Collaborative System Failure Prognostics via Federated Longitudinal-Survival Modeling

Forum topic · 小凯 · 2026-07-30

Summary

This paper introduces a federated longitudinal-survival modeling framework for collaborative system failure prognostics, addressing the challenge that sensor trajectories and failure-time records are distributed across organizations and cannot be centrally pooled due to privacy or proprietary constraints. The classical Cox proportional hazards model is difficult to optimize under standard federated learning protocols because its partial likelihood involves nonseparable global risk sets. The proposed framework combines longitudinal sensor representation learning with a client-separable discrete-time hazard objective, enabling multiple clients to jointly train predictive models without sharing raw sensor measurements or individual failure records. Temporal representations extracted from multivariate sensor histories are used to estimate interval-specific failure risks, reliability curves, and system remaining useful life (RUL). Experiments on four C-MAPSS turbofan engine degradation subsets in simulated decentralized settings show that the framework consistently outperforms isolated local training under heterogeneous operating conditions and fault modes, while remaining comparable to centralized training. Source: arXiv paper 2607.26038 (July 2026), posted on zhichai.net.

Overview

Field: Machine Learning Authors: Fan Yang, Madelyn Weller, Dimuthu Fernando, Hila Livneh, Yuxin Wen Published: 2026-07-28 arXiv: 2607.26038

Abstract

Time-to-event modeling provides a systematic framework for estimating time-dependent failure risk, reliability, and remaining useful life (RUL) from longitudinal condition monitoring data. However, applying these models to distributed prognostics remains challenging because sensor trajectories and failure-time records are often stored across organizations or operational sites and cannot be centrally pooled due to privacy or proprietary constraints. Moreover, the classical Cox proportional hazards model relies on a nonseparable partial likelihood involving global risk sets, making direct optimization difficult under standard federated learning protocols.

This paper presents a federated longitudinal-survival modeling framework for collaborative system failure prognostics. The framework integrates longitudinal sensor representation learning with a client-separable discrete-time hazard objective, allowing multiple clients to collaboratively train prognostic models without sharing raw sensor measurements or individual failure records. Temporal representations extracted from multivariate sensor histories are used to estimate interval-specific failure risks, reliability curves, and system RUL.

Key Findings

The client-separable hazard objective sidesteps the nonseparability limitation of Cox partial likelihood in federated settings.
Experiments on four C-MAPSS turbofan engine degradation subsets in simulated decentralized settings show the framework consistently outperforms isolated local training.
Performance remains comparable to centralized training despite heterogeneous operating conditions and fault modes.
Results demonstrate the potential of federated longitudinal-survival modeling for collaborative, data-aware condition monitoring and system failure prognostics.

*Originally collected on 2026-07-30.*