Intelligent control method for cutting path of scribing machine

Abstract

The invention relates to the technical field of precision machining control, and discloses an intelligent control method for a cutting path of a scribing machine, which comprises the following steps: acquiring a load current signal of a main shaft servo motor in real time and storing the load current signal into a sliding window data queue; performing weighted differential calculation to obtain a real-time load change rate representing a load trend, and calculating an average absolute deviation to obtain an oscillation state quantity representing load dispersion; querying based on a preset nonlinear gain scheduling relationship to obtain a first speed regulation coefficient and an oscillation suppression coefficient; according to the method, an orthogonal dual-channel feed-forward control mechanism is constructed in the same data window, and multiplication cooperation of weighted differential and dispersion statistics is utilized, so that a feed speed set value is corrected in real time; the problems of phase lag and response mismatch of conventional feedback control in the case of transient impact and microcosmic oscillation of hard and brittle materials are solved.

Description

Technical Field

[0001] This invention relates to an intelligent control method for the cutting path of a dicing machine, belonging to the field of precision machining control technology. Background Technology

[0002] In current dicing processes for semiconductor wafers, optical glass, and precision ceramics, constant feed rate motion control is the mainstream strategy. The spindle servo system is configured in a speed loop-dominated mode, adjusting the drive current based on the position command and encoder feedback deviation to maintain the tool's set linear speed relative to the workpiece. This control logic is based on the assumption of uniform internal texture of the processed material, maintaining a constant cutting removal rate to ensure processing efficiency. However, when processing high-hardness, non-homogeneous materials such as silicon carbide or sapphire, the linear feedback control strategy has limitations. The material has randomly distributed hard points or lattice defects, and the tool experiences a step-like load change upon contact. Existing servo control loops rely on current amplitude or position following error accumulation to trigger adjustment, resulting in phase lag. By the time the control system detects overload and issues a deceleration command, the tool edge has already caused a high-energy impact on the material, leading to edge chipping, microcracks, or chipping.

[0003] To address the aforementioned deviations and lags, industry research has attempted to introduce more complex sensing and prediction mechanisms to compensate for the cutting path. For example, Chinese invention patent CN118943061B discloses a method and device for compensating the cutting coordinates of a dicing machine. This method uses a high-speed laser scanner to monitor the wafer surface state in real time during the cutting process, acquires wafer surface deformation data, and uses a cutting twin model to predict future path offsets, thereby correcting the coordinates of the predetermined cutting path. However, this method essentially compensates for the geometric coordinates of the cutting path, and the surface deformation data it relies on is also a representation of the result after the cutting stress, rather than the stress itself. When the tool instantaneously impacts a hard point, the problem of excessive compensation delay that it aims to solve still exists in suppressing instantaneous impact. Before the deformation is detected and the model is predicted, edge chipping damage caused by transient impact energy has already occurred. This type of method focuses on ensuring trajectory accuracy, but it still does not provide a solution for how to actively suppress the impact load itself in real time to avoid edge chipping.

[0004] Therefore, the technical problem to be solved by this invention is how to achieve advanced perception and active nonlinear suppression of transient impact loads without sacrificing the system's noise immunity performance and avoiding lag in amplitude feedback control. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: An intelligent control method for the cutting path of a dicing machine, the method comprising: The load current signal of the spindle servo motor of the dicing machine is collected in real time at a predetermined sampling frequency, and N historical sampling points of the load current signal are stored in the sliding window data queue. Within the control cycle, the current feed speed setpoint of the dicing machine is acquired, and based on N sampling points in the sliding window data queue, the following calculation steps are performed in parallel: Step A: Using sliding window weighted differential logic, by setting a larger weight coefficient for sampling points closer to the current time, the trend slope of N sampling point data is fitted and calculated to obtain the real-time load change rate; based on the real-time load change rate, the first speed adjustment coefficient is obtained by querying the preset first nonlinear gain scheduling relationship; the first nonlinear gain scheduling relationship is configured such that when the absolute value of the real-time load change rate exceeds the first threshold, the first speed adjustment coefficient decreases exponentially with the increase of the absolute value of the real-time load change rate. Step B: Calculate the average absolute deviation between the N sampled data points and the mean of the N sampled data points to obtain the oscillation state quantity characterizing the dispersion of the load current signal; and based on the oscillation state quantity, query the preset second gain scheduling relationship to obtain the oscillation suppression coefficient; the second gain scheduling relationship is configured such that when the oscillation state quantity exceeds the oscillation threshold, the oscillation suppression coefficient decreases as the oscillation state quantity increases. Step C: Multiply and couple the first speed adjustment coefficient with the oscillation suppression coefficient to obtain the final speed adjustment coefficient; use the final speed adjustment coefficient to correct the feed speed setpoint in real time, and send the corrected speed command to the feed axis servo system for execution.

[0006] Preferably, in step A, the sliding window weighted differential logic calculates the real-time load change rate using the following discretization formula: Where K(t) is the real-time load change rate, I t-i To calculate the load current value at the i-th sampling point from the current moment, I t-N w represents the load current value at the earliest sampling point in the sliding window data queue. i The weights are preset and satisfy w0 > w1 > ... > w N-1 ΔT is the sampling time interval.

[0007] Preferably, the first nonlinear gain scheduling relationship specifically includes three continuous control intervals: a steady region, where a constant unity gain coefficient is output when the absolute value of the real-time load change rate is less than a first threshold; a suppression region, where a gain coefficient that decays exponentially according to a natural exponential function is output when the absolute value of the real-time load change rate is between the first threshold and a second threshold; and a protection region, where a constant minimum safe gain coefficient is output when the absolute value of the real-time load change rate is greater than the second threshold, and the minimum safe gain coefficient is not greater than 0.1.

[0008] Preferably, the method further includes a parameter drift compensation step based on the entire tool lifecycle: recording the cumulative cutting distance of the current tool of the dicing machine; establishing a positive linear mapping relationship between the cumulative cutting distance and the first threshold and the second threshold; and automatically increasing the values ​​of the first threshold and the second threshold according to the current cumulative cutting distance before each cutting task is started to offset the increase in background load fluctuation noise caused by tool wear.

[0009] Preferably, the method further includes a feedforward control step based on the defect association of adjacent paths: when the final speed adjustment coefficient is lower than a preset event recording threshold, the coordinate position of the current tool on the current cutting path is recorded as a risk node; when executing the operation of the next cutting path, the coordinates of the risk node are mapped to the next cutting path to define a virtual risk area; the tool position is monitored in real time, and when the tool position enters the pre-read range of the virtual risk area, the feed speed of the dicing machine is forcibly reduced to a preset safe speed, taking priority over the first nonlinear gain scheduling relationship; if the real-time load change rate does not exceed the first threshold during the passage through the virtual risk area, the marking of the virtual risk area is removed.

[0010] Preferably, the method further includes an edge effect compensation step based on the real-time position of the cutting tool: real-time acquisition of the position coordinates of the dicing machine cutting tool on the cutting path, and determination of whether the position coordinates are located within a preset cutting edge buffer zone; if located within the cutting edge buffer zone, the value of the first threshold is dynamically reduced according to the distance between the current position of the cutting tool and the cutting endpoint, based on a preset attenuation function, so that the sensitivity of triggering feed speed adjustment within the cutting edge buffer zone is higher than that outside the buffer zone.

[0011] Preferably, the method further includes an asymmetric speed recovery control step: after the feed speed of the dicing machine is reduced using the final speed adjustment coefficient, the control system enters a dissipation holding state; in the dissipation holding state, the feed speed is only allowed to be increased when the real-time load change rate is detected to fall below a first threshold and a preset stability confirmation condition is met; the stability confirmation condition includes: the absolute value of the real-time load current decreases by more than a preset hysteresis ratio or the duration after the feed speed is reduced exceeds a preset minimum time window.

[0012] Preferably, after the dissipation holding state is released, the process of increasing the feed speed follows the restricted climb logic, which limits the upper limit of the acceleration for speed recovery, so that the feed speed recovers to the feed speed set value according to a linear ramp, and the upper limit of the acceleration is less than the maximum physical acceleration of the dicing machine servo system.

[0013] Preferably, in step B, the oscillation suppression coefficient is applied to the multiplicative coupling only when the absolute value of the real-time load change rate is less than the first threshold, so as to ensure that the suppression action against high-frequency micro oscillations is performed independently when the load current signal is in a state of stable trend but severe micro-dispersion.

[0014] Preferably, the multiplicative coupling of the first speed adjustment coefficient and the oscillation suppression coefficient follows the minimum value priority principle: if the product of the first speed adjustment coefficient and the oscillation suppression coefficient is less than the minimum operating speed coefficient preset by the system, the final speed adjustment coefficient is forcibly set to the minimum operating speed coefficient to maintain the minimum torque required for the servo motor to overcome static friction.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. Transient Response and Noise Reduction Mechanism Based on Timing Characteristics: This invention constructs a control loop that includes sliding window weighted differential calculation and a nonlinear gain scheduling table. This solves the phase lag problem in conventional servo control systems when handling high-frequency step loads. By using a sliding window to perform weighted differential processing on the real-time acquired current signal, random electromagnetic noise interference is filtered out and the steepness characteristics of load change are extracted. These characteristics are then used as a pre-trigger signal and directly input into the feed gain scheduling table. This allows the control system to forcibly attenuate the feed speed command in the initial stage when the load amplitude has not reached the failure threshold but the rate of change shows a sudden trend. Based on the timing characteristics of the signal rather than a single amplitude response mechanism, the response time of the control system to hard point impacts is shortened, and the overload energy accumulation process is blocked within a safe window before the material fracture stress limit is reached.

[0016] 2. Reconstruction of System Damping Characteristics by Asymmetric Lag Logic: This invention introduces asymmetric control logic, including stability confirmation and limited climb, during the speed recovery phase. This eliminates the risk of secondary oscillations caused by the mismatch between the control system's response speed and the mechanical system's energy dissipation speed. After the feed speed is suppressed due to a sudden load change, the control system does not immediately recover the set speed as the load change rate falls back. Instead, it enters a dissipation holding state until the absolute value of the current falls back or the low-speed holding time meets the preset stability conditions. A one-way time lag and acceleration limit are artificially introduced into the control loop for the speed recovery direction. This is equivalent to injecting dynamic damping into the servo drive system at the moment of hard point cutting, ensuring that the tool and workpiece system are in a low-energy operating state during the mechanical relaxation period after completing high-stress cutting. This avoids system dynamic stiffness instability and cutting marks caused by the controller intervening too early to accelerate.

[0017] 3. Defect Memory and Feedforward Avoidance Based on Spatial Topological Association: This invention utilizes the spatial positional correlation between multiple cutting paths to construct a closed-loop control system that includes risk node recording, coordinate projection mapping, and feedforward pre-reading deceleration iteration. The system marks the trigger gain attenuation coordinate position in the preceding cutting path as a risk node and maps it to the corresponding control buffer of the subsequent adjacent cutting path. When executing subsequent cutting tasks, the control system compares the distance between the tool position and the virtual risk area in real time. Before the tool contacts the potential defect area, it prioritizes decoupling from the real-time feedback logic and switches to the preset safe speed. Utilizing the physical characteristic that internal defects of crystal materials are continuously distributed along the lattice or growth direction, the single random impact response is transformed into predictable spatial feedforward control. When dealing with continuously distributed defects, the inherent detection delay of feedback control is eliminated, achieving zero-impact passage through known risk areas. Attached Figure Description

[0018] Figure 1 This is a diagram showing the dual-channel logic architecture and data flow of the intelligent control method of the present invention. Figure 2 The performance comparison of the core algorithm of this invention is shown in the bar charts of edge collapse and load. Figure 3 This is a diagram showing the deployment architecture of the computing and execution nodes of the control system of this invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be described in detail below. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0020] This invention provides an intelligent control method for the cutting path of a dicing machine, which operates within the dicing machine's motion control system, such as a CNC or PLC controller, forming a real-time control and adjustment closed loop. In addition to the traditional position servo loop, a dual-channel feedforward control mechanism based on spindle load timing morphology analysis is employed. This mechanism includes a weighted differential channel for transient impact sensing and a discrete statistical channel for latent instability sensing. Within each control cycle, the control system uses the same set of load current sampling points in the sliding window data queue to calculate in parallel the real-time load change rate characterizing the load mutation trend and the oscillation state quantity characterizing the load fluctuation dispersion. The system converts these two state quantities into a first speed adjustment coefficient and an oscillation suppression coefficient respectively through a preset nonlinear gain scheduling relationship. By multiplying and coupling these two coefficients, the current feed speed setpoint of the dicing machine is corrected in real time. Without relying on high-cost sensors or adding complex physical models, this method solves the response lag and mismatch problems of conventional control strategies when facing the heterogeneity of hard and brittle materials. In its specific implementation, the execution flow of this intelligent control method is anchored to the servo control cycle of the dicing machine. The control system collects the load current signal I(t) of the dicing machine spindle servo motor in real time at a predetermined high sampling frequency, such as 10kHz or higher. This signal is a direct electrical characterization of the cutting resistance. The latest sampling point is stored in a data queue, maintaining a fixed length containing N historical sampling points to form a sliding window data queue. The choice of N value is an engineering trade-off. A smaller N value, such as 20, can improve the response sensitivity, while a larger N value, such as 100, helps to smooth high-frequency electromagnetic noise. Those skilled in the art can calibrate it according to the characteristics of the servo system and the characteristics of the processed material.

[0021] Within each control cycle, while acquiring the current feed speed setpoint of the dicing machine, the control system initiates two parallel calculation steps based on N sampling points in the sliding window data queue: Step A, the transient impact sensing channel, focuses on solving the phase lag problem of hard point impact; instead of using traditional mean filtering, it employs sliding window weighted differential logic, assuming that the closer the load data point is to the current moment, the more predictive value it has for impending sudden changes; therefore, this logic assigns a larger weighting coefficient w to sampling points closer to the current moment. i The trend slope of the N sampling points is fitted to obtain the real-time load change rate K(t). In the specific implementation, the real-time load change rate K(t) is calculated using the following discretization formula: , among which, I t-i To calculate the load current value at the i-th sampling point from the current moment, I t-N w represents the load current value at the earliest sampling point in the sliding window data queue. i The preset weight coefficients satisfy w0>w1>•••>wN-1 ΔT is the sampling time interval; the absolute value of the calculated K(t) directly represents the steepness of the load increase; the system uses the K(t) value to query the preset first nonlinear gain scheduling relationship to obtain the first speed adjustment coefficient; this scheduling relationship is designed as a nonlinear form that is highly sensitive to sudden changes, and can be divided into three continuous control intervals: the steady region: when When the value is less than the first threshold T1, the system determines it as normal cutting fluctuation and outputs a constant unity gain coefficient of 1.0; Suppression region: when When the value is between the first threshold T1 and the second threshold T2, the system determines it as a precursor to encountering a hard point or defect. At this time, the first speed adjustment coefficient follows... The increase of decreases exponentially, according to The form decreases rapidly to forcibly suppress feed before the impact reaches its peak; protected area: when When the value exceeds the second threshold T2, the system determines that a serious impact has occurred. At this time, the proportional adjustment is no longer performed, but a constant minimum safety gain coefficient is directly output. This coefficient is no greater than 0.1, such as 0.05, so that the system crawls through at the lowest speed or waits for operator intervention.

[0022] Step B, the micro-oscillation sensing channel, addresses the issue of high-frequency, intense micro-oscillations that may be overlooked in Step A, even if they lack a clear trend. It reuses N sampling points in a sliding window to calculate the average absolute deviation between these data and the mean within the window. This value represents the oscillation state quantity characterizing the dispersion of the load current signal. The average absolute deviation is chosen over variance to avoid computational overhead associated with squaring, thus improving the efficiency of the control system. After obtaining this oscillation state quantity, the oscillation suppression coefficient is retrieved from a preset second gain scheduling relationship. This relationship is configured such that when the oscillation state quantity is less than a preset oscillation threshold, the coefficient is 1.0, resulting in no suppression. When the oscillation state quantity exceeds the threshold, the oscillation suppression coefficient decreases as the oscillation state quantity increases, either linearly or inversely. In a preferred embodiment, the logical orthogonality of the two channels is ensured. The oscillation suppression coefficient in Step B is only allowed to participate in subsequent multiplicative coupling when Step A determines the system is in a stable region. This logical configuration is used when the overall trend is stable but... Under conditions of severe local dispersion, the system independently executes suppression actions against high-frequency micro-oscillations. Step C, the control command coupling and output stage, involves multiplying and coupling the first speed adjustment coefficient obtained in step A with the oscillation suppression coefficient obtained in step B to obtain the final speed adjustment coefficient. This multiplication design ensures that if any channel detects a risk (i.e., the coefficient is less than 1.0), it can suppress the final speed. Before output, the system also performs a safety check, i.e., the minimum value priority principle: if the final speed adjustment coefficient obtained after multiplication and coupling is lower than the minimum operating speed coefficient necessary to maintain the servo motor to overcome static friction (e.g., 0.02), the final speed adjustment coefficient is forcibly set to this minimum operating speed coefficient to prevent the feed axis from physically stalling due to excessively low speed commands. Finally, using this coupled and checked final speed adjustment coefficient, the original feed speed setting value obtained from the G code or host computer is corrected in real time, and the corrected speed command is sent to the feed axis servo system for execution.

[0023] To further enhance the adaptability and stability of the control system in real production environments, this invention may also include a series of additional control logic modules: parameter drift compensation based on the entire tool lifecycle; given that tools wear out after long-term use, leading to an overall increase in background load fluctuation noise during cutting, without compensation, fixed T1 and T2 thresholds will cause inconsistent system sensitivity on new and old tools, resulting in misjudgments; the system records the current cumulative cutting distance of the tool in real time; establishes a positive linear mapping relationship between the cumulative cutting distance and the first threshold T1 and the second threshold T2; before each cutting task is started, the control system automatically increases T1 and T2 based on the current cumulative cutting distance. The system dynamically compensates for control baseline drift caused by tool wear; it also compensates for edge effects based on the real-time position of the tool; hard and brittle materials have the weakest physical support structure and the lowest resistance to edge chipping at the cutting edge that is about to be cut through; therefore, the system obtains the position coordinates of the dicing machine tool on the current cutting path in real time; it determines whether the coordinates are located within the preset cutting edge buffer zone, which is set to the area 0.5mm away from the cutting endpoint; if the tool enters the buffer zone, the system will dynamically and in real time reduce the value of the first threshold T1 according to the distance between the current position of the tool and the cutting endpoint, based on a preset attenuation function; this makes the control system more sensitive to load fluctuations in the cutting edge area than in the center area of ​​the material.

[0024] Based on feedforward control using adjacent path defect correlation, considering internal defects in materials such as wafers, taking lattice defects or hard points as examples, these defects may be continuously distributed along specific crystal orientations in space, leading to repeated risks near the same coordinate position in adjacent cutting paths. Therefore, a defect memory mechanism is introduced. When executing the Nth path, if the final speed adjustment coefficient is lower than a preset event recording threshold, it is used to determine whether a serious risk has been encountered. The current tool coordinate position on that path is immediately recorded and marked as a risk node. When executing the N+1th cutting path, the coordinates of all risk nodes recorded on the Nth path are projected onto the N+1th path, defining several virtual risk regions. When the tool position enters the pre-read range of these virtual risk regions, the control system will prioritize real-time feedback logic and forcibly reduce the feed rate to a preset safe speed. Furthermore, this mechanism also has dynamic clearing capability: if the detected real-time load change rate K(t) does not exceed the first threshold T1 during the tool's passage through the virtual risk region at a safe speed, the system determines that the defect is not continuous at that point and immediately removes the mark of the virtual risk region, avoiding over-defense. Efficiency loss; Asymmetric speed recovery control: To address the secondary oscillation problem that may be caused by the mismatch between the energy dissipation rate of the mechanical system and the response rate of the control system, asymmetric logic is introduced in the speed recovery phase; When the system reduces the feed rate due to trigger suppression, the control system immediately enters the dissipation hold state; In this state, the system will not immediately recover the speed simply because K(t) falls back, but must wait for a strict stability confirmation condition to be met; This condition is configured to require that the real-time load change rate has fallen back below the first threshold T1 and one of the following two conditions is met: First, the absolute value of the real-time load current decreases by more than the preset lag ratio, taking 10% of the peak value as an example; second, the duration of the feed rate reduction exceeds the preset minimum time window, taking 30ms as an example; only after these combined conditions are met can the system allow the dissipation holding state to be released; after release, the process of increasing the feed rate is not a step, but follows the limited climb logic: the system limits the upper limit of the acceleration for speed recovery, so that the feed rate recovers to the original feed rate setting value according to a gentle linear ramp, and this upper limit of acceleration is set to be less than the maximum physical acceleration of the dicing machine servo system.

[0025] Example 1: This example illustrates the application of the intelligent control method in a typical system scenario: the cutting of polycrystalline silicon / silicon carbide composite materials. The core challenge in this scenario lies in the significant difference in physical properties between the hard SiC particles and the matrix Si, leading to a situation where the cutting load exhibits both drastic, random, step-like abrupt changes and high-frequency oscillations. In the specific cutting task, the dicing machine begins cutting at a set feed rate of 100 mm / s. Within each control cycle, taking 0.1 ms as an example, the control system collects the load current signal of the spindle servo motor in real time and stores it in a sliding window data queue with a length of N = 50 sampling points. In the initial stage of cutting, the tool is located in a uniform silicon matrix, and the load current signal has a relatively low average value. The surrounding area fluctuates smoothly, and in parallel operation, step A, namely the sliding window weighted differential, calculates a real-time load change rate K(t) whose absolute value is consistently lower than the first threshold T1. Simultaneously, step B, namely the mean absolute deviation, calculates an oscillation state quantity that is also lower than the oscillation threshold. Both the first speed adjustment coefficient and the oscillation suppression coefficient are constant at 1.0, and the final speed adjustment coefficient after their multiplicative coupling is also 1.0. The actual speed executed by the feed axis servo system is the set value of 100 mm / s. When the tool begins to contact a 0.8 mm diameter silicon carbide hard particle at the X = 10.5 mm coordinate, the cutting resistance increases instantaneously, and the load current signal begins to rise sharply. In the initial stage of the signal rise, taking the first 5 ms as an example... The absolute value of the load has not yet triggered any overload protection threshold of a traditional servo system. The weighted differential logic in step A, due to the high weighting of the latest data point, is sensitive to this upward trend. The calculated real-time load change rate K(t) quickly exceeds the first threshold T1, entering the suppression region of the first nonlinear gain scheduling relationship. The system looks up the table to obtain the first speed adjustment coefficient, which decays exponentially, such as 0.6. At this point, step B is bypassed because K(t) no longer meets the steady-state condition, and the final speed adjustment coefficient is 0.6. Before the load amplitude reaches its peak, the control system has already feedforwarded the actual feed rate command to 100 × 0.6 = 60 mm / s. As the tool further penetrates the hard point, K(t) continues to increase and may exceed the first nonlinear gain scheduling relationship. At threshold T2, the protection zone logic is triggered, forcing the first speed adjustment coefficient to be set to the minimum safety gain coefficient of 0.05, and the feed rate is suppressed to 5 mm / s. The system slowly rolls over the hard point with extremely low energy instead of impacting it at high speed. During the cutting process of this hard point, another specific working condition will occur: the tool does not cut smoothly, but instead induces high-frequency cutting chatter inside the hard phase. At this time, the load current signal may show an overall stable trend, that is, fluctuating on a certain high-level platform, with severe local dispersion. In this state, the weighted differential logic in step A may fall back to below the first threshold T1 due to the unclear trend, and is judged as a stable region, attempting to restore the first speed adjustment coefficient to 1.0; Step B's mean absolute deviation calculation captures this severe dispersion, causing the oscillation state quantity to exceed the oscillation threshold, thus outputting an oscillation suppression coefficient less than 1.0, such as 0.7; Since the control logic is set to activate the coupling of step B only in the stable region, the final speed adjustment coefficient will become 1.0 × 0.7 = 0.7, and the feed rate will be corrected to 70 mm / s; This dual-channel orthogonal collaboration overcomes the limitations of a single control logic. The weighted differential logic in step A addresses the overall trend-based step impact, while the dispersion statistical logic in step B addresses the micro-oscillations that the weighted differential logic cannot detect. Together, they ensure comprehensive suppression of complex loads.

[0026] When the tool completely passes the hard point and the load current drops, the system does not immediately resume a speed of 100 mm / s. At this point, the asymmetric speed recovery control logic intervenes, and the system enters a dissipative holding state. The control system continuously monitors the load signal until the real-time load change rate K(t) drops below T1, the absolute value of the load current decreases beyond the preset hysteresis ratio, or the low-speed holding time exceeds the preset minimum time window. Only when this composite stability confirmation condition is met does the system release the holding state and initiate the restricted ramp logic, causing the feed rate to linearly ramp back to 100 mm / s according to a preset lower acceleration upper limit. This avoids secondary impacts or oscillations to the tool or workpiece caused by premature or excessive re-acceleration. Since the event is triggered at X = 10.5 mm below the event recording threshold... The system has marked the coordinate as a risk node based on the adjustment coefficient. When the dicing machine completes the current path and moves to the next adjacent cutting path, the feedforward control logic based on the defect association of adjacent paths is activated. The system projects the coordinate X = 10.5 mm onto the new path, defining a virtual risk area. When the new path cutting is executed, when the tool position is about to enter the pre-read range of the virtual risk area, the control system will no longer rely on the real-time feedback calculation of steps A and B, but will prioritize the execution of the feedforward logic, forcibly reducing the feed speed to the preset safe speed in advance. This iterative learning based on spatial topology association transforms the passive response of the previous cutting into the active avoidance of this cutting, enabling the system to achieve a transition from hysteresis feedback to zero-delay feedforward control mode when facing defect clusters continuously distributed along the lattice.

[0027] Example 2: This example sets up the following comparative test to quantify the synergy of the dual-channel feedforward control mechanism of the present invention in suppressing transient impacts and high-frequency chatter; the test platform adopts an industrial-grade dicing machine, the motion controller CNC has the ability to run customized control logic at 10kHz, and high-speed acquisition of the load current signal of the spindle servo motor; the test workpiece is a silicon substrate with multiple rows of known coordinates and a diameter of 0.5mm to 0.8mm of alumina (Al2O3) hard particles embedded in it to simulate random hard point impacts and chatter caused by heterogeneous materials encountered during the cutting process.

[0028] The experiment set up five control groups, each cutting 10 parallel cutting paths on the same workpiece. The basic feed rate was set to a constant 80 mm / s. The controller monitored and recorded the peak spindle load current when cutting to the hard particle coordinates in real time. After cutting, the average chipping width of the cut edge at the hard particle position of each path was measured using a scanning electron microscope (SEM) as an evaluation index of processing quality. The control logic of the five groups is as follows: Control group 1: adopts standard constant feed rate control of existing technology, i.e., always running at 80 mm / s; Control group 2: adopts simplified amplitude feedback control, i.e., when the absolute value of the load current exceeds the preset overload threshold, this threshold is set to the value of Control group 1. The feed rate is reduced to 10 mm / s at 80% of the peak value. Group A (partially missing): Step A of this invention is enabled, i.e., the speed is adjusted only based on the real-time load change rate K(t) and the first nonlinear gain scheduling relationship; the oscillation suppression coefficient in step B is forcibly set to 1.0. Group B (partially missing): Step B of this invention is enabled, i.e., the speed is adjusted only based on the oscillation state quantity and the second gain scheduling relationship; the first speed adjustment coefficient in step A is forcibly set to 1.0. Sample group of this invention: The complete dual-channel control method of this invention is adopted, i.e., steps A and B are calculated in parallel, and the first speed adjustment coefficient and the oscillation suppression coefficient are multiplied and coupled to jointly correct the final feed rate. The key performance data obtained from the experiment are shown in Table 1.

[0029] Table 1: Performance Comparison Test Data under Different Control Logics

[0030] Analysis of Table 1 data: Control group 1 experienced a high-speed direct impact on the hard point, resulting in the highest peak load current of 7.93A and the most severe edge chipping of 24.6μm. Control group 2, relying on the absolute value of the load, inevitably experienced a lag in adjustment action after the impact. Although the peak current decreased to 6.21A, the damage had already occurred before deceleration, resulting in limited improvement in the edge chipping width of 18.3μm. The peak load current of partially missing group A (3.15A) was significantly lower than the control group, indicating that the K(t)-based feedforward logic could suppress the feed before the impact energy fully accumulated, i.e., through exponentially decaying gain, and the edge chipping width of 8.1μm was also significantly improved. The experimental data for partially missing group B showed a peak current of 7.79A and an edge chipping width of 23.9μm, similar to control group 1. This data suggests that the mean absolute deviation logic is mainly used to sense high-frequency oscillations, but for trend-based oscillations... The step impact is essentially ineffective, which also proves the necessity of step A in suppressing the impact. The data of the sample group of this invention shows that the peak load current of 3.22A is at the same level as that of the partially missing group A, indicating that the impact energy is also suppressed by step A, but the average chipping width of 4.7μm is better than that of 8.1μm in the partially missing group A. This data comparison shows that the dual-channel synergy of this invention is not a simple functional superposition. While step A detects the impact trend through K(t) and performs the main speed reduction, the oscillation state quantity in step B detects the high-frequency chatter generated when the tool cuts into the hard phase and outputs an additional oscillation suppression coefficient. The multiplicative coupling of the two coefficients enables the system to further suppress the micro chatter of the cutting edge during the slow rolling over of the hard point, thereby achieving suppression in both the step impact and high-frequency chatter dimensions, and finally obtaining the best machining quality.

[0031] Example 3: This example combines Figures 1 to 3 This describes an intelligent control method for the cutting path of a dicing machine, such as... Figure 1 As shown, the method begins with the spindle servo motor, which acquires the load current signal I(t) in real time. The signal is stored in a sliding window data queue containing N historical sampling points. The data in this queue is processed in parallel by two channels: the first channel performs sliding window weighted differentiation to fit the trend slope and obtain the load change rate K(t). The K(t) value is used to query the first nonlinear gain scheduling table to output the first speed adjustment coefficient, which decays exponentially. The second channel performs mean absolute deviation calculation to statistically analyze the load dispersion and obtain the oscillation state quantity. This state quantity is used to query the second gain scheduling relationship to output the oscillation suppression coefficient. The output coefficients of the two channels are coupled by multiplication and minimum value verification to achieve coefficient coupling and minimum running speed protection. The finally generated corrected speed command is sent to the feed axis servo system for execution. The load change feedback from the physical process forms a closed loop. In addition, the architecture also integrates four additional modules: an edge effect compensation module based on the real-time position of the tool, a parameter drift compensation module based on the entire tool life cycle, an asymmetric speed recovery control module, and a feedforward control module based on the defect association of adjacent paths.

[0032] like Figure 2 As shown in the figure, the vertical axis represents numerical values, and the legend distinguishes between peak load current A and average edge breakage width μm. The data shows that the sample group of this invention exhibits advantages over the supplementary control groups C and D in both key indicators: peak load current and average edge breakage width. Figure 3 As shown, the architecture is divided into computing nodes and execution nodes. The computing node is the motion control host, i.e., the CNC or PLC core, which runs under high-frequency real-time task cycles. It contains a data preprocessing area, a dual-channel intelligent analysis engine, a dynamic parameter mapping library, a speed decision and recovery controller, and a space defect memory component. The computing node receives the real-time load current signal from the spindle servo motor and outputs the corrected feed speed command. The execution node is the servo drive system, which contains current loop and speed loop control. It is responsible for receiving the command and driving the feed axis servo motor to perform variable speed motion.

[0033] Example 4: To verify the rationality of the specific algorithm selection in steps A and B of the present invention, based on the test platform and workpiece of Example 2, the following comparative test groups were set up; all test groups adopted complete dual-channel coupled logic, and the internal core algorithm of step A or step B was modified: supplementary control group C: step A adopted sliding window ordinary differential logic, that is, all weight coefficients w i All equal w i =1 / N, step B uses the mean absolute deviation (MAD) algorithm of the present invention; supplementary control group D: step A uses the sliding window weighted differential logic of the present invention, and step B uses the standard deviation (SD) algorithm instead of mean absolute deviation (MAD); the experimental results are compared with the sample group of the present invention (data is the same as in Example 2) as follows: Table 2: Comparison Test Data of Core Algorithms

[0034] Analysis of the data in Table 2: Comparing the sample group of this invention (4.7μm) with the supplementary control group (C10.9μm), there are differences in both peak current and chipping width. This indicates that the weighted differential logic, which assigns higher weight to recent data points, can capture the trend of impact earlier and more sensitively than ordinary differential logic, thereby triggering more timely velocity suppression, confirming the superiority of the weighted logic. Comparing the sample group of this invention (MAD) with the supplementary control group D (standard deviation), the peak current (3.22A vs 3.19A) and chipping width (4.7μm vs 4.9μm) are at the same level, with no statistical difference. This indicates that using the mean absolute deviation as the oscillation state quantity has a physical effect comparable to using the more computationally intensive standard deviation algorithm in suppressing high-frequency chatter and improving processing quality.

[0035] Example 5: This example discloses a parameter calibration procedure for a control system, used to determine dynamic compensation and recovery logic parameters, including parameters for tool wear compensation, edge effect compensation, and asymmetric recovery logic, for determining the length N of the sliding window data queue and the weighting coefficient w. i The calibration procedure includes: collecting multiple sets of load current signals as baseline noise samples when the dicing machine is idling or cutting smoothly at a constant speed in a standard homogeneous material; and identifying the main electromagnetic and mechanical noise frequencies f of the system by performing spectral analysis on the baseline noise samples. noise Set the length N, corresponding to the total time window Where ΔT is the sampling time interval, which is greater than twice the main noise period, i.e., 2 / f noise To smooth high-frequency noise, the weighting coefficient w i Generate using a normalized linear decay function or an exponential decay function, satisfying w0 > w1 > ... > w N-1 Those skilled in the art can adjust the specific shape of the decay function based on the autocorrelation characteristics of the baseline noise samples to optimize the signal-to-noise ratio of the trend fitting; this is to determine the oscillation threshold T in step B. osc The calibration procedure reuses the baseline noise samples collected during the calibration length N, or multiple sets of load current signals collected during smooth cutting at different feed rates in standard homogeneous material. For each set of smooth cutting signals, the average absolute deviation value sequence is calculated using a sliding window of length N, and the mean of the sequence is statistically analyzed. with standard deviation , oscillation threshold T osc Set as the high percentile of the statistical distribution of the mean absolute deviation under this stationary state, for example This threshold is set higher than the background oscillation level of normal cutting. The oscillation suppression coefficient is only adjusted when high-frequency micro-chatter causes a sharp increase in dispersion. A parameter drift compensation model based on tool wear is calibrated to determine the positive linear correlation between the cumulative cutting distance d and the first and second thresholds T1. A new tool is selected, and a benchmark cutting test is performed on a standard homogeneous workpiece. Load data during smooth cutting is collected, and the initial first threshold T is obtained using a statistical method of mean plus three times the standard deviation. 1,new The tool was continuously cut under production conditions, and its cumulative cutting distance d was recorded in stages, taking the nodes of d=1000m, d=2000m, and d=3000m as examples; at each distance node, the benchmark cutting test was repeated, and the stable load data under this wear state was collected. The background noise threshold T under this wear state was calculated using the same statistical method. 1,worn (d); Obtain a set of data pairs Then, linear regression was applied to fit the data points to obtain a linear mapping relationship. , where k is the drift coefficient; the compensation model for the second threshold T2 is also calibrated in this way.

[0036] Calibrate the edge effect compensation logic parameters based on the real-time tool position, including determining the width L between the cut edge buffers. buffer The shape of the attenuation function f(x) is determined; using a well-maintained tool, a cut is made in the central region of a standard workpiece, and its first threshold T1 is recorded; microcracks induced by the tool at the cut edge during the standard cutting task are observed using a high-power microscope, and their maximum propagation length L in the cutting direction is measured. carck If L carck The statistical average is 0.28 mm; the width L between the cut-out edge buffer zones is set. buffer If the length is greater than this and a safety margin is added, set L. buffer =0.6mm; Within this buffer zone, a linear decay function f(x) is set, which is applied at x = L buffer The threshold returns to normal, reaching its lowest point when x = 0, i.e., the cutting endpoint. Taking 0.5 × T1 as an example, this threshold is determined to be used to calculate the dynamic local first threshold T. 1,local A function of (x): Where x is the distance between the current position of the tool and the cutting endpoint; the parameters for calibrating the asymmetric speed recovery control logic are used to determine the preset hysteresis ratio P in the stability verification condition. lag Preset minimum time window t min And the upper limit of acceleration 'a' in the limited climb logic rec In impact testing, high-frequency data acquisition can be used, employing accelerometer or current loop signal spectrum analysis, to record the average time required from the peak of the load impact until the system's mechanical vibration decays and converges to the background noise level. If this time is statistically 38ms, then a preset minimum time window t is set. min =50ms; Simultaneously analyzing this batch of data, when the mechanical vibration converges, the absolute value of the real-time load current has fallen back to below 15% of the peak impact value. Based on this, the preset hysteresis ratio P is set. lag It is 15%, in the control logic, t min and P lag Satisfying any one of these conditions constitutes a stability confirmation condition; for the upper limit of acceleration a rec Set a rec Less than the maximum physical acceleration a of the servo system max , with a rec =0.2×a max For example, it is used to slow down the recovery speed.

[0037] Example 6: In a system deployment scenario, when the control method of the present invention is first applied to a new dicing machine with uncalibrated characteristics or a new batch of workpiece materials, a pre-system procedure can be executed to manage the initial entry boundary conditions of the control logic. Specifically, when the control system performs the initial cutting entry stage on the Z-axis (depth of cut axis), it will temporarily bypass or shield the weighted differential of step A and the discrete statistics of step B, so that the final speed adjustment coefficient is forced to remain at 1.0. The shielding state continues until the controller confirms that the tool has completely entered the workpiece to reach the preset cutting depth and that the X-axis or Y-axis (feed axis) has started executing the cutting path command. This boundary condition management is used to prevent the inevitable high-load transient change generated when the tool enters the material vertically from being misjudged as a cutting abnormality, and to avoid triggering unnecessary feed rate suppression before the cutting task has officially started.

[0038] After the control system enters the cutting path execution phase, the implementation procedure of the feedforward control logic based on the defect association of adjacent paths is as follows: The system maintains a dynamically updated list of risk node coordinates in the controller memory. When the final speed adjustment coefficient output in step C is lower than the preset event recording threshold during the cutting process (this threshold can be calibrated to 0.5), indicating that speed suppression has occurred, the controller immediately captures the coordinate position of the current tool on the cutting path and stores the coordinates in the risk node coordinate list. Before executing the next adjacent cutting path, the control system traverses the list, maps all recorded risk node coordinates to the new path, and defines the virtual risk area by applying an asymmetric pre-read range, setting it to 0 upstream of the risk node coordinates. Extending from 0.3mm downstream to 0.1mm downstream; when the tool travels along the new path under the guidance of the control system, once its real-time position enters the pre-read range of any virtual risk area, the feedforward control logic obtains the highest priority, forcibly switching the feed rate command to the preset safe speed, calibrated to 25% of the feed rate setting value, and maintaining this safe speed until the tool completely leaves the virtual risk area; in addition, the system performs dynamic verification on the list. If the real-time load change rate K(t) calculated in step A does not exceed the first threshold T1 throughout the entire process of the tool passing through the virtual risk area at a safe speed, the system determines that the defect is discontinuous or has disappeared on the current path, and then removes the risk node coordinates from the list.

[0039] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An intelligent control method for a dicing saw cutting path, characterized in that, The method comprises: Real-time acquisition of the load current signal of the main shaft servo motor of the dicing machine at a predetermined sampling frequency, and storage of N historical sampling points of the load current signal in a sliding window data queue; In a control period, the current feed speed setting value of the dicing machine is obtained, and based on the N sampling point data in the sliding window data queue, the following calculation steps are performed in parallel: Step A, using a sliding window weighted differential logic, by setting a weight coefficient that is larger for sampling points closer to the current time, the trend slope of the N sampling point data is fitted and calculated to obtain a real-time load change rate; Based on the real-time load change rate, a first speed adjustment coefficient is obtained by querying a preset first nonlinear gain scheduling relationship; the first nonlinear gain scheduling relationship is configured as: when the absolute value of the real-time load change rate exceeds a first threshold, the first speed adjustment coefficient exponentially decays with the increase of the absolute value of the real-time load change rate; Step B, the average absolute deviation value between the N sampling point data and the average of the N sampling point data is calculated to obtain a oscillation state quantity representing the dispersion of the load current signal; And based on the oscillation state quantity, an oscillation suppression coefficient is obtained by querying a preset second gain scheduling relationship; The second gain scheduling relationship is configured such that when the oscillation state quantity exceeds an oscillation threshold, the oscillation suppression coefficient decreases with the increase of the oscillation state quantity; Step C, the first speed adjustment coefficient and the oscillation suppression coefficient are multiplied to obtain a final speed adjustment coefficient; the final speed adjustment coefficient is used to modify the feed speed setting value in real time, and the modified speed command is sent to the feed shaft servo system for execution.

2. The intelligent control method of the cutting path of the dicing saw according to claim 1, wherein, In step A, the sliding window weighted differential logic calculates the real-time load change rate through the following discretization formula: wherein K(t) is the real-time load change rate, I t-i is the load current value of the i-th sampling point calculated from the current time, I t-N is the load current value of the earliest sampling point in the sliding window data queue, w i is the preset weight coefficient, and satisfies w0> w1> •••> w N-1 , and ΔT is the sampling time interval.

3. The intelligent control method of the cutting path of the dicing saw according to claim 1, wherein, The first nonlinear gain scheduling relationship specifically includes three consecutive control intervals: a stable interval, when the absolute value of the real-time load change rate is less than the first threshold, a constant unit gain coefficient is output; a suppression interval, when the absolute value of the real-time load change rate is between the first threshold and the second threshold, a gain coefficient that exponentially decays according to a natural exponential function is output; a protection interval, when the absolute value of the real-time load change rate is greater than the second threshold, a constant minimum safety gain coefficient is output, and the minimum safety gain coefficient is not greater than zero point one.

4. The intelligent control method of the cutting path of the scriber according to claim 3, characterized in that, The method further comprises a parameter drift compensation step based on the whole life cycle of the tool: recording the cumulative cutting distance of the current tool of the dicing machine; establishing a positive linear mapping relationship between the cumulative cutting distance and the first threshold and the second threshold; Before starting each cutting task, the values of the first threshold and the second threshold are automatically increased according to the current cumulative cutting distance to offset the increase of background load fluctuation noise caused by tool wear.

5. The intelligent control method of the cutting path of the scriber according to claim 1, characterized in that, The method further comprises a feedforward control step based on the correlation of adjacent path defects: when the final speed adjustment coefficient is lower than a preset event recording threshold, the coordinate position of the current tool on the current cutting path is recorded as a risk node; When performing the next cutting path operation, the coordinates of the risk node are mapped to the next cutting path to define a virtual risk area; Real-time monitoring of the tool position, when the tool position enters a pre-reading range of the virtual risk area, the feed speed of the dicing machine is forcibly reduced to a preset safety speed in priority to the first nonlinear gain scheduling relationship; If the real-time load change rate does not exceed the first threshold value during the virtual risk area, the marking of the virtual risk area is removed.

6. The intelligent control method of the cutting path of the dicing saw according to claim 1, wherein, The method further comprises an edge effect compensation step based on the real-time position of the tool: the position coordinates of the dicing machine tool on the cutting path are obtained in real time, and it is judged whether the position coordinates are located within the preset cutting edge buffer interval; if located within the cutting edge buffer interval, the value of the first threshold value is dynamically reduced according to the distance between the current position of the tool and the cutting end point, and the sensitivity of the feed speed adjustment within the cutting edge buffer interval is higher than that outside the buffer interval according to the preset attenuation function.

7. The intelligent control method of the cutting path of the dicing saw according to claim 1, characterized in that, The method further comprises an asymmetric speed recovery control step: after the feed speed of the dicing machine is reduced by using the final speed adjustment coefficient, the control system enters a dissipation holding state; in the dissipation holding state, only when the real-time load change rate falls below the first threshold value and the preset stability confirmation condition is met, the feed speed is allowed to be increased. The stability confirmation condition includes: the absolute value of the real-time load current decreases by more than a preset hysteresis ratio or the duration after the feed speed is reduced exceeds a preset minimum time window.

8. The intelligent control method of the cutting path of the scriber according to claim 7, characterized in that, After the dissipation holding state is removed, the process of increasing the feed speed follows a limited climb logic, limiting the upper limit of the acceleration of the speed recovery, so that the feed speed is recovered to the feed speed set value according to a linear slope, and the upper limit of the acceleration is less than the maximum physical acceleration of the dicing machine servo system.

9. The intelligent control method of the cutting path of the dicing saw according to claim 1, wherein, In step B, the oscillation suppression coefficient is applied to the multiplication coupling only when the absolute value of the real-time load change rate is less than the first threshold value, ensuring that the suppression action for high-frequency micro-oscillation is independently performed when the load current signal is in a state where the trend is stable but the micro-dispersion is intense.

10. The intelligent control method of the cutting path of the dicing saw according to claim 1, wherein, The multiplication coupling of the first speed adjustment coefficient and the oscillation suppression coefficient follows the minimum priority principle: if the product of the first speed adjustment coefficient and the oscillation suppression coefficient is less than the minimum running speed coefficient preset by the system, the final speed adjustment coefficient is forcibly set to the minimum running speed coefficient, maintaining the minimum torque required for the servo motor to overcome the static friction.

Images