What is Preliminary Hazard Analysis (PHA)

As part of the certification process for a new rail transport system, one of the essential steps is to examine, from a safety point of view, that the system (and its technical and human environment) satisfies the regulatory requirements of safety and in particular that it can in no case be a source of risk for travelers. This intellectual process by which a certification expert assesses a situation, predicts an event, or makes a decision is often difficult to model in the form of reliable and definitive algorithms. This difficulty can be partially overcome by using artificial intelligence (AI) techniques. In order to improve the usual approaches to analysis and evaluation of safety studies used in the context of certification, this chapter presents several methods and tools based on AI techniques and in particular on knowledge acquisition methods, supervised machine learning (SML), case-based reasoning (CBR), and knowledge-based systems (KBS).