Determination of Limiting Resistance for Dilatant Soil to Shearing While Using Artificial Neural Networks

Modern geotechnical pile manufacturing technologies allow to solve many engineering problems in construction with significant economic effect and reduction of work schedule. However a theoretical justification for these technologies is significantly lagging behind from their practical application. A constrained dilatancy factor is considered as the main reserve of pile bearing capacity in loose soils. Understanding of such approach opens a way to improvement of geotechnical pile manufacturing technologies that provide an active impact on base soil and its ultimate stress state which is determined by the ratio of soil resistance to shearing and normal stresses, or main stresses on a shear site at the moment of failure. Determination of total normal stresses in the shearing plane according to density of loose soil, its granulometric composition and constrained dilatancy conditions makes it possible to determine an ultimate resistance of the soil to shear, and, consequently, its strength. The proposed elastic-plastic model of dilatant soil in shear state being realized according to the adopted technology which was developed while using a special dilatometric shearing device has made it possible to obtain additional data on dilatant normal stresses and strength parameters of the soil depending on its density, granulometric composition and constrained dilatancy conditions. The use of artificial neural networks in the mathematical processing of experimental data has permitted to develop an analytical method for determining an ultimate resistance of loose soil to shearing process under constrained dilatancy conditions and carry out calibration of the calculated parameters of the adopted soil model. It has been proved that soil strength is a function of fracture conditions, which are determined by friction and dilatancy. In this case a conventional Coulomb-Mora strength theory for loose soils is valid both for free dilatation conditions and constrained shear but with due account of additional normal dilatant stresses.

1. Soils. Laboratory Methods for Determining the Strength and Strain Characteristics: GOST [All-Union State Standard] 12248–2010. Moscow, Standartinform, 2012. 78 (in Russian).

2. Sobolevsky D. Yu., Popov O. V. (1989) Device for Determination the Contact of Resistance to Bulk Material Displacement. Inventor’s Certificate No 1491143 USSR (in Russian).

3. Sobolevsky D. Yu. (1994) Strength and Bearing Capacity of Dilatant Soil. Minsk, Navuka i Tekhnika Publ. 232 (in Russian).

4. Sobolevsky D. Yu., Popov O. V. (1988) On Contact Settlement and Lateral Friction Along Pile Shafts in Sand Soil. Mechanized Non-Waste Technology for Creation of Pile Foundation. Vladivostok, 97–99 (in Russian).

5. Gersevanov N. M., Polshin D. E. (1948) Fundamentals of Soil Mechanics and its Practical Application. Moscow, Publishing House Stroyizdat. 248 (in Russian).

6. Wasserman F. (1989) Neural Computing: Theory and Practice. Coriolis Group. 230.

7. Training Paradigms for Neural Networks. Available at: http://apsheronsk.bozo.ru/Neural/Lec3.htm. (?ccessed 10 May 2017) (in Russian).

8. Popova Yu. B., Yatsynovich S. V. (2017) Training for Artificial Neural Networks While Using Back-Prop Method. Available at: http://www.bntu.by/news/67-conference-mido/4860-2016-11-18-15-47-40.html. (Accessed 1 August 2017) (in Russian).

9. Popova Yu. B., Yatsynovich S. V. (2017) Software Implementation of the Artificial Neural Network for Virtual Objects ?ontrol. Sistemny Analiz i Prikladnaya Informatika = System Analysis and Applied Informatics, (4), 72–78 (in Russian).

10. Popova Yu. B., Yatsynovich S. V. (2017) Simulation and Programming Realization of Artificial Neural Network. Matematicheskiye Metody v Tekhnike i Tekhnologiyakh. Sb. tr. Mezhdunar. Nauch. Konf. T. 12, Ch. 1 [Mathematical Methods for Equipment and Technologies. Collected Papers of International Scientific Conference. Vol. 12. Part. 1]. Saint-Petersburg, Publishing House of Polytechnical University, 168–171 (in Russian).