A load precision control method and system based on staged sampling and IIR filtering

Abstract

The application discloses a load precision control method and system based on phased sampling and IIR filtering, which divides a loading process into an empty load phase, a loading phase and a load maintaining phase; a high sampling frequency is adopted in the empty load phase, and a low sampling frequency is adopted in the loading phase and the load maintaining phase; a first-order IIR low-pass filtering is adopted to process a load signal in the whole loading process; and IIR filtering coefficients are adaptively matched in different phases; when initial contact is judged to occur in the empty load phase, double-link control is adopted to make an emergency stop; total demand of the present loading is calculated in advance when entering the loading phase to plan a loading function, the function is composed of three parts of starting smooth acceleration, front straight line loading and rear deceleration smoothness, and has feedforward and feedback capabilities; and an executing mechanism is propelled according to a predetermined loading speed curve of the function; and a step-by-step creep compensation mechanism based on equivalent stiffness estimation is adopted in the load maintaining phase. The application can effectively improve loading efficiency, control precision, signal stability and use safety of equipment.

Description

Technical Field

[0001] This invention belongs to the field of servo control technology and relates to a load precision control method and system based on staged sampling and IIR filtering, which is applicable to electric loading equipment, servo-driven loading equipment or mechanical testing equipment, etc. Background Technology

[0002] In existing electric loading equipment, load signals are typically collected by load sensors during load control, converted by A / D converters, and then transmitted to the controller. The controller then controls the servo driver and servo motor through pulse output or communication to load the workpiece being measured by the loading mechanism.

[0003] In practical applications, the loading process typically includes stages such as no-load approach, contact recognition, formal loading, and load holding. Different stages place different demands on the control system's response speed, noise suppression capability, control stability, and safety protection capabilities. However, existing technologies usually employ fixed sampling frequencies, uniform filtering parameters, and a single control strategy, making it difficult to meet the control requirements of the entire process. For example:

[0004] (1) When there is a large empty stroke between the pressure head and the workpiece being measured, it is usually necessary to approach quickly in order to improve efficiency. However, rapid approach can easily cause an impact at the moment of contact, resulting in damage to the sensor or measurement distortion.

[0005] (2) Fixed sampling frequency schemes cannot simultaneously achieve fast response and low noise stability. High sampling frequency is beneficial for timely detection of contact points, but the signal noise is large and it is easy to cause misjudgment; although low sampling frequency helps to stabilize the signal, it is easy to miss rapid changes.

[0006] (3) Traditional moving average filtering usually introduces a significant delay, making it difficult to balance contact detection sensitivity and loading process tracking.

[0007] (4) Existing loading strategies mostly rely on real-time load error for cycle-by-cycle feedback adjustment, lacking pre-planning for the entire loading process. When approaching the target load, traditional deceleration strategies are prone to problems such as overshoot, large fluctuations, or large errors before entering steady-state control, and sudden speed changes at the start-up stage can easily generate additional shocks.

[0008] (5) During the load holding phase, the load change of the electric loading system mainly comes from the creep of the load sensor, the creep of the structure and the slow deformation of the sample. Conventional continuous PID control is not suitable for such slow and discontinuous change processes, and is prone to frequent action or poor stability.

[0009] (6) Existing systems usually only stop normal pulse output after contact detection. If there is pulse residue, driver response lag or mechanical inertia, it may still cause continued pressure. The sensor protection capability is insufficient.

[0010] In summary, existing technologies still have room for improvement in terms of contact recognition accuracy, rapid protection capability, overall process control stability, and load stability. Therefore, this invention proposes a precision load control method and system based on staged sampling and IIR filtering. By employing different sampling frequencies and filtering parameters according to the different characteristics of each loading stage, the aforementioned problems are effectively solved, balancing response speed and signal stability. Summary of the Invention

[0011] The purpose of this invention is to address the shortcomings of existing technologies by providing a precision load control method and system based on staged sampling and IIR filtering. The entire loading process is divided into three stages: no-load (including contact detection), loading, and load holding. Different sampling frequencies are used in different control stages, and matching IIR filtering parameters are configured for each sampling frequency. This enables the system to approach the workpiece at a high speed during the no-load stage and accurately identify the initial contact point. Upon detection of contact, the motion is quickly cut off, avoiding residual pulses and additional impacts caused by inertia. By pre-planning the loading process rather than relying entirely on real-time errors for cycle-by-cycle testing, loading efficiency is ensured while achieving shock-free start-up and smooth, overshoot-free arrival at the target load. Furthermore, considering the creep-dominated load attenuation characteristics of electric loading systems, stable and efficient load holding control is achieved.

[0012] The technical solution adopted in this invention is as follows:

[0013] A precise load control method based on staged sampling and IIR filtering includes:

[0014] The entire loading process is divided into three stages: the no-load stage, which includes the approach and contact judgment of the no-load travel; the loading stage, which includes the formal loading; and the load maintenance stage, which includes the load holding stage after the target load is reached and the allowable error is reached.

[0015] High sampling frequency is used in the no-load stage, and low sampling frequency is used in the loading and load holding stages. The load signal is processed by first-order IIR low-pass filter throughout the loading process, and the IIR filter coefficients are adaptively matched in different stages.

[0016] During the no-load phase, when initial contact is detected, a dual-link control emergency stop is adopted. When entering the loading phase, the total loading demand is pre-calculated to plan the loading function. This function consists of three parts: smooth acceleration at the start, linear loading in the first stage, and smooth deceleration in the second stage. It also has feedforward and feedback capabilities, so that the actuator can advance according to the loading speed curve predetermined by the function. During the load holding phase, a step-by-step creep compensation mechanism based on equivalent stiffness estimation is adopted.

[0017] In the above scheme, the high sampling frequency is 200Hz~2000Hz and the low sampling frequency is 10Hz~100Hz.

[0018] Furthermore, the adaptive matching IIR filter coefficients for different stages are as follows: In the no-load stage, the IIR filter coefficient is 0.20~0.40 during the approach phase and 0.08~0.20 during the contact judgment phase; in the loading stage, the IIR filter coefficient is 0.20~0.40 during the smooth acceleration at start and the initial straight-line acceleration, and 0.03~0.10 during the smooth deceleration phase; in the load-maintaining stage, the IIR filter coefficient is 0.03~0.10. The IIR filter coefficient α for each stage is related to the sampling frequency of that stage. Matching settings to adjust the equivalent filter delay of the first-order IIR filter. To meet the needs of this stage, i.e. .

[0019] Furthermore, during the no-load phase, when the filtered load signal shows an upward trend that conforms to the contact characteristics at several consecutive sampling points, it is determined that the loading mechanism has made initial contact with the workpiece. At this time, the controller adopts dual-link control for emergency stop, including stopping the pulse output to the servo driver and simultaneously outputting an effective level signal directly to the independent digital input terminal of the servo driver through the controller's independent digital output channel, triggering it to directly execute rapid stop, enable cut-off, or motion prohibition operations. The several consecutive sampling points refer to 3 to 8 consecutive sampling points, and the upward trend that conforms to the contact characteristics is: the difference between adjacent filtered values ​​of the several consecutive sampling points is continuously positive, the increment at each step is greater than the preset noise threshold, and the cumulative increase exceeds the preset contact determination threshold.

[0020] Furthermore, after initial contact is determined, the loading phase begins, in which the initial smooth acceleration segment of the loading function smoothly increases the speed from zero to a constant speed according to a smooth function. Specifically, it starts with a quadratic function. Accelerate smoothly from the start, then use an exponential function. Smooth approaching uniform speed The initial straight-line loading segment is at a speed The loading is performed at a constant speed, and the subsequent deceleration smoothing section uses a speed that decreases exponentially with the remaining load demand. ;in, The acceleration phase of the quadratic function ends at time c, where c is the corresponding initial acceleration. For the exponential approach to the time constant, This is the difference between the target load and the current filtered measured load, representing the remaining load requirement. It is the exponential smoothing constant.

[0021] Furthermore, during the loading phase, the speed command for each control cycle is the smaller value between the initial acceleration envelope and the deceleration envelope:

[0022]

[0023] This refers to the loading speed.

[0024] Furthermore, the step-by-step creep compensation mechanism based on equivalent stiffness estimation specifically includes:

[0025] The controller continuously monitors the deviation between the current load and the target load. When the detected deviation exceeds the preset load protection threshold E, it will take action. h If this occurs, a step-by-step compensation is initiated. The controller calculates the theoretical compensation displacement based on the current load deviation and equivalent stiffness K, and converts it into the number of compensation pulses. Simultaneously, the single compensation displacement must not exceed the preset maximum value. ,in =E h / K; the load threshold E h The value must be greater than the residual noise band after filtering and not greater than the allowable load-preserving error.

[0026] Furthermore, during the load maintenance phase, after each compensation is completed, the controller does not immediately perform another compensation, but instead enters a stabilization waiting phase: the servo system stops further action and waits for a preset time. Then, the current load is resampled and judged. If the pre-set compensation judgment conditions are met, the next compensation is triggered; otherwise, the current state is maintained and monitoring continues.

[0027] The present invention also provides a load precision control system based on staged sampling and IIR filtering, including a controller, a load sensor, a servo driver, and a loading mechanism. The controller outputs a pulse control signal to the servo driver through a control link, and at the same time, a digital output terminal of the controller outputs a digital level signal to a designated input terminal of the servo driver to realize rapid shutdown protection. The load signal detected by the load sensor is converted by an A / D converter and input to the controller to form a control closed loop. The controller is used to implement the method described in any of the above embodiments.

[0028] The present invention also provides a computer-readable storage medium storing computer-executable instructions, which, when executed, are used to implement the method described in any of the preceding claims.

[0029] The beneficial effects of the present invention are at least as follows:

[0030] (1) By dividing the loading process into three major stages—no load, loading, and load holding—and using different sampling frequencies in different stages, the fast approach efficiency and stable control accuracy are balanced.

[0031] (2) By setting IIR filtering parameters for each sampling frequency in stages, the availability of the load signal is improved, which is beneficial for contact point identification and also takes into account the control tracking of the loading stage.

[0032] (3) By combining high sampling rate and IIR filtering during contact judgment, the accuracy of contact point detection can be improved and noise misjudgment can be reduced.

[0033] (4) By setting up an independent digital output emergency stop protection link, the servo motion can be stopped more quickly at the moment of contact, reducing the additional impact caused by pulse residue and inertial pressure, and improving the protection capability of the load sensor.

[0034] (5) By pre-calculating the loading requirements and executing the loading according to the pre-planned loading function (smooth acceleration at the start - uniform linear loading at the front - exponential deceleration and smooth arrival at the rear with the remaining amount), the loading efficiency is guaranteed while achieving no impact at the start and smooth arrival at the target load without overshoot; its feedforward plus feedback composite implementation is robust to the equivalent stiffness estimation error and eliminates the need for independent fine adjustment.

[0035] (6) By adopting a step-by-step compensation strategy for creep-dominated scenarios, the stability and control targeting of the load-holding phase can be improved.

[0036] (7) Overall, it can improve the loading efficiency, control accuracy, signal stability and safety of the equipment. Attached Figure Description

[0037] Figure 1 This is a hardware structure block diagram of the system to which this invention applies.

[0038] Figure 2 This is a schematic diagram of the conventional control link and the independent emergency stop protection link.

[0039] Figure 3 This is a schematic diagram of the closed-loop control principle of the system of the present invention.

[0040] Figure 4 This is a flowchart illustrating the overall control process of the three stages of loading in the method of the present invention: the no-load stage, the loading stage, and the load-maintaining stage.

[0041] Figure 5 This is a schematic diagram of the control flow in stage one, i.e., the no-load stage, of the method of the present invention.

[0042] Figure 6 This is a flowchart of the pre-planned loading function control in stage two, namely the loading stage, of the method of the present invention.

[0043] Figure 7 This is a schematic diagram of the loading function (velocity curve and load curve) during the loading stage in the method of the present invention.

[0044] Figure 8 This is a flowchart of the step compensation process in stage three, namely the load maintenance stage, of the method of the present invention. Detailed Implementation

[0045] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0046] This embodiment uses an electric servo-driven precision loading device to compress a small-range mechanical sample as an example. The device consists of a controller, load sensor, servo driver, servo motor, and a lead screw loading mechanism with differential speed reduction. The load sensor has a range of 200N, the lead screw pitch is 5mm, and the servo motor drives the lead screw via a differential speed reduction ratio of 1:10. Each revolution of the servo motor corresponds to 10,000 control pulses, so the displacement of the loading mechanism corresponding to each pulse is approximately 5mm / (10×10000) = 0.05μm (i.e., the pulse equivalent is 0.05μm). The initial idle stroke between the indenter and the sample is approximately 10mm. The target load for this loading is 100N, and the allowable error E is ±0.02N. The entire loading process is executed sequentially in three stages: no-load, loading, and holding. The sampling frequency and IIR filter coefficient for each stage are configured as described above. In this embodiment, a high sampling frequency of 480Hz is used in the no-load stage, and a low sampling frequency of 60Hz is used in the loading and holding stages. The following section will further explain the specific control process for each stage in this scenario.

[0047] This invention proposes a precise load control method and system based on staged sampling and IIR filtering. This scheme mainly includes two core aspects:

[0048] Staged sampling frequency and IIR filter parameter settings: Different sampling frequencies are used in different control stages, and IIR filter parameters are configured to match each sampling frequency. This ensures detection sensitivity in the rapidly changing stage and signal stability in the quasi-static stage, taking into account both response speed and noise resistance.

[0049] The loading process is divided into three main stages: the no-load stage (including contact detection), the loading stage, and the load holding stage (the overall process is as follows). Figure 4 As shown in the figure, a corresponding control strategy is adopted for each stage; at the same time, the sampling frequency and IIR filter parameter configuration are implemented in each stage.

[0050] According to some embodiments of the present invention, the present invention is applicable to loading control systems comprising the following components, such as Figure 1 As shown:

[0051] (1) Host computer or human-machine interface, used for parameter setting, status display and data recording.

[0052] (2) Control board or controller, used to perform sampling, filtering, stage identification, contact determination, loading planning, compensation calculation and control output.

[0053] (3) Load sensor, used to detect the actual load.

[0054] (4) A / D conversion module, used to convert the analog signal output by the load sensor into a digital signal.

[0055] (5) Servo driver and servo motor, used to drive the movement of the loading mechanism.

[0056] (6) Mechanical loading mechanism, including pressure head, clamp, loading frame, screw mechanism, etc., used to apply load to the workpiece being tested.

[0057] (7) Independent digital output link, used to send rapid stop, emergency stop or prohibition signals to the servo drive in an emergency.

[0058] The control board has dual links for controlling emergency stops, such as... Figure 2 As shown, the control board outputs pulse control signals to the servo driver through the normal control link and outputs digital level signals to the designated input terminal of the servo driver through an independent digital output link to achieve rapid shutdown protection. The load signal detected by the load sensor is input to the control board via the A / D conversion module, forming a control closed loop, as shown below. Figure 3 .

[0059] This invention does not use a fixed sampling frequency throughout the entire loading process, but switches the sampling frequency according to different control stages:

[0060] (1) During the no-load stage (including the rapid approach and contact judgment during the no-load stroke), a higher sampling frequency is used to improve the detection sensitivity of load change and facilitate the timely detection of the initial contact between the pressure head and the workpiece;

[0061] (2) During the loading stage, since the loading stage is mainly quasi-static and more attention is paid to signal stability, the sampling frequency is switched to a lower frequency to reduce the noise amplitude and data fluctuation caused by high frequency sampling.

[0062] (3) During the load maintenance phase, continue to use a lower sampling frequency and combine it with the corresponding filtering parameters to meet the requirements of stable maintenance.

[0063] In one embodiment, the high sampling frequency can be set to 480Hz or 960Hz, and the low sampling frequency can be set to 60Hz, but is not limited to these values. Typically, the high sampling frequency can also be any value between 200Hz and 2000Hz, and the low sampling frequency can be between 10Hz and 100Hz. The reason for this setting is that the high sampling frequency is sensitive but has higher noise, making it suitable for capturing fast-changing events such as contact; the low sampling frequency itself has lower noise and a more stable signal, making it suitable for quasi-static loading and load maintenance. Thus, by switching in stages, both sensitivity and stability can be balanced.

[0064] To overcome the problem of large delay in moving average filtering, this invention employs a first-order IIR low-pass filter to process the load signal, preserving the low-frequency effective load variation trend while suppressing high-frequency noise during sampling. The control board filters the digital signal from the load sensor after A / D conversion, with the following expression:

[0065]

[0066] In the formula, This represents the input load value from the k-th sample. This is the output value of the k-th filter. The output value is the previous filter value; α is the IIR filter coefficient. The larger α is, the greater the weight of the new sampled value, resulting in a faster response but weaker smoothing; the smaller α is, the stronger the smoothing but the greater the phase lag. For further explanation, the sampling frequency... The time delay (phase lag) between the IIR filter coefficients α and the filter can be achieved through an equivalent filter delay using a first-order IIR filter. Mutual matching conversion: ,on the contrary Accordingly, the sampling frequency and filter coefficients for each stage are determined and mutually consistent according to the following principles: First, the sampling frequency is determined based on the highest effective frequency of the event to be detected and according to the Nyquist criterion. —During the no-load phase, contact transients need to be captured, so a high sampling frequency (e.g., 480Hz, 960Hz) is used; during the loading and holding phases, which are quasi-static processes, a low sampling frequency (e.g., 60Hz) is used; then, the acceptable filtering delay for each phase is considered. The filter coefficient α is calculated from the above formula to ensure that the response speed and smoothing intensity meet the requirements of this stage. Taking the parameters used in this paper as an example, the equivalent filter delay is as follows: rapid approach without moving (480Hz, 0.30). Approximately 5ms, contact detection (480Hz, 0.12) Approximately 15ms, uniform linear loading (60Hz, 0.30) Approximately 39ms, smooth deceleration to the desired point (60Hz, 0.05). Approximately 317ms, load-keeping step compensation (60Hz, 0.08) Approximately 192ms; it can be seen that the filter delay increases as the control stage changes from a fast-changing process to a quasi-static process, and the smoothing intensity is correspondingly enhanced. The sampling frequency and the filter coefficient are consistent and coordinated with each other.

[0067] In this invention, the IIR filter parameters are not fixed, but are set separately to match the sampling frequency and the current stage. The correspondence between the sampling frequency and the IIR coefficients for each stage, and the reasons for their settings, are shown in the table below:

[0068]

[0069] The filtering parameters for each stage can be determined through pre-testing. This involves analyzing the static fluctuation range of the sensor under different sampling frequencies and filtering parameters to select the appropriate parameter combination for each stage. For example, the IIR filtering coefficient α for the five sub-processes—rapid approach during idle travel, contact judgment, uniform linear loading, smooth deceleration to position, and load-holding step compensation—can be set to 0.30, 0.12, 0.30, 0.05, and 0.08, respectively.

[0070] like Figure 4 As shown, this invention divides the entire control process into three main stages: the no-load stage (including contact judgment), the loading stage, and the load-maintaining stage. Each stage employs a corresponding control strategy based on the aforementioned sampling frequency and IIR filter parameters. Specifically:

[0071] Phase 1: No-load phase (including no-load approach and contact judgment)

[0072] like Figure 5 As shown, this stage includes rapid approach during idle travel, contact detection, and independent emergency stop protection after contact. Sampling and filtering configuration: A higher sampling frequency (such as 480Hz or 960Hz) is used; α is set to 0.3 during the rapid approach process to ensure tracking performance, and α is set to 0.12 during the contact judgment process to enhance smoothness and suppress noise, making it easier to distinguish between actual contact and sensor noise fluctuations within an acceptable delay.

[0073] The contact detection mechanism works as follows: During the rapid approach phase of the no-stroke operation, the pressure head approaches the workpiece at a high speed. The controller continuously monitors subtle changes in the load signal using a high sampling frequency and IIR filtering. When the filtered load signal shows an upward trend consistent with contact characteristics at several consecutive sampling points, it is determined that the pressure head and the workpiece have made initial contact. This detection mechanism does not rely solely on whether a single sample value is greater than zero, but combines high-frequency sampling, filtering, and fluctuation characteristic analysis to distinguish between the actual load change caused by contact and sensor noise fluctuations. Specifically, the "several consecutive sampling points" refers to 3 to 8 consecutive sampling points, preferably 4; the "upward trend consistent with contact characteristics" means that the following three conditions are met simultaneously:

[0074] ① Monotonicity, that is, the difference between adjacent filtered values ​​of a series of consecutive sampling points. The value remains positive.

[0075] ②Significance, meaning the increment at each step is greater than the noise band. Where σ is the pre-calibrated standard deviation of sensor noise, which can be taken as 0.1% of full scale;

[0076] ③ Cumulative, meaning that the cumulative increase ΣΔy of the several sampling points exceeds the preset contact determination threshold F. th F th =6σ.

[0077] This distinguishes the continuous monotonic rise caused by contact from the random noise fluctuations of the sensor's zero-mean value, avoiding misjudgments caused by single-point noise spikes. During the no-load phase, the rapid approach and contact judgment processes are automatically differentiated based on the degree of load rise relative to the baseline after filtering: a pre-trigger threshold F is set. pre =3σ and contact confirmation threshold F th When the filter load is within the noise band and does not exceed F... pre At that time, during the rapid approach process, α is set to 0.30 to ensure tracking and rapid response to load increases; when the filtered load first exceeds F... pre If a suspected lift occurs, the contact judgment process is initiated, α is reduced to 0.12 to enhance smoothness, and further confirmation is performed based on the three conditions for 3 to 8 consecutive sampling points as described above; if the three conditions are not met within the set number of sampling points and the load falls back to F... pre The following is considered a noise disturbance, and the rapid approach process is resumed.

[0078] Independent emergency stop protection mechanism: During the high-speed approach phase, even if the controller detects contact and stops pulse output, the pressure head may continue to descend due to reasons such as the preceding pulse not being fully executed, driver response delay, and mechanical inertia. Therefore, this invention sets up an independent fast stop protection channel in addition to the conventional control link from the control board to the servo driver: the digital output terminal DO of the control board is directly connected to the digital input terminal DI or DIA of the servo driver (see...). Figure 2 This input terminal does not participate in normal pulse position control; it is only used to receive emergency stop, prohibition, cut-off, or rapid stop signals.

[0079] Upon detecting initial contact, the controller, in addition to terminating normal pulse output, immediately outputs a valid level signal to the servo driver via this independent digital output channel, triggering it to directly execute rapid stop, enable cut-off, or motion prohibition operations. Because this protection signal does not require processing by the host computer software, communication transmission, or conventional pulse command queuing, its response path is short and its action speed is fast. It can operate independently of the software control link, thus more effectively truncating subsequent motion corresponding to residual pulses, significantly reducing the impact load at the moment of contact, and protecting small-range or high-precision load sensors.

[0080] Phase Two: Loading Phase – A Loading Method Based on Pre-planned Loading Functions

[0081] Unlike existing loading methods that rely on real-time load errors and iterative loading cycle by cycle, this invention pre-calculates the total loading requirements upon entering the loading phase and plans a loading function (loading speed planning curve) accordingly. This allows the actuator to advance according to the predetermined function, achieving shock-free start-up, rapid initial loading, and smooth, overshoot-free final loading. Figure 6 As shown.

[0082] (a) Pre-calculation of loading requirements

[0083] Upon entering the loading phase, i.e. after contact determination, let the contact point load be... (Approximately 0), target load is The system's equivalent stiffness is K. This equivalent stiffness can be obtained through initial micro-test loading calibration after contact or pre-calibration, i.e., calculated from the ratio of load increment to displacement increment at adjacent sampling times, K = ΔF / Δx. Therefore, the total load increment and total displacement requirement for this loading are:

[0084]

[0085]

[0086] in For contact point load, Let K be the target load, K be the equivalent stiffness of the system, ΔF be the load change, and Δx be the displacement change.

[0087] (ii) Loading function (loading speed planning curve)

[0088] The speed command for the loading process is planned as a pre-defined function, which consists of two parts: a straight-line loading phase and a smooth deceleration phase. A smooth acceleration is introduced at the start, specifically comprising three phases:

[0089] (a) Initial acceleration phase: To avoid the starting shock caused by a speed jump, the speed is increased from zero to a constant speed v according to a smooth function. maxIt can be expressed as an exponential function:

[0090]

[0091] That is, in an extremely short start-up time The inner function first uses a quadratic function to smooth the process, then approaches it using an exponential function. This makes the velocity and its derivative (acceleration) continuous.

[0092] in The acceleration phase of the quadratic function ends at time c, where c is the initial acceleration. The constant speed loading speed is the upper limit of the loading stage speed. For the exponential approach time constant, This refers to the initial speed.

[0093] (b) Initial uniform linear loading segment: After the start is completed, Uniform loading. When the equivalent stiffness K is approximately constant, the load increases approximately linearly with time, thus achieving rapid and efficient loading.

[0094] (c) Smooth deceleration phase at the end: depending on the remaining load requirements. As the velocity decreases, it smoothly declines according to a monotonically decreasing function until it approaches zero near the target, achieving a smooth, overshoot-free deceleration. Deceleration can be achieved by exponential decay with the remaining quantity.

[0095]

[0096] in For the remaining load, The current measured load (obtained by IIR filtering) Let λ be the speed during the deceleration phase and λ be the exponential smoothing constant.

[0097] The corresponding loading function velocity curve and load curve for this process are as follows: Figure 7 As shown.

[0098] (III) Comprehensive speed command

[0099] The speed command for each control cycle is the smaller value between the initial acceleration envelope and the deceleration envelope.

[0100]

[0101] This refers to the loading speed.

[0102] Note: When When larger ≈ This corresponds to the previous uniform linear loading segment; when When smaller The deceleration then decreases exponentially, corresponding to the subsequent smooth deceleration phase, and the two connect naturally without explicit switching. Here, λ is the length of the smooth deceleration phase, determining the starting position and the degree of smoothness of the deceleration segment; This determines the speed of acceleration from a standstill. The controller will [process the acceleration] each cycle. Convert the displacement increment and corresponding pulse number for that cycle and output them.

[0103] To further explain, the statement that "the speed command for each control cycle is the smaller value of the starting acceleration envelope and the deceleration envelope" is the definition of the loading function itself: the loading function is not a pre-calculated closed curve, but rather constructed by taking the smaller value of the two envelopes point by point for each control cycle, i.e. The initial acceleration envelope vinitial is driven by time (feedforward term), smoothly accelerating from zero to a constant speed v0. max After maintaining v max Deceleration envelope v 减速 =λ·ΔF is driven by the measured residual load (feedback term), and its value is greater than v when the target is far away. max When close to the target, it drops to v max Therefore, the smaller result before loading is v. max (Corresponding to the constant speed segment), when approaching the target, the smaller result is converted to v deceleration (corresponding to the deceleration segment). The two segments are naturally connected without explicit judgment of the switching point, and at the same time, the combination of feedforward and feedback is realized; therefore, the speed command is taken from the loading function, which is defined as the periodic minimization of the above two envelopes. The method for determining the values ​​of the exponential approach time constant τ and the exponential smoothing constant λ is as follows: τ determines the speed at which the initial segment smoothly transitions from secondary acceleration to constant speed. To ensure that the acceleration is continuous at the junction of the secondary segment and the exponential segment, τ is taken as (v max - v(t1)) / (c·t1) (t1 is the end time of the second acceleration segment, and c is the starting acceleration). In engineering, the starting is to be completed within 3% to 5% of the total loading time, that is, τ is approximately 2 to 5 control cycles, and it satisfies the condition of the maximum allowable acceleration a. max The defined constraint τ≥(v) max -v(t1)) / a max λ determines the starting position and smoothness of the subsequent deceleration, which is determined by the deceleration starting point, i.e., when the deceleration envelope decreases to v. max Deceleration begins at time, corresponding to the remaining load ΔF * =v max / λ, therefore we take λ=v max / ΔF * , where ΔF * The time constant for the final approximation can be 10% to 20% or (3 to 10) times the allowable error E, where τ is the total load increment. dec =1 / (K·λ), and it is advisable to take 3 to 5 control cycles to balance rapid positioning and no overshoot.

[0104] (iv) Combined feedforward and feedback

[0105] The above remaining amount The load signal, after being measured and filtered by IIR, is recalculated in real time, giving the loading function both feedforward planning and feedback correction capabilities: even if there is an estimation error in the equivalent stiffness K, the loading process can still automatically converge with the measured load; the exponential decay at the end naturally provides fine approximation capability, so there is no need to set up a separate trial-and-error fine adjustment stage.

[0106] (v) Exit Conditions

[0107] When the absolute value of the difference between the current load and the target load satisfies (E represents the allowable error) and when the system remains within this range for a preset time or for four consecutive sampling periods, the system exits the loading phase and enters the load maintenance phase.

[0108] Sampling and Filtering Configuration: In this stage, the sampling frequency is switched to a lower frequency (e.g., 60Hz). For the uniform linear loading segment, α is set to 0.3 to ensure tracking and avoid control lag. For the deceleration smoothing segment, α is set to 0.05 to enhance smoothness and suppress final-stage fluctuations and overshoot. It should be noted that although the loading stage includes three motion segments—starting smooth acceleration, initial uniform linear loading, and subsequent deceleration smoothing—only two sets of IIR filter coefficients are needed: the starting smooth acceleration segment and the initial uniform linear loading segment have similar load dynamic bandwidths and both require relatively fast tracking, so they share the same larger coefficient (0.20~0.40); the subsequent deceleration smoothing segment requires stronger smoothing to suppress final-stage fluctuations and overshoot, so a smaller coefficient (0.03~0.10) is used separately. Therefore, it is not necessary to set three sets of coefficients for each of the three motion segments.

[0109] Phase 3: Load Holding Phase Step Compensation

[0110] like Figure 8 As shown, the sampling and filtering configuration is as follows: a lower sampling frequency (e.g., 60Hz) is used in this stage, and α is set to 0.08 to enhance smoothness and suppress false triggering caused by creep signal noise.

[0111] After the system enters the load protection phase, the controller continuously monitors the current load. With target load The deviation between; when the detected load decrease meets the preset compensation trigger condition (i.e. Furthermore, when the current load is lower than the target load (a situation that is more common), a step-by-step compensation is initiated to avoid unnecessary actions for minor fluctuations or noise disturbances. E h The load protection threshold is E. hThe principle for determining its value is as follows: its lower limit should be greater than the residual noise band after filtering (e.g., take 3σ, where σ is the standard deviation of load noise during the load holding phase) to avoid noise causing false triggering of compensation, leading to frequent actions and oscillations; its upper limit should not be greater than the allowable load holding error E to ensure that the load drop does not exceed the accuracy requirements before triggering compensation; therefore, E h Satisfying 3σ < E h ≤E, typically taken as 0.5%~2% of the target load or (3~5) times the noise standard deviation. Furthermore, the maximum single-cycle compensation displacement is taken as Δx. max = E h / K, meaning that a single compensation amount exactly replenishes a threshold E. h The corresponding load deviation, in principle, avoids overshoot caused by excessive compensation in a single instance.

[0112] After compensation is triggered, the controller calculates the theoretical compensation displacement based on the current load error and the equivalent stiffness K. This can be further converted into the number of compensation pulses; to avoid overshoot caused by excessive compensation in a single instance, the displacement in a single compensation shall not exceed the preset maximum value. Where Δx is the theoretical compensation displacement, = E h / K.

[0113] After each compensation is completed, the controller does not immediately perform another compensation, but instead enters a stabilization waiting phase: the servo system stops further action and waits for a preset waiting time. Then, the current load is resampled and reassessed. If the compensation conditions are still met, the next compensation is triggered; if the error is less than the load protection threshold E... h The current state will be maintained and monitoring will continue. To improve stability, a continuous judgment condition can be added (compensation will only be performed when the trigger condition is met for four consecutive cycles) and the minimum compensation interval can be set to 10 cycles.

[0114] In this way, the system does not continuously perform high-frequency closed-loop adjustment during the load holding phase. Instead, it only performs discrete small-step compensation as needed when the load decreases substantially due to sensor creep, structural creep, or slow deformation of the sample. This makes it more suitable for handling the slow and discontinuous load decay process in electric loading systems, reducing frequent corrections, lowering the risk of oscillation, and improving the stability and accuracy of long-term load holding.

[0115] The method of the present invention is compared with a traditional loading control scheme that uses a fixed sampling frequency, uniform filtering parameters, and real-time load error feedback adjustment on a cycle-by-cycle basis, under the same device and operating conditions (200N range sensor, target load 100N, allowable error ±0.02N, pulse equivalent 0.05μm) in the above embodiments:

[0116] (1) In terms of contact protection, the traditional solution only stops the pulse output after detecting contact. Due to the influence of pulse residue, driver response lag and mechanical inertia, the pressure may continue to be applied at the moment of contact, resulting in a large impact overload, which can easily damage the 200N small range sensor. In this embodiment, the initial contact is accurately identified by a high sampling frequency of 480Hz combined with IIR filtering, and the servo emergency stop is directly triggered through an independent digital output link, which can significantly reduce the impact overload at the moment of contact and effectively protect the sensor.

[0117] (2) In terms of loading efficiency and positioning accuracy, traditional real-time trial deceleration is prone to overshoot, fluctuations in the final stage, or repeated adjustments when approaching the target. This embodiment executes according to the pre-planned loading function, with uniform and rapid loading in the first stage and exponential smooth positioning in the final stage according to the remaining load demand. While ensuring loading efficiency, it achieves no impact at the start and no overshoot in positioning. After positioning, the steady-state error is kept within the allowable error of ±0.02N. Moreover, thanks to the combined implementation of feedforward and feedback, it is robust to the estimation error of the equivalent stiffness K, eliminating the need for independent fine adjustment.

[0118] (3) In terms of load stability, traditional continuous PID control is not well adapted to slow, discontinuous load decay mainly characterized by creep, and is prone to frequent operation or oscillation. This embodiment adopts step-by-step creep compensation based on equivalent stiffness estimation and combined with post-compensation stability judgment and waiting, and only when the load substantially decreases beyond the load protection threshold E. h Discrete small-step compensation is performed only as needed to maintain load fluctuations within the threshold E during the load protection phase. h Within this range, it reduces frequent corrections, lowers the risk of oscillations, and improves the stability and accuracy of long-term load maintenance.

[0119] In summary, this embodiment has significant advantages over traditional fixed-parameter, real-time trial-and-error loading control schemes in terms of contact protection, loading efficiency and positioning accuracy, and load-holding stability.

[0120] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0121] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0122] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0123] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0124] The embodiments described above are merely some preferred embodiments of the present invention, and are not intended to limit the invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, all technical solutions obtained by equivalent substitution or equivalent transformation fall within the protection scope of the present invention.

Claims

1. A precise load control method based on staged sampling and IIR filtering, characterized in that, include: The entire loading process is divided into three stages: the no-load stage, which includes the approach and contact judgment of the no-load travel; the loading stage, which includes the formal loading; and the load maintenance stage, which includes the load holding stage after the target load is reached and the allowable error is reached. A high sampling frequency is used during the no-load phase, while a low sampling frequency is used during the loading and load-holding phases. A first-order IIR low-pass filter is used to process the load signal throughout the loading process. The IIR filter coefficients are adaptively matched for different phases: during the no-load phase, the IIR filter coefficient is 0.20~0.40 during the approach phase and 0.08~0.20 during the contact judgment phase; during the loading phase, the IIR filter coefficient is 0.20~0.40 during the initial smooth acceleration and initial straight-line acceleration phases, and 0.03~0.10 during the subsequent deceleration smoothing phase; and during the load-holding phase, the IIR filter coefficient is 0.03~0.

10. The IIR filter coefficient α for each phase is related to the sampling frequency of that phase. Matching settings to adjust the equivalent filter delay of the first-order IIR filter. To meet the needs of this stage, i.e. ; During the no-load phase, when initial contact is detected, a dual-link control emergency stop is adopted. When entering the loading phase, the total loading demand is pre-calculated to plan the loading function. This function consists of three parts: smooth acceleration at the start, linear loading in the first stage, and smooth deceleration in the second stage. It also has feedforward and feedback capabilities, so that the actuator can advance according to the loading speed curve predetermined by the function. During the load holding phase, a step-by-step creep compensation mechanism based on equivalent stiffness estimation is adopted.

2. The load precision control method based on staged sampling and IIR filtering according to claim 1, characterized in that, The high sampling frequency is 200Hz~2000Hz, and the low sampling frequency is 10Hz~100Hz.

3. The load precision control method based on staged sampling and IIR filtering according to claim 1, characterized in that, During the no-load phase, when the filtered load signal shows an upward trend that conforms to the contact characteristics at several consecutive sampling points, it is determined that the loading mechanism has made initial contact with the workpiece. At this time, the controller adopts dual-link control for emergency stop, including stopping the pulse output to the servo driver and simultaneously outputting an effective level signal directly to the independent digital input terminal of the servo driver through the controller's independent digital output channel, triggering it to directly execute rapid stop, enable cut-off, or motion prohibition operations. The several consecutive sampling points refer to 3 to 8 consecutive sampling points, and the upward trend that conforms to the contact characteristics is: the difference between adjacent filtered values ​​of the several consecutive sampling points is continuously positive, the increment at each step is greater than the preset noise threshold, and the cumulative increase exceeds the preset contact determination threshold.

4. The load precision control method based on staged sampling and IIR filtering according to claim 1, characterized in that, After initial contact is determined, the loading phase begins. In the loading function, the initial smooth acceleration phase accelerates the speed from zero to a constant speed using a smooth function. Specifically, it starts with a quadratic function. Accelerate smoothly from the start, then use an exponential function. Smooth approaching uniform speed The initial straight-line loading segment is at a speed The loading is performed at a constant speed, and the subsequent deceleration smoothing section uses a speed that decreases exponentially with the remaining load demand. ;in, The acceleration phase of the quadratic function ends at time c, where c is the corresponding initial acceleration. For exponentially approaching time constant, This is the difference between the target load and the current filtered measured load, representing the remaining load requirement. It is the exponential smoothing constant.

5. The load precision control method based on staged sampling and IIR filtering according to claim 4, characterized in that, During the loading phase, the speed command for each control cycle is the smaller value between the initial acceleration envelope and the deceleration envelope: ， This refers to the loading speed.

6. The load precision control method based on staged sampling and IIR filtering according to claim 1, characterized in that, The step-by-step creep compensation mechanism based on equivalent stiffness estimation specifically includes: The controller continuously monitors the deviation between the current load and the target load. When the detected deviation exceeds the preset load protection threshold E, it will take action. h If this occurs, a step-by-step compensation is initiated. The controller calculates the theoretical compensation displacement based on the current load deviation and equivalent stiffness K, and converts it into the number of compensation pulses. Simultaneously, the single compensation displacement must not exceed the preset maximum value. ,in =E h / K; the load threshold E h The value must be greater than the residual noise band after filtering and not greater than the allowable load-preserving error.

7. The load precision control method based on staged sampling and IIR filtering according to claim 6, characterized in that, During the load maintenance phase, after each compensation is completed, the controller does not immediately perform another compensation, but instead enters a stabilization waiting phase: the servo system stops further action and waits for a preset time. Then, the current load is resampled and judged. If the pre-set compensation judgment conditions are met, the next compensation is triggered; otherwise, the current state is maintained and monitoring continues.

8. A load precision control system based on staged sampling and IIR filtering, comprising a controller, a load sensor, a servo driver, and a loading mechanism, characterized in that, The controller outputs pulse control signals to the servo driver through the control link, and at the same time, a digital output terminal of the controller outputs a digital level signal to a designated input terminal of the servo driver to realize fast shutdown protection; the load signal detected by the load sensor is converted by A / D and input to the controller to form a control closed loop, and the controller is used to implement the method as described in any one of claims 1 to 7.

9. A computer-readable storage medium storing computer-executable instructions, which, when executed, are used to implement the method of any one of claims 1 to 7.

Images