Real-time pressure compensation method for multi-axis linkage precision press

Abstract

The invention discloses a real-time pressure compensation method for a multi-axis linkage precision press, which relates to the technical field of control of a press, and comprises the following steps of: when a coefficient is detected to exceed a warning threshold value, executing abnormal identification and generating abnormal axis information; a short-stroke reciprocating test is carried out on an abnormal shaft under a safe working condition, force-displacement and force-speed curve characteristics are recorded through high-frequency sampling, and friction model parameters are updated by utilizing a self-learning fitting algorithm. And calculating a life degradation index by comparing a historical parameter sequence with the updated friction model parameters, and formulating a pre-maintenance strategy including a replacement period, spare part requirements and priorities. If the pre-maintenance strategy is not executed in time and it is monitored that the friction increasing coefficient continues to rise or the compensation amount is close to the upper limit, real-time compensation is applied, and the speed-limiting and load-limiting or emergency shutdown process is triggered. Through a multi-level safety control strategy, sudden friction abnormity can be accurately captured, and reliability and maintainability are remarkably improved.

Description

Technical Field

[0001] The invention relates to the technical field of precision press control, and in particular to a real-time pressure compensation method for a multi-axis linkage precision press. Background Art

[0002] In the field of industrial production and equipment manufacturing, there are widespread multi-axis systems composed of key components such as lead screws, guide rails, hydraulic cylinders, etc., which are often used in various occasions such as precision molding, multi-station assembly, and high dynamic loads. During the long-term operation of such systems, wear, oil film damage, or aging of sealing components will occur between the moving pairs, resulting in unstable or even sudden changes in friction. If the friction characteristics suddenly deviate over time or load conditions, the actual speed or force-displacement response of the axis will often have a significant error with the original set model in a short period of time, which will have a great impact on synchronous control and production accuracy. On the other hand, due to the accelerating pace of modern production, equipment is more prone to problems such as lubrication decline and sudden leakage under continuous and high-load operation, resulting in potential threats to equipment operation safety and product quality. At the same time, the control logic of multi-axis systems is usually more complex. The traditional method of relying on linear friction models or regular manual inspections is difficult to capture potential nonlinear friction jumps in time and provide sufficient basis for subsequent maintenance and control adjustments.

[0003] In the Chinese invention patent with the authorization announcement number CN114055321B, a real-time compensation method for grinding and polishing pressure based on a numerical control system is disclosed, which includes the following steps: S1, when the spindle servo motor moves, the current load current of the spindle motor is obtained, and the current load current is fed back to the numerical control system controller; S2, the numerical control system controller continuously obtains the current load current value of the spindle motor in each cycle, and obtains the actual load current of the real-time spindle motor after filtering; S3, the numerical control system controller calculates the current compensation amount proportional value according to the actual load current; S4, calculates the spindle displacement compensation amount of the current moving trajectory of the spindle during the processing process; S5, assigns the axis displacement compensation amount to the feed servo electric axis movement amount, so as to keep the pressure constant; S6, repeats the above steps S1-S5 to realize the automatic real-time pressure compensation of the grinding wheel on the workpiece surface. The present invention has the advantages of not requiring the modification of mechanical equipment or compensation devices, and can be quickly applied to the occasions of plane grinding and polishing.

[0004] Combining the current application scenarios and the above existing technologies:

[0005] There is still a lack of a multi-level and closed-loop solution for timely identifying and predicting such nonlinear friction mutations in actual operation of multi-axis systems. Existing technologies mostly focus on a single link, such as simplifying analysis by only calibrating static friction parameters, or implementing fault alarms in a single threshold manner. It is difficult to take into account the asymmetric wear process when lubrication fails, and it is even more difficult to provide effective prediction and maintenance strategies before the fault approaches the critical point. In addition, conventional methods are not flexible enough for occasions with high requirements for multi-axis synchronization. It is impossible to rely on a unified model for real-time compensation when the friction states between different axes are inconsistent, which often leads to practical problems such as reduced system synchronization accuracy, increased wear of local bearings or screws, and a sharp increase in safety risks.

[0006] To this end, the present invention provides a real-time pressure compensation method for a multi-axis linkage precision press. Summary of the invention

[0007] 1. Technical issues to be solved

[0008] In view of the shortcomings of the prior art, the present invention provides a real-time pressure compensation method for a multi-axis linkage precision press. When it is detected that the coefficient exceeds the warning threshold, abnormality identification is performed and abnormal axis information is generated; a short-stroke reciprocating test is performed on the abnormal axis under safe working conditions, and the force-displacement and force-velocity curve characteristics are recorded through high-frequency sampling, and the friction model parameters are updated using a self-learning fitting algorithm.

[0009] By comparing the historical parameter sequence with the updated friction model parameters, the life degradation index is calculated, and a pre-maintenance strategy including replacement cycle, spare parts demand and priority is formulated. If the pre-maintenance strategy is not executed in time and it is monitored that the friction increase coefficient continues to increase or the compensation amount is close to the upper limit, real-time compensation will be applied to trigger speed and load limit or emergency shutdown process. Through multi-level safety control strategies, sudden friction anomalies can be accurately captured, significantly improving reliability and maintainability.

[0010] A comprehensive solution integrating online monitoring, dynamic friction model update, life assessment and safety compensation control has been established, so as to achieve early detection and refined control of friction mutations in the actual industrial environment of multi-axis systems, ensure the working accuracy and operation safety of the equipment, and quickly trigger preventive maintenance or emergency protection when necessary, solving the technical problems recorded in the background technology.

[0011] (II) Technical solution

[0012] To achieve the above objectives, the present invention is implemented through the following technical solutions: a real-time pressure compensation method for a multi-axis linkage precision press, when it is detected that the friction increase coefficient of each axis continuously exceeds the warning threshold within a limited time window, based on the collected feedback data, a rolling integral and multi-threshold judgment algorithm is used to perform abnormal identification and generate abnormal axis information;

[0013] After receiving the abnormal axis information, a short-stroke reciprocating motion test is performed on the specified axis under safe working conditions, and the force-displacement and force-velocity curve characteristics are recorded by high-frequency sampling. The friction model parameters are updated using a self-learning fitting algorithm to obtain the updated friction model parameters;

[0014] Compare the historical multi-cycle parameter sequence with the current updated friction model parameters, calculate the life degradation index according to the high-order deviation accumulation function, and formulate a pre-maintenance strategy including replacement cycle, spare parts demand and priority, and output the remaining usable life and maintenance period in the data layer;

[0015] When the predictive maintenance strategy is not executed in time and it is monitored that the friction increase coefficient continues to increase or the model compensation amount is close to the upper limit, real-time compensation is applied in the multi-axis synchronous control loop according to the updated friction model parameters, triggering the speed and load limit or emergency shutdown process.

[0016] Preferably, in the controller or host computer of the multi-axis system, a sampling period and a monitoring time window are predefined for each axis i; for each axis i′, its real-time pressure Pi(t), real-time speed vi(t), given reference pressure Pi(t) and model (t) and the reference speed vi ref (t), the friction increase coefficient Γi(t) is defined to evaluate the possible friction mutation of shaft i in the recent period, and the formula is as follows:

[0017]

[0018] τ is the integral variable; α 1 With α 2 are the weight coefficients of pressure deviation and velocity deviation, Pi(τ) and Pi model (τ) are the measured pressure and reference pressure of axis i, respectively; vi(τ) and vi ref (τ) are the measured speed and reference speed of axis i respectively; ε is a small positive constant.

[0019] Preferably, real-time data collection is performed according to the sampling period, and the friction increase coefficient Γi(t) of the axis i is updated at each sampling moment; one or more threshold curves Θi are set for each axis i, and when the friction increase coefficient Γi(t) continuously exceeds the warning threshold Θiwarn and is difficult to recover in a short time, an abnormal mark is automatically generated, and the abnormal axis information is recorded, including the axis number, trigger time t, and associated operating conditions; if the friction increase coefficient Γi(t) exceeds the severe threshold Θicrit, a higher level of warning is immediately triggered;

[0020] Preferably, according to the output abnormal axis information, the target axis that needs to perform a small reciprocating test is selected, recorded as axis number i, and its abnormal trigger time and abnormal operating condition are obtained. The controller moves the axis to a safe working area within a controllable time period; a reciprocating motion curve is planned according to a given stroke amplitude δpos and a test speed range δvel so that axis i repeats several short strokes in both the forward and reverse directions; the test force of axis i is collected in real time Actual displacement xi(t) and actual velocity vi(t); to obtain force-velocity and force-displacement curves;

[0021] Preferably, when reading the high frequency test force After calculating the displacement xi(t) and velocity vi(t), the noise is smoothed and filtered, and the key inflection point information of the force-displacement curve is recorded at the turning point; the segment features in the force-velocity curve are extracted according to different speed ranges; the generalized friction model is used to represent the comprehensive friction force of shaft i Defining a higher-order deviation accumulation function To measure the overall error between the test data and the model output, the following is the formula:

[0022]

[0023] Where: is the actual friction data observed in the test; is the friction force prediction value calculated based on the current friction model parameter vector Θi; β is the weighting coefficient of the deviation cubic term; [ts, te] is the start and end of the test data on the time axis;

[0024] The self-learning iterative algorithm is used to find the updated optimal friction model parameters Θi * , so that the higher-order deviation accumulation function When the minimum is reached, the optimal parameter vector is named as the updated friction model parameter and stored in association with the axis number i.

[0025] Preferably, the updated friction model parameters are denoted as Θi * , combined with the historical parameters of previous tests and actual operation to form a time series; with the updated friction model parameters Θi * Friction model parameters under the baseline condition The gap between them is used to construct the comprehensive degradation metric function Λi, and the formula is as follows:

[0026]

[0027] Where: θi,k(t) is the kth friction parameter corresponding to shaft i at time t; θi,k (0) is the calibration value of the same parameter under the reference state, γ 1 and γ 2is the weight coefficient of different parameter deviations;

[0028] After obtaining the comprehensive degradation metric function Λi corresponding to the current time point T, compare the historical sequence to analyze its growth rate and fit the degradation curve; if it is predicted in subsequent runs that Λi will exceed the threshold Λi at time T+Δt (crit) , then Δt is the estimated value of the remaining useful life;

[0029] Record and output the life assessment results, including the current wear level and predicted replacement period of each axis;

[0030] Preferably, the life assessment results of all axes are summarized. If the predicted remaining useful life RULi of some axes is close to zero or the wear rate is higher than expected, they are marked as high-priority maintenance objects, otherwise they are included in the periodic maintenance plan;

[0031] For high-priority maintenance objects, generate an immediate or near-term replacement demand list based on actual inventory conditions;

[0032] Establish regular maintenance cycles for the remaining axes, estimate the usage of corresponding seals, lubricants or other vulnerable parts, and finally output a preventive maintenance strategy, including a spare parts requirement list, a time schedule, and a priority description for each axis;

[0033] Preferably, after reading the pre-maintenance strategy, the priority information and expected replacement cycle of each axis are obtained, and the updated friction model parameters are obtained. and the friction increase coefficient Γ i (t);

[0034] In the main control cycle, differential budget is performed based on the multi-axis synchronization command and the actual value of sensor feedback;

[0035] The above difference results and the friction increase coefficient Γ i (t) and the updated friction model parameters Combine and calculate the instant compensation amount

[0036] The compensation amount calculated in real time Inject multi-axis synchronous controller to make the axis with abnormal friction get additional compensation or correction in force and speed;

[0037] Preferably, the shafts that may experience greater friction deterioration or seal failure in the near future are included in the list of high-risk shafts; if the friction increase coefficient Γ is detected i (t) continues to increase within a short time window, and the compensation amount If it is also close to the upper limit allowed, the speed limit and load limit strategy will be activated;

[0038] Execute the set speed limit curve for the high-risk axis, and set the maximum running speed of axis i Reduce to a certain preset ratio and limit the drive power output in the controller;

[0039] If the trend is not alleviated, continue to increase the restriction; if it is alleviated, maintain this speed limit until the next maintenance is completed.

[0040] Preferably, multiple safety thresholds are set; if any axis exceeds the threshold within a short time window, or the compensation amount has reached the limit, the early warning mechanism is triggered; if the early warning is a moderate risk, an alarm is issued to the operator and it is recommended to speed up the replacement operation in the preventive maintenance strategy; if the early warning is an extreme risk, the operating load of the axis or the equipment as a whole is immediately limited, and an emergency shutdown sequence is executed when necessary.

[0041] (III) Beneficial effects

[0042] The present invention provides a real-time pressure compensation method for a multi-axis linkage precision press, which has the following beneficial effects:

[0043] Combining online monitoring, self-learning models and multi-level safety control, from short-stroke reciprocating testing and real-time friction parameter correction, to life degradation prediction and flexible preventive maintenance, to multi-axis synchronous compensation, speed and load limits and emergency shutdown protection, a closed-loop management process with highly unified data and logic has been formed.

[0044] First, the friction coefficient T is increased by i The combination of (t) and the classification threshold can accurately capture sudden friction anomalies at an early stage; then, based on targeted high-frequency reciprocating tests, the self-learning algorithm continuously updates the friction model parameters and integrates the life degradation analysis, using the high-order deviation metric Λ i Accurate remaining life estimation is performed. Subsequently, spare parts demand and maintenance cycle are coordinated according to the predictive maintenance plan, making maintenance resource allocation more efficient and effectively avoiding downtime risks. Finally, with the help of multi-level safety control strategies, compensation In severe situations that are difficult to control, speed limit or emergency shutdown protection is automatically activated to prevent the spread of faults and ensure the safety of personnel and equipment.

[0045] It realizes global perception and dynamic decision-making of the operating status of the multi-axis system, greatly improving the reliability and maintainability of the system under the problem of nonlinear friction mutation. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a schematic flow chart of the real-time pressure compensation method of the multi-axis linkage precision press of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] See also Figure 1 The present invention provides a real-time pressure compensation method for a multi-axis linkage precision press, comprising:

[0049] Step 1: When it is detected that the friction increase coefficient of each axis exceeds the warning threshold continuously within a limited time window, based on the collected feedback data, the rolling integral and multi-threshold judgment algorithm is used to perform abnormal identification and generate abnormal axis information;

[0050] The step 1 includes the following contents:

[0051] Step 101: Online monitoring initialization and parameter definition

[0052] In the controller or host computer of the multi-axis system, a sampling period Δtsamp and a monitoring time window Δtwin are predefined for each axis i; the former is used to collect data at a fixed beat, and the latter is used to calculate the friction increase coefficient within a rolling time window;

[0053] For each axis i′, record its real-time pressure Pi(t), real-time speed vi(t), given reference pressure Pi model (t) (derived from the preliminary friction model or rated operating conditions) and the reference speed vi ref (t), the reference pressure can come from the preliminarily determined process or simulation model, and the reference speed comes from the main control instruction;

[0054] In order to effectively quantify the abnormal degree of friction characteristics, the friction increase coefficient Γi(t) is defined to evaluate the possible friction mutation of shaft i in the recent period of time. The formula is as follows:

[0055]

[0056] Γi(t) is the friction increase coefficient of shaft i, which is used to characterize the degree of deviation of the friction force or lubrication condition of the shaft relative to the reference model in the time period [t-Δtwin, t];

[0057] τ is the integral variable, and its value range is [t-Δtwin, t]; α 1 With α 2 are the weight coefficients of pressure deviation and velocity deviation, respectively, and their values ​​are between 0 and 1. They can be adjusted according to system characteristics to highlight the emphasis on pressure or velocity anomalies.model (τ) are the measured pressure and reference pressure of axis i, respectively; vi(τ) and vi ref (τ) are the measured speed and reference speed of axis i, respectively; ε is a small positive constant used to prevent calculation instability caused by division by zero or extreme conditions;

[0058] When used, by defining the friction increase coefficient Γi(t) and distinguishing between pressure and speed deviations, the potential friction anomaly of shaft i can be quickly captured, which helps to establish a refined diagnosis basis; and using the integral form with the deviation weighting coefficient can enhance the sensitivity to small and frequent fluctuations and improve the ability to capture early faults. This index processing method is flexible and can adjust α according to actual working conditions. 1 , α 2 and ε to adapt to different types of lubrication or sealing mechanisms; on the basis of conventional linear deviation, ε and reference value in the denominator are introduced to avoid numerical singularities and amplify anomalies under low speed and low pressure conditions; the weighted integral form combined with the rolling time window can balance the ability to detect fault signs between the short-term and the medium- and long-term.

[0059] Step 102: Real-time data collection and abnormal threshold determination

[0060] Real-time data collection is performed according to the sampling period Δtsamp, and the friction increase coefficient Γi(t) of axis i is updated at each sampling moment; one or more threshold curves Θi are set for each axis i, which are divided into two levels of warning threshold and critical threshold according to the actual tolerance, including Θiwarn and Θicrit. When the friction increase coefficient Γi(t) continuously exceeds the warning threshold Θiwarn and is difficult to recover in a short time, an abnormal mark is automatically generated, and abnormal axis information is recorded, including axis number, trigger time t, and associated operating conditions (such as load, temperature, etc.);

[0061] If the friction increase coefficient Γi(t) exceeds the severe threshold Θicrit, a higher level warning is immediately triggered;

[0062] When in use, the dual-threshold judgment mechanism can provide early warning of faults and screen out serious faults that really need priority attention. Based on the abnormal axis information, the specific axis and its timing can be quickly located to avoid large-scale shutdowns and blind troubleshooting. The abnormal data objects provided to subsequent steps are structured and traceable, making the micro-reciprocating test range more focused and the test efficiency higher.

[0063] The dual-threshold or multi-value judgment strategy is different from the simple single-threshold mode. It is more adaptable to the early fault screening and key fault identification under complex load conditions of multi-axis systems. It uses continuous over-threshold conditions in a rolling time window rather than a single over-threshold judgment to reduce the sensitivity to false alarms caused by instantaneous noise.

[0064] Step 2: After receiving the abnormal axis information, perform a short-stroke reciprocating motion test on the specified axis under safe working conditions, use high-frequency sampling to record the force-displacement and force-velocity curve characteristics, and use a self-learning fitting algorithm to update the friction model parameters to obtain the updated friction model parameters;

[0065] The step 2 includes the following contents:

[0066] Step 201: Small range reciprocating motion test

[0067] According to the output abnormal axis information, the target axis that needs to perform a small reciprocating test is selected and recorded as axis number i, and its abnormal trigger time and abnormal operating conditions are obtained for reference. To avoid impact on production, the controller moves the axis to a safe working area (usually away from the extreme position or the middle stroke with the smallest load) within a controllable time period to ensure minimal impact on other axes and equipment during the test.

[0068] In the safe working area, a reciprocating motion curve is planned according to the given stroke amplitude δpos and test speed range δvel so that axis i repeats several short strokes in both the positive and negative directions; during the test, the test force of axis i is collected in real time (which can be obtained from the force sensor or hydraulic system feedback), actual displacement xi(t) and actual speed vi(t); the required sampling frequency is higher than the monitoring frequency during normal production to obtain more precise force-speed and force-displacement curves;

[0069] All data in the test process are marked with test data, and a mapping relationship is established between the axis number i, the test period, and the abnormal identification of the first step to prevent confusion with the data in the normal production stage in the data file or database;

[0070] When in use, by collecting high-frequency force-velocity and force-displacement data within a controllable and safe reciprocating range, the friction characteristics of the shaft can be captured more accurately; the short-stroke forward and reverse motion can not only test the difference between static friction and dynamic friction, but also observe nonlinear phenomena such as stick-slip critical speed. Compared with the large-scale online monitoring in the first step, it can have higher resolution and more targeted detection capabilities. By using a short-range two-way reciprocating method, the friction characteristics of the shaft in the acceleration and deceleration sections in different directions are uniformly collected and packaged, which can be more comprehensive than a one-way uniform speed test. Through test data marking and production data, it can not only avoid confusion, but also provide a higher dimensional comparison and reuse value for subsequent analysis.

[0071] Step 202: Self-learning model update and parameter fitting

[0072] Reading high frequency test force After the displacement xi(t) and velocity vi(t) are obtained, the noise is smoothed and filtered, and the key inflection point information of the force-displacement curve is recorded at the turning point; according to different speed ranges (such as low speed, critical speed, medium and high speed), the segment features in the force-velocity curve are extracted to facilitate segmented or overall processing in the subsequent friction model fitting;

[0073] In order to be compatible with typical nonlinear friction phenomena, a generalized friction model is used to represent the comprehensive friction force of shaft i:

[0074]

[0075] Where θ_i=[θ i ,1,θ i ,2,θ i ,3] is the friction parameter vector to be updated, and the model can include the static friction term Ω static and the kinetic friction term Ω dynamic Etc., are used to describe the friction force changes in different speed ranges;

[0076] Define a higher-order deviation accumulation function To measure the overall error between the test data and the model output, the following is the formula:

[0077]

[0078] Where: is the actual friction force (or equivalent force) data observed in the test; is the friction force prediction value calculated based on the current friction model parameter vector Θi; β is the weighting coefficient of the deviation cube term, which is used to improve the sensitivity to large error areas and takes a value between 0 and 5; [ts, te] is the start and end of the test data on the time axis;

[0079] A self-learning iterative algorithm (such as adaptive gradient descent or nonlinear minimization strategy) is used to find the updated optimal friction model parameters Θi * , so that the higher-order deviation accumulation function After the update is completed, the optimal parameter vector is named as the updated friction model parameter and is stored in association with the axis number i.

[0080] When in use, by comparing the high-frequency data obtained in a small-scale reciprocating test with the generalized friction model, the key coefficients of static friction and dynamic friction can be corrected in real time to adapt to sudden changes in the operating state of the equipment; a high-order deviation accumulation function is used Replacing the traditional square or absolute error metrics helps to improve sensitivity to outliers and shorten convergence time; the self-learning update method can achieve dynamic iteration, gradually approaching the optimal friction parameters under actual working conditions, avoiding model mismatch caused by relying solely on offline calibration.

[0081] Step 3: Compare the historical multi-cycle parameter sequence with the currently updated friction model parameters, calculate the life degradation index according to the high-order deviation accumulation function, and formulate a pre-maintenance strategy including replacement cycle, spare parts demand and priority, and output the remaining usable life and maintenance period in the data layer;

[0082] The step three includes the following contents:

[0083] Step 301: Wear trend analysis and life degradation assessment

[0084] The updated friction model parameters are denoted as Θi * =[θi,1 * ,θi,2*,…] (where the subscript i represents the axis number), and combined with the historical parameters {θi (1) , Θi (2) , …,} form a time series; check whether there is a class change anomaly or data missing in each update of the axis; if it is found that there is a mismatch with the abnormal axis information in the first step or the test data mark in the second step, it will be supplemented or corrected at this stage;

[0085] With the updated friction model parameters Θi * Compared with the friction model parameters in the baseline state (initial equipment or after the last overhaul) The gap between them is used to construct the comprehensive degradation metric function Λi, and the formula is as follows:

[0086]

[0087] Where: Λi is the life degradation index, which is used to quantitatively describe the cumulative degree of deviation of the key friction parameters of shaft i from the reference value;

[0088] θi,k(t) is the kth friction parameter corresponding to axis i at time t (there may be multiple updates within this time, requiring interpolation or segment compensation); θi,k (0) is the calibration value of the same parameter under the reference state, γ 1 and γ 2 is the weight coefficient of different parameter deviations, with a value between 0 and 10, which is determined by the actual working conditions. It can also be extended to more parameters by adding corresponding weight items. The combination of exponents 4 and 2 can increase the sensitivity to large deviations and avoid the limitation that a single power is difficult to capture for nonlinear wear.

[0089] After obtaining the comprehensive degradation metric function Λi corresponding to the current time point T, compare the historical sequence {Λi (1) ,Λi (2) ,…} to analyze its growth rate and fit the degradation curve (such as using exponential or polynomial extrapolation).

[0090] Set a critical threshold Λi (crit) , if it is predicted in subsequent operation that Λi will exceed the threshold Λi at time T+Δt (crit) , then Δt is the estimated value of the remaining usable life (RUL); record and output the life assessment results, including the current wear level and predicted replacement period of each axis;

[0091] When used, the life degradation index Λi in the form of high-order deviation can enhance the sensitivity to nonlinear wear or sudden failure. By using the parameter change trend after multiple iterations of historical data, rather than just looking at a single test result, the wear evolution law can be grasped more accurately: Introducing the critical threshold Λi (crit) It is convenient to directly convert the degree of degradation into a remaining life estimate, simplifying the subsequent decision-making process.

[0092] Step 302: Preventive maintenance strategy generation and spare parts configuration

[0093] Summarize the life assessment results of all axes, including Λi(T), the predicted remaining useful life RULi and the compared threshold Λi for each axis (crit) Prioritize each axis; if the predicted remaining useful life RULi of some axes is close to zero or the wear rate is higher than expected, they will be marked as high-priority maintenance objects, otherwise they will be included in the periodic maintenance plan;

[0094] For high-priority maintenance objects, generate an immediate or near-term replacement demand list based on actual inventory conditions;

[0095] Establish regular maintenance cycles for the remaining axes, and estimate the usage of corresponding seals, lubricants or other wearing parts, and achieve a balance between cost and efficiency through bulk ordering and replacement in batches. If the system is a multi-machine parallel production, production scheduling can be adjusted at the overall level to switch machines with severe wear to low load or standby status to extend the actual operating time.

[0096] The final output is the preventive maintenance strategy, including the spare parts requirement list, time schedule and priority description of each axis; the strategy is archived together with the updated friction model parameters and life assessment results in this record;

[0097] When in use, by making priorities based on life degradation indicators and remaining life estimates, the equipment life can be maximized and unnecessary downtime can be reduced; the linkage planning of maintenance cycles and spare parts requirements avoids emergency out-of-stock or excessive inventory, achieving a balance between economy and equipment reliability.

[0098] Step 4: When the pre-maintenance strategy is not executed in time and it is monitored that the friction increase coefficient continues to increase or the model compensation amount is close to the upper limit, real-time compensation is applied in the multi-axis synchronous control loop according to the updated friction model parameters to trigger the speed limit, load limit or emergency shutdown process;

[0099] The step 4 includes the following contents:

[0100] Step 401: real-time compensation strategy loading and multi-axis synchronous adjustment

[0101] Read and call the pre-maintenance strategy output in the third step, obtain the priority information and expected replacement cycle of each axis (recorded as axis number i), and obtain the updated friction model parameters and the friction increase coefficient Γ i (t);

[0102] In the main control cycle, according to the multi-axis synchronization command (usually including the reference speed or reference displacement ) and the actual value fed back by the sensor (v i (t), x i (t)) performs differential operation;

[0103] The above difference results and the friction increase coefficient Γ i (t) and the updated friction model parameters Combined, calculate the instant compensation amount In order to improve the perception of sharp friction mutations, a high-order weighted form is introduced:

[0104]

[0105] is the comprehensive compensation for axis i, used to correct the driving force or servo command, Δt adj To compensate for the integration time window length;

[0106] ω 1 and 2 is a weight coefficient used to adjust the sensitivity to the deviation between the friction increase coefficient and the updated friction model parameters; the combination of powers 3 and 2 can amplify the impact of severe anomalies within the integral;

[0107] The compensation amount calculated in real time Inject into the multi-axis synchronous controller (for example, assign to servo drive instructions or hydraulic valve control instructions) so that the axis with abnormal friction can get additional compensation or correction in terms of force and speed;

[0108] When used, compared with conventional PID or cascade control, an additional Γ-based control is introduced into the control loop. i (t) The nonlinear term can quickly offset the sudden increase in friction and maintain the synchronization accuracy between the axes. By performing real-time compensation for abnormal friction with high-order weighted integration, the damage to accuracy and synchronization caused by sudden friction fluctuations can be quickly suppressed. Different from simple linear compensation, it combines the friction increase coefficient with the updated friction model parameters and has adaptive and nonlinear correction capabilities. The output compensation amount It can be recorded together with the axis number i and its maintenance priority to provide a decision reference for the safety mode or emergency processing of the subsequent steps.

[0109] Step 402: Dynamic speed limit and load limit execution

[0110] According to the remaining usable life RULi and priority determination results of each axis in the predictive maintenance strategy generated in the third step, the axes that may experience greater friction deterioration or seal failure in the near future are included in the list of high-risk axes; in the real-time monitoring process, once the friction increase coefficient Γ is detected i (t) continues to increase within a short time window, and the compensation amount If it is also close to the upper limit allowed, the speed limit and load limit strategy will be activated;

[0111] Execute the set speed limit curve for the high-risk axis, and set the maximum running speed of axis i Lower it to a preset ratio (such as 80% or even lower) and limit the drive power output in the controller; in the hydraulic or pneumatic system, the maximum output pressure at the shaft end can be lowered through the pressure limiting valve or valve core throttling to prevent rapid and large load impact and slow down wear and leakage spread.

[0112] After the speed limit and load limit take effect, continue to detect the friction increase coefficient Γ i (t) and compensation amount If the trend is still not alleviated, continue to increase the restriction; if it is alleviated, maintain this speed limit until the next maintenance is completed.

[0113] When in use, by implementing speed and load limits on high-risk axes, the probability of increased friction and fault spread can be greatly reduced, and the safe operation time of the equipment while waiting for maintenance can be extended. This complements the real-time compensation strategy: when compensation still cannot maintain stability, it automatically switches to speed and load limit mode to ensure that there is no risk of loss of control.

[0114] Step 403: Fault warning and emergency shutdown protection

[0115] When setting multiple safety thresholds, for example, for the friction increase coefficient Γ i (t) The warning threshold Θiwarn and the compensation amount of (maximum allowable compensation amount).

[0116] If any axis exceeds the threshold within a short time window, or the compensation amount reaches the limit, the warning mechanism is triggered;

[0117] If the warning is a medium risk, an alarm is issued to the operator and a suggestion is made to speed up the replacement operation in the preventive maintenance strategy;

[0118] If the warning is an extreme risk, indicating that the lubrication or seal may have failed, immediately limit the operating load of the axis or the entire equipment, and execute the emergency shutdown sequence (including pressure relief, power supply stop, etc.) if necessary.

[0119] All warnings and emergency stops will be logged by the system, along with the friction increase coefficient Γ at that time. i (t), compensation amount and speed and load limit status.

[0120] When in use, multi-level warnings and timely emergency shutdowns are used to avoid secondary disasters caused by extreme friction, loss of control or serious leakage. Real-time records can provide maintenance personnel and management systems with detailed post-analysis basis, and also have reference significance for the rapid identification and compensation algorithm optimization of similar faults in the future, forming the ultimate safety guarantee measure with the aforementioned compensation and load limiting strategies, ensuring that even if timely maintenance is not possible, the equipment and production line can be protected to the greatest extent.

[0121] Using multi-dimensional threshold (for friction increase coefficient Γ i (t) and compensation amount A two-pronged approach) is used to make safety judgments, emphasizing the flexibility and reliability of early warning triggering, and bidirectionally linking real-time logs with the third-step preventive maintenance strategy, so that the safety event itself can also be quantified and evaluated in the next round of maintenance cycle or life prediction, thereby enhancing the system's self-learning and iterative improvement capabilities.

[0122] Combining online monitoring, self-learning models and multi-level safety control, from short-stroke reciprocating testing and real-time friction parameter correction, to life degradation prediction and flexible preventive maintenance, to multi-axis synchronous compensation, speed and load limits and emergency shutdown protection, a closed-loop management process with highly unified data and logic has been formed.

[0123] 1. Discrete update and use of:

[0124] In the second step of abnormal axis micro-reciprocating test and self-learning model update, each time the system completes a small range reciprocating test and converges the self-learning algorithm, it will output a set of the latest friction parameters. Update frequency: Usually not continuously updated in real time, but triggered after anomalies are detected; therefore, It can be regarded as a piecewise constant or piecewise approximate linear state within the update interval. If it is necessary to appear in the integral form Then it can be expressed by step function or piecewise linear interpolation. At the update point τ = t u Previously, using the previous set of parameters, when τ ≥ t u Then switch to the new parameters.

[0125] 2. Gamma i (t) rolling calculation window:

[0126] In the first and fourth steps, Γ i (t) uses a sliding or rolling window Δt win To integrate or weight, it belongs to continuous or quasi-continuous real-time update; sampling and update cycle: it can be consistent with the main control cycle of the controller, or a shorter detection cycle can be set separately to improve the sensitivity of fault capture.

[0127] 3. Data collection and noise filtering:

[0128] Sampling frequency and noise characteristics: The sampling frequency during reciprocating testing is usually higher than the normal production monitoring frequency to capture small changes in the friction curve; the recommended frequency can reach hundreds to thousands of Hz, depending on the sensor and servo hardware. Data noise may come from the sensor itself (such as strain gauges, piezoresistive sensors) or hydraulic pulsation, mechanical vibration, etc.

[0129] Examples of filtering methods: Low-pass filtering: can smooth high-frequency noise and retain the main force-velocity / force-movement state Local fitting smoothing: such as Savitzky-Golay filtering, can suppress random noise as much as possible while retaining the inflection point information Kalman filtering: if the corresponding state equation is available, linear or extended Kalman filtering can be used to estimate the true force feedback

[0130] 4. Interpolation method and multiple update iterations:

[0131] Multiple updates: If multiple updates are made in different anomaly detection cycles Corresponding to multiple timestamps The third step is life prediction and preventive maintenance strategy, which requires these discrete Sequences and Benchmarks For comparison, linear interpolation or nearest point keeping can be used to process the parameters between update times; other interpolation or transition strategies: if a smoother transition is expected for the friction parameters, piecewise Bezier interpolation or spline interpolation can be used, but care should be taken to keep the model complexity controllable; within the control cycle, only when a new anomaly is detected and the test is completed, It will jump to avoid frequent parameter oscillations causing unstable control.

[0132] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0133] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0134] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only some logical function divisions. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0135] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0136] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A real-time pressure compensation method for a multi-axis linkage precision press, characterized in that: include, When it is detected that the friction increase coefficient of each axis exceeds the warning threshold continuously within the limited time window, based on the collected feedback data, the rolling integral and multi-threshold judgment algorithm is used to perform abnormal identification and generate abnormal axis information; After receiving the abnormal axis information, a short-stroke reciprocating motion test is performed on the specified axis under safe working conditions, and the force-displacement and force-velocity curve characteristics are recorded by high-frequency sampling. The friction model parameters are updated using a self-learning fitting algorithm to obtain the updated friction model parameters; Compare the historical multi-cycle parameter sequence with the current updated friction model parameters, calculate the life degradation index according to the high-order deviation accumulation function, and formulate a pre-maintenance strategy including replacement cycle, spare parts demand and priority, and output the remaining usable life and maintenance period in the data layer; When the predictive maintenance strategy is not executed in time and it is monitored that the friction increase coefficient continues to increase or the model compensation amount is close to the upper limit, real-time compensation is applied in the multi-axis synchronous control loop according to the updated friction model parameters, triggering the speed and load limit or emergency shutdown process.

2. The real-time pressure compensation method for a multi-axis linkage precision press according to claim 1, characterized in that: Predefine the sampling period and monitoring time window for each axis; record the real-time pressure, real-time speed, given reference pressure and reference speed for each axis; The friction increase coefficient is defined to evaluate the possible friction mutation of the shaft during the observation period; Real-time data collection is performed according to the sampling period, and the friction increase coefficient of axis i is updated at each sampling moment.

3. The real-time pressure compensation method for a multi-axis linkage precision press according to claim 2, characterized in that: One or more threshold curves are set for each axis. When the friction increase coefficient continuously exceeds the warning threshold and is difficult to recover in a short time, an abnormal mark is automatically generated and the abnormal axis information is recorded, including the axis number, trigger time and associated operating conditions; if the friction increase coefficient exceeds the serious threshold, a higher level of warning is immediately triggered.

4. The real-time pressure compensation method for a multi-axis linkage precision press according to claim 3, characterized in that: According to the output abnormal axis information, the target axis that needs to perform a small reciprocating test is selected, and its abnormal trigger time and abnormal operating conditions are obtained. The controller moves the axis to a safe working area within a controllable time period; a reciprocating motion curve is planned according to the given stroke amplitude and test speed range so that axis i repeats several short strokes in both the forward and reverse directions; the test force, actual displacement and actual speed of the axis are collected in real time.

5. The real-time pressure compensation method for a multi-axis linkage precision press according to claim 4, characterized in that: According to different speed ranges, the segment features in the force-speed curve are extracted, and the generalized friction model is used to represent the comprehensive friction force of shaft i; a high-order deviation accumulation function is defined to measure the overall error between the test data and the model output; A self-learning iterative algorithm is used to find the updated optimal friction model parameters so that the high-order deviation accumulation function is minimized. The optimal parameter vector is named as the updated friction model parameter and stored in association with the axis number.

6. The real-time pressure compensation method for a multi-axis linkage precision press according to claim 5, characterized in that: The updated friction model parameters are combined with historical parameters from previous tests and actual operation to form a time series; the gap between the updated friction model parameters and the friction model parameters under the baseline state is used to construct a comprehensive degradation measurement function; After obtaining the comprehensive degradation metric function corresponding to the current time point T, the historical sequence is compared to analyze its growth rate and fit the degradation curve; if it is predicted in subsequent operations that the comprehensive degradation metric function will exceed the corresponding threshold at time T+Δt, then Δt is the estimated value of the remaining useful life.

7. The real-time pressure compensation method for a multi-axis linkage precision press according to claim 6, characterized in that: Output the preventive maintenance strategy, including the spare parts requirement list, time schedule, and priority description of each axis, including: If the predicted remaining useful life of some shafts is close to zero or the wear rate is higher than expected, they will be marked as high-priority maintenance objects, otherwise they will be included in the periodic maintenance plan; for high-priority maintenance objects, a list of immediate or recent replacement needs will be generated based on the actual inventory situation; regular maintenance cycles will be established for the remaining shafts, and the usage of corresponding seals, lubricants or other wearing parts will be estimated.

8. The real-time pressure compensation method for a multi-axis linkage precision press according to claim 7, characterized in that: After reading the pre-maintenance strategy, the priority information and expected replacement cycle of each axis are obtained, and the updated friction model parameters and friction increase coefficient are obtained; in the main control cycle, differential budgeting is performed based on the actual values ​​of multi-axis synchronization instructions and sensor feedback; The above differential results are combined with the friction increase coefficient and the updated friction model parameters to calculate the instant compensation amount, and the real-time calculated compensation amount is injected into the multi-axis synchronous controller so that the axis with abnormal friction can obtain additional compensation or correction in force and speed.

9. The real-time pressure compensation method for a multi-axis linkage precision press according to claim 8, characterized in that: Add shafts that may experience greater friction deterioration or seal failure in the near future to the list of high-risk shafts; if it is detected that the friction increase coefficient continues to increase within a short time window and the compensation amount is also close to the upper limit allowed, the speed and load limit strategies are activated; Execute the set speed limit curve for the high-risk axis, reduce the maximum operating speed of axis i to a preset ratio, and limit the drive power output in the controller; if the trend is still not alleviated, continue to increase the restriction; if it is alleviated, maintain this speed limit state until the next maintenance is completed.

10. The real-time pressure compensation method for a multi-axis linkage precision press according to claim 9, characterized in that: Multiple safety thresholds are set; if any axis exceeds the threshold or the compensation amount reaches the limit within a short time window, the early warning mechanism is triggered; If the warning is a moderate risk, an alarm is issued and it is recommended to accelerate the replacement operation in the preventive maintenance strategy; if the warning is an extreme risk, the operating load of the axis or the entire equipment is immediately limited, and an emergency shutdown sequence is executed if necessary.

Images