A precision part numerical control milling machine processing monitoring control method and system

Abstract

The application discloses a precision spare part numerical control milling machine processing monitoring control method and system, relates to the precision numerical control processing and intelligent manufacturing technical field, and comprises the following steps: S1, collecting numerical control milling machine data, establishing milling data table through narrowband internet of things synchronous transmission; S2, performing data processing on the milling data table, combining industrial vision to complete regional positioning, and establishing a spare part feature library; S3, calculating the vibration, load, texture and wear change of the processing position, rearranging the feature relationship, and forming a milling state chain; S4, tracking the abnormal change of the milling state chain, and combining historical samples to determine abnormal influence factors; S5, optimizing processing parameters according to the influence factors, and determining control parameters through cloud computing; S6, executing parameter adjustment, recording processing data and updating the monitoring control process. Through multi-source data fusion, abnormal tracking and parameter optimization control, the application realizes the accurate monitoring and intelligent control of precision spare part processing.

Description

Technical Field

[0001] This invention relates to the field of precision CNC machining and intelligent manufacturing technology, and in particular to a method and system for monitoring and controlling the machining of precision parts using CNC milling machines. Background Technology

[0002] With the development of intelligent manufacturing technology, CNC milling machines are widely used in the machining of precision parts. They control spindle motion, tool path, and machining parameters through CNC programs to achieve high-precision forming of parts. Currently, monitoring of the CNC milling process mainly involves collecting operational data such as spindle vibration, tool stress, and machining parameters. This data is then combined with threshold judgments, state classifications, or manual experience analysis to detect tool wear, machining vibration, and changes in machining quality.

[0003] Existing CNC milling machine machining monitoring methods typically use a single sensor for data acquisition, which makes it difficult to simultaneously correlate vibration changes, load changes, surface texture changes, and tool condition changes. This results in an inaccurate analysis of the influence relationships between different abnormal factors during machining. Furthermore, traditional methods often rely on fixed thresholds or preset rules for anomaly judgment, only issuing alarms after an anomaly occurs. They cannot trace the influencing factors based on the anomaly formation process, making it difficult to determine the correlation between tool wear, machining vibration, and surface defects.

[0004] In addition, existing machining control methods usually adjust the spindle speed or feed rate directly based on the detection results. The adjustment process of control parameters relies on human experience and lacks dynamic analysis of the relationship between different parameter combinations and machining response, which can easily lead to reduced machining efficiency or fluctuations in machining accuracy. Furthermore, existing systems lack a closed-loop optimization mechanism based on historical machining data and real-time status data, making it impossible to achieve continuous monitoring and adaptive control during the machining of precision parts.

[0005] Therefore, how to provide a method and system for monitoring and controlling the machining of precision parts by CNC milling is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] One objective of this invention is to propose a monitoring and control method and system for CNC milling of precision parts. This invention fully utilizes multi-source data fusion, industrial vision, machine intelligence analysis, and cloud computing control technology to dynamically monitor vibration, stress, surface quality, and tool status during CNC milling. It also optimizes machining parameters based on abnormal influencing factors, achieving high-precision monitoring and adaptive control of the precision parts machining process. This invention has the advantages of high machining stability, accurate anomaly identification, and rapid control response.

[0007] A method for monitoring and controlling the machining of precision parts using a CNC milling machine according to an embodiment of the present invention includes the following steps: S1. Collect spindle vibration signals, tool force signals, machining trajectory data, and part surface image data during CNC milling. Use narrowband Internet of Things to complete multi-node data synchronous transmission and establish a milling data table according to tool position coordinates and sampling time. S2. Perform frequency domain filtering, noise correction and scale standardization on the milling data table, use industrial vision to enhance the texture and locate the edges of the part surface image, combine the tool trajectory coordinates to complete the division of the machining area, and establish a part feature library. S3. Calculate the vibration difference, load offset, surface texture change and tool wear trend of adjacent machining positions based on the component feature library, and rearrange the features according to the machining path direction to form a milling state chain. S4. Perform reverse tracing calculation on the milling state chain, extract the feature offset relationship before and after abnormal changes, and combine the parameter change law in historical machining samples to determine the influencing factors corresponding to tool abnormalities, machining vibration and surface defects. S5. Based on the influencing factors, the spindle speed, feed rate, depth of cut and tool path are adjusted in combination. The differences in machining response corresponding to different adjustment amounts are analyzed by cloud computing to determine the control parameters of the CNC milling machine. S6. Send the CNC milling machine control parameters to the distributed control node to perform machining adjustments, and record the adjusted vibration data, visual images and machining quality data through the cloud platform to update the precision parts machining monitoring and control process.

[0008] Optionally, S1 specifically includes: Collect spindle vibration signals, tool force signals, machining trajectory data, and part surface image data during CNC milling. Use narrowband Internet of Things to complete multi-node data synchronous transmission and establish a milling data table according to tool position coordinates and sampling time. A multi-node synchronous acquisition link is established based on narrowband IoT. A unified time identifier is set for different acquisition nodes. Timing compensation is performed based on data transmission delay, sampling period offset and node response difference. Vibration signal, force signal, trajectory data and visual image are matched according to the same tool position coordinate. Construct milling status marking units. Each milling status marking unit is an independent data record unit formed by binding different types of machining data with tool position coordinates, machining stage and sampling time as indexes. Each milling status marking unit corresponds to a local machining area. Based on the data changes between consecutive tool positions, the vibration change trend, force change trend and surface texture change trend of adjacent positions are calculated. The data sampling density is increased for positions with abrupt changes, and the original sampling frequency is maintained for stable positions, forming a milling data table with machining position correlation.

[0009] Optionally, the data processing procedure for the milling data table specifically includes: The spindle vibration signal and tool force signal in the milling data table are segmented according to the tool position coordinates, and the continuous machining process is divided into multiple local cutting windows; Within each local cutting window, the original sampling sequence is subjected to spectral transformation, and the low-frequency stable region, mid-frequency cutting response region, and high-frequency impact region are divided according to the energy ratio in different frequency ranges. For instantaneous peak data in the high-frequency impact zone, the amplitude change direction and duration between adjacent sampling points are combined to determine whether it belongs to cutting impact characteristics. Data with a duration less than the preset window length and not synchronized with the tool position change are marked as interference data, and the interference effect is reduced by frequency weight attenuation. For the mid-frequency cutting response zone, frequency components related to the tool cutting state are retained. The vibration frequency band offset is calculated by using the frequency migration relationship between adjacent machining windows. When there is a continuous frequency offset between adjacent windows, the corresponding frequency change trend is retained and recorded in the milling state marking unit. To address the low-frequency baseline shift caused by sensor drift, a dynamic reference value is established based on the stable sampling interval at the beginning of the processing stage. Baseline correction is then performed on subsequent sampling data to ensure that the vibration amplitude and force changes at different processing stages are within a unified reference range. After completing the frequency domain filtering, the spindle vibration features, tool force features, machining speed features, and visual extraction features are arranged according to the feature dimensions. The data change range within each machining window is calculated using a segmented scale mapping method. Data of different dimensions are converted into a unified feature scale, and the data increase and decrease relationship corresponding to the tool position change is retained to form standardized milling feature data.

[0010] Optionally, the process of image processing and processing area division of the component surface specifically includes: The surface images of parts acquired by industrial vision equipment are matched with regions according to the tool trajectory coordinates. The corresponding image regions are divided according to the processing section number. Local brightness compensation and texture gradient enhancement processing are performed on each image region. Pixel correction is performed on areas of uneven lighting in the image, and the local brightness distribution is adjusted according to the gray-scale change relationship of adjacent pixels to eliminate local overexposure areas caused by metal reflection during milling. Texture calculation is performed on the enhanced component surface image according to the pixel neighborhood relationship. Pixel gradient detection paths are established along the horizontal, vertical and 45-degree tilt directions respectively. The gray-level difference and the number of continuous changes between adjacent pixels are calculated in turn. The texture response value in the corresponding direction is determined based on the grayscale change amplitude in different directions. When the grayscale change direction of consecutive pixels is consistent, the corresponding area is marked as the tool motion texture area, and the texture extension direction, continuous length and change intensity are recorded. The texture response results obtained in the three directions are matched with their positions. The response values ​​in different directions are fused and compared, and the direction with the largest response value is retained as the main texture direction of the current region. The distribution of machining marks is determined based on the continuity of the main texture direction. Areas with continuous texture direction and stable intervals are marked as normal machining areas, while areas with abrupt changes in texture direction, continuous length decrease, or abnormal grayscale changes are marked as abnormal surface areas. Edge localization processing is performed on the enhanced image. By detecting continuous pixel change regions, the boundaries of parts, the edges of machining grooves, and the contours of surface defects are determined, and isolated noise points are filtered out based on edge connectivity. The detected edge position is spatially matched with the CNC milling machine tool trajectory coordinates, the positional offset between the tool movement path and the actual machining area is calculated, and the image area coordinates are corrected based on the offset. A visual feature index table for parts is established, and the contour coordinates, texture direction, edge length, grayscale variation range and defect area location corresponding to each processing area are associated and stored to form a part feature library.

[0011] Optionally, S3 specifically includes: Based on the machining area coordinates, tool position numbers and machining stage information in the component feature library, the continuous machining areas are sorted by path, and vibration features, load features, visual texture features and tool status features are bound to the corresponding tool position nodes. Read the vibration amplitude sequence, force data and visual texture data of adjacent tool position nodes, calculate the frequency band change, load offset, texture deflection and edge change, and extract the state difference features between the machining positions; Based on the changing trends of cutting pressure, vibration frequency, and surface texture within the continuous machining area, the synchronous changing relationship of multiple features is analyzed to determine the tool wear trend characteristics. The characteristics of various state changes are normalized, and the nodes of continuous change are connected according to the tool movement path. Abnormal offset positions are marked as state change nodes, forming a milling state chain that includes machining position and change trend.

[0012] Optionally, S4 specifically includes: Based on the state change nodes corresponding to each machining position in the milling state chain, trace the continuous machining nodes forward from the abnormal position in the opposite direction of the machining path, read the vibration change, load offset, surface texture change and tool wear trend of multiple machining positions before the abnormality occurred, and establish the abnormal forward feature sequence. Time alignment is performed on the feature change process of each feature in the abnormal forward feature sequence. The position, duration and direction of the first shift of different features are recorded. The position of the abnormal occurrence is matched with the historical stable processing state to extract the feature shift relationship before and after the abnormal change. Based on the data of tool usage, cutting load changes, vibration frequency changes, and part surface quality in historical machining samples, a machining state comparison table is established. The feature offset relationship in the current milling state chain is matched with the historical machining samples to select machining samples with the same trend. For the matched historical machining samples, the characteristic change patterns corresponding to tool wear state, machining vibration state and surface defect state are statistically analyzed, and the correlation changes of vibration enhancement, load increase, texture destruction and wear accumulation under different abnormality types are recorded. The influencing factors are sorted according to the degree of feature offset of the current abnormal location. Features that continuously affect the changes in the abnormal state are marked as the main influencing factors, and features with weak correlation are marked as auxiliary influencing factors. Establish an abnormal influence relationship table to associate and store the influencing factors corresponding to tool abnormalities, machining vibrations, and surface defects with the abnormal location, direction of change, and degree of influence.

[0013] Optionally, the process of establishing the anomaly forward feature sequence specifically includes: Based on the abnormal change nodes in the milling state chain, obtain the tool position coordinates, machining time and state change type corresponding to the abnormal node, and read the continuous machining nodes sequentially in the opposite direction of tool movement starting from the abnormal node. For each tracking node, the vibration change, load offset, surface texture change, and tool wear trend are extracted synchronously and rearranged according to the time of anomaly occurrence to form a characteristic change trajectory; Differences in characteristic changes between continuously tracked nodes are calculated, and the locations of the first increase in vibration amplitude, the first shift in load change, the first appearance of texture anomaly, and the first increase in wear trend are recorded respectively. The anomaly induction region is determined based on the distance relationship between the first change positions of multiple features. When multiple features change continuously in adjacent processing nodes, the corresponding nodes are divided into anomaly evolution regions. When a single feature undergoes a sudden change while other features remain stable, the corresponding node is marked as a local perturbation region; The sequence, direction, and duration of feature changes within the abnormal evolution region are encoded to form an abnormal forward feature sequence that contains the abnormal generation process.

[0014] Optionally, the process of establishing the anomaly impact relationship table specifically includes: Based on the feature offset relationship in the current milling state chain and the abnormal change records in historical machining samples, feature correlation analysis is performed on different anomaly types; Historical machining samples were categorized based on tool abnormalities, machining vibrations, and surface defects. Continuous variation intervals corresponding to vibration changes, load changes, texture changes, and wear changes were extracted from samples of each category. The feature offset data corresponding to the current anomaly location is matched with the historical samples of each category in terms of location and change trend. The order, duration and magnitude of change of each feature under different anomaly types are statistically analyzed. The influence relationship is determined based on the degree of synchronization of feature changes. When a feature change consistently appears before the formation of an anomaly and increases synchronously with the degree of anomaly, the feature is recorded as a highly correlated influence factor. When a feature changes only after an anomaly occurs, that feature is recorded as a result response factor. An anomaly impact relationship table is established based on the category of influencing factors, anomaly type, location of change, and degree of correlation, so that a corresponding relationship is formed between changes in tool condition, changes in cutting vibration, and changes in surface quality.

[0015] Optionally, S5 specifically includes: Based on the ranking of influencing factors in the abnormal influence relationship table, the abnormal types corresponding to tool abnormalities, machining vibrations and surface defects are associated and matched with spindle speed, feed rate, depth of cut and tool path to determine the control parameters to be adjusted; Extract the vibration changes, load offsets, texture changes, and wear trends before and after the anomaly occurred, and combine them with parameter response data from historical processing samples to establish processing response records corresponding to different combinations of control parameters. The response analysis of various control parameter combinations is performed through cloud computing. A comprehensive response score is calculated based on the degree of vibration suppression, load stability, surface quality change, and processing efficiency change to select the optimal parameter combination. The optimal parameter combination is verified by positive and negative disturbances. Data on the changes in machining state before and after the disturbance are collected. Based on the response differences, the adjustment direction and adjustment range of spindle speed, feed rate, depth of cut and tool path are determined.

[0016] A precision parts CNC milling machine machining monitoring and control system according to an embodiment of the present invention includes: The data acquisition module is used to collect spindle vibration signals, tool force signals, machining trajectory data and part surface image data during the CNC milling process. It uses narrowband Internet of Things to complete multi-node data synchronous transmission and establishes a milling data table according to the tool position coordinates and sampling time. The data processing module is used to perform frequency domain filtering, noise correction and scale normalization on the milling data table, use industrial vision to enhance the texture and locate the edges of the part surface image, combine the tool trajectory coordinates to complete the division of the machining area, and establish a part feature library. The state analysis module is used to calculate the vibration difference, load offset, surface texture change and tool wear trend between adjacent machining positions based on the component feature library, and rearrange the features according to the machining path direction to form a milling state chain. The anomaly identification module is used to perform reverse tracing calculations on the milling state chain, extract the feature offset relationship before and after the abnormal change, and combine the parameter change patterns in historical machining samples to determine the influencing factors corresponding to tool anomalies, machining vibrations and surface defects. The parameter optimization module is used to combine and adjust the spindle speed, feed rate, depth of cut and tool path according to the influencing factors. It uses cloud computing to analyze the differences in machining response corresponding to different adjustment amounts and determine the control parameters of the CNC milling machine. The monitoring and control module is used to send the control parameters of the CNC milling machine to the distributed control node to perform machining adjustments, and to record the adjusted vibration data, visual images and machining quality data through the cloud platform to update the monitoring and control process of precision parts machining.

[0017] The beneficial effects of this invention are: First, this invention collects spindle vibration, tool force, machining trajectory, and part surface image data, and combines this with narrowband IoT to achieve multi-node synchronous transmission. It also improves the consistency of machining data through filtering, correction, and standardization, and combines industrial vision to complete the positioning of the machining area, thereby improving the accuracy of obtaining the machining status of precision parts CNC milling machines.

[0018] Secondly, by establishing a component feature library, this invention performs correlation analysis on vibration changes, load offsets, surface texture changes, and tool wear trends corresponding to the machining position. By using the milling state chain to track abnormal changes, it can identify the influence relationship between tool abnormalities, machining vibrations, and surface defects, thereby improving the timeliness and reliability of machining abnormality detection.

[0019] Finally, this invention optimizes and adjusts the spindle speed, feed rate, depth of cut, and tool path based on abnormal influencing factors. By analyzing the machining response corresponding to different control parameters through cloud computing and combining it with distributed control nodes to perform closed-loop regulation, the invention achieves dynamic optimization of CNC milling machine machining parameters, thereby improving the machining accuracy, machining stability, and production efficiency of precision parts. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is an overall flowchart of a CNC milling machine monitoring and control method for precision parts proposed in this invention; Figure 2 This is a schematic diagram illustrating the data acquisition and synchronous transmission of a CNC milling machine monitoring and control method for precision parts proposed in this invention. Figure 3 This is a schematic diagram illustrating the data processing and component feature library establishment of a CNC milling machine machining monitoring and control method for precision components proposed in this invention; Figure 4 This is a schematic diagram of anomaly tracking and influencing factors in a CNC milling monitoring and control method for precision parts proposed in this invention. Figure 5 This is a schematic diagram of the CNC milling parameter optimization control of the CNC milling monitoring and control method for precision parts proposed in this invention; Figure 6 This is a module structure diagram of a precision parts CNC milling machine machining monitoring and control system proposed in this invention. Detailed Implementation

[0021] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0022] refer to Figures 1-5 A method for monitoring and controlling the machining of precision parts by CNC milling includes the following steps: S1. Collect spindle vibration signals, tool force signals, machining trajectory data, and part surface image data during CNC milling. Use narrowband Internet of Things to complete multi-node data synchronous transmission and establish a milling data table according to tool position coordinates and sampling time. S2. Perform frequency domain filtering, noise correction and scale standardization on the milling data table, use industrial vision to enhance the texture and locate the edges of the part surface image, combine the tool trajectory coordinates to complete the division of the machining area, and establish a part feature library. S3. Calculate the vibration difference, load offset, surface texture change and tool wear trend of adjacent machining positions based on the component feature library, and rearrange the features according to the machining path direction to form a milling state chain. S4. Perform reverse tracing calculation on the milling state chain, extract the feature offset relationship before and after abnormal changes, and combine the parameter change law in historical machining samples to determine the influencing factors corresponding to tool abnormalities, machining vibration and surface defects. S5. Based on the influencing factors, the spindle speed, feed rate, depth of cut and tool path are adjusted in combination. The differences in machining response corresponding to different adjustment amounts are analyzed by cloud computing to determine the control parameters of the CNC milling machine. S6. Send the CNC milling machine control parameters to the distributed control node to perform machining adjustments, and record the adjusted vibration data, visual images and machining quality data through the cloud platform to update the precision parts machining monitoring and control process.

[0023] In this embodiment, S1 specifically includes: Collect spindle vibration signals, tool force signals, machining trajectory data, and part surface image data during CNC milling. Use narrowband Internet of Things to complete multi-node data synchronous transmission and establish a milling data table according to tool position coordinates and sampling time. The spindle vibration signal is collected by a vibration sensor mounted on the spindle mount, which measures the triaxial vibration amplitude, vibration frequency, and impact change. The tool force signal is acquired by the tool holder strain sensor, which collects changes in cutting pressure, radial force, and axial force. The machining trajectory data is read by the CNC controller, which shows the current position of the tool, the direction of movement, the feed rate, and the machining segment number. The surface image data of the parts is collected by industrial vision equipment, which captures the surface texture, edge contours, and defect areas before and after processing. A multi-node synchronous acquisition link is established based on narrowband IoT. A unified time identifier is set for different acquisition nodes. Timing compensation is performed based on data transmission delay, sampling period offset and node response difference. Vibration signal, force signal, trajectory data and visual image are matched according to the same tool position coordinate. Construct milling status marking units. Each milling status marking unit is an independent data record unit formed by binding different types of machining data with tool position coordinates, machining stage and sampling time as indexes. Each milling status marking unit corresponds to a local machining area. Based on the data changes between consecutive tool positions, the vibration change trend, force change trend and surface texture change trend of adjacent positions are calculated. The data sampling density is increased for positions with abrupt changes, and the original sampling frequency is maintained for stable positions, forming a milling data table with machining position correlation.

[0024] In this embodiment, the timing compensation process performed according to the multi-node synchronous acquisition link specifically includes: After establishing a multi-node synchronous acquisition link based on narrowband IoT, node time identifiers are configured for vibration sensor nodes, tool holder strain sensor nodes, CNC controller nodes and industrial vision device nodes respectively, and acquisition sequence number, acquisition time and tool position number are added to the acquisition data of each node. Calculate the node response offset based on the arrival order of continuous data packets at each node, perform forward and backward displacement correction on data from different sources according to the response offset, delay and match the data collected in advance to the corresponding tool position, and replenish the data collected in delay to the corresponding sampling period. A time-coupling correction unit is constructed. The time-coupling correction unit is a data synchronization correction structure established based on the node acquisition time difference, data transmission delay, and tool movement distance. The time matching range is determined by analyzing the correspondence between vibration abrupt change, force change, and image edge change under the same processing position. Within the time matching range, sliding window recombination is performed on various types of data, dividing continuous sampling points into multiple local processing windows, and rearranging the data order within the window according to the tool movement direction, so that the vibration state, force state, trajectory state and visual state within the same local processing area maintain the corresponding relationship. The milling status marker unit is updated based on the reorganized local machining window, so that each milling status marker unit contains the corresponding tool position coordinates, machining stage, sampling time, vibration change, force change and surface image features, so as to achieve accurate correspondence of multi-source machining data.

[0025] In this embodiment, the data sampling process performed according to the milling status marking unit specifically includes: Based on the milling state marker units between consecutive tool positions, the vibration amplitude change, force direction change, texture edge change and tool motion change in adjacent machining areas are extracted to establish a machining state change evaluation sequence; A sampling adjustment unit is constructed, which is a data control structure that adjusts the data acquisition density according to the degree of change in processing status. When there is a rapid change in vibration amplitude, a change in force direction, or a sudden change in visual texture between adjacent milling status marker units, the corresponding processing area is divided into a status change area, and the sampling interval is shortened to increase the amount of local data. When the continuous milling status marker unit maintains a stable trend of change, the corresponding machining area is divided into a stable machining area, and the number of repeated samplings is reduced while maintaining the original sampling period. Differentiated data retention rules are set for different processing areas, including: The complete vibration waveform, force change trajectory, and surface image details are preserved in the region of state change; Key feature points and representative data are retained for stable processing areas; Based on the output of the sampling adjustment unit, the data priority of the milling status marking unit is rearranged to give higher data expression density to the areas corresponding to tool wear, machining vibration and surface defects, forming a milling data table containing the characteristics of key machining areas.

[0026] In this embodiment, the data processing procedure for the milling data table specifically includes: The spindle vibration signal and tool force signal in the milling data table are segmented according to the tool position coordinates, and the continuous machining process is divided into multiple local cutting windows; Within each local cutting window, the original sampling sequence is subjected to spectral transformation, and the low-frequency stable region, mid-frequency cutting response region, and high-frequency impact region are divided according to the energy ratio in different frequency ranges. For instantaneous peak data in the high-frequency impact zone, the amplitude change direction and duration between adjacent sampling points are combined to determine whether it belongs to cutting impact characteristics. Data with a duration less than the preset window length and not synchronized with the tool position change are marked as interference data, and the interference effect is reduced by frequency weight attenuation. For the mid-frequency cutting response zone, frequency components related to the tool cutting state are retained. The vibration frequency band offset is calculated by using the frequency migration relationship between adjacent machining windows. When there is a continuous frequency offset between adjacent windows, the corresponding frequency change trend is retained and recorded in the milling state marking unit. To address the low-frequency baseline shift caused by sensor drift, a dynamic reference value is established based on the stable sampling interval at the beginning of the processing stage. Baseline correction is then performed on subsequent sampling data to ensure that the vibration amplitude and force changes at different processing stages are within a unified reference range. After completing the frequency domain filtering, the spindle vibration features, tool force features, machining speed features, and visual extraction features are arranged according to the feature dimensions. The data change range within each machining window is calculated using a segmented scale mapping method. Data of different dimensions are converted into a unified feature scale, and the data increase and decrease relationship corresponding to the tool position change is retained to form standardized milling feature data.

[0027] In this embodiment, the process of image processing and processing area division of the component surface specifically includes: The surface images of parts acquired by industrial vision equipment are matched with regions according to the tool trajectory coordinates. The corresponding image regions are divided according to the processing section number. Local brightness compensation and texture gradient enhancement processing are performed on each image region. Pixel correction is performed on areas of uneven lighting in the image, and the local brightness distribution is adjusted according to the gray-scale change relationship of adjacent pixels to eliminate local overexposure areas caused by metal reflection during milling. Texture calculation is performed on the enhanced component surface image according to the pixel neighborhood relationship. Pixel gradient detection paths are established along the horizontal, vertical and 45-degree tilt directions respectively. The gray-level difference and the number of continuous changes between adjacent pixels are calculated in turn. The texture response value in the corresponding direction is determined based on the grayscale change amplitude in different directions. When the grayscale change direction of consecutive pixels is consistent, the corresponding area is marked as the tool motion texture area, and the texture extension direction, continuous length and change intensity are recorded. The texture response results obtained in the three directions are matched with their positions. The response values ​​in different directions are fused and compared, and the direction with the largest response value is retained as the main texture direction of the current region. The distribution of machining marks is determined based on the continuity of the main texture direction. Areas with continuous texture direction and stable intervals are marked as normal machining areas, while areas with abrupt changes in texture direction, continuous length decrease, or abnormal grayscale changes are marked as abnormal surface areas. Edge localization processing is performed on the enhanced image. By detecting continuous pixel change regions, the boundaries of parts, the edges of machining grooves, and the contours of surface defects are determined, and isolated noise points are filtered out based on edge connectivity. The detected edge position is spatially matched with the CNC milling machine tool trajectory coordinates, the positional offset between the tool movement path and the actual machining area is calculated, and the image area coordinates are corrected based on the offset. A visual feature index table for parts is established, and the contour coordinates, texture direction, edge length, grayscale variation range and defect area location corresponding to each processing area are associated and stored to form a part feature library.

[0028] In this embodiment, S3 specifically includes: Based on the machining area coordinates, tool position number and machining stage information in the component feature library, the continuous machining areas under the same tool movement path are sorted, and the vibration characteristics, load characteristics, visual texture characteristics and tool state characteristics corresponding to each machining area are bound to the corresponding tool position node. To investigate the vibration changes between adjacent tool positions, the vibration amplitude sequence within the corresponding time window of the two tool positions is read. The amplitude changes at the same frequency band are compared point by point, and the changes in each frequency band are accumulated. At the same time, the degree of vibration difference is determined by combining the number of times the impact peak occurs and the duration of the changes. For load offset, extract the radial cutting force, axial cutting force and spindle load data corresponding to adjacent tool positions, align and compare the force data of the current machining position with the historical force data of the corresponding tool position in the stable machining area, record the changing trend in each force direction, and determine the load offset based on the number and magnitude of force direction changes. To address surface texture changes, image texture features from the component feature library are called to perform position matching on the tool mark direction, edge continuity length, and grayscale change areas in adjacent processing areas. The number of texture direction deflections, the number of abnormal edges, and the range of pixel changes are counted to obtain the amount of surface texture change. To address the tool wear trend, the increase in cutting pressure, the shift in vibration frequency, and the degradation of surface texture within the continuous machining area are arranged in chronological order. The presence of synchronous growth relationships in multiple machining positions is detected. When multiple features change in the same direction, the wear trend weight is increased; when the direction of feature changes is inconsistent, the wear trend weight is decreased, thus forming a record of tool wear changes. The vibration difference, load offset, surface texture change and tool wear trend are normalized. The features are rearranged according to the order of each machining position in the machining path. Features with continuous change relationship between adjacent positions are connected. Positions with change amplitude exceeding the local stability range are marked as state change nodes. The state change nodes are associated sequentially according to the tool movement direction to form a milling state chain containing machining position, state change type and change trend.

[0029] In this embodiment, S4 specifically includes: Based on the state change nodes corresponding to each machining position in the milling state chain, trace the continuous machining nodes forward from the abnormal position in the opposite direction of the machining path, read the vibration change, load offset, surface texture change and tool wear trend of multiple machining positions before the abnormality occurred, and establish the abnormal forward feature sequence. Time alignment is performed on the feature change process of each feature in the abnormal forward feature sequence. The position, duration and direction of the first shift of different features are recorded. The position of the abnormal occurrence is matched with the historical stable processing state to extract the feature shift relationship before and after the abnormal change. Based on the data of tool usage, cutting load changes, vibration frequency changes, and part surface quality in historical machining samples, a machining state comparison table is established. The feature offset relationship in the current milling state chain is matched with the historical machining samples to select machining samples with the same trend. For the matched historical machining samples, the characteristic change patterns corresponding to tool wear state, machining vibration state and surface defect state are statistically analyzed, and the correlation changes of vibration enhancement, load increase, texture destruction and wear accumulation under different abnormality types are recorded. The influencing factors are sorted according to the degree of feature offset of the current abnormal location. Features that continuously affect the changes in the abnormal state are marked as the main influencing factors, and features with weak correlation are marked as auxiliary influencing factors. Establish an abnormal influence relationship table to associate and store the influencing factors corresponding to tool abnormalities, machining vibrations, and surface defects with the abnormal location, direction of change, and degree of influence.

[0030] In this embodiment, the process of establishing the anomaly forward feature sequence specifically includes: Based on the abnormal change nodes in the milling state chain, obtain the tool position coordinates, machining time and state change type corresponding to the abnormal node, and read the continuous machining nodes sequentially in the opposite direction of tool movement starting from the abnormal node. For each tracking node, the vibration change, load offset, surface texture change, and tool wear trend are extracted synchronously and rearranged according to the time of anomaly occurrence to form a characteristic change trajectory; Differences in characteristic changes between continuously tracked nodes are calculated, and the locations of the first increase in vibration amplitude, the first shift in load change, the first appearance of texture anomaly, and the first increase in wear trend are recorded respectively. The anomaly induction region is determined based on the distance relationship between the first change positions of multiple features. When multiple features change continuously in adjacent processing nodes, the corresponding nodes are divided into anomaly evolution regions. When a single feature undergoes a sudden change while other features remain stable, the corresponding node is marked as a local perturbation region; The sequence, direction, and duration of feature changes within the abnormal evolution region are encoded to form an abnormal forward feature sequence that contains the abnormal generation process.

[0031] In this embodiment, the process of establishing the abnormal influence relationship table specifically includes: Based on the feature offset relationship in the current milling state chain and the abnormal change records in historical machining samples, feature correlation analysis is performed on different anomaly types; Historical machining samples were categorized based on tool abnormalities, machining vibrations, and surface defects. Continuous variation intervals corresponding to vibration changes, load changes, texture changes, and wear changes were extracted from samples of each category. The feature offset data corresponding to the current anomaly location is matched with the historical samples of each category in terms of location and change trend. The order, duration and magnitude of change of each feature under different anomaly types are statistically analyzed. The influence relationship is determined based on the degree of synchronization of feature changes. When a feature change consistently appears before the formation of an anomaly and increases synchronously with the degree of anomaly, the feature is recorded as a highly correlated influence factor. When a feature changes only after an anomaly occurs, that feature is recorded as a result response factor. An anomaly impact relationship table is established based on the category of influencing factors, anomaly type, location of change, and degree of correlation, so that a corresponding relationship is formed between changes in tool condition, changes in cutting vibration, and changes in surface quality.

[0032] In this embodiment, S5 specifically includes: Based on the ranking of influencing factors in the abnormal influence relationship table, parameter correlation matching is performed on spindle speed, feed rate, depth of cut and tool path, and the abnormal type corresponding to the influencing factor is matched with the control parameter change record in the historical machining sample; Read the change characteristics of the main influencing factors under the current abnormal state, extract the changes in vibration amplitude, cutting load, surface texture and tool wear before and after the abnormality occurs, and determine the category of control parameters that need to be adjusted according to the direction of feature changes; When the vibration change factor continues to increase, read the vibration response data under different spindle speed and cutting depth combinations in the historical samples, and statistically analyze the range of vibration amplitude decrease and the duration of stability before and after each parameter adjustment. When the load offset factor continues to increase, extract the force change data under different feed rates and cutting depths, and compare the radial force, axial force and spindle load change trends before and after the parameter changes. When the surface texture change factor becomes abnormal, industrial vision inspection data is called to perform matching analysis on the surface texture continuity, edge integrity and number of defect areas under different tool paths and machining speeds. The above historical response data is uploaded to the cloud platform, and different combinations of control parameters are grouped and calculated using cloud computing. Parameter response records are established according to the vibration change, load change, texture change, and wear trend corresponding to the current abnormal state. For each set of parameter response records, the changes in vibration amplitude, load fluctuation, texture defect, and processing efficiency before and after adjustment are extracted. The parameters are weighted according to three dimensions: contribution to anomaly suppression, degree of quality retention, and impact on processing efficiency. The comprehensive response scores corresponding to each combination of control parameters are compared, and the parameter combination with the largest decrease in anomaly and the smallest quality deviation is determined as the optimal control parameters. Dynamic disturbance verification was performed on the selected parameter combinations by increasing and decreasing the spindle speed, feed rate, depth of cut, and tool path offset. Data on the changes in machining state before and after the disturbance were collected, and the response differences corresponding to parameter changes in different directions were compared to determine the adjustment direction and adjustment range of each control parameter.

[0033] In this embodiment, S6 specifically includes: The control parameters of the CNC milling machine are split into data according to the processing equipment node, control parameter type and execution priority, and then sent to the corresponding distributed control node via narrowband Internet of Things. After receiving the control parameters, the distributed control node performs parameter verification based on the execution status of the spindle control unit, feed control unit, and trajectory control unit, and detects the correspondence between the current equipment operating status and the control parameters. When the parameter adjustment range is detected to exceed the equipment safety boundary, the adjustment range is automatically corrected based on historical stable processing status. After the control node executes the parameter adjustment, it re-acquires the spindle vibration signal, tool force signal, part surface image data and machining quality inspection data, and performs time correspondence between the machining status data before and after the adjustment; Based on the changes in data before and after adjustment, calculate the vibration trend, load recovery degree, surface texture improvement degree, and processing deviation change, and mark the control effect as an effective adjustment state or an ineffective adjustment state; For effective adjustment states, the corresponding control parameter combinations, abnormal influencing factors, and processing response results are uploaded to the cloud platform to form a control sample library; For invalid adjustment states, the location of abnormal changes, the direction of parameter adjustment, and the difference in processing response are recorded in the correction data table; By using cloud computing to correlate and update the newly added control sample library and historical processing data, the matching relationship between abnormal influencing factors and control parameters is readjusted, and the precision parts processing monitoring and control process is updated, so that the subsequent CNC milling process can perform dynamic control based on the real-time processing status.

[0034] refer to Figure 6 A precision parts CNC milling machine machining monitoring and control system, comprising: The data acquisition module is used to collect spindle vibration signals, tool force signals, machining trajectory data and part surface image data during the CNC milling process. It uses narrowband Internet of Things to complete multi-node data synchronous transmission and establishes a milling data table according to the tool position coordinates and sampling time. The data processing module is used to perform frequency domain filtering, noise correction and scale normalization on the milling data table, use industrial vision to enhance the texture and locate the edges of the part surface image, combine the tool trajectory coordinates to complete the division of the machining area, and establish a part feature library. The state analysis module is used to calculate the vibration difference, load offset, surface texture change and tool wear trend between adjacent machining positions based on the component feature library, and rearrange the features according to the machining path direction to form a milling state chain. The anomaly identification module is used to perform reverse tracing calculations on the milling state chain, extract the feature offset relationship before and after the abnormal change, and combine the parameter change patterns in historical machining samples to determine the influencing factors corresponding to tool anomalies, machining vibrations and surface defects. The parameter optimization module is used to combine and adjust the spindle speed, feed rate, depth of cut and tool path according to the influencing factors. It uses cloud computing to analyze the differences in machining response corresponding to different adjustment amounts and determine the control parameters of the CNC milling machine. The monitoring and control module is used to send the control parameters of the CNC milling machine to the distributed control node to perform machining adjustments, and to record the adjusted vibration data, visual images and machining quality data through the cloud platform to update the monitoring and control process of precision parts machining.

[0035] Example 1: To verify the feasibility of this invention in practice, it was applied to a CNC milling machine machining scenario for a precision component. This scenario addresses the problems of tool wear accumulation, cutting load fluctuation, increased spindle vibration, and decreased surface quality during continuous milling of high-precision metal components. The machining object is a multi-structure precision component, which includes planar machining areas, curved transition areas, and groove machining areas. During machining, spindle speed, feed rate, depth of cut, and tool path are set as the main control parameters. Before machining, the operating status of the CNC milling machine is monitored, and data from the initial stable machining stage are collected. The effective value of spindle vibration is maintained within the range of 0.42 mm / s to 0.56 mm / s, the radial cutting force is maintained within the range of 86 N to 112 N, the axial cutting force is maintained within the range of 42 N to 63 N, the average continuous length of the surface texture of the component reaches 96.5%, and the machining dimensional deviation is controlled within 0.018 mm.

[0036] In practical applications, multiple types of sensor nodes are deployed near the spindle seat, tool holder connection position, and machining area of ​​the CNC milling machine. A distributed data synchronization link is established through narrowband IoT to synchronously collect spindle vibration signals, tool force signals, machining trajectory data, and part surface image data. During continuous machining, each machining area is numbered according to the tool position coordinates, and the corresponding collected data is bound to form a milling data table. During testing, data from different machining areas are continuously collected. In a single machining process, approximately 1.2 million sets of vibration sampling data, approximately 800,000 sets of tool force data, approximately 150,000 sets of machining trajectory data, and approximately 2,400 industrial vision images are obtained. The correspondence between multi-source data is completed through a unified time identifier.

[0037] For the collected milling data, frequency domain filtering was first performed on the vibration signal, dividing the original vibration data into a low-frequency stable region, a mid-frequency cutting response region, and a high-frequency impact region. Invalid interference signals generated by equipment operation were removed by frequency change detection. Before processing, approximately 13.6% of the vibration signal contained non-machining disturbance data; after filtering and correction, this was reduced to 2.1%. Simultaneously, offset correction was performed on the tool force data, reducing the force error caused by sensor zero-point drift from 5.8N to 1.2N. Industrial vision was used to enhance the texture and locate edges on the part surface image. By matching the tool trajectory coordinates to the actual machining area, the image positioning error was reduced from 0.34mm to 0.06mm. A part feature library containing machining area coordinates, texture direction, edge continuity length, and defect locations was established.

[0038] During the machining state analysis, the system performs correlation calculations on adjacent tool position nodes based on the component feature library. For example, in a certain continuous machining area, the system detected that the vibration amplitude between the 1200th and 1450th tool position nodes increased from 0.58 mm / s to 1.46 mm / s, an increase of 151.7%; simultaneously, the radial cutting force increased from 108 N to 176 N, an increase of 63.0%; and industrial vision detected that the continuous length of the tool mark texture in the corresponding area decreased from 95.8% to 81.4%, a decrease of 14.4 percentage points. Based on the above multi-source change relationship, the system marks this area as a state change node and traces the preceding machining nodes backward along the machining path. It finds that vibration frequency shift, load increase, and texture degradation had already appeared synchronously in about 36 machining nodes before the anomaly occurred, thus determining that tool wear is the main influencing factor.

[0039] During the anomaly impact analysis, the current machining state was matched with historical machining samples to establish a correspondence between anomaly types and control parameters. Test data showed that when the tool wear impact factor accounted for more than 68%, adjusting the feed rate had the most significant effect on improving the machining state; when the machining vibration impact factor accounted for more than 60%, adjusting the spindle speed could effectively reduce the vibration amplitude; and when the surface texture anomaly impact factor accounted for more than 55%, optimizing the toolpath could improve the continuity of surface texture. Based on the above analysis results, response analysis was performed on different combinations of spindle speed, feed rate, depth of cut, and toolpath using cloud computing, generating a total of 128 sets of control parameter combinations. The vibration change, load change, surface defect change, and machining efficiency change corresponding to each set of parameters were recorded.

[0040] In one case, under abnormal vibration machining conditions, the original machining parameters were a spindle speed of 4200 r / min, a feed rate of 680 mm / min, and a depth of cut of 0.35 mm, with a detected effective vibration value of 1.52 mm / s. After cloud computing analysis, an optimized combination of a spindle speed of 3900 r / min, a feed rate of 620 mm / min, and a depth of cut of 0.28 mm was selected. After adjustment, the effective vibration value decreased to 0.63 mm / s, a reduction of 58.6%; the radial cutting force decreased from 182 N to 126 N, a reduction of 30.8%; and the surface texture continuity length increased from 79.6% to 94.2%. Further verification through positive and negative directional perturbations, increasing and decreasing the spindle speed by 100 r / min respectively, revealed that the vibration reduction was greatest after decreasing by 100 r / min. Therefore, the direction of decreasing the spindle speed was determined as the final control direction.

[0041] Table 1 Comparison of Monitoring and Control Effects in CNC Milling Machining of Precision Parts

[0042] After the control parameters are determined, they are sent to the CNC milling machine for machining adjustments via a distributed control node, and machining data after adjustment is continuously collected. The cloud platform records vibration data, visual images, and machining quality data before and after adjustment, and performs feedback analysis on the control effect. After continuous testing of multiple machining processes, it was found that the method of this invention can identify tool wear trends in advance, complete parameter adjustments before surface defects form, reduce abnormal machining time, and improve the machining stability of parts. Specific test results are shown in Table 1.

[0043] As can be seen from the above embodiments, the present invention utilizes multi-source data synchronous acquisition, machining state chain analysis, abnormal influencing factor identification, and cloud computing parameter optimization control to achieve real-time monitoring and dynamic adjustment during the CNC milling process of precision parts. Compared with the traditional method that relies on single parameter detection and manual experience adjustment, it has achieved significant improvements in anomaly identification accuracy, machining quality stability, control response speed, and tool utilization efficiency.

[0044] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for monitoring and controlling the machining of precision parts using a CNC milling machine, characterized in that, Includes the following steps: S1. Collect spindle vibration signals, tool force signals, machining trajectory data, and part surface image data during CNC milling. Use narrowband Internet of Things to complete multi-node data synchronous transmission and establish a milling data table according to tool position coordinates and sampling time. S2. Perform frequency domain filtering, noise correction and scale standardization on the milling data table, use industrial vision to enhance the texture and locate the edges of the part surface image, combine the tool trajectory coordinates to complete the division of the machining area, and establish a part feature library. S3. Calculate the vibration difference, load offset, surface texture change and tool wear trend of adjacent machining positions based on the component feature library, and rearrange the features according to the machining path direction to form a milling state chain. S4. Perform reverse tracing calculation on the milling state chain, extract the feature offset relationship before and after abnormal changes, and combine the parameter change law in historical machining samples to determine the influencing factors corresponding to tool abnormalities, machining vibration and surface defects. S5. Based on the influencing factors, the spindle speed, feed rate, depth of cut and tool path are adjusted in combination. The differences in machining response corresponding to different adjustment amounts are analyzed by cloud computing to determine the control parameters of the CNC milling machine. S6. Send the CNC milling machine control parameters to the distributed control node to perform machining adjustments, and record the adjusted vibration data, visual images and machining quality data through the cloud platform to update the precision parts machining monitoring and control process.

2. The method for monitoring and controlling the machining of precision parts by CNC milling as described in claim 1, characterized in that, S1 specifically includes: Collect spindle vibration signals, tool force signals, machining trajectory data, and part surface image data during CNC milling. Use narrowband Internet of Things to complete multi-node data synchronous transmission and establish a milling data table according to tool position coordinates and sampling time. A multi-node synchronous acquisition link is established based on narrowband IoT. A unified time identifier is set for different acquisition nodes. Timing compensation is performed based on data transmission delay, sampling period offset and node response difference. Vibration signal, force signal, trajectory data and visual image are matched according to the same tool position coordinate. Construct milling status marking units. Each milling status marking unit is an independent data record unit formed by binding different types of machining data with tool position coordinates, machining stage and sampling time as indexes. Each milling status marking unit corresponds to a local machining area. Based on the data changes between consecutive tool positions, the vibration change trend, force change trend and surface texture change trend of adjacent positions are calculated. The data sampling density is increased for positions with abrupt changes, and the original sampling frequency is maintained for stable positions, forming a milling data table with machining position correlation.

3. The method for monitoring and controlling the machining of precision parts by CNC milling as described in claim 1, characterized in that, The data processing procedure for the milling data table specifically includes: The spindle vibration signal and tool force signal in the milling data table are segmented according to the tool position coordinates, and the continuous machining process is divided into multiple local cutting windows; Within each local cutting window, the original sampling sequence is subjected to spectral transformation, and the low-frequency stable region, mid-frequency cutting response region, and high-frequency impact region are divided according to the energy ratio in different frequency ranges. For instantaneous peak data in the high-frequency impact zone, the amplitude change direction and duration between adjacent sampling points are combined to determine whether it belongs to cutting impact characteristics. Data with a duration less than the preset window length and not synchronized with the tool position change are marked as interference data, and the interference effect is reduced by frequency weight attenuation. For the mid-frequency cutting response zone, frequency components related to the tool cutting state are retained. The vibration frequency band offset is calculated by using the frequency migration relationship between adjacent machining windows. When there is a continuous frequency offset between adjacent windows, the corresponding frequency change trend is retained and recorded in the milling state marking unit. To address the low-frequency baseline shift caused by sensor drift, a dynamic reference value is established based on the stable sampling interval at the beginning of the processing stage. Baseline correction is then performed on subsequent sampling data to ensure that the vibration amplitude and force changes at different processing stages are within a unified reference range. After completing the frequency domain filtering, the spindle vibration features, tool force features, machining speed features, and visual extraction features are arranged according to the feature dimensions. The data change range within each machining window is calculated using a segmented scale mapping method. Data of different dimensions are converted into a unified feature scale, and the data increase and decrease relationship corresponding to the tool position change is retained to form standardized milling feature data.

4. The method for monitoring and controlling the machining of precision parts by CNC milling as described in claim 1, characterized in that, The process of image processing and processing area division of the component surface specifically includes: The surface images of parts acquired by industrial vision equipment are matched with regions according to the tool trajectory coordinates. The corresponding image regions are divided according to the processing section number. Local brightness compensation and texture gradient enhancement processing are performed on each image region. Pixel correction is performed on areas of uneven lighting in the image, and the local brightness distribution is adjusted according to the gray-scale change relationship of adjacent pixels to eliminate local overexposure areas caused by metal reflection during milling. Texture calculation is performed on the enhanced component surface image according to the pixel neighborhood relationship. Pixel gradient detection paths are established along the horizontal, vertical and 45-degree tilt directions respectively. The gray-level difference and the number of continuous changes between adjacent pixels are calculated in turn. The texture response value in the corresponding direction is determined based on the grayscale change amplitude in different directions. When the grayscale change direction of consecutive pixels is consistent, the corresponding area is marked as the tool motion texture area, and the texture extension direction, continuous length and change intensity are recorded. The texture response results obtained in the three directions are matched with their positions. The response values ​​in different directions are fused and compared, and the direction with the largest response value is retained as the main texture direction of the current region. The distribution of machining marks is determined based on the continuity of the main texture direction. Areas with continuous texture direction and stable intervals are marked as normal machining areas, while areas with abrupt changes in texture direction, continuous length decrease, or abnormal grayscale changes are marked as abnormal surface areas. Edge localization processing is performed on the enhanced image. By detecting continuous pixel change regions, the boundaries of parts, the edges of machining grooves, and the contours of surface defects are determined, and isolated noise points are filtered out based on edge connectivity. The detected edge position is spatially matched with the CNC milling machine tool trajectory coordinates, the positional offset between the tool movement path and the actual machining area is calculated, and the image area coordinates are corrected based on the offset. A visual feature index table for parts is established, and the contour coordinates, texture direction, edge length, grayscale variation range and defect area location corresponding to each processing area are associated and stored to form a part feature library.

5. The method for monitoring and controlling the machining of precision parts by CNC milling as described in claim 1, characterized in that, S3 specifically includes: Based on the machining area coordinates, tool position numbers and machining stage information in the component feature library, the continuous machining areas are sorted by path, and vibration features, load features, visual texture features and tool status features are bound to the corresponding tool position nodes. Read the vibration amplitude sequence, force data and visual texture data of adjacent tool position nodes, calculate the frequency band change, load offset, texture deflection and edge change, and extract the state difference features between the machining positions; Based on the changing trends of cutting pressure, vibration frequency, and surface texture within the continuous machining area, the synchronous changing relationship of multiple features is analyzed to determine the tool wear trend characteristics. The characteristics of various state changes are normalized, and the nodes of continuous change are connected according to the tool movement path. Abnormal offset positions are marked as state change nodes, forming a milling state chain that includes machining position and change trend.

6. The method for monitoring and controlling the machining of precision parts by CNC milling as described in claim 1, characterized in that, S4 specifically includes: Based on the state change nodes corresponding to each machining position in the milling state chain, trace the continuous machining nodes forward from the abnormal position in the opposite direction of the machining path, read the vibration change, load offset, surface texture change and tool wear trend of multiple machining positions before the abnormality occurred, and establish the abnormal forward feature sequence. Time alignment is performed on the feature change process of each feature in the abnormal forward feature sequence. The position, duration and direction of the first shift of different features are recorded. The position of the abnormal occurrence is matched with the historical stable processing state to extract the feature shift relationship before and after the abnormal change. Based on the data of tool usage, cutting load changes, vibration frequency changes, and part surface quality in historical machining samples, a machining state comparison table is established. The feature offset relationship in the current milling state chain is matched with the historical machining samples to select machining samples with the same trend. For the matched historical machining samples, the characteristic change patterns corresponding to tool wear state, machining vibration state and surface defect state are statistically analyzed, and the correlation changes of vibration enhancement, load increase, texture destruction and wear accumulation under different abnormality types are recorded. The influencing factors are sorted according to the degree of feature offset of the current abnormal location. Features that continuously affect the changes in the abnormal state are marked as the main influencing factors, and features with weak correlation are marked as auxiliary influencing factors. Establish an abnormal influence relationship table to associate and store the influencing factors corresponding to tool abnormalities, machining vibrations, and surface defects with the abnormal location, direction of change, and degree of influence.

7. The method for monitoring and controlling the machining of precision parts by CNC milling as described in claim 6, characterized in that, The process of establishing the anomaly forward feature sequence specifically includes: Based on the abnormal change nodes in the milling state chain, obtain the tool position coordinates, machining time and state change type corresponding to the abnormal node, and read the continuous machining nodes sequentially in the opposite direction of tool movement starting from the abnormal node. For each tracking node, the vibration change, load offset, surface texture change, and tool wear trend are extracted synchronously and rearranged according to the time of anomaly occurrence to form a characteristic change trajectory; Differences in characteristic changes between continuously tracked nodes are calculated, and the locations of the first increase in vibration amplitude, the first shift in load change, the first appearance of texture anomaly, and the first increase in wear trend are recorded respectively. The anomaly induction region is determined based on the distance relationship between the first change positions of multiple features. When multiple features change continuously in adjacent processing nodes, the corresponding nodes are divided into anomaly evolution regions. When a single feature undergoes a sudden change while other features remain stable, the corresponding node is marked as a local perturbation region; The sequence, direction, and duration of feature changes within the abnormal evolution region are encoded to form an abnormal forward feature sequence that contains the abnormal generation process.

8. The method for monitoring and controlling the machining of precision parts by CNC milling as described in claim 6, characterized in that, The process of establishing the abnormal influence relationship table specifically includes: Based on the feature offset relationship in the current milling state chain and the abnormal change records in historical machining samples, feature correlation analysis is performed on different anomaly types; Historical machining samples were categorized based on tool abnormalities, machining vibrations, and surface defects. Continuous variation intervals corresponding to vibration changes, load changes, texture changes, and wear changes were extracted from samples of each category. The feature offset data corresponding to the current anomaly location is matched with the historical samples of each category in terms of location and change trend. The order, duration and magnitude of change of each feature under different anomaly types are statistically analyzed. The influence relationship is determined based on the degree of synchronization of feature changes. When a feature change consistently appears before the formation of an anomaly and increases synchronously with the degree of anomaly, the feature is recorded as a highly correlated influence factor. When a feature changes only after an anomaly occurs, that feature is recorded as a result response factor. An anomaly impact relationship table is established based on the category of influencing factors, anomaly type, location of change, and degree of correlation, so that a corresponding relationship is formed between changes in tool condition, changes in cutting vibration, and changes in surface quality.

9. The method for monitoring and controlling the machining of precision parts by CNC milling as described in claim 1, characterized in that, S5 specifically includes: Based on the ranking of influencing factors in the abnormal influence relationship table, the abnormal types corresponding to tool abnormalities, machining vibrations and surface defects are associated and matched with spindle speed, feed rate, depth of cut and tool path to determine the control parameters to be adjusted; Extract the vibration changes, load offsets, texture changes, and wear trends before and after the anomaly occurred, and combine them with parameter response data from historical processing samples to establish processing response records corresponding to different combinations of control parameters. The response analysis of various control parameter combinations is performed through cloud computing. A comprehensive response score is calculated based on the degree of vibration suppression, load stability, surface quality change, and processing efficiency change to select the optimal parameter combination. The optimal parameter combination is verified by positive and negative disturbances. Data on the changes in machining state before and after the disturbance are collected. Based on the response differences, the adjustment direction and adjustment range of spindle speed, feed rate, depth of cut and tool path are determined.

10. A monitoring and control system for CNC milling machining of precision parts, comprising the monitoring and control method for CNC milling machining of precision parts as described in any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to collect spindle vibration signals, tool force signals, machining trajectory data and part surface image data during the CNC milling process. It uses narrowband Internet of Things to complete multi-node data synchronous transmission and establishes a milling data table according to the tool position coordinates and sampling time. The data processing module is used to perform frequency domain filtering, noise correction and scale normalization on the milling data table, use industrial vision to enhance the texture and locate the edges of the part surface image, combine the tool trajectory coordinates to complete the division of the machining area, and establish a part feature library. The state analysis module is used to calculate the vibration difference, load offset, surface texture change and tool wear trend between adjacent machining positions based on the component feature library, and rearrange the features according to the machining path direction to form a milling state chain. The anomaly identification module is used to perform reverse tracing calculations on the milling state chain, extract the feature offset relationship before and after the abnormal change, and combine the parameter change patterns in historical machining samples to determine the influencing factors corresponding to tool anomalies, machining vibrations and surface defects. The parameter optimization module is used to combine and adjust the spindle speed, feed rate, depth of cut and tool path according to the influencing factors. It uses cloud computing to analyze the differences in machining response corresponding to different adjustment amounts and determine the control parameters of the CNC milling machine. The monitoring and control module is used to send the control parameters of the CNC milling machine to the distributed control node to perform machining adjustments, and to record the adjusted vibration data, visual images and machining quality data through the cloud platform to update the monitoring and control process of precision parts machining.

Images