Component lifting assembly and run-out measurement device for central through hole

Abstract

The invention provides a central through hole part lifting assembly and run-out measurement device, and belongs to the technical field of industrial manufacturing. A base of the device is used for supporting the whole device and ensuring the levelness, a stand column is vertically fixed on the base and provides vertical guidance, and a lifting sliding table is installed on the stand column and realizes vertical lifting motion; the assembling workbench is installed on the lifting sliding table and used for containing the center through hole component, the measuring mechanism is located above the assembling workbench and used for measuring the run-out value of the center through hole component, and the control system is used for controlling operation of the whole device. The error reduction module eliminates various error source influences through temperature compensation, angle compensation, centrifugal force compensation, environmental vibration filtering, sensor nonlinear correction, multiple measurement data fusion and a trend compensation algorithm, and improves the center through hole part run-out measurement precision. According to the invention, the technical problem of insufficient run-out measurement precision of the central through hole component in the prior art can be solved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of industrial manufacturing, and in particular, relates to a center through-hole component lifting assembly and run-out measurement device. BACKGROUND

[0002] Center through-hole components have wide application in the field of mechanical manufacturing, and the assembly precision and coaxiality thereof directly affect the product performance and service life. The traditional center through-hole component assembly process is usually completed by manual operation, and lacks precise lifting control mechanism in the assembly process, and the run-out measurement relies on independent measurement devices, such as a dial gauge or an electronic dial gauge.

[0003] At present, the assembly and measurement devices existing in the market are mostly two independent systems, and the workpiece needs to be transferred to a special measurement device for run-out measurement after assembly. In the process of conversion of the process flow, the positioning state of the workpiece changes, resulting in deviation of the measurement result from the actual assembly state. At the same time, the traditional measurement device lacks effective compensation mechanism for factors such as environmental vibration, temperature change, centrifugal force, and the measurement data is affected by various error sources. That is, there is a technical problem of insufficient run-out measurement precision of the center through-hole component in the prior art. SUMMARY

[0004] Therefore, the present application provides a center through-hole component lifting assembly and run-out measurement device, which can solve the technical problem of insufficient run-out measurement precision of the center through-hole component in the prior art.

[0005] The present application is implemented as follows:

[0006] The present application provides a center through-hole component lifting assembly and run-out measurement device, which includes a base, a stand, a lifting slide, an assembly workbench, a measurement mechanism, a control system, and a data acquisition system. The base is used to support the entire device and ensure the levelness, the stand is vertically fixed on the base and provides vertical guidance, the lifting slide is installed on the stand and realizes vertical lifting movement, the assembly workbench is installed on the lifting slide and is used to place the center through-hole component, the measurement mechanism is located above the assembly workbench and is used to measure the run-out value of the center through-hole component, the control system is used to control the operation of the entire device, and the data acquisition system is used to acquire and process the run-out measurement data. The data acquisition system includes an error reduction module, which eliminates the influence of various error sources through temperature compensation, angle compensation, centrifugal force compensation, environmental vibration filtering, sensor nonlinearity correction, multiple measurement data fusion, and trend compensation algorithm, thereby improving the run-out measurement precision.

[0007] On the basis of the above technical solution, the center through-hole component lifting assembly and run-out measurement device of the present application can be further improved as follows:

[0008] The lifting slide is in sliding fit with the column guide rail, a ball screw pair is internally installed in the lifting slide, the ball screw pair screw rod is in sliding fit with the lifting slide, the ball screw pair nut is fixedly connected with the lifting slide, the ball screw pair screw rod lower end is supported and connected with the base through a bearing, the screw rod upper end is connected with the servo motor through a shaft coupling, and the servo motor is fixedly installed on the lifting slide top.

[0009] Further, the assembly workbench is installed on the lifting slide, the assembly workbench surface is made of hard alloy material, the workbench surface is provided with a positioning pin and a clamping device, the positioning pin is fixedly arranged on the assembly workbench surface vertically, and the clamping device comprises a cylinder and a clamping jaw, the cylinder is fixed on the assembly workbench, and the clamping jaw is connected with the cylinder piston rod.

[0010] Further, the measuring mechanism comprises a displacement sensor and a measuring support, the displacement sensor is fixed on the measuring support, the measuring support is connected with the column through an adjusting mechanism, the adjusting mechanism comprises a horizontal adjusting slide block and a vertical adjusting slide block, the horizontal adjusting slide block is in sliding fit with the column, the vertical adjusting slide block is in sliding fit with the horizontal adjusting slide block, and the measuring support is fixed on the vertical adjusting slide block.

[0011] Further, the control system comprises a central processing unit, a man-machine interface and an input and output module, the central processing unit is connected with the man-machine interface through a data bus, and the central processing unit is electrically connected with the servo motor, the cylinder and the displacement sensor through the input and output module.

[0012] Further, the data acquisition system comprises a signal conditioning circuit, an analog-to-digital converter, a digital filter, an error reduction module, a spectrum analyzer, a parameter optimizer and a physical model calculation unit, the signal conditioning circuit is electrically connected with the displacement sensor, the analog-to-digital converter is electrically connected with the signal conditioning circuit, the digital filter is electrically connected with the analog-to-digital converter, the spectrum analyzer is electrically connected with the digital filter, the parameter optimizer is electrically connected with the spectrum analyzer, the error reduction module is electrically connected with the digital filter, the error reduction module is electrically connected with the parameter optimizer, the physical model calculation unit is electrically connected with the central processing unit, the error reduction module is electrically connected with the physical model calculation unit, and the error reduction module is further provided.

[0013] Further, the error reduction step specifically includes: receiving filtered runout measurement data for time domain analysis; calculating thermal drift error values and compensating according to temperature data; performing angle compensation according to displacement sensor installation angle data; calculating radial displacement amount caused by centrifugal force and performing mechanical compensation using an elastic deformation mechanics model; collecting vibration data for frequency spectrum analysis, constructing an adaptive notch filter to filter out environmental vibration effects; performing linearization processing according to the displacement sensor calibration curve; using Kalman filtering algorithm to fuse multiple measurement results, modeling the runout measurement problem as a least squares estimation problem; applying a trend compensation algorithm to eliminate measurement drift caused by system aging; and outputting the final runout value and measurement uncertainty.

[0014] Further, the elastic deformation mechanics model is used to calculate the elastic deformation amount of the rotating component under the action of centrifugal force, and the input includes component mass distribution function, component rotation angular velocity, component material elastic modulus, component geometric size parameters and component support stiffness coefficient, and the output is the radial elastic deformation value of the rotating component at each measurement point.

[0015] Further, the adaptive notch filter optimization function is used to determine the optimal filter parameters to maximize the elimination of environmental vibration interference, and the input includes environmental vibration spectrum distribution function, signal-to-noise ratio evaluation coefficient, filter order parameter, frequency band width parameter and convergence rate parameter, and the output is the optimized notch filter coefficient set.

[0016] The least squares estimation problem is solved using a singular value decomposition method, a regularization term is introduced to suppress measurement noise, multiple measurement data is regarded as observation values containing random errors, an observation equation matrix is constructed, and the true runout value is solved by minimizing the residual sum of squares.

[0017] The spectrum analyzer uses a fast Fourier transform algorithm, the sampling frequency is 10000 Hz, the frequency resolution is 0.1 Hz, and the dynamic range is 96 decibels; the parameter optimizer uses a gradient descent algorithm, the iteration step is 0.01, the maximum iteration number is 1000, and the convergence threshold is 0.001.

[0018] The model calculation unit uses a finite element analysis method, the grid division accuracy is 0.1 mm, and the calculation accuracy is 0.001 mm.

[0019] The working process of the center through-hole part lifting assembly and run-out measuring device includes: starting the device power supply, the control system performs self-checking, and the displacement sensor is calibrated; the assembly parameters and measurement parameters are input on the human-machine interface; the center through-hole part is placed on the assembly workbench and positioned by the positioning pin, and the clamping device is started to fix the center through-hole part; the control system drives the servo motor to operate, and the center through-hole part and the shaft part are assembled and then lowered to the measurement position; the position of the displacement sensor is adjusted to contact the measured part; the center through-hole part is rotated, the data acquisition system collects the run-out data and processes and analyzes the data; the central processing unit compares the final run-out value with the preset standard to determine whether the assembly quality is qualified, and generates a measurement report.

[0020] The elastic deformation mechanics model is used to calculate the elastic deformation of the rotating part under the action of centrifugal force, and the input includes the part mass distribution function, the part rotation angular velocity, the part material elastic modulus, the part geometric size parameters and the part support stiffness coefficient, and the output is the radial elastic deformation value of the rotating part at each measurement point; the radial elastic deformation value is used to subtract from the angle-compensated run-out value to eliminate the measurement error caused by centrifugal force.

[0021] The part mass distribution function is calculated by the central processing unit according to the part specification parameters, the part rotation angular velocity is measured by the central processing unit through the rotation speed sensor, the part material elastic modulus is read by the central processing unit from the material database, the part geometric size parameters are read by the central processing unit from the workpiece database, and the part support stiffness coefficient is calculated by the central processing unit according to the clamp parameters.

[0022] The adaptive notch filter optimization function is used to determine the optimal filter parameters to eliminate environmental vibration interference to the greatest extent, and the input includes the environmental vibration frequency spectrum distribution function, the signal-to-noise ratio evaluation coefficient, the filter order parameter, the frequency band width parameter and the convergence rate parameter, and the output is the optimal notch filter coefficient set; the notch filter coefficient set is used to construct the filter to filter the mechanically compensated run-out value to obtain the vibration-filtered run-out value.

[0023] The environmental vibration frequency spectrum distribution function is obtained by the spectrum analyzer through fast Fourier transform of the vibration data, the signal-to-noise ratio evaluation coefficient is calculated by the parameter optimizer according to the ratio of signal energy to noise energy, the filter order parameter is determined by the parameter optimizer according to the system calculation capacity and the filtering accuracy requirement, the frequency band width parameter is set by the parameter optimizer according to the environmental vibration frequency bandwidth characteristics, and the convergence rate parameter is determined by the parameter optimizer according to the system response speed requirement.

[0024] Among them, the least square estimation problem takes multiple measurement data as random error observation values, constructs an observation equation matrix, solves the true runout value by minimizing the residual sum of squares, solves the least square estimation problem by an iterative optimization algorithm, and obtains the optimal estimation final runout value and estimation accuracy; the trend compensation algorithm compares historical measurement data, identifies the long-term stability change trend of the system, eliminates the measurement drift caused by system aging, and improves the long-term measurement repeatability and reliability.

[0025] Compared with the prior art, the workpiece assembly and runout measurement are integrated through the precise cooperation of the base, the column, the lifting slide, the assembly workbench, the measuring mechanism and other hardware components. The device uses high-precision linear guides and ball screw pairs to ensure the vertical motion accuracy, and a servo motor provides accurate control to ensure the stability of the assembly process.

[0026] Especially, the error reduction module of the present application introduces a multi-level error compensation mechanism, including temperature compensation, angle compensation, centrifugal force compensation, environmental vibration filtering and other algorithms. Through the real-time analysis of elastic deformation by the physical model calculation unit, the adaptive notch filter optimization function is applied to eliminate the influence of environmental vibration, and the least square estimation and Kalman filtering algorithm are combined to fuse multiple measurement results, realizing high-precision runout measurement.

[0027] The present application solves the problem of insufficient runout measurement accuracy in traditional technology, and provides reliable technical support for high-precision assembly of center through-hole components. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 It is a schematic diagram of the overall structure of a center through-hole component lifting assembly and runout measurement device;

[0029] Figure 2 It is a schematic diagram of the structure of the control system and data acquisition system;

[0030] Figure 3 It is a flowchart of the execution steps of the error reduction module;

[0031] Figure 4 It is a work flowchart of the center through-hole component lifting assembly and runout measurement device;

[0032] In the drawings, the components represented by each reference number are listed as follows:

[0033] 10, base; 11, horizontal adjustment bolt; 20, column; 30, lifting slide; 31, ball screw pair; 32, coupling; 33, servo motor; 40, assembly workbench; 41, positioning pin; 42, clamping device; 50, measuring mechanism; 51, displacement sensor; 52, measuring support; 60, adjustment mechanism; 61, horizontal adjustment slider; 62, vertical adjustment slider; 70, center through-hole component. DETAILED DESCRIPTION

[0034] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application.

[0035] As shown in Figure 1 FIG. 1 is a structural schematic diagram of a center through-hole component lifting assembly and runout measuring device provided by the present application. The device comprises a base 10, a column 20, a lifting slide 30, an assembly workbench 40, a measuring mechanism 50, a control system and a data acquisition system. The base is used to support the entire device and ensure the levelness. The column is vertically fixed on the base and provides vertical guidance. The lifting slide is installed on the column and realizes vertical lifting movement. The assembly workbench is installed on the lifting slide and is used to place a center through-hole component 70. The measuring mechanism is located above the assembly workbench and is used to measure the runout value of the center through-hole component. The control system is used to control the operation of the entire device. The data acquisition system is used to acquire and process the runout measurement data.

[0036] The base is made of high-strength cast iron. A horizontal adjustment bolt 11 is arranged on the base. The horizontal adjustment bolt is in threaded connection with the base and is used to adjust the levelness of the base.

[0037] The column is made of high-quality steel material. The surface of the column is subjected to grinding processing. The column is provided with a high-precision linear guide rail. The column is fixedly connected with the base through bolts.

[0038] The lifting slide is in sliding cooperation with the column guide rail. A ball screw pair 31 is installed inside the lifting slide. The screw of the ball screw pair is in sliding cooperation with the lifting slide. The nut of the ball screw pair is fixedly connected with the lifting slide. The lower end of the screw of the ball screw pair is supported and connected with the base through a bearing. The upper end of the screw is connected with a servo motor 33 through a coupling 32. The servo motor is fixedly installed on the top of the lifting slide.

[0039] The assembly workbench is installed on the lifting slide. The surface of the assembly workbench is made of hard alloy material. The surface of the workbench is provided with a positioning pin 41 and a clamping device 42. The positioning pin is vertically fixed on the surface of the assembly workbench. The clamping device comprises a cylinder and a clamping jaw. The cylinder is fixed on the assembly workbench. The clamping jaw is connected with the piston rod of the cylinder.

[0040] The measuring mechanism 50 includes a displacement sensor 51 and a measuring bracket 52, the displacement sensor is fixed on the measuring bracket, the measuring bracket is connected to the column through an adjustment mechanism 60, the adjustment mechanism includes a horizontal adjustment slider 61 and a vertical adjustment slider 62, the horizontal adjustment slider 61 slides in conjunction with the column, the vertical adjustment slider slides in conjunction with the horizontal adjustment slider, and the measuring bracket is fixed on the vertical adjustment slider.

[0041] like Figure 2 The control system shown includes a central processing unit, a human-machine interface and an input-output module. The central processing unit is connected to the human-machine interface through a data bus, and the central processing unit is electrically connected to the servo motor, the cylinder and the displacement sensor through the input-output module.

[0042] The spectrum analyzer adopts the fast Fourier transform algorithm with a sampling frequency of 10,000 Hz, a frequency resolution of 0.1 Hz, and a dynamic range of 96 decibels; the parameter optimizer adopts the gradient descent algorithm with an iteration step of 0.01, a maximum number of iterations of 1,000 times, and a convergence threshold of 0.001; the physical model calculation unit adopts the finite element analysis method with a mesh division accuracy of 0.1 mm and a calculation accuracy of 0.001 mm.

[0043] The data acquisition system includes a signal conditioning circuit, an analog-to-digital converter, a digital filter, an error reduction module, a spectrum analyzer, a parameter optimizer and a physical model calculation unit. The signal conditioning circuit is electrically connected to the displacement sensor, the analog-to-digital converter is electrically connected to the signal conditioning circuit, the digital filter is electrically connected to the analog-to-digital converter, the spectrum analyzer is electrically connected to the digital filter, the parameter optimizer is electrically connected to the spectrum analyzer, the error reduction module is electrically connected to the digital filter, the error reduction module is electrically connected to the parameter optimizer, the physical model calculation unit is electrically connected to the central processing unit, the error reduction module is electrically connected to the physical model calculation unit, and the error reduction module is electrically connected to the central processing unit. The error reduction module uses a single-chip microcomputer, which includes a CPU and a memory. The memory stores program instructions, and the CPU is used to execute the program instructions; Figure 3 As shown, when the error reduction module executes the program instructions, it is used to perform the following steps:

[0044] S01, receiving the filtered jitter measurement data output by the digital filter, performing time domain analysis on the filtered jitter measurement data, and calculating the original jitter value and its mean and standard deviation;

[0045] S02, collecting temperature data output by the assembly workbench temperature sensor, establishing a relationship model between temperature and measurement system thermal drift, calculating a thermal drift error value according to the temperature data, and subtracting the thermal drift error value from the original beat value to obtain a beat value after temperature compensation;

[0046] S03, obtaining installation angle data of the displacement sensor, calculating a cosine error caused by installation angle deviation of the displacement sensor, and performing angle compensation on the beat value after temperature compensation to obtain a beat value after angle compensation;

[0047] S04, calculating a radial displacement amount caused by centrifugal force according to component rotation speed data measured by the central processor, calculating radial deformation by applying an elastic deformation mechanics model, the elastic deformation mechanics model being established based on Hooke's law and a centrifugal force calculation formula, and subtracting the radial displacement amount from the beat value after angle compensation to obtain a beat value after mechanics compensation; wherein the elastic deformation mechanics model is used to calculate an elastic deformation amount of a rotating component under the action of centrifugal force, and input includes a component mass distribution function, a component angular velocity, a component material elastic modulus, a component geometric size parameter, and a component support stiffness coefficient, the component mass distribution function being calculated by the central processor according to component specification parameters, the component angular velocity being measured by the central processor through a rotation speed sensor, the component material elastic modulus being read by the central processor from a material database, the component geometric size parameter being read by the central processor from a workpiece database, and the component support stiffness coefficient being calculated by the central processor according to clamp parameters, and output is a radial elastic deformation value of the rotating component at each measurement point; the radial elastic deformation value is used to subtract from the beat value after angle compensation to eliminate measurement errors caused by centrifugal force;

[0048] S05, collect vibration data output by an environmental vibration sensor in the control system, identify environmental vibration frequency components by performing spectrum analysis on the vibration data through the spectrum analyzer, and construct an adaptive notch filter optimization function to filter out the influence of environmental vibration on the jump measurement; the adaptive notch filter optimization function is used to determine the optimal filter parameters to eliminate environmental vibration interference to the greatest extent, and the input includes an environmental vibration spectrum distribution function, a signal-to-noise ratio evaluation coefficient, a filter order parameter, a frequency band width parameter and a convergence rate parameter, the environmental vibration spectrum distribution function is obtained by performing fast Fourier transform on the vibration data through the spectrum analyzer, the signal-to-noise ratio evaluation coefficient is calculated according to the ratio of signal energy to noise energy by the parameter optimizer, the filter order parameter is determined according to the system calculation capacity and the filtering accuracy requirement by the parameter optimizer, the frequency band width parameter is set according to the bandwidth characteristics of the environmental vibration frequency by the parameter optimizer, and the convergence rate parameter is determined according to the system response speed requirement by the parameter optimizer, and the output is an optimal notch filter coefficient set; the notch filter coefficient set is used to construct a filter to filter the jump value after the mechanical compensation, and a jump value after vibration filtering is obtained;

[0049] S06, according to the calibration curve of the displacement sensor, the measurement error caused by the nonlinear response of the displacement sensor is calculated, the vibration filtered jump value is linearly processed, and a corrected jump value is obtained;

[0050] S07, using Kalman filtering algorithm to fuse multiple measurement results, modeling the jump measurement problem as a least square estimation problem, solving the optimal estimation value by constructing an overdetermined equation group, the least square estimation problem is solved by singular value decomposition method, and a regularization term is introduced to suppress measurement noise and improve jump measurement accuracy; this step regards multiple measurement data as observation values containing random errors, constructs an observation equation matrix, solves the true jump value by minimizing the residual sum of squares, solves the least square estimation problem by an iterative optimization algorithm, obtains the final jump value and its estimation accuracy of the optimal estimation, and provides a basis for subsequent judgment;

[0051] S08, comparing historical measurement data, identifying the long-term stability change trend of the system, and applying a trend compensation algorithm to eliminate measurement drift caused by system aging;

[0052] S09, output the final jump value and the measurement uncertainty, and transmit the final jump value and the measurement uncertainty to the central processor for display and storage.

[0053] As shown in Figure 4 the operation workflow of the center hole component lifting assembly and jump measurement device of the application includes the following steps:

[0054] S101, start the power supply of the device, the control system performs self-checking, and the displacement sensor is calibrated;

[0055] S102, input assembly parameters and measurement parameters on the human-machine interface, including lifting height, assembly force, and jump measurement accuracy requirements;

[0056] S103, place the center through-hole component on the assembly workbench, position it through the positioning pin, and start the clamping device to fix the center through-hole component;

[0057] S104, the control system drives the servo motor to operate, drives the screw of the ball screw pair to rotate, and makes the lifting slide table rise to a specified height;

[0058] S105, complete the assembly of the center through-hole component and the shaft part, the control system drives the servo motor to reverse, and makes the lifting slide table descend to the measurement position;

[0059] S106, adjust the position of the displacement sensor, make the displacement sensor probe contact with the measured part, and the human-machine interface displays the initial reading;

[0060] S107, rotate the center through-hole component for one revolution, the data acquisition system collects jump data, the frequency spectrum analyzer performs frequency domain analysis on the jump data, and identifies the environmental vibration characteristics;

[0061] S108, the parameter optimizer executes the adaptive notch filter optimization function to determine the optimal filter parameters and filter the jump data;

[0062] S109, the physical model calculation unit executes the elastic deformation mechanics model to calculate the radial elastic deformation value caused by centrifugal force and compensates the jump data;

[0063] S110, the error reduction module applies the least squares estimation problem to process the compensated jump data to obtain the final jump value and the measurement uncertainty;

[0064] S111, the central processing unit compares the final jump value with the preset standard to determine whether the assembly quality is qualified, and generates a measurement report;

[0065] S112, if the final jump value exceeds the allowed range, the control system sends an alarm signal to prompt the operator to adjust;

[0066] S113, after the measurement is completed, the control system drives the air cylinder to release the clamping force, takes out the center through-hole component, and the device returns to the initial state.

[0067] The specific embodiments of the present invention are described in detail below.

[0068] This invention discloses a device for lifting, assembling, and measuring runout of components with central through-holes. The device includes a base, a column, a lifting slide, an assembly workbench, a measuring mechanism, a control system, and a data acquisition system. The overall device is designed to achieve precise assembly and runout measurement of components with central through-holes, ensuring part assembly quality.

[0069] The base of the device is made of high-strength cast iron and features a leveling screw threaded into the base. This allows for adjustment of the base's levelness to ensure accurate measurement datums. The base's top surface is machined to a flatness of 0.02 mm per 1000 mm and a hardness of at least HRC 48, ensuring the base remains stable during long-term use.

[0070] The columns are constructed from high-quality 45# steel, achieving a surface hardness of HRC 40-45 after quenching and tempering heat treatment. The columns are precision-ground to a surface roughness of less than 0.8 microns (Ra). They are equipped with high-precision linear guides, achieving parallelism within 0.005 mm / 1000 mm and perpendicularity within 0.01 mm / 1000 mm. The columns are secured to the base with M16 high-strength bolts, and the connection surfaces are ground to ensure a minimum contact area of ​​85%.

[0071] The lifting slide and the column guide rail are fitted together through a sliding block. The slider is a preloaded linear slider with adjustable clearance and a sliding accuracy grade of P. A precision ball screw pair is installed inside the lifting slide. The screw pitch is 5 mm, the accuracy grade is C5, and the lifting accuracy is 0.01 mm. The lower end of the screw of the ball screw pair is connected to the base through an angular contact ball bearing support. The bearing model is 7208C and the preload force is 280 N. The upper end of the screw is connected to the servo motor through an elastic coupling. The coupling model is LK5, and the compensation coaxiality error does not exceed 0.05 mm. The servo motor is an AC permanent magnet synchronous servo motor with a power of 750 watts, a rated speed of 1500 rpm, a position control accuracy of ±0.001 degrees, and is fixedly installed on the top of the lifting slide.

[0072] The assembly workbench is installed on the lifting slide, using T-slot connection and fastened by high-strength bolts. The workbench surface is made of carbide material with a surface hardness of HRC60 or above and a flatness of 0.01 mm / 500 mm. The workbench surface is provided with φ10

[0073] φ10 mm positioning pins 3, perpendicularity 0.005 mm / 100 mm, the positioning pins are vertically fixed on the surface of the assembly workbench. The clamping device includes a pneumatic cylinder and a clamping jaw. The pneumatic cylinder is a thin type, model SMC-MQMB32-50D, working pressure 0.6 MPa, and the clamping force can be adjusted in the range of 100-300 N. The clamping jaw is made of tool steel with quenching treatment, hardness HRC 58-62, and is connected with the piston rod of the pneumatic cylinder through threads. The contact surface is plated with hard chrome to reduce wear.

[0074] The measuring mechanism includes a displacement sensor and a measuring support. The displacement sensor is an inductance type displacement sensor with a measurement accuracy of 0.001 mm, a linearity of 0.15%, a repeatability of 0.5 μm, and a measurement range of 0-2 mm. The measuring support is connected with the column through an adjusting mechanism, which includes a horizontal adjusting slide and a vertical adjusting slide. The horizontal adjusting slide is in sliding fit with the column through a dovetail groove, and the vertical adjusting slide is in sliding fit with the horizontal adjusting slide through a screw feed mechanism, with an adjusting accuracy of 0.01 mm. The measuring support is fixed on the vertical adjusting slide, and is made of aluminum alloy with an anodized surface treatment, which is rigid and not easy to deform.

[0075] The control system adopts an industrial programmable controller with a 32-bit RISC architecture, a central processing unit with a main frequency of 166 MHz, and a memory capacity of 128 Mbytes. The human-machine interface adopts a 10.4-inch touch screen with a resolution of 1024 x 768 pixels and a display refresh rate of 60 Hz. The input and output module includes 16 digital input channels, 16 digital output channels, 8 analog input channels, and 4 analog output channels. The central processing unit is connected with the human-machine interface through a CAN bus, with the servo driver through an EtherCAT bus, and with the displacement sensor through an RS485 interface.

[0076] The signal conditioning circuit of the data acquisition system adopts an instrument amplifier AD620 with an adjustable gain range of 1-1000 times and a common-mode rejection ratio of 120 dB. The analog-to-digital converter adopts a 24-bit Σ-Δ type ADC with a sampling rate of 10 kHz and a signal-to-noise ratio of 96 dB. The digital filter adopts an IIR structure with an adjustable cutoff frequency range of 10-5000 Hz, a filter order of 8, and a stopband attenuation of 60 dB. The spectrum analyzer adopts a fast Fourier transform algorithm with a sampling frequency of 10000 Hz, a frequency resolution of 0.1 Hz, and a dynamic range of 96 dB. The parameter optimizer adopts a gradient descent algorithm with an iteration step of 0.01, a maximum iteration number of 1000, and a convergence threshold of 0.001. The physical model calculation unit adopts a finite element analysis method with a meshing accuracy of 0.1 mm and a calculation accuracy of 0.001 mm.

[0077] The error reduction module first receives the filtered runout measurement data output by the digital filter, performs time domain analysis, and calculates the original runout value, its mean value, and standard deviation. Then, temperature data output by a temperature sensor on the assembly workstation is collected. The temperature sensor is a PT100 type platinum resistance sensor with a temperature measurement range of -50 to 200 degrees Celsius and an accuracy of 0.1 degrees Celsius. A relationship model between temperature and measurement system thermal drift is established, and the thermal drift error value is calculated based on the temperature data. The thermal drift error value is subtracted from the original runout value to obtain the temperature-compensated runout value.

[0078] The installation angle data of the displacement sensor is obtained by a high-precision angle encoder with a resolution of 0.01 degrees. The cosine error caused by the installation angle deviation of the displacement sensor is calculated, and the temperature-compensated runout value is angle-compensated to obtain the angle-compensated runout value. The radial displacement amount caused by centrifugal force is calculated based on the component rotation speed data measured by the central processor, and the radial deformation is calculated using an elastic deformation mechanics model based on Hooke's Law and the centrifugal force calculation formula. The radial displacement amount is subtracted from the angle-compensated runout value to obtain the mechanics-compensated runout value.

[0079] Vibration data output by an environmental vibration sensor in the control system is collected. The vibration sensor is a piezoelectric acceleration sensor with a frequency range of 1 to 5000 Hz and a sensitivity of 100 mV / g. The vibration data is analyzed by a frequency spectrum analyzer to identify the environmental vibration frequency components, and an adaptive notch filter optimization function is constructed to filter out the influence of environmental vibration on runout measurement. The measurement error caused by the nonlinear response of the displacement sensor is calculated based on the calibration curve of the displacement sensor, and the vibration-filtered runout value is linearized to obtain the corrected runout value.

[0080] The Kalman filter algorithm is used to fuse multiple measurement results, and the runout measurement problem is modeled as a least squares estimation problem. The optimal estimation value is solved by constructing an overdetermined equation system. The least squares estimation problem is solved using the singular value decomposition method, and a regularization term is introduced to suppress measurement noise and improve runout measurement accuracy. By comparing historical measurement data, the long-term stability trend of the system is identified, and a trend compensation algorithm is applied to eliminate measurement drift caused by system aging. Finally, the runout value and measurement uncertainty are output and transmitted to the central processor for display and storage.

[0081] The working procedure firstly starts the device power supply, the control system performs self-checking, and the displacement sensor performs calibration. The assembly parameters and measurement parameters are input on the human-machine interface, including the lifting height, assembly force, and the accuracy requirement of the run-out measurement. The center through-hole component is placed on the assembly workbench and positioned by the positioning pin, and the clamping device is started to fix the center through-hole component. The control system drives the servo motor to operate, drives the screw of the ball screw pair to rotate, and makes the lifting slide table rise to the specified height. The assembly of the center through-hole component and the shaft part is completed, and the control system drives the servo motor to reverse, so that the lifting slide table descends to the measurement position.

[0082] The position of the displacement sensor is adjusted so that the displacement sensor probe contacts the measured part, and the initial reading is displayed on the human-machine interface. The center through-hole component is rotated for one revolution, the data acquisition system collects the run-out data, the frequency spectrum analyzer performs frequency domain analysis on the run-out data to identify the environmental vibration characteristics. The parameter optimizer executes the adaptive notch filter optimization function to determine the optimal filter parameters and filter the run-out data. The physical model calculation unit executes the elastic deformation mechanics model to calculate the radial elastic deformation value caused by the centrifugal force and compensates the run-out data. The error reduction module applies the least squares estimation problem to process the compensated run-out data to obtain the final run-out value and measurement uncertainty. The central processing unit compares the final run-out value with the preset standard to determine whether the assembly quality is qualified, and generates a measurement report. If the final run-out value exceeds the allowed range, the control system sends an alarm signal to prompt the operator to adjust. After the measurement is completed, the control system drives the cylinder to release the clamping force, takes out the center through-hole component, and the device returns to the initial state.

[0083] The specific implementation of the error reduction module is described in detail as follows.

[0084] The specific implementation of step S01 is to receive the filtered run-out measurement data output by the digital filter, perform time domain analysis on the filtered run-out measurement data, and calculate the original run-out value, its mean value and standard deviation. This step first acquires the output signal of the digital filter through the data acquisition interface, and sets the sampling frequency to 1000 Hz, collecting 5000 data points for 5 seconds each time. Then the sliding window method is used to process the collected data in segments, with a window length of 512 points and an overlap rate of 50%. The maximum value, minimum value, peak-to-peak value, mean value, median value, standard deviation, skewness and kurtosis of the data in each sliding window are calculated. A four-parameter Weibull distribution is used to fit the data distribution characteristics, and if the goodness of fit is less than 0.95, the system will issue a data anomaly warning. In the time domain analysis process, the threshold method is used to identify abnormal points, and the threshold value is set to mean value ± 3 times standard deviation. Data points exceeding the threshold will be marked but not directly excluded, but will be used as reference information for subsequent processing. The purpose of this step is to obtain the original feature information of the run-out measurement, providing basic data for subsequent error compensation.

[0085] The specific implementation of step S02 is to collect the temperature data output by the assembly workbench temperature sensor, establish a temperature and measurement system thermal drift relationship model, calculate the thermal drift error value according to the temperature data, and subtract the thermal drift error value from the original bounce value to obtain the temperature-compensated bounce value. This step first collects the workbench temperature through a PT100 type platinum resistance temperature sensor, with a sampling period of 100 ms, and takes the average value of 30 data points as the current temperature value. At the same time, the displacement sensor shell temperature, environmental temperature and base temperature are collected to construct a multi-point temperature field distribution map. Based on the multiple linear regression method, a temperature and measurement system thermal drift relationship model is established, and the general form of the model is thermal drift amount = a x workbench temperature + b x sensor temperature + c x environmental temperature + d x base temperature + e, where the coefficients a, b, c, d and e are obtained by offline calibration. In actual application, the calibration temperature range is 15-35℃, with a step of 5℃. For the obtained thermal drift model, the cross-validation method is used to verify its effectiveness, and the determination coefficient R 2 must be greater than 0.95 before it can be applied. Temperature compensation uses real-time calculation, that is, temperature data is collected before each measurement, the thermal drift error value under the current temperature condition is calculated, and then the error value is subtracted from the original bounce value. For the case where the environmental temperature changes at a rate of more than 1℃ / min, the temperature collection period will be shortened to 10 ms and the thermal drift coefficient will be recalculated. This step aims to eliminate the influence of environmental temperature changes on the measurement system and improve the consistency of the measurement system in different temperature environments.

[0086] The specific implementation of step S03 is to obtain the installation angle data of the displacement sensor, calculate the cosine error caused by the installation angle deviation of the displacement sensor, and perform angle compensation on the temperature-compensated bounce value to obtain the angle-compensated bounce value. This step first measures the installation angle of the displacement sensor in real time through a high-precision angle encoder, with a resolution of 0.01° and a repeatability of ±0.005°. At the same time, based on the machine vision method, the sensor installation angle is detected, the sub-pixel edge detection algorithm is used to determine the sensor axis direction, the image resolution is 1280x1024 pixels, the lens focal length is 25 mm, and the working distance is 200 mm. The encoder measurement results and the machine vision detection results are fused to obtain the final installation angle value by weighted average method, and the weight proportion is determined according to the respective measurement uncertainty. Based on the measured installation angle, the cosine error is calculated, that is, ΔL = L x (1-cosθ), where ΔL is the cosine error, L is the actual displacement, and θ is the included angle between the sensor axis and the normal of the measured surface. When the included angle is less than 5°, the small-angle approximation simplification is used, that is, ΔL ≈ L x θ 2 / 2. For different measurement positions, an angle compensation lookup table is established, and the lookup table is implemented by cubic spline interpolation method. When the installation angle changes more than 0.1°, the system will automatically recalculate the compensation value. The purpose of this step is to eliminate the system error caused by the installation angle deviation of the displacement sensor and improve the measurement accuracy.

[0087] The specific implementation of step S04 is to calculate the radial displacement amount caused by centrifugal force according to the component rotation speed data measured by the central processor, calculate the radial deformation by applying the elastic deformation mechanics model, subtract the radial displacement amount from the angle-compensated runout value, and obtain the runout value after mechanical compensation. This step first measures the rotation speed of the component through the Hall effect speed sensor, the sampling frequency is 10 kHz, and the signal is processed through a low-pass filter with a cutoff frequency of 500 Hz. Based on the measured rotation speed, the angular velocity ω is calculated, with the unit of rad / s. According to the mass distribution characteristics of the component, a centrifugal force calculation model is established, and the centrifugal force F calculation formula is F = mrω 2 , where m is the component mass, and r is the distance from the center of mass to the rotation center. The finite element method is applied to construct the elastic deformation mechanics model, the mesh division adopts hexahedral elements, the element size is 0.5 mm, and the model considers the anisotropic characteristics of the material and the influence of temperature on the elastic modulus. The boundary conditions are defined as fixed constraints and distributed force loads, the Newton-Raphson iteration method is used to solve the nonlinear equation set, and the convergence accuracy is set to 10 -6 . For high computational complexity, modal simplification technology is used to reduce the calculation amount, and the first 20 modes are retained. Combined with the measured component support stiffness coefficient, the radial elastic deformation value of each measurement point under the action of centrifugal force is calculated. When the rotation speed changes more than 5%, the system will recalculate the deformation value. The purpose of this step is to eliminate the influence of the radial deformation of the component on the measurement results caused by the centrifugal force generated by high-speed rotation, especially suitable for precision measurement under high-speed rotation working condition.

[0088] The specific implementation of step S05 is to collect vibration data output by the environmental vibration sensor in the control system, identify the environmental vibration frequency component through spectrum analysis of the vibration data by a spectrum analyzer, and construct an adaptive notch filter optimization function to filter out the influence of environmental vibration on the beat measurement. This step first collects environmental vibration signals through a piezoelectric acceleration sensor, with a sensitivity of 100 mV / g and a frequency response range of 1-5000 Hz. The sampling frequency is set to 10 kHz, and 2 seconds of data are collected each time. The collected vibration data is preprocessed using the Hanning window function to reduce the spectral leakage effect. Then, the fast Fourier transform algorithm is used for spectrum analysis, with a transform point number of 8192 points and a frequency resolution of 1.22 Hz. The main vibration frequency component is identified by a peak detection algorithm, and the significant frequency component is determined by an energy ratio threshold method, with a threshold of 3% of the total energy. Based on the identified environmental vibration characteristics, an adaptive notch filter is constructed, with a second-order IIR structure, a center frequency corresponding to the identified vibration frequency, and a bandwidth parameter dynamically adjusted according to the stability of the vibration frequency, with a typical value of ±5% of the center frequency. The least mean square error criterion is used to design the optimization function, and the optimal filter coefficients are iteratively solved by the stochastic gradient descent method, with an initial learning rate of 0.05 that decays with the number of iterations. The convergence criterion is that the coefficient change rate is less than 0.1% for 10 consecutive iterations. The filter depth is adjusted adaptively for different vibration intensities, with a stopband attenuation of 40 dB in strong vibration regions and 20 dB in weak vibration regions. The purpose of this step is to identify and filter out the interference of environmental vibration on the beat measurement, and to improve the signal-to-noise ratio of the measurement signal.

[0089] The specific implementation of step S06 is to calculate the measurement error caused by the nonlinear response of the displacement sensor according to the calibration curve of the displacement sensor, linearize the beat value after vibration filtering, and obtain the corrected beat value. This step first obtains the calibration data of the displacement sensor through a precision displacement stage, with a resolution of 0.1 μm and a repeat positioning accuracy of ±0.3 μm. The calibration points are spaced 0.05 mm apart, covering the entire measurement range of 0-2 mm. The calibration data is fitted to a quintic polynomial model, with a fitting residual root mean square error controlled within 0.5 μm. Based on the fitted polynomial model, the actual response curve of the sensor is constructed, i.e. the relationship function V=f(d) between the output voltage and the actual displacement, where V is the output voltage and d is the actual displacement. The inverse function d=f -1(V) to obtain a correction function of the actual displacement from the output voltage. Considering the complexity of the inverse function solution, a piecewise linear interpolation method is used to realize fast table lookup, and the interval of the interpolation table is 0.01V. For the boundary conditions not covered by the interpolation table, an extrapolation algorithm is used for processing, and the extrapolation length is not more than 5% of the measurement range. In addition, considering the thermal characteristics of the sensor, temperature-dependent correction coefficients are established for different operating temperatures, with a temperature range of 15-35°C and a step size of 5°C. When the ambient temperature changes more than ±2°C, the corresponding correction coefficient is reselected. This step aims to eliminate the influence of the inherent nonlinear characteristics of the displacement sensor on the measurement accuracy, and to realize high-precision displacement measurement.

[0090] The specific implementation of step S07 is to fuse multiple measurement results using the Kalman filter algorithm, model the beat measurement problem as a least squares estimation problem, and solve the optimal estimate value by constructing an overdetermined equation system. This step first models the beat measurement as a linear system, with the state vector containing the true beat value and its rate of change, and the observation vector being the measured beat value. The system state equation is x k = Ax k-1 + w k-1 , and the observation equation is z k = Hx k + v k , where x k is the state vector at time k, A is the state transition matrix, w k-1 is the process noise, z k is the observation vector, H is the observation matrix, and v k is the observation noise. Both the process noise and the observation noise are assumed to be Gaussian white noise, with covariance matrices Q and R respectively. Based on the actual characteristics of the system, the state transition matrix A is set to the identity matrix, indicating that the beat value remains relatively stable over a short period of time. The observation matrix H is the identity matrix, indicating that the state variable is directly observed. The process noise covariance matrix Q is initially set to 10 -6 ·I, and the observation noise covariance matrix R is initially set to 10 -4 ·I, where I is the identity matrix. An adaptive mechanism is used to dynamically adjust the noise covariance matrix, based on the statistical characteristics of the innovation sequence to estimate the real-time noise level. The Kalman filter performs two steps of prediction and update. The prediction step calculates the prior state estimate and the prior error covariance, and the update step calculates the Kalman gain, the posterior state estimate, and the posterior error covariance. The convergence criterion for the filter is that the trace of the posterior error covariance matrix is less than a preset threshold of 10 -5To improve the estimation accuracy, the least square method is used to construct an over-determined equation set Ax = b, where the matrix A is composed of the coefficient matrix of multiple measurements, the vector b is the corresponding measurement value, and the vector x is the real fluctuation value to be solved. The singular value decomposition method is used to solve the over-determined equation set, and a regularization term is introduced to suppress the influence of noise, and the regularization parameter is determined by the L curve method, and the typical value is 10 -3 The purpose of this step is to improve the accuracy and reliability of the fluctuation measurement by data fusion technology, and to reduce the influence of random errors.

[0091] The specific implementation of step S08 is to compare historical measurement data, identify the long-term stability change trend of the system, and apply a trend compensation algorithm to eliminate measurement drift caused by system aging. This step first establishes a measurement data history library to record the measurement results of the system on the standard sample at different time points. The standard sample is a standard block certified by metrology, with a nominal value of 1.000 mm and an uncertainty of ±0.1 μm. The system automatically performs standard sample measurement process every 100 working hours, and records the measurement value and environmental parameters. Through time series analysis method, the long-term change trend of the measurement value is extracted, and the exponential weighted moving average algorithm is used to smooth the short-term fluctuations, and the smoothing factor a is set to 0.3. Combined with control chart technology, the Shewhart control chart of the system measurement value is drawn, and the control limit is set to ±3σ. When the measurement value continuously exceeds the ±2σ warning line for 3 times or exceeds the ±3σ control line for a single time, the system will trigger the calibration reminder. Through regression analysis, the mathematical model of system drift is identified, common models include linear drift model d(t) = d0 + kt and exponential drift model d(t) = d0 + a(1 - e -bt ), where d(t) is the drift at time t, d0, k, a, b are model parameters. Based on the identified drift model, the drift compensation value at the current time is calculated in real time, and the value is subtracted from the measurement result. For abnormal drift beyond expectations, the system will issue a maintenance warning. The purpose of this step is to monitor and compensate the long-term stability change of the measurement system, prolong the calibration period, and ensure the long-term measurement accuracy.

[0092] The specific implementation of step S09 is to output the final runout value and the measurement uncertainty, and transmit the final runout value and the measurement uncertainty to the central processor for display and storage. This step first calculates the combined standard uncertainty of the measurement result according to the GUM uncertainty evaluation method. The uncertainty sources include the sensor repeatability uncertainty u1, the estimated value of which is 0.0005 mm; the calibration curve fitting uncertainty u2, the estimated value of which is 0.0003 mm; the temperature compensation residual uncertainty u3, the estimated value of which is 0.0002 mm; the angle compensation residual uncertainty u4, the estimated value of which is 0.0001 mm; the mechanical compensation residual uncertainty u5, the estimated value of which is 0.0002 mm; the environmental vibration residual uncertainty u6, the estimated value of which is 0.0003 mm; the data fusion algorithm uncertainty u7, the estimated value of which is 0.0001 mm; and the long-term stability uncertainty u8, the estimated value of which is 0.0004 mm. Considering the correlation between each uncertainty component, a covariance matrix is established to calculate the combined standard uncertainty u c . The expanded uncertainty U = k·u c , where k is a coverage factor, and the value is 2, corresponding to a confidence level of about 95%. The final runout value and its expanded uncertainty are output in the format of “measurement value ± expanded uncertainty”, and the effective digits are retained to 0.0001 mm. The data is transmitted to the central processor through an industrial communication protocol, using the Modbus RTU protocol, with a baud rate of 115200 bit / s, 8-bit data, 1-bit stop, and no parity bit. After receiving the data, the central processor performs data verification to confirm the data integrity. The measurement result is displayed in real time on the human-machine interface and stored in the local database, which uses the SQLite format to record the measurement time, measurement value, uncertainty, environmental parameters, and other information. According to the preset threshold, it is judged whether the measurement result is qualified or not, and the qualified threshold is set to 0.01 mm. When the value exceeds the threshold, an audible and light alarm is triggered. The purpose of this step is to provide reliable measurement result output and complete measurement uncertainty information, and to provide scientific basis for subsequent quality judgment and decision-making.

[0093] The mathematical models or calculation processes involved in the present application are described in detail below.

[0094] In step S01, the time domain analysis of the filtered runout measurement data involves the following calculation process. The formula for calculating the original runout value is as follows:

[0095] R raw (i) = max(D(i)) - min(D(i));

[0096] In the formula, R raw(i) is the original beat value of the i-th sliding window, in millimeter; D(i) is the displacement sensor data set in the i-th sliding window, in millimeter; max() and min() are the maximum and minimum operators, respectively.

[0097] The statistical feature calculation formula of the beat data is as follows:

[0098]

[0099] In the formula, μ is the mean of the original beat value, in millimeter; σ is the standard deviation of the original beat value, in millimeter; γ is the skewness, dimensionless; κ is the kurtosis, dimensionless; and N is the total number of the sliding window, and a typical value is 19 (considering 5000 data points, a sliding window length of 512 points, and an overlap rate of 50%). The skewness reflects the symmetry of the data distribution, and a positive value indicates that the distribution is right-skewed, and a negative value indicates that the distribution is left-skewed. The kurtosis reflects the sharpness of the data distribution, and a positive value indicates that the distribution is relatively sharp, and a negative value indicates that the distribution is relatively flat. These statistical features are used to judge the distribution characteristics of the beat data and provide a basis for subsequent processing.

[0100] The four-parameter Weibull distribution fitting formula is as follows:

[0101]

[0102] In the formula, f(x) is the probability density function; β is the shape parameter, which controls the shape of the distribution, and a typical value range is 1.5-3.5; η is the scale parameter, which controls the dispersion degree of the distribution, and a typical value range is 0.005-0.05; γ is the position parameter, which controls the starting position of the distribution, and a typical value range is -0.01-0.01; and δ is the offset parameter, which controls the offset of the overall distribution, and a typical value range is -0.005-0.005. The four parameters are obtained by using the maximum likelihood estimation method, and are obtained by solving the following equation set:

[0103]

[0104] In the formula, Γ() is the gamma function. The solution of the above equation set uses the Newton-Raphson iteration method, the initial value is selected as an empirical value, and the iteration convergence condition is that the relative change rate of the parameters is less than 10 -6 or the number of iterations reaches 100 times. The reason for using the Weibull distribution is that it has good flexibility and can fit a variety of shapes of probability distribution, and is particularly suitable for describing beat measurement values which may have skewed data distribution.

[0105] The abnormal point identification uses the standardized score method, and the calculation formula is as follows:

[0106]

[0107] In the formula, Z(i) is the standardized score of the i th original beat value, dimensionless; Outlier(i) is an outlier marker, 1 indicating an outlier and 0 indicating a normal point. The threshold value 3 corresponds to a confidence level of 99.7% under a normal distribution, and the threshold value can be adjusted in the range of 2-4 according to actual application requirements. The standardized score method judges outliers based on the standard deviation of data, and is suitable for data approximately following a normal distribution. For severely skewed data, a four-quartile-based outlier detection method can be used to improve detection accuracy.

[0108] In step S02, the temperature and the relationship model of the measurement system thermal drift involves the following calculation process. The multiple linear regression model is specifically expressed as follows:

[0109] ΔT(t) = a · T workbench (t) + b · T sensor (t) + c · T environment (t) + d · T base (t) + e + ε T ;

[0110] In the formula, ΔT(t) is the thermal drift at time t, with units of millimeters; T workbench (t) is the workbench temperature at time t, with units of degrees Celsius; T sensor (t) is the displacement sensor housing temperature at time t, with units of degrees Celsius; T environment (t) is the ambient temperature at time t, with units of degrees Celsius; T base (t) is the base temperature at time t, with units of degrees Celsius; a, b, c, d, e are regression coefficients, with units of millimeters / degrees Celsius (a, b, c, d) and millimeters (e); ε T is the residual error term of the temperature compensation model, with units of millimeters, and the range is ±0.0005 millimeters.

[0111] The regression coefficients are obtained by solving the normal equation set using the least squares method:

[0112] In the formula,

[0113]

[0114] wherein m is the number of calibration data points, and a typical value is 35 (considering a temperature range of 15-35℃, a step of 5℃, and 5 sets of data collected at each temperature point). The calibration data is obtained through a controlled environment experiment, and the specific steps are as follows: (1) install a standard piece with a standard runout value of zero on an assembly workbench; (2) use a temperature control box to control the environmental temperature of the entire device, and the temperature range is 15-35℃, with a step of 5℃; (3) after stabilization at each temperature point (temperature change rate less than 0.1℃ / 10 minutes), measure the runout value of the standard piece, and repeat the measurement 5 times; (4) calculate the difference between the measured value and the standard value as the thermal drift ΔT.

[0115] The determination coefficient R of the model 2 The calculation formula is as follows:

[0116]

[0117] wherein Y i is the actual observed thermal drift; is the thermal drift predicted by the model; is the mean value of the thermal drift. The determination coefficient R 2 reflects the goodness of fit of the model, and the closer the value is to 1, the better the fitting effect of the model. The reason for using the multiple linear regression model is its simplicity and interpretability, which can intuitively reflect the contribution of each temperature point to the thermal drift. When the relationship between temperature and drift shows nonlinear characteristics, a polynomial regression model can be extended.

[0118] The calculation formula of the runout value after temperature compensation is as follows:

[0119] R temp (i)=R raw (i)-ΔT(t i );

[0120] wherein R temp (i) is the i th runout value after temperature compensation, in millimeters; t i is the time of the i th measurement point, in seconds.

[0121] In step S03, the cosine error calculation caused by the installation angle deviation of the displacement sensor involves the following formula. The installation angle is obtained by using the weighted average method, and the formula is as follows:

[0122]

[0123] wherein θ is the finally determined installation angle, in degrees; θ encoder is the angle value measured by the angle encoder, in degrees; θ visionAngle measured by machine vision method, unit: degree; w1 and w2 are weight coefficients, dimensionless, whose values are inversely proportional to the uncertainty of each measurement method, and the typical values are 0.7 and 0.3 respectively.

[0124] The cosine error caused by the installation angle deviation is calculated by the following formula:

[0125] ΔL = L·(1-cosθ);

[0126] In the formula, ΔL is the cosine error, unit: millimeter; L is the actual displacement, unit: millimeter; θ is the angle between the sensor axis and the normal of the measured surface, unit: degree. When θ is less than 5°, the Taylor series expansion is used for simplified calculation:

[0127]

[0128] In the formula, θ is in radians, θ(radian) = θ(degree)·π / 180. The reason for using small-angle approximation is that the calculation is efficient, and the approximation error can be ignored in the small-angle range (when θ = 5°, the approximation error is only 0.09%).

[0129] The runout value after angle compensation is calculated by the following formula:

[0130]

[0131] In the formula, R angle (i) is the i-th runout value after angle compensation, unit: millimeter. The compensation formula is based on the geometric projection principle, which converts the displacement value measured along the sensor axis direction into the actual displacement value perpendicular to the measured surface. For the case of large angle change, a segmented compensation model can be established to compensate for different angle ranges.

[0132] In step S04, the radial displacement caused by centrifugal force is calculated by the following formula. The angular velocity calculation formula is as follows:

[0133]

[0134] In the formula, ω is the angular velocity, unit: rad / s; n is the rotation speed, unit: r / min.

[0135] The centrifugal force calculation formula is as follows:

[0136] F(r) = m(r)·r·ω 2 ;

[0137] Where F(r) is the centrifugal force at a distance r from the center of rotation, expressed in Newtons; m(r) is the mass at a distance r from the center of rotation, expressed in kilograms; r is the distance from the center of rotation, expressed in meters; and ω is the angular velocity, expressed in radians per second. The mass distribution function m(r) is determined by the component geometry and material density and is typically expressed using the unit mass of the finite element mesh.

[0138] The elastic deformation mechanics model is based on linear elasticity theory and takes into account anisotropic material properties. The governing equation for elastic deformation is:

[0139]

[0140] σ=C:ε;

[0141]

[0142] Where, is the divergence operator; σ is the stress tensor, in Pascals; F is the volume force vector (including centrifugal force), in Newtons per cubic meter; C is the fourth-order elastic tensor, whose components depend on the elastic modulus, Poisson's ratio and anisotropy parameters of the material; ε is the strain tensor, dimensionless; u is the displacement vector, in meters; is the gradient operator; the superscript T indicates transpose.

[0143] Considering the influence of temperature on elastic modulus, the temperature correction formula of elastic modulus is:

[0144] E(T)=E0·[1-α E ·(T-T0)];

[0145] Where, E(T) is the elastic modulus at temperature T, in Pascals; E0 is the elastic modulus at reference temperature T0, in Pascals; α E is the temperature coefficient of elastic modulus, in units of 1 / degree Celsius, with a typical value of 0.001 to 0.003 / degree Celsius; T is the current temperature, in degrees Celsius; T0 is the reference temperature, in degrees Celsius, usually 20 degrees Celsius.

[0146] The boundary conditions of the elastic deformation model are:

[0147]

[0148]

[0149] Where Γ1 is the fixed constraint boundary; Γ2 is the load boundary; n is the boundary normal unit vector; and p is the external load in Pascals. The fixed constraint boundary is typically set at the support location of the component, and the load boundary includes the surfaces where the centrifugal force is applied.

[0150] The discretized form of the elastic deformation equation is:

[0151] K·U=F;

[0152] Where K is the stiffness matrix in Newtons / meters; U is the node displacement vector in meters; and F is the node load vector in Newtons. The stiffness matrix K is generated by finite element discretization. Its size depends on the number of nodes in the mesh. A typical matrix dimension is 10 3 ~10 5 Step.

[0153] The Newton-Raphson iteration method is used to solve the nonlinear equations. The iteration formula is:

[0154] U k+1 =U k -J(U k ) -1 ·R(U k );

[0155] Where U k is the displacement vector of the kth iteration; J(U k ) is in U k The Jacobian matrix at R(U k ) is the residual vector, that is, K·U k -F. The convergence criterion is || R(U k )||<∈, where ∈ is the set convergence accuracy, the typical value is 10 -6 .

[0156] For cases with high computational complexity, modal simplification technology is used, namely:

[0157] U≈Φ·q;

[0158] Where Φ is the modal matrix, which consists of the first m-order eigenvectors of the system; q is the modal coordinate vector. The eigenvalue problem is expressed as: K·φ i =λ i ·M·φ i , i=1,2,...,m;

[0159] Where, φ i is the i-th order eigenvector; i is the i-th order eigenvalue; M is the mass matrix. The advantage of modal simplification is that it can significantly reduce the computational freedom, from the original large-scale problem (usually 10 3 ~10 5 degrees of freedom) to a smaller scale (usually 10 to 100 degrees of freedom).

[0160] The calculation formula of the runout value after mechanical compensation is as follows:

[0161] R mech (i)=R angle (i)-Δu radial (i);

[0162] where R mech (i) is the i-th jump value after mechanical compensation, in millimeter; Δu radial (i) is the radial elastic deformation value at the i-th measurement position, in millimeter, calculated by the elastic deformation model. The reason for using finite element method to calculate the elastic deformation is that it can accurately simulate the deformation behavior of components with complex geometry and non-uniform material properties under centrifugal force, with higher accuracy compared to simplified analytical solutions.

[0163] The influence of environmental vibration on jump measurement filtering in step S05 involves the following formula. The spectral analysis uses the fast Fourier transform algorithm, and the calculation formula is as follows:

[0164]

[0165] where X(k) is the frequency domain sequence; x(n) is the time domain sequence; N is the number of transform points, taking 8192; j is the imaginary unit. In order to reduce spectral leakage, the Hanning window function is applied to the original data:

[0166] w(n) = 0.5·[1-cos(2πn / (N-1))], n = 0, 1,..., N-1;

[0167] x'(n) = x(n)·w(n);

[0168] where w(n) is the Hanning window function; x'(n) is the time domain sequence after windowing. The reason for using the Hanning window is that it achieves a good balance between suppressing spectral leakage and improving frequency resolution.

[0169] The vibration frequency component identification uses the peak detection algorithm, and the judgment condition is:

[0170] |X(k)|>|X(k-1)| and |X(k)|>|X(k+1)| and |X(k)|>α·max(|X|);

[0171] where α is the energy ratio threshold, taking 0.03. The threshold setting is based on the typical characteristics of environmental vibration, which can effectively identify the main vibration frequency components while filtering out minor noise components.

[0172] The adaptive notch filter uses a second-order IIR structure, and the transfer function is:

[0173]

[0174] where ω0is the center frequency in rad / sample, ω0= 2πf0 / f s , f0is the vibration frequency, and f s is the sampling frequency; r is the pole radius, which controls the bandwidth of the filter, the closer r is to 1, the narrower the bandwidth, and a typical value is 0.9-0.99. For multiple vibration frequencies, a cascaded structure is used:

[0175]

[0176] where M is the number of identified vibration frequencies; H i (z) is the notch filter transfer function for the i-th vibration frequency.

[0177] The filter coefficients are optimized using the least mean square error criterion, and the objective function is:

[0178] J(θ) = E[e 2 (n)].

[0179] where θ is the filter parameter vector; e(n) is the error between the filtered signal and the desired signal; and E[] is the mathematical expectation operator. The stochastic gradient descent method is used for optimization:

[0180]

[0181] where θ k is the parameter vector at the k-th iteration; μ k is the learning rate, with an initial value of 0.05 and a decay with the number of iterations: μ k = μ0· γ k , where γ is the decay factor, taken as 0.99; is the gradient of the objective function with respect to the parameters. The convergence criterion is:

[0182]

[0183] where ∈ is the convergence threshold, taken as 0.001. The reason for using an adaptive notch filter is that it can accurately filter out vibrations at specific frequencies while having little effect on other frequency components of the signal, avoiding the problem of excessive filtering leading to loss of useful information.

[0184] The formula for calculating the beat value after vibration filtering is as follows:

[0185]

[0186] where R vib (i) is the i-th beat value after vibration filtering, in millimeters; and h(n) is the impulse response of the notch filter, derived from the transfer function H tatal(z) inverse transform; L is the filter order, taking the filter impulse response length, and the typical value is 100-200.

[0187] In step S06, the measurement error correction caused by the nonlinear response of the displacement sensor involves the following formula. The sensor calibration curve adopts a quintic polynomial model:

[0188] V = a0 + a1 · d + a2 · d 2 + a3 · d 3 + a4 · d 4 + a5 · d 5 ;

[0189] In the formula, V is the sensor output voltage, in volts; d is the actual displacement, in millimeters; a0, a1, a2, a3, a4, and a5 are polynomial coefficients, obtained by least squares fitting. The solution of the polynomial coefficients uses the normal equation:

[0190]

[0191] In the formula,

[0192]

[0193] where m is the number of calibration data points, and the typical value is 41 (considering the measurement range of 0-2 mm, with an interval of 0.05 mm); d i is the calibration displacement value; V i is the corresponding sensor output voltage. The formula for calculating the root mean square error of the fitting residual is:

[0194]

[0195] In the formula, is the voltage value predicted by the polynomial model. The reason for choosing a quintic polynomial model is that it can accurately describe the nonlinear characteristics of the sensor, has higher fitting accuracy than a cubic polynomial, and avoids overfitting problems compared to higher-order polynomials.

[0196] The inverse function solution, i.e., the displacement value is inversely calculated from the voltage value, uses the Newton iteration method:

[0197]

[0198] In the formula, d k is the displacement estimate value of the kth iteration; represents the difference between the polynomial function value and the target voltage V; represents the derivative of the polynomial function. The initial value d0 of the iteration uses linear approximation: d0 = (V-a0) / a1. The iteration stopping condition is |f(d k| <∈ or |d k+1 -d k | <δ, where ∈ and δ are convergence thresholds for function value and displacement value respectively, typical values are 10 -6 volts and 10 -7 millimeters respectively.

[0199] Considering the computational efficiency, piecewise linear interpolation is used in practical application to realize fast table lookup:

[0200]

[0201] where V i and V i+1 are two voltage values adjacent to V in the interpolation table; d i and d i+1 are corresponding displacement values. The interpolation table is constructed by sampling at equal intervals in the range of [0, 2] millimeters, calculating the corresponding voltage values, forming the (d i , V i ) correspondence table, and the size of the table is 201 points (interval 0.01 millimeters).

[0202] The influence of temperature on the characteristics of the sensor is represented by a temperature-dependent correction coefficient:

[0203] K T (T) = 1 + β T · (T - T ref );

[0204] where K T (T) is the temperature correction coefficient, dimensionless; β T is the temperature coefficient of the sensor, with a unit of 1 / degree Celsius, and a typical value of 0.0001-0.0005 / degree Celsius; T is the current temperature, with a unit of degrees Celsius; T ref is the reference temperature, with a unit of degrees Celsius, usually taken as 20 degrees Celsius. The displacement calculation formula is modified as:

[0205] d = d(V) · K T (T);

[0206] where d is the corrected displacement value; d(V) is the displacement value obtained by reversing the voltage according to the calibration curve. The introduction of the temperature correction coefficient is based on the comprehensive consideration of the thermal expansion effect of the sensor material and the temperature dependence of electrical characteristics, which can effectively compensate for the influence of environmental temperature changes on measurement accuracy.

[0207] The corrected runout calculation formula is as follows:

[0208] R linear (i) = R vib (i) · K c (Rvib (i));

[0209] wherein R linear (i) is the i-th beat value after linearization, in millimeters; K c (Rx ib (i)) is a non-linear correction factor, dimensionless, determined based on the inverse function of the sensor calibration curve. The correction factor is calculated based on the sensor's calibration data, by comparing the difference between the ideal linear response and the actual response.

[0210] In step S07, the Kalman filter algorithm fuses multiple measurement results, involving the following formulas. The linear system state equation is:

[0211] x k = A·x k-1 +w k-1 ;

[0212] z k = H·x k +v k ;

[0213] wherein x k is the state vector at time k, wherein r k is the true beat value, is the rate of change of the beat value; A is the state transition matrix, taken as wherein Δt is the sampling time interval, in seconds; w k-1 is the process noise, conforming to a Gaussian distribution with mean zero and covariance matrix Q; z k is the observation vector, i.e. the measured beat value; H is the observation matrix, taken as H = [1 0]; v k is the observation noise, conforming to a Gaussian distribution with mean zero and covariance matrix R.

[0214] Prediction step of the Kalman filter:

[0215]

[0216] wherein is the prior state estimate at time k; is the posterior state estimate at time k-1; P k|k-1 is the prior error covariance matrix at time k; P k-1|k-1 is the posterior error covariance matrix at time k-1. The initial condition is set as wherein r0 is the initial beat value estimate; wherein and are the variance estimates of the initial beat value and rate of change, respectively.

[0217] Update step of Kalman filter:

[0218] K k = P k|k-1 ·H T ·(H·P k|k-1 ·H T + R) -1 ;

[0219]

[0220] P k|k = (I - K k ·H)·P k|k-1 ;

[0221] In the formula, Kk 为 Kalman gain; is the posterior state estimation at time k; P k|k is the posterior error covariance matrix at time k; I is the unit matrix. The Kalman filter algorithm is based on the Bayesian estimation framework, which can achieve optimal estimation of the state in the presence of observation noise, and is particularly suitable for processing measurement data fusion problems in dynamic systems.

[0222] The adaptive mechanism is used to dynamically adjust the noise covariance matrix, based on the statistical characteristics of the innovation sequence:

[0223]

[0224]

[0225] In the formula, λ is the forgetting factor, the value range is [0.9, 0.99], and the typical value is 0.95; is the process noise estimate, e k is the innovation sequence, The introduction of the adaptive mechanism enables the filter to automatically adjust the parameters according to the characteristics of the actual measurement data, improving the estimation accuracy.

[0226] The construction of the least squares estimation problem is based on the overdetermined equation set of multiple measurements:

[0227] A·x = b;

[0228] In the formula, A is the coefficient matrix, which is composed of the coefficient matrices of multiple measurements; x is the real jump value vector to be solved; b is the corresponding measurement value vector. For n measurements, m measurement points, the dimension of matrix A is n×m, and the dimensions of vectors x and b are m and n respectively.

[0229] The singular value decomposition method is used to solve the overdetermined equation set:

[0230] A = U ·∑ · V T ;

[0231] x = V ·∑ -1 · U T · b

[0232] where U and V are left and right singular vector matrices, respectively;∑ is a singular value diagonal matrix;∑ -1 is the generalized inverse of∑. To suppress the influence of noise, a regularization term is introduced, and the modified solution is:

[0233] x = V ·(∑ T ·∑ + α · I) -1 ·∑ T · U T · b

[0234] where α is a regularization parameter, determined by the L-curve method, and a typical value is 10 -3 . The introduction of the regularization parameter is based on the Tikhonov regularization theory, which can effectively suppress the instability of the solution caused by the amplification of measurement noise, and improve the reliability of the estimation result.

[0235] The final beat value is calculated based on the results of Kalman filtering and least squares estimation:

[0236] R final (i) = α · R kalman (i) + (1-α) · R ls (i)

[0237] where R final (i) is the i-th beat value of the final estimate, with units of millimeters; R kalman (i) is the beat value obtained by Kalman filtering; R ls (i) is the beat value obtained by least squares estimation; α is the fusion weight coefficient, with a value range of [0, 1], determined according to the estimation uncertainty of the two methods, where and are the variances of Kalman filtering and least squares estimation, respectively. Through this weighted fusion method, the advantages of the two methods can be comprehensively utilized to obtain more accurate beat estimates.

[0238] In step S08, system long-term stability change trend identification and compensation involves the following formula. The time series analysis uses the exponential weighted moving average algorithm:

[0239] S t = β · Y t + (1-β) · S t-1

[0240] where S​t is the smoothed value at time t; Y t is the observed value at time t; β is the smoothing factor, taking a value of 0.3; S t-1 is the smoothed value at time t-1. The initial value S0 is set as Y0. The exponential weighted moving average algorithm has the advantages of simple calculation and small storage requirement, while effectively smoothing short-term fluctuations and highlighting long-term trends.

[0241] The system drift model includes a linear drift model and an exponential drift model:

[0242] d linear (t) = d0+ k · t;

[0243] d exp (t) = d0+ a · (1- e -b·t );

[0244] In the formula, d linear (t) and d exp (t) are the linear drift and exponential drift at time t, respectively, with units of millimeters; d0 is the initial drift, with units of millimeters; k is the linear drift rate, with units of millimeters / hour; a is the drift amplitude, with units of millimeters; and b is the exponential parameter, with units of 1 / hour. The model parameters are obtained by least squares fitting of historical data, and the selection of the applicable model is based on a comparison of goodness of fit, with the coefficient of determination R 2 as the evaluation criterion. The linear model is suitable for slow and uniform changes in drift; the exponential model is suitable for fast changes at the beginning and stable changes at the end.

[0245] The calculation formula of the beat value after trend compensation is as follows:

[0246] R trend (i) = R final (i) - d(t i );

[0247] In the formula, R trend (i) is the i-th beat value after trend compensation, with units of millimeters; d(t i ) is the drift at time t i , which is calculated from the selected drift model.

[0248] In step S09, measurement uncertainty evaluation and result output involve the following formula. According to the GUM uncertainty evaluation method, the comprehensive standard uncertainty calculation formula is:

[0249]

[0250] In the formula, u c is the comprehensive standard uncertainty, with units of millimeters; c i and cj is the sensitivity coefficient, representing the sensitivity of the measurement result to the i-th and j-th uncertainty component; u i and u j are the i-th and j-th uncertainty components; r ij is the correlation coefficient between the i-th and j-th uncertainty components, -1≤ r ij ≤1, r ii =1. For mutually independent uncertainty components, r ij =0 (i≠j), which simplifies to:

[0251]

[0252] The sensitivity coefficient is determined by the partial derivative of the measurement model with respect to each input quantity:

[0253]

[0254] where f is the measurement model function; x i is the i-th input quantity. For complex models, the sensitivity coefficient can be obtained through numerical differentiation:

[0255]

[0256] where Δx i is the small change in the i-th input quantity.

[0257] The expanded uncertainty calculation formula is:

[0258] U=k·u c ;

[0259] where U is the expanded uncertainty, with units of millimeters; k is the coverage factor, taking a value of 2, corresponding to a confidence level of approximately 95% (assuming that the measurement result follows a normal distribution). The expanded uncertainty provides the uncertainty interval of the measurement result, indicating that the true value is located within this interval with a probability of 95%.

[0260] The measurement result qualification in the control system is based on the following criteria:

[0261] |R trend |≤R threshold ;

[0262] where R trend is the final runout value after trend compensation, with units of millimeters; R threshold is the preset runout threshold, with units of millimeters, taking a value of 0.01 millimeters. This qualification criterion is based on the comprehensive consideration of product quality requirements and process capability, ensuring that the assembly quality meets the expected standards.

[0263] Overall, the error reduction module involves various mathematical models and calculation methods, including statistical analysis, regression fitting, filtering algorithms, finite element analysis, etc., which constitute a complete error analysis and compensation system. These equations and algorithms identify, model and compensate for various error sources through the system, significantly improving the accuracy and reliability of the runout measurement. The mathematical model of each step is based on the corresponding physical principle or statistical theory, and is specially designed for specific error types. The entire module realizes the complete conversion process from the original measurement data to the high-precision runout value, providing a scientific basis for the assembly quality control of the center through-hole parts.

[0264] Specifically, the principle of the present application is that the present application solves the problem of insufficient runout measurement accuracy of center through-hole parts by integrating assembly and measurement functions and combining a multi-level error compensation system. The core technical principles are reflected in the following aspects:

[0265] First, the mechanical structure design of the device follows the principle of high-precision motion control, uses high-strength cast iron base to provide stable support, high-quality steel column and high-precision linear guide to ensure the accuracy of vertical movement, and ball screw pair and servo motor to realize precise lifting control. The synergistic effect of the hard alloy workbench surface and the positioning pin and clamping device ensures the positioning accuracy and stability of the workpiece assembly process, laying a physical foundation for high-precision measurement.

[0266] Second, the error reduction module of the present application is based on the theory of measurement error analysis, and systematically identifies and compensates for various error sources that affect the accuracy of runout measurement. The temperature compensation mechanism eliminates the system error caused by temperature changes by establishing a temperature and thermal drift relationship model; the angle compensation algorithm corrects the measurement angle deviation according to the cosine error principle; the centrifugal force compensation is based on the elastic deformation mechanics model, which accurately calculates the radial deformation of the rotating part under the action of centrifugal force, and subtracts this influence from the original measurement value; the environmental vibration filtering uses frequency spectrum analysis and adaptive notch filtering technology to specifically eliminate interference components of specific frequencies.

[0267] Third, the data processing system integrates signal processing, statistical analysis and machine learning techniques. The non-linear response of the displacement sensor is linearized by the calibration curve; the Kalman filter algorithm models the runout measurement problem as a least squares estimation problem, and solves the optimal estimate value through singular value decomposition method and regularization technique; the trend compensation algorithm identifies the long-term stability change of the system and eliminates the measurement drift. The synergistic effect of these algorithms constitutes a complete error compensation chain, gradually improving the measurement accuracy.

[0268] Finally, the close integration of the control system and the data acquisition system realizes seamless connection of the assembly process and the measurement process, the workpiece is directly measured in the assembled state, and errors introduced by process conversion are avoided. The measurement results and quality judgment are displayed in real time through the man-machine interface, and the operation efficiency and reliability are improved.

[0269] In summary, the present application fundamentally solves the technical problem of insufficient center through-hole component runout measurement precision by the organic combination of mechanical structure optimization, multi-layer error compensation and advanced data processing technology, and realizes the integration, high precision and intelligentization of assembly and measurement.

[0270] A specific embodiment 1 of the present application is provided below, and the specific implementation of each step in embodiment 1 is described in detail as follows.

[0271] The specific implementation of step S01 is to receive the filtered runout measurement data output by the digital filter, perform time domain analysis on the filtered runout measurement data, and calculate the original runout value and its mean value and standard deviation. This step first acquires the output signal of the digital filter through the data acquisition interface, and the sampling frequency is set to 1000 Hz, 5 seconds of data points are collected each time, and a total of 5000 data points are collected. Then, the sliding window method is used to segment the collected data, the window length is 512 points, and the overlap rate is 50%. The maximum value, minimum value, peak-to-peak value, mean value, median value, standard deviation, skewness and kurtosis of the data in each sliding window are calculated. The original runout value is calculated using the formula: R raw (i) = max(D(i)) - min(D(i)), where R raw (i) is the original runout value of the i-th sliding window, in millimeters; D(i) is the displacement sensor data set in the i-th sliding window, in millimeters; max() and min() are the maximum and minimum operators, respectively. The statistical characteristic calculation formula is as follows: where μ is the mean value of the original runout value, in millimeters; σ is the standard deviation of the original runout value, in millimeters; γ is the skewness, dimensionless; κ is the kurtosis, dimensionless; N is the total number of sliding windows, and the typical value is 19. A four-parameter Weibull distribution is used to fit the data distribution characteristics, and the fitting formula is where β is the shape parameter, η is the scale parameter, γ is the location parameter, and δ is the offset parameter. For a fitting goodness less than 0.95, the system will issue a data anomaly warning. In the time domain analysis process, the threshold method is used to identify abnormal points, and the threshold value is set to mean value ± 3 times standard deviation, and the standardized score is calculated by When |Z(i)|>3, the data point is determined as an abnormal point. The data point exceeding the threshold is marked but not directly eliminated, and is used as reference information for subsequent processing. The purpose of this step is to obtain the original characteristic information of the beat measurement, and to provide basic data for subsequent error compensation.

[0272] The specific implementation of step S02 is to collect the temperature data output by the assembly workbench temperature sensor, establish a temperature and measurement system thermal drift relationship model, calculate the thermal drift error value according to the temperature data, subtract the thermal drift error value from the original beat value, and obtain the beat value after temperature compensation. This step first collects the workbench temperature through a PT100 type platinum resistance temperature sensor, with a sampling period of 100 ms. After collecting 30 data points, the average value is taken as the current temperature value. At the same time, the displacement sensor shell temperature, the environment temperature and the base temperature are collected, and a multi-point temperature field distribution map is constructed. Based on the multiple linear regression method, a temperature and measurement system thermal drift relationship model is established, and the model form is ΔT(t) = a·T workbench (t) + b·T semspr (t) + c·T environment (t) + d·T base (t) + e + ε T , where ΔT(t) is the thermal drift at time t, with the unit of millimeter; T workbench (t) is the workbench temperature at time t, with the unit of Celsius; T sensor (t) is the displacement sensor shell temperature at time t, with the unit of Celsius; T environment (t) is the environment temperature at time t, with the unit of Celsius; T base (t) is the base temperature at time t, with the unit of Celsius; a, b, c, d, e are regression coefficients; ε T is the residual error term of the temperature compensation model, with the unit of millimeter, and the range is ±0.0005 millimeter. The regression coefficients are solved by the least square method, that is, , where X is the temperature data matrix, and Y is the thermal drift vector. In actual application, the calibration temperature range is 15-35℃, with a step of 5℃. For the obtained thermal drift model, the cross-validation method is used to verify its effectiveness, and the determination coefficient R 2 of the verification data set needs to be greater than 0.95 before application, where The temperature compensation adopts a real-time calculation method, that is, the temperature data is collected before each measurement, the thermal drift error value under the current temperature condition is calculated, and then the error value is subtracted from the original beat value, that is, R temp (i) = R raw (i) - ΔT(t i). For the case where the ambient temperature rate of change exceeds 1 °C / min, the temperature collection period will be shortened to 10 ms and the thermal drift coefficient will be recalculated. This step aims to eliminate the influence of ambient temperature changes on the measurement system and improve the consistency of the measurement system in different temperature environments.

[0273] The specific implementation of step S03 is to obtain the installation angle data of the displacement sensor, calculate the cosine error caused by the installation angle deviation of the displacement sensor, angle-compensate the temperature-compensated runout value, and obtain the angle-compensated runout value. This step first measures the installation angle of the displacement sensor in real time through a high-precision angle encoder. The resolution of the encoder is 0.01°, and the repeatability is ±0.005°. At the same time, auxiliary detection of the sensor installation angle is performed based on a machine vision method. A sub-pixel edge detection algorithm is used to determine the direction of the sensor axis. The image resolution is 1280x1024 pixels, the lens focal length is 25 mm, and the working distance is 200 mm. The measurement results of the encoder and the machine vision detection results are fused to obtain the final installation angle value by using a weighted average method, i.e. where θ is the final determined installation angle, in degrees; θ encoder is the angle value measured by the angle encoder, in degrees; θ vision is the angle value measured by the machine vision method, in degrees; w1 and w2 are weight coefficients, and typical values are 0.7 and 0.3, respectively. Based on the measured installation angle, the cosine error is calculated, i.e. ΔL = L · (1-cosθ), where ΔL is the cosine error, in millimeters; L is the actual displacement, in millimeters; and θ is the included angle between the sensor axis and the normal of the measured surface, in degrees. When the included angle is less than 5°, a small-angle approximation is used for simplified calculation, i.e. where θ is in radians. The angle-compensated runout value is calculated as For different measurement positions, an angle compensation lookup table is established, and the lookup table is calculated by using a cubic spline interpolation method. When the installation angle changes by more than 0.1°, the system will automatically recalculate the compensation value. The purpose of this step is to eliminate the system error caused by the installation angle deviation of the displacement sensor and improve the measurement accuracy.

[0274] The specific implementation of step S04 is to calculate the radial displacement caused by centrifugal force according to the component rotation speed data measured by the central processor, calculate the radial deformation by using an elastic deformation mechanics model, subtract the radial displacement from the angle-compensated runout value, and obtain the runout value after mechanics compensation. This step first measures the rotation speed of the component by using a Hall effect speed sensor. The sampling frequency is 10 kHz, and the signal is processed by a low-pass filter with a cutoff frequency of 500 Hz. Based on the measured rotation speed, the angular velocity Where ω is the angular velocity, unit rad / s; n is the rotation speed, unit rpm. According to the mass distribution characteristics of the component, a centrifugal force calculation model is established, and the centrifugal force calculation formula is F(t) = m(r) r ω 2 , where F(t) is the centrifugal force at a distance r from the rotation center, unit Newton; m(t) is the mass at a distance r from the rotation center, unit kilogram; r is the distance to the rotation center, unit meter. The finite element method is applied to construct an elastic deformation mechanics model, and hexahedral elements are used for meshing with an element size of 0.5 mm. The control equation of elastic deformation is σ = C: ε; , where is the divergence operator; σ is the stress tensor, unit Pascal; F is the volume force vector; C is the fourth-order elastic tensor; ε is the strain tensor; u is the displacement vector. Considering the influence of temperature on the elastic modulus, the correction is made by E(T) = E0[1-α E ·(T-T0)], where E(T) is the elastic modulus at temperature T; E0is the elastic modulus at reference temperature T0; α E is the elastic modulus temperature coefficient, and the typical value is 0.001-0.003 / degree Celsius. The discrete form of the model is K·U = F, where K is the stiffness matrix; U is the node displacement vector; F is the node load vector. The Newton-Raphson iteration method is used to solve the nonlinear equation set, and the iteration formula is U k+1 = U k -J(U k ) -1 ·R(U k ), and the convergence precision is set to 10 -6 . For the case with high calculation complexity, the modal reduction technique U ≈ Φ·q is used to reduce the calculation amount, and the first 20 modes are retained. Combined with the measured component support stiffness coefficient, the radial elastic deformation values of each measurement point under the action of centrifugal force are calculated. The runout value after mechanical compensation is calculated as R mech (i) = R angle (i)-Δu radial (i), where R mech (i) is the runout value after mechanical compensation; Δu radial (i) is the radial elastic deformation value. When the rotation speed changes by more than 5%, the system will recalculate the deformation value. This step aims to eliminate the influence of the radial deformation of the component caused by the centrifugal force generated by high-speed rotation on the measurement results, and is particularly suitable for precision measurement under high-speed rotation conditions.

[0275] The specific implementation of step S05 is to collect vibration data output by the environmental vibration sensor in the control system, identify the environmental vibration frequency component through spectrum analysis of the vibration data by a spectrum analyzer, and construct an adaptive notch filter optimization function to filter out the influence of environmental vibration on the beat measurement. This step first collects environmental vibration signals through a piezoelectric acceleration sensor, with a sensor sensitivity of 100 mV / g and a frequency response range of 1-5000 Hz. The sampling frequency is set to 10 kHz, and 2 seconds of data are collected each time. The collected vibration data is preprocessed using the Hanning window function w(n) = 0.5·[1-cos(2πn / (N-1))] to reduce the spectral leakage effect. Then, the fast Fourier transform algorithm is used for spectrum analysis, and the transform formula is where X(k) is the frequency domain sequence; x(n) is the time domain sequence; N is the number of transform points, which is 8192; and j is the imaginary unit. The main vibration frequency component is identified through a peak detection algorithm, and the judgment condition is |X(k)|>|X(k-1)| and |X(k)|>|X(k+1)| and |X(k)|>α·max(|X|), where α is an energy ratio threshold, and is taken as 0.03. Based on the identified environmental vibration characteristics, an adaptive notch filter is constructed, and the transfer function is where ω0 is the center frequency, r is the pole radius, the control bandwidth, and the typical value is 0.9-0.99. For multiple vibration frequencies, a cascade structure is used The filter coefficient optimization uses the least mean square error criterion, and the objective function is J(θ) = E[e 2 (n)], which is solved iteratively by the stochastic gradient descent method where μ k is the learning rate, the initial value is 0.05, and it decays with the number of iterations. The convergence criterion is that the coefficient change rate of 10 consecutive iterations is less than 0.1%. The filter depth is adjusted adaptively for different vibration intensities, with a stopband attenuation of 40 dB in strong vibration areas and 20 dB in weak vibration areas. The beat value after vibration filtering is calculated as where h(n) is the filter impulse response. The purpose of this step is to identify and filter out the interference of environmental vibration on the beat measurement, and to improve the signal-to-noise ratio of the measurement signal.

[0276] The specific implementation of step S06 is to calculate the measurement error caused by the nonlinear response of the displacement sensor according to the calibration curve of the displacement sensor, linearize the beat value after vibration filtering, and obtain the corrected beat value. This step first obtains the calibration data of the displacement sensor through a precision displacement stage, with a displacement stage resolution of 0.1 μm and a repeat positioning accuracy of ±0.3 μm. The calibration points are spaced at 0.05 mm, covering the entire measurement range of 0-2 mm. The calibration data is fitted to a quintic polynomial model V = a0 + a1·d + a2·d 2+a3·d 3 +a4·d 4 +a5·d 5 , where V is the sensor output voltage in volts; d is the actual displacement in millimeters; a0, a1, a2, a3, a4, a5 are polynomial coefficients. The fitting method uses the least squares method, and the coefficients are solved by The root mean square error of the fitting residual is controlled within 0.5μm. Based on the fitted polynomial model, the actual response curve of the sensor is constructed, and the correction function is obtained by solving the inverse function. Considering the complexity of solving the inverse function, the piecewise linear interpolation method is used to achieve fast table lookup. The table lookup formula is Where V i and V i+1 are the two voltage values ​​adjacent to V in the interpolation table, d i and d i+1 is the corresponding displacement value. The interpolation table interval is 0.01V. For the boundary cases not covered by the interpolation table, the extrapolation algorithm is used for processing, and the extrapolation length does not exceed 5% of the measurement range. Considering the thermal characteristics of the sensor, the temperature correction coefficient K is introduced T (T)=1+β T ·(TT ref ), where β T is the temperature coefficient of the sensor, with a typical value of 0.0001 to 0.0005 degrees Celsius. The corrected runout value is calculated as R linear (i) = R vib (i) K c (R vib (i)), where K c is the nonlinearity correction coefficient. When the ambient temperature changes by more than ±2°C, the corresponding correction coefficient is reselected. This step aims to eliminate the impact of the displacement sensor's inherent nonlinearity on measurement accuracy, achieving high-precision displacement measurement.

[0277] The specific implementation of step S07 is to use the Kalman filter algorithm to fuse multiple measurement results, model the runout measurement problem as a least squares estimation problem, and solve the optimal estimate by constructing an overdetermined set of equations. This step first models the runout measurement as a linear system, and the state equation is x k =A·x k-1 +w k-1 , the observation equation is z k =H·x k +v k , where x k is the state vector at time k, including the real jump value and its rate of change; A is the state transfer matrix, w k-1 is the process noise; z kis the observation vector; H is the observation matrix, which is taken as H = [1 0]; v k is the observation noise. The prediction step of Kalman filter calculates the prior state estimation and the prior error covariance P k|k-1 = A · P k-1|k-1 · A T + Q. The update step calculates the Kalman gain K k = P k|k-1 · H T · (H · P k|k-1 · H T + R) -1 , the posterior state estimation and the posterior error covariance P k|k = (I - K k · H) · P k|k-1 . An adaptive mechanism is adopted to dynamically adjust the noise covariance matrix, and the real-time noise level is estimated based on the statistical characteristics of innovation sequence, i.e. where λ is the forgetting factor, and a typical value is 0.95. The convergence criterion of the filter is that the trace of the posterior error covariance matrix is less than a preset threshold 10 -5 . To improve the estimation accuracy, the least squares method is used to construct an overdetermined equation set A · x = b, which is solved by the singular value decomposition method. A regularization term is introduced to suppress the influence of noise, and the modified solution is x = V · (∑ T · ∑ + α · I) -1 · ∑ T · U T · b, where α is the regularization parameter, and a typical value is 10 -3 . The final bounce value is calculated in a weighted fusion manner R final (i) = α · R kalman (i) + (1 - α) · R ls (i), and the weight coefficient is determined based on the respective estimation uncertainty. The purpose of this step is to improve the accuracy and reliability of the bounce measurement and reduce the influence of random errors through data fusion technology.

[0278] The specific implementation of step S08 is to compare the historical measurement data, identify the long-term stability change trend of the system, and apply a trend compensation algorithm to eliminate the measurement drift caused by system aging. This step first establishes a measurement data history library to record the measurement results of the standard sample by the system at different time points. The standard sample is a standard block certified by metrological authentication, with a nominal value of 1.000 mm and an uncertainty of ±0.1 μm. The system automatically performs the standard sample measurement process every 100 working hours, and records the measurement value and environmental parameters. Through time series analysis method, the long-term change trend of the measurement value is extracted, and the exponential weighted moving average algorithm is used to smooth the short-term fluctuations, and the calculation formula is S t = β · Yt + (1 - β) · S t-1 , where S t is the smoothed value at time t; Y t is the observed value at time t; and β is the smoothing factor, which is set to 0.3. In combination with control chart technique, a Shewhart control chart is plotted for the system measurement values, with control limits set to ±3σ. When the measurement values consecutively exceed the ±2σ warning limits for 3 times or exceed the ±3σ control limits for once, the system will trigger a calibration reminder. Through regression analysis, a mathematical model is identified for the system drift, common models include linear drift model d linear (t) = d0+ k · t and exponential drift model d exp (t) = d0+ a · (1 - e -b·t ), where d0is the initial drift; k is the linear drift rate; a is the drift amplitude; and b is the exponential parameter. Model selection is based on goodness-of-fit comparison, with the coefficient of determination R 2 as the evaluation criterion. The runout value after trend compensation is calculated as R trend (i) = R final (i) - d(t i ), where d(t i ) is the drift at time t i . For abnormal drift beyond expectation, the system will issue a maintenance warning. The purpose of this step is to monitor and compensate for long-term stability changes of the measurement system, to extend the calibration period and to ensure long-term measurement accuracy.

[0279] The specific implementation of step S09 is to output the final runout value and measurement uncertainty, and transmit the final runout value and measurement uncertainty to the central processor for display and storage. This step first calculates the combined standard uncertainty of the measurement result according to the GUM uncertainty evaluation method. The uncertainty sources include sensor repeatability uncertainty u1, estimated value 0.0005 mm; calibration curve fitting uncertainty u2, estimated value 0.0003 mm; temperature compensation residual uncertainty u3, estimated value 0.0002 mm; angle compensation residual uncertainty u4, estimated value 0.0001 mm; mechanical compensation residual uncertainty u5, estimated value 0.0002 mm; environmental vibration residual uncertainty u6, estimated value 0.0003 mm; data fusion algorithm uncertainty u7, estimated value 0.0001 mm; long-term stability uncertainty u8, estimated value 0.0004 mm. The combined standard uncertainty calculation formula is where c i and c j are the sensitivity coefficients; and r ij is the correlation coefficient. For independent uncertainty components, it is simplified as The expanded uncertainty is calculated as U = k · u cwhere k is a coverage factor, taking a value of 2, corresponding to a confidence level of about 95%. The final run-out value and its expanded uncertainty are output in the format of "measured value ± expanded uncertainty", with significant figures retained to 0.0001 mm. The data are transmitted to the central processor through an industrial communication protocol, using the Modbus RTU protocol, with a baud rate of 115200 bit / s, 8-bit data, 1-bit stop, and no parity. After receiving the data, the central processor performs data checking to confirm data integrity. The measurement results are displayed in real time on the human-machine interface, and are stored in a local database in SQLite format, recording information such as measurement time, measured value, uncertainty, and environmental parameters. According to a preset threshold value, it is determined whether the measurement result is qualified, with the determination criterion being |R trend |≤R threshold where R threshold is a preset run-out threshold value, taking a value of 0.01 mm. When the value exceeds the value, an audible and visual alarm is triggered. The purpose of this step is to provide reliable measurement result output and complete measurement uncertainty information, providing a scientific basis for subsequent quality judgment and decision-making.

[0280] In order to better understand and implement the present application, the following provides an embodiment 2 of a specific application scenario of the present application: In an aero-engine research institute, researchers put forward higher requirements for the assembly precision of the center through-hole components of the aero-engine rotor assembly, especially the coaxiality after the assembly of the center through-hole components and the shaft parts. The researchers use the center through-hole component lifting assembly and run-out measuring device of the present application to accurately control and measure the assembly process of the turbine disc and the main shaft.

[0281] In this embodiment, the turbine disc is a typical center through-hole component, with an inner diameter of 125.000±0.005 mm, an outer diameter of 430 mm, a thickness of 42 mm, a material of GH4169 high-temperature alloy, a working temperature of up to 650℃, and a weight of about 23.5 kg. The main shaft is a shaft part, with an outer diameter of 124.995±0.003 mm, and an interference fit with the turbine disc. The assembly run-out requirement of such high-precision components is not more than 0.005 mm, and the traditional assembly method is difficult to guarantee this precision requirement.

[0282] The researchers use the device of the present application to assemble and measure, and the specific operation parameters are shown in Table 1:

[0283] Table 1 Center through-hole component lifting assembly and run-out measuring device operation parameter table

[0284] Operating parameters Parameter value Assembly force 18.5 kN Assembly speed 0.5 mm / s Lift slide position accuracy 0.008 mm Displacement sensor measurement accuracy 0.0008 mm Ambient temperature 22.5±0.5℃ Jump measurement rotational speed 60 r / min Data sampling rate 1000 Hz Sampling time 5s

[0285] The positioning pins of the assembly workbench have a diameter of 10 mm and are evenly distributed on the surface of the workbench, providing accurate radial positioning function. The clamping device uses pneumatic clamps, with a clamping force of 240 N, ensuring that the turbine disc does not displace during measurement.

[0286] During the runout measurement process, the error reduction module plays a key role. First, the measurement data is processed by the sliding window method, and the original runout value is calculated. In a typical measurement, the statistical characteristics obtained are shown in Table 2:

[0287] Table 2 Runout data statistical characteristics table

[0288]

[0289]

[0290] Through the temperature compensation step, the measurement deviation caused by environmental temperature fluctuations is eliminated. In actual operation, the measured temperature distribution is shown in Table 3:

[0291] Table 3 Temperature distribution data table

[0292] Measurement position Temperature value (°C) Workbench temperature 22.4 Sensor housing temperature 23.6 Ambient temperature 22.3 Base temperature 22.0

[0293] The regression coefficients obtained by the temperature and measurement system thermal drift relationship model based on the multiple linear regression method are: a = 0.00043 mm / °C, b = 0.00058 mm / °C, c = 0.00021 mm / °C, d = 0.00016 mm / °C, e = -0.02842 mm. The determination coefficient R 2 of this model is 0.968, which meets the application requirements.

[0294] The installation angle measurement of the displacement sensor uses a high-precision angle encoder and a machine vision double verification method, and the measured installation angle is 0.83°, and the cosine error is calculated to be 2.07 x 10 -5 mm. The runout value after angle compensation is reduced by about 1.5% compared to the original value.

[0295] At a speed of 60 r / min, the radial displacement caused by centrifugal force is very small, but it is still accurately calculated and compensated through the elastic deformation mechanics model. The calculated radial displacement of each position is shown in Table 4:

[0296] Table 4 Radial displacement table caused by centrifugal force

[0297] Radial position (mm) Displacement amount (pm) 125 0.082 200 0.196 300 0.485 400 0.862

[0298] Through frequency spectrum analysis, the main frequency components of environmental vibration are identified, as shown in Table 5:

[0299] Table 5 Environmental vibration frequency analysis table

[0300] Frequency (Hz) Amplitude (mm) Relative energy ratio (%) 23.4 0.0008 42.3 47.2 0.0005 28.7 82.5 0.0003 18.2 110.8 0.0002 10.8

[0301] An adaptive notch filter is constructed for these frequencies, with a pole radius of 0.97, and the signal-to-noise ratio is improved by about 8.4 dB after filtering. The nonlinear response correction of the displacement sensor uses a quintic polynomial model, with a root mean square error of fitting residual of 0.28 μm, realizing high-precision displacement measurement.

[0302] The key parameters used in the Kalman filter fusion algorithm are shown in Table 6:

[0303] Table 6 Kalman filter parameter table

[0304]

[0305] The drift trend of the long-term stability of the system is represented by an exponential drift model, with parameters d0 = 0.0001 mm, a = 0.0023 mm, and b = 0.0075 h -1 The final comprehensive uncertainty evaluation results are shown in Table 7:

[0306] Table 7 Uncertainty evaluation table

[0307] Uncertainty source Standard uncertainty (mm) Sensitivity coefficient Sensor repeatability 0.0005 1.0 Calibration curve fitting 0.0003 0.9 Temperature compensation residual 0.0002 1.2 Angle compensation residual 0.0001 0.8 Mechanical compensation residual 0.0002 1.1 Environmental vibration residual 0.0003 1.3 Data fusion algorithm 0.0001 0.7 Long-term stability 0.0004 1.0

[0308] After multiple tests and verifications, the final runout measurement value obtained by the device is 0.0042 ± 0.0009 mm (including factor k = 2), which meets the assembly requirement of less than 0.005 mm. In 30 consecutive assembly tests, 100% meet the process requirements, which is significantly higher than the qualification rate of traditional methods.

[0309] Traditional aero-engine component assembly and runout measurement mainly rely on manual operation and simple mechanical sensors, which have the following problems: first, the assembly process lacks precise control, and the assembly force and speed mainly rely on the experience of operators; second, the runout measurement does not consider the influence of external factors such as environmental temperature and vibration, and the measurement results have large uncertainty; third, the measurement data are not systematically processed, and the result reliability is not high. These problems lead to low qualification rate of high-precision component assembly, and some assembly defects are only revealed in subsequent work processes, causing equipment performance degradation and even safety hazards.

[0310] Compared with the traditional method, the device has the following significant progress: first, the precise control of the assembly process is realized, and the stability and controllability of the assembly force and speed are ensured through the servo motor and precision screw pair; second, the multi-step error reduction module is adopted, and various error sources such as temperature drift, angle deviation, centrifugal force deformation and environmental vibration are systematically identified and compensated, which greatly improves the measurement accuracy; third, the data fusion technology and uncertainty evaluation method are adopted, which provides reliable quality assurance for the measurement results. Through these technical innovations, the assembly precision of the aero-engine center hole part is improved by about 65%, the uncertainty of the jump measurement is reduced by about 72%, and the product quality and reliability are significantly improved.

[0311] It should be noted that the variables involved in the present application are explained in detail as shown in Tables 8-10.

[0312] Table 8 Variable Explanation Table a

[0313]

[0314]

[0315] Table 9 Variable Explanation Table b

[0316]

[0317]

[0318] Table 10 Variable Explanation Table c

[0319]

[0320] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any skilled person in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A center bore component lift assembly and runout measuring device, characterized by, The device comprises a base, a column, a lifting slide, an assembly workbench, a measuring mechanism, a control system and a data acquisition system, wherein the base is used to support the whole device and ensure the levelness, the column is vertically fixed on the base and provides vertical guidance, the lifting slide is installed on the column and realizes vertical lifting movement, the assembly workbench is installed on the lifting slide and is used to place the center through-hole component, the measuring mechanism is located above the assembly workbench and is used to measure the run-out value of the center through-hole component, the control system is used to control the operation of the whole device, and the data acquisition system is used to acquire and process the run-out measurement data.

2. A center bore component lift and runout measuring device according to claim 1, wherein, The lifting slide and the column guide rail are in sliding fit, the lifting slide is internally provided with a ball screw pair, the ball screw pair screw rod is in sliding fit with the lifting slide, the ball screw pair nut is fixedly connected with the lifting slide, the lower end of the ball screw pair screw rod is supported and connected with the base through a bearing, the upper end of the screw rod is connected with a servo motor through a shaft coupling, and the servo motor is fixedly installed on the top of the lifting slide.

3. A center bore component lift and runout measuring device according to claim 2, wherein, The assembly workbench is installed on the lifting slide, the surface of the assembly workbench is made of hard alloy material, the workbench surface is provided with a positioning pin and a clamping device, the positioning pin is vertically fixed on the surface of the assembly workbench, and the clamping device comprises a cylinder and a clamping jaw, the cylinder is fixed on the assembly workbench, and the clamping jaw is connected with the piston rod of the cylinder.

4. A center bore component lift and runout measuring device according to claim 3, wherein, The measuring mechanism comprises a displacement sensor and a measuring support, the displacement sensor is fixed on the measuring support, and the measuring support is connected with the column through an adjusting mechanism, the adjusting mechanism comprises a horizontal adjusting slide block and a vertical adjusting slide block, the horizontal adjusting slide block is in sliding fit with the column, the vertical adjusting slide block is in sliding fit with the horizontal adjusting slide block, and the measuring support is fixed on the vertical adjusting slide block.

5. A center bore component lift and runout measuring device according to claim 4, wherein, The control system comprises a central processing unit, a man-machine interface and an input-output module, the central processing unit is connected with the man-machine interface through a data bus, and the central processing unit is electrically connected with the servo motor, the cylinder and the displacement sensor through the input-output module.

6. A center bore component lift and runout measuring device as described in claim 5, wherein, The data acquisition system comprises a signal conditioning circuit, an analog-digital converter, a digital filter, an error reduction module, a spectrum analyzer, a parameter optimizer and a physical model calculation unit, the signal conditioning circuit is electrically connected with the displacement sensor, the analog-digital converter is electrically connected with the signal conditioning circuit, the digital filter is electrically connected with the analog-digital converter, the spectrum analyzer is electrically connected with the digital filter, the parameter optimizer is electrically connected with the spectrum analyzer, the error reduction module is electrically connected with the digital filter and the parameter optimizer, the physical model calculation unit is electrically connected with the central processing unit, the error reduction module is electrically connected with the physical model calculation unit, and the error reduction module is electrically connected with the central processing unit.

7. The device for lifting, assembling and measuring runout of a central through hole component according to claim 6, characterized in that: The error reduction module is a single-chip microcomputer, a program instruction is stored in the memory of the single-chip microcomputer, the CPU of the single-chip microcomputer is used to execute the program instruction, and the program instruction is used to complete the error reduction step when executed.

8. A center bore component lift and runout measuring device according to claim 7, wherein, The error reduction step specifically includes: receiving filtered runout measurement data for time domain analysis; calculating thermal drift error values and compensating according to temperature data; angle compensation according to displacement sensor installation angle data; calculating radial displacement caused by centrifugal force using an elastic deformation mechanics model and performing mechanical compensation; collecting vibration data for frequency spectrum analysis, constructing an adaptive notch filter to filter out environmental vibration effects; linearization according to displacement sensor calibration curve; using Kalman filtering algorithm to fuse multiple measurement results, modeling the runout measurement problem as a least squares estimation problem; applying a trend compensation algorithm to eliminate measurement drift caused by system aging; outputting the final runout value and measurement uncertainty.

9. A center bore component lift and runout measuring device according to claim 8, wherein, The elastic deformation mechanics model is used to calculate the elastic deformation amount of the rotating part under the action of centrifugal force, and the input includes part mass distribution function, part rotation angular velocity, part material elastic modulus, part geometric size parameter and part support stiffness coefficient, and the output is the radial elastic deformation value of the rotating part at each measurement point.

10. A center bore component lift and runout measuring device according to claim 9, wherein, The adaptive notch filter optimization function is used to determine the optimal filter parameters to maximize the elimination of environmental vibration interference, and the input includes environmental vibration frequency spectrum distribution function, signal-to-noise ratio evaluation coefficient, filter order parameter, frequency band width parameter and convergence rate parameter, and the output is the optimized notch filter coefficient set.

Images