Dynamic maintenance management method and system based on full life cycle of test equipment

Abstract

The invention relates to the technical field of industrial equipment control, and discloses a dynamic maintenance management method and system based on the full life cycle of test equipment, and the method comprises the steps: obtaining a time domain vibration signal and torque current data during the operation of the equipment through a high-frequency vibration sensor and a current loop feedback interface; calculating an accumulated damage gradient parameter representing the physical recession degree of the mechanical transmission chain; the parameters are input into a preset damage stress coupling model to solve the maximum radial impact force allowed to be borne by the damaged part, and the impact force is mapped into a dynamic capability boundary for carrying out logic constraint on the motion planning instruction; and the motion controller reconstructs an ideal speed curve by using the boundary in an interpolation period, and reshapes a torque load spectrum output to a servo motor by reducing the jump of acceleration and deceleration sections. The problem that an existing static control strategy is mismatched with the dynamic aging characteristic of equipment is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to a dynamic maintenance management method and system based on the whole life cycle of test equipment, belonging to the technical field of industrial equipment control. BACKGROUND

[0002] In current industrial manufacturing and precision testing scenarios, servo drive systems are used to achieve high-precision position, speed and torque control. Such systems usually use model-based or specific gain parameter motion control strategies. Control parameters such as position loop gain, speed feedforward coefficient and jerk limit value are set based on the mechanical impedance and frequency response characteristics of the new machine during the initial debugging stage of the equipment, and remain constant during subsequent operation. As the equipment runs over time, mechanical transmission components wear out, clearances increase, or lubrication conditions change, causing the physical model parameters of the controlled object to change nonlinearly and time-varyingly. If the control system still uses the initial set high dynamic control instructions to drive the physically degraded mechanical structure, the high-frequency response torque output will be converted into impact load on the damaged components, accelerating the expansion of micro-cracks or causing mechanical vibration, which will affect the stability of the equipment during the whole life cycle.

[0003] Existing technical means focus on maintenance scheduling or after-service repair, and there are shortcomings in real-time control intervention of software and hardware combination. For example, the maintenance management method and system disclosed in Chinese patent CN1607532A determines the maintenance work content by comparing the running time, environmental temperature and humidity, and image quality parameters of the medical equipment with the preset reference, and notifies the maintenance personnel. Although this scheme improves the generation of maintenance plans by digital means, the core logic still remains in the passive discrete mode of state monitoring, threshold triggering and manual intervention. The monitoring data is only used to generate external repair work orders or spare parts procurement lists, which are disconnected from the bottom servo drive loop of the equipment. Therefore, during the long running period before the fault alarm is triggered or between two regular maintenance, even if early micro-damage occurs in the mechanical transmission chain, the controller will still mechanically execute the high dynamic instructions set at the factory, accelerating the crack expansion and precision loss in the damaged full-speed running condition, and cannot achieve active life extension from the physical source.

[0004] Therefore, how to establish real-time perception of the degree of physical degradation of the equipment, and accordingly adaptively adjust the bottom control parameters and motion planning strategy to realize the transition from passive monitoring to active life control closed-loop regulation mechanism, becomes a technical problem to be solved by the present application. SUMMARY

[0005] To solve the problems raised in the background art, the technical solution of the present application is as follows: a dynamic maintenance management method based on the whole life cycle of test equipment, applied to an industrial equipment control system comprising a servo drive unit and a motion controller, the method comprising the following steps: In step S100, time-domain vibration signals and torque current data during equipment operation are obtained through a high-frequency vibration sensor and a current loop feedback interface of a servo drive unit, wavelet packet transformation is performed to extract energy features of specific frequency bands, and a cumulative damage gradient parameter representing a physical degradation degree of a mechanical transmission chain is calculated according to the energy features; In step S200, the cumulative damage gradient parameter is input into a preset damage stress coupling model, the maximum radial impact force allowed by a damaged component under a current working condition is calculated, and the maximum radial impact force is mapped to a dynamic capability boundary for logically constraining a motion planning instruction, the dynamic capability boundary at least including a maximum jerk limit value and a position loop gain correction coefficient. In step S300, a motion controller reconstructs an ideal speed curve generated by original machining codes by applying the dynamic capability boundary to the ideal speed curve in an interpolation period, reshapes a torque load spectrum output to a servo motor by reducing a jump degree of an acceleration / deceleration segment, and monitors a following error of the servo shaft in real time, and when the following error converges within a preset process precision threshold after reshaping, a current reshaping control strategy is maintained.

[0006] Preferably, the specific process of calculating the cumulative damage gradient parameter in step S100 includes: in step S201, the obtained time-domain vibration signals are divided into continuous time windows, and the data in each time window is subjected to detrend item processing; in step S202, a frequency spectrum energy proportion of the signal subjected to the detrend item processing at a mechanical natural frequency and a multiple frequency thereof is calculated, the frequency spectrum energy proportion is subjected to differential operation with a frequency spectrum energy distribution under a baseline health state to generate a differential feature vector; and in step S203, the differential feature vector is input into a pre-trained long short-term memory network model, a dimensionless scalar reflecting a micro crack propagation rate is output, the dimensionless scalar is defined as the cumulative damage gradient parameter, and the cumulative damage gradient parameter is transmitted to step S200.

[0007] Preferably, the process of mapping the maximum radial impact force to the maximum jerk limit value in step S200 follows a damage negative feedback control law as follows: wherein, is the calculated maximum jerk limit value of a current interpolation period, is a rated jerk value set by the equipment at factory, is the cumulative damage gradient parameter calculated in step S100, is a preset damage sensitivity coefficient, is a position following error value of a previous control period, is a preset error suppression weight factor; and the motion controller reduces a dynamic response rate of an output instruction in real time according to the damage negative feedback control law, so as to ensure that mechanical stress does not exceed the dynamic capability boundary.

[0008] Preferably, the specific way of reconstructing the ideal velocity curve generated by the original machining code in step S300 includes: step S401, parsing the linear or circular arc interpolation instructions in the original machining code to generate a velocity planning curve; step S402, reading the maximum jerk limit value determined in step S200 at the acceleration and deceleration stage of the velocity planning curve; step S403, when the original jerk demand of the velocity planning curve exceeds the maximum jerk limit value, keeping the target velocity unchanged, reducing the jerk peak value by prolonging the time length of the acceleration and deceleration stage, and generating a smoothed actual execution velocity curve; step S404, discretizing the actual execution velocity curve into position control instructions, and sending the position control instructions to the position loop of the servo driver.

[0009] Preferably, the dynamic capability boundary further includes a torque feedforward compensation coefficient; step S300 further includes a nonlinear compensation step for mechanical clearance: step S501, when the cumulative damage gradient parameter indicates that there is a reverse clearance in the transmission chain, the system superimposes a pulse torque consistent with the motion direction of the servo motor at the zero crossing point of the servo motor speed according to the torque feedforward compensation coefficient; step S502, the amplitude of the pulse torque is positively correlated with the cumulative damage gradient parameter, and the duration of the pulse torque is determined by the response bandwidth of the current loop of the servo drive unit, so as to offset the motor idle hysteresis caused by mechanical wear clearance.

[0010] Preferably, after monitoring the follow-up error of the servo shaft in step S300, a secondary derating convergence step is further included: step S601, if the follow-up error still exceeds the preset process accuracy threshold for consecutive N interpolation periods after applying the dynamic capability boundary, a secondary derating strategy is triggered; step S602, the secondary derating strategy is executed, the maximum operating speed of the servo motor is gradually reduced by a preset step, until the follow-up error re-converges to within the preset process accuracy threshold or the current operating speed is lower than the minimum speed threshold required to maintain the production process; step S603, when the current operating speed is lower than the minimum speed threshold, a maintenance instruction containing the type of component to be replaced and the fault position information is generated.

[0011] Preferably, the method further includes a model iteration step based on edge computing: step S701, after each device performs physical maintenance and replaces components, record the initial vibration signal and current data after replacement; step S702, use the initial vibration signal and current data as new baseline health state data to update the baseline parameters in the damage stress coupling model; step S703, calculate the change rate of the cumulative damage gradient parameters before and after maintenance, and correct the mapping weight of the dynamic capability boundary accordingly, so that the response characteristics of the control system adapt to the actual mechanical impedance after replacing the components.

[0012] Preferably, the determination logic of the position loop gain correction coefficient in step S200 comprises: step S801, monitoring the current noise level of the servo motor in the steady state operation stage; step S802, when the current noise level exceeds the preset noise threshold, determining that the system has the risk of resonance caused by the decrease of mechanical stiffness; step S803, calculating the critical proportional gain value capable of suppressing resonance, and setting the position loop gain correction coefficient as the attenuation ratio that makes the current position loop gain lower than the critical proportional gain value, the method is applied to the control system with multi-axis cooperation function, and the reconstruction in step S300 further comprises multi-axis synchronization constraint processing: step S901, when a certain axis prolongs the acceleration / deceleration time due to the application of dynamic capability boundary, the motion state of other axes related to the axis is obtained; step S902, the motion planning parameters of the other axes are synchronously adjusted according to the prolonged acceleration / deceleration time of the axis, so that the synthesized trajectory of multi-axis interpolation motion is consistent with the preset trajectory.

[0013] Preferably, the method further comprises a load spectrum discretization storage step for specific working conditions: step S1001, dividing the whole life cycle of the equipment into a running-in period, a stable period and a degradation period; step S1002, in the degradation period, collecting and recording the actual torque load spectrum data shaped by step S300 according to different motion types; step S1003, establishing an association matrix between the motion type and the mechanical damage propagation rate, and when a motion type with a high damage propagation rate is identified, further superimposing a preset safety redundancy coefficient on the basis of the dynamic capability boundary.

[0014] A dynamic maintenance management system based on the whole life cycle of a test equipment, the system comprising: a data acquisition and feature extraction unit, a boundary calculation and decision unit, and an adaptive shaping control unit; The data acquisition and feature extraction unit is configured to acquire time-domain vibration signals and torque current data of the equipment during operation through a high-frequency vibration sensor and a current loop feedback interface, perform wavelet packet transform to extract energy features of specific frequency bands, and calculate a cumulative damage gradient parameter representing the physical degradation degree of a mechanical transmission chain according to the energy features; The boundary calculation and decision unit is configured to input the cumulative damage gradient parameter into a preset damage stress coupling model, calculate the maximum radial impact force that a damaged component can withstand under the current working condition, and map the maximum radial impact force to a dynamic capability boundary for logically constraining the motion planning instruction, wherein the dynamic capability boundary at least comprises a maximum jerk limit value and a position loop gain correction coefficient; Adaptive shaping control unit integrated in motion controller, used to reconstruct ideal speed curve generated by original machining code with dynamic capability boundary in interpolation period, reshape torque load spectrum output to servo motor by reducing the jump of acceleration and deceleration section, and monitor servo shaft following error in real time, maintain the current shaping control strategy when the following error converges within the preset process accuracy threshold after shaping.

[0015] Compared with the prior art, the beneficial effects of the present application are: 1. In the dynamic maintenance of the test equipment throughout its life cycle, a dynamic mapping mechanism of the physical state of the equipment and the control parameters is constructed, solving the problem of accelerated wear caused by the mismatch between the static control strategy and the dynamic aging characteristics of the equipment in the prior art, and limiting the instantaneous impact load on the worn parts by adaptively reshaping the motion planning curve of the servo driver and the torque output characteristics according to the real-time calculation of the dynamic capability boundary; the load spectrum shaping control based on the cumulative damage gradient physically delays the mechanical crack propagation or the expansion of the clearance, without human intervention, and substantially extends the service life of the equipment by actively intervening with the control algorithm.

[0016] 2. Establishing a control parameter adaptive compensation loop for the nonlinear degradation of mechanical structure, overcoming the precision drift of machining or testing caused by physical wear in the later stage of the traditional equipment life cycle, and dynamically offsetting the mechanical error caused by bearing wear or transmission chain looseness by real-time correction of the position loop gain or torque feedforward parameter at the control level in microseconds; ensuring that the equipment is in a non-ideal mechanical state, and still maintaining the stability of the core process indicators through accurate adjustment of the electronic flow, to avoid batch quality fluctuations caused by equipment aging.

[0017] 3. Implementing a derating operation and non-stop intervention mode based on the health of the equipment, breaking through the rigid limitation of traditional operation and maintenance system between full-speed operation and shutdown maintenance, automatically transitioning to a low dynamic operation state that meets the current residual capacity of the equipment when early fault features are monitored, ensuring production continuity while preventing fault deterioration; a time-based flexible control strategy that extends the survival space of the equipment eliminates the risk of production line paralysis caused by sudden shutdown, and reserves a certain buffer window for orderly adjustment of production plans; by correlating actual maintenance data and historical control intervention records in a closed loop, a self-iterative fault feature library and a life prediction model are constructed, the multi-dimensional alignment of vibration spectrum changes before and after maintenance, replacement component model parameters, and control strategy suppression effects is performed, and the weight coefficients of the state evaluation algorithm are corrected; based on the assimilation mechanism of real physical feedback data, the prediction deviation of the theoretical model under complex working conditions is eliminated. BRIEF DESCRIPTION OF DRAWINGS

[0018] Fig. 1 The adaptive shaping control flowchart of the cumulative damage gradient of the present application; Fig. 2A torque load spectrum suppression comparison chart for application of dynamic capability boundary of the application; Fig. 3 A system function module interaction and model iteration closed loop schematic diagram of the application. DETAILED DESCRIPTION

[0019] The technical solutions of the application will be described in detail below with reference to the specific embodiments of the application. It should be understood that the specific embodiments described herein are only used to explain the application and do not limit the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the application.

[0020] The application provides a dynamic maintenance management method and system based on the whole life cycle of a test device, which comprises an Internet of Things data acquisition module, a cloud data processing module, a boundary calculation and decision unit, and an adaptive shaping control unit integrated in the bottom layer of a motion controller. The Internet of Things data acquisition module is deployed at key parts of the device, including a high-frequency vibration sensor and a current loop feedback interface of a servo drive unit. The cloud data processing module is responsible for running damage assessment algorithms. The boundary calculation and decision unit is responsible for calculating dynamic physical constraints. The adaptive shaping control unit is responsible for converting the constraints into servo control instructions at the bottom layer. In the data acquisition stage, the system acquires time-domain vibration signals in real time when the device is running through the high-frequency vibration sensor at a sampling frequency of no less than 2 kHz, and reads torque current data of the servo motor at a current loop update period through the communication interface of the servo drive unit, such as an EtherCAT bus. For the acquired time-domain vibration signals, the processor performs a wavelet packet transform algorithm, selects a db4 wavelet basis function to decompose the signals to the 3rd layer or higher, calculates the energy norm of each node reconstruction coefficient, extracts the energy features of a specific frequency band, and inputs the real-time calculated energy features and the frequency spectrum energy distribution under the baseline health state of the device into a pre-trained long short-term memory network model. The model is supervised and learned through the whole life cycle data, and outputs a dimensionless scalar normalized to the interval of 0 to 1, i.e. a cumulative damage gradient parameter , which quantitatively represents the current physical degradation degree of the mechanical transmission chain. The closer the value is to 1, the more serious the degradation is. The damage stress coupling model construction needs to be preformed destructive physical calibration: select samples of the same batch of transmission components and place them on a servo hydraulic test bench, apply a frequency to sweep frequency excitation force, record the critical radial force value when the surface microhardness of the component decreases , and the corresponding vibration energy features; map the vibration energy features to the cumulative damage gradient parameter , establish and Discrete set of mapping points; least square method is used to perform third order polynomial fitting on the set of points to generate the degradation function wherein is the structure stiffness degradation coefficient, which is written into the controller's non-volatile memory for real-time table lookup call through offline calibration.

[0021] The boundary calculation and decision unit calculates the maximum radial impact force that the damaged component can withstand in the current interpolation period according to the cumulative damage gradient parameter, and maps this physical constraint to the dynamic capability boundary of motion control, wherein the maximum jerk limit value is determined according to the following damage negative feedback control law: wherein, is the calculated maximum jerk limit value of the current interpolation period; is the rated jerk value set by the factory, for example, it is set to ; is the cumulative damage gradient parameter calculated above; is the preset damage sensitivity coefficient, whose value is determined through offline calibration experiment, and is usually set between 0.5 and 0.8, which is used to define the response degree of the control strategy to damage; is the absolute value of the position following error feedback by the servo driver in the last control period; is the preset error suppression weight factor; the adaptive shaping control unit in the motion controller reconstructs the ideal speed curve generated by the original machining code in the interpolation period after receiving the updated dynamic capability boundary, and when the jerk requested by the original instruction exceeds the calculated maximum jerk limit value , the controller keeps the target position and target speed unchanged, clamps the acceleration to , and correspondingly prolongs the time length of the acceleration and deceleration stage or increases the smoothing time constant of the S-shaped speed curve; for the reverse backlash caused by the wear of the mechanical transmission chain, the system executes a nonlinear compensation procedure, when the cumulative damage gradient parameter indicates that the transmission chain has reverse backlash and odd harmonic components appear in the vibration spectrum, the system obtains the basic compensation torque amplitude from the table lookup value of , at the zero crossing point of the servo motor speed command, the system superimposes a pulse torque consistent with the motion direction on the current loop given end, the amplitude of the pulse torque is set as the product of the basic compensation torque amplitude and the preset torque feedforward compensation coefficient, and its duration is determined by the response bandwidth of the current loop of the servo drive unit, for example, it is set to 62.5 microseconds to 187.5 microseconds, this mechanism makes the motor rotor accelerate in advance before the mechanical backlash is completely eliminated, offsetting the idle hysteresis.

[0022] And for the resonance risk caused by mechanical stiffness reduction, the system performs position loop gain adaptive correction, the system continuously monitors the torque current root mean square value and its high frequency noise component of the servo motor in the steady state operation stage, the current noise extraction and critical gain determination procedure is as follows: configure a fourth order Butterworth high pass filter, the cutoff frequency is set to be the current loop bandwidth times, filter out the fundamental torque component; perform length sliding window root mean square operation on the filtered residual, output real-time noise value ; servo static state injection amplitude rated current white noise sequence, identify open loop transfer function amplitude margin crossing through fast Fourier transform line frequency point , calculate critical gain , as the physical upper limit of gain adjustment, when the current noise level exceeds the preset noise threshold, for example, 5% of the rated current, it is determined that there is a resonance risk, the system calculates the critical proportional gain value that can suppress the resonance , and sets the position loop gain correction coefficient to a decay ratio of 0.7 to 0.9, and then adjusts the current position loop gain to the product of the critical proportional gain value and the decay ratio. The adjustment instruction is written to the control register of the servo driver in real time through the field bus; in the multi-axis linkage scene, the system performs synchronization constraint processing, when a certain axis causes its acceleration and deceleration time to be prolonged due to the application of dynamic capability boundary, the motion controller reads the motion parameters of other axes that have interpolation association with it, and synchronously adjusts the motion planning parameters of other axes based on the prolonged acceleration and deceleration time, to ensure that the multi-axis synthesized trajectory is consistent with the preset trajectory. In addition, the system monitors the servo axis following error after shaping in real time, if the following error still exceeds the preset process precision threshold, for example, 0.02mm, in continuous 50 interpolation periods, the system triggers the secondary derating strategy, reduces the maximum running speed by 5% each time, when the running speed is lower than the minimum speed threshold required to maintain the production process, the system generates a maintenance instruction containing the component model to be replaced and the fault position information.

[0023] Example 1: In the application scenario of a high-frequency reciprocating motion platform for fatigue testing of key automotive components, the servo drive system needs to drive the load to perform continuous reciprocating loading motion with a frequency of 5Hz and a stroke of 50mm. Under this condition, the ball screw and support bearing in the mechanical transmission chain are subjected to long-term alternating loads, resulting in physical degradation. The time-domain signal collected by the high-frequency vibration sensor shows a high-energy transient impact response at the moment of screw reversal. The torque current data fed back by the servo drive unit shows nonlinear spike fluctuations at the zero-crossing point. The frequency band energy characteristics extracted by the cloud data processing module through wavelet packet transform indicate that the energy ratio in the 120Hz to 150Hz frequency band has shifted positively relative to the baseline healthy state. Based on this difference feature vector, the long short-term memory network model calculates the cumulative damage gradient parameter characterizing the current physical degradation degree of the mechanical transmission chain. The value is 0.55, and the boundary calculation and decision-making unit then applies a preset damage negative feedback control law. The control parameters are dynamically reconstructed to resolve the physical coupling contradiction between constant high-dynamic commands and time-varying mechanical clearances, and a damage sensitivity coefficient is set. The error suppression weighting factor is 0.6. The value is 1.0, which is combined with the normalized position following error from the previous cycle. The system calculates the maximum jerk limit value for the current interpolation cycle. for The adaptive shaping control unit applies this constraint value to reconstruct the S-shaped velocity curve, smoothing out the peaks of the acceleration curve by extending the jump change time of the acceleration and deceleration segments.

[0024] After the multi-scale collaborative control strategy was adjusted, the amplitude of the commutation impact vibration monitored by the external sensor decreased by 60%, and the position following error of the servo axis was stably converged within the process accuracy threshold of 0.02mm. This process minimized the impact of mechanical structure damage on machining accuracy through the active adaptation of the control algorithm, so that the equipment could continue to maintain the predetermined production process indicators without replacing hardware components.

[0025] Example 2: This embodiment constructs a precision motion control test platform based on dual-axis linear motor drive, simulates the actual operating conditions of high-precision chip mounter, aims to reproduce the mechanical loss characteristics in the process of high-frequency start-stop and high-speed micro-motion, the core driving unit of the test platform includes two linear motors, the position feedback is provided by the grating ruler with a resolution of 10 nm, and the driver communicates with the upper motion controller in real time through the EtherCAT bus. In order to simulate the real industrial field environment, controllable external disturbance is introduced in the test: an eccentric mass exciter is installed on the mechanical base to generate background vibration noise with adjustable amplitude and frequency of 20-200 Hz, which simulates the ground microseismicity transmitted by adjacent heavy equipment in the workshop, and in the current sampling loop of the servo driver, a signal generator is used to inject a Gaussian white noise with a signal-to-noise ratio of 25 dB to simulate the influence of electromagnetic interference on the feedback signal.

[0026] Three core performance indicators are selected as evaluation criteria: root mean square error (RMSE) of position following error, peak value of mechanical structure vibration acceleration, and effective action cycle number per unit time, among which the position following error directly reflects the processing or testing accuracy of the device; the peak value of vibration acceleration represents the influence of mechanical impact on the service life of the component; and the effective action cycle number measures the production efficiency of the system. There is an inherent engineering constraint among the three indicators: pursuing extremely high precision often requires limiting dynamic response to reduce efficiency, and simply pursuing efficiency may lead to increased mechanical impact and deteriorated precision. The verification focus of this embodiment is to investigate whether the present solution can find a better dynamic balance point in the above-mentioned contradictions. Therefore, three comparison schemes are designed for the test. The control group A adopts the traditional fixed gain PID control strategy, all control parameters such as position loop gain , speed feedforward coefficient , maximum acceleration are set according to the new machine state when the device is shipped, and remain constant during the entire test process. The control group B adopts a passive maintenance strategy based only on vibration threshold alarm, that is, the control parameters remain unchanged, and the alarm is triggered when the vibration amplitude exceeds the preset safety threshold, and the inventive sample group completely deploys the dynamic load spectrum shaping control mechanism based on cumulative damage gradient, including damage negative feedback control law and nonlinear compensation logic for gap.

[0027] The test process was divided into three stages, corresponding to the break-in period, stabilization period, and degradation period of the equipment's entire life cycle. To accelerate the simulation of the degradation process, different degrees of pre-set damage were artificially introduced into the mechanical transmission chain: some balls were gradually removed from the linear guide slider to simulate increased clearance, and the thickness of the guide rail lubricating oil film was artificially reduced to simulate increased friction coefficient. The first stage (simulated break-in period): At this time, the mechanical structure was in a near-healthy state with only minor assembly errors. In this stage, the eccentric vibrator output low-amplitude background noise. All three schemes ran the same test trajectory: point-to-point movement with an amplitude of 10mm and a frequency of 10Hz. Data showed that the position following error RMSE of all three schemes remained within 2. The vibration peak values ​​were all below 0.5g at approximately m, and the cumulative damage gradient parameters calculated for the sample group of this invention were [data missing]. The value is approximately 0.05, indicating that the system basically maintains its rated parameters and does not trigger significant derating or compensation.

[0028] Phase Two (Simulating the transition from stable period to early degradation): 5% of the balls were removed from the guide rail slider, and a moderate amplitude background disturbance was output from the exciter. At this point, the RMSE of the tracking error monitored by control groups A and B increased to 5.8. m, and a significant current spike appears at the moment of motor commutation, indicating that mechanical backlash begins to affect control stability. In contrast, the sample group of this invention... Once the value climbs to 0.35, the system detects this change and automatically fine-tunes the maximum jerk limit based on the damage negative feedback control law. Remove it from the rated Reduced to approximately Although the cycle time of a single action is slightly extended by 1.5ms, the jump in motor output torque is reduced. Actual measurement data shows that the RMSE (Real-Time Error Sequence) of the sample group in this invention is controlled within 3.2. m, the peak vibration acceleration was reduced by about 25% compared with the control group.

[0029] Phase 3 (Simulated Decay Period): An additional 15% of the balls were removed, increasing the mechanical clearance, and the background noise was increased to its highest level. At this point, the following error of control groups A and B deteriorated sharply to 18.5. m, exceeding 15 The process parameters were within the allowable range of m, and the peak vibration acceleration frequently exceeded the safety limit of 2.0g, causing the control group B to frequently trigger alarm shutdowns and significantly reduce the number of effective operation cycles. In contrast, the sample group of this invention... When the value reaches 0.72, the system fully activates the deep derating and nonlinear compensation strategies, with maximum jerk. Further restrictions were imposed on Simultaneously, torque feedforward compensation is initiated, and Table 1 details the comparison of key performance data for the three schemes in this stage.

[0030] Table 1: Comparison of Key Performance Data During the Decline Phase (Simulated) Data analysis shows that, under severe degradation conditions, although the prototype of this invention actively reduced the operating frequency by about 10% (from the theoretical 600 times / min to 540 times / min) to adapt to the mechanical capability boundary, it successfully suppressed the position following error to 9.4. Within the acceptable range of m, and with a vibration peak of only 1.15g, no destructive resonance was triggered. This result confirms that by sacrificing a small amount of instantaneous dynamic response speed, continuous and stable processing capability under non-ideal mechanical conditions can be obtained, avoiding the loss of the entire production capacity caused by sudden failure shutdown. In stark contrast, control group A suffered batch scrap due to its inability to adapt to changes in physical properties, while control group B lost its production value due to frequent passive shutdowns. Further long-term operation tests showed that after the same mechanical damage was introduced, the sample of this invention could run continuously and stably for more than 72 hours without manual intervention, while control group A was forced to shut down after about 4 hours of operation due to overheating of the guide rail (caused by high-frequency vibration friction).

[0031] Example 3: This example combines Figs. 1 to 3 This document describes a dynamic maintenance management method and system based on the entire lifecycle of testing equipment. Fig. 1 As shown, time-domain vibration signals and servo drive torque and current data during equipment operation are acquired through a high-frequency vibration sensor and a current loop feedback interface, respectively. The data acquisition and feature extraction unit performs wavelet packet transform on the signals to extract specific frequency band energy features and calculates the cumulative damage gradient parameter characterizing the physical degradation degree of the mechanical transmission chain. The parameter is transmitted to the boundary calculation and decision unit, which calculates the maximum radial impact force through the damage stress coupling model and generates a dynamic capability boundary that includes the maximum jerk limit and the position loop gain correction coefficient. On the one hand, this boundary is called by the adaptive shaping control unit to reconstruct the ideal velocity curve and reduce the jump, thereby driving the servo drive unit to output the shaped torque load spectrum and nonlinear compensation. At the same time, the system performs real-time tracking error monitoring. If the error converges to the process accuracy threshold, the current strategy is maintained. If it does not converge, a secondary derating strategy is triggered and feedback adjustment is performed. On the other hand, the system includes a model iteration path based on edge computing, which is used to correct the dynamic capability boundary mapping weight and update the parameters.

[0032] like Fig. 2As shown in the figure, the torque load spectrum under different states is shown, the horizontal axis represents time in seconds, and the vertical axis represents torque value percentage, wherein the dashed line represents the rated torque load spectrum of the new machine state, showing a higher peak torque requirement, the dotted line represents the unadjusted torque of the recession period, showing the original response of the device in the aging state, and the solid line represents the torque adjusted by the present application, which can be clearly seen that the torque amplitude at the peak is reduced by smoothing processing; as Fig. 3 As shown, the system logic architecture includes high-frequency vibration sensors and servo drive units as input sources, respectively used to obtain time-domain vibration signals and torque current data, which flow to the calculation module to perform wavelet packet transform to extract energy features and calculate cumulative damage gradient parameters , solve the damage stress coupling model and generate dynamic capability boundary, on the one hand, guide to reconstruct the ideal speed curve to reduce the jump and combine the nonlinear gap compensation pulse torque to superimpose on the servo drive unit, on the other hand, when the running speed is lower than the minimum threshold, execute the secondary derating strategy and generate maintenance instructions to replace the components and send to the operation and maintenance engineer, the operation and maintenance engineer triggers the correction model benchmark parameter process after physical maintenance, thereby updating the parameters to complete the closed-loop management.

[0033] Embodiment 4: This embodiment constructs a hardware-in-the-loop simulation test environment for the core algorithm module of the system, which is composed of a real-time simulator, a real motion controller and a physical signal injection interface, wherein the real-time simulator runs a high-fidelity mechanical dynamics model, which not only integrates the nonlinear friction, gap hysteresis and stiffness time-varying characteristics of the ball screw, but also implants a microscopic damage evolution sub-model based on the Archard wear theory, which is used to generate physically decaying signals dynamically accumulated with running time. The test process sets an accelerated aging full life cycle scenario, covering the whole process from the new machine running-in period to the wear period. On the simulation time axis, the system injects virtual vibration signals and torque current data containing specific frequency band energy features to the motion controller through the physical interface. In the initial stage, the input signal simulates the new machine state, and the feature vector is characterized by low energy and stable distribution. With the passage of simulation time, according to the preset damage evolution law, the energy proportion of the injected signal in the 120Hz to 150Hz frequency band presents a nonlinear growth, while the torque disturbance with random amplitude fluctuation is superimposed at the zero crossing point to simulate the real wear gap effect.

[0034] During this process, the correlation between the cumulative damage gradient parameter calculated by the motion controller and the theoretical damage value set by the simulation model is monitored , the test data shows that in the whole life cycle, the change trend can be tracked with very high followability, the Pearson correlation coefficient of the two reaches 0.985, especially in the early stage of simulating microscopic crack propagation, The successful capture of the vibration energy in a specific frequency band of the weak positive deviation, about 15% of the life cycle time earlier than the traditional alarm mechanism based on the total value (RMS) of vibration; and for the wear period (set ), the adaptive adjustment effect of the maximum jerk limit value is verified by test, when the damage degree represented by the injection signal aggravates, the system adjusts down in real time according to the damage negative feedback control law, the simulation results show that when the dynamic boundary limit is not enabled, the torque waveform of the virtual motor output appears severe shock at the commutation moment, and the peak torque reaches 2.5 times of the rated value; after the algorithm of the application is enabled, is automatically limited to about 40% of the rated value, the output torque waveform becomes smooth, the commutation impact peak is reduced to within 1.2 times of the rated value, and the position following error always converges within the process allowed range.

[0035] Embodiment 5: In order to verify the stability of the control system under extreme parameter boundaries, this embodiment systematically fills data and tests boundaries for key parameters in the damage stress coupling model, in order to ensure that the calculation accuracy of the cumulative damage gradient parameter is not affected by the discreteness of the physical characteristics of different batches of mechanical components, the engineering team performs a standardized baseline calibration process for the newly assembled transmission chain, including driving the servo shaft at 5%, 50% and 100% of the rated speed under no load conditions for full stroke reciprocating motion, and synchronously collecting raw data of vibration sensors and torque current, the system uses these data to construct an initial health state feature vector matrix specific to the device, as the only reference for subsequent differential operation, by introducing a set of standardized simulated damage samples, including known size guide rail scratches and bearing pre-tightening force deviations, the system performs a series of gradient loading experiments, the system records the mapping relationship between the value and the physical damage measured value of the model output under different damage levels, thereby completing the fine calibration of the internal weight parameters of the model, ensuring the consistency and reproducibility of the damage assessment results throughout the device life cycle.

[0036] And for the difference of environment that the device may face in different production sites, the embodiment performs a pre-site deployment calibration procedure, before the device is officially put into production, the control system automatically executes an environmental noise self-learning program, by driving the servo motor to keep still and closing the active control loop, the vibration signals and electromagnetic interference background noise in the environment are collected, the system automatically adjusts the filter cutoff frequency and wavelet packet transform threshold parameters in the signal processing module according to the collected noise spectrum characteristics, to maximize the signal-to-noise ratio, in addition, the system also performs a dynamic response test, by sending a step excitation signal and monitoring the transient response curve of the system, the resonance frequency point of the current mechanical structure based on the actual installation is identified, based on the identification result, the system automatically configures the center frequency and bandwidth of the notch filter, eliminates the risk of unstable control caused by environmental factors, and ensures that the dynamic maintenance management system can maintain the designed performance indicators in different deployment environments.

[0037] Embodiment 6: This embodiment constructs an offline generation and verification process of adaptive parameter matrix, in order to solve the problem of control performance fluctuation caused by the difference of physical parameters such as motor rotor inertia, pole pair number and stator inductance, the system executes a parameter identification subprogram, in the no-load state of the motor and the load, the controller injects a set of pseudo-random binary sequence (PRBS) covering 1Hz to 500Hz as excitation signal, and synchronously collects the speed response and current feedback of the motor, using the collected input and output data, the system identifies the motor's rotational inertia , viscous friction coefficient and electromagnetic time constant based on the least square method algorithm, the system enters the load coupling calibration stage, after connecting the mechanical load, the controller drives the motor to execute a set of preset trapezoidal speed curve motion, by comparing the torque current integral values in no-load and load states, the load inertia and coulomb friction torque reduced to the motor shaft are solved , based on the identified and , the system automatically calculates the inertia ratio , according to the pre-set expert rule base, when is less than 5, the system sets the position loop gain to the reference value of high dynamic response mode; when is between 5 and 15, the system selects medium gain and activates the vibration suppression filter; when is greater than 15, the system automatically switches to large inertia control mode, reduces the speed loop integral gain and enhances the feedforward compensation.

[0038] Finally, the system performs a full stroke dynamic verification test, the controller drives the load to reciprocate at 80% of the maximum design speed, and monitors the peak value of the position following error and the settling time in real time. If the monitoring indicators exceed the pre-set qualified threshold, for example, the following error exceeds 0.05 mm or the settling time exceeds 50 ms, the system will automatically trigger the parameter fine-tuning logic to iteratively adjust the speed loop proportional gain with a step size of 0.5 until the performance indicators converge within the qualified range.

[0039] It is apparent for those skilled in the art that the present application is not limited to the details of the foregoing exemplary embodiments, and the present application can be implemented in other concrete forms without departing from the spirit or essential characteristics of the present application.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application.​

Claims

1. A dynamic maintenance management method based on the full life cycle of a test device, characterized by, The method is applied to an industrial equipment control system comprising a servo drive unit and a motion controller, and comprises the following steps: In step S100, time-domain vibration signals and torque current data during equipment operation are obtained through a high-frequency vibration sensor and a current loop feedback interface of the servo drive unit, wavelet packet transformation is performed to extract energy features of specific frequency bands, and a cumulative damage gradient parameter representing the degree of physical degradation of the mechanical transmission chain is calculated according to the energy features; In step S200, the cumulative damage gradient parameter is input into a preset damage stress coupling model, the maximum radial impact force that the damaged component can withstand under the current working condition is calculated, and the maximum radial impact force is mapped as a dynamic capability boundary for logically constraining the motion planning instruction, the dynamic capability boundary at least comprising a maximum jerk limit value and a position loop gain correction coefficient; In step S300, the motion controller reconstructs an ideal speed curve generated by the original machining code by applying the dynamic capability boundary to the ideal speed curve in an interpolation period, reshapes the torque load spectrum output to the servo motor by reducing the jerk of the acceleration and deceleration segments, and monitors the following error of the servo shaft in real time, and when the following error converges within a preset process precision threshold after reshaping, the current reshaping control strategy is maintained.

2. The dynamic maintenance management method based on the whole life cycle of a test device according to claim 1, characterized by, The specific process of calculating the cumulative damage gradient parameter in step S100 comprises the following steps: in step S201, the obtained time-domain vibration signals are divided into continuous time windows, and the data in each time window is subjected to detrending processing; in step S202, the spectral energy proportion of the signal subjected to the detrending processing at the mechanical natural frequency and its multiple frequencies is calculated, the spectral energy proportion is subjected to differential operation with the spectral energy distribution under the baseline health state to generate a differential feature vector; in step S203, the differential feature vector is input into a pre-trained long short-term memory network model, a dimensionless scalar reflecting the microscopic crack propagation rate is output, the dimensionless scalar is defined as the cumulative damage gradient parameter, and the cumulative damage gradient parameter is transmitted to step S200.

3. The method of claim 1, wherein the method further comprises: The process of mapping the maximum radial impact force to the maximum jerk limit value in step S200 follows a damage negative feedback control law as follows: wherein, is the calculated maximum jerk limit value for the current interpolation period, is the rated jerk value set by the factory, is the cumulative damage gradient parameter calculated in step S100, is the preset damage sensitivity coefficient, is the position following error value of the previous control period, is the preset error suppression weight factor; the motion controller reduces the dynamic response rate of the output command in real time according to the damage negative feedback control law.

4. The method of claim 1, wherein the method further comprises: The specific way of reconstructing the ideal speed curve generated by the original machining code in step S300 comprises the following steps: in step S401, straight line or circular arc interpolation instructions in the original machining code are analyzed to generate a speed planning curve; in step S402, the maximum jerk limit value determined in step S200 is read in the acceleration and deceleration stage of the speed planning curve; in step S403, when the original jerk demand of the speed planning curve exceeds the maximum jerk limit value, the target speed is kept unchanged, the peak jerk is reduced by prolonging the time length of the acceleration and deceleration stage, and a smoothed actual execution speed curve is generated; in step S404, the actual execution speed curve is discretized into position control instructions, and the position control instructions are sent to the position loop of the servo driver.

5. The method of claim 1, wherein the method further comprises: The dynamic capability boundary further comprises a torque feedforward compensation coefficient; the step S300 further comprises a nonlinear compensation step for mechanical clearance: step S501, when the cumulative damage gradient parameter indicates that the transmission chain has reverse clearance, the system superimposes a pulse torque consistent with the motion direction at the zero crossing point of the servo motor speed according to the torque feedforward compensation coefficient; step S502, the amplitude of the pulse torque is positively correlated with the cumulative damage gradient parameter, and the duration of the pulse torque is determined by the response bandwidth of the current loop of the servo drive unit.

6. The method of claim 1, wherein, The step S300 further comprises a secondary derating convergence step after monitoring the follow-up error of the servo shaft: step S601, if the follow-up error still exceeds the preset process accuracy threshold after the continuous N interpolation periods after applying the dynamic capability boundary, the secondary derating strategy is triggered; Step S602, execute the secondary derating strategy, gradually reduce the maximum running speed of the servo motor by a preset step, until the follow-up error converges to the preset process accuracy threshold or the current running speed is lower than the minimum speed threshold required to maintain the production process; step S603, when the current running speed is lower than the minimum speed threshold, generate a maintenance instruction containing the type of the component to be replaced and the fault position information.

7. The method of claim 1, wherein, The method further comprises a model iteration step based on edge computing: step S701, after each device performs physical maintenance and replaces the component, record the initial vibration signal and current data after replacement; Step S702, update the reference parameters in the damage stress coupling model by taking the initial vibration signal and current data as new reference health state data; step S703, calculate the change rate of the cumulative damage gradient parameters before and after maintenance, and correct the mapping weight of the dynamic capability boundary accordingly, so that the response characteristics of the control system adapt to the actual mechanical impedance after replacing the component.

8. The method of claim 1, wherein the method further comprises: The determination logic of the position loop gain correction coefficient in step S200 includes: step S801, monitor the current noise level of the servo motor in the steady state running stage; step S802, when the current noise level exceeds the preset noise threshold, it is determined that the system has resonance risk caused by the decrease of mechanical stiffness; step S803, calculate the critical proportional gain value that can suppress resonance, and set the position loop gain correction coefficient to the attenuation ratio that makes the current position loop gain lower than the critical proportional gain value; the method is run in a control system with multi-axis cooperation function, and the reconstruction in step S300 further comprises multi-axis synchronization constraint processing: step S901, when a certain axis prolongs the acceleration and deceleration time due to the application of the dynamic capability boundary, the motion state of other axes related to the axis is obtained; step S902, according to the acceleration and deceleration time of the axis after prolonging, the motion planning parameters of other axes are adjusted synchronously to maintain the consistency of the synthesized trajectory of multi-axis interpolation motion with the preset trajectory.

9. The method of claim 1, wherein the method further comprises: The method further comprises a load spectrum discretization storage step for specific working conditions: step S1001, dividing the whole life cycle of the equipment into a running-in period, a stable period and a degradation period; step S1002, in the degradation period, collecting and recording the actual torque load spectrum data shaped by step S300 according to different action types; step S1003, establishing a correlation matrix between the action type and the mechanical damage propagation rate, and when a high damage propagation rate action type is identified, further superimposing a preset safety redundancy coefficient on the basis of the dynamic capability boundary.

10. A dynamic maintenance management system based on the full life cycle of test equipment for implementing the method of claim 1, characterized in that, The system comprises a data acquisition and feature extraction unit, a boundary calculation and decision unit, and an adaptive shaping control unit. The data acquisition and feature extraction unit is used to acquire time-domain vibration signals and torque current data of the equipment during operation through a high-frequency vibration sensor and a current loop feedback interface, perform wavelet packet transformation to extract energy features of specific frequency bands, and calculate a cumulative damage gradient parameter representing the physical degradation degree of the mechanical transmission chain according to the energy features; The boundary calculation and decision unit is used to input the cumulative damage gradient parameter into a preset damage stress coupling model, calculate the maximum radial impact force that the damaged component can withstand under the current working condition, and map the maximum radial impact force to a dynamic capability boundary for logically constraining the motion planning instruction, wherein the dynamic capability boundary at least includes a maximum jerk limit value and a position loop gain correction coefficient; The adaptive shaping control unit is integrated in a motion controller and is used to reconstruct an ideal speed curve generated by original machining code by applying the dynamic capability boundary within an interpolation period, shape the torque load spectrum output to the servo motor by reducing the jerk of the acceleration and deceleration segments, and monitor the following error of the servo shaft in real time, and when the following error converges within a preset process precision threshold after shaping, maintain the current shaping control strategy.

Images