岩石工程相互作用矩阵的神经网络编码方法

岩石工程系统RES(Rock Engineering System)中现行的相互作用矩阵的编码方法均是静态的,无法反映岩石工程现场各参数间的动态的非线性关系.本文基于人工神经网络的应用提出了一种新的建立岩石工程相互作用矩阵的相对作用强度RSE(RelativeStrength of Effect)方法,并进一步探讨了将RSE应用于建立RES的相互作用矩阵的可行性与现实性.结果表明,基于人工神经网络的RSE与RES有很多共同点,将RSE应用于RES能够建立起动态的非线性的相互作用矩阵,使得RES相互作用矩阵更为切合实际,其应用也更为方便和可靠.     

From Neural Networks to Qualitative Models in Environmental Engineering

As an alternative to physical models, artificial neural networks (ANNs) are a valuable forecast tool in environmental sciences. They can be used effectively due to their learning capabilities and their low computational costs. Once all relevant variables of the system are identified and put into the network, it works quickly and accurately. However, one of the major shortcomings of neural networks is that they do not reveal causal relationships between major system components and thus are unable to improve the explicit knowledge of the user. Another problem is due to the fact that reasoning is only done from the inputs to the outputs. In cases where the opposite is requested (i.e., deriving inputs leading to a given output), neural networks can hardly be used. To overcome these problems, we introduce a novel approach for deriving qualitative information out of neural networks. Some of the resulting rules can directly be used by a qualitative simulator for producing possible future scenarios. Because of the explicit representation of knowledge, the rules should be easier to understand and can be used as a starting point for creating models wherever a physical model is not available. Moreover, the resulting rules are well adapted to be used in decision support systems. We illustrate our approach by introducing a network for predicting surface ozone concentrations and show how rules can be derived from the network and how the approach can be naturally extended for use in decision support systems.     

Neural Networks in Structural Engineering

In the past few years literature on computational civil engineering has concentrated primarily on artificial intelligence (Al) applications involving expert system technology. This article discusses a different Al approach involving neural networks. Unlike their expert system counterparts, neural networks can be trained based on observed information. These systems exhibit a learning and memory capability similar to that of the human brain, a fact due to their simplified modeling of the brain's biological function. This article presents an introduction to neural network technology as it applies to structural engineering applications. Differing network types are discussed. A back-propagation learning algorithm is presented. The article concludes with a demonstration of the potential of the neural network approach. The demonstration involves three structural engineering problems. The first problem involves pattern recognition; the second, a simple concrete beam design; and the third, a rectangular plate analysis. The pattern recognition problem demonstrates a solution which would otherwise be difficult to code in a conventional program. The concrete beam problem indicates that typical design decisions can be made by neural networks. The last problem demonstrates that numerically complex solutions can be estimated almost instantaneously with a neural network.     

基于粒子群算法和广义回归神经网络的岩爆预测

 岩爆是岩石深部开挖中一种常见的工程地质灾害。为评价岩爆发生的可能性,提出一种基于粒子群算法和广义回归神经网络模型(PSO-GRNN模型)的岩爆预测方法。该方法利用已有岩爆数据,通过神经网络技术建立回归模型,采用粒子群算法对模型参数进行优化,减少人为因素对神经网络设计的影响。据此方法,在能量理论的基础上,选取洞壁围岩最大切向应力、岩石单轴抗压强度、抗拉强度和弹性能量指数作为主要影响因素,利用国内外26组已有工程数据建立岩爆预测的PSO-GRNN模型。通过对苍岭隧道和冬瓜山铜矿岩爆预测的工程实例分析验证该方法的可行性和适用性。所提方法可为类似工程的岩爆预测提供参考。     

Nonlinear Identification of Dynamic Systems Using Neural Networks

A neural-network-based method is proposed for the modeling and identification of a discrete-time nonlinear hysteretic system during strong earthquake motion. The learning or modeling capability of multilayer neural networks is explained from the mathematical point of view. The main idea of the proposed neural approach is explained, and it is shown that a multilayer neural network is a general type of NARMAX model and is suitable for the extreme nonlinear input-output mapping problems. Numerical simulation of a three-story building and a real structure (a bridge in Taiwan) subjected to several recorded earthquakes are used here to demonstrate the proposed method. The results illustrate that the neural network approach is a reliable and feasible method.     

岩石工程稳定性控制参数的直觉分析

基于BP神经网络，定义了衡量网络输入对输出作用大小的相对作用强度RSE（RelativeStrengthofEffect），并结合实际的岩石工程实例数据用RSE分析了其各个作用参数对工程稳定性的影响的相对大小与作用方式。实际分析的结果表明，所提出的方法能够较全面地反映岩石工程现场的复杂实际情况，具有易于处理不确定性、动态与非线性问题等优点，是对岩石工程进行参数分析的有效工具。     

StructNet: A Neural Network for Structural System Selection

Abstract: This paper describes StructNet, a computer application developed to select the most effective structural member materials given a building project's attributes. The system analyzes 15 parameters of a building project (e.g., available site space, budget, height) and determines the most appropriate structural system for the beam, column, and slab structural members. This paper first describes the process for selecting a structural system for a building. It was very important to understand this process before determining the best type and structure for the computer application. Then a comparison between a neural network approach and a rule-based expert-system approach for this application is presented. A discussion of the reasons for selecting a neural network approach is given. The StructNet application is described in detail, including the testing of the network. Along with the testing of the network is a discussion of how varying the learning rate and error limit affect the performance of the neural network application. The testing of the network shows that the program can reasonably select the same structural system types as the expert used to collect the training project data. Since the system will be used only as a preliminary tool to limit the number of possible structural systems for a project, the accuracy of the system is acceptable. However, additional experimentation needs to be conducted to determine the accuracy and practical use of this application. The final sections of the paper discuss the lack of adequate testing procedures for neural networks used in applications for unstructured or ill-defined decision making. The use of these types of networks and their relevance to the civil engineering computer field are also discussed.     

Machine Learning in Engineering Analysis and Design: An Integrated Fuzzy Neural Network Learning Model

Applying neural network computing to structural engineering problems has received increasing interest, with particular emphasis placed on a supervised neural network with the backpropagation (BP) learning algorithm. In this article, we present an integrated fuzzy neural network (IFN) learning model by integrating a newly developed unsupervised fuzzy neural network (UFN) reasoning model with a supervised learning model in structural engineering. The UFN reasoning model is developed on the basis of a single-layer laterally connected neural network with an unsupervised competing algorithm. The IFN learning model is compared with the BP learning algorithm as well as with a counterpropagation learning algorithm (CPN) using two engineering analysis and design examples from the recent literature. This comparison indicates not only a superior learning performance in solved instances but also a substantial decrease in computational time for the IFN learning model. In addition, the IFN learning model is applied to a complicated engineering design problem involving steel structures. The IFN learning model also demonstrates superior learning performance in a complicated structural design problem with a reasonable computational time.     

Neural Networks as Inference and Learning Engines

Abstract: This paper presents an abridgment of a neural network constructive methodology and applications with real data. The neural network can be considered as the learning core and inference engine of an expert system that produces either different network designs or simulations as output, its input being data sequences. Basically, it consists of additive structural learning, limiting it by a cross-validation technique.
Considerations about uncertainty treatment in neural networks are also presented, including uncertainty in data, in neuron activation, in outputs, and combination of several uncertainty sources.
Applications include three different sets of data, all of them related to the energy field. First, river streamflow estimation is discussed. Then CO₂ concentration prediction from gas injection rate is studied. Finally, the program learns to imitate a feedwater control system in a nuclear reactor. All tests show good results, as can be seen when compared with other standard methods.     

岩石力学与工程专家系统研究新进展

本文内容包括：岩石力学与工程知识表示、不确定性推理、知识获取与学习、神经网络专家系统、综合集成系统等。本文基本上反映了岩石力学与工程专家系统领域最新的研究成果及其发展趋势。     

A fracture sliding potential index for wellbore stability analysis

Sliding failure along the fractures intersecting a wellbore is one of the major wellbore instability mechanisms. This kind of failure is similar to the slope instabilities, a well-known phenomenon in mining and civil engineering. During drilling operations the drilling fluid can penetrate through fractures and lead to fracture reactivation and wellbore instability. The rock engineering systems (RES), initially introduced in the mining- and civil-related geomechanics problems, is an approach to analyze the interrelationship between the parameters affecting rock engineering activities. In this study, after discussing the sliding mechanism along a fracture in a wellbore during drilling, and identifying all the effective parameters, an interaction matrix is introduced to study the sliding failure mechanism. Thereafter, the interaction intensity and dominance of each parameter in the system is determined to classify these parameters. A systematic approach was used to determine the relative interactive intensity and value of each contributing parameter in the fracture sliding mechanism. As a result, an index is presented to estimate the fracture sliding potential. The results indicate the ability of this method to analyse wellbore instability due to fracture reactivation mechanism. This will assist in finding a better engineering action to mitigate or eliminate potential fracture sliding during drilling. The results show a good agreement with those obtained using Mohr–Coulomb failure analysis and field observations.     

Deep convolutional neural networks for uncertainty propagation in random fields

The development of a reliable and robust surrogate model is often constrained by the dimensionality of the problem. For a system with high‐dimensional inputs/outputs (I/O), conventional approaches usually use a low‐dimensional manifold to describe the high‐dimensional system, where the I/O data are first reduced to more manageable dimensions and then the condensed representation is used for surrogate modeling. In this study, a new solution scheme for this type of problem based on a deep learning approach is presented. The proposed surrogate is based on a particular network architecture, that is, convolutional neural networks. The surrogate architecture is designed in a hierarchical style containing three different levels of model structures, advancing the efficiency and effectiveness of the model in the aspect of training. To assess the model performance, uncertainty quantification is carried out in a continuum mechanics benchmark problem. Numerical results suggest the proposed model is capable of directly inferring a wide variety of I/O mapping relationships. Uncertainty analysis results obtained via the proposed surrogate have successfully characterized the statistical properties of the output fields compared to the Monte Carlo estimates.     

New artificial neural networks for true triaxial stress state analysis and demonstration of intermediate principal stress effects on intact rock strength

Simulations are conducted using five new artificial neural networks developed herein to demonstrate and investigate the behavior of rock material under polyaxial loading. The effects of the intermediate principal stress on the intact rock strength are investigated and compared with laboratory results from the literature. To normalize differences in laboratory testing conditions, the stress state is used as the objective parameter in the artificial neural network model predictions. The variations of major principal stress of rock material with intermediate principal stress, minor principal stress and stress state are investigated. The artificial neural network simulations show that for the rock types examined, none were independent of intermediate principal stress effects. In addition, the results of the artificial neural network models, in general agreement with observations made by others, show （a） a general trend of strength increasing and reaching a peak at some intermediate stress state factor, followed by a decline in strength for most rock types; （b） a post-peak strength behavior dependent on the minor principal stress, with respect to rock type; （c） sensitivity to the stress state, and to the interaction between the stress state and uniaxial compressive strength of the test data by the artificial neural networks models （two-way analysis of variance; 95% confidence interval）. Artificial neural network modeling, a self-learning approach to polyaxial stress simulation, can thus complement the commonly observed difficult task of conducting true triaxial laboratory tests, and/or other methods that attempt to improve two-dimensional （2D） failure criteria by incorporating intermediate principal stress effects.     

Retrospective case example using a comprehensive suitability index (CSI) for siting the Shisan-Ling power station, China

The purpose of this paper is to present a retrospective case example of using the rock engineering systems (RES) methodology to site a pumped storage power station in China. With such a siting problem, there are many interacting factors governing both the site and the specific underground position of the powerhouse. The RES approach, based on an interaction matrix for semi-quantitative characterization of the factors and their interactions, is used to develop a comprehensive suitability index (CSI). In this way, the factors governing the geological and rock mechanics related factors are structured and evaluated. Additionally, the complexity of the decision making process is condensed to the CSI values for different potential sites and underground locations, a higher CSI value indicating a more suitable site.The retrospective analysis uses information from the investigations made during the actual site investigation and design work for the Shisan-Ling Power Station, China — which has been constructed. For the first stage, site selection for the overall engineering arrangement, 11 parameters concerned with geology, geomorphology, engineering layout, environment, cost and construction are taken into account. After comparison of site options in conglomerate, andesite and limestone formations, the conglomerate formation proved to be most favorable — because it has the highest CSI value. For the second stage, specifically locating the underground powerhouse, another seven parameters associated with faults, joints, groundwater, etc. are evaluated. As a result, position II proved to be the most favorable location for the powerhouse. Since these were the same conclusions reached during the actual investigations, this retrospective application of the CSI demonstrates the value of the RES methodology and associated indices for assisting in rock engineering design.     

Prediction of roadheaders' performance using artificial neural network approaches (MLP and KOSFM)

Application of mechanical excavators is one of the most commonly used excavation methods because it can bring the project more productivity, accuracy and safety. Among the mechanical excavators, roadheaders are mechanical miners which have been extensively used in tunneling, mining and civil industries. Performance prediction is an important issue for successful roadheader application and generally deals with machine selection, production rate and bit consumption. The main aim of this research is to investigate the cutting performance(instantaneous cutting rates(ICRs)) of medium-duty roadheaders by using artificial neural network(ANN) approach. There are different categories for ANNs, but based on training algorithm there are two main kinds: supervised and unsupervised. The multi-layer perceptron(MLP) and Kohonen self-organizing feature map(KSOFM) are the most widely used neural networks for supervised and unsupervised ones, respectively. For gaining this goal, a database was primarily provided from roadheaders' performance and geomechanical characteristics of rock formations in tunnels and drift galleries in Tabas coal mine, the largest and the only fullymechanized coal mine in Iran. Then the database was analyzed in order to yield the most important factor for ICR by using relatively important factor in which Garson equation was utilized. The MLP network was trained by 3 input parameters including rock mass properties, rock quality designation(RQD), intact rock properties such as uniaxial compressive strength(UCS) and Brazilian tensile strength(BTS), and one output parameter(ICR). In order to have more validation on MLP outputs, KSOFM visualization was applied. The mean square error(MSE) and regression coefficient(R) of MLP were found to be 5.49 and 0.97, respectively. Moreover, KSOFM network has a map size of 8 5 and final quantization and topographic errors were 0.383 and 0.032, respectively. The results show that MLP neural networks have a strong capability to predict and evaluate the performance of medium-duty roadheaders in coal measure rocks. Furthermore, it is concluded that KSOFM neural network is an efficient way for understanding system behavior and knowledge extraction. Finally, it is indicated that UCS has more influence on ICR by applying the best trained MLP network weights in Garson equation which is also confirmed by KSOFM.     

利用神经元网络预测岩石或岩石工程的力学性态

本文将人工智能中的神经元网络引入岩石力学领域,用以预测岩石或岩石工程的力学性态。其用法类似经验公式,但其优点是影响岩体力学性态的各种描述性地质因素,均可做为变量输入,故可求得离散性较小的结果。文中附有两个实例,说明此方法的实用性。     

偏压双连拱公路隧道围岩稳定性动态预测分析

由于高速公路偏压双连拱隧道的复杂地质条件,会给隧道安全施工带来严重威胁,提出在加强隧道开挖现场监控量测的基础上,以位移量测结果作为学习样本,应用BP神经网络预测隧道围岩位移的大小,分析围岩的稳定性。由于BP神经网络能综合考虑隧道围岩节理、裂隙等对围岩位移的影响,所以与有限元反分析法计算隧道围岩位移结果比较,显示BP神经网络预测结果的误差较小,预测值与实际测量值趋于一致,因此应用BP网络预测偏压双连拱隧道围岩位移,超前分析其稳定性是安全可靠的,该预测方法的预测结果可以指导现场的施工。     

A comparative study on the application of various artificial neural networks to simultaneous prediction of rock fragmentation and backbreak

In blasting operation, the aim is to achieve proper fragmentation and to avoid undesirable events such as backbreak. Therefore, predicting rock fragmentation and backbreak is very important to arrive at a technically and economically successful outcome. Since many parameters affect the blasting results in a complicated mechanism, employment of robust methods such as artificial neural network may be very useful. In this regard, this paper attends to simultaneous prediction of rock fragmentation and backbreak in the blasting operation of Tehran Cement Company limestone mines in Iran. Back propagation neural network (BPNN) and radial basis function neural network (RBFNN) are adopted for the simulation. Also, regression analysis is performed between independent and dependent variables. For the BPNN modeling, a network with architecture 6-10-2 is found to be optimum whereas for the RBFNN, architecture 6-36-2 with spread factor of 0.79 provides maximum prediction aptitude. Performance comparison of the developed models is fulfilled using value account for (VAF), root mean square error (RMSE), determination coefficient (R²) and maximum relative error (MRE). As such, it is observed that the BPNN model is the most preferable model providing maximum accuracy and minimum error. Also, sensitivity analysis shows that inputs burden and stemming are the most effective parameters on the outputs fragmentation and backbreak, respectively. On the other hand, for both of the outputs, specific charge is the least effective parameter.     

Dynamic Wavelet Neural Network for Nonlinear Identification of Highrise Buildings

Abstract:   Recently, the authors presented a multiparadigm dynamic time-delay fuzzy wavelet neural network (WNN) model for nonparametric identification of structures using the nonlinear autoregressive moving average with exogenous inputs. Compared with conventional neural networks, training of a dynamic neural network for system identification of large-scale structures is substantially more complicated and time consuming because both input and output of the network are not single valued but involve thousands of time steps. In this article, an adaptive Levenberg–Marquardt least-squares algorithm with a backtracking inexact linear search scheme is presented for training of the dynamic fuzzy WNN model. The approach avoids the second-order differentiation required in the Gauss–Newton algorithm and overcomes the numerical instabilities encountered in the steepest descent algorithm with improved learning convergence rate and high computational efficiency. The model is applied to two highrise moment-resisting building structures, taking into account their geometric nonlinearities. Validation results demonstrate that the new methodology provides an efficient and accurate tool for nonlinear system identification of high-rising buildings.     

Prediction of thermal conductivity of rock through physico-mechanical properties

The transfer of energy between two adjacent parts of rock mainly depends on its thermal conductivity. Present study supports the use of artificial neural network (ANN) and adaptive neuro fuzzy inference system (ANFIS) in the study of thermal conductivity along with other intrinsic properties of rock due to its increasing importance in many areas of rock engineering, agronomy and geo environmental engineering field. In recent years, considerable effort has been made to develop techniques to determine these properties. Comparative analysis is made to analyze the capabilities among six different models of ANN and ANFIS. ANN models are based on feedforward backpropagation network with training functions resilient backpropagation (RP), one step secant (OSS) and Powell–Beale restarts (CGB) and radial basis with training functions generalized regression neural network (GRNN) and more efficient design radial basis network (NEWRB). A data set of 136 has been used for training different models and 15 were used for testing purposes. A statistical analysis is made to show the consistency among them. ANFIS is proved to be the best among all the networks tried in this case with average absolute percentage error of 0.03% and regression coefficient of 1, whereas best performance shown by the FFBP (RP) with average absolute error of 2.26%. Thermal conductivity is predicted using P-wave velocity, porosity, bulk density, uniaxial compressive strength of rock as input parameters.