Risk identification and assessment methods of offshore platform equipment and operations

Abstract

Risk management is crucial for the safety of offshore platforms. With advancements in safety management, much data on hidden dangers has been generated. The data contains a wealth of risk information, which realistically reflects the current state of safety on offshore platforms. However, this data is unstructured natural language, which cannot be analyzed statistically or queried with structured fields. Moreover, the manual identification of risk factors from hidden danger data is time-consuming and subject to subjective biases. A method for risk identification and assessment for offshore platform equipment and operations based on text mining of hidden danger data is proposed. From a big data perspective, a framework for expressing risk factors is established, which completes the transition from an unstructured text about the hidden danger to a structured expression of risk information. An automated method for risk identification is developed that combines machine learning, deep learning, and natural language processing techniques. The identified risk factors are used to build the Bayesian network model for data-driven risk assessment. Data from China’s Bohai oil field is used to demonstrate the effectiveness of this method. The result shows that it improves the identification of risk factors and the automation of the assessment.