• Àüü
  • ÀüÀÚ/Àü±â
  • Åë½Å
  • ÄÄÇ»ÅÍ
´Ý±â

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ³í¹®Áö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ³í¹®Áö > Çѱ¹Á¤º¸°úÇÐȸ ³í¹®Áö > Á¤º¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)

Á¤º¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)

Current Result Document : 13 / 15

ÇѱÛÁ¦¸ñ(Korean Title) ž籤 Áֱ⼺¿¡ ÀûÇÕÇÑ ½Ã°£ º¯¼ö¸¦ ¹Ý¿µÇÑ SHAP ±â¹ÝÀÇ Å¾籤 ¹ßÀü·® ¿¹Ãø
¿µ¹®Á¦¸ñ(English Title) SHAP based Solar Power Generation Forecasting Scheme Reflecting the Solar Periodic Time Variable
ÀúÀÚ(Author) Á¤¿ø¿ë   ¹Ú¼º¿ì   ¹®Àç¿í   ȲÀÎÁØ   Wonyong Chung   Sungwoo Park   Jaeuk Moon   Eenjun Hwang  
¿ø¹®¼ö·Ïó(Citation) VOL 28 NO. 03 PP. 0196 ~ 0201 (2022. 03)
Çѱ۳»¿ë
(Korean Abstract)
ÃÖ±Ù, Àü·Â»ý»ê °úÁ¤¿¡¼­ ȯ°æ ¿À¿°À» ¹ß»ý½ÃÅ°´Â È­·Â ¹ßÀüÀÇ ÀÇÁ¸µµ¸¦ ÁÙÀÌ°íÀÚ ½ÅÀç»ý¿¡³ÊÁö Áß ³ôÀº »ç¾÷¼ºÀ» °¡Áø ž籤 ¹ßÀüÀÇ ºñÀ²À» Áõ°¡½ÃÅ°°í ÀÖ´Ù. µû¶ó¼­ È¿À²ÀûÀÎ ¿¡³ÊÁö ¿î¿ë°ú ¾ÈÁ¤ÀûÀÎ Àü·Â °ø±ÞÀ» À§ÇØ Å¾籤 ¹ßÀü·®À» Á¤È®ÇÏ°Ô ¿¹ÃøÇÏ´Â ´Ù¾çÇÑ ¿¬±¸°¡ ¼öÇàµÇ°í ÀÖ´Ù. ÇÏÁö¸¸ ±âÁ¸ÀÇ Å¾籤 ¹ßÀü·® ¿¹Ãø ¿¬±¸µéÀº ¹ßÀü·®ÀÌ ¾ø´Â ¹ã ½Ã°£´ëÀÇ µ¥ÀÌÅ͵µ ÇÔ²² »ç¿ëÇÏ¿© Á¤È®µµ°¡ ³·°í ±â°èÇнÀÀ» ÅëÇØ ¿¹ÃøµÈ °ªÀÇ µµÃâ °úÁ¤À» ¼³¸íÇϱ⠾î·Á¿ü´Ù. ÀÌ¿¡, º» ³í¹®¿¡¼­´Â ž籤ÀÇ ½Ã°£ ÆÐÅÏÀ» ¹Ý¿µÇÑ ¼³¸í°¡´ÉÇÑ ÀΰøÁö´É ±â¹ÝÀÇ Å¾籤 ¹ßÀü·® ¿¹Ãø ±â¹ýÀ» Á¦¾ÈÇÑ´Ù. º» ±â¹ý¿¡¼­´Â ž籤 ¹ßÀü·®°ú ³ôÀº »ó°ü°ü°è¸¦ °®´Â ½Ã°£ º¯¼ö¸¦ ¼±ÅÃÇÏ°í Áֱ⼺À» ¹Ý¿µÇÏ´Â ÇüÅ·Πº¯ÇüÇÏ¿© »ç¿ëÇßÀ¸¸ç, ¹ã ½Ã°£´ë µ¥ÀÌÅ͸¦ Á¦¿ÜÇÏ°í ÇнÀÇß´Ù. ½ÇÇè °á°ú, ³· ½Ã°£´ë µ¥ÀÌÅ͸¦ »ç¿ëÇÑ LightGBMÀÇ ¼º´ÉÀÌ °¡Àå ¶Ù¾î³­ °Í°ú, LightGBMÀ» ¼³¸í°¡´ÉÇÑ ÀΰøÁö´ÉÀ¸·Î Çؼ®ÇßÀ» ¶§ Àϻ緮ÀÇ ¿µÇâÀÌ °¡Àå Å« °ÍÀ» È®ÀÎÇß´Ù.
¿µ¹®³»¿ë
(English Abstract)
Recently, environmental pollution caused by massive usage of fossil fuels in electric power production has become a critical issue in the world. According to this issue, the proportion of solar energy is expanding due to its high business potential among renewable energies, and various works have been proposed to precisely predict solar power generation for efficient energy management and stable power supply. However, previous works include data of night time when solar power is not generated, and this method leads to low forecasting accuracy. Also, they have limitations in explaining how the forecasting results are derived. To solve these problems, we propose a solar power generation forecasting scheme based on explainable AI(Artificial Intelligence) reflecting time period patterns of the sun. We selected time pattern variables showing high correlation with solar power generation and transformed them into a pattern reflecting the solar cycle. We then excluded data of night time and compared the performances of various machine learning models. Experimental results proved that utilizing LightGBM with only daytime input variables showed the best forecasting performance, and solar radiation had the most significant influence among the input variables.
Å°¿öµå(Keyword) LightGBM   ¼³¸í °¡´ÉÇÑ ÀΰøÁö´É   ž籤 ¹ßÀü·®   ½Ã°è¿­ µ¥ÀÌÅÍ ¿¹Ãø   LightGBM   XAI   solar power generation   time series forecasting  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå