Search

CN-121995851-A - Time grating position precision compensation device and method for numerical control turntable

CN121995851ACN 121995851 ACN121995851 ACN 121995851ACN-121995851-A

Abstract

The application discloses a time grating position precision compensation device and a time grating position precision compensation method for a numerical control turntable, which aim to solve the problems of control lag and insufficient positioning precision caused by time-space dyssynchrony, time dyssynchrony and data incompatibility when an absolute time grating displacement sensor is applied to an incremental grating numerical control system; the application takes an absolute time grating based on space-time conversion theory as a position detection core, builds a BP neural network position prediction model through multi-source data synchronous acquisition and preprocessing, takes the current and historical absolute angle values of the time grating as core input, is assisted with turntable working condition data, predicts and outputs the continuous angle value of the next interpolation period, converts absolute discrete measurement data into an incremental continuous pulse signal, directly adapts to the existing incremental numerical control system, and simultaneously is provided with an online compensation calibration unit to realize long-term operation precision self-compensation.

Inventors

  • ZHAO XIANGYU

Assignees

  • 江苏古田自动化股份有限公司

Dates

Publication Date
20260508
Application Date
20260408

Claims (6)

  1. 1. The time grating position precision compensation device for the numerical control turntable is characterized by comprising an absolute time grating displacement sensor, a multi-source data acquisition unit, a data preprocessing unit, a BP neural network position prediction unit, an incremental pulse conversion unit, a numerical control system closed-loop control unit and an online compensation calibration unit, wherein the absolute time grating displacement sensor, the multi-source data acquisition unit, the data preprocessing unit, the BP neural network position prediction unit, the incremental pulse conversion unit and the online compensation calibration unit are arranged on a numerical control turntable body; The absolute time grating displacement sensor is designed based on a time grating displacement measurement principle of a space-time conversion theory, is a round time grating coaxially arranged between a numerical control turntable workbench and a turntable bearing, is coaxially arranged with the rotation center of the numerical control turntable, has an angular resolution of less than or equal to 0.0001 DEG and an absolute position measurement precision of less than or equal to +/-0.5 DEG s, has a sampling frequency of more than or equal to 10kHz, and is used for converting airspace angular displacement information of the numerical control turntable into time domain electric signals to be output by a time-space coordinate conversion mechanism, and is used for acquiring absolute angle position data of the numerical control turntable workbench and electrically connected with a multi-source data acquisition unit; The multi-source data acquisition unit is internally provided with a space-time synchronization module, and is used for synchronizing the angle data of the absolute time grating displacement sensor, the rotating speed data of the rotating speed encoder, the temperature data of the distributed PT100 temperature sensor and the vibration data of the triaxial MEMS vibration sensor to the same time axis, wherein the synchronization error of the multi-source data is less than or equal to 1ms, and the multi-source data is electrically connected with the data preprocessing unit; The data preprocessing unit comprises an adaptive Kalman filtering module, a space-time alignment module and a data normalization module, wherein the adaptive Kalman filtering module is used for eliminating random noise and cutting impact abrupt change interference in measured data, the space-time alignment module is used for aligning a grid sampling period with a time sequence of an interpolation period of a numerical control system during matching, and the data normalization module maps multi-source data to a [0,1] interval, adapts to the input requirement of a BP neural network, outputs a standardized time sequence data set and is electrically connected with the BP neural network position prediction unit; The BP neural network position prediction unit is internally provided with a trained BP neural network prediction model, the BP neural network prediction model is a three-layer or more feedforward neural network and comprises an input layer, a hidden layer and an output layer, 6-10 neurons are arranged on the input layer, input parameters comprise time grid absolute angle values of current and previous 4 measurement periods, real-time rotating speed, average temperature and vibration peak values of a numerical control rotary table, 2 layers are arranged on the hidden layer, 16 neurons and 8 neurons are respectively arranged on the hidden layer, a ReLU function is adopted as an activation function, 1 neuron is arranged on the output layer, a continuous angle prediction value of the next interpolation period is output, the model is trained by adopting an L-M optimization algorithm, training samples cover actual measurement data of the full rotating speed, full load and full temperature conditions of the numerical control rotary table, the absolute angle values of the current and previous N measurement periods in a time sequence data set are used as core input, the continuous angle prediction value of the next interpolation period of the numerical control system is predicted and output by the aid of rotating speed, temperature and vibration working condition data; The incremental pulse conversion unit is internally provided with a pulse frequency division and quadrature coding module, converts a continuous angle predicted value into an AB phase quadrature pulse signal and a zero Z phase pulse signal which are compatible with an incremental grating interface, the pulse equivalent is completely compatible with the incremental grating, the highest pulse output frequency is more than or equal to 1MHz, and the continuous angle predicted value is directly connected to a position feedback interface of the existing incremental grating numerical control system and is electrically connected with a closed-loop control unit of the numerical control system; The closed-loop control unit of the numerical control system is a three-loop nested control system adapting to the cam type numerical control turntable, and comprises a position loop, a speed loop and a torque loop, wherein the position loop receives an orthogonal pulse signal output by an incremental pulse conversion unit as position feedback, the control bandwidth of the position loop is more than or equal to 200Hz, the dynamic tracking error is less than or equal to 0.3 angular seconds, the received orthogonal pulse signal is compared with a processing target instruction to generate a driving control instruction, and a cam driving mechanism of the numerical control turntable is driven to operate to form full-closed loop position control; The on-line compensation calibration unit takes a high-precision laser interferometer or an absolute grating ruler arranged on the numerical control turntable as a reference measurement element, acquires the deviation between a reference angle value and a BP neural network prediction angle value in real time, triggers on-line updating of a model when the deviation is more than or equal to 0.3 corner seconds, adjusts the weight and bias parameters of the neural network in an incremental learning mode, and performs long-time running precision self-compensation.
  2. 2. The time grating position accuracy compensation device for the numerical control turntable according to claim 1, wherein the absolute time grating displacement sensor replaces an original incremental grating ruler as a position detection element of the numerical control turntable.
  3. 3. The time grating position precision compensation device for the numerical control turntable according to claim 1, wherein the numerical control turntable body is a four-axis numerical control turntable, a five-axis numerical control turntable or an APC horizontal cam exchange table with no clearance transmission of cam rollers, the positioning error of the numerical control turntable after precision compensation is less than or equal to +/-0.5 angular seconds, and the repeated positioning precision is less than or equal to +/-2 angular seconds.
  4. 4. The time grating position accuracy compensation device for the numerical control turntable according to claim 1, wherein the training samples of the BP neural network prediction model are 10000 sets of actual measurement data covering the working conditions of full rotating speed, full load and full temperature of the numerical control turntable, and the predicted mean square error after training is less than or equal to 0.1 angular second.
  5. 5. The time grating position precision compensation device for the numerical control turntable according to claim 1, wherein the AB phase orthogonal pulse signal and the zero Z phase pulse signal output by the incremental pulse conversion unit are completely consistent with the pulse format of the original incremental grating ruler.
  6. 6. A time grating position accuracy compensation method for a numerical control turntable, characterized by being realized based on the time grating position accuracy compensation device for a numerical control turntable according to any one of claims 1-5, comprising the following steps: S1, synchronously acquiring data, namely acquiring absolute angle position data of a numerical control turntable through an absolute time grating displacement sensor coaxially arranged on the numerical control turntable, synchronously acquiring rotating speed, temperature and vibration working condition data of the turntable, and aligning multi-source data to the same time axis through a space-time synchronization module, wherein the synchronization error of the multi-source data is less than or equal to 1ms; S2, preprocessing data, namely eliminating random noise and cutting impact abrupt change interference of the acquired multi-source data through self-adaptive Kalman filtering, completing time sequence alignment of a grid sampling period and an interpolation period of a numerical control system through a time-space alignment module, mapping the multi-source data to a [0,1] interval through a data normalization module, and generating a standardized time sequence data set; S3, position prediction modeling, namely inputting a time grating absolute angle value, a numerical control turntable real-time rotating speed, an average temperature and a vibration peak value of the current and the previous 4 measurement periods in a time sequence data set into a pre-trained BP neural network prediction model to predict and output a continuous angle prediction value of the interpolation period of the next numerical control system, wherein the BP neural network prediction model is a three-layer or more feedforward neural network, training is carried out by adopting an L-M optimization algorithm, and training samples cover actual measurement data of the full rotating speed, the full load and the full temperature working condition of the numerical control turntable; S4, incremental pulse conversion, namely converting the continuous angle predicted value into an AB phase orthogonal pulse signal and a zero Z phase pulse signal which are compatible with an incremental grating interface through an incremental pulse conversion unit, wherein the pulse equivalent is completely compatible with the incremental grating, and directly inputting the pulse equivalent into a closed-loop control unit of the numerical control system; And S5, closed loop control and online compensation, wherein the numerical control system generates a driving control instruction to drive the numerical control turntable to operate based on the received orthogonal pulse signal and the processing target instruction, and simultaneously acquires the deviation of the reference angle value and the predicted angle value in real time through an online compensation calibration unit, and when the deviation is more than or equal to 0.3 angle seconds, the weight and the bias parameter of the BP neural network model are updated online in an incremental learning mode to perform long-time running precision self-compensation.

Description

Time grating position precision compensation device and method for numerical control turntable Technical Field The invention relates to the technical field of precision measurement and control of numerical control equipment, in particular to a time grating position precision compensation device and method for a numerical control turntable, which are particularly suitable for a cam roller gapless transmission four-axis/five-axis numerical control turntable and an APC horizontal cam exchange platform developed by Jiangsu old field automation Co., ltd. Background The position detection element of the numerical control turntable is a core component for determining the positioning precision and control performance of the numerical control turntable, and the high-end numerical control turntable at present commonly adopts an imported incremental grating ruler as a position feedback element, but the numerical control turntable has the problems of high cost, weak anti-interference capability and severe requirements on the installation environment, and is used for restricting the development of the domestic high-end numerical control turntable. The time grating displacement sensor is a novel displacement measuring element independently researched and developed in China, based on a space-time conversion theory, space domain displacement information is converted into time domain electric signals through a space-time coordinate conversion technology, absolute position measurement is realized, and the time grating displacement sensor has the outstanding advantages of high precision, strong anti-interference capability, low manufacturing cost, good environmental adaptability and the like, and is a core technical scheme for the localization substitution of grating scales. However, in practical engineering application, the following core technical bottlenecks still exist in the large-scale application of the time grating sensor in the numerical control turntable: The prior main stream numerical control system is a control framework adapting to an incremental grating ruler, continuous orthogonal pulse signals are adopted as position feedback, real-time closed loop control is realized based on interpolation periods, absolute position data of discrete periods are output by an absolute time grating sensor, the signal output modes and data sequential logic of the two are essentially different, and direct access can generate the problems of time-space asynchronism (mismatch between airspace discrete measurement and time domain continuous control) and real-time asynchronism (asynchronism between a time grating sampling period and the interpolation period of the numerical control system), so that control lag and dynamic tracking errors are large, even system oscillation occurs, and the high-speed precision machining requirement cannot be met. In the prior art, the time grating discrete data are converted into continuous data by adopting a simple linear interpolation and curve fitting mode, the low-speed and steady-state working conditions can be only adapted, the fitting error is rapidly increased under the dynamic working conditions of high-speed start-stop, variable load cutting and the like of the numerical control turntable, the dynamic control requirement of high-speed precise transmission of the cam type turntable cannot be met, meanwhile, the coupling compensation for working condition factors such as temperature, vibration and rotating speed is lacked, and the time grating measuring precision is easy to drift due to environmental influence during long-time continuous operation, and the compensation effect is continuously attenuated. Furthermore, the prior art does not form an integrated time grating application scheme adapting to a domestic numerical control turntable, especially the structural characteristics and control requirements of a cam roller gapless transmission turntable of ancient field company are met, and a special precision compensation and system compatibility scheme is lacked, so that the performance of a time grating sensor cannot be fully exerted, and direct substitution of an imported grating ruler is difficult to realize. The prior art (CN 120734769A) discloses an encircling type numerical control turntable intelligent brake system based on AI predictive control, which only optimizes a brake control scene of a numerical control turntable, does not relate to the adaptation problem of an absolute time grating displacement sensor and an incremental numerical control system, does not provide a related solution for time grating position precision compensation, and cannot solve the core pain point of large-scale application of a time grating on a high-end numerical control turntable. Aiming at the problems, the invention provides a time grating precision compensation scheme based on BP neural network prediction measurement aiming at the application scene of the cam type numerical control turntable, solves the