Effects of mineral admixtures on fresh and hardened properties of self-compacting concretes: binary,ternary and quaternary systems

The paper presented herein investigates the effects of using supplementary cementitious materials in binary, ternary, and quaternary blends on the fresh and hardened properties of self-compacting concretes (SCCs). A total of 22 concrete mixtures were designed having a constant water/binder ratio of 0.32 and total binder content of 550 kg/m³. The control mixture contained only portland cement (PC) as the binder while the remaining mixtures incorporated binary, ternary, and quaternary cementitious blends of PC, fly ash (FA), ground granulated blast furnace slag (GGBFS), and silica fume (SF). After mixing, the fresh properties of the concretes were tested for slump flow time, L-box height ratio, V-funnel flow time, setting time, and viscosity. Moreover, compressive strength, ultrasonic pulse velocity, and electrical resistivity of the hardened concretes were measured. Test results have revealed that incorporating the mineral admixtures improved the fresh properties and rheology of the concrete mixtures. The compressive strength and electrical resistivity of the concretes with SF and GGBFS were much higher than those of the control concrete.     

Mechanical strength and setting times estimation of hydroxyapatite cement by using neural network

In this study, the mechanical strength, the initial and the final setting times in hydroxyapatite (HA) bone cement are estimated by designing a back-propagation neural network (BPNN) which has 2 inputs and 3 outputs. Firstly, some experimental samples have been prepared to train the BPNN to get it to estimate the output parameters. Then BPNN is tested using some experimental samples that have not been used in the training stage. To prepare the training and testing data sets, some experiments were performed. In these experiments, the β-tricalcium phosphate (β-TCP), the calcium carbonate and the dicalcium phosphate are used to prepare the powder part of the HA bone cement. Also the liquid part of the cement consists of the NaH₂PO₄⋅2H₂O solution with different concentrations. The effects of liquid phase concentration and the liquid/powder ratio of the cement, as input parameters, have been investigated on the setting times and the mechanical strength of the cement, as output parameters. The comparison of the predicted values and the experimental data indicates that the developed model has an acceptable performance to estimation of the setting times and the mechanical strength in HA bone cement. Also three neural networks with 2-inputs and 1-output was developed, similar to above method, and were compared with 3-outputs model. It is found that the prediction accuracy of 3-outputs model is better than those of other 1-output models.     

Artificial neural network model of the strength of thin rectangular plates with weld induced initial imperfections

Probabilistic assessment of post-buckling strength of thin plate is a difficult problem because of computational effort needed to evaluate single collapse load. The difficulties arise from the nonlinear behaviour of an in-plane loaded plate showing up multiple equilibrium states with possible bifurcations, snap-through or smooth transitions of states. The plate strength depends heavily on the shape of geometrical imperfection of the plate mid-surface. In this paper, an artificial neural network (ANN) is employed to approximate the collapse strength of plates as a function of the geometrical imperfections. For the training set, mainly theoretical imperfections with the corresponding collapse loads of plate calculated by FEM are considered. The ANN validation is based on the measured imperfections of ship plating and FEM strength.     

Integration of rough set and neural network ensemble to predict the configuration performance of a modular product family

Configuration performance prediction (CPP) is critical in the whole process of configuration design for a modular product family. Its aim is to estimate the key performance parameter values in advance, thus evaluating if the product variant can satisfy the customers’ personalised requirements or not. In this paper, we propose a novel prediction approach based on the integration of rough set and neural network ensemble through discovering the knowledge from the historical configuration information table. The minimal hitting set is introduced and its equivalence relationship with the minimal attribute reduction is proven. A genetic algorithm is designed to perform the approximate reduction of the condition attributes. A neural network ensemble model used for regression prediction is constituted by means of the variant bagging method based on error clustering. This methodology can reuse the discovered configuration rules and knowledge efficiently, as well as reduce the effort of experimental measurement to some extent. Finally, the applicability of this prediction method is verified on a newly developed refrigerator family.     

Determination of deformation and failure properties of ductile materials by means of the small punch test and neural networks

This paper describes an approach to identify plastic deformation and failure properties of ductile materials. The experimental method of the small punch test is used to determine the material response under loading. The resulting load displacement curve is transferred to a neural network, which was trained using load displacement curves generated by finite element simulations of the small punch test and the corresponding material parameters. The simulated material behavior of the specimen is based on the ductile elastoplastic damage theory of Gurson, Tvergaard and Needleman. During a training process the neural network generates an approximated function for the inverse problem relating the material parameters to the shape of the load displacement curve of the small punch test. This technique was tested for three different materials (ductile steels). The identified parameters are verified by testing and simulating notched tensile specimens.     

Super‐resolution of the undersampled and subpixel shifted image sequence by a neural network

Numerous approaches to super‐resolution (SR) of sequentially observed images (image sequence) of low resolution (LR) have been presented in the past two decades. However, neural network methods are almost ignored for solving SR problems. This is because the SR problem traditionally has been regarded as the optimization of an ill‐posed large set of linear equations. A designed neural network based on this has a large number of neurons, thereby requiring a long learning time. Also, the deduced cost function is overly complex. These defects limit applications of a neural network to an SR problem. We think that the underlying meaning of the SR problem should refer to super‐resolving an imaging system by image sequence observation, instead of merely improving the image sequence itself. SR can be regarded as a pattern mapping from LR to SR images. The parameters of the pattern mapping can be learned from the imaging process of the image sequence. This article presents a neural network for SR based on learning from the imaging process of the image sequence. In order to speed up the convergence, we employ vector mapping to train the neural network. A mapping vector is composed of some neighbor subpixels. Such a well‐trained neural network has powerful generalization ability so that it can be used directly to estimate the SR image of the other image sequences without learning again. Our simulations show the effectiveness of the proposed neural network. © 2004 Wiley Periodicals, Inc. Int J Imaging Syst Technol 14, 8–15, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ima.20001     

Prediction of the concrete compressive strength by means of core testing using GMDH-type neural network and ANFIS models

Assessment of insitu concrete strength by means of cores cut from hardened concrete is accepted as the most common method, but may be affected by many factors. Group method of data handling (GMDH) type neural networks and adaptive neuro-fuzzy inference systems (ANFIS) were developed based on results obtained experimentally in this work along with published data by other researchers. Genetic algorithm (GA) and singular value decomposition (SVD) techniques are deployed for optimal design of GMDH-type neural networks. Samples incorporated six parameters with core strength, length-to-diameter ratio, core diameter, aggregate size and concrete age considered as inputs and standard cube strength regarded as the output. The results show that a generalized GMDH-type neural network and ANFIS have great ability as a feasible tool for prediction of the concrete compressive strength on the basis of core testing. Moreover, sensitivity analysis has been carried out on the model obtained by GMDH-type neural network to study the influence of input parameters on model output.     

Artificial neural network analysis of the effect of matrix size and milling time on the properties of flake Al-Cu-Mg alloy particles synthesized by ball milling

In this study, the effect of matrix size and milling time on the particle size, apparent density, and specific surface area of flake Al-Cu-Mg alloy powders was investigated both by experimental and artificial neural networks model. Four different matrix sizes (28, 60, 100, and 160?µm) and five different milling times (0.5, 1, 1.5, 2, and 2.5?h) were used in the fabrication of the flake Al-Cu-Mg alloy powders. A feed forward back propagation artificial neural network (ANN) system was used to predict the properties of flake Al-Cu-Mg alloy powders. For training process, the ANN models of the flake size, apparent density, and specific surface area have the mean square error of 0.66, 0.004, and 0.01%. For testing process, it was obtained that the R² values were 0.9984, 0.9998, and 0.9932 for the flake size, apparent density, and specific surface area, respectively. The degrees of accuracy of the prediction models were 95.145, 99.705, and 94.25% for the flake size, apparent density, and specific surface area, respectively.     

An analysis of the influence of hand hole and ventilation hole design on compressive strength of corrugated fiberboard boxes by an artificial neural network model

This research aimed at developing a high-performing corrugated fiberboard box compression strength prediction model and to analyze the influences of ventilation and hand hole designs for these containers on the box compression test (BCT) by applying artificial neural network (ANN) modeling. The input variables considered in this study are composed of nine parameters including box dimension as well as shapes, sizes, positions, and locations of ventilations and hand holes of a regular slotted container (RSC, FEFCO 0201). Back propagation algorithms (BPNs) of ANN models were developed from 970 BCT testing data points (single wall boards, C flute, 205/112/205 g/m²). Tested data was randomly broken into three groups for the model development as 80:10:10 for the training set, testing set, and validating set. According to the analysis performed, a BPN 9-13-1 model reflected the highest prediction performance with R² = 0.97. According to the analysis, BCT was significantly affected by the hand hole location followed by the geometrical dimensions of the box (height, length, and width) and the ventilation factors (shape, number, and location) in that order. Hand holes at the top flaps caused a lower BCT reduction compared with those at the vertical locations of the box. Slight changes to the eliminated board area for both hand holes and ventilation (±5%) contributed to less BCT reduction compared with its locations and shapes. Interestingly, increasing the box height significantly increased the BCT, and this was found to be limited only to shorter boxes fabricated from a high stiffness corrugated board.