A high-fidelity weather time series generator using the Markov Chain process on a piecewise level

1st Conference of Computational Methods in Offshore Technology

计算机科学

Hersvik, K.^1 ; Endrerud, O.-E.V.^1

Shoreline AS, Stavanger, Norway^1

关键词: High-fidelity;    Markov chain process;    Markov model;    Offshore operations;    Piece-wise;    Seasonal characteristics;    Statistical quality;    Time-series generators;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/276/1/012003/pdf
DOI  :  10.1088/1757-899X/276/1/012003

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

A method is developed for generating a set of unique weather time-series based on an existing weather series. The method allows statistically valid weather variations to take place within repeated simulations of offshore operations. The numerous generated time series need to share the same statistical qualities as the original time series. Statistical qualities here refer mainly to the distribution of weather windows available for work, including durations and frequencies of such weather windows, and seasonal characteristics. The method is based on the Markov chain process. The core new development lies in how the Markov Process is used, specifically by joining small pieces of random length time series together rather than joining individual weather states, each from a single time step, which is a common solution found in the literature. This new Markov model shows favorable characteristics with respect to the requirements set forth and all aspects of the validation performed.