, C4.5, and CART. Information gain, information gain rate, and Gini index indicators are used for feature selection. A decision tree is generated from top to bottom in a recursive manner, and optimized by pre-pruning or post-pruning to improve the generalization ability of the model. When the cutting process parameters and tool parameters change, the changing trend of material utilization is predicted, and the risk of material waste is warned in advance.
[0076] In view of the low material utilization rate, a mathematical model with the goal of maximizing material utilization rate is established: maxf(x)=(Ww(x)) / W, where W is the weight of the battery protection board, w(x) is the weight of the scrap material, and x is the cutting process parameter vector. At the same time, considering the cutting quality and production efficiency constraints, a multi-objective optimization model is established: maxf1(x)=(Ww(x)) / W, maxf2(x)=Q(x), maxf3(x)=V(x), stg1(x)≤ε1, g2(x)≤ε2, …, where f1(x) is the material utilization rate target, f2(x) is the cutting quality target, f3(x) is the production efficiency target, Q(x) is the cutting quality evaluation function, V(x) is the production efficiency evaluation function, g1(x) and g2(x) are constraints, and ε1 and ε2 are constraint thresholds;
[0077] Genetic algorithm and particle swarm algorithm heuristic algorithm are used to solve the problem. Through selection, crossover and mutation operations, iterative optimization is continuously performed until the convergence conditions are met or the maximum number of iterations is reached. The Pareto optimal solution set is obtained, from which a better combination of cutting process parameters is selected. On the premise of meeting the cutting quality requirements, the number and weight of scraps are minimized, the material utilization efficiency is improved, and the material cost is saved. The optimized cutting plan is then sent to the CNC system and operators to guide actual production.
[0078] In one embodiment, a high-precision gravity sensor is installed at the feeding port of the cutting equipment, and a resistance strain sensor with a measuring range of 0-10kg and an accuracy of 0.1g is used. Data is transmitted every 100ms via the RS485 bus using the Modbus-RTU protocol, and the weight data of the scraps dropped after cutting is collected in real time. The weight data is smoothed by the Kalman filter algorithm, wherein the state equation of the Kalman filter is:
[0079] x(k)=Ax(k-1)+Bu(k)+w(k);
[0080] The observation equation is:
[0081] z(k)=Hx(k)+v(k);
[0082] The historical data is fitted by the least square method to estimate the parameters of the A, B, and H matrices, including A=
[95] , B=
[05] , and H=[0]. According to the fluctuation characteristics of the weight data, the initial values of the Q and R matrices are set, including Q=
[01] and R=[1]. Then, through the prediction and update process of the Kalman filter, the optimal value of the weight state is recursively estimated to eliminate high-frequency noise and burr interference, thereby improving the stability and reliability of weight measurement.
[0083] Before cutting, the weight of the battery protection board to be cut is measured by a high-precision electronic weighing sensor. The weighing sensor adopts a resistance strain type or electromagnetic sensor with a range of 0-5kg and an accuracy of 0.1g. The 24-bit AD acquisition card is used for sampling at a frequency of 10 times per second, and the multiple weighing data are fitted by the least square method, including fitting 10 weighing data [205, 203, 208, 201, 207, 204, 206, 209, 202, 205], and the average weight value of the battery protection board is 205kg, which is used as the benchmark value for subsequent material utilization calculation;
[0084] During the cutting process, the weight of the scraps from cutting is monitored in real time. The weight of the scraps from a certain cutting is 123kg, and it is associated with the weight of the battery protection board 205kg before cutting through the database system. The material utilization rate is calculated by the ratio of the weight of the scraps to the weight of the battery protection board, that is, material utilization rate = (205-123) / 205 = 40%. The utilization rate data is used as a key indicator to evaluate the material utilization efficiency of the cutting process and is compared with the preset target value of 95%. Material waste problems are discovered and warned in a timely manner. Statistical process control analysis is performed on the material utilization rate data. The I-MR control chart is used to monitor the change trend and abnormal conditions of the material utilization rate, and the upper and lower control limits are set, including taking the truncation position as μ±3σ. When 5 consecutive sample points: [82%, 89%, 85%, 83%, 81%] fall outside the control limit, the material utilization rate warning is triggered, and relevant personnel are notified by email or SMS to analyze the cause and formulate improvement measures;
[0085] Data mining technology is used to perform association analysis on material utilization data and multi-source heterogeneous data of cutting process parameters, tool parameters, and material batches. The key factors affecting material utilization are discovered through the Apriori association rule algorithm. First, according to the minimum support threshold of 10%, frequent item sets are generated in an iterative manner, that is, the combination of data items that meet the minimum support. Secondly, for each frequent item set, according to the minimum confidence threshold of 80%, association rules are generated, that is, the reasoning relationship between the elements in the frequent item set, including {tool model = A, material batch = B}=>{material utilization <90%}, with a support of 15% and a confidence of 85%. By setting the minimum support and minimum confidence, strong association rules are screened out. At the same time, a material utilization prediction model is established through decision tree algorithms including ID3, C5, and CART. Information gain, information gain rate, and Gini index indicators are used for feature selection. A decision tree is generated from top to bottom in a recursive manner, and optimized by pre-pruning or post-pruning to improve the generalization ability of the model. The generated decision tree is as follows:
[0086] Tool diameter ≤ 10mm: Material utilization = 92%;
[0087] Feed speed ≤ 50mm / min: Material utilization = 95%;
[0088] Feed speed>50mm / min: material utilization = 90%;
[0089] When cutting process parameters and tool parameters change, predict the changing trend of material utilization and warn of the risk of material waste in advance;
[0090] In view of the low material utilization rate, a mathematical model with the goal of maximizing material utilization rate is established: maxf(x)=(Ww(x)) / W, where W is the weight of the battery protection board, w(x) is the weight of the scrap, and x is the cutting process parameter vector;
[0091] At the same time, considering the cutting quality and production efficiency constraints, a multi-objective optimization model is established: maxf1(x)=(Ww(x)) / W, maxf2(x)=Q(x), maxf3(x)=V(x), stg1(x)≤ε1, g2(x)≤ε2, …:
[0092] Cutting defect rate ≤ 1%g2(x);
[0093] Cutting size deviation ≤ 05mmg3(x);
[0094] Cutting efficiency ≥ 6000 pieces / hour;
[0095] Where f1(x) is the material utilization rate target, f2(x) is the cutting quality target, f3(x) is the production efficiency target, Q(x) is the cutting quality evaluation function, V(x) is the production efficiency evaluation function, g1(x), g2(x), g3(x) are constraint conditions, ε1 and ε2 are constraint thresholds;
[0096] Genetic algorithm and particle swarm algorithm heuristic algorithm are used to solve the problem. Through selection, crossover and mutation operations, iterative optimization is continuously performed until the convergence condition is met or the maximum number of iterations is 500 generations. The Pareto optimal solution set is obtained, from which a better combination of cutting process parameters is selected, including a tool diameter of 8mm, a feed speed of 45mm / min, and a spindle speed of 6000rpm. Under the premise of meeting the cutting quality requirements, the number and weight of scraps are minimized, and the material utilization rate can reach more than 96%, realizing material cost savings. The optimized cutting plan is then sent to the CNC system and operators to guide actual production.
[0097] Monitor the surface image of the cutting surface through the visual sensor, extract the surface features of the cutting surface, including roughness, and quantify the cutting quality;
[0098] A high-resolution digital camera is installed on the cutting equipment as a visual sensor. A CCD camera with a resolution of 8 million pixels and a frame rate of 60 frames per second is selected. The image data is transmitted to the industrial computer in real time through the Gigabit Ethernet interface, and an image acquisition card is used to synchronously collect and cache the image data to ensure the integrity and real-time performance of the image data. The collected cutting surface image is preprocessed. Through image enhancement, denoising and correction operations, the noise and distortion interference factors in the image are eliminated to improve the image quality and provide a good data basis for subsequent feature extraction. Among them, median filtering is used to remove impulse noise, histogram equalization is used to enhance image contrast, and perspective transformation is used for geometric correction of the image. Through these preprocessing operations, a grayscale image of the cutting surface is obtained;
[0099] Based on the texture features of the image, the surface roughness features of the cut surface are extracted. The gray-level co-occurrence matrix (GLCM) algorithm is used to calculate the gray-level co-occurrence matrix of the image in different directions (0°, 45°, 90°, 135°) and different distances (1, 2, 3, 4 pixels). The contrast, correlation, and entropy texture features are extracted from the matrix. The comprehensive roughness index is obtained by weighted average to quantify the surface quality of the cut surface.
[0100] The calculation steps of the GLCM algorithm are to first calculate the gray-level co-occurrence matrix according to the selected direction and distance, and then calculate the contrast, correlation, and entropy texture features according to the matrix. For different cutting materials and processes, the best direction and distance parameter combination is selected through experiments to improve the discrimination and stability of the roughness feature;
[0101] According to the roughness characteristics of the cutting surface under different materials and different cutting processes, a cutting quality evaluation model is established. The support vector regression (SVR) algorithm is used to train the SVR model with roughness characteristics as input and manually scored cutting quality scores as output. The SVR algorithm minimizes the distance from the training sample to the hyperplane by finding the optimal hyperplane, while considering the complexity of the hyperplane to improve the generalization ability of the model.
[0102] When training the SVR model, the kernel function (linear kernel, Gaussian kernel) and penalty factor hyperparameters of the SVR are optimized through grid search and cross-validation to improve the prediction accuracy of the regression model and achieve quantitative evaluation of the cutting quality;
[0103] During the cutting process, the cutting surface image is collected in real time, the roughness features are extracted, and input into the SVR model for cutting quality prediction. If the predicted quality score is lower than the preset threshold, and the cutting quality score is lower than 60 points (out of 100), the cutting quality warning is triggered, and a warning signal is issued through the three-color light and sound and light alarm equipment to remind the operator to adjust the cutting parameters or perform equipment maintenance in time;
[0104] Adopt the statistical process control (SPC) method to control and analyze the cutting quality evaluation results, draw the XR control chart, monitor the changing trend and abnormal situation of cutting quality, and automatically identify the out-of-control state of cutting quality through the Western Electric rule or Nelson rule. The Western Electric rule includes 8 rules: 3 consecutive points exceed 2 standard deviations on both sides of the center line, and 7 consecutive points are on the same side of the center line. The Nelson rule includes 8 rules: 9 consecutive points are on the same side of the center line, and 6 consecutive points increase or decrease. Through these rules, abnormal fluctuations and trend changes in cutting quality can be sensitively identified. Combined with process parameters and equipment status data, root cause analysis is carried out, and corrective and preventive measures are formulated to achieve continuous improvement of cutting quality.
[0105] The cutting surface image, roughness characteristics, and quality assessment result data are stored in the MySQL relational database, and integrated with the MES and ERP production management systems. Data sharing and exchange are achieved through Web services and APIs. The correlation between cutting quality and material properties, tool parameters, and environmental factors is explored, and a cutting quality prediction model is established. This model serves as a supplement and extension of online quality assessment, guides cutting process optimization, improves cutting efficiency and product qualification rate, and realizes intelligent and digital cutting quality management.
[0106] In one embodiment, a Cognex In-Sight 8000 series smart camera is installed on the cutting equipment. The camera uses a 1 / 1.8-inch, 8-megapixel CCD sensor with a frame rate of up to 60 frames per second. The image data is transmitted to the industrial computer through a Gigabit Ethernet interface. The HALCON image processing software is used to collect and cache the image to ensure the integrity and real-time performance of the image data.
[0107] For the collected cutting surface image, median filtering is first performed, and the filter window size is set to 5×5 to eliminate the impulse noise in the image. Then histogram equalization is performed to map the image grayscale value to the entire range of 0-255 to improve the image contrast. Finally, perspective transformation is performed to correct the perspective distortion of the image through the coordinate mapping relationship of the four control points to obtain the grayscale image of the cutting surface.
[0108] The roughness characteristics of the cutting surface were extracted using the GLCM algorithm. The direction parameters were set to 0°, 45°, 90°, and 135°, and the distance parameter was set to 1 pixel. The seven texture features of contrast, correlation, and entropy were calculated based on the gray-level co-occurrence matrix. The comprehensive roughness index was obtained by weighted average, and the weight coefficient was determined by principal component analysis.
[0109] 500 cutting samples of different materials (45 steel, 6061 aluminum alloy) and different cutting processes (plasma cutting, laser cutting) were selected, and the roughness of the cutting surface was manually detected and scored to obtain the quality score (0-100 points). The roughness characteristics and quality scores were used as training data, and the LIBSVM toolbox was used to train the SVR model. The Gaussian kernel function was selected, and the optimal penalty factor C=10 and kernel function parameter g=0.1 were obtained through 10-fold cross validation and grid search. The prediction accuracy of the trained SVR model on the test set reached more than 95%;
[0110] During the cutting process, the cutting surface image is collected every 1 second, the roughness features are extracted, and the SVR model is input for quality prediction. When three consecutive prediction values are lower than 60 points, a quality warning is triggered, and the operator is prompted by a three-color light and voice broadcast. At the same time, the warning information is uploaded to the MES system, and an XR control chart is drawn for the cutting quality evaluation results. The UCL and LCL control limits are set to μ+3σ and μ-3σ respectively. The out-of-control state of the cutting quality is identified through the Western Electric rule. Three of the four consecutive points exceed the control limit. Combined with the cutting parameters and tool status data, the key factors causing the decline in cutting quality are found through data mining algorithms, including Apriori association rules and decision tree algorithms, including Apriori association rules and decision tree algorithms. Corresponding process adjustment suggestions are generated and fed back to operators and process engineers to continuously optimize cutting process parameters;
[0111] The cutting quality data is stored in the MySQL database and connected to the MES system through the Web service interface to achieve real-time sharing and traceability of quality data. A cutting quality prediction model is established through machine learning algorithms, including support vector machines and random forests, to achieve quality prediction based on material properties, tool parameters, and cutting parameter factors. The prediction accuracy reaches more than 90%, guiding the quality control and optimization of the cutting production process.
[0112] Specifically, in step S106, the size and shape of the scraps generated after the battery protection board is cut are identified, and the identification result is combined with the material type and the brand of the battery corresponding to the battery protection board, and recorded in the material management system.
[0113] A Keyence CV-X500 series smart camera is installed on the battery protection board cutting equipment. The camera uses a 5-megapixel CMOS sensor and transmits the scrap image to the industrial computer at a speed of 60 frames per second through the CameraLink interface. The HALCON20.11 software is used to collect and cache the image in real time. The image cache size is set to 1GB to ensure the integrity and real-time performance of the image data during high-speed cutting. For the collected scrap image, Gaussian filtering is first performed with a filter size of 5×5 and a standard deviation of 1.5 to remove high-frequency noise. Then histogram equalization is performed to improve the image contrast. Finally, the OTSU algorithm is used for adaptive threshold segmentation. The maximum inter-class variance method is used to search for the optimal threshold in the grayscale range of 0-255, and the grayscale image is converted into a black and white binary image. Morphological filtering is performed on areas with a size less than 100 pixels to eliminate small noise in the binary image. The advanced shape matching algorithm of HALCON is used to extract the geometric invariant moment (Hu moment) of the scrap contour, including The algorithm includes seven features, including area, center of gravity, and direction. The shape category with the highest matching degree is searched in the standard shape template library, including rectangle, triangle, and circle. The matching degree threshold is set to 0.8. If the threshold is exceeded, it is considered to be a known standard shape, otherwise it is judged to be an irregular shape. For rectangular scraps, the length and width of the circumscribed rectangle are calculated by HALCON's minimum circumscribed rectangle algorithm as the size of the scrap, and the dimensional accuracy can reach 0.1mm; for irregular scraps, the convex hull of the scraps is estimated by the α-shape algorithm, and then the circumscribed rectangle of the convex hull is calculated to obtain the approximate length and width of the scrap, and the dimensional accuracy can reach 0.5mm. The geometric shape and size information of the scraps are associated with the cutting process parameters, material batch, and product model production information to build a scrap information database. Using MySQL8.0 database, 10 data tables including scrap information table, material information table, and process parameter table are designed. The table association is realized through the key fields of material ID and process ID to support complex data query and statistical analysis.
[0114] Specifically, in the battery parts production process in step S107, the scrap data associated with the brand of the corresponding battery is queried in the material management system, and the scraps are matched for the battery parts through the shape matching algorithm. The material block layout and re-cutting instructions are calculated using the heuristic optimization algorithm and sent to the battery parts production system to execute the reuse of the matched scraps.
[0115] The battery brand BT-2023-A01 is queried in the material management system, and the data of 5 scraps associated with the brand are obtained, including the scrap number, size, and shape information. The shape matching algorithm is used to match the rectangular shape of the battery part with the shape data of the 5 scraps, and the matching degrees are 8, 6, 9, 7, and 5 respectively. The candidate scrap with the highest matching degree is the third one, with a matching degree of 9. The preset threshold is 85. Since the matching degree 9 of the candidate scrap with the highest matching degree exceeds the threshold, the third scrap is determined as the target scrap. The size of the target scrap is 100 cm × 80 cm. The heuristic optimization algorithm is used to calculate the optimal material block layout, which is to place 6 30 cm × 20 cm material blocks on the scrap. According to the optimal material block layout, the heuristic optimization algorithm is used to calculate The re-cutting instructions for the target scraps are to cut three times along the long side, each time at an interval of 30 cm; then cut four times along the wide side, each time at an interval of 20 cm. The target scrap number, size, optimal block layout and re-cutting instructions are sent to the battery parts production system. The battery parts production system cuts the target scraps according to the block layout and re-cutting instructions, and uses the cut 6 30 cm × 20 cm blocks to produce battery parts. After the battery parts are produced, the target scraps are marked as utilized in the material management system, and the scrap data is updated. The utilization of the scraps within one month is counted, and it is found that the utilization rate has increased by 15%. Then the scrap data in the material management system is sorted, the utilized scrap information is deleted, and the scrap storage method is optimized to improve the reuse efficiency.
[0116] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.
[0117] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and rules of the present invention shall be included in the protection scope of the present invention.
Claims
1. A method for automatic trajectory planning of battery protection board blanking, characterized in that: The mechanical strength and thermal conductivity of the preset composite materials and metal materials are evaluated by mechanical properties testers and thermal conductivity instruments, including evaluating mechanical strength through tensile strength, yield strength, and fracture toughness, and evaluating thermal conductivity through thermal conductivity and thermal diffusivity, to obtain material performance parameters, manage the test results using the material database, and classify the materials according to performance parameters: According to the preset types of composite materials and metal materials, the tensile strength, yield strength and fracture toughness of the materials are tested using a mechanical properties tester to obtain mechanical strength performance parameters; According to the preset types of composite materials and metal materials, thermal conductivity and thermal diffusivity of the materials are tested using a thermal conductivity instrument to obtain thermal conductivity performance parameters; Take mechanical strength performance parameters and thermal conductivity performance parameters as material properties, manage the test results through the material database, and establish the corresponding relationship between material performance parameters and material types; According to the numerical range of mechanical strength performance parameters, materials are divided into three categories: high-strength materials, medium-strength materials and low-strength materials, and the classification results are recorded in the material database; According to the numerical range of thermal conductivity parameters, materials are divided into three categories: high thermal conductivity materials, medium thermal conductivity materials and low thermal conductivity materials, and the classification results are recorded in the material database; If a material has high levels of mechanical strength and thermal conductivity, it is labeled as a high-quality material; If a material has low mechanical strength and thermal conductivity, it is labeled as ordinary material; If the rest occurs, it is marked as medium material; By querying the material database, the quantitative statistical information of high-quality materials, medium-quality materials and ordinary materials of various types of materials can be obtained, and a statistical report on material performance can be generated; According to the statistical report of material properties, judge the advantages and disadvantages of various materials in terms of mechanical strength and thermal conductivity, and generate guidance plans for material application; The material performance parameters, classification results, and statistical report information are stored in an associated manner to form a material performance evaluation and application knowledge base; Through the plastic processing heat treatment model, the processing difficulty of each type of material is analyzed according to the material classification results, including cutting force, tool wear rate, and processing temperature. The cutting strategy module generates a cutting process roadmap for different types of materials with different processing difficulties, and annotates the cutting speed, angle, cutting tool type, including laser cutting machine or plasma cutting machine, cooling and lubrication methods: According to the material classification results, the mechanical properties parameters of each type of material, including hardness, strength and toughness, are statistically analyzed, and a prediction model between material properties and cutting parameters is established using a machine learning algorithm; For continuous numerical performance indicators, support vector regression or neural network algorithms are used; For discrete indicators, use decision tree or random forest algorithms; Through data collection, feature selection, algorithm training and verification steps, the quantitative evaluation results of the cutting force, tool wear rate and processing temperature of the material type are obtained to determine the processing difficulty of each type of material; According to the material types with different processing difficulty, the cutting strategy module uses the cutting strategy optimization algorithm, including genetic algorithm or ant colony optimization algorithm, to optimize the cutting process route by setting the value range and constraint conditions of the cutting speed, feed rate, cutting depth, and angle process parameters, and generates a cutting process route map containing cutting order, cutting direction and cutting parameter information; Select the type of cutting tool based on the physical and chemical properties of the material, including melting point, thermal conductivity and light absorption; For high melting point materials, laser cutting is used to control the cutting energy density and heat affected zone range by adjusting the laser power, spot diameter and pulse frequency parameters; For materials with good thermal conductivity, plasma cutting is used by adjusting the plasma gas flow, arc current and cutting speed parameters; At the same time, the cooling and lubrication methods are determined according to the thermophysical properties of the material; For difficult-to-cut materials, finite element simulation technology is used to establish a three-dimensional physical model of the cutting process; Establish the finite element model through the steps of geometric modeling, meshing, material properties and constitutive relationship definition, boundary conditions and load setting; Adopting the incremental solution method and considering the influence of material damage, chip breaking and friction contact nonlinear factors, the stress-strain distribution and temperature field evolution in the cutting process are simulated, and the influence of cutting parameters on cutting force, tool wear and machining deformation is analyzed; In view of the high temperature and residual stress generated during the cutting process, multi-physics field coupling simulation technology, such as thermal-structural coupling analysis, is used to establish the control equations of multiple physical fields such as heat conduction, elastic-plastic deformation and damage evolution. The distribution results of temperature field, stress-strain field and damage field are obtained through finite element discretization and solution, and coupling analysis is performed to predict the deformation and stress distribution of the cut parts, optimize the cutting process and cutting sequence, and set the cooling process parameters, including the type of cooling medium and cooling time, to control the dimensional accuracy and surface quality of the cut parts. Establish a cutting process knowledge base and use ontology and semantic network methods to represent knowledge; Use machine learning algorithms to establish a mapping relationship between material properties and processing requirements and cutting process parameters; Analyze the cutting process roadmap, combine material performance parameters and structural characteristics, including shape, size and thickness, use genetic algorithm to optimize the cutting path, generate cutting operation plans, and use taboo search module to adjust the cutting sequence; Receiving a cutting operation plan through a cutting device, and adjusting cutting parameters of the cutting device according to the cutting operation plan, including a motion path of a cutting knife and a battery protection board, power output, and a feeding speed of the battery protection board; The data acquisition module is used to connect the sensors of the cutting equipment to monitor the cutting process in real time, calculate the material utilization rate and cutting quality, and use the statistical process control module to analyze the monitoring data. The principal component analysis module is used to calculate the cutting efficiency and material loss. If the cutting efficiency is less than the preset threshold or the material loss is greater than the preset threshold, the cutting strategy module is dynamically optimized. The size and shape of the scraps generated after the battery protection board is cut are identified, and the identification results are combined with the material type and the brand of the battery corresponding to the battery protection board, and recorded in the material management system; During the production process of battery parts, the scrap data associated with the corresponding battery brand is queried in the material management system, and the scraps are matched to the battery parts through the shape matching algorithm. The heuristic optimization algorithm is used to calculate the material block layout and re-cutting instructions, which are sent to the battery parts production system to execute the reuse of the matched scraps.
2. The method for automatic trajectory planning of battery protection board blanking according to claim 1, characterized in that: The cutting process roadmap is analyzed, and the cutting path is optimized by using a genetic algorithm in combination with material performance parameters and structural characteristics, including shape, size and thickness, to generate a cutting operation plan, and a taboo search module is used to adjust the cutting sequence, including: According to the cutting process roadmap, the geometric shape, size and thickness structural feature information of the parts to be cut are extracted, and the digital model of the parts is established in combination with the material performance parameters, including the hardness, strength and thermal conductivity of the material; Using graph theory algorithm, according to the geometric features of the cutting contour, including straight lines, arcs and free curves, the cutting path discretization method is selected, equal distance sampling, equal angle sampling and key point sampling, and the geometric contour of the part to be cut is converted into a graph structure composed of nodes and edges, where the nodes represent the starting and ending points of the cutting, and the edges represent the feasible cutting paths. The topological structure of the cutting path is preliminarily generated through the shortest path algorithm; The genetic algorithm is used to optimize the initially generated cutting path, with the optimization goal of minimizing the total cutting time and cutting cost. The objective function is the weighted sum method and the weighted product method. The cutting sequence, tool turning and cooling time constraints are considered at the same time. The cutting path is searched through chromosome encoding, fitness function, selection operator, crossover operator and mutation operator step calculation, and through convergence analysis and parameter tuning, the cutting operation plan is obtained. For complex structural parts, a layered cutting strategy is adopted. According to the thickness characteristics of the parts, the cutting path is divided into multiple levels, and the genetic algorithm is applied to optimize the cutting path in each level. By setting the layering threshold and cutting parameters, including cutting height and cutting speed, the cutting sequence and transition between layers are optimized, and a layered cutting algorithm flow chart is given; The taboo search module is introduced to perform local search and improvement on the cutting operation scheme obtained by genetic algorithm optimization; By setting the taboo table length, taboo objects and contempt criteria parameters, cutting turning and tool change nonlinear constraints are simultaneously implemented, the cutting operation scheme is obtained, and the algorithm convergence analysis and parameter tuning method are given. In view of the situation of multi-station cutting, a mathematical model of multi-objective optimization is constructed. The objective function minimizes the total cutting time and maximizes the equipment utilization rate. The processing capacity of different stations, the processing sequence of workpieces and the cutting path constraints of workpieces are considered. Intelligent scheduling algorithms are used, including scheduling optimization methods based on ant colony algorithm or particle swarm algorithm, to optimize the configuration of the cutting sequence and cutting resources of different parts. By dynamically balancing the cutting load, the total cutting time and equipment idle time are minimized. It also includes: calculating an edge threshold of a cutting line based on performance parameters and thickness of the material, and optimizing a cutting path based on the edge threshold so that a distance between the cutting path and an edge of the material exceeds the edge threshold; Specifically include: According to the material performance parameters, including hardness, strength, and toughness, a machine learning algorithm is used to establish a prediction model between material performance and cutting parameters; by collecting experimental data of different materials under different cutting parameters, characteristic variables are selected, including material thickness, cutting speed, and cutting power, and support vector regression, neural network, and decision tree algorithms are used to evaluate the performance of the prediction model through cross-validation methods, and parameter tuning is performed to obtain the cutting parameters of different materials, including cutting speed and cutting power; Obtain the 3D CAD model of the material to be cut, extract the material thickness information, and calculate the edge threshold of the cutting line according to the material thickness and performance parameters, that is, the minimum safe distance between the cutting path and the material edge. Use the formula D=k×t, where D is the edge threshold, t is the material thickness, and k is the material performance coefficient to determine the value range of k; Or use finite element simulation method to establish a mechanical model of the material cutting process, analyze the cutting stress distribution, determine the edge threshold, discretize the cutting path, convert the cutting path into a digital model composed of discrete points, use B-spline curve and NURBS curve mathematical tools to fit and smooth the discrete points, and generate a cutting path curve; Computational geometry algorithms, including offset algorithms, are used to offset the optimized cutting path curve to generate an offset curve parallel to the edge of the material; The steps of the bias algorithm are as follows: First, calculate the normal vector of each discrete point on the cutting path, then offset a certain distance in the direction of the normal vector to obtain the offset point, and finally connect the offset points to form an offset curve; For different cutting path types, including straight lines, arcs, and free curves, different offset strategies and parameter settings are used, and then the distance between the offset curve and the edge of the material is calculated, and the distance is compared with the edge threshold. If the distance is less than the threshold, the offset curve is adjusted; For cutting paths with complex shapes, the strategy of path decomposition and reorganization is adopted to decompose the cutting path into several simple sub-paths, including straight line segments and arc segments. Each sub-path is subjected to offset processing and distance judgment respectively, and then the processed sub-paths are recombined into a cutting path.
3. The method for automatic trajectory planning of battery protection board blanking according to claim 1, characterized in that: The cutting device receives the cutting operation plan, and adjusts the cutting parameters of the cutting device according to the cutting operation plan, including the motion path of the cutting knife and the battery protection board, the power output, and the feeding speed of the battery protection board, including: After receiving the cutting operation plan, the cutting device extracts the cutting parameters of the cutting knife and the battery protection board by parsing the cutting operation plan file, including the motion path, power output and the battery protection board feed speed, converts them into the instruction format recognized by the cutting device controller, including G code and M code, and transmits the instruction to the cutting device controller; The cutting equipment controller configures and adjusts the hardware of the cutting equipment according to the received cutting parameter instructions, including the motion trajectory of the cutting knife and the feeding mechanism of the battery protection board, and adjusts the output power of the laser and plasma power cutting energy sources according to the power output parameters through the servo motor and stepper motor actuators; During the cutting process, the cutting equipment controller monitors the position and speed of the cutting knife and the battery protection board in real time, and adjusts and optimizes the movement of the cutting knife and the battery protection board in real time through position feedback and speed feedback control algorithms, including PID control; Among them, the PID control algorithm includes position loop and speed loop, and the Ziegler-Nichols method and Cohen-Coon method are used to adjust the PID parameters; The cutting equipment controller uses sensors, including temperature sensors and photoelectric sensors, to collect temperature and cutting head position status information in real time during the cutting process, performs filtering and denoising preprocessing on the collected data, and extracts key characteristic parameters, including extracting temperature mean and cutting head vibration frequency characteristic parameters by using time domain analysis, frequency domain analysis, and wavelet transform signal processing methods, and judges whether the cutting state is normal by comparing with the preset threshold value; If an abnormality is detected, the alarm mechanism is triggered, and the operator is warned through voice synthesis, SMS push and email notification. At the same time, the cutting parameters are automatically adjusted according to the type of abnormality, and the cutting speed and power output parameters are adjusted using fuzzy control and neural network intelligent control methods; Optimization parameter module The cutting equipment controller integrates the optimization parameter module and machine learning algorithm to automatically generate cutting parameter combinations based on the type, thickness and cutting process parameters of the cutting material; Among them, the machine learning algorithm uses support vector machine, decision tree and random forest algorithm through the training and optimization process of the algorithm; The cutting equipment controller exchanges and integrates data with the MES system and ERP system through the network communication interface, and uploads the status information and processing data of the cutting equipment to the enterprise management system in real time; The cutting equipment controller integrates fault diagnosis and predictive maintenance algorithms. The health assessment model adopts statistical methods including multivariate regression analysis and principal component analysis. The fault prediction model adopts machine learning methods including neural networks, support vector machines and hidden Markov models through model training and verification, including cross-validation and leave-one-out method.
4. The method for automatic trajectory planning of battery protection board blanking according to claim 1, characterized in that: The data acquisition module is connected to the sensor of the cutting equipment to monitor the cutting process in real time, calculate the material utilization rate and cutting quality, and use the statistical process control module to analyze the monitoring data. The principal component analysis module is used to calculate the cutting efficiency and material loss. If the cutting efficiency is less than the preset threshold or the material loss is greater than the preset threshold, the cutting strategy module is dynamically optimized, including: The data acquisition module communicates with various sensors on the cutting equipment, including thermocouple temperature sensors using AD converters and RS485 communication protocols, and pressure sensors using 4-20mA current acquisition and Modbus-RTU communication protocols. The module collects temperature, pressure and vibration state parameters in the cutting process in real time, sets the sampling frequency according to the changing characteristics of the parameters, and sets the temperature sampling frequency to 10Hz and the vibration sampling frequency to 1kHz. The collected data is preprocessed, including limiting filtering, median filtering and Kalman filtering, to extract time domain features, frequency domain features and time-frequency domain features. During the cutting process, the laser distance sensor measures the size of the cut piece in real time, calculates the area and volume of the cut piece, obtains the size information of the raw material, calculates the material utilization rate, that is, the ratio of the volume of the cut piece to the volume of the raw material, and compares the utilization rate data with the preset target value; The cutting quality is evaluated by combining online and offline evaluation. The online evaluation collects cutting surface information in real time through image sensors and laser sensors to extract surface roughness and crack characteristic parameters. Offline evaluation: Use three-dimensional coordinate measuring machine and roughness tester to measure the cut parts offline, obtain dimensional accuracy and surface quality data, and establish a quantitative index system for cutting quality; Based on the collected quality data, a cutting quality evaluation model was constructed using support vector classification, support vector regression, decision tree and random forest machine learning algorithms; Perform statistical process control analysis on the collected cutting state parameters, select control charts according to data type and process characteristics, including mean-standard deviation control charts and individual-moving range control charts, monitor in real time whether temperature and pressure parameters exceed the upper and lower control limits, and calculate process capability indexes Cp and Cpk; Where, Cp=(USL-LSL) / (6σ), Cpk=min((USL-μ) / (3σ, (μ-LSL) / (3σ)), USL and LSL are the upper and lower specification limits, respectively, and μ and σ are the process mean and standard deviation, respectively; When abnormal parameters are detected or the process capability index is lower than the requirement, adjust the cutting parameters; For various cutting state parameters, the principal component analysis method is used for dimensionality reduction processing. The number of principal components is selected through cross-validation and cumulative variance contribution rate methods, the influencing factors of the cutting process are extracted, and the weight of each influencing factor is calculated. The correlation analysis is performed between the principal component scores and the cutting efficiency and material loss, and a regression model is established to quantify the influence of the principal components on efficiency and loss. A mathematical model of cutting efficiency and material loss is constructed, and the cutting efficiency and material loss values are calculated in real time. When the cutting efficiency is lower than the preset threshold of 0.8 or the material loss is higher than the preset threshold of 0.1, the cutting strategy optimization process is triggered; The cutting strategy optimization adopts multi-objective optimization algorithms, including non-dominated sorting genetic algorithm, multi-objective evolutionary algorithm based on decomposition and intensity Pareto evolutionary algorithm, setting population size, crossover probability and mutation probability parameters, balancing the algorithm's search ability and convergence speed, and setting and adjusting constraints and objective functions in a targeted manner for different cutting tasks and material characteristics; It also includes: using a gravity sensor to monitor the weight of the scraps that fall after cutting, and combining it with the weight of the battery protection board before cutting to calculate the material utilization rate; using a visual sensor to monitor the surface image of the cutting surface, extracting the surface features of the cutting surface, including roughness, and quantifying the cutting quality; The method monitors the weight of the scraps dropped after cutting by a gravity sensor, and calculates the material utilization rate in combination with the weight of the battery protection board before cutting, specifically including: installing a gravity sensor at the feed port of the cutting equipment, using a resistance strain sensor with a range of 0-10kg and an accuracy of 0.1g, transmitting data once every 100ms via the RS485 bus using the Modbus-RTU protocol, collecting the weight data of the scraps dropped after cutting in real time, smoothing the weight data using the Kalman filter algorithm, fitting the historical data using the least squares method, and setting the initial values of the Q and R matrices according to the fluctuation characteristics of the weight data, and then recursively estimating the optimal value of the weight state through the prediction and update process of the Kalman filter; Before cutting, the weight of the battery protection board to be cut is measured by an electronic weighing sensor. The weighing sensor adopts a resistance strain type or electromagnetic sensor with a range of 0-5kg and an accuracy of 0.1g. The 24-bit AD acquisition card is used to sample at a frequency of 10 times per second, and the multiple weighing data are fitted by the least squares method to obtain the average weight value of the battery protection board as the benchmark value for material utilization calculation; During the cutting process, the weight of the scraps from cutting is monitored in real time, and is associated with the weight data of the battery protection board before cutting through the database system. The material utilization rate is calculated by the ratio of the scrap weight to the battery protection board weight, that is, material utilization rate = (battery protection board weight - scrap weight) / battery protection board weight. The utilization rate data is used as a key indicator to evaluate the material utilization efficiency of the cutting process and is compared with the preset target value. The material utilization rate data is subjected to statistical process control analysis, and the I-MR control chart is used to monitor the changing trend and abnormal conditions of the material utilization rate. The upper and lower control limits are set, and the cutoff position is taken as μ±3σ. When 5 consecutive sample points fall outside the control limit, the material utilization rate warning is triggered, and the relevant personnel are notified by email or SMS to analyze the cause and formulate improvement measures; Data mining technology is used to perform association analysis on material utilization data and cutting process parameters, tool parameters, and material batch multi-source heterogeneous data. The key factors affecting material utilization are discovered through the Apriori association rule algorithm. First, according to the minimum support threshold, frequent item sets are generated in an iterative manner, that is, the combination of data items that meet the minimum support. Secondly, for each frequent item set, according to the minimum confidence threshold, association rules are generated, that is, the reasoning relationship between the elements in the frequent item set. By setting the minimum support and minimum confidence, strong association rules are screened out. At the same time, a material utilization prediction model is established through decision tree algorithms including ID3, C4.5, and CART. Information gain, information gain rate, and Gini index indicators are used for feature selection. A decision tree is generated from top to bottom through recursion, and optimized by pre-pruning or post-pruning. In view of the low material utilization rate, a mathematical model with the goal of maximizing material utilization rate is established: maxf(x)=(Ww(x)) / W, where W is the weight of the battery protection board, w(x) is the weight of the scrap, and x is the cutting process parameter vector; At the same time, considering the cutting quality and production efficiency constraints, a multi-objective optimization model is established; Genetic algorithm and particle swarm algorithm heuristic algorithm are used to solve the problem. Through selection, crossover and mutation operations, the Pareto optimal solution set is obtained until the convergence condition is met or the maximum number of iterations is reached. The better combination of cutting process parameters is selected from the solution to minimize the number and weight of scraps. The method of monitoring the surface image of the cutting surface by a visual sensor, extracting the surface features of the cutting surface, including roughness, and quantifying the cutting quality, specifically includes: A digital camera is installed on the cutting equipment as a visual sensor. A CCD camera with a resolution of 8 million pixels and a frame rate of 60 frames per second is selected. The image data is transmitted to the industrial computer in real time through the Gigabit Ethernet interface, and an image acquisition card is used to synchronously collect and cache the image data. Preprocess the collected cutting surface images through image enhancement, denoising and correction operations. The image preprocessing algorithms include median filtering, histogram equalization and perspective transformation. Among them, median filtering is used to remove impulse noise, histogram equalization is used to enhance image contrast, and perspective transformation is used for geometric correction of images. Through preprocessing operations, the grayscale image of the cutting surface is obtained. Based on the texture features of the image, the surface roughness features of the cutting surface are extracted. The grayscale co-occurrence matrix algorithm is used to calculate the grayscale co-occurrence matrix of the image at different directions of 0°, 45°, 90°, 135° and different distances of 1, 2, 3, and 4 pixels. The contrast, correlation, and entropy texture features are extracted from the matrix. The comprehensive roughness index is obtained by weighted average to quantify the surface quality of the cutting surface. The calculation steps of the GLCM algorithm are to first calculate the gray-level co-occurrence matrix according to the selected direction and distance, and then calculate the contrast, correlation, and entropy texture features based on the matrix; According to the roughness characteristics of the cutting surface under different materials and different cutting processes, a cutting quality evaluation model is established. The support vector regression algorithm is used to train the SVR model with the roughness characteristics as input and the manually scored cutting quality scores as output. The SVR algorithm minimizes the distance from the training sample to the hyperplane by finding the hyperplane. When training the SVR model, the kernel function and penalty factor hyperparameters of SVR are optimized through grid search and cross-validation; During the cutting process, the cutting surface image is collected in real time, the roughness features are extracted, and input into the SVR model for cutting quality prediction. If the predicted quality score is lower than the preset threshold and the cutting quality score is lower than 60 points, the cutting quality warning is triggered, and a warning signal is issued through the three-color light and sound and light alarm equipment; Use statistical process control methods to manage and analyze the cutting quality evaluation results, draw XR control charts, monitor the changing trends and abnormal conditions of cutting quality, and automatically identify the out-of-control state of cutting quality through Western Electric rules or Nelson rules; The cutting surface images, roughness characteristics, and quality assessment result data are stored in the MySQL relational database, and integrated with the MES and ERP production management systems. Data is shared and exchanged through Web services and APIs to explore the correlation between cutting quality and material properties, tool parameters, and environmental factors, and to establish a cutting quality prediction model.
5. The method for automatic trajectory planning of battery protection board blanking according to claim 1, characterized in that: The scraps generated after cutting the battery protection board are identified in size and categorized in shape, and the identification results are combined with the material type and the brand of the battery corresponding to the battery protection board, and recorded in the material management system, including: By installing a visual sensor on the cutting equipment, the images of the scraps produced after the battery protection board is cut are collected. An industrial camera is selected to transmit the image data to the industrial computer through the CameraLink interface, and the HALCON image processing software is used to collect and cache the images. The collected scrap images are preprocessed by image enhancement, denoising and binarization operations, and an adaptive threshold segmentation algorithm, including the OTSU algorithm, is used to automatically determine the optimal binarization threshold of the image by maximizing the inter-class variance. The algorithm steps include: Calculate the image grayscale histogram, calculate the inter-class variance, traverse all thresholds and find the threshold with the largest inter-class variance as the optimal threshold to obtain the binary image of the scraps; Based on the binary image of the scraps, the contour extraction algorithm, including contour tracing and region growing, is used to extract the contour information of the scraps, including the coordinates of the contour points, the perimeter of the contour and the area geometric parameters. The geometric shape of the scraps is fitted through the contour fitting algorithm, including the least squares method and Hough transform. The geometric invariant moment feature extraction method is used to calculate the 7 invariant moments of the contour and extract the geometric features of the contour. Then, the Euclidean distance and angle cosine measurement method are used to match it with the standard shape template. For regular-shaped scraps, the length, width and height dimension parameters of the scraps are calculated through geometric calculations, including the minimum circumscribed rectangle and the maximum inscribed circle; For irregularly shaped scraps, the bounding box method is used to extract the minimum circumscribed rectangle of the scrap outline and calculate the length and width of the rectangle as the estimated size of the scrap, compared with the pixel counting method; The shape classification results and size recognition results of the scraps are associated with the material type of the battery protection board and the brand information of the battery to which it belongs. The ER diagram is used to design the database table structure, and the scraps, materials and battery entities and relationships are abstracted into ER diagram models, scrap information tables, material information tables and battery information tables. MySQL and Oracle relational database systems are used to store the associated information in the material management system to record the material type, size, shape, quantity, generation time and battery brand attributes of the scraps.
6. The method for automatic trajectory planning of battery protection board blanking according to claim 1, characterized in that: In the battery parts production process, the scrap data associated with the brand of the corresponding battery is queried in the material management system, and the scraps are matched for the battery parts through the shape matching algorithm. The material block layout and re-cutting instructions are calculated using the heuristic optimization algorithm and sent to the battery parts production system to execute the reuse of the matched scraps, including: According to the battery brand of the battery parts, query and obtain the scrap data and scrap shape data associated with the battery brand in the material management system; Using shape matching algorithm, the shape of battery parts is matched with the shape data of scraps, and the scraps with the highest matching degree are obtained as candidate scraps; If the matching degree of the candidate scrap with the highest matching degree exceeds a preset threshold, the candidate scrap is determined as the target scrap, otherwise it is determined that there is no scrap for the current battery part production; According to the determined target scrap shape and size, a heuristic optimization algorithm is used to calculate the optimal block layout on the target scrap; According to the optimal material block layout, a heuristic optimization algorithm is used to calculate the re-cutting instructions of the target scraps; Send the target scrap information, optimal block layout and re-cutting instructions to the battery parts production system; After receiving the relevant information, the battery parts production system cuts the target scraps according to the material block layout and re-cutting instructions, and uses the cut material blocks for battery parts production; After the production of battery parts is completed, the scrap data in the material management system is updated and the used scrap information is marked.
Citation Information
Patent Citations
Automatic sheet metal part integrated nesting system and method thereof
CN102521459A
Material cutting method and device, control system, storage medium and electronic equipment
CN115167274A
Auxiliary system of laser cutting machine
CN116833582A
