Proceedings of the Eighth International Conference on Probabilistic Safety Assessment & Management (PSAM)
Transportation: Approach and Methodology
Download citation file:
- Ris (Zotero)
- Reference Manager
For the tracked vehicle, basing on the comprehensive analysis of its performance and operational characteristics, research the management and technical ways of improving vehicle's operating reliability, such as improving operating environment, optimizing maintenance decision and applying advanced test methods.
Port State Control (PSC) is an effective measure to verify and ensure that the condition and equipment of foreign ships visiting national ports are in compliance with the requirements of international safety and anti-pollution standards. Ships that are found to be in serious violation of standards are detained in port until their deficiencies have been rectified. The objective of PSC is to detect substandard ships and eliminate the threat that they pose to life, property and the marine environment. In order to ensure the effectiveness of PSC, regional port state control organizations are established and agreements on port state control — “Memoranda of Understanding” or MOUs — have been signed covering all of the world's oceans. For Asia and Pacific region, “Tokyo MOU” is signed by 18 authorities. One of the effective methods that authorities adhere to the MOUs is to develop a common targeting system that determines which ships in port should be inspected.
There are a number of targeting systems proposed and introduced into PSC, however, most of the systems are based on experiential measures or simple comparison of a few attributes of the ships.
In this research we introduce discriminant analysis as a risk assessment method, in order to promote the efficiency of PSC by developing a more accurate and flexible targeting system for ships-in-port classification. Discriminant analysis generates classification functions for a sample of observations from pre-determined categories. A large number of variables can be adapted to the functions. We applied Linear Discriminant Function to historical inspection data. Sixteen attributes recorded in Tokyo MOU database, e.g. ship type, ship age, flag state, and previous inspection results, are utilized as the variables of the function. Coefficients of variables are produced by 32,121 inspection data set collected through the years 2002 and 2003. The developed discriminant function estimates the possibility of detention for each arriving ship. However general solution of the Linear Discriminant Function maximizes the percentage of correct classification, adjustment of constant term enables us to calculate an overall detention rate for a given inspection ability. Among the attributes, flag state, classification society, and the number of detentions during last 4 inspections are the three major factors that influence the estimation. By extending the function, we also developed a function that estimates the required inspection rate to achieve a given rate of detention. Authorities can adopt it as a management measure because it helps to determine the number of inspection officers or other resources for their ports. The functions we proposed can be updated to adjust to the newly obtained data in MOU inspection database. Finally, this paper shows the comparison between the proposed targeting system and the 2003 revised targeting system developed by Tokyo MOU. The comparison result shows that the former provides better capture rates in the range of 10–60% inspection rates.