From Winding-Fault Geometry to Reliability: Estimation and Prognosis of Stator Inter-Turn Faults
summary
The gist
This paper develops an integrated framework for estimation, prognosis, and reliability assessment of stator inter-turn faults in induction motors The resulting framework provides a unified connection…
In short
This work creates a complete framework to detect, estimate, and predict failures in induction motor windings caused by inter-turn faults. It uses an augmented particle filter to simultaneously estimate the motor's operational state and the unknown fault severity. This allows for online diagnosis, forecasting degradation trends using multiple models, and assessing long-term reliability based on various failure criteria.
Key concepts
- Geometric Fault Representation (ηcc, γcc)
- Inter-turn faults are visualized using two parameters: ηcc (magnitude of the fault's effect) and γcc (spatial orientation). This geometric model simplifies the complex physics by showing how the fault current relates to its size and direction within the stator structure, making it physically intuitive for analysis.
- Augmented-State Particle Filter
- A sophisticated estimation tool used to find both a motor's current operational status (like speed and currents) and the unknown fault severity (ηcc). It works by simulating many possible states simultaneously, providing an online estimate of how severe the fault is while also quantifying the uncertainty in that estimate.
- Prognosis Mechanisms
- Four different mathematical models—linear trend, Holt exponential smoothing, Bayesian degradation, and particle forecasting—are applied to the estimated fault severity history. These mechanisms allow the system to generate multiple predictions about how fast the winding will degrade over time, offering both certain and probabilistic forecasts.
- Reliability Assessment
- This step uses the degradation predictions to determine if a motor will fail under different conditions. Formulations include calculating a predicted failure time (threshold crossing), using statistical models like Weibull for lifetime, or assessing risk based on how the fault severity evolves over time.
Terminology used across episodes
This episode discusses
- From Winding-Fault Geometry to Reliability: Estimation and Prognosis of Stator Inter-Turn Faults · Paper Radio
- Neural-Enhanced Micro-Kalman Filtering for Satellite Tracking: A Comparative Study
- Distributed Estimation with Decentralized Control for Quadruple-Tank Process
- On the Comparison of Reinforcement Learning and Adaptive Control for Linear Systems under Packet Loss and Uncertainty · Paper Radio
The paper
From Winding-Fault Geometry to Reliability: Estimation and Prognosis of Stator Inter-Turn Faults · Read on arXiv
Institut Teknologi Sepuluh Nopember · Universitas Pembangunan Nasional Veteran Jawa Timur
This paper develops an integrated framework for estimation, prognosis, and reliability assessment of stator inter-turn faults in induction motors. The fault is characterized by its severity, defined as the fraction of short-circuited turns, and its spatial orientation. A geometric fault model shows that the resulting output signature is affine in the fault severity and exhibits a fundamental spatial periodicity. Exploiting this structure, an augmented-state particle filter jointly estimates the nonlinear electromechanical state and the unknown fault severity while quantifying posterior uncertainty. The estimated degradation is then propagated using four prognostic models: linear trend, Holt exponential smoothing, Bayesian degradation, and particle-based forecasting. Their predictions are connected to threshold-crossing remaining useful life (RUL), first-passage reliability, degradation-dependent hazard reliability, and a Weibull lifetime benchmark, thereby providing both deterministic and probabilistic health assessments. Numerical results demonstrate accurate online fault estimation and output reconstruction, characterize the effects of degradation pattern and prediction horizon on prognosis, and show consistent reliability and threshold-crossing predictions. Moreover, estimation accuracy remains comparable across the distinct fault orientations induced by the spatial periodicity. The resulting framework provides a unified connection from physics-based inter-turn fault modeling to online diagnosis, degradation prognosis, and reliability assessment.
Transcript
Introduction to the show: ident: Robotics Radio. Generated commentary on the latest robotics and control papers.
Rosa: We're looking at the paper: "From Winding-Fault Geometry to Reliability".
Dev: Here's the gist of the paper. This paper develops an integrated framework for estimation, prognosis, and reliability assessment of stator inter-turn faults in induction motors The resulting framework provides a unified connection from physics-based inter-turn fault modeling to online diagnosis, degradation prognosis, and reliability assessment Fault Characterization and Modeling The paper introduces a geometric representation of stator inter-turn faults in terms of the physically interpretable pair (ηcc, γcc) The fault-induced current is represented through a geometric orientation model and factorized into the product of ηcc and an orientation- and operating-condition-dependent signature This yields a measured-output model that is affine in the unknown fault severity The spatial orientation γcc is fixed with respect to the stator, whereas the synchronous (d, q) reference frame rotates during operation The fault-induced current in the synchronous (d, q) frame is given by icc,dq = 2ηcc / 3Rs P(-pγ)Q(pγcc)P(pγ)Vdq This equation separates the roles of the two fault parameters: ηcc determines the magnitude of the fault-induced current, whereas γcc determines its spatial orientation through Q(pγcc) The short-circuit contribution admits the compact factorization icc,dq = ηccG(γ, γcc, Vdq) Estimation Framework An augmented-state particle filter is developed for joint estimation of the nonlinear electromechanical state and the constrained fault severity ηcc ∈
0, 1: The augmented state zk:= xk ηcc,k ∈ Z:= R 5 ×
0, 1: contains the five electromechanical states governing the nominal motor dynamics together with the scalar degradation variable ηcc,k governing the fault contribution to the measured output The measurement noise satisfies vk ∼ N (0, R), where R ∈ R 3×3 and is positive definite The posterior particle population provides both an online severity estimate and its uncertainty, rather than only a point-valued fault indicator The estimated degradation history is available for prognosis, where the estimated fault severity is propagated forward to predict its future evolution and subsequent reliability Prognosis Mechanisms Four prognosis mechanisms are formulated on the same PF-based health information: linear trend, Holt exponential smoothing, Bayesian degradation, and particle-based forecasting The linear-trend model predicts future winding degradation by approximating the recent evolution of the estimated fault severity with a straight line The Holt exponential smoothing updates these quantities whenever a new estimate ηbcc,k becomes available The Bayesian degradation model introduces the latent degradation process ηecc,k+1 = ηecc,k + µ + ξk, where µ ∈ R is the unknown mean degradation increment per sampling interval and ξk ∼ N (0, σ 2d), with σ 2d > 0 The particle-based forecast propagates an empirical representation of the current severity uncertainty through propagated particles Reliability Assessment Reliability assessment uses these predictions to quantify whether the winding degradation remains below a prescribed failure condition Four reliability formulations are considered below: deterministic threshold crossing, Weibull lifetime reliability, Bayesian degradation reliability, and degradation-dependent hazard reliability The threshold-crossing formulation maps each point forecast ηb qcc,k+rk directly into a predicted failure time The Weibull model describes the motor lifetime statistically using a two-parameter Weibull distribution with scale λW > 0 and shape βW > 0 The Bayesian reliability measures survival at a particular future instant, whereas first-passage reliability requires the complete predicted trajectory to remain below the critical severity Degradation-dependent hazard reliability maps the evolving fault severity into accumulated failure risk by defining the degradation-dependent hazard h q k(r) = h0 exp αηb qcc,k+rk/ηcrit!
Title and authors: Dev: So, let's look at the actual authors and what this paper is trying to achieve in its title, "From Winding-Fault Geometry to Reliability: Estimation and Prognosis of Stator Inter-Turn Faults." The title suggests a complete journey from understanding the physical shape of the fault right through to making reliability predictions.
Rosa: Right. It’s not just about finding the fault; it’s about using that geometric information—the shape—to build a system that can estimate how bad it is, predict how fast it's getting worse, and finally tell you when the motor is going to fail.
Taro: So, if I understand correctly, the authors are trying to bridge the gap between a physical observation of damage and a usable health indicator for long-term planning.
Dev: Precisely. They are proposing an integrated framework that links the physics of how faults occur in the winding structure to online diagnosis and future reliability assessment. It’s about making that connection explicit so it’s not just some theoretical exercise in a lab.
Rosa: And this paper is important because it moves beyond just detecting a fault to quantifying its severity in a way that respects the physical structure of the motor itself, which is what Taro mentioned earlier.
Taro: I guess that means if we use this framework, we can get more reliable predictions about the machine’s degradation over time.
The paper's summary: Dev: So, looking at the summary of "From Winding-Fault Geometry to Reliability: Estimation and Prognosis of Stator Inter-Turn Faults," it boils down to them developing a unified method. They start with a geometric model that shows the fault signature is affine in the fault severity and has a spatial periodicity.
Rosa: That's the core concept. They use this structure—that geometric relationship—to build an augmented state particle filter that estimates both the motor’s mechanical and electrical state, along with that unknown fault severity variable, which they call eta cc.
Taro: So, they aren't just estimating the motor’s current condition; they are simultaneously tracking how severe the fault itself is. That joint estimation is a big step forward for real-time health monitoring.
Dev: Right. And once they have that joint estimate, they feed it into four different prognostic models: linear trend, Holt exponential smoothing, Bayesian degradation, and particle-based forecasting. These are all using the same information to predict future winding degradation.
Rosa: So what this means for us on the ground is that instead of just getting a single point about fault severity at one moment, we get an estimate with uncertainty attached to it, which Taro mentioned earlier is really valuable for planning.
The paper's improvements: Taro: Now, looking at what they suggest as improvements to their own work in "From Winding-Fault Geometry to Reliability: Estimation and Prognosis of Stator Inter-Turn Faults," they focus on making the estimation process more robust and interpretable.
Dev: They introduce a fault signature vector, psi t gamma cc, which is designed specifically to isolate how the measured output depends on the unknown fault severity variable eta cc. This lets them calculate that severity variable directly from a fault residual, which is basically the difference between what they expected and what they actually measured.
Rosa: That sounds like they are creating a direct pathway from noise in the measurement back to an estimate of the physical degradation, which is powerful for diagnosis.
Taro: And they also add a constraint on that estimated severity variable eta cc, ensuring it stays within a physical range, specifically between zero and one. This keeps the degradation variable physically interpretable throughout the entire online estimation process.
Dev: That constraint is crucial for maintaining physical meaning; without it, you could get some weird number that doesn't make sense in a real motor winding. It’s about keeping the model grounded in reality as it runs live.
Conclusion: Rosa: So to wrap up on this paper, the authors show how they connect the physical geometry of stator inter-turn faults to a full estimation and prognosis framework that includes joint state estimation and multiple prognostic methods for winding degradation.
Dev: It’s a comprehensive approach because it doesn't just stop at detection; it lets you use those predictions for reliability measures like deterministic threshold crossing, Weibull lifetime, and degradation-dependent hazard reliability.
Taro: I think the most important part is that they provide these different ways to look at the same predicted winding condition, which gives maintenance planners a lot of options for how much risk they want to take.
Rosa: Right. They give you deterministic failure times based on those forecasts, and probabilistic measures like survival probability or hazard rates, depending on what kind of assessment you need.
Dev: And the authors also acknowledge that while they show this framework works across nine different spatial orientations in their numerical results, the practical challenge is ensuring this high-level estimation remains accurate when the world misbehaves in a real-world setting.
More episodes
- 2608.13438-ContactGuard: Pre-Contact Execution Monitoring with Action-Conditioned Latent World Models
- 2610.34593-Future Prediction as Training-Only Video Supervision for Pretrained VLA Policies
- 2610.35385-EmbodiedLab: Beyond Manipulation, Benchmarking Embodied Agents That Act to Know
- 2610.36378-Data-Driven Hamiltonian for Neural Optimal Control
- 2610.37275-A Heuristic for Scheduling Autonomous Vehicles at Intersections
- 2610.37288-Dynamic Coordination of Autonomous Vehicles at Intersections
- 2610.37632-Real-Time Coordination of Autonomous Vehicles at Intersections
- 2611.13779-Softness is something you also need! Why compliance helps control: information metrics of a passive walker with compliant ankle
- 2611.13411-Residual-Corrected Physics-Informed Neural Network KKL Observers with Input-to-State Stability Guarantees
- 2611.14488-Is Your Robot's Memory Just a Clock? Temporal Shortcuts in Memory-Augmented VLA Policies