Intelligent control double contact probe head numerical control machine tool on-machine measurement system

Abstract

This invention relates to the field of CNC machine tool measurement technology, specifically to an intelligent control dual-contact probe CNC machine tool in-machine measurement system, aiming to solve problems such as fixed measurement paths and lack of collaborative logic between the two probes. The system includes hardware sensing, wireless transmission, edge intelligent processing, a CNC system interface, and an adaptive closed-loop control center. Through hierarchical collaboration between the first and second probes, combined with FPGA coordinate latching and piezoelectric ceramic sensing, the system dynamically corrects the machining G-code using adaptive task scheduling and real-time error compensation algorithms. By adopting the above scheme, this invention achieves a deep closed loop of machining, measurement, and compensation, significantly improving the flexibility, efficiency, and accuracy of in-machine measurement under complex dynamic conditions.

Description

Technical Field

[0001] This invention relates to the field of CNC machine tool measurement technology, and in particular to an intelligent control dual-contact probe CNC machine tool in-machine measurement system. Background Technology

[0002] With the continuous development of intelligent manufacturing and precision machining technologies, the on-machine measurement system of CNC machine tools, as a key link in realizing integrated closed-loop control of machining and measurement, directly affects product quality and production costs in terms of accuracy, efficiency, and intelligence. Dual-contact probes, due to their high precision and good repeatability, are widely used for on-machine measurement of complex workpieces. However, existing dual-contact probe measurement systems still have significant shortcomings in intelligent control, measurement process automation, and system integration, making it difficult to meet the demands of modern high-end manufacturing for high efficiency, adaptability, and end-to-end closed-loop feedback.

[0003] A search revealed an in-machine measurement method for CNC machine tools based on dual-contact workpiece probes, disclosed in publication number CN113681352B. This method employs two contact probes with different diameter probe balls, used for workpiece datum positioning and feature measurement respectively, and updates system parameters through two measurements to improve machining consistency. However, this technical solution relies on manually preset measurement logic and fixed procedures, lacking the ability to make real-time intelligent judgments and dynamic adjustments to the measurement process. Furthermore, it lacks integrated intelligent control algorithms, failing to automatically optimize probe paths, contact speeds, or data processing strategies based on workpiece status or environmental changes. This results in low efficiency and susceptibility to human-induced errors in measuring complex curved surfaces or multi-feature workpieces.

[0004] A search revealed a method for measuring the stationary blade groove of a steam turbine cylinder, published under publication number CN117001424B. This method involves issuing commands from a remote control center, using a wireless probe to perform in-machine measurements of a specific structure (such as the stationary blade groove), and achieving remote monitoring. While a remote communication mechanism is introduced, the probe is a single wireless probe, lacking a dual-contact structure, and the measured object is highly specialized, resulting in poor versatility. More importantly, the system only implements a one-way control mode of "command-execution-feedback," lacking intelligent feedback and adaptive compensation mechanisms based on measurement results. It cannot dynamically correct machine tool parameters or optimize subsequent processing strategies during the measurement process, making it difficult to support the requirements of high-precision closed-loop manufacturing.

[0005] The aforementioned problems indicate that existing technologies have significant shortcomings in areas such as intelligent collaborative control of dual-contact probes, adaptive optimization of the measurement process, and deep integration with CNC systems. Therefore, this invention proposes an intelligent control system for on-machine measurement of CNC machine tools using dual-contact probes. This system aims to achieve autonomous decision-making, dynamic optimization, and high-precision adaptive compensation in the measurement process by integrating intelligent algorithms, dual-probe collaborative strategies, and closed-loop feedback mechanisms. This will improve the intelligence, versatility, and reliability of on-machine measurement, meeting the urgent needs of high-end manufacturing for efficient, accurate, and flexible online inspection. Summary of the Invention

[0006] This invention provides an intelligent control dual-contact probe CNC machine tool in-machine measurement system, which aims to solve the technical problems in existing in-machine measurement technology, such as fixed measurement paths, lack of perception of complex dynamic working conditions, lack of dual-probe collaborative scheduling logic, and inability to form real-time deep closed-loop feedback between measurement and machining processes.

[0007] To achieve the above-mentioned objectives, the technical solution adopted by the present invention is as follows:

[0008] An intelligent control dual-contact probe CNC machine tool in-machine measurement system, the system includes a hardware sensing module, a wireless signal transmission module, an edge intelligent processing module, a CNC system interface module, and an adaptive closed-loop control center.

[0009] The hardware sensing module includes a first contact probe and a second contact probe symmetrically arranged within the machine tool spindle probe magazine. The first contact probe is configured as a coarse and positioning probe, with a ruby ​​probe ball of 4mm to 6mm in diameter fixed to its probe end. The stiffness coefficient of the first contact probe is set between 0.5N / μm and 0.8N / μm, used for rapid coarse alignment and large-area topographic scanning of the workpiece's initial pose. The second contact probe is configured as a fine and feature probe, with a tungsten carbide or ruby ​​probe ball of 0.5mm to 2mm in diameter fixed to its probe end. The trigger sensitivity of the second contact probe is set between 0.01N and 0.03N, used for high-precision detection of deep holes, narrow grooves, and minute geometric features of the workpiece. Both the first and second contact probes integrate piezoelectric ceramic sensing units, with a sampling frequency of not less than 50kHz, used to capture microscopic pressure vector fluctuations at the moment of probe ball contact.

[0010] The wireless signal transmission module includes a signal transmitter fixed to the probe handle and a signal receiver mounted on the top of the inner wall of the machine tool. The signal transmitter uses frequency hopping spread spectrum technology to transmit data in the 2.4GHz to 2.4835GHz frequency band, with a frequency switching rate of 1600 times per second to suppress electromagnetic noise interference in the machine tool cutting environment. The signal transmitter and signal receiver use an authentication-based encryption protocol, and the signal transmission delay is controlled within 0.5ms to ensure the real-time nature of the trigger signal.

[0011] The edge intelligent processing module is communicatively connected to the signal receiver. Its internal circuit board integrates a central processing unit based on a multi-core architecture and a field-programmable gate array (FPGA) coprocessor. The FPGA coprocessor has a preset hard-core timer with a clock frequency of 200MHz, which is specifically used to latch the synchronous coordinate data of each axis of the machine tool at the moment the trigger signal is received. The latching accuracy of the coordinate data reaches 0.1μs. The central processing unit runs a deep neural network inference engine, which is loaded with a pre-trained probe accuracy attenuation prediction model.

[0012] The CNC system interface module achieves bidirectional data interaction with the machine tool CNC system through a bus protocol; the bus protocol adopts the industrial Ethernet protocol and supports the OPCUA or MTConnect standard; the CNC system interface module writes the compensation vector and measurement macro instructions processed by the edge intelligent processing module into the CNC system register in real time.

[0013] The adaptive closed-loop control center is the logical control core of this system, and it runs an adaptive task scheduling algorithm, a dynamic path planning algorithm, and a real-time error compensation algorithm.

[0014] The adaptive task scheduling algorithm automatically divides the measurement task into a global coarse measurement stage and a local fine measurement stage based on the computer-aided design model information of the workpiece to be measured. In the global coarse measurement stage, the scheduling algorithm activates the first contact probe and controls it to perform discrete sampling at a feed rate of 1000 mm / min to 2000 mm / min to construct a macroscopic pose model of the workpiece. In the local fine measurement stage, the scheduling algorithm automatically switches to the second contact probe based on the deviation between the macroscopic pose model and a preset tolerance threshold, and automatically adjusts the contact feed rate according to the feature dimensions. for:

[0015] ;

[0016] in, As the reference feed rate, As an environmental damping factor, The design tolerance for this feature point, Let be the local radius of curvature of the characteristic surface.

[0017] The dynamic path planning algorithm dynamically corrects the measurement path based on the real-time sensing of workpiece residual stress deformation and machine tool thermal elongation. The algorithm calculates the distance field gradient between the current measurement point and the known obstacle avoidance space boundary, and uses the potential field method to update the ingress vector of the next sampling point in real time, ensuring the safe obstacle avoidance of the probe during the measurement process.

[0018] The real-time error compensation algorithm is based on a six-degree-of-freedom spatial error model, and its calculation process is as follows:

[0019] First, temperature distribution fields of the spindle box, worktable, and bed are collected in real time using temperature sensor arrays distributed at key nodes of the machine tool structure. ;

[0020] Secondly, the initial deviation vector between the machine tool coordinate system and the workpiece coordinate system fed back by the first contact probe is combined with the initial deviation vector between the machine tool coordinate system and the workpiece coordinate system. Calculate the comprehensive error matrix :

[0021] ;

[0022] in, For the machine tool Jacobian matrix, Let be the geometric positioning error vector for each motion axis. This is a thermal error compensation function based on the temperature field. This is a compensation term for the inherent eccentricity and stress deformation of the probe system;

[0023] Finally, the adaptive closed-loop control center feeds back the generated correction instructions to the CNC system, dynamically modifying the offset values ​​in the machining G-code.

[0024] Furthermore, both the first and second contact probes are equipped with a self-cleaning mechanism; the self-cleaning mechanism is achieved through a compressed air nozzle and a high-pressure coolant nozzle installed on the side of the machine tool tool changer; before each probe switch or measurement task begins, the CNC system controls the spindle to move to the cleaning position, and the nozzle sprays dry compressed air at a pressure of 0.6MPa to 0.8MPa for 3 seconds to remove cutting fluid residue and small chips from the probe surface.

[0025] Furthermore, the hardware sensing module also includes an infrared tool setting device for in-machine online calibration of the probe length and ball radius of the first and second contact probes; the calibration accuracy is controlled within 0.001mm, and the calibration results are updated in real time to the storage unit of the edge intelligent processing module as a reference input for error compensation calculation.

[0026] Furthermore, the wireless signal transmission module has a real-time signal strength monitoring function; when the signal receiver detects that the received power level is lower than -85dBm or the message loss rate exceeds 0.5%, the edge intelligent processing module automatically triggers the machine tool pause command and controls the spindle to move the probe to the signal enhancement area, and continues the measurement task after the signal quality is restored.

[0027] As a preferred embodiment of the present invention, the edge intelligent processing module also integrates a probe collision early warning mechanism; by real-time monitoring of the non-contact high-frequency acoustic emission signal returned by the piezoelectric ceramic sensing unit, the short-time Fourier transform is used to identify the airflow whistling characteristics generated when the probe approaches the workpiece surface; when the characteristic frequency amplitude reaches the preset threshold and a sudden change occurs outside the safe distance from the model prediction point, the system immediately sends an emergency stop signal to the machine tool driver, with a response time of less than 10ms.

[0028] In a preferred embodiment of the present invention, the adaptive closed-loop control center of the system also includes a data mining-assisted decision-making subsystem. This subsystem stores the workpiece error data obtained from each measurement in a non-volatile database, and identifies the trend of machine tool accuracy evolution with ambient temperature and running time through multivariate linear regression analysis, thereby automatically recommending the optimal machine tool geometric accuracy calibration cycle.

[0029] In this invention, the specific workflow of the system is as follows:

[0030] The first step is that the CNC system receives the on-machine measurement start command, drives the spindle to grab the first contact probe from the tool magazine, executes the self-cleaning program, and then performs feature sampling of 5 to 10 points on the workpiece on the worktable according to the preset coarse measurement path.

[0031] The second step is for the edge intelligent processing module to calculate the three-dimensional translation and rotation deviations of the workpiece in the machine tool coordinate system and establish a rough measured model of the workpiece.

[0032] The third step involves the adaptive closed-loop control center identifying the key precision feature areas that need to be detected based on the comparison between the measured coarse model and the CAD design model, and generating the precision measurement path instructions for the second stage.

[0033] Fourth step: The spindle automatically switches to the second contact probe, performs self-cleaning, and then performs high-density sampling of the precision feature area according to the feed speed determined by the dynamic path planning algorithm.

[0034] The fifth step involves the edge intelligent processing module combining real-time temperature data to calculate a global compensation vector that includes thermal and geometric errors.

[0035] The sixth step involves injecting the compensation vector into the machine tool register through the CNC system interface module to complete the real-time correction of the subsequent cutting path, thereby achieving closed-loop control of machining-measurement-compensation.

[0036] Compared with the prior art, the present invention has the following significant advantages:

[0037] 1. This invention breaks through the limitations of fixed measurement paths in traditional methods by employing a hierarchical collaborative strategy with dual-contact probes and combining it with a task adaptive scheduling algorithm. The system can dynamically identify the actual pose and shape deviation of the workpiece based on the initial measurement results, and autonomously plan the precision measurement path, significantly improving the adaptability of on-machine measurement to complex, irregular, and stress-deformed workpieces.

[0038] 2. By combining the large-feed coarse measurement of the first contact probe with the fine detection of the second contact probe, this invention significantly reduces the overall measurement time while ensuring the accuracy of key feature detection. Simultaneously, the nanosecond-level coordinate latching technology based on FPGA and the high-frequency sensing of piezoelectric ceramics eliminate the hysteresis effect in traditional contact triggering, resulting in an order-of-magnitude improvement in the repeatability of the measurement system.

[0039] 3. This invention no longer treats measurement as a separate post-inspection process. Instead, through deep integration of the edge intelligent processing module with the machine tool's CNC system, it transforms comprehensive errors, including thermal errors, geometric errors, and stress deformation, into correction vectors for the machining path in real time. This "measure-and-complement" mode effectively offsets the temperature rise drift of the machine tool during long-term machining, ensuring the consistency of part processing and reducing the scrap rate.

[0040] 4. Employing frequency-hopping spread spectrum wireless transmission technology and a collision warning mechanism based on acoustic emission characteristics, the measurement system maintains extremely high communication reliability even in harsh machining environments with strong electromagnetic interference and multiple coolant splashes. Physical-level multi-sensor redundancy and algorithm-level collision prevention significantly reduce the risk of damage to expensive probe components, ensuring automated operation of CNC machine tools in unattended environments.

[0041] 5. By accumulating and deeply analyzing historical measurement data, the system can perceive the degradation trend of machine tool performance, provide data support for preventive maintenance, and improve the intelligent maintenance level of the entire manufacturing system.

[0042] At the hardware implementation level, the circuit design of the edge intelligent processing module is as follows: The core processor is an industrial-grade SoC chip, which integrates a quad-core Cortex-A53 processor and an FPGA architecture with 250K logic units. The FPGA is internally configured with a dedicated high-speed input capture unit, which receives pulse signals from the signal receiver via a high-speed optocoupler isolation circuit. When the first or second contact probe touches the workpiece and generates a level transition, this transition signal is converted into an LVDS signal via a differential transformer and enters the FPGA. At the instant the FPGA captures the edge trigger, it initiates an interrupt request to the CNC system via its internal AXI4 bus and simultaneously reads the coordinate feedback data stored in the high-speed buffer register.

[0043] At the software logic level, the adaptive closed-loop control center runs on a real-time operating system. The implementation logic of the task adaptive scheduling algorithm adopts a hierarchical state machine architecture. The first state is the "environment initialization state," where the system performs probe calibration and machine tool zero-point calibration; the second state is the "pose perception state," which drives the first contact probe to perform rapid multi-point sampling, with sampling points selected based on the vertices and geometric centers of the workpiece's maximum containment box; the third state is the "feature refinement state," which uses an octree algorithm to subdivide the CAD model, identify areas with drastic changes in normal vectors or curvature radii less than 5mm, and mark them as high-priority sampling areas; the fourth state is the "data fusion and compensation state," which generates a compensation surface by fitting the sampling results using the least squares method.

[0044] In terms of signal processing, the output signal of the piezoelectric ceramic sensing unit is preprocessed by a low-noise charge amplifier before entering a 16-bit analog-to-digital converter. The edge intelligent processing module uses a Kalman filter algorithm to reduce noise in the original sensing signal. The state equation of the filter is described as follows:

[0045] ;

[0046] in, for The filtered pressure estimate at time 10:00. For Kalman gain, The raw pressure observations are collected by the ADC. Using this algorithm, the system can effectively distinguish between momentary spurious triggering caused by coolant splashing and genuine workpiece contact signals.

[0047] In terms of wireless communication protocol encapsulation, each frame of measurement data message consists of a preamble, synchronization word, device ID, payload data, and a 16-bit cyclic redundancy check code. To cope with factory environments with multiple machine tools operating in parallel, the system supports a time-division multiple access-based channel allocation scheme to ensure that multiple measurement systems in the same space do not interfere with each other.

[0048] Furthermore, regarding the thermal error compensation, the compensation model established in this invention also considers the influence of spindle speed on temperature rise. The thermal characteristic transfer function of the machine tool is obtained through preliminary experiments at different speed gradients. In actual measurement, the adaptive closed-loop control center dynamically updates the compensation matrix by acquiring historical spindle speed data in real time and using a recursive least squares algorithm. The thermal hysteresis parameter in the data ensures that the compensation remains accurate even under non-equilibrium thermal conditions.

[0049] Furthermore, the CNC system interface module is equipped with a dedicated API protocol conversion layer, which can automatically convert complex compensation vectors into G10 (data setting) or G52 (local coordinate system setting) commands that the CNC system can recognize. During the machining cycle, the system triggers the measurement program through M-codes, and after the measurement is completed, the compensation value is directly applied to the tool compensation register through the system variable feedback mechanism, realizing a fully automated closed loop without manual intervention.

[0050] In terms of system safety, the second contact probe is equipped with a mechanical overload protection device. When the force on the probe exceeds a preset safety threshold, the magnetic or mechanical locking mechanism inside the probe holder will automatically disengage, simultaneously triggering a physical circuit breaker signal. This signal is transmitted through the highest priority channel of the wireless signal transmission module, forcing the machine tool feed driver into a zero-speed lock state.

[0051] In summary, the intelligent control dual-contact probe CNC machine tool in-machine measurement system described in this invention forms a complete, efficient, and highly reliable in-machine inspection technology system through hardware dual-head redundancy, high-frequency frequency hopping in communication, edge processing and depth compensation in algorithms, and a closed-loop logic throughout the entire process. It not only solves the problem of limited functionality of a single probe but also, through intelligent control strategies, enables the machine tool to possess self-learning and self-correcting capabilities in complex process environments, providing solid technical support for the digital transformation of precision manufacturing.

[0052] The system described in this invention has clearly defined connections between its components, a rigorous logical flow, and specific and definite technical means. There are no ambiguous or optional descriptions, enabling it to be implemented by those skilled in the art based on the detailed engineering description above. By fundamentally reconstructing the measurement logic at its core, this invention eliminates the uncertainties caused by human intervention, transforming on-machine measurement in CNC machine tools from "auxiliary quality inspection" to "core machining closed loop," demonstrating significant inventiveness and industrial application value. Attached Figure Description

[0053] Figure 1 This is a block diagram of the module connection architecture of the system described in this invention;

[0054] Figure 2This is a schematic diagram showing the installation layout of the hardware sensing module, machine tool spindle, and cleaning mechanism of the present invention.

[0055] Figure 3 This is a flowchart of the system described in this invention. Detailed Implementation

[0056] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0057] like Figure 1 As shown, the construction of an intelligent control dual-contact probe CNC machine tool in-machine measurement system embodies the deep integration of high-precision machining and real-time quality inspection. Through precise hardware configuration and complex software algorithms, it achieves closed-loop control of the CNC machine tool machining process. At the physical level, the system consists of a hardware sensing module, a wireless signal transmission module, an edge intelligent processing module, a CNC system interface module, and an adaptive closed-loop control center. These modules are logically connected through standardized industrial communication protocols and a high-bandwidth data bus, forming an organic whole with environmental perception, autonomous decision-making, and real-time compensation capabilities.

[0058] like Figure 2 As shown, the hardware sensing module constitutes the physical detection foundation of the system, comprising a first contact probe and a second contact probe symmetrically arranged within the machine tool spindle probe magazine. The first contact probe, serving as a coarse measurement and positioning unit, is designed with a focus on structural rigidity and stability over a wide scanning range. A ruby ​​probe ball with a diameter of 4mm to 6mm is fixed to the end of the probe's stylus; the use of ruby ​​material ensures wear resistance and chemical stability at the contact point. In terms of mechanical performance, the internal return spring and flexible support mechanism of the first contact probe are calibrated to a stiffness coefficient between 0.5N / μm and 0.8N / μm. This stiffness setting allows the probe to effectively resist airflow disturbances and the impact of residual cutting fluid when performing initial workpiece orientation alignment at feed rates of 1000mm / min to 2000mm / min, thereby constructing the macroscopic geometric topology of the workpiece in the machine tool coordinate system. Complementing this is the second contact probe, configured as a precision measurement and feature detection unit. Its probe tip is fixed with an extremely small diameter probe ball ranging from 0.5mm to 2mm, made of tungsten steel or ruby, to meet the detection requirements of deep holes, narrow grooves, and minute step features on workpieces. The triggering mechanism of the second contact probe has undergone micro-force balancing adjustments, with its triggering sensitivity strictly set within the range of 0.01N to 0.03N. This parameter setting greatly reduces the pre-stroke error during the detection process, ensuring the ability to capture precise geometric features.

[0059] Furthermore, to obtain physical information at the moment of contact at a microscopic level, both the first and second contact probes integrate piezoelectric ceramic sensing units within their internal housings. These piezoelectric ceramic sensing units employ a stacked structure and are connected to a high-impedance charge amplification circuit, with a sampling frequency set to no less than 50kHz. When the probe ball contacts the workpiece surface, the resulting microscopic pressure vector fluctuations are transmitted through the probe to the piezoelectric ceramic, generating a piezoelectric effect and outputting a millivolt-level voltage signal. After processing by the filtering circuit within the edge intelligent processing module, this signal can accurately lock onto the physical coordinates of the contact point.

[0060] In terms of signal transmission, the wireless signal transmission module employs a communication strategy with extremely strong anti-interference capabilities. This module includes a signal transmitter mounted on the probe shank and a signal receiver fixed to the top of the inner wall of the machine tool's machining area. Considering the strong electromagnetic noise generated by the motor driver and the multipath reflection caused by the metal body in the machine tool cutting environment, the signal transmitter uses frequency-hopping spread spectrum technology. Its operating frequency band is selected from 2.4GHz to 2.4835GHz, commonly used in industrial, scientific, and medical applications. Frequency hopping is performed using a preset pseudo-random sequence, with a frequency switching rate of up to 1600 times per second. This high-frequency frequency hopping mechanism ensures that even when some sub-channels are occupied, the measurement trigger signal can still be uploaded in real time through other clean channels. Simultaneously, an authentication-based encrypted handshake protocol is established between the signal transmitter and receiver to prevent signal aliasing when multiple machine tools are operating in parallel. The end-to-end signal transmission delay of the system is compressed to within 0.5ms, providing a time deterministic guarantee for coordinate latching under high-speed measurement.

[0061] The edge intelligent processing module serves as the data preprocessing hub for the entire system. Its core circuit board houses a multi-core central processing unit (CPU) and a field-programmable gate array (FPGA) coprocessor. The FPGA coprocessor, through parallel logic design, implements a hard-core timer with a clock frequency of 200MHz, synchronized with the position feedback encoder signals of each axis of the machine tool's CNC system. When the signal receiver captures a probe trigger pulse, the capture unit within the FPGA latches the current coordinate register value with an accuracy of 0.1μs. Furthermore, the CPU of the edge intelligent processing module runs a deep neural network inference engine, which loads a pre-trained probe accuracy attenuation prediction model. This model is based on a long short-term memory (LSTM) network structure. Input parameters include the number of historical measurement triggers, changes in ambient temperature and humidity, residual probe battery voltage, and the contact impact intensity induced by the piezoelectric ceramic. The output is the current system deviation prediction value of the probe, used for real-time dynamic correction of the measurement results.

[0062] The CNC system interface module acts as a translator between the edge intelligent processing module and the machine tool CNC system. Its hardware layer supports the Industrial Ethernet protocol, and its physical interface conforms to the OPCUA or MTConnect standards, ensuring cross-platform data interoperability. Through this interface, error compensation vectors and measurement macro instructions generated by the adaptive closed-loop control center can be written into the CNC system's system variables and registers at sub-millisecond intervals.

[0063] As the logical hub of the system, the adaptive closed-loop control center integrates an adaptive task scheduling algorithm, a dynamic path planning algorithm, and a real-time error compensation algorithm. The adaptive task scheduling algorithm enables intelligent hierarchical classification of measurement tasks. After acquiring the computer-aided design model of the workpiece to be measured, the algorithm automatically calculates the geometric weights of the features and divides it into a global coarse measurement stage and a local fine measurement stage. In the first stage, the scheduling algorithm activates the first contact probe and performs rapid scatter sampling. Based on the sampled data, the system uses the least squares method to fit the measured pose model of the workpiece. Subsequently, the algorithm calculates the residual deformation or machining allowance distribution based on the deviation between the measured model and the theoretical CAD model, and autonomously generates the fine measurement path for the second stage accordingly. In the fine measurement stage, the second contact probe is activated; its contact feed speed is not a constant value but is calculated in real time according to a formula:

[0064] ;

[0065] Specifically, Set as the baseline feed rate, with a typical value of 200 mm / min; This is the environmental damping factor, which is determined based on the machine tool lubrication condition and the level of environmental vibration. This represents the design tolerance for that measurement point; the tighter the tolerance, the slower the speed. This is based on the local radius of curvature extracted from the CAD model. This dynamic adjustment scheme for the feed rate balances measurement efficiency and point accuracy.

[0066] The dynamic path planning algorithm focuses on safety and path optimization during the measurement process. The algorithm acquires real-time data from temperature sensors mounted on the machine tool's structural components, calculates the thermal elongation of the machine tool, and combines this with the workpiece's deformation trend to correct the measurement path in real time. The path planning employs obstacle avoidance logic based on the potential field method, treating the probe as a point mass, the workpiece surface and fixture as repulsive force sources, and the target sampling point as an attractive force source. By calculating the gradient field direction and updating the probe's entry vector in real time, it avoids interference and collisions that may occur in complex internal cavity structures.

[0067] The real-time error compensation algorithm is based on a six-degree-of-freedom spatial error model to dynamically compensate for the machining process. The calculation process first involves acquiring the temperature field using platinum resistance temperature sensors distributed across the spindle box, worktable, and bed. Subsequently, the initial deviation vector fed back by the first contact probe was analyzed. Calculate the comprehensive error matrix :

[0068] ;

[0069] In the formula, The Jacobian matrix of the machine tool is obtained through kinematic modeling of the mechanism; The geometric positioning error of each motion axis in the current coordinate system is usually measured in advance by a laser interferometer and stored in a lookup table; As the thermal error compensation function, the mapping relationship between the temperature field and the thermal deformation of the machine tool is established using multivariate regression analysis. This covers the probe's ball head eccentricity error and bending deformation under stress. The adaptive closed-loop control center will use the obtained... It is converted into compensation instructions in real time, and the tool compensation amount or local coordinate system offset value in the machining G code is modified through the CNC system interface module.

[0070] Furthermore, to ensure the reliability of the measurement system in a cutting fluid splashing environment, both the first and second contact probes are equipped with a self-cleaning mechanism. This mechanism includes a compressed air nozzle and a high-pressure coolant spray nozzle mounted on the side of the machine tool's tool changer. Before executing a measurement command, the spindle carries the probe to a specific cleaning position. The nozzle sprays dry compressed air at a controlled pressure of 0.6 MPa to 0.8 MPa for 3 seconds, using the physical impact of the high-speed airflow to thoroughly remove the cutting fluid and adsorbed micro-chips from the probe surface.

[0071] Furthermore, the hardware sensing module is equipped with an infrared calibration device. This device uses a high-precision infrared beam to capture the edge of the probe, enabling on-the-fly online calibration. The calibration process dynamically updates the effective length of the probe and the radius of the probe ball, maintaining a calibration accuracy within 0.001 mm. The calibrated parameters are pushed in real time to the static random access memory of the edge intelligent processing module, serving as the basic zero point for all subsequent measurement calculations.

[0072] Furthermore, the wireless signal transmission module incorporates a signal quality monitoring mechanism. The edge intelligent processing module reads the RSSI value returned by the signal receiver. If it detects that the level is below -85dBm, or that the data packet loss rate exceeds 0.5% due to the large-area metal shielding of the machine tool, the system will immediately trigger protection logic, suspend the current measurement sequence, and issue a command to move the spindle to an enhanced area with better signal coverage, thereby avoiding measurement logic deadlock caused by communication interruption.

[0073] In a preferred embodiment of the present invention, the edge intelligent processing module embeds a collision warning mechanism based on acoustic emission technology. This mechanism, by performing a short-time Fourier transform on the high-frequency raw signal returned by the piezoelectric ceramic sensing unit, can identify the characteristic airflow whistling frequency generated by air compression when the probe approaches the workpiece surface at extremely close range. When the amplitude of this characteristic frequency undergoes an abnormal abrupt change, and the system determines that this abrupt change occurs outside the preset theoretical contact point safety range, the edge intelligent processing module bypasses the conventional logic of the CNC system and directly sends a highest-priority hardware emergency stop signal to the machine tool driver. The delay of the entire response chain is less than 10ms, effectively protecting the probe body from damage due to misoperation.

[0074] In a preferred embodiment of the present invention, the adaptive closed-loop control center also includes a data mining-assisted decision-making subsystem. This subsystem utilizes the non-volatile storage space of the edge processor to construct a local database, recording the historical measurement residuals of the machine tool under different operating loads and ambient temperatures. Through multivariate linear regression analysis, the subsystem can plot the drift curve of the machine tool's geometric accuracy. When it is predicted that the mechanical backlash wear of the machine tool due to long-term operation has exceeded the automatic compensation threshold, the system automatically generates maintenance suggestions, reminding the operator to perform physical-level machine tool accuracy calibration.

[0075] like Figure 3 As shown, the operational process in actual engineering applications is as follows: First, after receiving the measurement macro program trigger command, the machine tool CNC system instructs the tool-changing robot to retrieve the first contact probe from the tool magazine. After loading, the probe first undergoes surface cleaning at the self-cleaning station, and then, according to the pre-generated coarse measurement path, it rapidly samples 5 to 10 feature points on the workpiece's reference surface and main contour points. Upon receiving these trigger coordinates, the edge intelligent processing module immediately calculates the workpiece's translation vector and rotation Euler angles, and constructs a measured coarse model of the workpiece in memory. The adaptive closed-loop control center locates severely deformed areas or key precision feature positions by performing Boolean operations and comparisons between this measured coarse model and the CAD standard model. Next, the system controls the spindle to return the first contact probe and exchange it for the second contact probe. After cleaning and online calibration, the second contact probe is used according to the formula... The calculated optimized feed rate is used for high-density probing of precision areas. After the sampled data is aggregated, it is combined with real-time temperature sensor array feedback and analyzed using a formula. The global compensation vector is calculated. Finally, this vector is written to the CNC register via the bus, and the machine tool will automatically apply these offset values ​​in the subsequent cutting cycle, thereby achieving in-machine closed-loop machining.

[0076] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An intelligent control dual contact probe head CNC machine tool on-machine measurement system, characterized in that, The system includes a hardware sensing module, a wireless signal transmission module, an edge intelligent processing module, a CNC system interface module, and an adaptive closed-loop control center. The hardware sensing module includes a first contact probe and a second contact probe symmetrically arranged in the probe magazine of the machine tool spindle. Both the first contact probe and the second contact probe have integrated piezoelectric ceramic sensing units inside, which are used to capture the micro pressure vector fluctuations at the moment the probe ball contacts the workpiece. The wireless signal transmission module includes a signal transmitter fixed to the probe handle and a signal receiver installed on the top of the inner wall of the machine tool. The signal transmitter and the signal receiver use frequency hopping spread spectrum technology to wirelessly transmit measurement trigger signals and sensor data. The edge intelligent processing module is communicatively connected to the signal receiver. It integrates a central processing unit and a field-programmable gate array (FPGA) coprocessor. The FPGA coprocessor is equipped with a hard-core timer, which is used to latch the synchronous coordinate data of each motion axis of the machine tool the instant the trigger signal is received. The CNC system interface module achieves bidirectional data interaction with the machine tool CNC system through the industrial Ethernet bus protocol, and is used to write compensation vectors and measurement macro instructions into the CNC system registers in real time. The adaptive closed-loop control center is logically connected to the edge intelligent processing module and the CNC system interface module, respectively. It runs a task adaptive scheduling algorithm, a dynamic path planning algorithm and a real-time error compensation algorithm. The task adaptive scheduling algorithm controls the first contact probe to perform a global coarse measurement stage and controls the second contact probe to perform a local fine measurement stage based on the computer-aided design CAD model information of the workpiece to be measured. It also generates correction instructions through the real-time error compensation algorithm to dynamically modify the machining code of the machine tool.

2. The intelligent control dual-contact probe CNC machine tool in-machine measurement system according to claim 1, characterized in that: The first contact probe is configured as a coarse measurement and positioning probe. A ruby ​​probe ball with a diameter of 4mm to 6mm is fixed to the end of its probe. The first contact probe has a flexible support mechanism inside, and its stiffness coefficient is set between 0.5N / μm and 0.8N / μm. It is used for rapid coarse alignment of the initial position of the workpiece and large-area topographic scanning. The second contact probe is configured as a precision and feature probe, with a tungsten carbide or ruby ​​probe ball with a diameter of 0.5 mm to 2 mm fixed to the end of its probe. The trigger sensitivity of the second contact probe is set to 0.01 N to 0.03 N, and it is used to detect deep holes, narrow grooves and small geometric features of the workpiece. The piezoelectric ceramic sensing unit adopts a stacked structure and is connected to a charge amplification circuit. Its sampling frequency is not less than 50kHz, which is used to transmit millivolt-level pressure fluctuation signals to the edge intelligent processing module.

3. The intelligent control dual-contact probe CNC machine tool in-machine measurement system according to claim 1, characterized in that: The wireless signal transmission module's signal transmitter uses frequency hopping spread spectrum technology to transmit data in the 2.4GHz to 2.4835GHz frequency band, with its frequency switching rate set to 1600 times per second to suppress electromagnetic noise interference in the machine tool cutting environment. The signal transmitter and the signal receiver use an authentication-based encryption protocol, and the signal transmission delay is controlled within 0.5ms. The wireless signal transmission module also has a real-time signal strength monitoring function. When the edge intelligent processing module detects that the received power level of the signal receiver is lower than -85dBm or the message loss rate exceeds 0.5%, it automatically triggers the machine tool pause command and controls the machine tool spindle to move the probe to the signal enhancement area. Measurement will continue after the signal quality is restored.

4. The intelligent control dual-contact probe CNC machine tool in-machine measurement system according to claim 1, characterized in that: The edge intelligent processing module's FPGA coprocessor is pre-loaded with a hard-core timer with a clock frequency of 200MHz, and the latching accuracy of the coordinate data reaches 0.1μs; The central processing unit runs a deep neural network inference engine, which is loaded with a pre-trained probe accuracy attenuation prediction model. This model is based on a long short-term memory network structure and predicts the current system deviation value of the probe based on the number of historical measurement triggers, ambient temperature and humidity, battery voltage and contact impact intensity. The edge intelligent processing module also employs a Kalman filter algorithm to reduce noise in the original sensing signal returned by the piezoelectric ceramic sensing unit. The state equation of the filter is as follows: ； wherein, is the filtered pressure estimate at time k, is the Kalman gain, is the raw pressure observation, used to distinguish between false triggers due to coolant splashes and real workpiece contact signals.

5. The intelligent control dual-contact probe CNC machine tool in-machine measurement system according to claim 1, characterized in that: In the global coarse measurement stage, the task adaptive scheduling algorithm controls the first contact probe to perform discrete sampling at a feed rate of 1000 mm / min to 2000 mm / min to construct a macroscopic pose model of the workpiece. During the local precision measurement phase, the scheduling algorithm automatically switches to the second contact probe based on the deviation between the macroscopic pose model and the preset tolerance threshold, and automatically adjusts the touch feed speed according to the feature size. The feed rate Determined by the formula: ； in, As the reference feed rate, As an environmental damping factor, The design tolerance for this feature point, Let be the local radius of curvature of the characteristic surface.

6. The intelligent control dual-contact probe CNC machine tool in-machine measurement system according to claim 1, characterized in that: The dynamic path planning algorithm corrects the measurement path based on the real-time sensing of workpiece residual stress deformation and machine tool thermal elongation. It updates the cutting vector of the next sampling point in real time by calculating the distance field gradient between the current measuring point and the known obstacle avoidance space boundary and using the potential field method. The real-time error compensation algorithm is based on a six-degree-of-freedom spatial error model. First, it collects the temperature distribution field of the spindle box, worktable, and bed through a group of temperature sensors distributed on the machine tool structural nodes. Secondly, combined with the initial deviation vector Calculate the comprehensive error matrix : ； in, For the machine tool Jacobian matrix, Let be the geometric positioning error vector for each motion axis. This is a thermal error compensation function based on the temperature field. This is a compensation item for the eccentricity and stress deformation of the probe system; The thermal error compensation function By acquiring historical data of machine tool spindle speed in real time, the thermal hysteresis parameters are dynamically updated using a recursive least squares algorithm.

7. The intelligent control dual-contact probe CNC machine tool in-machine measurement system according to claim 1, characterized in that: Both the first contact probe and the second contact probe are equipped with a self-cleaning mechanism, which includes a compressed air nozzle and a high-pressure coolant nozzle installed on the side of the machine tool tool changer. Before each probe switching, the machine tool spindle moves to the cleaning position, and the compressed air nozzle sprays dry compressed air at a pressure of 0.6MPa to 0.8MPa for 3 seconds to remove cutting fluid residue from the probe surface. The hardware sensing module also includes an infrared tool setting device for performing on-machine online calibration of the probe length and ball radius of the first and second contact probes. The calibration accuracy is controlled within 0.001 mm, and the calibration results are updated in real time to the storage unit of the edge intelligent processing module as a compensation calculation benchmark.

8. The intelligent control dual-contact probe CNC machine tool in-machine measurement system according to claim 1, characterized in that: The edge intelligent processing module integrates a probe collision warning mechanism. By monitoring the high-frequency raw signal returned by the piezoelectric ceramic sensing unit in real time, it uses short-time Fourier transform to identify the characteristic frequency of the airflow whistling generated when the probe approaches the workpiece surface. When the characteristic frequency amplitude reaches the preset threshold and a sudden change occurs outside the safe distance from the model prediction point, the edge intelligent processing module sends an emergency stop signal to the machine tool driver, with a response time of less than 10ms. The second contact probe is also equipped with a mechanical overload protection device. When the force on the probe exceeds the preset safety threshold, the probe holder automatically disengages and triggers a physical circuit breaker signal, forcing the machine tool feed driver to enter a zero-speed lock state.

9. The intelligent control dual-contact probe CNC machine tool in-machine measurement system according to claim 1, characterized in that: The adaptive closed-loop control center also includes a data mining-assisted decision-making subsystem. The subsystem stores the workpiece error data obtained from each measurement in a non-volatile database, identifies the trend of machine tool accuracy evolution with ambient temperature and running time through multivariate linear regression analysis, and automatically recommends the machine tool geometric accuracy calibration cycle. The CNC system interface module is equipped with an API protocol conversion layer, which converts the generated compensation vector into G10 data setting instructions or G52 local coordinate system setting instructions that the CNC system can recognize. By modifying the tool compensation register of the CNC system, the subsequent cutting path can be corrected in real time.

Images