pith. sign in

arxiv: 2607.01215 · v1 · pith:YUD54UJ4new · submitted 2026-07-01 · 🧮 math.OC · cs.SY· eess.SY

Computationally Efficient Near-Optimal Control for Current Ripple Reduction and Optimization of Three-Phase Motors via LMIs

Huu-Thinh Do , Trung B. Tran , Jing Sun , Ilya Kolmanovsky This is my paper

Pith reviewed 2026-07-02 07:39 UTC · model grok-4.3

classification 🧮 math.OC cs.SYeess.SY
keywords controlcomputationallycostcurrentfunctionmotorsoptimaloptimization
0
0 comments X

The pith

LMI-based quadratic approximation of the value function via iterated Bellman inequalities yields a tractable offline convex program for horizon-one near-optimal control of PMSMs with performance close to FCS-MPC.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Electric motors in cars and machines need fast control to cut energy waste and vibration from uneven current flow. Full optimal control is hard because the motor equations are nonlinear and the possible control actions are discrete switches. The authors replace the true long-term cost with a simple quadratic function whose parameters are found by solving a set of linear matrix inequalities offline. These inequalities come from repeatedly applying the Bellman equation in a relaxed convex form. Once computed, the quadratic serves as an extra cost term in a cheap one-step lookahead controller. Simulations indicate the resulting switching pattern reduces current ripple while keeping computational load far below that of standard model-predictive control.

Core claim

The computed quadratic function can be obtained efficiently offline and used online as a tail cost in a horizon-one optimal control law, achieving a favorable trade-off between switching effort and current ripple with performance comparable to finite-control-set MPC but with significantly lower computational cost.

Load-bearing premise

The quadratic parameterization combined with iterated Bellman inequalities produces a sufficiently tight upper bound on the true infinite-horizon value function for the nonlinear PMSM dynamics and discrete input set.

Figures

Figures reproduced from arXiv: 2607.01215 by Huu-Thinh Do, Ilya Kolmanovsky, Jing Sun, Trung B. Tran.

Figure 1
Figure 1. Figure 1: PMSM schematic and the finite control set in (6). Despite the deceptively simple linear structure, the dynam￾ics (1) pose a significant control challenge when powered by a VSI. Specifically, let us define the relation between the control input u r dq and the VSI’s discrete output as follows. Let: Sabc ∈ N 3 [0,Nlevel−1], (2) be the switch states of the inverter taking values from the finite set N 3 [0,Nlev… view at source ↗
Figure 2
Figure 2. Figure 2: Horizon-one MPC with the approximate value function for different values of Nlevel with fs = 50kHz. 50kHz has been assumed. As observed, the THD decreases as the number of inverter voltage levels increases. This behavior is consistent with the expected trade-off between hardware complexity and achievable performance. From another per￾spective, this effect can also be interpreted as a quantization phenomeno… view at source ↗
Figure 4
Figure 4. Figure 4: Top: Comparison of performance trade-off between the pro￾posed setting, FCS-MPC (Np = 3) and PI-SVM. yielding 90µs on average3 for LMI-ADP compared to 5.5µs for PI-SVM and 13400µs of FCS-MPC (see [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The approximation function Vˆ from different objective functions in (25) and the interpolated function from value iteration (red points) with θre = 0 (The right figure shows a close-up of the boxed region). cost is chosen as c(ΘM) = 0 and only the LMI conditions are enforced as hard constraints, the resulting approximation is the most conservative, as no mechanism encourages the under-approximation to incr… view at source ↗
read the original abstract

The optimal control of three-phase permanent-magnet synchronous motors (PMSMs) is challenging due to their nonlinearity and the discrete nature of the control set. Existing approaches either rely on mixed-integer trajectory optimization or require computationally intensive value-iteration procedures. This paper proposes a Linear Matrix Inequality (LMI)-based method for approximating the infinite-horizon value function using a quadratic parameterization and iterated Bellman inequalities, yielding a tractable convex program. The computed function can be obtained efficiently offline and used online as a tail cost in a horizon-one optimal control law. Simulation results show that the proposed approach achieves a favorable trade-off between switching effort and current ripple, with performance comparable to that of finite-control-set MPC but with a significantly lower computational cost.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes an LMI-based convex program that approximates the infinite-horizon value function of a nonlinear PMSM via a quadratic parameterization and iterated Bellman inequalities. The resulting quadratic is computed offline once and then deployed online as a terminal cost in a horizon-one optimal control law over the discrete voltage set, with the claim that this yields a favorable switching-ripple trade-off comparable to finite-control-set MPC at far lower online cost.

Significance. If the quadratic indeed supplies a valid and sufficiently tight upper bound on the true infinite-horizon cost, the approach supplies a practical, offline-computable tail-cost construction that sidesteps both mixed-integer trajectory optimization and repeated value iteration for discrete-input motor control. The explicit use of a single convex program solved once is a clear methodological strength.

major comments (2)
  1. [§4.2] §4.2, the LMI relaxation of the iterated Bellman inequality (presumably Eq. (12) or (15)): because the plant f(x,u) is nonlinear, the conversion to LMIs necessarily employs a relaxation (local linearization, polytopic embedding, or S-procedure). The manuscript does not demonstrate that the obtained quadratic V satisfies the exact pointwise inequality V(x) ≥ ℓ(x,u) + V(f(x,u)) for every state and every admissible discrete voltage; without this verification the tail-cost argument is not guaranteed to be valid.
  2. [Simulation results] Simulation section (and abstract claim of “comparable performance”): no quantitative metrics, error bars, motor parameters, sampling time, discretization method, or explicit description of the FCS-MPC baseline are supplied. The central performance claim therefore rests on unverified simulation outcomes and cannot be assessed.
minor comments (2)
  1. [§2] Notation for the discrete input set and the stage cost ℓ(x,u) should be introduced once and used consistently; several symbols appear to be redefined between §2 and §4.
  2. [Figures 3-5] Figure captions for the current-ripple and switching-frequency plots should state the exact motor parameters, horizon length, and weighting matrices used.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and will incorporate the suggested improvements in the revised manuscript.

read point-by-point responses

  1. Referee: [§4.2] §4.2, the LMI relaxation of the iterated Bellman inequality (presumably Eq. (12) or (15)): because the plant f(x,u) is nonlinear, the conversion to LMIs necessarily employs a relaxation (local linearization, polytopic embedding, or S-procedure). The manuscript does not demonstrate that the obtained quadratic V satisfies the exact pointwise inequality V(x) ≥ ℓ(x,u) + V(f(x,u)) for every state and every admissible discrete voltage; without this verification the tail-cost argument is not guaranteed to be valid.

    Authors: We acknowledge that the LMI provides a sufficient (but potentially conservative) condition via relaxation, and the manuscript does not include an explicit post-solution verification that the quadratic satisfies the exact nonlinear Bellman inequality pointwise. In the revision we will add a numerical validation procedure: a dense grid of states within the operating region will be sampled and the inequality checked for every admissible discrete voltage. Any detected violations will be reported together with their magnitude and a discussion of implications for the tail-cost guarantee. revision: yes

  2. Referee: [Simulation results] Simulation section (and abstract claim of “comparable performance”): no quantitative metrics, error bars, motor parameters, sampling time, discretization method, or explicit description of the FCS-MPC baseline are supplied. The central performance claim therefore rests on unverified simulation outcomes and cannot be assessed.

    Authors: We agree that the simulation section is insufficiently detailed for reproducibility and assessment of the performance claim. The revised manuscript will supply the PMSM parameters, sampling time, discretization method, quantitative metrics (average and standard deviation of current ripple and switching count, with error bars from repeated runs), and a precise description of the FCS-MPC baseline (horizon length, cost function, and implementation details). revision: yes

Circularity Check

0 steps flagged

No circularity; offline LMI program yields independent tail cost

full rationale

The abstract and reader's summary describe an offline convex LMI program that computes a quadratic approximation to the infinite-horizon value function via iterated Bellman inequalities; this quadratic is then deployed as a fixed tail cost inside a separate horizon-one controller. No quoted step shows the output being defined in terms of itself, a fitted parameter renamed as a prediction, or a load-bearing self-citation chain. The derivation chain remains self-contained against external benchmarks because the convex program is solved once from the system model and the resulting quadratic is used without further fitting to the online performance metric.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is provided, so no concrete free parameters, axioms, or invented entities can be extracted.

pith-pipeline@v0.9.1-grok · 5674 in / 1052 out tokens · 22366 ms · 2026-07-02T07:39:03.906415+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

21 extracted references · 21 canonical work pages

  1. [1]

    300 years of optimal control: From the brachystochrone to the maximum principle,

    H. Sussmann and J. Willems, “300 years of optimal control: From the brachystochrone to the maximum principle,”IEEE Control Systems Magazine, vol. 17, no. 3, pp. 32–44, 1997

    work page 1997

  2. [2]

    Dynamic programming,

    R. Bellman, “Dynamic programming,”Science, vol. 153, no. 3731, pp. 34–37, 1966

    work page 1966

  3. [3]

    Bertsekas,Dynamic programming and optimal control: V olume I

    D. Bertsekas,Dynamic programming and optimal control: V olume I. Athena scientific, 2012, vol. 4

    work page 2012

  4. [4]

    Long prediction horizon FCS-MPC for power converters and drives,

    E. Zafra, S. Vazquez, T. Geyer, R. P. Aguilera, and L. G. Franquelo, “Long prediction horizon FCS-MPC for power converters and drives,” IEEE Open Journal of the Industrial Electronics Society, vol. 4, pp. 159–175, 2023

    work page 2023

  5. [5]

    Guidelines for the design of finite control set model predictive controllers,

    P. Karamanakos and T. Geyer, “Guidelines for the design of finite control set model predictive controllers,”IEEE Transactions on Power Electronics, vol. 35, no. 7, pp. 7434–7450, 2019

    work page 2019

  6. [6]

    Finite-control-set model predictive control of permanent magnet synchronous motor drive systems—an overview,

    T. Li, X. Sun, G. Lei, Y . Guo, Z. Yang, and J. Zhu, “Finite-control-set model predictive control of permanent magnet synchronous motor drive systems—an overview,”IEEE/CAA Journal of Automatica Sinica, vol. 9, no. 12, pp. 2087–2105, 2022

    work page 2087

  7. [7]

    Nonlin- ear differential flatness-based speed/torque control with state-observers of permanent magnet synchronous motor drives,

    P. Thounthong, S. Sikkabut, N. Poonnoy, P. Mungporn, B. Yodwong, P. Kumam, N. Bizon, B. Nahid-Mobarakeh, and S. Pierfederici, “Nonlin- ear differential flatness-based speed/torque control with state-observers of permanent magnet synchronous motor drives,”IEEE Transactions on Industry Applications, vol. 54, no. 3, pp. 2874–2884, 2018

    work page 2018

  8. [8]

    Port controlled Hamilton with dissipation-based speed control of IPMSM drive,

    M. N. Uddin, Z. Zhai, and I. K. Amin, “Port controlled Hamilton with dissipation-based speed control of IPMSM drive,”IEEE Transactions on Power Electronics, 2019

    work page 2019

  9. [9]

    Trajectory tracking for a class ofθ-periodic switched systems,

    G. S. Deaecto, L. C. Costanzo, and L. N. Egidio, “Trajectory tracking for a class ofθ-periodic switched systems,”IEEE Transactions on Automatic Control, vol. 69, no. 3, pp. 1874–1881, 2023

    work page 2023

  10. [10]

    Trajec- tory tracking for a class of switched nonlinear systems: Application to PMSM,

    L. N. Egidio, G. S. Deaecto, J. P. Hespanha, and J. C. Geromel, “Trajec- tory tracking for a class of switched nonlinear systems: Application to PMSM,”Nonlinear Analysis: Hybrid Systems, vol. 44, p. 101164, 2022

    work page 2022

  11. [11]

    Finite control set model predictive control with limit cycle stability guarantees,

    D. Xu and M. Lazar, “Finite control set model predictive control with limit cycle stability guarantees,”Automatica, vol. 181, p. 112507, 2025

    work page 2025

  12. [12]

    Global optimal predictive control of pmsm using dynamic programming: An offline benchmarking tool,

    Q. Zhu, G. Ozkan, M. Figueroa-Santos, M. Barron, C. S. Edrington, and R. Prucka, “Global optimal predictive control of pmsm using dynamic programming: An offline benchmarking tool,”IEEE Access, vol. 12, pp. 169 720–169 732, 2024

    work page 2024

  13. [13]

    Optimal switching of dc–dc power converters using approx- imate dynamic programming,

    A. Heydari, “Optimal switching of dc–dc power converters using approx- imate dynamic programming,”IEEE transactions on neural networks and learning systems, vol. 29, no. 3, pp. 586–596, 2016

    work page 2016

  14. [14]

    Approximate dynamic pro- gramming via iterated Bellman inequalities,

    Y . Wang, B. O’Donoghue, and S. Boyd, “Approximate dynamic pro- gramming via iterated Bellman inequalities,”International Journal of Robust and Nonlinear Control, vol. 25, no. 10, pp. 1472–1496, 2015

    work page 2015

  15. [15]

    High-speed finite control set model predictive control for power electronics,

    B. Stellato, T. Geyer, and P. J. Goulart, “High-speed finite control set model predictive control for power electronics,”IEEE Transactions on Power Electronics, vol. 32, no. 5, pp. 4007–4020, 2016

    work page 2016

  16. [16]

    Simultaneous identification and adaptive torque control of permanent magnet synchronous machines,

    D. M. Reed, J. Sun, and H. F. Hofmann, “Simultaneous identification and adaptive torque control of permanent magnet synchronous machines,” IEEE Transactions on Control Systems Technology, vol. 25, no. 4, pp. 1372–1383, 2016

    work page 2016

  17. [17]

    Boyd and L

    S. Boyd and L. Vandenberghe,Convex optimization. Cambridge University Press, 2004

    work page 2004

  18. [18]

    Practical stability of discrete-time switched affine systems,

    G. S. Deaecto and L. N. Egidio, “Practical stability of discrete-time switched affine systems,” in2016 European Control Conference (ECC). IEEE, 2016, pp. 2048–2053

    work page 2016

  19. [19]

    Space vector modulation: an engineering review,

    P. Handley and J. Boys, “Space vector modulation: an engineering review,” in1990 F ourth International Conference on Power Electronics and V ariable-Speed Drives. IET, 1990, pp. 87–91

    work page 1990

  20. [20]

    Geyer,Model predictive control of high power converters and industrial drives

    T. Geyer,Model predictive control of high power converters and industrial drives. John Wiley & Sons, 2016

    work page 2016

  21. [21]

    Approximate dynamic programming for constrained linear systems: A piecewise quadratic approximation approach,

    K. He, S. Shi, T. van den Boom, and B. De Schutter, “Approximate dynamic programming for constrained linear systems: A piecewise quadratic approximation approach,”Automatica, vol. 160, p. 111456, 2024

    work page 2024