{"title":"基于物理性质的膨胀土渗透性人工神经网络预测","authors":"","doi":"10.7454/mst.v27i2.1623","DOIUrl":null,"url":null,"abstract":"Permeability is a soil parameter related to the construction industry to understand the processes of infiltration, runoff, and settlement. The risk of testing errors is inevitable in permeability investigations, especially in expansive soils. Trial and error in permeability testing becomes difficult due to soils with small pore sizes and large shrinkage expansion. Several studies related to soil physical properties that affect permeability have been conducted. However, the correlation results obtained still have poor accuracy. Artificial neural networks (ANN) are machine learning systems that can change their structure to solve problems that are included in the system. The use of ANNs in data learning is applied to help the established model predict future output values with a small error value. This research aims to study the correlation between the physical properties of expansive soil that affect its permeability using ANN correlation and then produce correlation equations for future inputs. The research was conducted with input data in the form of soil liquid limit, soil plasticity index (IP), %fine grains, and soil permeability as output data. Results demonstrated a good correlation between soil physical properties and permeability, revealing high accuracy in the output regression equation.","PeriodicalId":42980,"journal":{"name":"Makara Journal of Technology","volume":"86 1","pages":"0"},"PeriodicalIF":0.2000,"publicationDate":"2023-08-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Permeability Prediction for Expansive Soil Based on Physical Properties Using Artificial Neural Networks\",\"authors\":\"\",\"doi\":\"10.7454/mst.v27i2.1623\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Permeability is a soil parameter related to the construction industry to understand the processes of infiltration, runoff, and settlement. The risk of testing errors is inevitable in permeability investigations, especially in expansive soils. Trial and error in permeability testing becomes difficult due to soils with small pore sizes and large shrinkage expansion. Several studies related to soil physical properties that affect permeability have been conducted. However, the correlation results obtained still have poor accuracy. Artificial neural networks (ANN) are machine learning systems that can change their structure to solve problems that are included in the system. The use of ANNs in data learning is applied to help the established model predict future output values with a small error value. This research aims to study the correlation between the physical properties of expansive soil that affect its permeability using ANN correlation and then produce correlation equations for future inputs. The research was conducted with input data in the form of soil liquid limit, soil plasticity index (IP), %fine grains, and soil permeability as output data. Results demonstrated a good correlation between soil physical properties and permeability, revealing high accuracy in the output regression equation.\",\"PeriodicalId\":42980,\"journal\":{\"name\":\"Makara Journal of Technology\",\"volume\":\"86 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.2000,\"publicationDate\":\"2023-08-30\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Makara Journal of Technology\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.7454/mst.v27i2.1623\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"ENGINEERING, MULTIDISCIPLINARY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Makara Journal of Technology","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.7454/mst.v27i2.1623","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"ENGINEERING, MULTIDISCIPLINARY","Score":null,"Total":0}
Permeability Prediction for Expansive Soil Based on Physical Properties Using Artificial Neural Networks
Permeability is a soil parameter related to the construction industry to understand the processes of infiltration, runoff, and settlement. The risk of testing errors is inevitable in permeability investigations, especially in expansive soils. Trial and error in permeability testing becomes difficult due to soils with small pore sizes and large shrinkage expansion. Several studies related to soil physical properties that affect permeability have been conducted. However, the correlation results obtained still have poor accuracy. Artificial neural networks (ANN) are machine learning systems that can change their structure to solve problems that are included in the system. The use of ANNs in data learning is applied to help the established model predict future output values with a small error value. This research aims to study the correlation between the physical properties of expansive soil that affect its permeability using ANN correlation and then produce correlation equations for future inputs. The research was conducted with input data in the form of soil liquid limit, soil plasticity index (IP), %fine grains, and soil permeability as output data. Results demonstrated a good correlation between soil physical properties and permeability, revealing high accuracy in the output regression equation.