Die machining self-adaptive control method and system based on machining state feedback

Abstract

The application provides a mold machining adaptive control method and system based on machining state feedback, and relates to the technical field of machining control. The method comprises the following steps: continuously collecting machining state data streams and constructing a time sequence machining state set; performing feature compression and state coding to obtain a first machining condition state vector; obtaining a tool structure parameter vector of a second tool and geometric and process constraint information; constructing a cross-tool mapping relationship to generate a target machining parameter distribution; extracting a target parameter combination to drive the second tool to perform machining on a mold region to be machined; constructing a second machining condition state vector in real time, calculating a machining condition state deviation; continuously adjusting to form a mold machining closed-loop adaptive control. The application solves the technical problem that the parameters of mold machining in the prior art are usually set according to experience or preliminary test, cannot be adjusted in real time and adapt to changes in the machining process, and result in unstable machining quality.

Description

Technical Field

[0001] This invention relates to the field of machining control technology, and more specifically to an adaptive control method and system for mold machining based on machining status feedback. Background Technology

[0002] With the development of modern manufacturing, the complexity and precision requirements of mold processing are gradually increasing. In order to improve the efficiency and quality of mold processing, more and more production lines have adopted advanced automated control technology. However, in the existing technology, the control parameters of mold processing are usually set based on experience or preliminary tests, and these control parameters are fixed throughout the processing. Since the materials, tools and processes involved in mold processing are different, these statically set processing parameters cannot cope with the changes that occur in actual processing. This leads to unstable processing quality. In particular, when problems such as tool wear and cutting force fluctuations are not detected and adjusted in time, it often leads to a decrease in processing accuracy or even tool damage. Summary of the Invention

[0003] This application provides an adaptive control method and system for mold processing based on processing status feedback, which aims to solve the technical problem that the parameters of mold processing in the prior art are usually set based on experience or preliminary tests, and cannot be adjusted and adapted to changes in the processing process in real time, resulting in unstable processing quality.

[0004] The first aspect disclosed in this application provides an adaptive control method for mold processing based on processing state feedback. The method includes: continuously acquiring processing state data streams during the processing of the mold's processing area by a first tool, constructing a time-series processing state set, wherein the processing state data streams include cutting load, vibration response, tool wear state, and processing stability indicators; performing feature compression and state encoding on the time-series processing state set to obtain a first processing condition state vector; acquiring the tool structure parameter vector of a second tool, and geometric and process constraint information of the current mold processing area; constructing a cross-tool mapping relationship based on the processing condition state vector, tool structure parameter vector, and geometric and process constraint information, combined with a parameter migration function, to generate a target processing parameter distribution; extracting a target parameter combination from the target processing parameter distribution based on stability and efficiency constraints, and driving the second tool to perform processing on the mold's processing area based on the target parameter combination; constructing a second processing condition state vector in real time during the second tool's processing, and calculating the processing condition state deviation based on the mold processing target state domain; continuously adjusting the target parameter combination based on the processing condition state deviation to form a closed-loop adaptive control for mold processing.

[0005] The second aspect of this application discloses an adaptive control system for mold processing based on processing state feedback. The system is used in the aforementioned adaptive control method for mold processing based on processing state feedback. The system includes: a processing state construction module, used to continuously collect processing state data streams during the processing of the mold's processing area by a first tool, constructing a time-series processing state set, wherein the processing state data stream includes cutting load, vibration response, tool wear state, and processing stability indicators; a state encoding module, used to perform feature compression and state encoding on the time-series processing state set to obtain a first processing condition state vector; and a constraint information acquisition module, used to acquire the tool structure parameter vector of a second tool, and the geometric and process constraint information of the current mold processing area. The system comprises: a mapping relationship construction module, used to construct a cross-tool mapping relationship based on the machining condition state vector, tool structure parameter vector, geometric and process constraint information, and combined with a parameter migration function, to generate a target machining parameter distribution; a machining execution module, used to extract a target parameter combination from the target machining parameter distribution based on stability and efficiency constraints, and drive the second tool to perform machining on the mold's machining area based on the target parameter combination; a state deviation calculation module, used to construct a second machining condition state vector in real time during the machining process of the second tool, and calculate the machining condition state deviation based on the mold machining target state domain; and a continuous adjustment module, used to continuously adjust the target parameter combination based on the machining condition state deviation, forming a closed-loop adaptive control for mold machining.

[0006] One or more technical solutions provided in this application have at least the following beneficial effects: By continuously collecting data streams of multiple indicators in real time during the machining process, a comprehensive understanding of the actual situation can be obtained. This real-time monitoring enables timely identification of potential problems. During machining, feature extraction and encoding techniques transform the raw machining state data into concise state vectors. These vectorized machining conditions provide an efficient expression of the current machining state, making subsequent calculations, comparisons, and optimizations more efficient. By combining the differences in machining characteristics between the first and second tools, as well as the geometric and process constraints of the current mold, cross-tool mapping is achieved. Based on this, a target machining parameter distribution suitable for the second tool is generated. This mapping effectively ensures that the adjusted machining parameters remain effective even when the tool is changed. Maintaining machining stability and accuracy; based on machining stability and efficiency constraints, automatically selecting the optimal parameter combination and driving the second tool, this process significantly improves machining efficiency and ensures that machining quality meets requirements, avoiding fluctuations in machining quality due to different tools; during the second tool machining process, the deviation between the machining state and the preset target state domain is calculated in real time. This deviation calculation reflects the difference between the actual machining state and the target state during the machining process, providing data support for subsequent adjustments; when the machining condition deviation exceeds the preset threshold, parameters are automatically adjusted. This closed-loop adaptive control mechanism can effectively correct deviations in the machining process and achieve self-optimization of the machining process.

[0007] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0008] Figure 1 This is a schematic flowchart of an adaptive control method for mold processing based on processing status feedback, provided in an embodiment of this application.

[0009] Figure 2 This is a schematic diagram of the mold processing adaptive control system based on processing status feedback provided in an embodiment of this application.

[0010] Explanation of reference numerals in the attached diagram: Processing state construction module 10, state encoding module 20, constraint information acquisition module 30, mapping relationship construction module 40, processing execution module 50, state deviation calculation module 60, and continuous adjustment module 70. Detailed Implementation

[0011] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0012] Example 1, as Figure 1 As shown in the embodiments of this application, an adaptive control method for mold processing based on processing status feedback is provided, the method comprising: During the machining process of the first tool on the area to be machined in the mold, the machining status data stream is continuously collected to construct a time series machining status set. The machining status data stream includes cutting load, vibration response, tool wear status and machining stability index.

[0013] While the first cutting tool is machining the area to be processed in the mold, various machining status data are collected in real time. Cutting load refers to the load borne by the tool during the cutting process, obtained by monitoring the cutting force through sensors; vibration response refers to the response caused by vibration during machining, acquired through sensors such as accelerometers, which helps monitor whether machining instability or vibration has occurred; tool wear status is assessed by monitoring the tool wear condition, such as wear depth and morphology, to evaluate tool life and machining accuracy; machining stability indicators include monitoring the stability of the machining process, such as vibration amplitude and cutting force fluctuations, reflecting whether the machining process is stable. The real-time collected machining status data stream is processed in a time-series manner to form a series of data sets arranged in chronological order. This data is used for subsequent analysis and control to understand the evolution of the machining process.

[0014] The time series processing state set is subjected to feature compression and state encoding to obtain the first processing condition state vector.

[0015] Feature compression is performed on the collected time-series processing state set to reduce redundant data and extract useful information. For example, principal component analysis is used to reduce the dimensionality of the original features and extract the principal components. The compressed features are then encoded using methods such as state vector encoding to transform complex processing state information into concise numerical representations. These encoded state vectors can reflect the key features of the processing. Through feature compression and state encoding, a state vector representing the processing condition is finally obtained. This vector is a concise description of the current processing state and contains all the key parameter information of the processing process.

[0016] Obtain the tool structure parameter vector of the second tool, as well as the geometric and process constraint information of the current mold processing area.

[0017] Obtain the structural parameters of the second tool, such as its geometry, number of cutting edges, material, and coating type. This information affects the tool's cutting performance and durability. Collect information on the geometry, dimensions, material properties, and machining accuracy requirements of the area to be machined in the current mold, as well as constraints during the machining process, such as maximum depth of cut and minimum feed rate. These constraints are used to determine feasible machining parameters.

[0018] Based on the machining condition state vector, tool structure parameter vector, and geometric and process constraint information, a cross-tool mapping relationship is constructed by combining the parameter migration function to generate the target machining parameter distribution.

[0019] By using the obtained first machining condition state vector, second tool structural parameter vector, and mold geometric and process constraint information, combined with the parameter transfer function, a mapping relationship is established between different tools. This means that, considering the structural differences between different tools, the machining parameters of the first tool are transformed into applicable parameters for the second tool. The parameter transfer function is a mathematical model used to describe the parameter transfer of different tools under the same machining state. For example, the first and second tools may differ in geometry, number of cutting edges, and cutting performance. The parameter transfer function uses these differences to generate machining parameters suitable for the second tool. The core of parameter transfer is mapping the machining state of the first tool to the corresponding machining parameters of the second tool. By generating a target machining parameter distribution, optimized parameter selection can be provided for subsequent mold machining processes.

[0020] Based on stability and efficiency constraints, a target parameter combination is extracted from the target machining parameter distribution, and the second tool is driven to perform machining on the mold's machining area based on the target parameter combination.

[0021] The stability of the machining process includes vibration stability and cutting force stability. In mold machining, stability is crucial to machining quality. Vibration or excessive fluctuations in cutting force during machining can lead to decreased machining accuracy or accelerated tool wear. Efficiency constraints consider the production efficiency of the machining process, including machining speed, machining time, and material removal rate. High-efficiency machining can improve production efficiency and reduce costs. However, increased efficiency may lead to decreased stability; therefore, a balance needs to be struck between stability and efficiency to find an optimal combination.

[0022] Within the target machining parameter distribution, based on stability and efficiency constraints, combinations of machining parameters that satisfy these two constraints are selected. A set of parameters best suited to the current machining conditions is then selected based on multi-level constraints. The optimization process simultaneously considers the balance between machining stability and efficiency, ensuring that both factors are effectively controlled during machining. The extracted target parameter combinations are used to drive the second tool to perform the actual machining operation on the mold's work area.

[0023] During the second tool machining process, a second machining condition state vector is constructed in real time, and the machining condition state deviation is calculated based on the target state domain of mold machining.

[0024] During the second tooling process, the machining status is continuously monitored, data is collected and processed, and a second machining condition state vector is constructed. The target state domain for mold machining is a preset ideal machining state range, including desired machining accuracy, machining stability, tool life, and other requirements. By comparing the real-time constructed second machining condition state vector with the target state domain, the machining condition state deviation is calculated. The state deviation is calculated through multi-dimensional matching, where each dimension represents different machining characteristics, such as vibration amplitude, cutting force, and tool wear. For example, if the vibration amplitude exceeds the target state domain, the deviation is large, indicating machining instability.

[0025] Based on the deviation of the processing condition, the target parameter combination is continuously adjusted to form a closed-loop adaptive control for mold processing.

[0026] Based on the deviations in machining conditions, control parameters are constructed, including the adjustment direction and magnitude of multiple deviation components, such as feed rate, spindle speed, and depth of cut. During adjustment, multiple machining parameters are considered simultaneously, and a collaborative adjustment method is used to optimize the combination of these parameters. The goal of collaborative adjustment is to minimize state deviations and ensure a smooth and efficient machining process. By continuously monitoring the machining process and adjusting based on real-time feedback, a closed-loop adaptive control is formed. This allows for automatic parameter adjustment during machining to adapt to different machining conditions, ensuring optimal machining quality and efficiency.

[0027] Furthermore, constructing a time-series processing state set includes: After performing multi-source synchronization alignment on the processing state data stream, data preprocessing is performed to form an original time series data stream; the original time series data stream is segmented and sliced ​​based on a sliding time window to generate multiple continuous time series state subsets; multidimensional statistical features and dynamic change features are extracted from the multiple continuous time series state subsets to construct a high-dimensional state feature matrix; feature fusion is performed based on the high-dimensional state feature matrix to generate the time series processing state set.

[0028] During mold processing, the collected data comes from multiple sensors or data streams, such as cutting load, vibration response, tool wear status, and machining stability. Since these data streams originate from different sensors, their timestamps and acquisition frequencies may differ, necessitating synchronization and alignment. The goal of synchronization and alignment is to align the timestamps of all data streams, ensuring correct pairing of data from multiple sensors at the same time point. After synchronization and alignment, data preprocessing is performed to remove noise and correct outlier data. Once synchronization, alignment, and preprocessing are complete, the time-series data from multiple data sources are integrated into a unified raw time-series data stream.

[0029] To extract local features from the original time series, a sliding time window is used. A sliding window is a fixed-length time range that slides across the original time series over time to capture data from different time intervals. The size of the sliding window determines the length of time for each data extraction and is set according to the characteristics of the processing procedure; the time step of each movement of the sliding window can be set to a fixed time interval. Using the sliding window technique, the entire original time series data stream is divided into multiple continuous time series state subsets. Each state subset contains processing state data over a period of time, and these subsets can be considered as individual samples, facilitating subsequent analysis.

[0030] Each time-series state subset contains multidimensional state information during the machining process, such as cutting load and vibration response. Multidimensional statistical features extracted from these subsets include mean, standard deviation, maximum and minimum values, skewness, and kurtosis. These statistical features reflect the overall trend and stability of the machining process. In addition to static statistical features, dynamic change features are also extracted to capture the changing trends of the state during machining, such as rate of change and trend. These dynamic features help to understand unstable factors or potential anomalies in the machining process.

[0031] By extracting multidimensional statistical features and dynamic change features from each continuous time series state subset, a high-dimensional state feature matrix is ​​finally obtained. Each row in the matrix represents the processing state of a time period, and the columns represent various features extracted from each subset. This matrix is ​​a mathematical description of the entire processing state and is used for subsequent analysis and modeling.

[0032] Feature fusion combines features from different sources or dimensions to improve their expressive power. For example, dimensionality reduction techniques map a high-dimensional feature space to a lower dimension, removing redundant features and retaining those that best reflect the processing state. Through feature fusion, different features are combined into a complete feature set representing the processing state. This feature set provides a comprehensive description of the processing state, including both static statistical features and dynamically changing information, which is used for subsequent model training, state recognition, and control decisions.

[0033] Furthermore, generating the target processing parameter distribution includes: Based on the parameter migration function, multiple candidate parameter sets are output, and a corresponding parameter candidate space is constructed. For the first candidate parameter set, the cutting load response, vibration response, and machining efficiency index are simulated and predicted to obtain the first simulation prediction result. Based on the first simulation prediction result, a multi-dimensional evaluation index system is constructed, including stability index, efficiency index, and tool life index. The multi-dimensional evaluation index system is input into a comprehensive evaluation function to score the first candidate parameter set, and the first parameter simulation evaluation result and the first probability distribution are output. Based on the first parameter simulation evaluation result and the first probability distribution, the target machining parameter distribution is constructed, which includes multiple mold machining areas.

[0034] Parameter transfer functions (PPFs) describe the parameter mapping relationships under different tools, machining stages, or process constraints. This function transforms the parameters of a tool under a certain machining state into parameters suitable for another tool or different machining stages. This is because different tools have different geometric parameters, number of cutting edges, cutting characteristics, etc. Based on known machining conditions, mold geometric constraints, and process requirements, the PPF outputs multiple candidate parameter sets. These candidate parameter sets include different combinations of machining parameters, such as feed rate, depth of cut, and tool spindle speed. Through the PPF, a set of possible machining parameters can be obtained for subsequent optimization and selection. These candidate parameter combinations form a parameter candidate space, representing the range of all possible parameter selections. This space contains multi-dimensional data on various parameter combinations, which can be explored in subsequent simulation prediction, evaluation, and optimization processes.

[0035] For the first set of candidate parameters, simulations are performed to predict key responses during the machining process, including cutting load, vibration response, and machining efficiency. Cutting load refers to the force exerted on the tool during machining. Simulating the cutting load response predicts whether different machining parameters will lead to tool overload or wear, affecting machining quality. Vibration response reflects the degree of vibration that may occur during machining. Excessive vibration can lead to machining instability, thus affecting workpiece accuracy or tool life. Simulating vibration response allows for the assessment of machining stability. Machining efficiency involves indicators such as material removal rate and machining time. Simulating machining efficiency predicts production efficiency under different parameter combinations and assesses whether it can meet production requirements.

[0036] Based on the initial simulation results, a multi-dimensional evaluation index system is constructed. This system comprehensively evaluates the performance of each candidate parameter set. For example, stability is measured based on vibration response and cutting load stability; a lower stability index indicates larger vibration or load variations during machining. Efficiency is measured based on machining efficiency and material removal rate; a higher efficiency index indicates faster material removal without affecting accuracy. Tool wear directly affects tool life and is evaluated using data such as cutting force, temperature, and wear. By combining these various indicators, a multi-dimensional evaluation index system is formed. Each candidate parameter set will be scored based on these indicators to facilitate the selection of the optimal parameters.

[0037] The comprehensive evaluation function combines multiple evaluation indicators, using weighted summation or other statistical methods to generate a single score. The comprehensive evaluation function assigns different weights to each indicator based on its importance. Through this function, the results of the multi-dimensional evaluation indicator system are transformed into a single score, representing the overall performance of the candidate parameter set. Furthermore, since parameter selection often involves uncertainty, a first probability distribution is calculated based on the score, representing the probability of selecting that parameter set. This probability distribution provides further information for subsequent optimization, helping to select the optimal solution.

[0038] Based on the simulation evaluation results of the first parameter and the corresponding first probability distribution, a target processing parameter distribution is generated. This target processing parameter distribution refers to the optimal selection of all candidate parameters and the corresponding parameter space, which corresponds to the specific requirements of multiple mold processing areas. Different areas may have different processing requirements, such as dimensions, materials, and processing accuracy. Therefore, when generating the target processing parameter distribution, these differences are considered to ensure that the processing requirements of each area are met.

[0039] Furthermore, extracting the target parameter combination from the target processing parameter distribution includes: In the target machining parameter distribution, multi-level constraint screening is performed based on the multiple mold machining requirements of the multiple mold machining areas. The optimal parameter combination is extracted from the screening results as the target parameter combination, while several alternative parameter combinations are retained as a dynamic adjustment candidate set. The multi-level constraint screening includes: S1: eliminating infeasible parameter sets based on machine tool power constraints and tool strength constraints; S2: screening stable parameter sets based on vibration stability constraints and machining accuracy constraints; S3: performing multi-objective optimization ranking based on the trade-off between machining efficiency and tool life.

[0040] The main purpose of multi-level constraint screening is to gradually eliminate parameter sets that do not meet the requirements and finally extract the optimal combination of target parameters. The screening process depends on different constraints, each of which is specific to a particular processing requirement, to ensure that the parameters finally selected can achieve the best processing effect while satisfying all constraints.

[0041] Every machine tool has its power limitations. Machining parameters that exceed the machine tool's power tolerance, such as excessively high depth of cut or feed rate, will cause machine tool overload or even malfunction. Therefore, when selecting machining parameters, it is necessary to ensure that the power requirements of all candidate parameter combinations do not exceed the machine tool's maximum power. Tool strength is another important limiting factor. The cutting force and wear of the tool during machining are all affected by tool strength. If the selected parameter combination leads to excessive tool wear or excessive cutting force, it will severely shorten tool life. By calculating the power and tool strength for each candidate parameter combination, those parameter combinations that exceed the machine tool power or tool strength limits are eliminated. These infeasible parameter combinations will not meet the actual machining requirements and are therefore removed from the candidate space.

[0042] Excessive vibration during machining can lead to instability, affecting machining accuracy and even damaging equipment or cutting tools. If a combination of parameters causes excessive vibration, that parameter set should be eliminated. Machining accuracy constraints ensure that the machined workpiece meets design requirements. Machining accuracy varies under different machining parameters. By calculating and simulating errors during machining, the machining accuracy of each candidate parameter combination can be evaluated, and parameter sets that guarantee machining accuracy can be selected. By evaluating the vibration stability and machining accuracy of candidate parameters, stable parameter combinations are selected that ensure the stability of the machining process and meet machining accuracy requirements.

[0043] Machining efficiency and tool life are often mutually restrictive. High-efficiency machining, such as with large depths of cut and high feed rates, can lead to accelerated tool wear and shorten tool life, while choosing overly conservative machining parameters may result in low machining efficiency. Therefore, a trade-off must be struck between machining efficiency and tool life to ensure that production efficiency is met while also extending tool life and reducing downtime maintenance costs. Through multi-objective optimization, candidate parameter combinations are ranked according to their comprehensive scores. Finally, the parameter combination with the highest score is selected as the optimal parameter combination, while those that are relatively suboptimal are retained as alternative parameter combinations.

[0044] Furthermore, it also includes: The second tool is driven to process the area to be processed by the mold based on the target parameter combination. The second processing condition state vector is acquired in real time, and the processing condition state deviation between the second processing condition state vector and the target state domain is calculated. When the processing condition state deviation exceeds the preset deviation threshold, a candidate parameter combination is selected from the dynamic adjustment candidate set. The current processing parameters are gradually adjusted based on the candidate parameter combination.

[0045] The second tool is driven to actually machine the area to be processed on the mold according to the selected combination of target parameters, and the machining status is monitored in real time. Data is collected by sensors and converted into a second machining condition state vector. The preset target state domain includes the ideal range of machining parameters and state ranges, such as the expected machining accuracy, cutting force range, and vibration stability. By comparing the real-time acquired second machining condition state vector with the target state domain, the machining condition state deviation is calculated. This deviation measures the difference between the current machining state and the ideal state. If the deviation is large, it indicates that the machining state does not meet expectations and adjustments are needed.

[0046] A preset deviation threshold is set, which is based on experience or machining requirements. The deviation of the machining condition is compared with the preset deviation threshold. If the deviation exceeds the preset threshold, it indicates that the current machining state has deviated from expectations, which may lead to unqualified machining quality or tool damage. At this time, alternative parameter combinations are selected from the dynamic adjustment candidate set. The dynamic adjustment candidate set consists of alternative solutions retained from the previous screening process. Although they are not the optimal solutions, they can still meet the machining requirements under the current conditions.

[0047] The current machining parameters are gradually adjusted based on alternative parameter combinations. Gradual adjustment means that the machining parameters are not changed drastically at once, but rather adjusted step by step to avoid excessive changes and reduce instability or errors in the machining process. This method can ensure machining quality while reducing the impact on the existing machining process, allowing the machining to gradually return to the target state range.

[0048] Furthermore, constructing cross-tool mapping relationships by combining parameter transfer functions includes: The differences in geometric parameters, number of cutting edges, and cutting characteristics between the first and second cutting tools at different processing stages of the mold are obtained. Based on the differences in geometric parameters, a proportional mapping relationship between feed rate and cutting width is established. Based on the differences in the number of cutting edges and cutting characteristics, an equivalent conversion relationship for unit cutting edge load is established. Based on the correlation between cutting thickness and material removal rate, a cutting depth compensation relationship is constructed. Based on the proportional mapping relationship, the equivalent conversion relationship for unit cutting edge load, and the cutting depth compensation relationship, the output result of the parameter migration function is corrected to obtain a parameter mapping result adapted to the second cutting tool.

[0049] The geometric parameters of a cutting tool include its cutting edge shape, cutting edge angle, diameter, and length. These parameters vary significantly between different tools, affecting cutting forces, cutting efficiency, and machining accuracy. The number of cutting edges refers to the number of cutting edges on the tool, including single-edge and multi-edge tools. Multi-edge tools offer higher cutting efficiency but also face different load distribution and tool wear patterns. Cutting characteristics include the tool's adaptability and properties to the material during cutting, such as cutting forces, cutting heat, tool life, tool material, and coatings.

[0050] The geometry of the cutting tool has a significant impact on the feed rate during machining. Larger tool diameters require lower feed rates to avoid overload, while smaller tools require higher feed rates to maintain machining efficiency. The width of cut also varies with the tool's geometry; generally, the larger the tool diameter, the larger the width of cut. The width of cut directly affects the cutting area and tool load. Based on the differences in geometry, a proportional mapping relationship between feed rate and width of cut is established. This mapping relationship allows for the reasonable conversion and adaptation of machining parameters such as feed rate and width of cut for both the first and second cutting tools.

[0051] The number of cutting edges of a cutting tool directly affects the cutting load per unit cutting edge. Tools with more cutting edges bear a smaller load per cut, thus distributing cutting force more evenly and improving cutting efficiency and tool life. Differences in cutting characteristics also affect the distribution of cutting force. For example, coated tools reduce friction and lower cutting forces, while carbide tools are more suitable for high-load cutting. By analyzing the differences between the number of cutting edges and cutting characteristics, an equivalent conversion relationship for the load per unit cutting edge is established, allowing for the equivalent conversion of the load per unit cutting edge between the first and second cutting tools.

[0052] Depth of cut and material removal rate are important parameters affecting machining efficiency. The depth of cut directly determines the amount of material removed in each cut. A larger depth of cut can improve the material removal rate, but it may also increase tool load and vibration, affecting machining stability and tool life. By establishing a depth of cut compensation relationship, the material removal rate can be adjusted according to the cutting thickness of different tools. This compensation relationship helps maintain high material removal efficiency and machining stability under different tools and machining conditions.

[0053] By applying the aforementioned proportional mapping relationship, unit cutting edge load equivalent conversion relationship, and cutting depth compensation relationship, the output of the parameter migration function is corrected. These corrections ensure that the machining parameters of the first tool can reasonably adapt to the characteristics of the second tool. Based on the corrections, a set of parameters suitable for the second tool is generated. These parameters include feed rate, cutting depth, and tool speed, which have been adjusted through the aforementioned mapping relationship and compensation to ensure that the second tool can achieve the expected machining effect during machining.

[0054] Furthermore, during the second tool machining process, a second machining condition state vector is constructed in real time, and the machining condition state deviation is calculated based on the mold machining target state domain, including: During the second tool machining process, the machining status data stream is continuously updated, and a dynamic data buffer is constructed; based on the dynamic data buffer, continuous status data segments are extracted using a sliding window method; the continuous status data segments are subjected to feature extraction and encoding processing to generate a second machining condition status vector; based on the second machining condition status vector and a preset target status domain, a multi-dimensional matching is performed, and the machining condition status deviation is calculated based on the multi-dimensional matching result.

[0055] During the machining process with the second tool, machining status data, such as cutting force, vibration response, and tool wear, are continuously collected. This data reflects the real-time status of the machining process and helps determine whether adjustments are needed. The dynamic data buffer is a temporary storage area used to store the most recently collected machining status data. The purpose of the data buffer is to store a certain amount of historical data to facilitate the analysis of machining trends.

[0056] In the dynamic data buffer, a sliding window approach is used to process the stored processing status data. By sliding a fixed-size window within the data buffer, continuous time-segment data is extracted. The sliding window technique extracts continuous status data segments from the dynamic data buffer; each segment includes processing status data within a certain time range, describing the processing procedure within that time period.

[0057] In each extracted continuous state data segment, key information reflecting the processing state is extracted using feature extraction methods. The extracted features are then encoded and transformed into a numerical form that is easy to understand and process. After feature extraction and encoding, a second processing condition state vector is generated. This state vector contains all the key features of the current processing process and can represent the processing state at that moment.

[0058] The target state domain is a pre-defined ideal machining state range, including desired machining accuracy, vibration level, cutting force range, etc., reflecting the ideal values ​​or allowable fluctuation ranges of various indicators during machining. A multi-dimensional matching process is performed between the second machining condition state vector and the target state domain. This multi-dimensional matching means comparing each feature in the second machining condition state vector with the corresponding feature in the target state domain and calculating the matching degree. Based on the results of the multi-dimensional matching, the machining condition state deviation is calculated, reflecting the difference between the current machining state and the target state.

[0059] Furthermore, continuously adjusting the target parameter combination based on the deviation of the processing condition includes: Based on the deviation of the machining condition, a control adjustment quantity is constructed. The control adjustment quantity includes multiple adjustment directions and adjustment amplitudes of multiple deviation components. The multiple deviation components include feed rate, spindle speed and depth of cut. The feed rate, spindle speed and depth of cut are adjusted in a multi-variable coordinated manner. During the coordinated adjustment process, constraints are introduced to limit the rate of change of parameters.

[0060] Control adjustments are used to adjust machining parameters to restore the machining process to the target state range. Control adjustments include not only the adjustment range of each parameter but also the direction of adjustment. Deviation components correspond to different machining parameters, including feed rate, spindle speed, and depth of cut; each deviation component has its corresponding adjustment direction and range.

[0061] In actual machining processes, feed rate, spindle speed, and depth of cut are interrelated. Adjusting one parameter alone may affect the others. The goal of multivariate coordinated adjustment is to optimize the adjustment by comprehensively considering the relationships between all parameters. When performing multivariate coordinated adjustment, constraints are introduced to prevent rapid parameter changes from causing machining instability or equipment failure. These constraints include machine tool limitations, such as the maximum adjustable range of feed rate and spindle speed; and tool durability, as excessive depth of cut and excessively fast feed rate can lead to rapid tool wear, thus requiring limits on the rate of change of these parameters. These constraints ensure that each adjustment does not exceed the capacity of the machining equipment and tools, while also making the adjustment process smooth and controllable.

[0062] Furthermore, this includes: In the closed-loop adaptive control of mold processing, the data of the entire processing process is recorded. The data of the entire processing process includes the processing state sequence, parameter migration input and output, parameter adjustment trajectory and processing quality results. The data of the entire processing process is structured and organized to construct a training sample dataset. The parameter migration function is updated based on the training sample dataset and deployed for subsequent tool switching processes.

[0063] In mold processing, to ensure effective subsequent optimization and adjustment, the entire processing process is recorded. Specifically, data points at various states during processing, such as cutting force, vibration, and tool wear, are recorded to form a processing state sequence for analyzing fluctuations and instabilities. Inputs to the parameter migration function, such as processing conditions and tool parameters, and outputs, such as target processing parameters and actual processing parameters, are recorded. This helps in understanding how to map the parameter relationships between different tools and conditions. The trajectory of parameter adjustments during processing is recorded, including the adjustment time, the magnitude of parameter changes, and their effects, to help evaluate the effectiveness of adjustment strategies. Finally, processing quality-related indicators, such as dimensional accuracy and surface quality, are recorded to determine the final effect of the entire processing process.

[0064] The collected data from the entire processing stage is structured and organized, meaning the data is classified and labeled according to certain formats and standards. This facilitates subsequent data analysis and model training. The organized data serves as the training sample dataset for subsequent machine learning or model optimization.

[0065] By utilizing the training sample dataset, the original parameter transfer function is updated. Machine learning or statistical methods are used to learn the relationships between parameters from historical machining data, further improving the accuracy and adaptability of the transfer function. The updated transfer function can more accurately map the machining parameters of the first tool to the second tool, thereby improving the stability and efficiency of the machining process. Through continuous learning and optimization, the transfer function can adapt to changes in different tools and machining conditions. Deploying the updated parameter transfer function into actual tool change processes allows for automatic adjustment of machining parameters during tool switching, ensuring that the machining process maintains optimal performance even after tool changes.

[0066] Example 2 is based on the same inventive concept as the mold processing adaptive control method based on processing state feedback in the previous examples, such as... Figure 2 As shown in the embodiment of this application, an adaptive control system for mold processing based on processing status feedback is provided. The system includes: The machining state construction module 10 is used to continuously collect machining state data streams during the machining process of the first tool on the mold's machining area, and construct a time-series machining state set. The machining state data stream includes cutting load, vibration response, tool wear state, and machining stability indicators. The state encoding module 20 is used to perform feature compression and state encoding on the time-series machining state set to obtain a first machining condition state vector. The constraint information acquisition module 30 is used to acquire the tool structure parameter vector of the second tool, as well as the geometric and process constraint information of the current mold's machining area. The mapping relationship construction module 40 is used to construct a mapping relationship based on the machining condition state vector, tool structure parameter vector, and... Geometric and process constraint information, combined with parameter migration functions, are used to construct cross-tool mapping relationships and generate a target machining parameter distribution. Machining execution module 50 extracts target parameter combinations from the target machining parameter distribution based on stability and efficiency constraints, and drives the second tool to perform machining on the mold's processing area based on these target parameter combinations. State deviation calculation module 60 constructs a second machining condition state vector in real time during the second tool's machining process and calculates the machining condition state deviation based on the mold machining target state domain. Continuous adjustment module 70 continuously adjusts the target parameter combinations based on the machining condition state deviation, forming a closed-loop adaptive control for mold machining.

[0067] Furthermore, the processing status construction module 10 is used to perform the following operation steps: After performing multi-source synchronization alignment on the processing state data stream, data preprocessing is performed to form an original time series data stream; the original time series data stream is segmented and sliced ​​based on a sliding time window to generate multiple continuous time series state subsets; multidimensional statistical features and dynamic change features are extracted from the multiple continuous time series state subsets to construct a high-dimensional state feature matrix; feature fusion is performed based on the high-dimensional state feature matrix to generate the time series processing state set.

[0068] Furthermore, the mapping relationship construction module 40 is used to perform the following operation steps: Based on the parameter migration function, multiple candidate parameter sets are output, and a corresponding parameter candidate space is constructed. For the first candidate parameter set, the cutting load response, vibration response, and machining efficiency index are simulated and predicted to obtain the first simulation prediction result. Based on the first simulation prediction result, a multi-dimensional evaluation index system is constructed, including stability index, efficiency index, and tool life index. The multi-dimensional evaluation index system is input into a comprehensive evaluation function to score the first candidate parameter set, and the first parameter simulation evaluation result and the first probability distribution are output. Based on the first parameter simulation evaluation result and the first probability distribution, the target machining parameter distribution is constructed, which includes multiple mold machining areas.

[0069] Furthermore, the processing execution module 50 is used to perform the following operation steps: In the target machining parameter distribution, multi-level constraint screening is performed based on the multiple mold machining requirements of the multiple mold machining areas. The optimal parameter combination is extracted from the screening results as the target parameter combination, while several alternative parameter combinations are retained as a dynamic adjustment candidate set. The multi-level constraint screening includes: S1: eliminating infeasible parameter sets based on machine tool power constraints and tool strength constraints; S2: screening stable parameter sets based on vibration stability constraints and machining accuracy constraints; S3: performing multi-objective optimization ranking based on the trade-off between machining efficiency and tool life.

[0070] Furthermore, the processing execution module 50 is used to perform the following operation steps: The second tool is driven to process the area to be processed by the mold based on the target parameter combination. The second processing condition state vector is acquired in real time, and the processing condition state deviation between the second processing condition state vector and the target state domain is calculated. When the processing condition state deviation exceeds the preset deviation threshold, a candidate parameter combination is selected from the dynamic adjustment candidate set. The current processing parameters are gradually adjusted based on the candidate parameter combination.

[0071] Furthermore, the mapping relationship construction module 40 is used to perform the following operation steps: The differences in geometric parameters, number of cutting edges, and cutting characteristics between the first and second cutting tools at different processing stages of the mold are obtained. Based on the differences in geometric parameters, a proportional mapping relationship between feed rate and cutting width is established. Based on the differences in the number of cutting edges and cutting characteristics, an equivalent conversion relationship for unit cutting edge load is established. Based on the correlation between cutting thickness and material removal rate, a cutting depth compensation relationship is constructed. Based on the proportional mapping relationship, the equivalent conversion relationship for unit cutting edge load, and the cutting depth compensation relationship, the output result of the parameter migration function is corrected to obtain a parameter mapping result adapted to the second cutting tool.

[0072] Furthermore, the state deviation calculation module 60 is used to perform the following operation steps: During the second tool machining process, the machining status data stream is continuously updated, and a dynamic data buffer is constructed; based on the dynamic data buffer, continuous status data segments are extracted using a sliding window method; the continuous status data segments are subjected to feature extraction and encoding processing to generate a second machining condition status vector; based on the second machining condition status vector and a preset target status domain, a multi-dimensional matching is performed, and the machining condition status deviation is calculated based on the multi-dimensional matching result.

[0073] Furthermore, the continuous adjustment module 70 is used to perform the following operation steps: Based on the deviation of the machining condition, a control adjustment quantity is constructed. The control adjustment quantity includes multiple adjustment directions and adjustment amplitudes of multiple deviation components. The multiple deviation components include feed rate, spindle speed and depth of cut. The feed rate, spindle speed and depth of cut are adjusted in a multi-variable coordinated manner. During the coordinated adjustment process, constraints are introduced to limit the rate of change of parameters.

[0074] Furthermore, the continuous adjustment module 70 is used to perform the following operation steps: In the closed-loop adaptive control of mold processing, the data of the entire processing process is recorded. The data of the entire processing process includes the processing state sequence, parameter migration input and output, parameter adjustment trajectory and processing quality results. The data of the entire processing process is structured and organized to construct a training sample dataset. The parameter migration function is updated based on the training sample dataset and deployed for subsequent tool switching processes.

[0075] Through the foregoing detailed description of the mold processing adaptive control method based on processing state feedback, those skilled in the art can clearly understand the mold processing adaptive control system based on processing state feedback in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section.

[0076] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An adaptive control method for mold processing based on processing status feedback, characterized in that, The method includes: During the machining process of the first tool on the mold's machining area, a machining status data stream is continuously collected to construct a time-series machining status set. The machining status data stream includes cutting load, vibration response, tool wear status, and machining stability indicators. The time-series processing state set is subjected to feature compression and state encoding to obtain the first processing condition state vector; Obtain the tool structure parameter vector of the second tool, as well as the geometric and process constraint information of the current mold processing area; Based on the machining condition state vector, tool structure parameter vector, geometric and process constraint information, a cross-tool mapping relationship is constructed by combining the parameter migration function to generate the target machining parameter distribution. Based on stability and efficiency constraints, a target parameter combination is extracted from the target machining parameter distribution, and the second tool is driven to perform machining on the mold area to be machined based on the target parameter combination. During the second tool machining process, a second machining condition state vector is constructed in real time, and the machining condition state deviation is calculated based on the mold machining target state domain; Based on the deviation of the processing condition, the target parameter combination is continuously adjusted to form a closed-loop adaptive control for mold processing.

2. The adaptive control method for mold processing based on processing state feedback as described in claim 1, characterized in that, Construct a time series processing state set, including: After performing multi-source synchronization alignment on the processing status data stream, data preprocessing is performed to form the original time series data stream; The original time series data stream is segmented and sliced ​​based on a sliding time window to generate multiple continuous time series state subsets. Multidimensional statistical features and dynamic change features are extracted from the multiple continuous time series state subsets to construct a high-dimensional state feature matrix; Feature fusion is performed based on the high-dimensional state feature matrix to generate the time series processing state set.

3. The adaptive control method for mold processing based on processing state feedback as described in claim 1, characterized in that, Generate the target processing parameter distribution, including: Based on the parameter transfer function, multiple candidate parameter sets are output, and a corresponding parameter candidate space is constructed; For the first candidate parameter set, the cutting load response, vibration response and machining efficiency index are simulated and predicted to obtain the first simulation prediction result; Based on the first simulation prediction results, a multi-dimensional evaluation index system is constructed, including stability index, efficiency index and tool life index. The multidimensional evaluation index system is input into the comprehensive evaluation function to score the first candidate parameter set and output the simulation evaluation result and the first probability distribution of the first parameter. The target processing parameter distribution is constructed based on the simulation evaluation results of the first parameter and the first probability distribution. The target processing parameter distribution includes multiple mold processing areas.

4. The adaptive control method for mold processing based on processing status feedback as described in claim 3, characterized in that, Extracting target parameter combinations from the target processing parameter distribution includes: In the target processing parameter distribution, multi-level constraint screening is performed based on multiple mold processing requirements of multiple mold processing areas. The optimal parameter combination is extracted from the screening results as the target parameter combination, while several alternative parameter combinations are retained as a dynamic adjustment candidate set. The multi-level constraint screening includes: S1: Eliminate infeasible parameter sets based on machine tool power constraints and tool strength constraints; S2: Screening a set of stable parameters based on vibration stability constraints and machining accuracy constraints; S3: Perform multi-objective optimization ranking based on the trade-off between machining efficiency and tool life.

5. The adaptive control method for mold processing based on processing status feedback as described in claim 4, characterized in that, Also includes: Based on the target parameter combination, the second tool is driven to perform machining on the area to be machined in the mold. The second machining condition state vector is acquired in real time, and the machining condition state deviation between the second machining condition state vector and the target state domain is calculated. When the deviation of the processing condition exceeds a preset deviation threshold, a candidate parameter combination is selected from the dynamic adjustment candidate set. The current processing parameters are gradually adjusted based on the alternative parameter combinations.

6. The adaptive control method for mold processing based on processing state feedback as described in claim 1, characterized in that, Constructing cross-tool mapping relationships using parameter transfer functions includes: The differences in geometric parameters, number of cutting edges, and cutting characteristics between the first and second cutting tools at different stages of mold processing are obtained. Based on the differences in the geometric parameters, a proportional mapping relationship between feed rate and cutting width is established. Based on the differences in the number of cutting edges and cutting characteristics of the aforementioned tools, an equivalent conversion relationship for the unit cutting edge load is established. Based on the correlation between cutting thickness and material removal rate, a cutting depth compensation relationship is constructed; Based on the aforementioned proportional mapping relationship, the equivalent conversion relationship of unit cutting edge load, and the cutting depth compensation relationship, the output result of the parameter migration function is corrected to obtain a parameter mapping result adapted to the second tool.

7. The adaptive control method for mold processing based on processing state feedback as described in claim 1, characterized in that, During the second tool machining process, a second machining condition state vector is constructed in real time, and the machining condition state deviation is calculated based on the mold machining target state domain, including: During the second tool machining process, the machining status data stream is continuously updated, and a dynamic data buffer is constructed; Based on the dynamic data buffer, continuous state data segments are extracted using a sliding window method. The continuous state data segment is subjected to feature extraction and encoding to generate a second processing condition state vector; Multidimensional matching is performed between the second processing condition state vector and the preset target state domain, and the processing condition state deviation is calculated based on the multidimensional matching results.

8. The adaptive control method for mold processing based on processing state feedback as described in claim 1, characterized in that, Continuous adjustment of the target parameter combination based on the deviation of the processing condition includes: Based on the deviation of the machining condition, a control adjustment amount is constructed. The control adjustment amount includes multiple adjustment directions and adjustment amplitudes of multiple deviation components. The multiple deviation components include feed rate, spindle speed and depth of cut. The feed rate, spindle speed and depth of cut are adjusted in a multi-variable coordinated manner, and constraints are introduced during the coordinated adjustment process to limit the rate of change of parameters.

9. The adaptive control method for mold processing based on processing state feedback as described in claim 1, characterized in that, include: In closed-loop adaptive control of mold processing, data of the entire processing process is recorded. The data of the entire processing process includes processing state sequence, parameter migration input and output, parameter adjustment trajectory and processing quality results. The data from the entire processing process are structured and organized to construct a training sample dataset; The parameter transfer function is updated based on the training sample dataset, and the updated parameter transfer function is deployed for subsequent tool switching processes.

10. A mold processing adaptive control system based on processing status feedback, characterized in that, For implementing the mold processing adaptive control method based on processing status feedback as described in any one of claims 1-9, the system comprises: The machining state construction module is used to continuously collect machining state data streams during the machining process of the first tool on the area to be machined of the mold, and construct a time series machining state set. The machining state data streams include cutting load, vibration response, tool wear state and machining stability index. The state encoding module is used to perform feature compression and state encoding on the time series processing state set to obtain the first processing condition state vector; The constraint information acquisition module is used to acquire the tool structure parameter vector of the second tool, as well as the geometric and process constraint information of the current mold processing area; The mapping relationship construction module is used to construct cross-tool mapping relationships based on the machining condition state vector, tool structure parameter vector, geometric and process constraint information, and in combination with parameter migration functions, and generate the target machining parameter distribution. The machining execution module is used to extract a target parameter combination from the target machining parameter distribution based on stability constraints and efficiency constraints, and drive the second tool to perform machining on the mold area to be machined based on the target parameter combination. The state deviation calculation module is used to construct the second machining condition state vector in real time during the machining process of the second tool, and calculate the machining condition state deviation based on the target state domain of mold machining. The continuous adjustment module is used to continuously adjust the target parameter combination based on the deviation of the processing condition, forming a closed-loop adaptive control for mold processing.

Images