Intelligent control method and system for CNC machine tools used in machining aviation titanium alloy structural parts

Abstract

The present application relates to the technical field of machine tool control, and in particular to an intelligent control method and system for CNC machine tools used for machining aviation titanium alloy structural parts. The present application first collects three types of working condition data: cutting force, temperature, and vibration, and then extracts the working condition feature vector to input into the pre-training model to obtain the optimal combination value of feed speed and spindle speed, and finally adjusts the machine tool trajectory in real time based on the optimized parameters; the continuous optimization of the deep learning model improves the adaptability to complex working conditions and improves the machining accuracy and surface quality of aviation titanium alloy structural parts.

Description

Technical Field

[0001] The present application relates to the technical field of machine tool control, and in particular to an intelligent control method and system for CNC machine tools used in machining aviation titanium alloy structural parts. Background Art

[0002] Aviation titanium alloy structural parts are used in the aerospace field due to their high specific strength, excellent corrosion resistance and good high temperature performance. However, due to the poor thermal conductivity and high chemical activity of titanium alloy materials, problems such as excessive cutting temperature and rapid tool wear are easily generated during processing, which places extremely high demands on processing accuracy and surface quality.

[0003] At present, when machining aviation titanium alloy structural parts, CNC machine tools mainly use preset fixed machining parameters for control, collect machining process data through real-time monitoring systems, adjust machining parameters based on preset rules, and achieve adaptive control. However, the existing control methods are difficult to cope with complex and changeable machining conditions, and cannot accurately predict the optimal parameter combination at different machining stages, resulting in unstable cutting conditions during machining, affecting machining accuracy and efficiency. This situation needs to be further improved. Summary of the invention

[0004] In order to solve the problem that the existing CNC machine tool control method is difficult to cope with complex and changeable processing conditions, resulting in unstable cutting state during processing, affecting processing accuracy and efficiency, the present application provides a CNC machine tool intelligent control method and system for processing aviation titanium alloy structural parts, adopting the following technical solutions:

[0005] In a first aspect, the present application provides an intelligent control method for a CNC machine tool for machining aviation titanium alloy structural parts, comprising the following steps:

[0006] Collect cutting force, temperature and vibration data during machining to obtain real-time working condition data;

[0007] Extracting cutting process characteristic parameters according to the real-time working condition data to obtain working condition characteristic vectors;

[0008] Based on the working condition characteristic vector, the optimal machining parameters are calculated through a pre-trained deep neural network model to obtain an optimized combination value of the feed speed and the spindle speed;

[0009] According to the optimized combination value, the motion trajectory of the CNC machine tool is adjusted in real time to obtain a compensated processing path;

[0010] Based on the compensated machining path, the tool is controlled to perform a machining operation.

[0011] By adopting the above technical scheme, the prior art usually adopts preset fixed processing parameters for CNC machining, but due to the poor thermal conductivity and high chemical activity of titanium alloy materials, problems such as sudden changes in cutting temperature and fluctuations in cutting force often occur in the actual machining process. For example, when machining a certain type of engine blade, even if the same processing parameters are used, there are still significant differences in the machining quality of different areas, especially in the transition section of complex curved surfaces, which are prone to defects such as excessive surface roughness and local burns. The present application first collects three types of working condition data: cutting force, temperature and vibration, and then extracts the working condition feature vector and inputs it into the pre-trained model to obtain the optimized combination value of feed speed and spindle speed, and finally adjusts the machine tool trajectory in real time based on the optimized parameters. The continuous optimization of the deep learning model improves the adaptability to complex working conditions, and significantly improves the machining accuracy and surface quality of aviation titanium alloy structural parts.

[0012] Optionally, according to the real-time working condition data, the cutting process characteristic parameters are extracted to obtain the working condition characteristic vector, which specifically includes the following steps:

[0013] Dividing the real-time working condition data into a cutting entry segment, a stable cutting segment and a cutting exit segment according to the cutting stage;

[0014] Performing mean filtering on the data of the stable cutting segment to obtain filtered steady-state data;

[0015] Calculating the root mean square value of cutting force, temperature and vibration according to the filtered steady-state data to obtain characteristic parameters;

[0016] Based on the characteristic parameters, a working condition characteristic vector is constructed.

[0017] By adopting the above technical scheme, since the working condition data in the processing of aviation titanium alloy structural parts contains a large amount of transient fluctuations and noise interference, directly using the original data for feature extraction will lead to inaccurate parameter optimization. For example, the violent fluctuations in the feed and retract stages are confused with the data in the stable cutting stage. The application first divides the processing process into three stages: feed segment, stable cutting segment and retract segment, and then applies mean filtering to the stable cutting segment to eliminate random fluctuations, and then calculates the root mean square value of the filtered data to obtain feature parameters, and finally constructs a working condition vector containing cutting force, temperature and vibration characteristics. The phased processing strategy effectively avoids the influence of transient interference, and the root mean square feature extraction method is used to improve the stability of feature expression, which significantly improves the accuracy of processing parameter optimization.

[0018] Optionally, the deep neural network model includes a first pre-trained deep neural network model and a second pre-trained deep neural network model; based on the working condition feature vector, the optimal machining parameters are calculated by the pre-trained deep neural network model to obtain the optimized combination value of the feed speed and the spindle speed, which specifically includes the following steps:

[0019] According to the geometric features of the processing area, the working condition feature vector is divided into a straight line processing segment and a curved surface processing segment;

[0020] Inputting the working condition feature vector corresponding to the straight line processing segment into a first pre-trained deep neural network model to obtain a parameter prediction value of the straight line processing segment;

[0021] Inputting the working condition feature vector corresponding to the curved surface processing section into a second pre-trained deep neural network model to obtain a parameter prediction value of the curved surface processing section;

[0022] Based on the predicted parameter values, the optimal combination values ​​of the feed speed and the spindle speed are generated in sections, and a transition interval is set at the junction of the processing sections.

[0023] By adopting the above technical scheme, when a unified model is used for parameter prediction, problems such as too slow feeding of straight segments or too heavy cutting of curved segments often occur, especially at the transition position between straight segments and curved segments, where processing marks are easily generated; the present application first divides the working condition feature vector into straight line processing segments and curved surface processing segments according to geometric features, and then respectively inputs the corresponding pre-trained models to obtain parameter prediction values, and finally realizes smooth switching between different processing segments by setting transition intervals; it solves the problem that a single model is difficult to adapt to multiple types of processing features, and ensures the continuous change of processing parameters through the design of transition intervals, which significantly improves the overall processing quality of aviation titanium alloy structural parts, especially improves the processing effect of the feature transition area.

[0024] Optionally, according to the geometric features of the processing area, the working condition feature vector is divided into a straight line processing segment and a curved surface processing segment, which specifically includes the following steps:

[0025] Based on the trajectory data of the machining program, the curvature change rate between adjacent tool positions is calculated to obtain the original curvature change rate data;

[0026] According to the original curvature change rate data, a smoothing process is performed through a sliding window to obtain a smoothed curvature change rate;

[0027] Comparing the curvature change rate after the smoothing process with a preset discrimination threshold and a minimum segment length threshold to determine a candidate segmentation area;

[0028] According to the candidate segmentation regions, the continuity and length characteristics of each segment are calculated to obtain a preliminary segmentation result;

[0029] Based on the preliminary segmentation result and the topological relationship of the machining features, error compensation is performed to obtain the final working condition feature vector segmentation result.

[0030] By adopting the above technical scheme, due to the complex and changeable geometric characteristics of aviation titanium alloy structural parts, when dividing the processing segments, relying solely on simple curvature threshold judgment can easily cause inaccurate segmentation results or excessive segmentation fragmentation; the application first calculates the curvature change rate of the trajectory and performs sliding window smoothing, and then determines the candidate segmentation area based on the preset threshold, and then analyzes the continuity and length characteristics of each segment to obtain the preliminary segmentation results, and finally performs error compensation based on the topological relationship to obtain the final segmentation scheme; the reliability of the segmentation results and the consistency of the processing quality are improved.

[0031] Optionally, according to the candidate segmentation regions, the continuity and length features of each segment are calculated to obtain a preliminary segmentation result, which specifically includes the following steps:

[0032] Based on the tool position coordinate sequence, the direction angle change between adjacent points is calculated to obtain the angle change sequence;

[0033] Cumulatively summing the angle change sequence to obtain a cumulative value of direction change;

[0034] Determining the geometric type of each segment according to a comparison result of the cumulative value of the direction change and a set threshold;

[0035] Based on the length of the continuous segment of the geometric type, an adaptive weight coefficient is set, wherein when the length of the continuous segment is greater than a preset length threshold, the weight coefficient increases logarithmically with the segment length, and when the length of the continuous segment is less than the preset length threshold, the weight coefficient decreases linearly with the segment length;

[0036] The segmentation boundary is locally adjusted according to the weight coefficient to obtain a preliminary segmentation result, wherein the segmentation with a weight coefficient less than a preset weight threshold is merged with the adjacent segment.

[0037] By adopting the above technical scheme, the determination of the boundaries of each segment during the processing of aviation titanium alloy structural parts directly affects the optimization effect of subsequent processing parameters. The application first calculates the change in the angular orientation between the tool positions and accumulates the sum, then determines the geometric type according to the change trend, and then sets the adaptive weight coefficient based on the length of the continuous segment. The long segment adopts a logarithmic growth weight to maintain integrity, and the short segment adopts a linear reduction weight to promote merging. Finally, the segment merging optimization is performed according to the weight threshold. The rationality of the segmentation results is improved by differentiated processing of long and short segments, especially while maintaining the integrity of key features, effectively reducing fragmented segmentation.

[0038] Optionally, according to the optimized combination value, the motion trajectory of the CNC machine tool is adjusted in real time to obtain a compensated processing path, which specifically includes the following steps:

[0039] Calculate the difference between the optimized feed speed and spindle speed and the current machining parameters to obtain the parameter adjustment amount;

[0040] According to the magnitude of the parameter adjustment amount, the trajectory adjustment is divided into a rapid adjustment area and a gradual adjustment area;

[0041] In the fast adjustment area, the motion trajectory is adjusted directly according to the optimized combination value;

[0042] In the progressive adjustment area, the motion trajectory is gradually adjusted by piecewise linear interpolation;

[0043] The trajectories of the rapid adjustment area and the gradual adjustment area are combined to obtain a compensated processing path.

[0044] By adopting the above technical solution, the processing parameters need to be adjusted in real time during the processing of aviation titanium alloy structural parts. The traditional parameter synchronization adjustment method is prone to cause unstable machine tool motion or over-response. The present application first calculates the difference between the optimized parameters and the current parameters, and then determines the adjustment area type according to the size of the difference. For the rapid adjustment area with a large difference, the target parameters are directly used. For the gradual adjustment area with a small difference, a piecewise linear interpolation method is used to gradually adjust, and finally the trajectories of the two areas are seamlessly combined. Through differentiated adjustment strategies, both processing efficiency and process stability are guaranteed, thereby improving the processing quality and efficiency of aviation titanium alloy structural parts.

[0045] In the second aspect, the present application provides an intelligent control system for CNC machine tools for machining aviation titanium alloy structural parts, including:

[0046] Real-time working condition data acquisition module, used to collect cutting force, temperature and vibration data during the processing to obtain real-time working condition data;

[0047] A working condition characteristic vector acquisition module is used to extract cutting process characteristic parameters according to the real-time working condition data to obtain a working condition characteristic vector;

[0048] An optimized combination value acquisition module is used to calculate the optimal processing parameters based on the working condition feature vector through a pre-trained deep neural network model to obtain an optimized combination value of the feed speed and the spindle speed;

[0049] A machine tool motion trajectory adjustment module is used to adjust the motion trajectory of the CNC machine tool in real time according to the optimized combination value to obtain a compensated processing path;

[0050] The tool control module is used to control the tool to perform a machining operation based on the compensated machining path.

[0051] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned intelligent control method for CNC machine tools for machining aviation titanium alloy structural parts when executing the computer program.

[0052] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned intelligent control method for CNC machine tools for machining aviation titanium alloy structural parts.

[0053] In summary, the present application includes at least one of the following beneficial technical effects:

[0054] This application first collects three types of working condition data: cutting force, temperature and vibration, then extracts the working condition feature vector and inputs it into the pre-trained model to obtain the optimized combination value of feed speed and spindle speed, and finally adjusts the machine tool trajectory in real time based on the optimized parameters; through the continuous optimization of the deep learning model, the adaptability to complex working conditions is improved, and the processing accuracy and surface quality of aviation titanium alloy structural parts are significantly improved;

[0055] Since the working condition data during the machining of aviation titanium alloy structural parts contains a large amount of transient fluctuations and noise interference, directly using the original data for feature extraction will lead to inaccurate parameter optimization. For example, the violent fluctuations in the feed and retract stages are confused with the data in the stable cutting stage. The present application first divides the machining process into three stages: feed stage, stable cutting stage and retract stage. Then, mean filtering is applied to the stable cutting stage to eliminate random fluctuations. Then, the root mean square value of the filtered data is calculated to obtain the characteristic parameters. Finally, a working condition vector containing cutting force, temperature and vibration characteristics is constructed. The influence of transient interference is effectively avoided through a staged processing strategy. The root mean square feature extraction method is used to improve the stability of feature expression and significantly improve the accuracy of machining parameter optimization.

[0056] When a unified model is used for parameter prediction, problems such as slow feeding of straight segments or heavy cutting of curved segments often occur, especially at the transition position between straight segments and curved segments, where processing marks are easily generated. The present application first divides the working condition feature vector into straight line processing segments and curved surface processing segments according to geometric features, and then inputs the corresponding pre-trained models to obtain parameter prediction values, and finally achieves smooth switching between different processing segments by setting transition intervals. The problem that a single model is difficult to adapt to multiple types of processing features is solved, and the continuous change of processing parameters is ensured through the design of transition intervals, which significantly improves the overall processing quality of aviation titanium alloy structural parts, especially improves the processing effect of the feature transition area. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a flow chart of an intelligent control method of a CNC machine tool for machining aviation titanium alloy structural parts according to an embodiment of the present application;

[0058] Figure 2It is a flow chart of step S200 in a method for intelligently controlling a CNC machine tool for machining aviation titanium alloy structural parts according to an embodiment of the present application;

[0059] Figure 3 It is a flow chart of step S300 in a method for intelligently controlling a CNC machine tool for machining aviation titanium alloy structural parts according to an embodiment of the present application;

[0060] Figure 4 It is a flow chart of step S310 in a method for intelligently controlling a CNC machine tool for machining aviation titanium alloy structural parts according to an embodiment of the present application;

[0061] Figure 5 It is a flow chart of step S314 in a method for intelligently controlling a CNC machine tool for machining aviation titanium alloy structural parts according to an embodiment of the present application;

[0062] Figure 6 It is a flow chart of step S400 in a method for intelligently controlling a CNC machine tool for machining aviation titanium alloy structural parts according to an embodiment of the present application;

[0063] Figure 7 This is a module schematic diagram of an intelligent control system for a CNC machine tool used for machining aviation titanium alloy structural parts according to an embodiment of the present application;

[0064] Figure 8 It is a diagram of the internal structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0065] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification and appended claims of the present application, the singular expressions "one", "a kind of", "said", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more listed items.

[0066] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0067] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0068] In the first aspect, the present application provides an intelligent control method for a CNC machine tool for machining aviation titanium alloy structural parts, referring to Figure 1 , including the following steps:

[0069] S100, collecting cutting force, temperature and vibration data during the machining process to obtain real-time working condition data.

[0070] Among them, the aviation titanium alloy structural parts in this embodiment mainly include typical parts such as engine casings, blades, landing gear, etc. These parts have the characteristics of thin walls, large sizes, and complex curved surfaces, and have high requirements for processing accuracy and surface quality. The CNC machine tool is a five-axis linkage machining center equipped with a high-precision spindle system and a high-response servo system, which can achieve efficient processing of complex curved surfaces. The cutting force is collected by a three-axis dynamometer installed on the tool handle, the temperature data is measured non-contact by an infrared thermal imager, and the vibration signal is collected by a combination of an acceleration sensor and a displacement sensor.

[0071] Specifically, during the machining process, the synchronous collection of multi-source information is achieved by rationally arranging the sensor network on the tool system. The cutting force signal collects the component data in three directions through the dynamometer, the temperature measurement focuses on monitoring the temperature distribution in the tool edge area, and the vibration signal collects acceleration and displacement information at the same time. These signals are transmitted to the controller in real time through the data acquisition system. For example, when machining a certain type of titanium alloy casing, this monitoring solution can effectively capture the changing characteristics of physical quantities such as force, heat, and vibration during the cutting process, providing reliable data support for subsequent intelligent control.

[0072] S200, extracting cutting process characteristic parameters according to real-time working condition data to obtain working condition characteristic vectors.

[0073] In this embodiment, a signal analysis method is used to extract features from real-time working condition data. The characteristic parameters include the mean and fluctuation characteristics of the cutting force, the temperature change trend characteristics, and the amplitude characteristics of the vibration. The working condition feature vector is constructed by combining these characteristic parameters to characterize the stability of the current processing state.

[0074] Specifically, the collected original signal is first subjected to noise reduction and smoothing to remove interference components. Then the characteristic values ​​of the cutting force, temperature and vibration signals are calculated, and these characteristics are combined into characteristic vectors according to preset rules. For example, in the processing of thin-walled casing parts, by analyzing the fluctuation characteristics of the cutting force and the changing trend of the vibration amplitude, abnormal changes in the cutting state can be identified in a timely manner.

[0075] S300, based on the working condition feature vector, the optimal machining parameters are calculated through the pre-trained deep neural network model to obtain the optimized combination value of the feed speed and the spindle speed.

[0076] In this embodiment, a deep neural network model is used to analyze the working condition feature vector and establish a mapping relationship between the feature vector and the optimal processing parameters. The deep neural network model is pre-trained with a large amount of offline test data and can output the optimized feed speed and spindle speed in real time according to the current working condition characteristics.

[0077] Specifically, the working condition feature vector is input into the pre-trained deep neural network model, and the model automatically calculates the optimal feed rate and spindle speed combination according to the current processing status. For example, when a sudden increase in cutting force is detected, the model will output a lower feed rate to ensure the stability of the processing process.

[0078] S400, according to the optimized combination value, the motion trajectory of the CNC machine tool is adjusted in real time to obtain a compensated processing path.

[0079] In this embodiment, the motion trajectory of the machine tool is compensated in real time based on the optimized feed speed and spindle speed. The compensation process takes into account the stability of parameter adjustment to avoid drastic changes that affect the processing quality.

[0080] Specifically, different adjustment strategies are adopted according to the difference between the optimized parameters and the current parameters. When the parameter changes greatly, a segmented adjustment method is adopted, and when the parameter changes slightly, a continuous and gradual adjustment is performed. For example, at the corner of the machining surface, the system will steadily reduce the feed speed and return to the normal level after the transition is completed.

[0081] S500: Control the tool to perform a machining operation based on the compensated machining path.

[0082] In this embodiment, the compensated machining path is converted into tool motion instructions through a numerical control system to achieve precise control of the tool motion.

[0083] Specifically, the control system converts the compensated path data into motion instructions for each axis and drives the tool to perform processing through the servo system. For example, in the processing of a certain type of titanium alloy structural part, the use of this control method can control the deviation between the actual processing trajectory and the theoretical trajectory within the required range, ensuring the processing quality.

[0084] In one embodiment, referring to Figure 2 In step S200, according to the real-time working condition data, the cutting process characteristic parameters are extracted to obtain the working condition characteristic vector, which specifically includes the following steps:

[0085] S210, dividing the real-time working condition data into a cutting entry segment, a stable cutting segment and a cutting exit segment according to the cutting stage.

[0086] In this embodiment, the working condition data is processed in sections according to the physical characteristics of the cutting process. The cutting force rises rapidly in the cutting section, the stable cutting section is a stage where all parameters are relatively stable, and the cutting force drops rapidly in the retracting section.

[0087] Specifically, different cutting stages are identified by analyzing the changing trend of the cutting force signal. When the cutting force exceeds the set threshold and shows an upward trend, it is determined to have entered the cutting stage; when the cutting force is basically stable, it is determined to have entered the stable cutting stage; when the cutting force begins to continuously decrease, it is determined to have entered the cutting stage.

[0088] S220, performing mean filtering on the data of the stable cutting segment to obtain filtered steady-state data.

[0089] In this embodiment, the mean filtering method is used to process the data of the stable cutting segment to eliminate the influence of random fluctuations. The mean filtering realizes data smoothing by calculating the average value of the data in the sliding window, and can effectively extract the steady-state cutting features.

[0090] Specifically, an appropriate sliding window length is set to filter the force, heat and vibration data of the stable cutting segment. The filter window moves sequentially on the data sequence, and the mean of the data in the window is calculated as the filtering result.

[0091] S230. Calculate the root mean square value of the cutting force, temperature and vibration according to the filtered steady-state data to obtain characteristic parameters.

[0092] In this embodiment, the root mean square value is selected as the characteristic parameter to characterize the processing state. The root mean square value can reflect the energy level of the signal and has a good indication effect on the stability of the processing process.

[0093] Specifically, the root mean square values ​​of the filtered cutting force, temperature and vibration signals are calculated respectively. For the cutting force, the root mean square values ​​of the force components in three directions are calculated; for the temperature and vibration signals, the root mean square values ​​of their time series are calculated.

[0094] S240. Construct a working condition characteristic vector based on the characteristic parameters.

[0095] In this embodiment, the calculated characteristic parameters are organized into a characteristic vector in a predetermined order.

[0096] In one embodiment, referring to Figure 3 The deep neural network model includes a first pre-trained deep neural network model and a second pre-trained deep neural network model; in step S300, based on the working condition feature vector, the optimal processing parameters are calculated by the pre-trained deep neural network model to obtain the optimal combination value of the feed speed and the spindle speed, which specifically includes the following steps:

[0097] S310. Divide the working condition feature vector into a straight line processing segment and a curved surface processing segment according to the geometric features of the processing area.

[0098] In this embodiment, the working condition feature vectors are classified by analyzing the geometric features of the processing path. The straight processing segment refers to the area where the tool moves along the straight trajectory, and the curved surface processing segment refers to the area where the tool moves along the curved surface trajectory.

[0099] Specifically, the straight line segment and the curved surface segment are identified according to the path information in the processing program. When the path curvature is less than a preset threshold, it is determined to be a straight line processing segment; when the path curvature is greater than a preset threshold, it is determined to be a curved surface processing segment. For example, in the processing of the casing wall, the plane area can be divided into a straight line processing segment, and the chamfer and transition area can be divided into a curved surface processing segment.

[0100] S320. Input the working condition feature vector corresponding to the straight line processing segment into the first pre-trained deep neural network model to obtain the parameter prediction value of the straight line processing segment.

[0101] The first pre-trained deep neural network model in this embodiment is specifically used to process the eigenvectors of the straight line processing segment. The model is trained by a large amount of straight line processing condition data and has a strong ability to predict straight line processing parameters. The model first collects straight line processing test data, including cutting force, temperature and vibration data under different feed speed and spindle speed combinations. These data are then cleaned and standardized, and divided into training sets and validation sets. During the training process, an optimizer is used and a learning rate decay strategy is used. Through multiple rounds of iterative training, the model training is completed when the loss function value on the validation set tends to be stable. For example, in the straight line milling test of a certain type of titanium alloy plate, the model obtained by training 2000 sets of condition data can accurately predict the optimal combination of processing parameters. Furthermore, for different aviation titanium alloy structural parts, specific model training strategies are set. According to the material properties and processing difficulty of structural parts, the training data are divided into high-strength titanium alloy group and medium-strength titanium alloy group, and corresponding training data sets are constructed respectively. For high-strength titanium alloy structural parts, more low-speed and heavy-load cutting condition data are collected for model training to enhance the model's prediction ability under heavy-load conditions; for medium-strength titanium alloy structural parts, the focus is on collecting high-speed and light-load cutting condition data.

[0102] Specifically, the working condition feature vector of the straight line processing segment is input into the first pre-trained model, and the model predicts the feed speed and spindle speed suitable for straight line processing based on the feature vector. For example, in the plane processing of a certain type of titanium alloy part, the model can automatically adjust the processing parameters according to the cutting state to maintain the processing efficiency.

[0103] Furthermore, the system acquires the tool flank wear width VB value, cutting force spectrum characteristics and vibration signal RMS value in real time through the high-frequency sampling system to construct tool health status evaluation indicators; when it is detected that the tool wear exceeds the safety threshold, the model will automatically adjust the cutting parameters to extend the tool life. At the same time, the model pre-establishes a material microstructure characteristic database, which includes characteristic parameters such as grain distribution, anisotropy coefficient and hardened layer depth in different processing areas, and dynamically adjusts the cutting strategy based on these parameters, especially in sensitive areas of material organization, by controlling the cutting temperature and strain rate to avoid the formation of undesirable organizations.

[0104] S330. Input the working condition feature vector corresponding to the curved surface processing segment into the second pre-trained deep neural network model to obtain the parameter prediction value of the curved surface processing segment.

[0105] The second pre-trained deep neural network model in this embodiment is specially designed for surface processing features, taking into account the curvature change factor in surface processing, and can more accurately predict the processing parameters required for surface processing.

[0106] Specifically, the model pre-establishes a calculation model for the instantaneous contact arc length between the tool and the workpiece, discretizes the surface into several micro-elements, and obtains the contact arc length and actual cutting depth of each discrete point based on the interference calculation between the tool envelope surface and the workpiece surface; the stress distribution in different curvature areas is predicted based on the model to identify potential high wear risk areas. Then, the cutting process signal is collected, the frequency domain features are extracted through fast Fourier transform, and the amplitude changes of the fundamental frequency and its frequency multiple components related to the tool speed are analyzed; at the same time, the vibration signal of the acceleration sensor is collected, and the time-frequency analysis is performed through wavelet transform to extract energy entropy. The system performs correlation analysis on the frequency domain features of the contact arc length, actual cutting depth, force signal, and the time-frequency features of the vibration signal to establish a local wear prediction model for the tool. For example, when the vibration energy entropy of a certain area exceeds the preset threshold, and the position coincides with the predicted high stress area, it is determined to be a local wear aggravation area; then, according to the wear prediction results, the feed speed is automatically reduced in the area, and the original feed speed is maintained in the low-risk area to achieve differentiated parameter adjustment. Similarly, a higher material removal rate is maintained in low-risk areas, thereby achieving a dynamic balance between machining efficiency and tool life.

[0107] S340, based on the parameter prediction value, generate the optimal combination value of the feed speed and the spindle speed in segments, and set a transition interval at the junction of the processing segments.

[0108] In this embodiment, the prediction results of the two models are integrated to generate a complete optimization scheme for machining parameters. A transition interval is set at the junction of the straight line segment and the curved surface segment to ensure a smooth transition of the machining parameters and avoid the influence of sudden changes on the machining quality.

[0109] Specifically, the processing parameters of the straight line segment and the curved surface segment are generated respectively according to the predicted values ​​of the two models. A transition interval of appropriate length is set between the segments, and the interpolation algorithm is used to achieve the gradual transition of the parameters.

[0110] In one embodiment, referring to Figure 4 In step S310, the working condition feature vector is divided into a straight line processing segment and a curved surface processing segment according to the geometric features of the processing area, which specifically includes the following steps:

[0111] S311. Based on the trajectory data of the machining program, the curvature change rate between adjacent tool position points is calculated to obtain original curvature change rate data.

[0112] In this embodiment, the curvature change information is obtained by analyzing the tool trajectory data in the machining program using a numerical calculation method.

[0113] Specifically, the three-point method is used to calculate the curvature at each tool position, and then the curvature change rate between adjacent points is calculated. For each tool position, a point before and after it is selected to form a calculation unit, and the curvature value is obtained by solving the differential equation.

[0114] S312: Perform sliding window smoothing on the original curvature change rate data to obtain a smoothed curvature change rate.

[0115] In this embodiment, the sliding window method is used to smooth the original curvature change rate data. By setting a suitable window length and weight coefficient, local fluctuations in the data are eliminated and the main curvature change trend is highlighted.

[0116] Specifically, a sliding window of length N is set, and the smoothing value is calculated by weighted average in the window. The window moves point by point on the data sequence, and the data at each position is smoothed. For example, in the machining trajectory analysis of the casing rib plate, a sliding window of 15 points can effectively eliminate the curvature fluctuation caused by interpolation error.

[0117] S313: Compare the curvature change rate after smoothing with a preset discrimination threshold and a minimum segment length threshold to determine a candidate segmentation region.

[0118] In this embodiment, two key parameters are set: a curvature change rate discrimination threshold and a minimum processing segment length threshold. The discrimination threshold is used to distinguish between straight line segments and curved surface segments, and the minimum segment length threshold is used to avoid excessive segmentation and ensure the continuity of processing.

[0119] Specifically, the smoothed curvature change rate is compared with the discrimination threshold to preliminarily mark possible segmentation points. Then check whether the distance between adjacent segmentation points meets the minimum segment length requirement and merge the segments that are too short. For example, in the processing of a certain type of bracket parts, the boundary point between the plane area and the transition area can be accurately identified by setting a reasonable threshold.

[0120] S314. Calculate the continuity and length features of each segment based on the candidate segmentation regions to obtain a preliminary segmentation result.

[0121] In this embodiment, feature analysis is performed on the candidate segmented regions to determine the continuity between the segments and the length distribution features within the segments.

[0122] Specifically, the tangential continuity at the endpoints of each segment is calculated to evaluate the transition characteristics between adjacent segments. At the same time, the length distribution of each segment is counted to identify unreasonable short or long segments.

[0123] S315. Based on the preliminary segmentation result and the topological relationship of the processing features, error compensation is performed to obtain the final working condition feature vector segmentation result.

[0124] In this embodiment, an error compensation model based on topological relationships is pre-constructed. The model first establishes a feature adjacency graph to describe the connection relationship between processing features; then sets compensation weights based on feature types, where the weight coefficient of the structural transition area is higher, which is used to strengthen the adjustment of the segmentation points in the transition area; finally, the segmentation point positions are fine-tuned through an iterative optimization method.

[0125] Specifically, based on the feature model information of the parts, identify the continuous feature areas that need to be processed uniformly, and adjust the segmentation results in these areas. For the segmentation points at the junction of features, the position is corrected according to the topological relationship. First, the topological relationship of the features is extracted, and a feature connection matrix is ​​established. The matrix elements represent the connection strength between the features. The connection strength is determined according to the feature type (such as plane, cavity, rib plate, etc.) and the transition method (such as tangent transition, arc transition, etc.); secondly, calculate the distance between each segmentation point and the feature boundary. When the distance is less than the preset threshold, adjust the segmentation point position according to the feature connection strength, and the adjustment amount decays exponentially with the increase of distance; finally, perform local smoothing on the adjusted segmentation points to ensure the continuity of the segmentation boundary.

[0126] In one embodiment, referring to Figure 5 In step S314, the continuity and length features of each segment are calculated according to the candidate segmentation regions to obtain a preliminary segmentation result, which specifically includes the following steps:

[0127] S3141. Based on the tool position point coordinate sequence, calculate the direction angle change between adjacent points to obtain the angle change sequence.

[0128] In this embodiment, a vector analysis method is used to calculate the direction change between tool position points, and the change in direction angle is calculated through the vectors formed by adjacent tool position points.

[0129] Specifically, for any three adjacent tool position points, the angle between the two vectors formed by these points is calculated. The direction angle is calculated using the inverse tangent function, and the quadrant judgment is considered to obtain complete angle information.

[0130] S3142. Cumulatively sum the angle change sequence to obtain a cumulative value of the direction change.

[0131] In this embodiment, the overall change trend of the trajectory is analyzed by cumulative summation. The change rate of the cumulative value can reflect the curvature characteristics of the trajectory, which is helpful for identifying different types of processing sections.

[0132] Specifically, starting from the starting point of the trajectory, the angle changes are accumulated in sequence to form a cumulative curve of direction changes. By analyzing the slope changes of the cumulative curve, the characteristic points of the trajectory can be identified. For example, in the processing trajectory of the casing wall, the straight line segment appears as a flat area of ​​the cumulative curve, while the curved surface segment appears as a steep area.

[0133] S3143. Determine the geometric type of each segment based on the comparison result between the cumulative value of the direction change and the set threshold.

[0134] In this embodiment, a threshold criterion of the cumulative value of the direction change is set to distinguish different geometric types. By comparing the relationship between the cumulative value and the threshold, the trajectory segment can be divided into different types such as straight line segment, transition segment and curved surface segment.

[0135] Specifically, when the change rate of the cumulative value is less than the lower limit threshold, it is determined to be a straight line segment; when the change rate is between the upper and lower limit thresholds, it is determined to be a transition segment; when the change rate exceeds the upper limit threshold, it is determined to be a curved surface segment.

[0136] S3144. Set an adaptive weight coefficient based on the length of the continuous segment of the geometry type.

[0137] When the length of the continuous segment is greater than the preset length threshold, the weight coefficient increases logarithmically with the segment length; when the length of the continuous segment is less than the preset length threshold, the weight coefficient decreases linearly with the segment length.

[0138] In this embodiment, an adaptive weight calculation model based on segment length is established. The model uses a piecewise function to describe the relationship between the weight coefficient and the segment length based on a preset length threshold.

[0139] Specifically, when the segment length exceeds the preset threshold, the weight coefficient increases according to a logarithmic function, reflecting a tendency to retain long segments; when the segment length is less than the preset threshold, the weight coefficient decreases according to a linear function, reflecting a suppression effect on short segments.

[0140] S3145. Locally adjust the segmentation boundary according to the weight coefficient to obtain a preliminary segmentation result.

[0141] Among them, the segments whose weight coefficients are less than the preset weight threshold are merged with the adjacent segments.

[0142] In this embodiment, the merging process of the segments is controlled by a weight threshold. When the weight coefficient of the segment is lower than the preset threshold, the merging operation is triggered to merge the short segment into the adjacent main segment to optimize the overall segmentation result.

[0143] Specifically, the weight coefficient of each segment is compared with a preset threshold to identify the short segments that need to be merged. In the merging process, adjacent segments with higher similarity to geometric features are preferentially selected for merging.

[0144] In one embodiment, referring to Figure 6 In step S400, according to the optimized combination value, the motion trajectory of the CNC machine tool is adjusted in real time to obtain a compensated processing path, which specifically includes the following steps:

[0145] S410, calculating the difference between the optimized feed speed and spindle speed and the current machining parameters to obtain the parameter adjustment amount.

[0146] In this embodiment, the processing parameters of the numerical control system are monitored in real time and compared with the parameters output by the optimization algorithm. A parameter difference calculation model is established to calculate the adjustment amounts of the feed speed and the spindle speed.

[0147] Specifically, the sliding time window method is used to collect the current processing parameters at fixed time intervals and calculate the difference with the optimized parameters. The difference calculation takes into account the change trend of the parameters and eliminates the influence of instantaneous fluctuations through weighted averaging.

[0148] S420 , dividing the trajectory adjustment into a fast adjustment area and a gradual adjustment area according to the size of the parameter adjustment amount.

[0149] In this embodiment, the system pre-sets a partition threshold for parameter adjustment to distinguish different adjustment strategies. When the parameter difference exceeds the preset threshold, it is divided into a fast adjustment zone; when the difference is small, it is divided into a gradual adjustment zone.

[0150] Specifically, for feed speed adjustment, when the difference exceeds 30% of the current value, it is divided into the rapid adjustment area; for spindle speed adjustment, when the difference exceeds 20% of the current value, it is divided into the rapid adjustment area. All other situations are divided into the gradual adjustment area.

[0151] S430. In the fast adjustment area, directly adjust the motion trajectory according to the optimized combination value.

[0152] In this embodiment, the processing parameters are updated by direct assignment in the fast adjustment area. In order to ensure the safety of the adjustment process, an upper limit constraint on the parameter change rate is set to prevent processing abnormalities caused by sudden changes in parameters.

[0153] Specifically, under the premise of ensuring that the parameter change rate does not exceed the safety threshold, the optimized parameters are directly written into the parameter register of the CNC system. At the same time, the real-time monitoring mechanism is started to observe the change trend of working condition signals such as cutting force and vibration.

[0154] S440: In the gradual adjustment area, the motion trajectory is gradually adjusted using a piecewise linear interpolation method.

[0155] In this embodiment, a piecewise linear interpolation algorithm is used in the gradual adjustment area to decompose the parameter adjustment process into a plurality of gradual processes with small steps. By controlling the interpolation step size and the update frequency, a smooth transition of the parameters is achieved to avoid drastic fluctuations during the processing.

[0156] Specifically, the interpolation step length is dynamically calculated based on the parameter difference and the current processing status. The adjustment process is generally divided into 5-10 steps. After each interpolation step is completed, the stability of the working condition signal is checked, and the interpolation parameters of the subsequent steps are adjusted in a timely manner based on the feedback results.

[0157] S450, combining the trajectories of the rapid adjustment area and the gradual adjustment area to obtain a compensated processing path.

[0158] In this embodiment, a buffer section is set at the junction of two areas, and a smooth transition of parameters is achieved through a cubic spline curve. The length of the buffer section is dynamically adjusted according to the parameter change amount, and is generally controlled within 1-2 interpolation cycles.

[0159] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0160] In the second aspect, the present application provides an intelligent control system for CNC machine tools for processing aviation titanium alloy structural parts. The intelligent control system for CNC machine tools for processing aviation titanium alloy structural parts of the present application is described below in combination with the above-mentioned intelligent control method for CNC machine tools for processing aviation titanium alloy structural parts.

[0161] Reference Figure 7 , an intelligent control system for CNC machine tools used for machining aviation titanium alloy structural parts, comprising:

[0162] Real-time working condition data acquisition module, used to collect cutting force, temperature and vibration data during the processing to obtain real-time working condition data;

[0163] A working condition characteristic vector acquisition module is used to extract cutting process characteristic parameters according to real-time working condition data to obtain a working condition characteristic vector;

[0164] The optimized combination value acquisition module is used to calculate the optimal processing parameters based on the working condition feature vector through the pre-trained deep neural network model to obtain the optimized combination value of the feed speed and the spindle speed;

[0165] The machine tool motion trajectory adjustment module is used to adjust the motion trajectory of the CNC machine tool in real time according to the optimized combination value to obtain the compensated processing path;

[0166] The tool control module is used to control the tool to perform machining operations based on the compensated machining path.

[0167] In one embodiment, the operating condition feature vector acquisition module is specifically used for:

[0168] The real-time working condition data is divided into the cutting stage, the stable cutting stage and the cutting stage according to the cutting stage;

[0169] Perform mean filtering on the data of the stable cutting section to obtain filtered steady-state data;

[0170] According to the filtered steady-state data, the root mean square values ​​of cutting force, temperature and vibration are calculated to obtain characteristic parameters;

[0171] Based on the characteristic parameters, the operating condition characteristic vector is constructed.

[0172] In one embodiment, the deep neural network model includes a first pre-trained deep neural network model and a second pre-trained deep neural network model; the optimization combination value acquisition module is specifically used to:

[0173] According to the geometric characteristics of the processing area, the working condition feature vector is divided into straight line processing segment and surface processing segment;

[0174] Inputting the working condition feature vector corresponding to the straight line processing segment into the first pre-trained deep neural network model to obtain the parameter prediction value of the straight line processing segment;

[0175] Inputting the working condition feature vector corresponding to the curved surface processing section into the second pre-trained deep neural network model to obtain the parameter prediction value of the curved surface processing section;

[0176] Based on the predicted parameter values, the optimal combination values ​​of feed speed and spindle speed are generated in sections, and transition intervals are set at the junctions of machining sections.

[0177] In one embodiment, the optimization combination value acquisition module is specifically used to divide the operating condition feature vector:

[0178] Based on the trajectory data of the machining program, the curvature change rate between adjacent tool positions is calculated to obtain the original curvature change rate data;

[0179] According to the original curvature change rate data, a smoothed curvature change rate is obtained by performing a sliding window smoothing process;

[0180] The curvature change rate after smoothing is compared with the preset discrimination threshold and the minimum segment length threshold to determine the candidate segmentation area;

[0181] According to the candidate segmentation regions, the continuity and length characteristics of each segment are calculated to obtain the preliminary segmentation results;

[0182] Based on the preliminary segmentation results and the topological relationship of the machining features, error compensation is performed to obtain the final working condition feature vector segmentation results.

[0183] In one embodiment, the optimization combination value acquisition module is specifically used to calculate the continuity and length characteristics of each segment:

[0184] Based on the tool position coordinate sequence, the direction angle change between adjacent points is calculated to obtain the angle change sequence;

[0185] Cumulatively sum the angle change sequence to obtain the cumulative value of direction change;

[0186] Determine the geometric type of each segment according to the comparison result between the cumulative value of the direction change and the set threshold;

[0187] Based on the length of the continuous segment of the geometric type, an adaptive weight coefficient is set, wherein when the length of the continuous segment is greater than a preset length threshold, the weight coefficient increases logarithmically with the segment length, and when the length of the continuous segment is less than the preset length threshold, the weight coefficient decreases linearly with the segment length;

[0188] The segment boundaries are locally adjusted according to the weight coefficient to obtain a preliminary segmentation result, wherein the segments whose weight coefficients are less than a preset weight threshold are merged with the adjacent segments.

[0189] In one embodiment, the machine tool motion trajectory adjustment module is specifically used to:

[0190] Calculate the difference between the optimized feed speed and spindle speed and the current machining parameters to obtain the parameter adjustment amount;

[0191] According to the size of parameter adjustment, the trajectory adjustment is divided into a fast adjustment area and a gradual adjustment area;

[0192] In the quick adjustment area, adjust the motion trajectory directly according to the optimized combination value;

[0193] In the gradual adjustment area, the motion trajectory is gradually adjusted using piecewise linear interpolation;

[0194] The trajectories of the rapid adjustment area and the progressive adjustment area are combined to obtain the compensated processing path.

[0195] In one embodiment, the present application provides an electronic device, which may be a server, and its internal structure diagram may be as follows: Figure 8 As shown. The electronic device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the electronic device is used to store data. The network interface of the electronic device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an intelligent control method for a CNC machine tool for machining aviation titanium alloy structural parts is implemented.

[0196] Those skilled in the art will understand that Figure 8 The structure shown in the figure is merely a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0197] In one embodiment, an electronic device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0198] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the above-mentioned computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0199] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. An intelligent control method for a CNC machine tool used for machining aviation titanium alloy structural parts, characterized in that: The steps include: Collect cutting force, temperature and vibration data during machining to obtain real-time working condition data; Extracting cutting process characteristic parameters according to the real-time working condition data to obtain working condition characteristic vectors; Based on the working condition characteristic vector, the optimal machining parameters are calculated through a pre-trained deep neural network model to obtain an optimized combination value of the feed speed and the spindle speed; According to the optimized combination value, the motion trajectory of the CNC machine tool is adjusted in real time to obtain a compensated processing path; Based on the compensated machining path, controlling the tool to perform a machining operation; The deep neural network model includes a first pre-trained deep neural network model and a second pre-trained deep neural network model; based on the working condition feature vector, the optimal machining parameters are calculated by the pre-trained deep neural network model to obtain the optimized combination value of the feed speed and the spindle speed, which specifically includes the following steps: According to the geometric features of the processing area, the working condition feature vector is divided into a straight line processing segment and a curved surface processing segment; Inputting the working condition feature vector corresponding to the straight line processing segment into a first pre-trained deep neural network model to obtain a parameter prediction value of the straight line processing segment; Inputting the working condition feature vector corresponding to the curved surface processing section into a second pre-trained deep neural network model to obtain a parameter prediction value of the curved surface processing section; Based on the predicted parameter values, the optimal combination values ​​of the feed speed and the spindle speed are generated in sections, and a transition interval is set at the junction of the processing sections.

2. The intelligent control method for CNC machine tools for machining aviation titanium alloy structural parts according to claim 1, characterized in that: According to the real-time working condition data, the cutting process characteristic parameters are extracted to obtain the working condition characteristic vector, which specifically includes the following steps: Dividing the real-time working condition data into a cutting entry segment, a stable cutting segment and a cutting exit segment according to the cutting stage; Performing mean filtering on the data of the stable cutting segment to obtain filtered steady-state data; Calculating the root mean square value of cutting force, temperature and vibration according to the filtered steady-state data to obtain characteristic parameters; Based on the characteristic parameters, a working condition characteristic vector is constructed.

3. The intelligent control method for CNC machine tools for machining aviation titanium alloy structural parts according to claim 1, characterized in that: According to the geometric features of the processing area, the working condition feature vector is divided into a straight line processing segment and a curved surface processing segment, which specifically includes the following steps: Based on the trajectory data of the machining program, the curvature change rate between adjacent tool positions is calculated to obtain the original curvature change rate data; According to the original curvature change rate data, a smoothing process is performed through a sliding window to obtain a smoothed curvature change rate; Comparing the curvature change rate after the smoothing process with a preset discrimination threshold and a minimum segment length threshold to determine a candidate segmentation area; According to the candidate segmentation regions, the continuity and length characteristics of each segment are calculated to obtain a preliminary segmentation result; Based on the preliminary segmentation result and the topological relationship of the machining features, error compensation is performed to obtain the final working condition feature vector segmentation result.

4. The intelligent control method for CNC machine tools for machining aviation titanium alloy structural parts according to claim 3 is characterized in that: According to the candidate segmentation regions, the continuity and length characteristics of each segment are calculated to obtain a preliminary segmentation result, which specifically includes the following steps: Based on the tool position coordinate sequence, the direction angle change between adjacent points is calculated to obtain the angle change sequence; Cumulatively summing the angle change sequence to obtain a cumulative value of direction change; Determining the geometric type of each segment according to a comparison result of the cumulative value of the direction change and a set threshold; Based on the length of the continuous segment of the geometric type, an adaptive weight coefficient is set, wherein when the length of the continuous segment is greater than a preset length threshold, the weight coefficient increases logarithmically with the segment length, and when the length of the continuous segment is less than the preset length threshold, the weight coefficient decreases linearly with the segment length; The segmentation boundary is locally adjusted according to the weight coefficient to obtain a preliminary segmentation result, wherein the segmentation with a weight coefficient less than a preset weight threshold is merged with the adjacent segment.

5. The intelligent control method for CNC machine tools for machining aviation titanium alloy structural parts according to claim 1, characterized in that: According to the optimized combination value, the motion trajectory of the CNC machine tool is adjusted in real time to obtain a compensated processing path, which specifically includes the following steps: Calculate the difference between the optimized feed speed and spindle speed and the current machining parameters to obtain the parameter adjustment amount; According to the magnitude of the parameter adjustment amount, the trajectory adjustment is divided into a rapid adjustment area and a gradual adjustment area; In the fast adjustment area, the motion trajectory is adjusted directly according to the optimized combination value; In the progressive adjustment area, the motion trajectory is gradually adjusted by piecewise linear interpolation; The trajectories of the rapid adjustment area and the gradual adjustment area are combined to obtain a compensated processing path.

6. An intelligent control system for CNC machine tools used for machining aviation titanium alloy structural parts, characterized in that: include: Real-time working condition data acquisition module, used to collect cutting force, temperature and vibration data during the processing to obtain real-time working condition data; A working condition characteristic vector acquisition module is used to extract cutting process characteristic parameters according to the real-time working condition data to obtain a working condition characteristic vector; An optimized combination value acquisition module is used to calculate the optimal processing parameters based on the working condition feature vector through a pre-trained deep neural network model to obtain an optimized combination value of the feed speed and the spindle speed; A machine tool motion trajectory adjustment module is used to adjust the motion trajectory of the CNC machine tool in real time according to the optimized combination value to obtain a compensated processing path; A tool control module, used for controlling the tool to perform a machining operation based on the compensated machining path; The deep neural network model includes a first pre-trained deep neural network model and a second pre-trained deep neural network model; the optimization combination value acquisition module is specifically used for: According to the geometric features of the processing area, the working condition feature vector is divided into a straight line processing segment and a curved surface processing segment; Inputting the working condition feature vector corresponding to the straight line processing segment into a first pre-trained deep neural network model to obtain a parameter prediction value of the straight line processing segment; Inputting the working condition feature vector corresponding to the curved surface processing section into a second pre-trained deep neural network model to obtain a parameter prediction value of the curved surface processing section; Based on the predicted parameter values, the optimal combination values ​​of the feed speed and the spindle speed are generated in sections, and a transition interval is set at the junction of the processing sections.

7. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the intelligent control method for a CNC machine tool for machining aviation titanium alloy structural parts according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the intelligent control method of a CNC machine tool for machining aviation titanium alloy structural parts according to any one of claims 1 to 5 are implemented.

Images