A Bayesian network structural learning algorithm for calculating the failure probabilities of complex engineering systems with limited data

The failure of complex engineering systems is easy to lead to disastrous consequences. To prevent the failure, it is necessary to model complex engineering systems using probabilistic techniques with limited data which is a major feature of complex engineering systems. It is a good choice to perform such modeling using Bayesian network because of its advantages in probabilistic modeling. However, few Bayesian network structural learning algorithms are designed for complex engineering systems with limited data. Therefore, an algorithm for learning the Bayesian network structure of them should be developed. Based on the process of self-purification of water, a complex engineering system is segmented into three components according to the degree of difficulty in solving them. And then a Bayesian network learning algorithm with three components (TC), including PC algorithm, MIK algorithm which is originated by the paper through combining Mutual Information and K2 algorithm, and the Hill-Climbing method, is developed, i.e. TC algorithm. To verify its effectiveness, TC algorithm, K2 algorithm, and Max-Min Hill-Climbing are respectively used to learn Alarm network with different sizes of samples. The results imply that TC algorithm has the best performance. Finally, TC algorithm is applied to study tank spill accidents with 220 samples.

Understanding design change propagation in complex engineering systems using a digital twin and design structure matrix

PurposeAs the engineering design process becomes increasingly complex, multidisciplinary teams need to work together, integrating diverse expertise across a range of disciplinary models. Where changes arise, these design teams often find it difficult to handle these design changes due to the complexity and interdependencies inherent in engineering systems. This paper aims to develop an innovative approach to clarifying system interdependencies and predicting the design change propagation at the asset level in complex engineering systems based on the digital-twin-driven design structure matrix (DSM).Design/methodology/approachThe paper first defines the digital-twin-driven DSM in terms of elements and interdependencies, where the authors have defined three types of interdependency, namely, geospatial, physical and logical, at the asset level. The digital twin model was then used to generate the large-scale DSMs of complex engineering systems. The cluster analysis was further conducted based on the improved Idicula–Gutierrez–Thebeau algorithm (IGTA-Plus) to decompose such DSMs into modules for the convenience and efficiency of predicting design change propagation. Finally, a design change propagation prediction method based on the digital-twin-driven DSM has been developed by integrating the change prediction method (CPM), a load-capacity model and fuzzy linguistics. A section of an infrastructure mega-project in London was selected as a case study to illustrate and validate the developed approach.FindingsThe digital-twin-driven DSM has been formally defined by the spatial algebra and Industry Foundation Classes (IFC) schema. Based on the definitions, an innovative approach has been further developed to (1) automatically generate a digital-twin-driven DSM through the use of IFC files, (2) to decompose these large-scale DSMs into modules through the use of IGTA-Plus and (3) predict the design change propagation by integrating a digital-twin-driven DSM, CPM, a load-capacity model and fuzzy linguistics. From the case study, the results showed that the developed approach can help designers to predict and manage design changes quantitatively and conveniently.Originality/valueThis research contributes to a new perspective of the DSM and digital twin for design change management and can be beneficial to assist designers in making reasonable decisions when changing the designs of complex engineering systems.

The Need for Systems Awareness to Support Early-Phase Decision-Making—A Study from the Norwegian Energy Industry

In this paper, we explore the need to improve systems awareness to support early-phase decision-making. This research uses the Norwegian energy industry as context. This industry deals with highly complex engineering systems that shall operate remotely for 25+ years. Through an in-depth study in a systems supplier company, we find that engineers are not sufficiently aware of the systems operational context and do not focus on the context in the early phase. We identified the lack of a holistic mindset and the challenge of balancing internal strategy and customers’ needs as the prevalent barriers. To support the concept evaluation, the subsea system suppliers need to raise systems awareness in the early phase. The study identifies four aspects that are important to consider when developing and implementing approaches to improve systems awareness in the early phase.

Multistream Sensor Fusion-Based Prognostics Model for Systems Under Multiple Operational Conditions

The residual useful lifetime prediction of complex engineering systems operating under dynamic multi-operational conditions is crucial yet challenging. One of the key challenges is that sensing signals consist of both degradation information that resulted from the physical deterioration of the system and amplitude jumps due to the switch of operational conditions. To accurately predict the failure times, we need to separate the degradation information from sensing signals and exclude jumps. Another challenge stems from the fact that complex engineering systems are typically monitored by multiple sensors that generate multi-stream sensing signals but not all the signals are informative for failure time prediction. Therefore, a systematical sensor selection method needs to be developed to identify informative sensors. The third challenge is that signals from the informative sensors are high-dimensional and may provide redundant information. As a result, a data fusion method is needed to fuse the multi-stream signals from the informative sensors. To address the aforementioned challenges, this paper focuses on developing a prognostic framework for systems that operate under dynamic multiple operational conditions. The framework will first extract the degradation signal from each sensor by removing the jump information resulted from the change of operational conditions. Next, informative sensors are selected using a sensor selection method. Finally, the degradation signals from the selected sensors are fused to predict the failure time of a partially degraded system. The effectiveness of the proposed prognostic framework is validated using a degradation data set of aircraft turbofan engines from NASA repository.