The complexity of industrial processes imposes a lot of challenges in building accurate and representative causal models for abnormal events diagnosis, control and maintenance of equipment and process units. This paper presents an innovative data-driven causality modeling approach using interpretable machine learning and process mining techniques, in addition to human expertise, to efficiently and automatically capture the complex dynamics of industrial systems. The approach tackles a significant challenge in the causality analysis community, which is the discovery of high-level causal models from low-level continuous observations. It is based on the exploitation of event data logs by analyzing the dependency relationships between events to generate accurate multi-level models that can take the form of various state-event diagrams. Highly accurate and trustworthy patterns are extracted from the original data using interpretable machine learning integrated with a model enhancement technique to construct event data logs. Afterward, the causal model is generated from the event log using the inductive miner technique, which is one of the most powerful process mining techniques. The causal model generated is a Petri net model, which is used to infer causality between important events as well as a visualization tool for real-time tracking of the system’s dynamics. The proposed causality modeling approach has been successfully tested based on a real industrial dataset acquired from complex equipment in a Kraft pulp mill located in eastern Canada. The generated causality model was validated by ensuring high model fitness scores, in addition to the process expert’s validation of the results.

Existing research on vulnerability discovery models shows that the existence of vulnerabilities inside an application may be linked to certain features, e.g., size or complexity, of that application. However, the applicability of such features to demonstrate the relative security between two applications is not well studied, which may depend on multiple factors in a complex way. In this paper, we perform the first large-scale empirical study of the correlation between various features of applications and the abundance of vulnerabilities. Unlike existing work, which typically focuses on one particular application resulting in limited successes, we focus on the more realistic issue of assessing the relative security level among different applications. To the best of our knowledge, this is the most comprehensive study of 780 real world applications involving 6,498 vulnerabilities. We apply seven feature selection methods to nine feature subsets selected among 34 collected features, which are then fed into six types of machine learning models, producing 523 estimations. The predictive power of important features is evaluated using four different performance measures. Our study reflects that the complexity of applications is not the only factor in vulnerability discovery, and that human related factors contribute to explaining the number of discovered vulnerabilities in an application.

This paper proposes a multi-agent-based methodology for real-time fault diagnosis in industrial processes. This work aims to build a decision support tool that helps process operators identify and better manage abnormal situations. Supervised and semi-supervised machine learning methods are widely used to develop such tools. Despite their accuracy in classifying faults, supervised methods have a major limitation: they cannot diagnose novel faults. Semi-supervised methods can detect and isolate novel faults, but cannot disclose their root causes. The proposed methodology combines both supervised and semi-supervised methods in a parallel-serial structure, exploiting their respective strengths. Moreover, it provides the process expert with meaningful explanations of detected novel faults or otherwise. Two case studies are used in this paper to demonstrate the effectiveness of the proposed methodology. The first case is the Tennessee Eastman Process benchmark. The second one uses real data collected from a heat recovery system in a thermomechanical pulp mill.