³í¹®¸í |
Åͺ¸ ³Ãµ¿±â ±â°èÇнÀ ¸ðµ¨°ú °¡¿ì½Ã¾È È¥ÇÕ ¸ðµ¨À» ÀÌ¿ëÇÑ À¯È¿¼º °ËÁõ / Machine Learning Model for a Turbo Chiller and Validity Check of the Model by Gaussian Mixture Model / ´ëÇлýºÎ¹® |
ÀúÀÚ¸í |
¹Ú¼ºÈ£(Park, Sung-Ho) ; ¾È±â¾ð(Ahn, Ki Uhn) ; À̵¿Çõ(Yi, Dong-Hyuk) ; Ȳ½ÂÈ£(Hwang, Aaron) ; ÃÖ¼±±Ô(Choi, Sunkyu) ; ¹Úö¼ö(Park, Cheol Soo) |
¼ö·Ï»çÇ× |
´ëÇѰÇÃàÇÐȸ Çмú¹ßÇ¥´ëȸ ³í¹®Áý, Vol.37 No.2 (2017-10) |
ÆäÀÌÁö |
½ÃÀÛÆäÀÌÁö(550) ÃÑÆäÀÌÁö(2) |
ÁÖÁ¦¾î |
Àΰø½Å°æ¸Á ; °¡¿ì½Ã¾È È¥ÇÕ ¸ðµ¨ ; ±â°èÇнÀ ; Artificial Neural Network ; Gaussian Mixture Model ; Machine Learning |
¿ä¾à1 |
ÃÖ±Ù ±Þ¼ÓÇÏ°Ô ¹ßÀüÇÑ ±â°èÇнÀÀÌ ±âÁ¸ °ÇÃ๰ BEMS µ¥ÀÌÅÍ¿¡ Àû¿ëµÇ¾î, ¼³ºñ ½Ã½ºÅÛ µ¿Àû °Åµ¿ ¹× ÃÖÀûȸ¦ ´Þ¼ºÇϰíÀÚ ÇÏ´Â ³ë·ÂÀÌ È°¹ßÇÏ´Ù(Kim et al., 2016). ¿Àü´Þ ¹× ¿¿ªÇÐÀ» ±â¹ÝÀ¸·Î ÇÏ´Â Á¦ 1¹ýÄ¢ ¸ðµ¨(¿¹:µ¿Àû ½Ã¹Ä·¹À̼Ç)°ú ´Þ¸®, ±â°èÇнÀ ¸ðµ¨Àº ÀÔÃâ·Â µ¥ÀÌÅÍÀÇ »ó°ü°ü°è ¸¸À¸·Î ±¸¼ºµÈ ¸ðµ¨ÀÌ´Ù. ¸ÕÀú º» ¿¬±¸¿¡¼´Â ¼¿ï½Ã Áß±¸ ¼ÒÀç ¾÷¹«¿ë °Ç¹°ÀÇ Åͺ¸ ³Ãµ¿±âÀÇ COP¸¦ ¿¹ÃøÇÏ´Â Àΰø ½Å°æ¸Á ¸ðµ¨À» °³¹ßÇÏ¿´´Ù. ¶ÇÇÑ, Àΰø ½Å°æ¸Á ¸ðµ¨À» Àû¿ëÇÒ ¶§, À¯È¿ÇÑ ÀԷº¯¼ö ¹üÀ§¸¦ ƯÁ¤ ½Å·Ú±¸°£ ¹üÀ§ ³»(¿¹: 95%)¿¡¼ Á¦½ÃÇÏ´Â °ÍÀÌ Áß¿äÇÏ´Ù. À̸¦ À§ÇØ, °¡¿ì½Ã¾È È¥ÇÕ ¸ðµ¨(Gaussian Mixture Model) ¹æ¹ýÀ» Àû¿ëÇÏ¿´´Ù. ³Ãµ¿±â COP¿¹Ãø Àΰø½Å°æ¸Á ¸ðµ¨À» °³¹ßÇϰí, ±× Á¤È®ÇÔÀ» ¾Ë ¼ö ÀÖ¾úÀ¸¸ç, °¡¿ì½Ã¾È È¥ÇÕ ¸ðµ¨À» Àû¿ëÇÏ¿©, ±â°èÇнÀ ¸ðµ¨ÀÇ À¯È¿¼º ¹üÀ§¸¦ Á¤·®ÀûÀ¸·Î È®ÀÎÇÏ¿´´Ù. |
¿ä¾à2 |
In this study, the authors developed an artificial neural network (ANN) model to predict the COP of a turbo chiller. The BEMS data gathered in a real office building was used for the training of the ANN model. However, it is also important to have a validity check for the use of the ANN model. For this purpose, Gaussian Mixture Model (GMM) was introduced. It has been shown in the paper that (1) the ANN model is good enough (MBE: ?0.29%, CVRMSE: 6.28%) and (2) the validity check is necessary when using the ANN model. |