Abstract
Under high torque and extended duty cycles, current-only control can become less effective as magnetic saturation limits further torque increase, temperature rise weakens the MR-fluid yield stress, and hysteresis affects repeatability. This study develops a coordinated gap-current torque control strategy for a variable-gap multi-disc MR brake, in which bracket displacement and coil current are scheduled together. The control framework integrates weighted multi-criteria gap allocation, feedforward torque-current mapping, and an improved golden jackal optimization (IGJO)-tuned PID loop used only for local current-loop correction. Bench and real-time implementation tests were conducted to evaluate the integrated strategy. For a 60 N·m step command, relative to a current-only IGJO-PID baseline, the proposed controller reduced the time to 90% of the target from 0.942 s to 0.607 s, shortened the 2% settling time from 2.308 s to 0.848 s, and reduced the integral of time-weighted absolute error from 18.30 to 8.90. Step tests at 120, 240, and 360 rpm showed no monotonic dependence of the short-duration closed-loop response on rotational speed within the tested range. During loading–unloading staircase tracking over 0–70 N·m, stable tracking was maintained, with directional asymmetry mainly concentrated in the 70–60 N·m unloading transition. In 0.125 Hz sinusoidal tests, the delay remained about 0.336 s across the tested peak-to-peak torque levels, while amplitude attenuation increased moderately at higher loads. In a 300 s continuous braking test at 60 N·m and 120 rpm, the controller maintained a mean torque of 59.923 N·m over the main quasi-steady interval with 100% coverage within the 2% error band, although ripple increased toward the end as temperature rose. These results show that coordinated gap-current control improves transient response, tracking consistency, and thermal-drift compensation under the tested operating conditions, while the IGJO-tuned PID acts as a local current-loop corrector within the overall architecture.
Funding Information
  • Data Center of Management Science, National Natural Science Foundation of China - Peking University (52375412)
  • Liuzhou Polytechnic University Project (2023KB04)

This publication has 35 references indexed in Scilit: