A magnetic base drill hole depth self-adaptive control method

Abstract

This application relates to the field of adaptive control technology, and in particular to an adaptive control method for drilling depth of a magnetic drill bit. The method includes acquiring multiple real-time signals from the magnetic drill bit during the drilling process; for any given moment, using the ratio of the magnetic reaction force to the rated magnetic attraction force as the magnetic attraction preload margin; estimating the remaining wall thickness based on a preset a priori upper bound of the plate thickness and a feed displacement estimate; calculating the dimensionless thrust slope, and using the product of the dimensionless thrust slope and the power ratio of the drill bit diameter relative to the remaining wall thickness as the penetration proximity; constructing a predicted penetration depth based on the penetration proximity and its time derivative; synthesizing a predictive adaptive correction factor based on the magnetic attraction preload margin and the predicted penetration depth; and using the predictive adaptive correction factor to scale the feed rate to obtain the feed rate. This application has the effect of reducing the impact load generated at the moment of penetration during drilling.

Description

Technical Field

[0001] This application relates to the field of adaptive control technology, and in particular to an adaptive control method for drilling depth of a magnetic drill. Background Technology

[0002] A magnetic drill is a portable drilling device that uses electromagnetic or permanent magnet adsorption to fix itself to the surface of a metal workpiece. It is widely used in steel structure construction, bridge manufacturing, shipbuilding, rail transit equipment, and the installation and maintenance of engineering machinery. Compared to traditional bench drills, magnetic drills offer advantages such as flexible installation, adaptability to complex spatial operations, and the ability to drill in vertical, inclined, and even inverted conditions, making them widely applicable in on-site machining.

[0003] When performing drilling operations, existing magnetic drills typically preset the drilling depth based on the operator's experience, or control the drill bit's feed stroke through mechanical limits or displacement detection. When the workpiece thickness is known, these methods can meet basic machining requirements. However, in applications such as steel structure installation, equipment maintenance, and on-site modifications, the actual thickness of the workpiece is often difficult to accurately obtain, and the drilling process is usually done blindly. In this situation, as the drill bit approaches the bottom of the workpiece and is about to penetrate, the cutting load changes significantly. If the original feed rate is maintained, a large impact load can easily be generated at the moment of penetration.

[0004] In existing technologies, drilling depth control for magnetic drills can be broadly categorized into three types: mechanical limiting, segmented current feed, and closed-loop cutting force control. Mechanical limiting solutions rely on limit switches or mechanical stops, such as multi-faceted adsorption structures and anchor drilling depth control devices with a single displacement criterion, offering a simple depth determination. Segmented current feed solutions use speed and current monitoring to achieve segmented switching between rapid traverse, feed, and retraction, with segment thresholds set empirically. Closed-loop cutting force control solutions use constant thrust or constant torque as the single control objective, and attempts are made in areas such as bone drill penetration detection, CFRP stack interface recognition, and PCB current-thrust mapping to support closed-loop control through real-time sensing of thrust or current.

[0005] However, most of the above solutions use a single detection parameter for control, which makes it difficult to fully reflect the correlation between the magnetic adsorption state, the change of cutting load and the critical state of drill bit penetration during the magnetic drilling process. When drilling with unknown workpiece thickness, the magnetic drill cannot accurately identify the working state when the drill bit is close to penetration, which makes it easy for the drill bit to generate a large impact when penetrating the workpiece, affecting the drilling quality and the stability of equipment operation. Summary of the Invention

[0006] To address the issues of poor drilling quality and equipment stability in related technologies, this application provides an adaptive control method for drilling depth of a magnetic base drill.

[0007] Firstly, this application provides an adaptive control method for drilling depth of a magnetic drill bit, employing the following technical solution:

[0008] An adaptive control method for drilling depth of a magnetic base drill includes acquiring multiple real-time signals during the drilling process of the magnetic base drill; the real-time signals include: spindle current, magnetic base reaction force, axial drilling reaction force and feed displacement, no-load current and magnetic base rated adsorption force.

[0009] At any given moment, the ratio of the magnetic seat reaction force to the magnetic seat rated adsorption force is taken as the magnetic adsorption preload margin; the remaining wall thickness is estimated based on the preset a priori upper limit of the plate thickness and the feed displacement estimate; the dimensionless thrust slope is calculated, and the product of the dimensionless thrust slope and the power ratio of the drill bit diameter relative to the remaining wall thickness is taken as the penetration proximity.

[0010] Based on the penetration proximity and its time derivative, a predicted penetration is constructed. Based on the magnetic adsorption preload margin and the predicted penetration, a predictive adaptive correction factor is synthesized. The predictive adaptive correction factor is used to simultaneously scale the feed rate output by the PID.

[0011] Simultaneously, multiple real-time signals, including spindle current, magnetic seat reaction force, axial drilling reaction force, feed displacement, no-load current, and rated magnetic seat adsorption force, are collected to construct multi-dimensional state parameters such as magnetic adsorption preload margin, remaining wall thickness, penetration proximity, and predicted penetration. A predictive adaptive correction factor is used to synchronously adjust the feed rate and PID proportional gain, enabling the control system to comprehensively reflect the magnetic seat adsorption state, cutting load changes, and dynamic characteristics as the drill approaches penetration. This allows for a preemptive reduction in feed rate and control gain just before the drill penetrates the workpiece. Compared to existing technologies that adjust feed rate based on a single detection, this application reduces the impact load generated at the moment of penetration, improves drilling quality and equipment operational stability, and enhances adaptability to unknown plate thickness conditions.

[0012] Optionally, the process of obtaining the axial drilling reaction force includes:

[0013] The current difference is obtained by subtracting the no-load current from the current spindle current.

[0014] The axial drilling reaction force is obtained by multiplying the current difference by a pre-calibrated current thrust calibration coefficient.

[0015] It is understandable that the current and the reaction force on the spindle are linearly related. Therefore, the current thrust calibration coefficient can be determined in advance through experiments, and then the axial drilling reaction force estimate can be derived from the change of current in the actual production process.

[0016] Optionally, the current thrust calibration factor is determined experimentally, characterizing the relationship between the increment of the current relative to the no-load current and the axial drilling reaction force.

[0017] Establish the correspondence between the spindle current increment and the axial drilling reaction force so that the current signal can be accurately mapped to the drilling load change, thereby improving the accuracy of drilling reaction force estimation based on current and reducing measurement errors caused by changes in working conditions.

[0018] Optionally, the steady-state value of the magnetic seat reaction force when the magnetic seat is attached to the workpiece and is stable under no-load is recorded as the rated adsorption force of the magnetic seat.

[0019] When the magnetic base is attached to the workpiece and is stable under no-load, the steady-state value of the magnetic base reaction force is recorded as the rated adsorption force of the magnetic base. This allows the magnetic adsorption pre-tightening margin calculated subsequently to truly reflect the change in the current adsorption state relative to the standard adsorption state, thereby improving the accuracy of the magnetic adsorption state assessment.

[0020] Optionally, estimating the remaining wall thickness based on a preset a priori upper bound of the plate thickness and the feed displacement includes: multiplying the difference between the a priori upper bound and the feed displacement by a preset magnetic margin sensitivity index power of the magnetic adsorption preload margin to obtain the remaining wall thickness estimate.

[0021] The difference between the prior upper limit of the plate thickness and the feed displacement is taken as the geometric remaining thickness. Nonlinear correction is performed in combination with the magnetic adsorption preload margin to realize the dynamic estimation of the remaining wall thickness. This makes the remaining wall thickness estimation result not only consider the actual feed depth of the drill bit, but also the impact of changes in the magnetic circuit state on the penetration risk, thereby improving the accuracy of judging the state of the drill bit approaching the bottom of the workpiece.

[0022] Optionally, calculating the dimensionless thrust slope includes: calculating the absolute value of the least squares first-order difference of the sliding window of the axial drilling reaction force, and using the product of the absolute value and the ratio of the reference time constant to the working condition calibration thrust as the dimensionless thrust slope.

[0023] The axial drilling reaction force is processed by sliding window least squares first-order difference, and normalized by using reference time constant and working condition calibration thrust to obtain dimensionless thrust slope, which can effectively characterize the trend of drilling load change and reduce the influence of absolute thrust value difference under different working conditions and different materials.

[0024] Optionally, constructing a predicted penetration degree based on the penetration proximity and its time derivative includes: performing a first-order difference on the penetration proximity to obtain the time derivative of the penetration proximity;

[0025] The step response identification result of the feed actuator is used as the equivalent time constant of the actuator;

[0026] The predicted penetration is obtained by multiplying the time derivative of the penetration proximity by the equivalent time constant of the actuator and the prediction gain, and then summing the product with the penetration proximity.

[0027] By performing time derivative analysis on the penetration proximity and constructing a predicted penetration rate in conjunction with the actuator's equivalent time constant, the penetration state can be predicted in advance over a period of time. This allows the system to reserve control response time before actual penetration occurs, thereby improving the foresight of the control system and reducing the overshoot risk caused by actuator response lag.

[0028] Optionally, synthesizing a predictive adaptive correction factor based on the magnetic adsorption preload margin and the predicted penetration includes: multiplying the magnetic adsorption preload margin by a negative exponential function of the product of the attenuation sensitivity coefficient and the predicted penetration to obtain the predictive adaptive correction factor.

[0029] The magnetic adsorption preload margin and the predicted penetration rate are introduced into the predictive adaptive correction factor and adjusted using a negative exponential decay method. This allows the control parameters to change smoothly as the penetration risk increases or the adsorption capacity decreases, avoiding the abrupt adjustment problem caused by traditional threshold control and improving the continuity and stability of the feed control process.

[0030] Optionally, it also includes: when the increasing trend and duration of the penetration proximity meet the conditions, sequentially entering the penetration warning state and the penetration confirmation state, and in the penetration confirmation state, performing a pecking drill retraction and recording the final drilling depth.

[0031] This application has the following technical effects:

[0032] By fusing magnetic adsorption state with feed displacement information, the remaining wall thickness is dynamically estimated, and a penetration proximity degree is constructed by combining thrust variation trends. This more accurately characterizes the actual state of the drill bit as it approaches the bottom of the workpiece. Furthermore, the penetration proximity degree and its variation trends are used to predict future penetration risks, generating a predicted penetration degree. This is combined with the magnetic adsorption preload margin to form a predictive adaptive correction factor, which synchronously and continuously adjusts the feed rate and PID control gain, allowing the drill bit to gradually decelerate rather than suddenly brake before penetration. This achieves accurate identification and early intervention during drilling of unknown plate thicknesses, effectively reducing the impact load at the moment of drill bit penetration, improving drilling quality and equipment operational stability. This solves the technical problem in existing technologies where a single detection parameter is insufficient to accurately identify the penetration state, leading to significant penetration impact. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of an adaptive control method for drilling depth of a magnetic base drill according to an embodiment of this application. Detailed Implementation

[0034] This application discloses an adaptive control method for drilling depth of a magnetic base drill.

[0035] Reference Figure 1 An adaptive control method for drilling depth of a magnetic base drill includes steps S1-S3.

[0036] S1: Collects multiple real-time signals during the drilling process of the magnetic base drill; real-time signals include: spindle current, magnetic base reaction force, axial drilling reaction force and feed displacement, no-load current and magnetic base rated adsorption force.

[0037] The real-time signals specifically include spindle current, magnetic base reaction force, axial drilling reaction force, and feed displacement. The spindle current is acquired by the spindle motor phase current transformer and then amplified and filtered using differential amplification and anti-aliasing. The magnetic base reaction force is synthesized by four strain gauges on the bottom surface of the magnetic base. These four strain gauges are connected to the strain conditioning module via a Wheatstone bridge circuit, and the bridge circuit output is linearly related to the normal pressure on the bottom surface of the magnetic base. The feed displacement is acquired by an incremental encoder at the end of the feed screw, with a resolution of at least ten micrometers.

[0038] A unified time base is used for all four signals, with a sampling period of 1 millisecond. Each signal is first filtered by a second-order Butterworth low-pass filter with a preset cutoff frequency of 50Hz to suppress tool cutting ripple. Then, it is filtered by a multi-point median filter to suppress occasional spikes. The median filter window is set to five sampling points in this embodiment. The unified time base filtering ensures the synchronization of subsequent calculations of coupling characteristics such as magnetic adsorption preload margin and penetration proximity.

[0039] Next, the no-load current and the rated adsorption force of the magnetic base are calibrated.

[0040] The no-load current is determined by the average phase current during the spindle idling period one second before power-on, together with a temperature rise correction table based on the spindle temperature sensor. Specifically, the spindle temperature at power-on is used as the lookup index to obtain the temperature rise correction amount from the temperature rise correction table, and the average phase current is added to the temperature rise correction amount to obtain the current no-load current. The temperature rise correction table is pre-calibrated through factory bench testing.

[0041] The rated adsorption force of the magnetic seat is determined by the steady-state value of the magnetic seat reaction force when it is first attached to the workpiece and under stable no-load conditions. Specifically, the magnetic seat reaction force is continuously sampled when the spindle is idling and there is no feed. The average value is taken as the rated adsorption force of the magnetic seat after the signal variance is less than the preset fluctuation threshold.

[0042] The axial drilling reaction force is estimated based on the change in current: the current difference is obtained by subtracting the no-load current from the current spindle current; the difference between the current current and the no-load current is multiplied by a pre-calibrated current thrust calibration factor to obtain the axial drilling reaction force. The current thrust calibration factor is pre-calibrated through factory bench testing.

[0043] Specifically, the formula for calculating the axial drilling reaction force is as follows:

[0044] ;

[0045] in For the estimated axial drilling reaction force, the unit ; The current-thrust calibration coefficient characterizes the relationship between current and axial drilling reaction force. It is mainly used to map current to axial drilling reaction force. The unit is N / A. It is obtained from the factory bench test. Specifically, it refers to the slope of the straight line formed by fitting the axial drilling reaction force measured on the test bench with the current measured during the test and the axial drilling reaction force measured during the test. For a moment Effective value of spindle current, in units ; For temperature The no-load current, in amperes (A).

[0046] S2: For any given moment, the ratio of the magnetic seat reaction force to the magnetic seat rated adsorption force is taken as the magnetic adsorption preload margin; the remaining wall thickness is estimated based on the preset a priori upper limit of the plate thickness and the feed displacement estimate; the dimensionless thrust slope is calculated, and the product of the dimensionless thrust slope and the power ratio of the drill bit diameter relative to the remaining wall thickness is taken as the penetration proximity.

[0047] The magnetic adsorption preload margin is defined as the ratio of the current magnetic seat reaction force to the rated magnetic seat adsorption force. When the magnetic adsorption preload margin approaches 1, the magnetic seat adsorption is sufficient; when the magnetic adsorption preload margin approaches zero, the magnetic seat is about to be lifted off.

[0048] The formula for calculating the preload margin of magnetic adsorption can be expressed as:

[0049] ;

[0050] In the formula, Indicates time The magnetic adsorption preload margin is dimensionless. This represents the current magnetic base reaction force, measured in Newtons (N). This indicates the rated magnetic attraction force, measured in Newtons (N).

[0051] To prevent the denominator from being too small and causing numerical overflow, a minimum value protection is applied to the denominator; at the same time, the magnetic adsorption preload margin is clamped in the normalized range of 0.05 to 1.0.

[0052] The remaining wall thickness is estimated by back-calculating the magnetic adsorption preload margin and the feed displacement.

[0053] The estimated remaining wall thickness is determined by both the workpiece's geometric remaining thickness and the change in the magnetic circuit air gap. Specifically, the difference between the a priori upper limit of the plate thickness and the feed displacement is taken as the geometric remaining thickness term, and then multiplied by the power of the preset magnetic margin sensitivity index of the magnetic adsorption preload to obtain the estimated remaining wall thickness.

[0054] The magnetic margin sensitivity index is determined by referring to a table based on the material grade: 0.5 for carbon steel, 0.4 for stainless steel, and 0.6 for cast iron.

[0055] For any given moment, the formula for estimating the remaining wall thickness can be expressed as:

[0056] ;

[0057] In the formula, Indicates time The estimated remaining wall thickness, in millimeters; This indicates the a priori upper limit of the plate thickness, which is input by the user or given by the rangefinder, and the unit is millimeters. Indicates time The feed displacement, in millimeters; This represents the preset magnetic margin sensitivity index, which is dimensionless and ranges from 0.3 to 0.8, with 0.5 for carbon steel. It is obtained by referring to a table based on the material grade. Indicates time The magnetic adsorption preload margin.

[0058] When the feed displacement approaches the a priori upper limit of the plate thickness, the geometric residual thickness converges; the decrease in magnetic adsorption preload margin is mapped to the nonlinear contraction of the residual wall thickness through the multiplicative correction of the magnetic margin sensitivity exponent, which is equivalent to the increase in the magnetic circuit air gap causing the residual wall thickness to approach the critical value.

[0059] Calculate the dimensionless thrust slope. The dimensionless thrust slope is normalized by the first-order difference of the thrust using the calibrated thrust under operating conditions and the reference time constant. Specifically, the absolute value of the least squares first-order difference of the sliding window of the axial drilling reaction force is multiplied by the preset reference time constant and divided by the calibrated thrust under operating conditions to obtain the dimensionless thrust slope. The sliding window is 20 milliseconds in this embodiment, the reference time constant is 0.1 seconds in this embodiment, and the working condition calibration thrust is the average value of the estimated axial drilling reaction force during the stable cutting phase (e.g., the first minute of drilling the workpiece).

[0060] Finally, the penetration proximity is calculated using the dimensionless thrust slope. The penetration proximity is obtained by multiplying the dimensionless thrust slope by the ratio of the drill bit diameter to the geometrical power of the remaining wall thickness. The specific formula is as follows:

[0061] ;

[0062] In the formula, Indicates the degree of penetration proximity, which is dimensionless; Indicates time The dimensionless thrust slope; This indicates the drill bit diameter, in millimeters. This represents the geometric magnification index, which is dimensionless. In this embodiment, it is 1.2 for carbon steel. Indicates time The estimated remaining wall thickness.

[0063] As the remaining wall thickness decreases, the ratio of the drill bit diameter to the remaining wall thickness increases. The dimensionless thrust slope synchronously reflects this instantaneous change. The increase in penetration proximity can only originate from the contraction of the remaining wall thickness or the increase in the thrust slope. In this embodiment, the upper limit of penetration proximity is clamped at 10 to prevent numerical overflow.

[0064] By embedding the estimated remaining wall thickness into the magnetic adsorption preload margin, and then coupling the remaining wall thickness with the normalized thrust slope in the form of a geometric amplification exponent to obtain the penetration proximity, this step achieves the serial fusion of the magnetic adsorption boundary, the evolution of the remaining wall thickness, and the thrust slope, thereby reducing the false triggering rate across operating conditions.

[0065] S3: Construct a predicted penetration degree based on the penetration proximity and its time derivative, and synthesize a predictive adaptive correction factor based on the magnetic adsorption preload margin and the predicted penetration degree. The predictive adaptive correction factor simultaneously scales the feed rate output by the PID.

[0066] Based on the penetration proximity and its time derivative, a predicted penetration is constructed:

[0067] The predicted penetration is a first-order Taylor extrapolation of the penetration proximity after applying the equivalent time constant of the actuator. It represents the prediction of the penetration proximity at the future actuator response time, allowing for braking stroke for pre-braking.

[0068] Specifically, the first-order difference of the penetration proximity is performed to obtain the time derivative of the penetration proximity; the step response identification result of the feed actuator is used as the equivalent time constant of the actuator; the product of the time derivative of the penetration proximity, the equivalent time constant of the actuator, and the prediction gain is summed with the penetration proximity to obtain the result.

[0069] The formula for predicting the penetration index can be expressed as:

[0070] ;

[0071] In the formula, Indicates time The predictive penetration is dimensionless; Indicates the degree of penetration proximity, which is dimensionless; This represents the time derivative of the proximity degree, in seconds; This represents the equivalent time constant of the actuator, in seconds, and is obtained from the step response of the feed actuator. In this embodiment, it is taken as 30 milliseconds to 80 milliseconds. This represents the predicted gain, which is dimensionless; in this embodiment, it is taken as 1.2.

[0072] Next, a predictive adaptive correction factor is synthesized. This predictive adaptive correction factor uses the magnetic adsorption preload margin as a multiplicative constraint and the predicted penetration rate as an exponential decay term. When the magnetic adsorption preload margin approaches 1 and the predicted penetration rate approaches 0, the predictive adaptive correction factor approaches 1, and the PID controller maintains its original performance. When the predicted penetration rate increases or the magnetic adsorption preload margin decreases, the predictive adaptive correction factor decays exponentially. Its calculation formula is as follows:

[0073] ;

[0074] In the formula, Indicates time The predictive adaptive correction factor is dimensionless. Indicates time The magnetic adsorption preload margin is dimensionless. This represents the attenuation sensitivity coefficient, which is dimensionless. In this embodiment, it is taken as 5 for carbon steel. Indicates time Predictive penetration.

[0075] The feed rate is scaled using a predictive adaptive correction factor.

[0076] In some embodiments, the initial feed rate under the working condition is multiplied by a predictive adaptive correction factor and then subjected to a maximum value calculation with a safety lower limit to obtain the feed rate. The safety lower limit prevents chip clogging and stagnation.

[0077] Specifically, feed rate synthesis:

[0078] ;

[0079] In the formula, For a moment The feed rate, in mm / s; The initial feed for this operating condition, in mm / s, is given by the process table; This is the safe lower limit for feed rate, in mm / s, to prevent chip clogging; Indicates time The predictive adaptive correction factor.

[0080] In other embodiments, the basic proportional gain of the PID can also be corrected simultaneously by a predictive adaptive correction factor, that is, by multiplying the predictive adaptive correction factor by the basic proportional gain to obtain the proportional gain, while keeping the integral gain and derivative gain unchanged.

[0081] The basic proportional gain, integral gain, and derivative gain are pre-tuned on a standard 20 mm carbon steel plate using the Ziegler-Nichols method.

[0082] By introducing the time derivative of the penetration proximity into the exponential decay term of the predictive adaptive correction factor, this step enables the PID controller to begin decaying the feed command before the penetration proximity reaches the threshold. Through dual-layer scaling of the feed command and proportional gain, this step upgrades the controller from abrupt switching to continuous softening, thereby reducing the number of step commutations of the feed motor.

[0083] In some embodiments, a three-state finite state machine can be introduced as a discrete fallback mechanism based on the continuous softening of the predictive adaptive correction factor to handle extreme cases and the drill-pecking closed loop. The three-state finite state machine includes a normal cutting state, a penetration warning state, and a penetration confirmation state. In addition to the above three states, a magnetic seat protection state is triggered when the magnetic adsorption preload margin is lower than the magnetic adsorption protection threshold.

[0084] During normal cutting, the feed rate is calculated as the product of the predictive adaptive correction factor synthesized in the third step and the initial feed rate of the working condition. When the time derivative of the penetration proximity is not less than the warning slope threshold, and the penetration proximity is not less than the low-level threshold, and the preset number of warning judgment cycles is maintained continuously, the system transitions from the normal cutting state to the penetration warning state. In this embodiment, the warning slope threshold is 8 for carbon steel; the low-level threshold is 0.3; and the number of warning judgment cycles is 5.

[0085] After entering the penetration warning state, the feed speed is limited to a preset proportion of the initial feed speed of the working condition, which is 50% in this embodiment, thereby providing braking stroke in advance for insufficient actuator bandwidth.

[0086] During the penetration warning state, the predicted penetration rate is continuously monitored. When the predicted penetration rate is not less than the penetration confirmation threshold and remains at the preset confirmation period, the system transitions from the penetration warning state to the penetration confirmation state. In this embodiment, the penetration confirmation threshold is set to 1.0; the confirmation period is set to 3.

[0087] After entering the penetration confirmation state, the feed rate is smoothly reduced to zero according to the exponential decay law of the preset time constant. The exponential decay expression is the feed rate multiplied by the natural exponent of the negative ratio of time to time constant. The time constant is determined by referring to a table based on the material grade. Then, the drill retracts at a constant speed to a preset safety displacement to complete chip breaking. In this embodiment, the preset safety displacement is one to two millimeters. After chip breaking, it is determined whether the remaining feed displacement reaches the through hole. If it does not reach the through hole, the process returns to normal cutting mode and continues drilling; if it reaches the through hole, the process enters the completed state.

[0088] The conditions for reverting from the penetration warning state to the normal cutting state are as follows: When the time derivative of the penetration proximity is not greater than a preset proportion of the warning slope threshold, and the penetration proximity is not greater than the low-level threshold, and a preset number of reverting judgment cycles are maintained continuously, the system switches from the penetration warning state to the normal cutting state. The preset proportion of the warning slope threshold is taken in this embodiment as... The rollback determination cycle number is 10 in this embodiment.

[0089] In some embodiments, when the magnetic adsorption preload margin is lower than the magnetic adsorption protection threshold, the system enters a magnetic base protection state. The calibrated thrust is limited according to the ratio of the magnetic adsorption preload margin to the magnetic adsorption protection threshold, and an audible and visual alarm is triggered. In this embodiment, the magnetic adsorption protection threshold is set to 0.2. The specific limiting process is as follows: the calibrated thrust is multiplied by the ratio of the magnetic adsorption preload margin to the magnetic adsorption protection threshold (…). This serves as the new upper limit of thrust for the operating condition calibration; at the same time, the audible and visual alarm unit issues an audible and visual warning to remind the operator to check the magnetic seat adsorption status.

[0090] After the drill bit is successfully cut and the cutting is completed in the penetration confirmation state, if the change in feed displacement within the preset observation time is less than the displacement convergence threshold and the axial drilling reaction force is less than the no-load thrust threshold, the current feed displacement is recorded as the final drilling depth and the feed is stopped. In this embodiment, the preset observation time is 0.5; the displacement convergence threshold is 10 micrometers; and the no-load thrust threshold is given as 10% of the steady-state average of the first layer of cutting.

[0091] By employing continuous softening with a predictive adaptive correction factor and discrete fallback with a three-state finite state machine, this step provides both continuous and discrete safety nets. Under blind drilling conditions where the workpiece thickness is unknown, the continuous softening layer handles conventional overshoot, while the discrete fallback layer handles extreme cases and the pecking drill loop. The warning state provides pre-braking stroke, the confirmation state provides final hard protection, the normal state provides a return path, and the magnetic base protection state provides adsorption boundary protection. The entire chain is driven serially by magnetic adsorption preload margin, estimated remaining wall thickness, penetration proximity, predicted penetration, and predictive adaptive correction factor, controlling feed speed and state switching. Each stage output serves only as the input for the next stage.

[0092] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for adaptive control of drilling depth using a magnetic drill bit, characterized in that, Collect multiple real-time signals during the drilling process of the magnetic base drill; the real-time signals include: spindle current, magnetic base reaction force, axial drilling reaction force and feed displacement, no-load current and magnetic base rated adsorption force. At any given moment, the ratio of the magnetic seat reaction force to the magnetic seat rated adsorption force is taken as the magnetic adsorption preload margin; the remaining wall thickness is estimated based on the preset a priori upper limit of the plate thickness and the feed displacement estimate; the dimensionless thrust slope is calculated, and the product of the dimensionless thrust slope and the power ratio of the drill bit diameter relative to the remaining wall thickness is taken as the penetration proximity. Based on the penetration proximity and its time derivative, a predicted penetration is constructed. Based on the magnetic adsorption preload margin and the predicted penetration, a predictive adaptive correction factor is synthesized. The predictive adaptive correction factor is used to simultaneously scale the feed rate output by the PID.

2. The adaptive control method for drilling depth of a magnetic base drill according to claim 1, characterized in that, The process of obtaining the axial drilling reaction force includes: The current difference is obtained by subtracting the no-load current from the current spindle current. The axial drilling reaction force is obtained by multiplying the current difference by a pre-calibrated current thrust calibration coefficient.

3. The adaptive control method for drilling depth of a magnetic base drill according to claim 2, characterized in that, The current thrust calibration coefficient is determined by experiment and characterizes the relationship between the increment of current relative to no-load current and the axial drilling reaction force.

4. The adaptive control method for drilling depth of a magnetic base drill according to claim 1, characterized in that, The steady-state value of the magnetic seat reaction force when the magnetic seat is attached to the workpiece and is stable under no-load is recorded as the rated adsorption force of the magnetic seat.

5. The adaptive control method for drilling depth of a magnetic base drill according to claim 1, characterized in that, The estimation of the remaining wall thickness based on the preset a priori upper bound of the plate thickness and the feed displacement includes: multiplying the difference between the a priori upper bound and the feed displacement by the preset magnetic margin sensitivity index power of the magnetic adsorption preload margin to obtain the remaining wall thickness estimate.

6. The adaptive control method for drilling depth of a magnetic base drill according to claim 1, characterized in that, Calculating the dimensionless thrust slope includes: calculating the absolute value of the least squares first-order difference of the sliding window of the axial drilling reaction force, and using the product of the absolute value and the ratio of the reference time constant to the working condition calibration thrust as the dimensionless thrust slope.

7. The adaptive control method for drilling depth of a magnetic base drill according to claim 1, characterized in that, Constructing a predicted penetration rate based on the penetration proximity and its time derivative includes: performing a first-order difference on the penetration proximity to obtain the time derivative of the penetration proximity; The step response identification result of the feed actuator is used as the equivalent time constant of the actuator; The predicted penetration is obtained by multiplying the time derivative of the penetration proximity by the equivalent time constant of the actuator and the prediction gain, and then summing the product with the penetration proximity.

8. The adaptive control method for drilling depth of a magnetic base drill according to claim 1, characterized in that, The predictive adaptive correction factor is synthesized based on the magnetic adsorption preload margin and the predicted penetration, including: multiplying the magnetic adsorption preload margin by a negative exponential function of the product of the attenuation sensitivity coefficient and the predicted penetration to obtain the predictive adaptive correction factor.

9. The adaptive control method for drilling depth of a magnetic base drill according to claim 1, characterized in that, Also includes: When the increasing trend and duration of the penetration proximity meet the conditions, the system sequentially enters the penetration warning state and the penetration confirmation state. In the penetration confirmation state, the system performs a pecking drill retraction and records the final drilling depth.

Images