Research in the area of geotechnical construction has been conducted by a scientist from South Ural State University (SUSU) and his foreign colleagues. They proposed a hybrid intelligence system that can help Civil Engineers to get the most accurate pile bearing capacity data. The results of the work have been published in the highly rated journal “Artificial Intelligence Review” (Q1).
A research group, that is led by SUSU scientist Danial Jahed Armaghani, has presented a renewed way to estimate pile bearing capacity. This property reflects the maximum load that a pile can safely carry. Bearing capacity data is necessary for builders in designing deep foundations during construction. Nowadays, there are four ways to estimate this parameter; three of them need actual tests on the site that are expensive, difficult, and time-consuming. The other one is theoretical calculation and needs adjustment based on the actual data on the characteristics of the soil.
Scientists have suggested a new technique of an advance hybrid intelligence system based on the adaptive neuro-fuzzy inference system (ANFIS)-group method of data handling (GMDH) optimized by the imperialist competitive algorithm (ICA), ANFIS-GMDH-ICA for forecasting pile bearing capacity.
“In this advanced system, the imperialist competitive algorithm role is to optimize the membership functions obtained by ANFIS-GMDH technique for receiving a higher accuracy level and lower error. It receives more accurate predicted values of pile bearing capacity compared to those obtained by ANFIS and GMDH predictive models,” says Danial Jahed Armaghani, PhD, the senior researcher of the Town Planning, Engineering Systems and Networks department (Institute of Architecture and Construction, SUSU).
The research group has checked the effectiveness of this technique by analysing 257 high strain dynamic load tests. The data of this work has been gathered and prepared in different construction sites of Indonesia. Scientists have used five independent parameters including ram weight, drop height, length and diameter of pile, and pile set as predictors or model inputs. The results predicted by the model are very close to the measured values of pile bearing capacity which can solve the problem related to measuring the actual parameters in the construction sites.
The developed model can be applied to relevant projects with similar conditions, and civil engineers instead of conducting the actual tests in the site can apply the proposed model and get very close results with minimum error.