• Smart Structures and Systems
• /
• 제29권3호
• /
• pp.513-522
• /
• 2022

Due to the importance of accurate analysis of bearing capacity in civil engineering projects, this paper studies the efficiency of two novel metaheuristic-based models for this objective. To this end, black hole algorithm (BHA) and multi-verse optimizer (MVO) are synthesized with an artificial neural network (ANN) to build the proposed hybrid models. Based on the settlement of a two-layered soil (and a shallow footing) system, the stability values (SV) of 0 and 1 (indicating the stability and failure, respectively) are set as the targets. Each model predicted the SV for 901 stages. The results indicated that the BHA and MVO can increase the accuracy (i.e., the area under the receiving operating characteristic curve) of the ANN from 94.0% to 96.3 and 97.2% in analyzing the SV pattern. Moreover, the prediction accuracy rose from 93.1% to 94.4 and 95.0%. Also, a comparison between the ANN's error decreased by the BHA and MVO (7.92% vs. 18.08% in the training phase and 6.28% vs. 13.62% in the testing phase) showed that the MVO is a more efficient optimizer. Hence, the suggested MVO-ANN can be used as a reliable approach for the practical estimation of bearing capacity.

• 한국물환경학회지
• /
• 제29권1호
• /
• pp.19-28
• /
• 2013

Artificial neural networks were used for time series modelling of algal dynamics of whole year and by season at the Paldang dam station (confluence area). The modelling was based on comprehensive weekly water quality data from 1997 to 2004 at the Paldang dam station. The results of validation of seasonal models showed that the timing and magnitude of the observed chlorophyll a concentration was predicted better, compared with the ANN model for whole year. Internal weightings of the inputs in trained neural networks were obtained by sensitivity analysis for identification of the primary driving mechanisms in the system dynamics. pH, COD, TP determined most the dynamics of chlorophyll a, although these inputs were not the real driving variable for algal growth. Short-term prediction models that perform one or two weeks ahead predictions of chlorophyll a concentration were designed for the application of Harmful Algal Alert System in Lake Paldang. Short-term-ahead ANN models showed the possibilities of application of Harmful Algal Alert System after increasing ANN model's performance.

• 한국식품과학회지
• /
• 제28권5호
• /
• pp.916-921
• /
• 1996

감자칩의 생산 공정에서 속슬렛 추출법에 의한 지방 함량의 측정은 측정 시간이 길어서 공정 제어에 이용하기에 적절하지 못하기 때문에, 근적외선 분광 분석법에 의한 감자칩의 지방 함량의 신속한 측정에 대한 연구를 수행하였다. 분쇄된 시료와 분쇄되지 않은 시료에 대한 스펙트럼과 지방 함량과의 상관 관계를 각각 분석하였고, 분석된 상관 관계를 토대로 하여 예측 모형을 개발하였으며, 개발된 예측 모형에 대해 검증하였다. 예측 모형의 개발에는 회귀 분석과 신경회로망 기법을 이용하였다. 분쇄된 시료에 대한 지방 함량의 예측 모형은 회귀모형과 신경 회로망 기법에 의해서, 결정 계수가 각각 093, 0.92, SEP는 각각 1.29%, 1.17%였다. 이와 같은 결과에 의해서, 근적외 분광 분석법에 의한 분쇄된 감자칩의 지방 함량의 예측 모형 개발이 가능하다고 판단되었다. 또한 SEP를 비교할 때, 회귀 모형보다는 신경 회로망 기법에 의한 예측 모형이 더 우수한 것으로 나타났다. 분쇄되지 않은 시료에 대한 흡광도와 지방 함량과의 상관 관계 분석 결과, 근적외 분광 분석법에 의한 지방 함량의 예측 모형의 개발은 적절하지 못하다고 판단되었다.

• 한국환경복원기술학회지
• /
• 제11권3호
• /
• pp.28-38
• /
• 2008

The purpose of this study are to compare existing presence-absence predictive models and to predict suitable habitat for Goral (Nemorhaedus caudatus raddeanus) that is an endangered and protected species in Seoraksan national park using the best model among existing predictive models. The methods of this study are as follows. First, 375 location data and 9 environmental data layers were implemented to build a model. Secondly, 4 existing presence-absence models : Generalized Linear Model (GLM), Generalized Addictive Model (GAM), Classification and Regression Tree (CART), and Artificial Neural Network (ANN) were tested to predict the Goal habitat. Thirdly, ROC (Receiver Operating Characteristic) and Kappa statistics were used to calculate a model performance. Lastly, we verified models and created habitat suitability maps. The ROC AUC (Area Under the Curve) and Kappa values were 0.697/0.266 (GLM), 0.729/0.313 (GAM), 0.776/0.453 (CART), and 0.858/0.559 (ANN). Therefore, ANN was selected as the best model among 4 models. The models showed that elevation, slope, and distance to stream were the significant factors for Goal habitat. The ratio of predicted area of ANN using a threshold was 31.29%, but the area decreased when human effect was considered. We need to investigate the difference of various models to build a suitable wildlife habitat model under a given condition.