Energy-Aware Performance Evaluation of Nonlinear Mechatronic Systems Under Matched-Tracking Conditions

arXiv:2608.23578 · eess.SY, cs.RO, cs.SY · Submitted 2026-07-08 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Robotics Radio. Generated commentary on the latest robotics and control papers.

Rosa: I'm Rosa.

Dev: And I'm Dev, here with Taro.

Rosa: We're looking at the paper: "Energy-Aware Performance Evaluation of Nonlinear Mechatronic Systems Under Matched-Tracking Conditions".

Dev: Here's the gist of the paper. Trajectory-tracking metrics alone may not fully capture differences in system effort, and that including energy provides a more informative basis for performance evaluation.

Rosa: Energy-Aware Evaluation Framework The paper introduces an energy-aware evaluation framework that combines tracking metrics with a measure of cumulative actuator energy to enable comparison between systems with approximately matched performance.

Dev: This framework utilizes a simple index (EAPI) to capture both aspects in a single measure, allowing for joint evaluation of tracking performance and actuator energy.

Rosa: The index is defined as EAPI = λJe/Jref e + (1 - λ)Ju/Jref u.

Title and authors: Rosa: So we're looking at this paper, "Energy-Aware Performance Evaluation of Nonlinear Mechatronic Systems Under Matched-Tracking Conditions." It's about how we measure control performance when systems are nonlinear.

Dev: Exactly. Most of us use things like root-meansquare error and settling time to judge if a robot or a machine is doing its job, but those metrics only tell you about the path tracking itself, not how much energy the motors actually used to get there.

Taro: So it's like judging a car by how fast it goes instead of looking at the fuel efficiency. If you only look at speed, you miss a lot about how hard the engine worked and what kind of drag it was fighting.

Rosa: Right, and this paper tackles that problem directly by introducing an energy-aware evaluation framework. They use something called an index, which they call EAPI, to combine tracking error with cumulative actuator energy into one single number.

Dev: That's the core idea: you can compare two different systems that track the same path really closely using this EAPI index. It lets you see if one system is just more efficient than another for the exact same tracking accuracy.

Taro: But the paper doesn't stop there; it dives into how nonlinear dynamics actually make a difference in energy usage, even when they look similar on paper.

Rosa: They show that comparable tracking performance does not mean comparable actuator energy usage in nonlinear systems. They prove that the way stiffness and friction behave in a nonlinear system adds extra work to the motors that isn't present in a simpler linear system.

Dev: That is important because it means if you just look at tracking metrics, you might think two systems are equal when they are actually using vastly different amounts of energy over time.

Taro: And they quantify this difference by looking at the stored mechanical energy and the dissipated energy between a nonlinear system and its linear counterpart. They found that the difference in stored mechanical energy is proportional to alpha times x to the fourth power, where alpha is that nonlinear stiffness term.

Rosa: That alpha term, which represents nonlinear stiffness, means that as you move further away from zero in a nonlinear system, you build up more stored energy than you would expect from just the linear part.

Title and authors: Dev: And on top of that, there's the dissipated energy difference between the two systems over time. The paper shows this difference is related to friction and damping terms multiplied by some integral of velocity squared over time, which is basically how much energy gets lost through viscous forces and Coulomb friction in each case.

Taro: So, when you put it all together, the nonlinear system consistently requires more actuator energy than the linear one across matched operating conditions. The simulation results they showed were pretty stark.

Rosa: They found a really strong example where the linear system used an energy of one hundred ninety-four point seven hundred twenty-seven compared to about four hundred two point six hundred ninety for the nonlinear system, which is an increase of about one hundred six point eight percent in actuator energy.

Dev: That jump in energy is what they call hidden energetic inefficiency that trajectory-based metrics completely miss. The EAPI index they proposed successfully captures this, showing that the nonlinear system consistently gave larger EAPI values than the linear system even when tracking was nearly identical.

Taro: They set up a very specific test to make sure this wasn't just a fluke; they had to find a nonlinear operating point where the tracking error difference was minimal, limited by an epsilon of one-thousandth of a meter.

Rosa: That matching procedure is key because it ensures you aren't comparing two systems that are fundamentally different in their required control effort, but rather two systems performing similarly under the same constraints. It’s about finding that specific nonlinear operating point where the tracking error difference is minimized relative to the linear system.

Dev: And they did this by mapping controller parameters to pairs of tracking error and energy values for both systems, which essentially lets you predict how increasing control gains will affect both accuracy and total actuator energy requirements simultaneously.

Taro: What this means for real-world deployment is that if you're designing a nonlinear system, you can't just optimize the tracking error; you have to optimize that EAPI index too, or your system will be surprisingly inefficient.

Rosa: So the main thing here is that including energy alongside trajectory metrics gives us a much more informative basis for evaluating performance in complex systems. It moves us beyond just looking at how closely we follow the path.

Title and authors: Dev: I think what they're saying is that for any system with nonlinear dynamics, you have to account for those dynamic effects on energy, because they are not negligible when you're pushing the limits of tracking accuracy.

Taro: And from a failure mode perspective, it suggests that if we ignore energy costs in nonlinear control design, we might end up with systems that track well but are completely impractical because the required actuator power is too high for real hardware.

Rosa: So, to wrap up on this paper, "Energy-Aware Performance Evaluation of Nonlinear Mechatronic Systems Under Matched-Tracking Conditions," the authors show how to use a simple index called EAPI to jointly evaluate tracking performance and energy cost. They clearly demonstrate that nonlinear stiffness and friction introduce additional work that isn't in linear systems, leading to significant increases in actuator energy when tracking is matched.

Dev: And they provide a method for pairing up those operating points so you can actually compare them fairly. It gives designers a concrete way to see the energetic inefficiency hidden by trajectory-only metrics.

Taro: The implication for future work, which they touch on, is extending this to higher-dimensional systems and doing experimental validation outside of simulation to confirm these results in the real world.

Rosa: Exactly. We need more than just simulation results to really trust these kinds of energetic comparisons when we move these concepts into actual hardware testing.

Dev: So, the big picture here is that for complex mechatronic systems, especially those with nonlinearities, you have to look at both how accurately you track and how much physical energy you're actually consuming to do it.

Taro: It’s about making sure that when we design a system for the real world, we don't just create something that looks good on paper but drains the battery or burns out the motors because of hidden nonlinear energy costs.

Rosa: That’s what this paper is all about, showing that energy is not just a byproduct; it's a critical part of performance evaluation when you deal with nonlinear dynamics.

Dev: We'll keep an eye on how researchers apply this EAPI index in their real-time control loops to make sure we aren't just optimizing for tracking at the expense of the system's physical endurance.

The paper's summary: Rosa: So, we just talked about how we use standard tracking metrics to judge robot performance, and now we're looking at this paper that says those metrics are incomplete when you have nonlinear systems.

Dev: Right. This paper introduces this whole energy-aware framework, using something they call the EAPI index to combine the error you track with the actual energy your actuators spend.

Rosa: Exactly. The main point is that you can now compare two systems that track a path nearly identically on paper, but see if one is actually more efficient in how much power it uses.

Dev: They show that this difference comes from the nonlinear stiffness and friction terms in the system dynamics, which just don't exist in a simple linear model.

Rosa: They give us some concrete numbers where they compare a linear system and a nonlinear one, and even when tracking is very similar, the nonlinear version uses substantially more actuator energy—like over one hundred percent more.

Dev: That discrepancy comes from the difference in stored mechanical energy between those two systems, which scales with the fourth power of position because of that nonlinearity.

Rosa: It’s not just a small difference; it means that if you design a nonlinear system based only on tracking error, you could end up with something physically impractical because it demands too much power.

Dev: They set up this specific procedure where they find the exact operating point where the tracking error difference is minimal between the linear and nonlinear systems.

Rosa: That’s smart because it stops you from comparing two systems that are fundamentally different, and instead compares them when they are doing almost exactly the same thing.

Dev: And this matching procedure lets them map controller parameters to both performance metrics—the tracking error and the energy cost—so you can predict how a change in your controller will affect both things at once.

Rosa: So what this means for us on the ground, it suggests that when we’re building systems with real nonlinear physics, we have to worry about the physical effort required to move things, not just how close we are to the desired path.

Dev: It shifts the focus from just making a robot follow a line perfectly to making it follow that line efficiently within its physical energy constraints.

Rosa: And this opens up a whole new way of thinking about control design—not just optimizing for accuracy, but optimizing for energy efficiency too.

The paper's improvements: Rosa: So, we've looked at how this paper introduces that EAPI index to combine tracking error and energy cost into one number for comparison. Now we're talking about what they suggest to make this method better moving forward.

Dev: They point out that they need a way to predict how much energy you’ll use based on your control settings, which is called mapping controller parameters to pairs of tracking error and energy values.

Rosa: That’s a big one. It moves the evaluation from just looking at what happened during the trial to being able to predict the cost before you even run the system.

Dev: Exactly. It lets you see how increasing your control gains will affect both how accurately you track and how much energy you'll consume overall for that specific nonlinear system.

Rosa: That’s useful because if a design requires more power, we can see that immediately when tuning the controller parameters, instead of waiting for a long simulation run to find out.

Dev: It gives you a proactive safety case; you know the energetic trade-off before you deploy it in real time loops where latency matters.

Rosa: And they also suggest extending this idea to higher dimensional systems, which means applying this energy awareness to things with way more degrees of freedom than what they tested.

Dev: That’s a big leap. The current analysis is on a simple system, but the principle—that you must account for nonlinear energy costs—is clearly applicable to complex robotics or vehicle control.

Rosa: So the implication is that future work needs to focus on applying this framework to systems that are physically much more challenging than the ones they modeled here.

Dev: The real challenge will be making sure these predictive models actually work in real-time without adding too much computational load to the control loop itself.

Rosa: That's a fair point. So while this paper gives us a solid foundation for understanding nonlinear energy usage, we still have to figure out how to implement it efficiently in live systems.

Conclusion: Rosa: So, to wrap this up, this paper on "Energy-Aware Performance Evaluation of Nonlinear Mechatronic Systems Under Matched-Tracking Conditions" shows that energy isn't just a side effect; it’s a critical part of how we judge performance in nonlinear systems.

Dev: Right. The main idea is that you need to look at both the tracking error and the actuator energy, using that EAPI index, especially when dealing with nonlinear stiffness and friction.

Taro: I think this matters because it forces us to be realistic about what we're designing; if a system tracks well but drains its battery too fast, it's not a good design for real-world use.

Rosa: It really changes how we evaluate controllers; instead of just chasing the lowest error number, we’re now looking at the most balanced trade-off between accuracy and physical cost.

Dev: The matching procedure they used to pair up linear and nonlinear operating points is a key methodology because it ensures you're comparing systems that are actually performing similarly under those specific constraints.

Taro: I like how they showed the difference in stored mechanical energy, delta H, which is proportional to alpha times x to the fourth power; that tells us exactly where the nonlinear behavior starts contributing extra work.

Rosa: So even when you’re trying to keep things close to a linear model, those nonlinear terms create this hidden energy drain that we can finally quantify.

Dev: The implication for real-time control is that if you want a safe system, you need those energy metrics built into your performance evaluation criteria, not just the tracking error.

Taro: It pushes us to think about graceful recovery too; if the world misbehaves and we have to correct for that nonlinear stiffness, we need to know how much extra energy that correction will cost.

Rosa: So while this paper gives us a better tool for evaluation, the authors also point out they’re still working on extending this framework to higher dimensional systems and getting experimental validation outside of simulation.

Dev: That's the next step; moving from proving it in a simulation to making sure it holds up when we put it on physical hardware with real-world noise and latency.

Taro: I agree; seeing how those nonlinear effects play out under actual disturbance is where the true test for autonomy comes in.

Sachintha Bhanuka Dayawansa

Department of Electronic and Telecommunication Engineering, University of Moratuwa

eess.SY, cs.RO, cs.SY

Submitted: 2026-07-08

Updated: 2026-07-08

Comments: Accepted at MERCon 2026. To be presented at MERCon 2026. 6 pages, 5 figures

Journal ref: 2026 Moratuwa Engineering Research Conference (MERCon), Moratuwa, Sri Lanka, 2026, pp. 862-867

DOI: 10.1109/MERCon71835.2026.11691375

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 82/100

The gist: Trajectory-tracking metrics alone may not fully capture differences in system effort, and that including energy provides a more informative basis for performance evaluation.

Key concepts

EAPI (Energy-Aware Performance Index)
A simple index used to combine two performance aspects: tracking accuracy and actuator energy usage. It is calculated as a weighted average of normalized tracking error ($J_e$) and normalized actuator energy ($J_u$), allowing for a single measure to assess overall system efficiency.
Delta H(x) (Difference in Stored Mechanical Energy)
This term quantifies the difference in total stored mechanical energy between a nonlinear system and its linear counterpart at the same state. It is calculated as $1/4 imes x^4$, showing how nonlinear stiffness and friction introduce extra stored energy that linear systems do not have.
Matched-Performance Evaluation Procedure
This procedure ensures fair comparisons by finding operating points where the tracking error difference between a nonlinear system and a linear system is nearly zero. This isolates the comparison to show that energy differences are due to system dynamics, not just tracking accuracy.

Terminology

Summary

Trajectory-tracking metrics alone may not fully capture differences in system effort, and that including energy provides a more informative basis for performance evaluation. p. 0

Energy-Aware Evaluation Framework

The paper introduces an energy-aware evaluation framework that combines tracking metrics with a measure of cumulative actuator energy to enable comparison between systems with approximately matched performance. This framework utilizes a simple index (EAPI) to capture both aspects in a single measure, allowing for joint evaluation of tracking performance and actuator energy. p. 0 The index is defined as EAPI = λJe/Jref e + (1 - λ)Ju/Jref u. p. 0

Energetic Differences Under Nonlinear Dynamics

The analysis examines how nonlinear dynamics influence actuator energy requirements relative to linear systems, demonstrating that comparable tracking performance does not guarantee comparable actuator energy usage. The difference in stored mechanical energy between the nonlinear and linear systems is given by ∆H(x) = 1/4 αx 4. p. 1 Furthermore, the cumulative dissipated energy difference between the two systems is DN(T) - DL(T) = Fc/Z T 0 x˙ dt. p. 2 This shows that nonlinear stiffness and friction introduce additional contributions to actuator work that are not present in the linear system. p. 2

Matched-Performance Evaluation Procedure

To ensure a meaningful comparison, actuator energy is evaluated under approximately equivalent tracking conditions by defining a performance-constrained comparison where Je,1 − Je,2 ≤ ϵm. A parameterized controller exploration is used to generate comparable operating conditions by mapping controller parameters to (Je(θ), E(θ)) pairs for both linear and nonlinear systems. p. 3 The procedure involves selecting a nonlinear operating point θN such that the tracking error difference is minimized subject to the matching condition Je,N (θN) − Je L(θL) ≤ ϵm. p. 3

Results and Discussion

Simulation results consistently demonstrate that nonlinear systems can require substantially more actuator energy than linear counterparts even when tracking performance is nearly identical. For the strongest matched-performance case, the linear system achieved an Energy = 194. p. 4727 while the nonlinear system required Energy = 402. p. 4690, corresponding to an increase of approximately 106. p. 48% in actuator energy. p. 0 The proposed EAPI successfully reveals the hidden energetic inefficiency associated with the nonlinear dynamics, showing that the nonlinear system consistently produced larger EAPI values than the linear system across all matched operating conditions. p. 5

Conclusion

The results suggest that including energy alongside trajectory metrics provides a more informative basis for comparing system performance. The authors conclude that nonlinear stiffness and friction can increase energy requirements relative to linear systems, even when tracking accuracy is similar. p. 5 Future work will consider extension to higher-dimensional systems and experimental validation. p. 5

REFERENCES

[1] G. L. Hailu and T. p. 5 T. Yatayew, “Intelligent trajectory tracking in autonomous plantation robots using PSO-tuned nonlinear fuzzy PID, fuzzy PID, and conventional PID controllers,” Scientific Reports, vol. p. 5 p. 5 15, Art. no. 45357, 2025

[2] M. Y. Alwardat, O. p. 5 E.-L. M’bolo, Y. p. 5 Benslimane, and H. p. 5 M. Alwan, “Intelligent control of rigid-link manipulators: A systematic review of recent advances and future trends,” International Journal of Robotics and Control Systems, vol. p. 5 p. 5 5, no. 3, pp. 1949–1974, 2025

[3] S. Abd-Elhaleem, A. p. 5 H. El-Garawany, and M. p. 5 El-Brawany, “Adaptive intelligent controller for a lower limb rehabilitation robot using QAOAbased online membership optimization,” Scientific Reports, vol. p. 5 p. 5 16, 2026

[4] W. Sun, A. D. Ames, and others, “Tube-certified trajectory tracking for nonlinear systems with robust control contraction metrics,” IEEE Robotics and Automation Letters, vol. p. 5 7, no. 2, pp. 1–8, Apr. p. 5 2022

[5] A. M. Eissa, S. A. Mohamed, M. p. 5 I. Awad, H. E. Abd El Munim, and D. p. 5 Emad, “Trajectory tracking control of autonomous underwater vehicles: A review from classical methods to AI-based approaches,” International Journal of Dynamics and Control, vol. p. 5 14, Art. no. 32, 2026

[6] D. Zeng, S. Pan, Y. Yu, Y. Hu, J. Yang, P. Zhang, L. Xiong, G. Carbone, and L Gao. p. 5, “A comparative study on trajectory tracking control methods for automated vehicles,” Scientific Reports, vol. p. 5 15., 2025

[7] M. Hosseini, A. p. 5 Ranjbar Noei, and S. p. 5 J. S. Rostami, “Trajectory tracking control of an underwater vehicle in the presence of disturbance, measurement errors, and actuator dynamic and nonlinearity,” Robotica, vol. p. 5 41., no 10, pp 3059–3078, 2023

[8] Y. Satoh, “Trajectory tracking of nonlinear systems with convex input constraints based on tracking control Lyapunov functions,” Applied Sciences, vol. p. 5 p. 5 14., no 11, Art. no. 4377, 2024

[9] P. Zhao, Z. Guo, Y. Cheng, A. Gahlawat, H Kang, and N Hovakimyan. p. 5 p. 5, “Guaranteed trajectory tracking under learned dynamics with contraction metrics and disturbance estimation,” arXiv preprint arXiv:2112. p. 5 p. 508222., 2021

[10] Y. Kim and T Singh, “Energy-time optimal control of wheeled mobile robots,” arXiv preprint arXiv:2206. p. 502732., 2022

[11] D Yang, X Wei, and M Han, “Research on energy consumption optimization strategies of robot joints based on NSGA-II and energy consumption mapping,” Robotics, vol. 14., no 10, Art. no. 138., 2025

[12] B Laschowski and J McPhee, “Energy-efficient actuator design principles for robotic leg prostheses and exoskeletons: A review of series elasticity and backdrivability,” Journal of Computational and Nonlinear Dynamics, vol. 18., no 6, Art. no. 060801, 2023

[13] R B Hill, S Briot, A Chriette, and P Martinet., “Performing energyefficient pick-and-place motions for high-speed robots by using variable stiffness springs,” Journal of Mechanisms and Robotics, vol. p. 5 14., no 5, Art. no. 051004, 2022

[14] G L Hailu and T T Yatayew., “Intelligent trajectory tracking in autonomous plantation robots using PSO-tuned nonlinear fuzzy PID, fuzzy PID, and conventional PID controllers,” Scientific Reports, vol. 15, Art. no. 45357, 2025

[15] M Y Alwardat, O E.-L M’bolo, Y Benslimane, and H M Alwan. p. 5, “Intelligent control of rigid-link manipulators: A systematic review of recent advances and future trends,” International Journal of Robotics and Control Systems, vol. 5, no. 3, pp 1949–1974, 2025

[16] S Abd-Elhaleem, A H El-Garawany, and M El-Brawany., “Adaptive intelligent controller for a lower limb rehabilitation robot using QAOAbased online membership optimization,” Scientific Reports, vol. 16, 2026

[17] Y Satoh., “Trajectory tracking of nonlinear systems with convex input constraints based on tracking control Lyapunov functions,” Applied Sciences, vol. 14, no 11, Art. no. 4377, 2024

[18] P Zhao, Z Guo, Y Cheng, A Gahlawat, H Kang., and N Hovakimyan., “Guaranteed trajectory tracking under learned dynamics with contraction metrics and disturbance estimation,” arXiv preprint arXiv:2112.08222., 2021

[19] Y Kim and T Singh, “Energy-time optimal control of wheeled mobile robots,” arXiv preprint arXiv:2206.02732., 2022

[No token on page 1 for the reference list]

[No token on page 1 for the reference list]

[No token on page 1 for the reference list]

[No token on page 1 for the reference list]

[No token on page 2 for the reference list]

[No token on page 2 for the reference list]

[No token on page 3 for the reference list]

[No token on page 3 for the reference list]

[No token on page 4 for the reference list]

[No token on page 4 for the reference list]

[No token on page 5 for the reference list]

[No token on page 5 for the reference list]

[No token on page 6 for the reference list]

--- Page 1 ---

Energy-Aware Performance Evaluation of Nonlinear Mechatronic Systems Under Matched-Tracking Conditions <ref:2608.23578#pg2>

Sachintha Bhanuka Dayawansa Department of Electronic and Telecommunication Engineering University of Moratuwa Katubedda, Sri Lanka dayawansasb.22@uom.lk <ref:2608. p. 023578#pg2>

Abstract—Trajectory-tracking metrics such as root-meansquare error (RMSE), overshoot, and settling time are widely used to evaluate control performance in mechatronic systems <ref:2608.23578#pg2>

However, these measures describe output tracking alone and do not account for the actuator effort required to produce the observed motion <ref:2608. p. 023578#pg2> This limitation becomes more pronounced in nonlinear systems, where stiffness and dissipation depend on the system state <ref:2608. p. 0 p. 023578#pg2> This paper examines how these effects influence actuator energy under similar tracking conditions <ref:2608. p. 023578#pg2>. An energy-aware evaluation framework is introduced that combines tracking error with cumulative actuator energy and enables comparison between systems with approximately matched performance <ref:2608. p. 0 p. 023578#pg2>. A simple index (EAPI) is used to capture both aspects in a single measure <ref:2608. p. 0 p. 023578#pg2>.

Pruning redundant reasoning tokens

Trajectory-tracking metrics such as root-meansquare error (RMSE), overshoot, and settling time are widely used to evaluate control performance in mechatronic systems <ref:2608.23578#pg2>. However, these measures describe output tracking alone and do not account for the actuator effort required to produce the observed motion <ref:2608.23578#pg2>. This limitation becomes more pronounced in nonlinear systems, where stiffness and dissipation depend on the system state <ref:2608.23578#pg2>.

How the training signal is built

The paper examines how these effects influence actuator energy under similar tracking conditions <ref:2608.23578#pg2>. An energy-aware evaluation framework is introduced that combines tracking error with cumulative actuator energy and enables comparison between systems with approximately matched performance <ref:2608.23578#pg2>. A simple index (EAPI) is used to capture both aspects in a single measure <ref:2608.23578#pg2>.

Where the gains come from

Simulation results for linear and nonlinear systems under proportional–derivative control show that comparable tracking accuracy can correspond to significantly different actuator energy <ref:2608.23578#pg2>. The nonlinear system consistently requires more energy across matched operating points, reflecting the influence of nonlinear stiffness and friction <ref:2608. p. 023578#pg2>. These results suggest that trajectory-based metrics alone may not fully capture differences in system effort, and that including energy provides a more informative basis for performance evaluation <ref:2608. p. 023578#pg2>.

Energy-Aware Nonlinear Framework

Consider the nonlinear mechatronic system mx¨ + D(˙x) + R(x) = u, (1) <ref:2608.23578#pg2> To capture nonlinear energetic effects, the dissipative and restoring components are modeled as D(˙x) = cx˙ + Fc sgn(˙x), R(x) = kx + αx3, (2) <ref:2608. p. 123578#pg2>. The resulting dynamics become mx¨ + cx˙ + Fc sgn(˙x) + kx + αx3 = u. p. 1 (3) <ref:2608. p. 023578#pg2>.

Mechanical Energy Representation

The total stored mechanical energy can be defined as H(x, x˙) = T(˙x) + V (x), (4) <ref:2608.23578#pg2>. The kinetic energy is given by T(˙x) = 1/2 mxdot squared, (5) <ref:2608. p. 123578#pg2>. The potential energy satisfies dV/dx = R(x). p. 1 (6) <ref:2608. p. 023578#pg2>. Integrating the restoring force yields V (x) = 1/2 kx squared + 1/4 αx 4. p. 1 (7) <ref:2608. p. 023578#pg2>. Hence, the total energy becomes H(x, x˙) = 1/2 mxdot squared + 1/2 kx squared + 1/4 αx 4. p. 1 (8) <ref:2608. p. 023578#pg2>.

Energy Balance

Differentiating H with respect to time gives H˙ = mxdotx¨ + kxx˙ + αx3x.˙ (9) <ref:2608. p. 023578#pg2>. Substituting the system dynamics results in H˙ = xdot(u − cxdot - Fc sgn(˙x)), (10) <ref:2608. p. 123578#pg2>. This expression represents the energy balance of the system, where actuator input contributes power uxdot, while viscous damping and Coulomb friction dissipate energy <ref:2608. p. 123578#pg2>.

Actuator Energy

The instantaneous actuator power is defined as P(t) = u(t) ˙x(t). (12) <ref:2608. p. 023578#pg2>. The cumulative actuator energy over a time interval [0, T] is given by E(T) = Z T 0 u(t) ˙x(t) dt. p. 1 (13) <ref:2608. p. 023578#pg2>. This quantity measures the total energy transferred by the actuator, independent of direction, and depends on both control input and system response <ref:2608. p. 123578#pg2>.

Tracking Metrics and Energetic Limitation

Let xd(t) denote the desired trajectory and define the tracking error e(t) = xd(t) − x(t). (14) <ref:2608. p. 023578#pg2>. A commonly used performance measure is the root-meansquare error RMSE = s 1/T Z T 0 e 2(t) dt. p. 1 (15) <ref:2608. p. 023578#pg2>. However, this metric depends only on trajectory deviation and does not include actuator input, velocity, or energy-related quantities <ref:2608. p. 123578#pg2>.

Combined Performance Measure

To evaluate both tracking accuracy and energetic cost, define Je = Z T 0 e 2(t) dt, Ju = Z T 0 u(t) ˙x(t) dt. (16) <ref:2608. p. 023578#pg2>. Using reference values Jref e and Jref u, a normalized performance index is introduced as EAPI = λJe/Jref e + (1 - λ)Ju/Jref u, 0 ≤ λ ≤ 1. p. 1 (17) <ref:2608. p. 023578#pg2>.

Consider the nonlinear system mx¨ + cx˙ + Fc sgn(˙x) + kx + αx3 = uN, (18) and the corresponding linear system mx¨ + cxdot squared = uL. (19) <ref:2608. p. 023578#pg2>. The associated stored mechanical energies are HN = 1/2 mxdot squared + 1/2 kx squared + 1/4 αx 4, (20) <ref:2608. p. 0 p. 223578#pg2>. HL = 1/2 mxdot squared + 1/2 kx squared. (HL) <ref:2608. p. 023578#pg2>. The difference in stored energy is therefore ∆H(x) = HN − HL = 1/4 αx 4. p. 2 (20) <ref:2608.23578#pg2>.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Energy-Aware Performance Evaluation Framework Implementation: The system can now perform joint evaluation of tracking accuracy and energetic cost by using a normalized performance index (EAPI) to capture both aspects in a single measure, enabling comparison between systems with approximately matched performance.

  2. Nonlinear System Effort Quantification: The AI system can detect and quantify differences in actuator energy usage between linear and nonlinear counterparts, even when trajectory-tracking metrics appear similar, by analyzing the difference in stored mechanical energy (Delta H) and dissipated energy (DN - DL).

  3. Matched-Performance Pairing for Comparison: The system can generate scientifically meaningful comparisons by selecting a nonlinear operating point that minimizes the tracking error difference with a linear operating point, ensuring that differences in actuator energy are not simply due to differences in tracking accuracy.

  4. Identification of Nonlinear System Effects: The AI can specifically attribute observed energy differences to the nonlinear stiffness and friction mechanisms, as the analysis shows these terms introduce additional contributions to actuator work that are not present in the linear system.

  5. Predictive Energy Cost Assessment: By mapping controller parameters to pairs of tracking error and energy values (Phi(theta)), the AI can predict how increasing control gains will affect both trajectory accuracy and total actuator energy requirements for a given nonlinear system.

Sources

Related papers