An artificial intelligence-based CNC machining optimization method for mobile phone middle frames

Abstract

The present application provides an artificial intelligence-based CNC machining optimization method for a mobile phone middle frame, comprising: performing real-time scanning on the surface of the mobile phone middle frame workpiece to obtain initial data on the material hardness distribution, forming a hardness distribution map, and obtaining a regional division result of hardness change; identifying the hardness grade distribution and the cutting force mutation amplitude at a specific location where the cutting force suddenly changes, calculating adjustment parameters for the tool feed speed based on the hardness grade distribution and the cutting force mutation amplitude, and generating a speed adjustment instruction based on the adjustment parameters; dynamically matching the cutting depth parameter based on the speed adjustment instruction to determine the target cutting depth configuration, and matching the cutting path deviation monitoring value, the vibration amplitude index, and the machining accuracy deviation in real time through the target cutting depth configuration.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to an artificial intelligence-based CNC machining optimization method for a mobile phone middle frame. Background Art

[0002] In modern manufacturing, the machining quality of mobile phone midframes is directly related to the product's appearance and performance, making it a crucial research area in the field of intelligent manufacturing. With the introduction of artificial intelligence (AI) technology, optimized cutting path planning is considered a key approach to improving machining efficiency and surface quality, and holds irreplaceable value in driving the manufacturing industry's transition toward high-precision, high-efficiency manufacturing. However, current approaches still have significant shortcomings when addressing complex machining environments. Many solutions often overlook the dynamic impact of varying workpiece material properties on the machining process. In particular, when faced with uneven material hardness distribution, the lack of real-time adaptive adjustment of machining parameters leads to uncontrollable risks during machining, which in turn affects the quality consistency of the final product. Focusing on specific challenges, uneven material hardness distribution significantly interferes with cutting forces. These sudden force fluctuations can easily cause tool deflection and disrupt machining path stability. Fluctuating cutting forces increase the difficulty of chip removal during machining, leading to more scratches on the machined surface and severely compromising surface quality. For example, when a tool cuts from a softer aluminum alloy region into a harder, reinforced region, accumulated metal chips cannot be quickly removed, resulting in drag marks on the workpiece surface and a sharp deterioration in surface roughness. These two factors are interrelated and together constitute difficult-to-overcome technical difficulties in the machining process, making how to maintain machining accuracy and surface consistency in a dynamic environment a core issue that needs to be solved urgently.

[0003] Therefore, how to effectively deal with the problems of sudden changes in cutting force and difficulty in chip discharge by dynamically adjusting the tool feed speed and cutting depth parameters under the condition of uneven material hardness distribution, while taking into account the overall processing efficiency and consistency of surface roughness, becomes a key issue. Summary of the Invention

[0004] The present invention provides an artificial intelligence-based CNC machining optimization method for a mobile phone middle frame, which mainly includes:

[0005] The surface of the mobile phone middle frame workpiece is scanned in real time to obtain initial data on the material hardness distribution, form a hardness distribution map, and obtain the regional division results of hardness changes; a cutting force monitoring benchmark is established based on the regional division results of hardness changes, and cutting force fluctuation data during processing is collected. The abnormal fluctuation area is identified based on the cutting force monitoring benchmark, and the specific location and time point of the cutting force mutation is determined based on the material property differences and the cutting force threshold standard in the hardness distribution map; the hardness grade distribution and cutting force mutation amplitude at the specific location of the cutting force mutation are identified, and the adjustment parameters of the tool feed speed are calculated based on the hardness grade distribution and cutting force mutation amplitude. The speed adjustment instruction is generated based on the adjustment parameters; the cutting depth parameter is dynamically matched according to the speed adjustment instruction to determine the target cutting depth configuration. The cutting path deviation monitoring value, vibration amplitude index and processing accuracy deviation are matched in real time through the target cutting depth configuration; the offset risk in the processing process is predicted based on the regional division results of the hardness distribution map and the depth change law of the target cutting depth configuration, the tool path is corrected, and a corrected tool path trajectory is obtained. The complexity of the corrected trajectory and the cutting load change are analyzed by combining the cutting path deviation monitoring value and vibration amplitude index; The auxiliary system parameters related to chip discharge efficiency are adjusted according to the corrected trajectory of the tool path. The cleaning frequency and airflow intensity of the chip discharge channel are optimized according to the complexity of the corrected trajectory and the change of cutting load. The chip accumulation risk is evaluated by monitoring the chip accumulation amount and discharge efficiency. If the chip accumulation risk is lower than the preset risk threshold, the target machining accuracy correction plan is determined in combination with the machining accuracy deviation.

[0006] Furthermore, the real-time scanning of the surface of the mobile phone middle frame workpiece to obtain initial data of the material hardness distribution, forming a hardness distribution map, and obtaining a regional division result of hardness variation includes:

[0007] The surface of the mobile phone middle frame workpiece is detected point by point through a sensor array to obtain the hardness value of each position, forming a data set containing position coordinates and hardness values; the hardness values ​​in the data set are spatially interpolated, and the hardness values ​​between the detection points are calculated using the inverse distance weighted method to generate continuous hardness distribution data; a hardness distribution map is generated based on the continuous hardness distribution data, and the hardness difference between adjacent positions in the hardness distribution map is calculated to determine the hardness area boundary, and the boundary points are connected to form the outline of the hardness change area.

[0008] Furthermore, the method includes establishing a cutting force monitoring benchmark based on the regional division results of hardness changes, collecting cutting force fluctuation data during machining, identifying abnormal fluctuation areas based on the cutting force monitoring benchmark, and determining the specific location and time of the cutting force mutation based on the material property differences in the hardness distribution map and the cutting force threshold standard, including:

[0009] According to the regional division results of the hardness distribution map, the cutting force reference value of each area is set, and a monitoring benchmark table containing position coordinates and cutting force range is generated; the cutting force time series data is collected by a force sensor, and the cutting force time series data is compared with the cutting force range in the monitoring benchmark table to identify abnormal points and record the position and time of the abnormal points; according to the abnormal points and the hardness distribution map, the cutting force change rate is calculated to determine the position of the cutting force mutation.

[0010] Furthermore, the identifying of abnormal fluctuation areas according to the cutting force monitoring benchmark includes:

[0011] The cutting force component data of the tool is collected through a three-dimensional force sensor, and the three-dimensional cutting force reference value of each area is set in combination with the regional division result of the hardness distribution map; the cutting force component data is compared with the three-dimensional cutting force reference value, the deviation ratio is calculated, and the position and time of the abnormal point exceeding the threshold are recorded; the cutting force change amplitude is calculated based on the abnormal point, the abnormal point density per unit area is counted, and the cutting force abnormal area is divided.

[0012] Furthermore, the identification of the hardness grade distribution and the cutting force mutation amplitude at a specific location where the cutting force mutation occurs, calculating an adjustment parameter for the tool feed speed based on the hardness grade distribution and the cutting force mutation amplitude, and generating a speed adjustment instruction based on the adjustment parameter, includes:

[0013] The hardness grade and cutting force variation range of the cutting force mutation position are extracted, and the hardness and speed correspondence table is queried to determine the feed speed adjustment ratio; the acceleration limit value is calculated according to the cutting force variation range, and the regional transition distance is calculated according to the hardness grade difference; and an adjustment instruction including speed value and acceleration parameters is generated according to the feed speed adjustment ratio, the acceleration limit value and the regional transition distance.

[0014] Furthermore, the method of dynamically matching cutting depth parameters according to the speed adjustment instruction, determining a target cutting depth configuration, and matching cutting path deviation monitoring values, vibration amplitude indicators, and machining accuracy deviations in real time through the target cutting depth configuration includes:

[0015] According to the speed value in the speed adjustment instruction, the speed and depth correspondence table is queried, the cutting depth value is calculated, and the cutting depth configuration data is generated; the path deviation and vibration threshold are set according to the cutting depth configuration data, the tool path deviation and vibration amplitude are measured by the sensor, and the monitoring value matching the cutting depth configuration data is generated.

[0016] Furthermore, the method predicts the offset risk during the machining process based on the regional division result of the hardness distribution map and the depth variation law of the target cutting depth configuration, corrects the tool path, obtains the corrected trajectory of the tool path, and combines the cutting path deviation monitoring value and the vibration amplitude index analysis to obtain the complexity of the corrected trajectory and the cutting load change, including:

[0017] The hardness change rate is calculated according to the hardness distribution map, the depth change rate is calculated according to the cutting depth configuration, and the offset risk coefficient is generated by combining the hardness change rate and the depth change rate to mark the high-risk points; based on the high-risk points and the theoretical path, the path compensation vector is calculated, and the pre-compensation path is generated by superposition, and the corrected trajectory is generated by real-time correction; the curvature change rate is calculated according to the corrected trajectory, and the trajectory complexity is quantified by combining the cutting path deviation and the vibration amplitude.

[0018] Furthermore, the auxiliary system parameters related to chip removal efficiency are adjusted according to the corrected trajectory of the tool path, the cleaning frequency and airflow intensity of the chip removal channel are optimized according to the complexity of the corrected trajectory and the change in cutting load, and the chip accumulation risk is evaluated by monitoring the chip accumulation amount and discharge efficiency, including:

[0019] According to the curvature data of the corrected trajectory, the curvature and chip generation rate table is queried to determine the chip generation rate, and the coolant flow, vacuum cleaner power and chip collector frequency are adjusted; according to the complexity of the corrected trajectory, the cleaning time interval and air flow velocity are calculated; the chip accumulation rate is measured by a detection device, and the chip accumulation rate is compared with a threshold value to evaluate the accumulation risk.

[0020] Furthermore, if the chip accumulation risk is lower than a preset risk threshold, a target machining accuracy correction scheme is determined in combination with the machining accuracy deviation, including:

[0021] According to the accumulation risk being lower than the threshold, a correction coefficient is calculated in combination with the machining accuracy deviation, and a correction coefficient and compensation parameter table is queried to generate a target machining accuracy correction plan including a compensation value and an adjustment ratio.

[0022] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0023] The present invention discloses an artificial intelligence-based CNC machining optimization method for mobile phone midframes. This method uses a sensor array to scan the workpiece surface in real time to obtain a hardness distribution map, establish a cutting force monitoring benchmark, and identify areas of abnormal fluctuations. The method then calculates tool feed rate adjustment parameters based on the hardness distribution and cutting force mutations. The method dynamically matches the cutting depth and predicts offset risks, corrects the tool path, and optimizes chip removal efficiency. This method can adjust machining parameters in real time based on differences in workpiece material properties, effectively addressing cutting force fluctuations caused by hardness changes, improving machining accuracy and efficiency, reducing tool wear, and enabling intelligent machining process control. By continuously updating the parameter database, the present invention can continuously optimize machining plans, improving the machining quality and production efficiency of complex workpieces. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a flow chart of an artificial intelligence-based CNC machining optimization method for a mobile phone middle frame of the present invention.

[0025] Figure 2 This is a schematic diagram of an artificial intelligence-based CNC machining optimization method for a mobile phone middle frame according to the present invention. DETAILED DESCRIPTION

[0026] To help those skilled in the art better understand the technical solutions in this specification, the following will provide a clear and complete description of the technical solutions in the embodiments of this specification, in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. All other embodiments derived by those skilled in the art based on the embodiments in this specification without creative effort shall fall within the scope of protection of this specification.

[0027] like Figure 1-2 In this embodiment, a method for optimizing CNC machining of a mobile phone middle frame based on artificial intelligence may specifically include:

[0028] In step S101 , the surface of the mobile phone middle frame workpiece is scanned in real time to obtain initial data of the material hardness distribution, form a hardness distribution map, and obtain a regional division result of hardness variation.

[0029] The sensor array performs point-by-point inspection of the surface of the mobile phone's midframe workpiece along a preset scanning path. Each sensor node obtains a material hardness feedback value at the corresponding position through piezoelectric effect or electromagnetic induction. The pressure value or induced current value measured by the sensor is directly converted into a hardness value, forming an original data set containing position coordinates and hardness values. The hardness values ​​of each detection point in the original data set are spatially interpolated. Based on the hardness values ​​and spatial distances of adjacent detection points, the inverse distance weighted interpolation method is used to calculate the hardness estimates of the positions between the detection points, generating continuous hardness distribution data covering the entire workpiece surface. A hardness distribution map is constructed based on the continuous hardness distribution data, and the hardness difference between each position point in the map and its eight adjacent positions is calculated. If the hardness difference exceeds the preset hardness change threshold, the position is determined to be a boundary point of the hardness region. All boundary points are connected to form a closed hardness region contour, and the regional division result of the hardness change is obtained.

[0030] Specifically, the scanning detection of the sensor array is achieved through the collaborative work of multiple sensor nodes.

[0031] Specifically, the sensor array uses a matrix layout, with each sensor node maintaining a fixed spacing of 2-5 mm. When the sensor contacts the workpiece surface, the piezoelectric sensor generates a charge signal through the force applied to the piezoelectric crystal. The amount of charge is proportional to the applied pressure, which in turn reflects the hardness characteristics of the material. Electromagnetic induction sensors, on the other hand, emit high-frequency electromagnetic fields and determine the hardness of the material based on its response to the electromagnetic field. The harder the material, the greater its impedance to the electromagnetic field, and the smaller the induced current.

[0032] In one possible implementation, the scanning path uses a serpentine trajectory, starting from one end of the workpiece, moving horizontally to the other end, then moving longitudinally by one sensor spacing, and then scanning in the opposite direction horizontally, repeating this process until the entire surface is covered. This scanning method ensures the continuity and integrity of the detection. The hardness value of each detection point is obtained by looking up a pre-calibrated conversion table, which records the hardness value relationship corresponding to different pressure values ​​or induced current values. The core of the spatial interpolation processing is to use the data of known detection points to infer the hardness value of undetected locations. The basic principle of the inverse distance weighted interpolation method is that the closer the points are, the greater the influence on the interpolation result.

[0033] For example, to calculate the hardness of an untested point P, first determine the eight closest test points around it, calculate the distance from point P to each test point, use the inverse of the distance as a weight, multiply the hardness value of each test point by the corresponding weight, and then sum them to obtain the estimated hardness value of point P. This method ensures a smooth transition in the interpolation result and avoids sudden changes.

[0034] It should be noted that the construction of a hardness distribution map actually converts discrete hardness data into a continuous two-dimensional image. Each pixel in the map corresponds to a tiny area on the workpiece surface, and the pixel value represents the hardness of that area. Color coding, such as dark blue for low-hardness areas, dark red for high-hardness areas, and intermediate colors for medium-hardness, provides a visual representation of the hardness distribution. Hardness differences are calculated using an eight-neighborhood comparison method, which calculates the absolute difference in hardness between the center point and its eight surrounding points. The preset hardness change threshold is typically set at 10%-15% of the average hardness value; when the difference exceeds this threshold, a significant hardness change is indicated. Boundary points are connected using an edge tracking algorithm. Starting from a boundary point, adjacent boundary points are searched in a clockwise or counterclockwise direction until a closed contour is formed. This region segmentation method accurately identifies areas with different hardness characteristics, providing precise location information for subsequent differentiated processing, improving processing efficiency and product quality.

[0035] Step S102: Establish a cutting force monitoring benchmark based on the regional division results of hardness changes, collect cutting force fluctuation data during processing, identify abnormal fluctuation areas based on the cutting force monitoring benchmark, and determine the specific location and time point of the cutting force mutation based on the material property differences in the hardness distribution map and the cutting force threshold standard.

[0036] Based on the hardness zone classification results, a corresponding cutting force reference value is set for each hardness zone. A pre-established database of hardness and cutting force correspondences is used to obtain standard cutting force values ​​corresponding to different hardness values. A cutting force monitoring benchmark table containing upper and lower cutting force limits for each zone is established. This monitoring benchmark table records the position coordinate range of each zone and the corresponding allowable cutting force fluctuation range. A force sensor is used to collect cutting force values ​​during machining in real time, acquiring cutting force time series data at a rate of 100 times per second. Based on the tool's current machining position, the allowable cutting force range for the corresponding zone is read from the cutting force monitoring benchmark table. If the measured cutting force exceeds the allowable range, the moment is identified as a fluctuation anomaly. The location coordinates and timestamp of the fluctuation anomaly point are recorded to form a fluctuation anomaly zone distribution map. Based on the fluctuation anomaly point data in the fluctuation anomaly zone distribution map and the material hardness values ​​at the corresponding location in the hardness distribution map, the cutting force change rate at the fluctuation anomaly point is calculated by dividing the cutting force difference between adjacent moments by the time interval. If the change rate exceeds 50% of the upper limit of the allowable cutting force fluctuation range, it is identified as a sudden change in cutting force. The location coordinates and timestamp of the fluctuation anomaly point are used to determine the specific location and time of the sudden change in cutting force.

[0037] Specifically, the establishment of the database corresponding to hardness and cutting force is based on the principles of material mechanics and a large amount of processing experimental data.

[0038] Specifically, materials of varying hardness exhibit varying shear strength during cutting, with harder materials requiring greater cutting force for effective cutting. This database records the standard cutting forces at varying hardness levels by conducting cutting experiments on common mobile phone midframe materials, including aluminum alloy, stainless steel, and titanium alloy.

[0039] For example, the standard cutting force of aluminum alloy material with a hardness of HRC20 is about 50N, while the standard cutting force of stainless steel material with a hardness of HRC40 can reach 150N.

[0040] In one possible implementation, the cutting force monitoring benchmark table is constructed by gridding the workpiece surface according to hardness zones. Each grid cell corresponds to a set of position coordinate ranges, such as a rectangular area measuring 10-15 mm on the X axis and 20-25 mm on the Y axis. Based on the average hardness value within this area, the corresponding standard cutting force is retrieved from a database, with a 20% fluctuation range as the allowable fluctuation range. This approach ensures that each machining position has a clear cutting force reference standard. The real-time force sensor acquisition process relies on the operating principle of a piezoelectric sensor. When the tool cuts the workpiece, the reaction force is transmitted through the tool holder to the sensor's piezoelectric crystal, which generates a charge signal proportional to the pressure. A sampling frequency of 100 times per second means that the cutting force value is recorded every 10 milliseconds. This high-frequency sampling can capture transient changes during the cutting process. The collected data forms a time series, with each data point containing a timestamp, a cutting force value, and the corresponding tool position coordinates.

[0041] It should be noted that the process of identifying fluctuation anomalies is actually a real-time comparison process. When the tool moves to a certain position, the control system immediately extracts the corresponding allowable range of cutting force from the monitoring reference table. If the allowable range at a certain position is 80-120N, and the measured value is 135N, then this point is marked as a fluctuation anomaly. These anomalies form a distribution map in the two-dimensional coordinate system. Areas with dense anomalies often indicate local changes in material properties or mismatches in machining parameters. The cutting force change rate is calculated using a differential method.

[0042] For example, if the cutting force at time t1 is 100N and at time t2 is 140N, with a time interval of 0.01 seconds, the rate of change is 4000N / s. When this rate of change exceeds 50% of the upper limit of the allowable fluctuation range—that is, 50% of 60N is 30N, and a change of more than 30N in 0.01 seconds—it is determined to be a sudden change. This sudden change often occurs when the tool enters a hard area from a soft area or encounters internal material defects. Through this multi-level monitoring and judgment mechanism, the location and time of the sudden change in cutting force can be precisely determined. This information is of great value for optimizing subsequent processing parameters, preventing tool damage, and improving processing quality. This real-time monitoring and anomaly identification capability significantly improves the controllability and stability of the machining process, especially when processing complex workpieces with uneven hardness distribution.

[0043] Continuously monitor the real-time changes of the three-dimensional cutting force components during the tool cutting process, set the cutting force monitoring benchmark values ​​corresponding to different hardness levels based on the hardness change area division results, compare the collected cutting force data with the monitoring benchmark to obtain the degree of deviation, identify the time period and spatial position where the cutting force value exceeds the benchmark range, mark the abnormal fluctuation points where the cutting force suddenly increases or decreases, screen out the areas with abnormally increased cutting resistance and unstable cutting force, and form the position coordinates of the fluctuation abnormal area and the abnormality degree classification.

[0044] A three-axis force sensor continuously collects cutting force components along the X, Y, and Z axes. Based on the hardness grade of each region in the hardness variation region classification results, the corresponding three-axis cutting force monitoring benchmark values ​​are obtained from a hardness-cutting force correspondence table established based on experimental data, forming a monitoring benchmark dataset containing position coordinates and three-axis force benchmark values. The real-time three-axis cutting force data are compared with the benchmark values ​​at the corresponding locations in the monitoring benchmark dataset. The deviation ratio of the cutting force in each direction is calculated by subtracting the benchmark value from the measured value and then dividing it by the benchmark value. If the deviation ratio in any direction exceeds a preset deviation threshold, the timestamp and tool position coordinates at that moment are recorded to form a set of abnormal data points outside the benchmark range. Based on the position and time information in the abnormal data point set, the variation in cutting force between adjacent sampling moments is calculated. If the variation exceeds a preset ratio threshold of the current benchmark value, it is marked as an abnormal fluctuation point. The abnormal density is determined by counting the number of abnormal fluctuation points per unit area. Regions of abnormally increased cutting resistance and unstable cutting force are identified based on the abnormal density and the degree of cutting force deviation. Based on the identified areas of abnormally increased cutting resistance and unstable cutting force, the cutting force deviation ratio in each area is compared with the preset grading threshold. Deviation ratios less than the first grading threshold are marked as mild abnormalities, deviation ratios between the first grading threshold and the second grading threshold are marked as moderate abnormalities, and deviation ratios exceeding the second grading threshold are marked as severe abnormalities, forming a complete record containing the location coordinates of the fluctuation abnormal area and the abnormality degree classification.

[0045] Specifically, the working principle of the three-axis force sensor is based on strain gauge measurement technology, and strain gauge groups are arranged inside the tool handle along the three orthogonal directions of X, Y, and Z.

[0046] Specifically, when cutting forces act on the tool, the toolholder deforms slightly, causing the strain gauge resistance to change accordingly. This resistance change is converted into a voltage signal via a Wheatstone bridge circuit, enabling independent measurement of the cutting force components in three directions. The X-axis primarily reflects the feed resistance, the Y-axis corresponds to the main cutting force, and the Z-axis represents the axial force. Together, these three components constitute the complete force state of the cutting process.

[0047] In one possible implementation, the process of establishing a hardness-cutting force correspondence table involves systematic cutting experiments. For material samples of different hardness grades, such as HRC20, HRC30, and HRC40 metal materials, processing experiments are carried out under the same cutting parameters, and the average values ​​of the three-axis forces under stable cutting conditions are recorded. Experimental data show that for every HRC10 increase in hardness, the main cutting force increases by approximately 35%, the feed resistance increases by 25%, and the axial force increases by 20%. This differentiated growth pattern forms the basis of the monitoring benchmark data set. The deviation ratio is calculated using the relative deviation method. The specific calculation process is: first obtain the measured cutting force value at the current position, find the benchmark value corresponding to that position, then calculate the difference between the measured value and the benchmark value, and then divide it by the benchmark value to obtain the deviation ratio.

[0048] For example, if the Y-axis reference value at a certain location is 100 N and the measured value is 135 N, the deviation ratio is 0.35. This relative deviation can eliminate the influence of the difference in reference values ​​in different hardness areas and achieve a unified abnormality judgment standard.

[0049] It should be noted that the statistical method for determining abnormal density directly impacts the accuracy of identifying abnormal areas. A unit area is typically defined as a 10mm x 10mm grid, within which the number of abnormal fluctuation points is counted. When the number of abnormal points exceeds five, the area is considered to have a cutting anomaly. Areas of abnormally increased cutting resistance typically manifest as a simultaneous increase in all three forces, while areas of unstable cutting force manifest as frequent force fluctuations exceeding 20% ​​of the baseline value. The thresholds for grading abnormality severity are based on extensive machining experience. The first grading threshold is typically set at 0.3, and the second grading threshold at 0.6. Mild anomalies are often caused by localized hardness fluctuations in the material, moderate anomalies may stem from internal inclusions or structural inhomogeneities, and severe anomalies indicate the presence of defects such as cracks and pores. This grading mechanism provides a clear basis for subsequent adjustments to machining parameters. This multi-dimensional monitoring and analysis approach enables precise monitoring of the cutting process. Accurate identification of abnormal areas not only helps prevent tool breakage but also guides optimization of cutting parameters and improves machining quality. Especially when processing high-value workpieces, this real-time monitoring and graded early warning mechanism significantly reduces processing risks and improves production efficiency.

[0050] Step S103, identifying the hardness grade distribution and the cutting force mutation amplitude at the specific position of the cutting force mutation, calculating the adjustment parameters of the tool feed speed based on the hardness grade distribution and the cutting force mutation amplitude, and generating a speed adjustment instruction based on the adjustment parameters.

[0051] The corresponding position coordinates are extracted from the records of specific locations and time points where the cutting force suddenly changes. The hardness grade at that location is retrieved from the hardness distribution map. The cutting force values ​​before and after the sudden change are obtained from the cutting force monitoring data. The sudden change amplitude of the cutting force is calculated by subtracting the value before the sudden change from the value after the sudden change. This generates a feature dataset containing the sudden change position coordinates, hardness grade, and sudden change amplitude. Based on the hardness grade in the feature dataset, the feed rate reduction ratio is determined by querying an experimentally established table of hardness and speed correspondences. The acceleration limit value is calculated by dividing the sudden change amplitude by the baseline cutting force and then multiplying it by a preset coefficient. The hardness zone transition buffer distance is calculated by multiplying the hardness grade difference between adjacent hardness zones by a preset distance coefficient. Based on the feed rate reduction ratio, acceleration limit value, and hardness zone transition buffer distance, the current feed rate is multiplied by the reduction ratio to obtain the target feed rate. The acceleration limit value is converted into the upper limit of the speed change per unit time. The speed gradient value is calculated based on the buffer distance and speed difference. A speed adjustment instruction containing the speed value, acceleration limit, and gradient parameters is generated according to the standard CNC code format.

[0052] Specifically, the identification of the location of the cutting force mutation depends on the monitoring system established in the early stage.

[0053] Specifically, when the tool enters a high hardness area from a low hardness area, the cutting resistance increases sharply in a very short time, and this change is captured in real time by the force sensor.

[0054] For example, when machining an aluminum alloy substrate, the cutting force is 80N. Upon entering the embedded stainless steel reinforcement area, the cutting force instantly rises to 200N, resulting in a sudden change of 120N. This sudden change not only reflects the difference in material properties but also indicates the need for subsequent adjustment of machining parameters. The hardness-speed correspondence table is established based on cutting theory and experimental verification.

[0055] In one possible implementation, a series of cutting tests are conducted on materials of different hardnesses to record the optimal feed rate while ensuring machining quality. The experiments show that when the material hardness increases from HRC20 to HRC40, the reasonable feed rate should decrease from 300 mm per minute to 150 mm per minute, forming a corresponding rule that the feed rate decreases by 25% for every HRC10 increase in hardness. This decreasing ratio ensures the cutting stability of the tool in high-hardness areas. The calculation process of the acceleration limit value involves the principle of dynamic balance. The ratio of the mutation amplitude to the baseline cutting force reflects the severity of the cutting conditions. The larger the ratio, the more stringent speed control is required. The preset coefficient is usually between 0.5 and 0.8, which is used to convert the dimensionless ratio into a physically meaningful acceleration value.

[0056] For example, the mutation amplitude is 120N, the baseline cutting force is 100N, the ratio is 1.2, and multiplying by the coefficient 0.6 gives an acceleration limit value of 0.72 meters per square second, which means that the speed change must be completed within this limit.

[0057] It should be noted that the setting of the buffer distance for hardness zone transitions directly impacts the smoothness of the machining process. The greater the difference in hardness levels between adjacent zones, the longer the required buffer distance. The distance coefficient is selected based on factors such as tool diameter and material properties, and is generally between 2 and 5 mm per HRC unit. If the hardness difference between adjacent zones is HRC 15 and the distance coefficient is 3 mm per HRC, the buffer distance is 45 mm. Within this distance, the feed rate is gradually adjusted to avoid the impact of sudden speed changes on machining quality. The calculation of the speed gradient ensures the continuity of speed adjustment. The speed change per unit distance is calculated by dividing the difference between the target speed and the current speed by the buffer distance. The standard CNC code format follows ISO standards, and the generated instructions consist of a combination of G codes and F codes. G01 represents linear interpolation, and the F value defines the feed rate. By inserting multiple intermediate speed values ​​within the buffer zone, a smooth speed transition is achieved. This adaptive speed adjustment mechanism, based on hardness distribution and sudden changes in cutting forces, significantly improves the machining quality of workpieces made of complex materials. By precisely controlling the change process of the feed speed, it not only protects the tool from impact damage but also ensures the consistency of the processed surface. It is particularly suitable for the manufacture of precision parts such as the middle frames of high-end mobile phones.

[0058] Step S104, dynamically matching cutting depth parameters according to the speed adjustment instruction, determining the target cutting depth configuration, and matching the cutting path deviation monitoring value, vibration amplitude index and machining accuracy deviation in real time through the target cutting depth configuration.

[0059] Based on the feed rate value in the speed adjustment command, the recommended cutting depth at that speed is obtained by querying a speed-depth correspondence table established based on cutting experiments. The actual cutting depth is calculated by multiplying the ratio of the current feed rate to the standard feed rate by the recommended cutting depth, generating target cutting depth configuration data containing the cutting depth values ​​for each machining position. Monitoring parameter thresholds are set based on the target cutting depth configuration data. A laser displacement sensor is used to measure the distance deviation between the tool cutting trajectory and the theoretical path. As the cutting depth increases, the allowable deviation range is increased accordingly to obtain a cutting path deviation monitoring value that matches the cutting depth. An acceleration sensor is used to collect vibration signals and calculate their root mean square value. The vibration amplitude warning line is adjusted based on the cutting depth to obtain a vibration amplitude index. Based on the real-time data of the cutting path deviation monitoring value and the vibration amplitude index, combined with the current cutting depth value, the machined surface dimensions are detected using an online measurement device. The difference between the measured dimension and the designed dimension is divided by the cutting depth to obtain the relative deviation rate. When the relative deviation rate exceeds the preset threshold, it is recorded as the machining accuracy deviation, thus achieving a dynamic matching relationship between the target cutting depth configuration and the three monitoring indicators.

[0060] Specifically, the establishment of the speed-depth correspondence table is based on the basic principles of cutting processing.

[0061] Specifically, as feed rate decreases, the amount of material removed per unit time decreases. To maintain machining efficiency, the depth of cut can be appropriately increased. This relationship was established through a large number of cutting experiments, in which the cutting force was kept constant and the optimal cutting depth at different speeds was recorded.

[0062] For example, a standard feed rate of 200 mm / min corresponds to a cutting depth of 0.5 mm. When the speed is reduced to 100 mm / min, the cutting depth can increase to 0.8 mm, forming an inverse proportional relationship.

[0063] In one possible implementation, the ratio calculation method reflects the continuity of parameter adjustment. The current feed rate is 150 mm / min, the standard speed is 200 mm / min, and the ratio is 0.75. The recommended cutting depth of 0.5 mm is multiplied by the inverse of the ratio 1.33, which gives an actual cutting depth of 0.67 mm. This calculation method ensures the relative stability of the material removal rate and avoids large fluctuations in processing efficiency due to speed changes. The working principle of the laser displacement sensor is based on triangulation. The sensor emits a laser beam to the surface of the workpiece, and the reflected light is captured by the receiver. The distance is calculated by the change in the spot position. During the cutting process, the sensor measures the distance from the bottom of the tool to the surface of the workpiece in real time, and compares it with the theoretical path height to obtain the deviation value. The greater the cutting depth, the larger the allowable path deviation range. This is because the amount of material removed during deep cutting is large, and small deviations have relatively little effect on the final accuracy.

[0064] It should be noted that calculating the RMS value of vibration amplitude involves signal processing. Accelerometers collect vibration acceleration data at a high frequency, typically collecting thousands of data points per second. For a data series over a period of time, the square of each data point is calculated, the sum is divided by the number of data points, and the square root is taken to obtain the RMS value. This value reflects the average level of vibration energy and is more representative of the overall vibration state than the instantaneous peak value. The introduction of the relative deviation rate achieves a unified evaluation standard for different cutting depths.

[0065] For example, when the cutting depth is 0.5 mm, the difference between the measured and designed dimensions is 0.01 mm, and the relative deviation rate is 2%; when the cutting depth is 1 mm, the same 0.01 mm dimension difference has a relative deviation rate of only 1%. This normalization process makes the accuracy evaluation under different processing conditions comparable. The core of the dynamic matching relationship lies in establishing a functional mapping between the cutting depth and various monitoring indicators. As the cutting depth increases, the allowable range of path deviation expands linearly, the vibration amplitude warning line increases according to the square root relationship, and the accuracy deviation threshold remains relatively constant.

[0066] In step S105, the offset risk in the machining process is predicted based on the regional division results of the hardness distribution map and the depth variation law of the target cutting depth configuration, the tool path is corrected, and a corrected trajectory of the tool path is obtained. The complexity of the corrected trajectory and the cutting load change are obtained by combining the cutting path deviation monitoring value and the vibration amplitude index analysis.

[0067] Based on the hardness grade values ​​of each area in the hardness distribution map, the hardness change rate is obtained by calculating the hardness difference between adjacent areas and dividing it by the area spacing. At the same time, the depth change rate is obtained by calculating the depth difference between adjacent positions and dividing it by the position spacing from the target cutting depth configuration. The offset risk coefficient is obtained by multiplying the hardness change rate and the depth change rate. If the offset risk coefficient exceeds the preset threshold, the position is marked as a high-risk offset point, forming risk distribution data containing the offset risk level. Based on the high-risk offset point position in the risk distribution data and the preset theoretical processing path, the cubic spline interpolation algorithm is used to calculate the path compensation vector at the high-risk point based on the coordinates and tangent direction of the previous and next normal path points. The pre-compensated path is obtained by superimposing the compensation vector on the theoretical path coordinates. At the same time, according to the deviation between the actual position of the tool obtained during the processing and the pre-compensated path, real-time correction is performed to generate a corrected trajectory for the tool path. The local curvature is obtained by calculating the inverse of the radius of the arc formed by three consecutive path points in the corrected trajectory. The curvature change rate is obtained by dividing the difference between adjacent curvature values ​​by the path length. The curvature change rate is used as an indicator of trajectory complexity. Combined with the cutting path deviation monitoring value and the vibration amplitude indicator, the dynamic change of the cutting load is quantified by multiplying the curvature change rate by the vibration amplitude increment to obtain the complexity of the corrected trajectory and the cutting load change data.

[0068] Specifically, the calculation of hardness change rate and depth change rate reflects the coupling relationship between material properties and processing parameters.

[0069] Specifically, when the tool transitions from an aluminum alloy zone with a hardness of HRC20 to a stainless steel zone with a hardness of HRC35, with a 5 mm interval, the hardness changes at a rate of 3 HRC / mm. Simultaneously, the depth of cut increases from 0.5 mm to 0.8 mm, with a rate of change of 0.06 mm / mm. Multiplying these two factors yields a deflection risk factor of 0.18, reflecting the likelihood of deflection due to the sudden change in tool force in this transition zone.

[0070] In one possible implementation, the risk level is divided based on a large amount of machining experimental data. An offset risk coefficient below 0.1 is low risk, 0.1 to 0.3 is medium risk, and more than 0.3 is high risk. High-risk areas often appear at the junction of hard and soft materials, or where the cutting depth changes sharply. This risk assessment method makes subsequent path compensation more targeted and avoids overcompensation of the entire path. The application of the cubic spline interpolation algorithm in path compensation is due to its good smoothing properties. The algorithm constructs a cubic polynomial curve through known front and back normal path points to ensure that the compensated path has continuous first-order and second-order derivatives at the connection points.

[0071] For example, three normal path points are taken before and after a high-risk point. Using the coordinates of these six points and the tangent direction constraints, a smooth curve passing through the high-risk point is calculated. The direction of the compensation vector is perpendicular to the original path, and its magnitude is proportional to the offset risk factor.

[0072] It should be noted that the real-time correction process relies on position feedback during machining. The CNC system obtains the actual tool position via an encoder and compares the deviation between the actual position and the pre-compensated path during each interpolation cycle. If the deviation exceeds the allowable range, the control system generates a correction command to adjust the target position of the next interpolation point. This dual compensation mechanism combines predictive compensation with real-time correction capabilities. Curvature calculation uses a discrete point approximation method. A circular arc passing through three consecutive path points, P1, P2, and P3, is constructed. The inverse of the arc radius is the curvature at that position. The difference in curvature between adjacent positions reflects the degree of path curvature variation. A greater curvature variation indicates a more complex path and higher requirements for the machine tool's dynamic performance. The dynamic variation of the cutting load is comprehensively evaluated using multiple indicators. The vibration amplitude increment reflects the stability of the cutting process. When the curvature variation rate is 0.5 per millimeter, the vibration amplitude increases from 0.02 mm to 0.05 mm, an increment of 0.03 mm. Multiplying these two values ​​yields a load variation index of 0.015. This indicator intuitively reflects the impact of path complexity on machining stability. By establishing a corresponding relationship between load changes and processing quality, it is possible to predict which areas may have processing defects, so as to adjust the processing parameters in advance and ensure the consistency of the overall processing quality.

[0073] Step S106: Adjust the auxiliary system parameters related to chip removal efficiency according to the corrected trajectory of the tool path, optimize the cleaning frequency and airflow intensity of the chip removal channel according to the complexity of the corrected trajectory and the change of cutting load, and evaluate the chip accumulation risk by monitoring the chip accumulation amount and discharge efficiency.

[0074] Based on the curvature change data in the tool path correction trajectory, the experimentally established trajectory curvature and chip generation rate correspondence table is queried to determine the chip generation rate of different trajectory segments. The coolant flow control valve opening is adjusted by the ratio of the chip generation rate to the standard chip generation rate. The vacuum cleaner power level is set according to the complexity value of the corrected trajectory, and the chip collector opening interval is determined using the cutting load change data. Based on the determined coolant flow control valve opening, vacuum cleaner power level, and chip collector opening interval, combined with the complexity value of the corrected trajectory, the cleaning time interval of the chip discharge channel is calculated by multiplying the complexity value by the cleaning cycle adjustment coefficient. The purge airflow velocity value is determined by multiplying the ratio of the cutting load change data to the standard cutting load by the basic airflow velocity, resulting in the setting values ​​for the cleaning frequency and airflow intensity parameters. The chip height is measured by a detection device installed in the discharge channel, and the chip accumulation rate is calculated based on the ratio of the chip height to the total channel height. The discharge efficiency is obtained by measuring the mass of chips discharged per unit time and dividing the discharged chip mass by the theoretical generation mass calculated based on the chip generation rate during the same period. If the chip accumulation rate exceeds the preset accumulation threshold or the discharge efficiency is lower than the preset efficiency standard, it is determined that there is a risk of cutting accumulation.

[0075] Specifically, the correspondence between trajectory curvature and chip generation rate is based on the physical principles of cutting mechanism.

[0076] Specifically, when the tool moves along a path with significant curvature, the cutting angle constantly changes, altering the way the material is sheared, which in turn affects chip morphology and generation rate. While straight-line cutting produces continuous chips with a relatively stable generation rate, curved cutting tends to break into smaller pieces, increasing the generation rate by 20%-30%. This correspondence table was developed through cutting experiments on paths with varying curvatures, recording chip generation rates for curvatures ranging from 0 to 0.5 mm.

[0077] In one possible implementation, the adjustment of the coolant flow rate directly affects the chip flushing effect. The standard chip generation rate is set at 5 grams per second. When the actual rate reaches 7 grams, the ratio is 1.4, and the control valve opening is increased by 40%, increasing the coolant flow rate from 10 liters per minute to 14 liters per minute. The increased flow rate not only removes more heat, but more importantly, it provides a stronger flushing force to prevent chips from accumulating between the tool and the workpiece. The setting of the vacuum cleaner power level takes into account the impact of path complexity on chip flying. Complex trajectories mean frequent acceleration and deceleration and changes in direction, and the movement trajectory of the chips becomes unpredictable. When the complexity value is 0.8, the vacuum cleaner power is increased from the standard 500 watts to 800 watts, ensuring sufficient suction to cover a larger chip flying range. The opening interval of the chip collector is determined by the change in cutting load. The greater the load, the more intensive the chip generation, and the opening interval is shortened accordingly.

[0078] It should be noted that the cleaning cycle adjustment coefficient is determined based on statistical analysis of a large amount of production data. This coefficient is usually between 0.5 and 2. The higher the complexity, the smaller the coefficient and the more frequent the cleaning.

[0079] For example, if the basic cleaning interval is 300 seconds, the complexity is 0.6, and the adjustment factor is 0.8, the actual cleaning interval is 144 seconds. This dynamic adjustment avoids the drawbacks of fixed-cycle cleaning, preventing production efficiency from being affected by over-frequent cleaning and preventing blockages from excessive intervals. The calculation of the purge air velocity reflects the relationship between cutting load and chip weight. Under standard cutting load, a basic air velocity of 15 meters per second is sufficient to remove typical chips. When the cutting load increases by 50%, the chips become thicker and heavier, requiring an air velocity 1.5 times higher, or 22.5 meters per second, for effective removal. This proportional relationship ensures effective chip cleaning under varying machining conditions. The principle of conservation of mass is used to evaluate removal efficiency. The theoretical generated mass is calculated by multiplying the chip generation rate by the time. For example, at a rate of 5 grams per second, 50 grams are theoretically generated in 10 seconds. Actual removal of 45 grams results in an efficiency of 90%. When the efficiency falls below the preset standard of 85%, it indicates that chips are trapped in the channel. Combined with the accumulation rate indicator, if the accumulation rate exceeds the 60% threshold, a blockage risk is identified and an early warning mechanism is triggered. This dual monitoring mechanism improves the accuracy of risk identification and effectively prevents machining interruptions caused by chip accumulation.

[0080] In step S107, if the chip accumulation risk is lower than the preset risk threshold, the target machining accuracy correction plan is determined in combination with the machining accuracy deviation; the hardness distribution, cutting force fluctuation and parameter adjustment records can also be integrated and stored, and the machining parameter dynamic database is continuously updated to form a parameter optimization file.

[0081] If the chip accumulation risk value is lower than the preset risk threshold, a precision correction coefficient is calculated based on the ratio of the machining accuracy deviation to the allowable deviation of the workpiece design dimensions. By querying a table of correction coefficients and compensation parameters based on historical machining data, the tool radial compensation value and feed speed fine-tuning percentage are obtained. These values ​​are combined to form a target machining accuracy correction scheme containing the compensation value and adjustment ratio. Based on the compensation value and adjustment ratio in the target machining accuracy correction scheme, the hardness distribution map data, cutting force fluctuation monitoring value series, and feed speed adjustment records generated during the machining process are collected. A correspondence between these data is established based on timestamps and machining position coordinates. Lossless compression is used to reduce the data storage volume, generating a structured record file containing information about the entire machining process. This structured record file is written into a dynamic machining parameter database in a predetermined format. A multi-dimensional retrieval index is established based on the workpiece material type, hardness grade range, and machining accuracy requirements. An automatic database update mechanism regularly integrates new machining data and removes outdated records, creating a parameter optimization archive that is categorized by machining characteristics and continuously updated, resulting in a complete parameter optimization archive.

[0082] Specifically, the calculation of the precision correction coefficient reflects the evaluation principle of relative error.

[0083] Specifically, when the workpiece design size is 50 mm, the allowable deviation is ±0.05 mm, and the measured machining accuracy deviation is 0.03 mm, the correction factor is 0.03 divided by 0.05, which is 0.6. This factor reflects the ratio of the actual deviation to the allowable deviation. A larger value indicates that the machining accuracy is closer to the tolerance limit and requires greater compensation.

[0084] In one possible implementation, a table of correspondences between correction coefficients and compensation parameters is established based on statistical analysis of extensive historical machining data. Regression analysis of the machining results of thousands of workpieces revealed a near-linear relationship between the correction coefficients and the required compensation values. When the correction coefficient is 0.6, a table lookup indicates a radial tool compensation value of 0.02 mm, requiring a 15% reduction in feed rate. This correspondence ensures that appropriate compensation parameters are obtained for varying precision deviations. The combination of radial compensation values ​​and speed fine-tuning percentages creates a dual safeguard mechanism. Radial compensation directly corrects the tool position, bringing the cutting trajectory closer to the ideal path; speed fine-tuning reduces dynamic errors in the cutting process by reducing the feed rate. The combination of these two methods not only addresses static position deviations but also controls dynamic machining errors.

[0085] It should be noted that the generation of structured record files involves the integration of data from multiple sources. The hardness distribution map is stored as a two-dimensional array, with each element containing a position coordinate and a hardness value. The cutting force fluctuation monitoring value sequence is arranged in chronological order, recording the three-dimensional force data at each sampling moment. The feed rate adjustment record includes the adjustment time, original speed, and adjusted speed. These heterogeneous data are linked via timestamps and position coordinates to form a complete description of the machining process. Lossless compression uses run-length encoding to compress consecutive identical data.

[0086] For example, during the stable cutting phase, the cutting force varies very little. For 100 consecutive sampling points, the values ​​remain within the range of 80 ± 1 Newton. This can be compressed into "80 Newtons repeated 100 times," significantly reducing storage space. The compressed data volume is typically only 30%-40% of the original data. The establishment of multi-dimensional search indexes improves data query efficiency. The material type index categorizes aluminum alloys, stainless steel, and titanium alloys; the hardness grade index divides HRC values ​​into multiple intervals; and the precision requirement index groups tolerance levels. For example, when searching for machining parameters such as "HRC 30-35 stainless steel, tolerance ±0.02 mm," the intersection of the three indexes quickly locates the relevant records. An automatic database update mechanism ensures information currency. Each time new machining data is written, the system automatically calculates the similarity between the data and existing records. Older data with high similarity is marked as a candidate for deletion. When the number of records for a particular machining parameter category exceeds the set limit, the latest and most effective records are retained, while outdated and less effective data are deleted. This dynamic update mechanism enables the database to always maintain the optimal parameter combination, providing a reliable reference for subsequent processing.

[0087] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A CNC machining optimization method for mobile phone middle frame based on artificial intelligence, characterized in that: The method includes: performing real-time scanning on the surface of a mobile phone middle frame workpiece to obtain initial data on the material hardness distribution, forming a hardness distribution map, and obtaining a regional division result of hardness variation; establishing a cutting force monitoring benchmark based on the regional division result of hardness variation, collecting cutting force fluctuation data during processing, identifying abnormal fluctuation regions based on the cutting force monitoring benchmark, determining a specific location and time point of a cutting force mutation based on material property differences and a cutting force threshold standard in the hardness distribution map; identifying a hardness grade distribution and a cutting force mutation amplitude at a specific location of the cutting force mutation, calculating a tool feed speed adjustment parameter based on the hardness grade distribution and the cutting force mutation amplitude, and generating a speed adjustment instruction based on the adjustment parameter; Dynamically match cutting depth parameters according to speed adjustment instructions, determine target cutting depth configuration, and match cutting path deviation monitoring value, vibration amplitude index and machining accuracy deviation in real time through target cutting depth configuration; predict offset risk in machining process according to regional division result of hardness distribution map and depth variation law of target cutting depth configuration, correct tool path, obtain corrected trajectory of tool path, analyze complexity of corrected trajectory and cutting load change in combination with cutting path deviation monitoring value and vibration amplitude index, including: calculating hardness change rate according to hardness distribution map, calculating depth change rate according to cutting depth configuration, generating combined hardness change rate and depth change rate. Offset risk coefficient, mark high-risk points; calculate path compensation vectors based on the high-risk points and theoretical paths, superimpose to generate pre-compensated paths, and perform real-time correction to generate corrected trajectories; calculate curvature change rate based on the corrected trajectories, and quantify the complexity of the trajectories in combination with the cutting path deviation and the vibration amplitude; adjust auxiliary system parameters related to chip discharge efficiency based on the corrected trajectories of the tool paths, optimize the cleaning frequency and airflow intensity of the chip discharge channels based on the complexity of the corrected trajectories and the changes in cutting loads, and evaluate the chip accumulation risk by monitoring the chip accumulation amount and discharge efficiency; if the chip accumulation risk is lower than the preset risk threshold, determine the target machining accuracy correction plan in combination with the machining accuracy deviation.

2. The method for CNC machining optimization of a mobile phone middle frame based on artificial intelligence according to claim 1, characterized in that: The method comprises performing real-time scanning on the surface of the mobile phone middle frame workpiece to obtain initial data on the material hardness distribution, forming a hardness distribution map, and obtaining a regional division result of hardness variation, including: detecting the surface of the mobile phone middle frame workpiece point by point through a sensor array, obtaining hardness values ​​at each position, and forming a data set including position coordinates and hardness values; performing spatial interpolation on the hardness values ​​in the data set, calculating hardness values ​​between detection points using an inverse distance weighting method, and generating continuous hardness distribution data; generating a hardness distribution map based on the continuous hardness distribution data, calculating hardness differences between adjacent positions in the hardness distribution map, determining hardness region boundaries, and connecting boundary points to form a hardness variation region outline.

3. The method for CNC machining optimization of a mobile phone middle frame based on artificial intelligence according to claim 1, characterized in that: The method establishes a cutting force monitoring benchmark based on the regional division results of hardness changes, collects cutting force fluctuation data during processing, identifies fluctuation abnormal areas based on the cutting force monitoring benchmark, and determines the specific position and time of the cutting force mutation based on the material property differences and cutting force threshold standards in the hardness distribution map, including: setting cutting force reference values ​​for each area based on the regional division results of the hardness distribution map, and generating a monitoring benchmark table containing position coordinates and cutting force ranges; collecting cutting force time series data through a force sensor, comparing the cutting force time series data with the cutting force range in the monitoring benchmark table, identifying abnormal points, and recording the position and time of the abnormal points; calculating the cutting force change rate based on the abnormal points and the hardness distribution map, and determining the position of the cutting force mutation.

4. The method for CNC machining optimization of a mobile phone middle frame based on artificial intelligence according to claim 1, characterized in that: The method of identifying abnormal fluctuation areas based on the cutting force monitoring benchmark includes: collecting tool cutting force component data through a three-axis force sensor, and setting the three-axis cutting force benchmark value for each area in combination with the regional division result of the hardness distribution map; comparing the cutting force component data with the three-axis cutting force benchmark value, calculating the deviation ratio, and recording the position and time of the abnormal point exceeding the threshold; calculating the cutting force change amplitude based on the abnormal point, counting the abnormal point density per unit area, and dividing the cutting force abnormal area.

5. The method for CNC machining optimization of a mobile phone middle frame based on artificial intelligence according to claim 1, characterized in that: The method for identifying the hardness grade distribution and the cutting force mutation amplitude at a specific position where the cutting force suddenly changes, calculating the adjustment parameters of the tool feed speed based on the hardness grade distribution and the cutting force mutation amplitude, and generating a speed adjustment instruction based on the adjustment parameters includes: extracting the hardness grade and the cutting force variation amplitude at the cutting force sudden change position, querying a hardness and speed correspondence table to determine the feed speed adjustment ratio; calculating the acceleration limit value based on the cutting force variation amplitude, and calculating the regional transition distance based on the hardness grade difference; and generating an adjustment instruction including the speed value and the acceleration parameter based on the feed speed adjustment ratio, the acceleration limit value, and the regional transition distance.

6. The method for CNC machining optimization of a mobile phone middle frame based on artificial intelligence according to claim 1, characterized in that: The method dynamically matches cutting depth parameters according to the speed adjustment instruction, determines the target cutting depth configuration, and matches the cutting path deviation monitoring value, vibration amplitude index and machining accuracy deviation in real time through the target cutting depth configuration, including: querying the speed and depth correspondence table according to the speed value in the speed adjustment instruction, calculating the cutting depth value, and generating cutting depth configuration data; setting the path deviation and vibration threshold according to the cutting depth configuration data, measuring the tool path deviation and vibration amplitude through a sensor, and generating a monitoring value matching the cutting depth configuration data.

7. The method for CNC machining optimization of a mobile phone middle frame based on artificial intelligence according to claim 1, characterized in that: The auxiliary system parameters related to the chip discharge efficiency are adjusted according to the corrected trajectory of the tool path, the cleaning frequency and airflow intensity of the chip discharge channel are optimized according to the complexity of the corrected trajectory and the change of the cutting load, and the chip accumulation risk is evaluated by monitoring the chip accumulation amount and discharge efficiency, including: according to the curvature data of the corrected trajectory, querying the curvature and chip generation rate table, determining the chip generation rate, adjusting the coolant flow rate, vacuum cleaner power and chip collector frequency; according to the complexity of the corrected trajectory, calculating the cleaning time interval and airflow velocity; measuring the chip accumulation rate by a detection device, comparing the chip accumulation rate with a threshold value, and evaluating the chip accumulation risk.

8. The method for CNC machining optimization of a mobile phone middle frame based on artificial intelligence according to claim 1, characterized in that: If the chip accumulation risk is lower than the preset risk threshold, the target machining accuracy correction plan is determined in combination with the machining accuracy deviation, including: calculating the correction coefficient based on the accumulation risk being lower than the threshold and combining it with the machining accuracy deviation, querying the correction coefficient and compensation parameter table, and generating a target machining accuracy correction plan including the compensation value and the adjustment ratio.

Images