ﺗﺨﻤﯿﻦ ﺗﺮاواﯾﯽ ﺑﺎ ﺑﮑﺎرﮔﯿﺮي ﻧﮕﺎرهﻫﺎي ﭘﺘﺮوﻓﯿﺰﯾﮑﯽ و روشﻫﺎي ﻫﻮش ﻣﺼﻨﻮﻋﯽ: ﻣﻄﺎﻟﻌﻪ ﻣﻮردي در ﻣﺨﺰن آﺳﻤﺎري ﯾﮑﯽ از ﻣﯿﺎدﯾﻦ ﻧﻔﺘﯽ ﺟﻨﻮب ﻏﺮﺑﯽ اﯾﺮان

Permeability estimation using petrophysical logs and Aartificial iIntelligence methods: A case study in Asmari reservoir of one of the oil fields of southwestern Iran

در اﯾﻦ ﭘﮋوﻫﺶ، اﺑﺘﺪا ﺗﺮاواﯾﯽ ﻧﮕﺎره ﺗﺸﺪﯾﺪ ﻣﻐﻨﺎﻃﯿﺴﯽ ﻫﺴﺘﻪ¬اي ﺑﺎ اﺳﺘﻔﺎده از دو روش ﻣﺮﺳﻮم ﻣﺪل ﺳﯿﺎل آزاد) )Coatesو ﻣﺪل ﺷﻠﻤﺒﺮژه ﯾﺎ ﻣﯿﺎﻧﮕﯿﻦ SDR) T2(1 ﻣﺤﺎﺳﺒﻪ ﺷﺪ. ﺳﭙﺲ ﯾﮏ ﻣﺪل ﺳﺎده ﺷﺒﮑﻪ ﻋﺼﺒﯽ ﻣﺼﻨﻮﻋﯽ ﺑﺎ ﻓﺮاﯾﻨﺪ آﻣﻮزش از ﻧﻮع اﻟﮕﻮرﯾﺘﻢ ﭘﺲ اﻧﺘﺸﺎر ﺧﻄﺎ، ﻃﺮاﺣﯽ ﮔﺮدﯾﺪ، در اداﻣﻪ ﺑﺎ اﺳﺘﻔﺎده از اﻟﮕﻮرﯾﺘﻢ ﺑﻬﯿﻨﻪ ﺳﺎزي رﻗﺎﺑﺖ اﺳﺘﻌﻤﺎري )ANN-ICA( و اﻟﮕﻮرﯾﺘﻢ ازدﺣﺎم ذرات )ANN-PSO( اﯾﻦ ﻣﺪل ﺑﻬﯿﻨﻪ ﺷﺪ و از آن ﺑﺮاي ﺗﺨﻤﯿﻦ ﭘﺎراﻣﺘﺮ ﺗﺮاواﯾﯽ اﺳﺘﻔﺎده ﺷﺪ. در ﻧﻬﺎﯾﺖ ، ﻧﺘﺎﯾﺞ ﺑﺎ ﻣﻘﺎﯾﺴﻪ ﻧﻔﻮذﭘﺬﯾﺮي ﺗﺨﻤﯿﻦ زده ﺷﺪه ﺑﺎ ﻣﻘﺪار واﻗﻌﯽ ﻣﻮرد ﺗﺠﺰﯾﻪ و ﺗﺤﻠﯿﻞ ﻗﺮار ﮔﺮﻓﺖ و دﻗﺖ ﺑﺮآورد از ﻧﻈﺮدو ﭘﺎراﻣﺘﺮ ﺧﻄﺎي ﻣﯿﺎﻧﮕﯿﻦ ﻣﺮﺑﻊ و ﺿﺮﯾﺐ ﻫﻤﺒﺴﺘﮕﯽ ﻣﻘﺎﯾﺴﻪ ﺷﺪ. ﻧﺘﺎﯾﺞ ، ﺑﯿﺎﻧﮕﺮ دﻗﺖ ﺑﺎﻻي ﻣﻘﺎدﯾﺮ ﺗﺮاواﯾﯽ ﺗﺨﻤﯿﻦ زده ﺷﺪه ﺑﺎ اﺳﺘﻔﺎده از ﺗﺮﮐﯿﺐ ﺷﺒﮑﻪ ﺳﺎده ﻋﺼﺒﯽ ﺑﺎ اﻟﮕﻮرﯾﺘﻢﻫﺎي ﺑﻬﯿﻨﻪﺳﺎزي اﺳﺖ. ﻧﺘﺎﯾﺞ ﺣﺎﺻﻞ از ﺗﺮﮐﯿﺐ اﻟﮕﻮرﯾﺘﻢﻫﺎي ﺑﻬﯿﻨﻪﺳﺎزي در اﯾﻦ ﻣﻄﺎﻟﻌﻪ ﻣﯽﺗﻮاﻧﺪ ﺑﻪ ﻋﻨﻮان روﺷﯽ ﻗﺪرﺗﻤﻨﺪ و ﻣﻔﯿﺪ در ﺟﻬﺖ ﺑﺪﺳﺖ آوردن ﺳﺎﯾﺮ ﭘﺎراﻣﺘﺮﻫﺎ، از ﺟﻤﻠﻪ ﭘﺎراﻣﺘﺮﻫﺎي ﻣﺨﺰﻧﯽ، ﭘﺘﺮوﻓﯿﺰﯾﮑﯽ و ژﺋﻮﻣﮑﺎﻧﯿﮑﯽ اﺳﺘﻔﺎده ﺷﻮد.

In this study, first, the permeability of the magnetic resonance imaging of the nucleus was calculated using two conventional methods, the free fluid model (Coates) and the Schlumberger model or the mean T2 (SDR). Then, a simple model of artificial neural network was designed with the training process of the backpropagation algorithm, then using the Imperialist competition optimization algorithm (ANN-ICA) and particle swarm algorithm (ANN-PSO) this model was optimized and It was used to estimate the permeability parameter. Finally, the results were analyzed by comparing the estimated permeability with the actual value and the estimation accuracy was compared in terms of two parameters of mean-square error and correlation coefficient. The results indicate the high accuracy of the permeability values estimated using a combination of simple neural network with optimization algorithms. The results of combining optimization algorithms in this study can be used as a powerful and useful method to obtain other parameters, including reservoir, petrophysical and geomechanical parameters.