Assessing changes in reliability methods over time: An unsupervised text mining approach

• The study systematically reviewed 30,543 reliability engineering papers, identifying 279 topics, which were reduced to eight main topics and three domain-specific ones.
• Findings suggest a shift in the focus of reliability engineering literature towards topics related to later stages in the development process, like prognostics and maintenance.
• The field of reliability engineering is evolving, showing a trend towards more practical and reactive interventions due to increased system complexity and uncertainty.

 The study by Charles K. Brown and Bruce G. Cameron analyzes how reliability engineering methods have evolved by performing a systematic literature review on 30,543 reliability engineering papers. Using topic modeling, they identified 279 topics, which were reduced to eight top-level method topics (prognostics, statistics, maintenance, quality control, management, physics of failure, modeling, and risk assessment) and three domain-specific topics (nuclear, infrastructure, and software). This comprehensive review offers insights into how the focus of reliability engineering literature has shifted, particularly towards topics associated with later stages of product development.

Reliability engineering faces the same fundamental challenge today as in the 1950s: the uncertainty in system representation. This challenge is exacerbated by the increasing complexity of modern systems. The study reveals that literature in this field increasingly addresses topics like prognostics, maintenance, and quality control, which are more relevant during the later stages of product development. This shift suggests a response to the difficulty in predicting reliability early in the development process due to the high level of uncertainty and the need to manage complex systems.

The authors highlight that the field has moved from early prediction efforts toward more practical and reactive approaches. For instance, the popularity of topics like prognostics and maintenance has grown over time, reflecting the industry’s adaptation to the challenges posed by system complexity and model uncertainty. By mapping the evolution of these topics, the study provides a window into how reliability engineering has adapted to the needs of modern complex systems and suggests that the field may continue to evolve toward more practical applications.

In conclusion, the study shows a significant growth in reliability engineering literature, with a notable trend toward addressing later-stage development processes. This evolution reflects the industry’s need to manage increasing complexity and uncertainty, highlighting a shift from early-stage predictive models to more reactive and practical interventions. The findings suggest that reliability engineering focuses more on practical applications that address real-world challenges in maintaining system performance and reliability over time.

Read the full article