TY - JOUR
T1 - ALK-PE
T2 - An efficient active learning Kriging approach for wave energy converter power matrix estimation
AU - Ren, Chao
AU - Tan, Jian
AU - Xing, Yihan
PY - 2023
Y1 - 2023
N2 - Wave energy is considered one of the most potential renewable energy. In the last two decades, many wave energy converters (WECs) have been designed to harvest energy from the ocean. Different power take-off systems are developed to maximize the power generation of WECs. However, the estimation of the power matrix of the WECs and annual power generation on the different sites is much more complex. A lot of simulations or experiments are required to obtain the power matrix of one specific WEC. To solve this problem, this paper proposes an active learning Kriging approach to estimate the WEC power matrix with less computational cost or experiment test. The efficiency of the proposed approach is demonstrated by two analytic problems and a point absorber WEC. The results show the proposed approach can efficiently and accurately estimate the power matrix of the WECs. Using the proposed ALK-PE approach, less than one-fifth of simulations or experiments are required to construct the whole power matrix of WECs at all the sea states, and the mean absolute percentage error is around 1%.
AB - Wave energy is considered one of the most potential renewable energy. In the last two decades, many wave energy converters (WECs) have been designed to harvest energy from the ocean. Different power take-off systems are developed to maximize the power generation of WECs. However, the estimation of the power matrix of the WECs and annual power generation on the different sites is much more complex. A lot of simulations or experiments are required to obtain the power matrix of one specific WEC. To solve this problem, this paper proposes an active learning Kriging approach to estimate the WEC power matrix with less computational cost or experiment test. The efficiency of the proposed approach is demonstrated by two analytic problems and a point absorber WEC. The results show the proposed approach can efficiently and accurately estimate the power matrix of the WECs. Using the proposed ALK-PE approach, less than one-fifth of simulations or experiments are required to construct the whole power matrix of WECs at all the sea states, and the mean absolute percentage error is around 1%.
KW - Active learning
KW - Kriging surrogate model
KW - Wave energy converter
KW - WEC power matrix estimation
UR - http://www.scopus.com/inward/record.url?scp=85168583940&partnerID=8YFLogxK
U2 - 10.1016/j.oceaneng.2023.115566
DO - 10.1016/j.oceaneng.2023.115566
M3 - Article
AN - SCOPUS:85168583940
SN - 0029-8018
VL - 286
JO - Ocean Engineering
JF - Ocean Engineering
M1 - 115566
ER -