A rudder priority control method based on double position loop feedback

Abstract

The application relates to the technical field of steering engine control, in particular to a steering engine priority control method based on double position loop feedback, which comprises the following steps: synchronously collecting potentiometer angle readings and magnetic encoder angle readings of a steering engine, and calculating an instantaneous deviation amount; processing the instantaneous deviation amount based on an adaptive smoothing algorithm to obtain a deviation baseline estimated value; calculating a deviation residual based on the instantaneous deviation amount and the deviation baseline estimated value, and performing fault identification based on the deviation residual, and triggering priority switching when a potentiometer fault is determined; in response to the priority switching, extracting a deviation baseline snapshot value at the switching trigger moment to compensate the magnetic encoder angle readings, and sending an equivalent position feedback value after compensation into a position loop. The application effectively eliminates the step jump injected into the position loop at the switching moment, and reduces the transient overshoot amplitude.

Description

Technical Field

[0001] This application relates to the field of servo control technology, specifically to a servo priority control method based on dual position loop feedback. Background Technology

[0002] In applications with extremely high reliability requirements, such as aerospace, weaponry, and industrial automation, servo motors, as core electric actuators, play a crucial role in precise angle control. The stability of their position feedback system directly determines the overall control accuracy and mission success rate. In harsh service environments with high vibration and wide temperature ranges, single-sensor feedback architectures are prone to single-point failure. Once a sensor fails, the position loop loses its feedback basis, leading to servo motor malfunction.

[0003] To address the single-point failure problem, the industry has gradually introduced dual-sensor redundant feedback schemes. A typical approach is to combine a diaphragm potentiometer and a magnetic encoder, using the potentiometer as the high-priority primary feedback and the magnetic encoder as the low-priority backup feedback, and integrating a fault identification program into the control algorithm: when the system detects an anomaly in the high-priority sensor, it automatically switches to the low-priority sensor to continue control.

[0004] However, due to differences in physical installation location, thermal expansion coefficient, and contact wear characteristics between diaphragm potentiometers and magnetic encoders, a dynamic drift deviation accumulates between their position readings over service time. Current solutions do not track or model this deviation during normal operation; instead, they directly inject the original magnetic encoder reading into the PID position loop upon triggering a switch. Since the feedback input of the position loop at the moment of switching generates a step jump equal to the current accumulated deviation, this step disturbance is amplified by both the proportional and derivative terms of the PID controller. In applications with large load inertia or stringent tracking accuracy requirements, this can easily cause transient overshoot or even oscillation of the output shaft, leading to mechanical damage or mission failure. Summary of the Invention

[0005] This application aims to solve the technical problem that existing dual-sensor redundant feedback schemes cause a step jump in the position loop due to accumulated deviation at the moment of switching, which in turn leads to transient overshoot of the output shaft.

[0006] This application provides a servo priority control method based on dual position loop feedback, comprising: synchronously acquiring potentiometer angle readings and magnetic encoder angle readings of the servo, and calculating instantaneous deviation; processing the instantaneous deviation based on an adaptive smoothing algorithm to obtain a deviation baseline estimate; calculating a deviation residual based on the instantaneous deviation and the deviation baseline estimate, and performing fault identification based on the deviation residual, triggering priority switching when a potentiometer fault is determined; responding to the priority switching, extracting the deviation baseline snapshot value at the switching trigger time to compensate the magnetic encoder angle reading, and sending the compensated equivalent position feedback value into the position loop.

[0007] This application transforms the dynamic deviation baseline between the two sensors into an actively managed system state quantity by continuously tracking the deviation baseline during normal operation. It also compensates the magnetic encoder readings with a baseline snapshot when priority switching is triggered. This ensures the smoothness of the switching process while effectively eliminating the step jump injected into the PID position loop at the moment of switching, and significantly reducing the transient overshoot of the output shaft.

[0008] Preferably, the step of processing the instantaneous deviation based on the adaptive smoothing algorithm to obtain the deviation baseline estimate includes: calculating the instantaneous rate of change of the instantaneous deviation; dynamically selecting a corresponding smoothing factor based on the comparison result of the instantaneous rate of change and a preset rate threshold; and using the selected smoothing factor to perform exponentially weighted moving average processing on the instantaneous deviation to obtain the deviation baseline estimate.

[0009] Preferably, the step of dynamically selecting a corresponding smoothing factor based on the comparison result between the instantaneous change rate and the preset rate threshold includes: selecting a first smoothing factor when the instantaneous change rate is greater than the preset rate threshold; and selecting a second smoothing factor when the instantaneous change rate is less than or equal to the preset rate threshold; wherein the first smoothing factor is greater than the second smoothing factor.

[0010] Compared to traditional fixed-parameter filtering methods, the adaptive smoothing factor dynamic selection design of this invention provides sufficient noise suppression capability when the deviation baseline is in a normal, slowly changing condition, and provides fast following capability when the deviation accelerates and drifts.

[0011] Preferably, the step of fault identification based on the deviation residual and triggering priority switching when a potentiometer fault is determined includes: comparing the absolute value of the deviation residual with a preset fault judgment threshold; when the absolute value is greater than the preset fault judgment threshold, performing continuous trigger counting; when the number of continuous triggers exceeds a preset count limit, determining that the potentiometer has malfunctioned, and finally triggering the priority switching.

[0012] While effectively filtering out false triggers caused by vibration and shock, it can maintain the shortest response delay for real persistent faults.

[0013] Preferably, before comparing the absolute value of the deviation residual with the preset fault decision threshold, the method further includes: collecting a preset number of deviation residual samples during the calibration stage and calculating the mean square error of the deviation residual samples; determining the preset fault decision threshold based on the mean square error; counting the maximum number of consecutive triggering cycles caused by mechanical vibration during the calibration stage and determining the preset count upper limit based on the maximum number of consecutive triggering cycles, and finally obtaining the preset fault decision threshold and the preset count upper limit for fault identification.

[0014] Preferably, the step of extracting the deviation baseline snapshot value at the switching trigger time to compensate the magnetic encoder angle reading and sending the compensated equivalent position feedback value into the position loop includes: recording the switching trigger time when the priority switching is triggered; extracting the deviation baseline estimate value corresponding to the switching trigger time as the deviation baseline snapshot value; and superimposing the deviation baseline snapshot value with the magnetic encoder angle reading at the current time to obtain the equivalent position feedback value.

[0015] Preferably, after sending the compensated equivalent position feedback value into the position loop, the method further includes: obtaining the first noise mean square error of the potentiometer and the second noise mean square error of the magnetic encoder; calculating the ratio of the first noise mean square error to the second noise mean square error; and using the ratio to perform proportional correction on the original differential gain of the position loop to finally obtain the corrected differential gain.

[0016] Preferably, the step of proportionally correcting the original differential gain of the position loop using the ratio to obtain the corrected differential gain includes: multiplying the original differential gain by the ratio to obtain the corrected differential gain; keeping the proportional gain and integral gain of the position loop unchanged, applying the corrected differential gain to the differential channel of the position loop, and finally achieving equal-amplitude noise correction of the differential gain after switching.

[0017] The differential gain is corrected by equal amplitude based on the ratio of the root mean square error of the two sensor noises, which effectively suppresses high-frequency jitter caused by sudden changes in sensor noise characteristics and ensures the stable operation of the position loop after switching.

[0018] Preferably, the synchronous acquisition of the potentiometer angle reading and the magnetic encoder angle reading of the servo motor, and the calculation of the instantaneous deviation, includes: synchronously acquiring the analog voltage signal output by the potentiometer and the digital angle signal output by the magnetic encoder in each control cycle; converting the analog voltage signal to obtain the potentiometer angle reading, and parsing the digital angle signal to obtain the magnetic encoder angle reading; subtracting the magnetic encoder angle reading from the potentiometer angle reading to finally calculate the instantaneous deviation.

[0019] Preferably, after sending the compensated equivalent position feedback value into the position loop, the method further includes: using the magnetic encoder as the main feedback source; stopping the update of the deviation baseline estimate; and continuously superimposing the deviation baseline snapshot value as a fixed compensation amount onto subsequent magnetic encoder angle readings, ultimately maintaining continuous compensation before system reset.

[0020] This application continuously tracks the dynamic deviation baseline between two sensors during normal operation and compensates the magnetic encoder reading with the deviation baseline snapshot value when switching is triggered. This eliminates the step jump injected into the position loop at the moment of switching, effectively suppresses the synchronous amplification of the step disturbance by the proportional and derivative terms of the PID controller, significantly reduces the transient overshoot of the output shaft, and ensures the continuity and stability of the servo position control under harsh conditions of high vibration and wide temperature range. Attached Figure Description

[0021] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent upon reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of the invention are illustrated by way of example and not limitation, and like or corresponding reference numerals denote like or corresponding parts, wherein: Figure 1 This is a flowchart illustrating a servo priority control method based on dual position loop feedback according to the present invention.

[0022] Figure 2 This is a schematic diagram of dual-sensor deviation baseline tracking and priority switching triggering in an embodiment of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0025] This invention discloses a servo priority control method based on dual position loop feedback, referring to... Figure 1 This includes steps S1-S4: S1. Simultaneously acquire signals from both sensors and track the deviation baseline.

[0026] In an optional embodiment, the control system synchronously acquires the analog voltage signal output by the diaphragm potentiometer and the digital angle signal output by the magnetic encoder MT6701 within each control cycle. The analog voltage signal from the potentiometer is processed via ADC conversion to obtain the current potentiometer angle reading, and the digital angle signal from the magnetic encoder is processed via communication parsing to obtain the current magnetic encoder angle reading. Both signals are expressed in degrees (°). The sampling frequency of both signals is synchronized with the main control loop to ensure timestamp alignment.

[0027] Furthermore, within each control cycle, the system calculates the instantaneous deviation between the two sensors:

[0028] in, For potentiometer angle readings, For the angle reading of the magnetic encoder, This comprehensively reflects the combined effects of installation eccentricity, thermal drift, and contact wear at the current moment. Since all of these factors are gradually changing processes, Under normal operating conditions, it exhibits low-frequency, slowly varying characteristics, with short-term fluctuations primarily originating from the discrete noise and electrical noise of each of the two sensors. To address this... After removing high-frequency noise and extracting the true slowly varying deviation baseline, the deviation sequence can be smoothed using an adaptive exponential weighted moving average algorithm, the standard form of which is:

[0029] in, This is the baseline deviation estimate at the current moment. As a smoothing factor, During the calibration phase after the system is powered on, the initial value is assigned by the initial sample mean. The calibration phase requires the servo to be stationary at a known angle position.

[0030] Next, the system dynamically adjusts the smoothing factor based on the instantaneous rate of change of the deviation. The rate of change of deviation During the calibration phase, statistics were collected on normal operating conditions. upper quartile As a preset rate threshold, this value is calculated from historical sampling data of the servo motor completing its full-stroke reciprocating motion within the rated temperature range, reflecting the upper limit of the normal gradually changing rate. When When this occurs, it indicates that the deviation is undergoing accelerated changes beyond the normal thermal drift rate, at which point the processor applies a larger smoothing factor. As the first smoothing factor, it enables the baseline to quickly track the accelerated drift of the deviation; when When the deviation rate is within the normal, gradual range, the processor selects a smaller smoothing factor. As a second smoothing factor, it makes the baseline smoothing window longer, thus effectively suppressing noise.

[0031] and The specific value is determined by the control cycle. The target time constant is derived using the existing exponentially weighted moving average design formula:

[0032] The slow tracking time constant The typical timescale for thermal drift is directly determined by the thermal response time parameter specified in the sensor device datasheet, reflecting the physical timescale of the gradually changing deviation. Fast tracking time constant. The determination method is as follows: During the calibration phase, the potentiometer contact jitter fault is manually reproduced, and the results are recorded. Measured maximum value The baseline tracking lag per cycle at the maximum rate of change must not exceed the normal residual noise amplitude. That is, the following must be satisfied:

[0033] Conversely, this leads to... The upper bound is taken as the upper bound. The design values ​​ensure that the baseline can effectively follow changes in deviation even when they accelerate without generating hysteresis errors exceeding the noise level. The specific calibration method is given in the fault identification stage. The derivation depends on this value, and both must be completed in sequence during the calibration process.

[0034] Thus, the dynamic switching of the dual-speed adaptive smoothing factor enables the deviation baseline to have sufficient noise suppression capability under normal slowly changing operating conditions, while effectively reducing tracking lag when the deviation accelerates and drifts.

[0035] S2. Calculate the deviation residual and identify potentiometer faults.

[0036] In an optional embodiment, at the deviation baseline While continuously updating, the system simultaneously calculates the deviation residuals within each control cycle:

[0037] in, This represents the high-frequency deviation of the current instantaneous deviation relative to the gradually changing baseline. A larger value indicates a greater likelihood of faults in the potentiometer, such as contact wear and open circuit, localized damage to the resistive film, or poor soldering of the leads. This will cause abrupt changes or continuous drift in the potentiometer angle reading, leading to… The amplitude significantly exceeds the normal noise range; the smaller the value, the more normally the potentiometer is working. The amplitude is limited by the inherent noise levels of the two sensors, exhibiting bounded random fluctuation characteristics.

[0038] Furthermore, in order to establish objective fault judgment boundaries, the system collects a sufficient number of control cycles during the power-on calibration phase. The sample size, specifically, is determined based on the required statistical confidence level and control cycle duration. For example, the sampling quantity can be no less than 500 control cycles. During the calibration phase, the servo motor is required to perform a low-speed, full-stroke scan, and the root mean square error of the collected samples is calculated. , It objectively reflects the upper bound of the residual amplitude after the noise from the two sensors is superimposed under normal operating conditions, and is strongly tied to specific hardware components. The preset fault judgment threshold is based on... The product with the preset multiplier is determined, that is... The multiplier The false alarm rate requirement is determined during the calibration phase. For example, The value of can be 4.

[0039] Next, a single exceedance of the threshold does not trigger a switchover; the system uses continuous counting for decision-making: in response to The number of consecutive triggers exceeds the preset count limit. The system determines that the potentiometer has malfunctioned and triggers a priority switch.

[0040] Furthermore, during the calibration phase, a rated vibration impact load was applied to the servo motor, and the effects of purely mechanical vibration were statistically analyzed. Maximum number of consecutive triggers ,make:

[0041] because Only one more than the maximum number of consecutive trigger cycles that can be caused by vibration and shock, any false triggers caused by vibration and shock are completely filtered out, and the decision delay for real faults is compressed to the theoretical minimum. This addresses the issue that occurs during the counting accumulation process. In cases of sampling points, the counter is reset and re-accumulated to ensure that intermittent noise is not misjudged as a continuous fault.

[0042] Thus, the fault identification method based on the statistical characteristics of deviation residuals has its decision threshold and count upper limit derived from the measured data in the calibration stage, which is strongly bound to specific hardware. While effectively filtering out false triggers caused by vibration and shock, it maintains the shortest response delay for real persistent faults.

[0043] S3. Extract the deviation baseline snapshot and perform a non-disruptive priority switch.

[0044] In an optional embodiment, the system records the switching trigger time in response to the state where the fault determination condition is met. The baseline deviation estimate corresponding to that moment is extracted and used as the baseline deviation snapshot value. While switching the position loop feedback source from the potentiometer to the magnetic encoder, compensation is applied to the magnetic encoder readings:

[0045] in, The compensated equivalent position feedback value is obtained by superimposing the deviation baseline snapshot value with the current magnetic encoder angle reading. This final equivalent position feedback value directly replaces the original potentiometer reading and is fed into the PID position loop. Compared with the last valid potentiometer reading before the switch The difference between them is compressed to the level of residual noise, i.e. This eliminates the step jump injected into the PID loop during switching, thus achieving bumpless switching.

[0046] Furthermore, after the switch is complete, the system uses the magnetic encoder as the main feedback source and stops estimating the deviation baseline. Update and change the baseline snapshot value. As a fixed compensation amount, it is continuously added to the subsequent magnetic encoder angle readings, ultimately maintaining continuous compensation before the system reset, until the system is reset or the potentiometer is replaced and recalibrated.

[0047] In this way, by compensating the magnetic encoder readings with the deviation baseline snapshot, the step jump injected into the PID position loop at the moment of switching is effectively eliminated, and the transient overshoot of the output shaft is significantly reduced.

[0048] S4. Correct the differential gain of the position loop after the switching.

[0049] In one optional embodiment, diaphragm potentiometers and magnetic encoders differ fundamentally in their signal characteristics. The potentiometer outputs an analog voltage, which, after ADC conversion, contains high levels of discretization noise and contact noise, with an equivalent angular resolution typically on the order of 0.1° to 0.5°. The MT6701 magnetic encoder, on the other hand, boasts an angular resolution of up to 14 bits, approximately 0.022°, and its signal noise is significantly lower than that of the potentiometer. Since the actual output amplitude of the PID derivative term is proportional to the product of the feedback signal noise amplitude and the derivative gain, if the PID parameters tuned for the potentiometer's noise characteristics are still used after switching, the PID derivative term will experience excessive gain due to the sudden drop in feedback signal noise level, leading to high-frequency chattering.

[0050] Therefore, it is necessary to calculate the root mean square error of the angle reading noise of the two sensors in a static state during the calibration phase, and obtain the first root mean square error of the potentiometer noise. The second noise mean square error of the magnetic encoder After switching, the ratio of the first noise mean square error to the second noise mean square error is calculated. This ratio is then used to proportionally correct the original differential gain of the position loop. The derivation logic is as follows: To ensure that the noise-driven output amplitude of the differential term remains consistent before and after switching, the following condition must be met: From this, the correction relation can be directly derived:

[0051] in This is the original differential gain tuned when the potentiometer is used as the feedback source. The corrected value after switching is the differential gain after correction. The ratio of the root mean square error of the noise of the two sensors is calculated from the calibration data. The original differential gain is multiplied by this ratio to obtain the corrected differential gain.

[0052] Furthermore, keeping the proportional gain and integral gain of the position loop unchanged, the corrected differential gain is applied to the differential channel of the position loop, ultimately achieving constant-amplitude noise correction of the differential gain after switching. The proportional gain and integral gain reflect the steady-state tracking characteristics of the position loop and are weakly correlated with the sensor noise level, remaining unchanged after switching. The correction only applies to the differential channel, so that the position loop maintains its original steady-state accuracy after switching while suppressing high-frequency jitter caused by changes in sensor noise characteristics.

[0053] Figure 2 This is a schematic diagram of dual-sensor deviation baseline tracking and priority switching triggering in an embodiment of the present invention. As can be seen, the deviation baseline estimate has a good suppression effect on the high-frequency noise of the instantaneous deviation during the normal gradual change phase, and can quickly follow during the deviation acceleration drift phase, demonstrating the dynamic switching characteristics of the dual-speed adaptive smoothing factor.

[0054] In this way, by performing equal-amplitude correction on the differential gain, the high-frequency jitter caused by the sudden change in sensor noise characteristics is effectively suppressed, ensuring the stable operation of the position loop after switching.

Claims

1. A servo motor priority control method based on dual position loop feedback, characterized in that, include: The potentiometer angle readings and magnetic encoder angle readings of the servo motor are collected synchronously, and the instantaneous deviation is calculated. The instantaneous deviation is processed using an adaptive smoothing algorithm to obtain a baseline deviation estimate. The deviation residual is calculated based on the instantaneous deviation and the deviation baseline estimate, and fault identification is performed based on the deviation residual. When a potentiometer fault is determined, priority switching is triggered. In response to the priority switch, the deviation baseline snapshot value at the switch trigger moment is extracted to compensate the magnetic encoder angle reading, and the compensated equivalent position feedback value is sent into the position loop.

2. The method according to claim 1, characterized in that, The process of processing the instantaneous deviation using an adaptive smoothing algorithm to obtain a baseline deviation estimate includes: Calculate the instantaneous rate of change of the instantaneous deviation; Based on the comparison between the instantaneous rate of change and the preset rate threshold, the corresponding smoothing factor is dynamically selected. The instantaneous deviation is processed by an exponentially weighted moving average using the selected smoothing factor to obtain the baseline deviation estimate.

3. The method according to claim 2, characterized in that, The dynamic selection of the corresponding smoothing factor based on the comparison result between the instantaneous rate of change and the preset rate threshold includes: When the instantaneous rate of change is greater than the preset rate threshold, the first smoothing factor is selected; When the instantaneous rate of change is less than or equal to the preset rate threshold, a second smoothing factor is selected; Wherein, the first smoothing factor is greater than the second smoothing factor.

4. The method according to claim 1, characterized in that, The fault identification based on the deviation residual, and the triggering of priority switching when a potentiometer fault is determined, includes: The absolute value of the deviation residual is compared with a preset fault judgment threshold; When the absolute value is greater than the preset fault judgment threshold, continuous trigger counting is performed; When the number of consecutive triggers exceeds the preset count limit, the potentiometer is determined to be faulty, and the priority switch is eventually triggered.

5. The method according to claim 4, characterized in that, Before comparing the absolute value of the deviation residual with the preset fault judgment threshold, the method further includes: During the calibration phase, a predetermined number of deviation residual samples are collected, and the mean square error of the deviation residual samples is calculated. The preset fault judgment threshold is determined based on the mean square error. During the calibration phase, the maximum number of consecutive triggering cycles caused by mechanical vibration is counted, and the preset count upper limit is determined based on the maximum number of consecutive triggering cycles. Finally, the preset fault judgment threshold and the preset count upper limit are obtained for fault identification.

6. The method according to claim 1, characterized in that, The step of extracting the deviation baseline snapshot value at the switching trigger moment to compensate the magnetic encoder angle reading and sending the compensated equivalent position feedback value into the position loop includes: Record the switching trigger time when the priority switch is triggered; Extract the baseline deviation estimate corresponding to the switching trigger time as the baseline deviation snapshot value; The deviation baseline snapshot value is superimposed with the current magnetic encoder angle reading to obtain the equivalent position feedback value.

7. The method according to claim 1, characterized in that, After sending the compensated equivalent position feedback value into the position loop, the method further includes: Obtain the first noise mean square error of the potentiometer and the second noise mean square error of the magnetic encoder; Calculate the ratio of the first noise root mean square error to the second noise root mean square error; The original differential gain of the position loop is proportionally corrected using the ratio to obtain the corrected differential gain.

8. The method according to claim 7, characterized in that, The step of proportionally correcting the original differential gain of the position loop using the ratio to obtain the corrected differential gain includes: Multiplying the original differential gain by the ratio yields the corrected differential gain; By keeping the proportional gain and integral gain of the position loop constant, the corrected differential gain is applied to the differential channel of the position loop, ultimately achieving equal-amplitude noise correction of the differential gain after switching.

9. The method according to claim 1, characterized in that, The synchronous acquisition of the potentiometer angle readings and magnetic encoder angle readings of the servo motor, and the calculation of the instantaneous deviation, include: During each control cycle, the analog voltage signal output by the potentiometer and the digital angle signal output by the magnetic encoder are acquired synchronously. The analog voltage signal is converted to obtain the potentiometer angle reading, and the digital angle signal is analyzed to obtain the magnetic encoder angle reading. The instantaneous deviation is calculated by subtracting the angle reading of the magnetic encoder from the angle reading of the potentiometer.

10. The method according to claim 1, characterized in that, After sending the compensated equivalent position feedback value into the position loop, the method further includes: The magnetic encoder is used as the main feedback source; Stop updating the baseline deviation estimate; The deviation baseline snapshot value is continuously superimposed on subsequent magnetic encoder angle readings as a fixed compensation amount, ultimately maintaining continuous compensation before system reset.

Images