Comparison of Computational Prognostic Methods for Complex Systems Under Dynamic Regimes: A Review of Perspectives

Abstract

Complex systems are expected to play a key role in the progress of Prognostics Health Management but the breadth of technologies that will highlight gaps in the dynamic regimes are expected to become more prominent and likely more challenging in the future. The design and implementation of sophisticated computational algorithms have become a critical aspect to solve problems in many prognostic applications for multiple regimes. In addition to a wide variety of conventional computational and cognitive paradigms such as machine learning and data mining fields, specific applications in prognostics have led to a wealth of newly proposed methods and techniques. This paper reviews practices for modeling prognostics and remaining useful life applications in complex systems working under multiple operational regimes. An analysis is provided to compare and combine the findings of previously published studies in the literature, and it assesses the effectiveness of techniques for different stages of prognostic development. The paper concludes with some speculations on the likely advances in fusion of advanced methods for case specific modeling.

Predictive Maintenance for Pump Systems and Thermal Power Plants: State-of-the-Art Review, Trends and Challenges

“…The weakness of digital twins is the time they consume before being established [21]. Knowledge-based models enable the use of domain expert knowledge and deploy it in models, this can be beneficial as it can assist the prediction process [22]. Data-driven models mainly apply machine learning, soft computing, and statistical theory to establish a model that takes in historical operational or condition data.…”

“…Stochastic data-driven models within prognostics are often considered within the Bayesian category. Rather than giving a single estimated output on the current system health, it gives a probability distribution of possible likely options [22]. In this way, the Bayesian method can present the current state of the system, but can also evaluate future trends before a given threshold.…”

“…Where stochastic algorithms give a probability distribution as an output, the statistical algorithms give a single specified output. A common algorithm is ARIMA, but it is also possible to apply regression techniques [22]. SVM sees its use in both classification and regression problems.…”

“…As technology has developed and ANN has taken on increasingly complex tasks, a new term has been given to more complex ANN, which is called Deep Neural Network. The most distinct difference between the two terms is the complexity, referring to the architecture of the networks and the number of layers that it may contain [22,63,64].…”

“…Table 3. An overview of some of the general advantages and limitations of the algorithms within the given categories [22,63]. As individual algorithms present their own set of strengths and weaknesses the reader is recommended to study the individual algorithm as well.…”