Artificial Neural Network Applications for Geotechnical Engineering

Dear Colleagues,

Over the past few decades algorithms using biologically inspired signal transformation—artificial neural networks (ANNs)—has undergone a remarkable and exciting development path. From simple algorithms for classification and approximation to deep learning-based procedures whose operation is almost indistinguishable from the assistance of an intelligent being.

At all stages of this journey, ANNs have found and continue to find their applications in geotechnics and soil mechanics more often than in other disciplines of civil engineering and mechanics. The soil description usually contains a larger number of parameters. Many of these parameters are correlated. These correlations are phenomenological and are established mostly through experimental research in the laboratory and in situ. In soil description, the use of ANNs is natural because they are universal approximators of functions of many variables, built on the basis of a large set of experimental data. This is where the attractiveness of ANN comes from in applications such as site characterization, classification of soils, and modeling various aspects of soil behavior, including soil compaction, soil swelling, coupled hydro-thermo-mechanical processes in soils, liquefaction, and many others. The same properties of ANNs allow this technique to be used to predict the load-bearing capacity of piles, columns, and the effectiveness of soil strengthening methods. The source of information here is usually an extensive database on the behavior of these structures, collected both by geotechnical companies and laboratories, or as a result of theoretical research and numerical simulations. Particularly important is the ease of solving inverse problems by neural networks. In geotechnics, back-calculations are often performed to elaborate the results of experiments, such as with the Falling Weight Deflectometer used to assess the mechanical properties of layered subsoils. The great advantage of ANN applications is the possibility to increase the reliability of the existing soil description or improve the accuracy of predictions as new data are collected and even as the data concerning geometric and mechanical parameters arrives in real time while an engineering structure is being constructed. In this way, the ANNs support the empirical and observational methods of geotechnical design.

In applications to build numerical models in geotechnical engineering, as in other fields, artificial ANNs are used as surrogates of calculations performed by FEM programs or as an element of the constitutive description of the soil within the FEM procedure. Modern concepts of physically informed neural networks are also used in geotechnical models. Almost all types of ANN are proposed. In most applications, these are classical neural networks with hidden layers trained by the backward error propagation method; however, the use of convolutional networks or Long Short-Term Memory (LSTM) nets has appeared recently as promising practical applications (mostly to interpret data coming from the CPTU cone or Flat Dilatometer sensors).

We would like this collection of articles to include papers that document the use of ANNs in issues related to soil mechanics, but also to the mechanics of the soil-structure system and foundation techniques. The following specific topics, mentioned in the foreword above, are particularly welcome:

• ANN as a tool of elaboration of in situ or laboratory tests in soil mechanics;
• ANN for site characterization;
• Classification of soils with ANN;
• ANN as a numerical model of soil compaction, soil swelling, and coupled hydro-thermo-mechanical processes in soils;
• ANN for prediction of liquefaction risk;
• ANN for prediction of risk of landslides;
• ANN-aided prediction of the load-bearing capacity of piles, columns, and other engineering structures;
• ANN for assessment of the effectiveness of soil strengthening methods;
• ANN as a surrogate procedure replacing some elements of FE computations in numerical modeling of soil-structure interactions;
• ANN description of constitutive law in FE model of soils;
• Physically informed ANNs in numerical modeling in geotechnics and soil mechanics;
• ANN used for back calculations in solving inverse problem in geotechnics and soil mechanics;
• Deep learning ANNs (convolutional, LSTM) in applications in geotechnics and soil mechanics;
• Expert programs based on ANNs and data mining in geotechnics.

Prof. Dr. Marek Lefik
Guest Editor

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Applied Sciences is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2400 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• artificial neural networks
• approximation of experimental data with ANN
• ANN as a surrogate for finite elements
• correlations in soil mechanics with ANN
• deep learning
• ANN–aided prediction of structural properties