the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Hysteretic switching active disturbance rejection control for the servo drives of a skating-training cable robot
Abstract. Whether a skating-training robot reproduces the prescribed motion depends on whether its servo layer can follow the motor-angle reference amplified by the screw transmission and the cable geometry. Switching active disturbance rejection control improves the large-error response by switching between linear and nonlinear corrections, but the switching frequency and the tracking accuracy lack a quantitative relation, and the hysteresis width is selected without a stated rule. This paper builds an integrated model from cable geometry and screw transmission to the permanent-magnet synchronous motor and gives the torque, speed and voltage margins of each drive; it then proposes a state-preserving hysteretic implementation in which two normalized switching variables act on the observer and feedback corrections separately, with a hysteresis band added. Six controller families are compared at fixed gains with equal tuning effort over 154 cases: the 30 rad step overshoot falls from 52.171% to 0.311%, the maximum error in the load interval falls from 0.101 rad to 0.0447 rad, and the number of mode actions falls from 46 to 18 at a hysteresis width of 0.2. With the switching frequency taken as a design variable, the servo layer can set the switching count to its target while preserving tracking quality, which offers a reusable parameterized method for the servo design of comparable cable robots.
- Preprint
(1225 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 17 Dec 2026)
- CC1: 'Comment on ms-2026-193', Shuangqing Yu, 02 Nov 2026 reply
-
RC1: 'Comment on ms-2026-193', Anonymous Referee #1, 03 Nov 2026
reply
The manuscript proposes a state-preserving hysteretic switching ADRC implementation for the servo drives of a skating-training cable robot. The application is relevant, and the inclusion of fixed-gain comparisons, paired trials and switching-path ablation is useful. However, the executable controller is not specified consistently, the claimed switching-frequency design relation is not established, and the simulated drive results do not validate the stated robot-level engineering contribution. Important contradictions in tuning, feasibility and interpretation further weaken the evidence. Resolving these issues would require reconstructing the formulation and conducting a substantially new, independently verifiable evaluation. My main concerns are as follows.
- The proposed controller cannot be reconstructed unambiguously from Sections 3.1-3.2. Equation (17), introduced as the definition of two normalized switching variables, actually gives the disturbance-compensated input; Eq. (18), introduced as the mode-update rule, gives the switching variables; and Eq. (19), introduced as the drive command, gives the mode-update rule. Equation (13) specifies only a linear ESO, without the claimed nonlinear observer injection, and the switched correction in Eq. (16) is not clearly connected to the applied input using the nominal command. Moreover, the feedback switching variable divides a position error by D, while Section 5.5 assigns D units of rad/s². These inconsistencies prevent verification of the actual two-path control law and its dimensional consistency.
- The central relation between hysteresis and a prescribed switching frequency is not established. The text and Eq. (19) use entry and exit thresholds of 1 and 1-h, whereas Fig. 3(b) uses 1+h and 1-h and Eq. (20) assumes a width of 2h. The dwell-time bound therefore does not correspond consistently to the stated implementation. No usable rate bound, inverse design rule or prospective test against specified switching-count targets is provided. The manuscript acknowledges that noise jumps and saturation can violate the bound assumptions. A conditional dwell estimate and a retrospective count-versus-width plot do not establish target-frequency design or closed-loop stability of the sampled, saturated switching system.
- The claimed integration with an eight-cable training robot is not demonstrated by the model. Equations (1)-(4) supply kinematics and a static wrench relation, while Fig. 2 feeds prescribed cable-force proxies into single-drive dynamics. No explicit coupled platform dynamics or tension-allocation law establishes that the eight realized motor trajectories maintain compatible cable lengths, positive tensions and the desired platform motion. Section 5.3 attributes channel errors to geometric singularities without showing Jacobian-conditioning histories or a causal test. Although Eq. (23) correctly labels its output as a motor-layer diagnostic, motor-angle overshoot is subsequently interpreted as platform overshoot. All Table 1 parameters are assumed, and no calibrated robot or hardware validation closes these gaps.
- The comparative evaluation is insufficiently specified to support the claimed equal tuning effort. Section 3.4 and the abstract state that all families receive equal effort, but Table 2 lists 54 candidates and runs for the independently tuned families and zero for H-SADRC. Inheriting gains could be reasonable, but the inheritance and the selection of thresholds and hysteresis are not sufficiently documented. The actual candidate grids, score weights, final gains and relevant implementation settings are also missing. Consequently, the large headline improvement over PID cannot be separated confidently from differences in parameterization and tuning quality, and the fairness of the comparison cannot be independently assessed.
- The feasibility analysis and the subsequent controller ranking are not reconciled. Section 5.1 reports a torque margin of -12.12% and approximately 10% worst-channel saturation at nominal cadence, yet also states that the speed margin fails first and that saturated cases should not be used to rank controllers. Section 5.4 then includes large-error, capability-limited parameter realizations in the headline pooled comparison without a separate common-feasible analysis. A defined admissibility criterion, controller-specific saturation diagnostics and a separation of feasible tracking performance from actuator-limited failures are necessary before these error reductions can be interpreted as evidence for the control method.
- The reported ablation and electrical proxies do not establish the proposed structural and engineering advance. Section 5.5 gives the same pooled error, 2.241 rad, for both-switch and feedback-only configurations, while Fig. 12 reports zero actions for the observer-only configuration. This does not demonstrate an effective contribution from the observer-switch path. The paired H-SADRC-minus-SADRC difference is only -0.00158 rad with an interval containing zero. At h=0.2, a reduction from 46 to 18 logical mode actions is accompanied by changes of only -0.516% in current total variation and -0.173% in the copper-loss proxy. Logical controller-mode transitions are not demonstrated to be inverter switching events; device-lifetime, EMI and tension-impact benefits remain unsupported. The continuous variant also requires a meaningful engineering comparison.
- The practical-performance interpretation does not adequately address counterexamples in the results. Under the stated encoder-noise and 2 ms delay condition, Section 5.4 reports pooled RMSE of 2.110 rad for the switching family and 1.615 rad for PID. Calling this difference small by comparison with the earlier step and load errors mixes different operating conditions and error metrics. A task-based tolerance is needed, but Table 3 contains unrendered and truncated formulas, intervals and practical-threshold entries, so the stated practical-significance rule cannot be applied or reproduced. The results therefore do not establish the claimed reusable engineering advantage under the measurement conditions that motivate hysteresis.
- The breadth and numerical reliability of the 154-case evaluation cannot be independently checked. Section 4.1 announces a 64-point threshold/hysteresis scan and an integration-step check, but does not provide the full scan, refinement results or a complete case inventory. The parameter distributions, noise generation, seeds, sampling and integration settings, and bootstrap implementation are not sufficiently specified. The reference Fourier coefficients and load-proxy construction are also needed to reproduce the drive demands. Statements that configurations are retained and figures trace to data do not substitute for accessible case definitions, executable models and data. These omissions affect the central validation, rather than only presentation.
Citation: https://doi.org/10.5194/ms-2026-193-RC1
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 200 | 42 | 21 | 263 | 25 | 38 |
- HTML: 200
- PDF: 42
- XML: 21
- Total: 263
- BibTeX: 25
- EndNote: 38
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript investigates hysteretic switching active disturbance rejection control for a skating-training cable robot. Although the topic has potential engineering relevance, the manuscript has substantial limitations in originality and completeness. The contribution is not sufficiently established to warrant publication. I therefore recommend rejection for the following reasons:
The proposed approach primarily combines existing active disturbance rejection control, linear–nonlinear switching, and hysteresis mechanisms. The manuscript does not convincingly establish a substantive advance over existing methods. The reported improvements appear to reflect controller integration and parameter selection rather than a clearly demonstrated methodological or theoretical contribution.
The mathematical formulation, theoretical analysis, and simulation validation do not form a sufficiently complete and consistent argument. The operating conditions, performance guarantees, and parameter-selection principles of the proposed switching mechanism are inadequately established, limiting confidence in the method and its reproducibility.
 The comparisons do not adequately isolate the contribution of hysteresis from those of the other controller components. Consequently, the reported performance improvements cannot be clearly attributed to the claimed innovation. The evidence supporting the method’s general applicability and practical value is also limited.
The connection between the proposed controller and the cable-robot application remains underdeveloped. Validation focuses primarily on motor-level servo performance, without adequately assessing overall robot motion or application-specific cable constraints. The broader conclusions regarding the robotic application are therefore insufficiently supported.
Overall, substantial redevelopment and additional validation would be required to establish the manuscript’s originality, methodological soundness, and application-level contribution. I do not recommend acceptance.
Â
Â