Repetitive Control¶
Feed error from a modeled recurrence period back into an operating controller to track periodic references or reject periodic disturbances.
Core Idea¶
Repetitive control is a feedback method that remembers error from a modeled period and uses it to correct later cycles of an operating plant. It is suited to commands that repeat, such as a robot's trajectory, or disturbances with a repeating component, such as output-voltage distortion in an inverter. The controller measures an output, compares it with a target, retains corresponding-phase error and returns correction to later input.[ref-0207c71dd507][ref-3fa1c9780425][^ref-e29f203b8a36]
Its name identifies that period-memory architecture, not a guarantee of zero error. Hara's ideal delay model gives a conditional tracking result when the closed loop is stable and the signal belongs to the modeled periodic class. Actual filters, plant dynamics and a wrong period can leave error. A controller can still be recognized as repetitive control when its attempted implementation misses a performance goal.[^ref-0207c71dd507]
Scope of Application¶
Omata and colleagues apply the method to a nonlinear three-link manipulator following a periodic trajectory. Their original publisher abstract reports position and velocity feedback, nonlinear compensation and low tracking error, but does not reveal numeric error or exact memory-block settings. Zhang and colleagues use a digital repetitive branch in an SPWM UPS inverter to reduce repeating output-voltage distortion under nonlinear loads. In both cases prior-period error changes later plant input, although the measured outputs and design choices differ.[ref-3fa1c9780425][ref-e29f203b8a36]
This is a control-engineering method. A repeating signal without a feedback plant, or an ordinary controller with no stored period-linked error, does not instantiate it. Nor does a periodic calendar task or a human learning exercise merely because it repeats.[^ref-0207c71dd507]
Clarity¶
Ask three separate questions. Architecture: does measured error return through a memory of the modeled period to later plant input? Match: does the actual command or disturbance follow that period? Result: is the implemented loop stable and accurate in a test? The first question identifies the method; the other two determine whether a performance claim is warranted. An ideal harmonic result does not establish perfect rejection by every filtered implementation.[^ref-0207c71dd507]
A reference-tracking robot and disturbance-rejecting inverter also differ. The robot follows a periodic command; the inverter preserves its output despite repeating distortion. Both use the same period-error correction relation, but their error measures cannot be treated as interchangeable.[ref-3fa1c9780425][ref-e29f203b8a36]
Manages Complexity¶
A design may include sensors, plant dynamics, sampling, phase, filters, compensation and loads. Four checks organize it: what output is measured, what period is modeled, what error is stored, and how does it change later input? After identifying the method, assess stability, period match and the particular output measure. A low-pass element, notch, DSP and exact memory size belong to specific designs rather than the all-instance definition.[ref-0207c71dd507][ref-e29f203b8a36]
This simplification does not supply one universal recurrence equation. Hara analyzes ideal and modified loops; Zhang implements one digital variant. State the variant and conditions before claiming a quantitative reduction.[ref-0207c71dd507][ref-e29f203b8a36]
Abstract Reasoning¶
If distortion remains after a repetitive branch is enabled, first check that the stored error covers the intended output cycle and actually returns to later input. If so, the architecture is present. Then inspect whether the disturbance repeats at the modeled period, the loop is stable, and the chosen filters preserve correction where needed. If the branch is absent, ordinary feedback or open-loop output generation may explain the remaining behavior; Zhang's same SPWM inverter without the repetitive branch is not an instance of the method.[ref-0207c71dd507][ref-e29f203b8a36]
The live Feedback Prime is the strict parent: measured output returns to affect input. Period-linked retained error is the narrower differentia. An ordinary feedback controller can lack it.[^ref-0207c71dd507]
Knowledge Transfer¶
Carry the role test from robot motion to UPS output regulation, but specify new sensors, periods, errors, plant dynamics and stability measures for each case. Omata's nonlinear compensation does not become a universal inverter component; Zhang's notch and reported voltage THD do not become robot-tracking claims. The broader Feedback Prime travels wherever a measured output changes later input, while this named method remains the documented period-memory specialization.[ref-3fa1c9780425][ref-e29f203b8a36]
Example¶
Three-link manipulator: A robot repeatedly follows a commanded trajectory. Omata's original abstract describes nonlinear compensation, position and velocity feedback, and a three-link experiment reporting low tracking error. The plant and output are the manipulator and observed motion; the modeled period is the repeated command; the error is the difference from target motion; the named repetitive branch returns earlier-cycle error. That last mapping follows the method identity and Hara's description: the abstract does not expose the exact robot memory block or coefficients.[ref-3fa1c9780425][ref-0207c71dd507]
SPWM UPS inverter: Zhang's inverter supplies a nonlinear rectifier load. The fundamental output cycle sets the controller's digital memory horizon; the load is a disturbance source, not the timing clock. The plant output is measured voltage; the error compares that output with the reference; the DSP returns stored prior-period error through a shaped correction branch. In the tested apparatus the authors report 1.4–1.7% voltage THD under nonlinear loads and error settling after roughly three to five fundamental periods. Those numbers describe their test, not every repetitive controller. Without the repetitive branch, the same SPWM inverter remains but lacks period-memory correction.[^ref-e29f203b8a36]
Relationships to Other Abstractions¶
Current abstraction Repetitive Control Domain-specific
Parents (1) — more general patterns this builds on
-
Repetitive Control is a kind of Feedback Prime
A measured error returns through period-linked memory to change later plant input.
Hierarchy path (1) — routes to 1 parentless root
- Repetitive Control → Feedback
Neighborhood in Abstraction Space¶
Repetitive Control sits in a sparse region of the domain-specific corpus (94th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Unclustered & Miscellaneous (2551 abstractions)
Nearest neighbors
- Sampled Data System — 0.86
- Approximate Computing — 0.80
- Dynamic Amplification Factor — 0.78
- Delay-Locked Loop — 0.77
- Simulation Decomposition — 0.77
Computed from structural-signature embeddings · 2026-10-08
Not to Be Confused With¶
Periodicity is a signal property and can exist with no controller. Ordinary Feedback or PID control need not retain earlier-period error. Hara historically contrasts a continuous repetitive loop with specific betterment methods that restart separate trials; that is not a universal division of all later iterative learning. The ideal internal-model theorem supplies conditional accuracy, while actual tracking error or voltage THD is a separately measured outcome.[ref-0207c71dd507][ref-e29f203b8a36]
References¶
[^ref-0207c71dd507]: Shinji Hara, Yutaka Yamamoto, Tohru Omata, and Michio Nakano, “Repetitive Control System, A New Type Servo System for Periodic Exogenous Signals” (title punctuation transcribed as a comma for the reference binder; the original prints a colon), IEEE Transactions on Automatic Control 33, no. 7 (1988): 659–668, https://doi.org/10.1109/9.1274. Full original at Kyoto University, https://repository.kulib.kyoto-u.ac.jp/server/api/core/bitstreams/09b05ef9-9e66-4326-9cba-6d3bc017d8ec/content; especially printed pp.659–664 (PDF pp.1–6). [^ref-3fa1c9780425]: Tohru Omata, Shinji Hara, and Michio Nakano, “Nonlinear Repetitive Control with Application to Trajectory Control of Manipulators,” Journal of Robotic Systems 4, no. 5 (1987): 631–652, https://doi.org/10.1002/rob.4620040505. Original publisher abstract at https://onlinelibrary.wiley.com/doi/abs/10.1002/rob.4620040505; full article not inspected. [^ref-e29f203b8a36]: Kai Zhang, Yong Kang, Jian Xiong, and Jian Chen, “Direct Repetitive Control of SPWM Inverter for UPS Purpose,” IEEE Transactions on Power Electronics 18, no. 3 (2003): 784–792, https://doi.org/10.1109/TPEL.2003.810846. Original author-uploaded full article at https://www.researchgate.net/publication/3280382_Direct_repetitive_control_of_SPWM_inverter_for_UPS_purpose; especially printed pp.786–790, §§III–V.