Propagation Artificial Neural Network Research Articles

Artificial Neural Networks for Modeling Rheological Properties under High-Temperature and High-Pressure in Gas Well Cement Slurries

Artificial neural networks (ANN) was used to predict the rheological properties of high temperature high pressure gas well cement slurries. Seven different materials were used as additives which includes: Fresh water, dyckerhoff, silica flour, antifoam, extender, fluid loss, dispersant, retarder, anti-settling agent, gas control agent, dry viscosifier, potassium chloride and accelerator. Four recipes were prepared using these additives in different mixtures. Recipe four have all the additives. The rheological properties were investigated at different temperatures in the range of 23 to 60ºC using an advanced shear-stress/shear-strain controlled rheometer. Experimental data thus obtained were used to develop predictive models based on back- propagation artificial neural networks. It was found that ANN depicted good agreement with the experimental data, with ANN achieving more accurate predictions. The developed models could effectively predict the rheological properties of new slurries designed within the range of input parameters of the experimental database with an absolute error of 3.43, 3.17, and 2.82%, in the case of ANN, for the different recipes. The flow curves developed using ANN allowed predicting the Bingham parameters (yield stress and plastic viscosity) of the oil well slurries with adequate accuracy. The goal of the process is to choose the network that minimizes the prediction errors/RMSEs. There is however need to avoid an over-trained network. The result showed that over-training of the networks sets in around the scenario when the number of hidden layer neurons exceeds 9. It also demonstrates that the network with 9 hidden layer neurons gave the least RMSEs, and it is this network that has been adopted as the network for the final model development in this work.

Exhaustive investigation of drug delivery systems to achieve optimal condition of drug release using non-linear generalized artificial neural network method: feedback from the loading step of drug

Drug delivery systems are potential systems with ability to release drugs with a variety of mechanisms. Some mechanisms may consist of multiple steps, where the release rate of each step can vary from the others. Moreover, a drug release is a kinetic process intrinsically and the initial amount of the loaded drug may influence the amount of the release. Therefore, to achieve the desirable release, the loading and release processes should be investigated simultaneously, while researchers have considered these processes independently so far. Considering the fact that functional dependence between loading and release is not obvious, we proposed the combination of experimental design and powerful non-linear generalized artificial neural network (G-ANN) methods for the simultaneous optimization of these processes. Here, the functionalized PEGylated KIT-6 ([β-CD@PEGylated KIT-6]) NPs and curcumin were selected as a nano-carrier and drug, respectively. The curcumin release was optimized by G-ANN by considering the the feedback from the loading step. The obtained optimal parameters were as follows: (in the release process) 1.80 of the weight ratio of drug to nano-carrier, 5.70 of pH and 120 h of release time; and (in the loading process) 43 h of loading time and 2.2 of weight ratio of drug/nano-carrier.

Generalized wavelet neural networks for evapotranspiration modeling in India

This study focuses on the application of generalized wavelet neural network (GWNN) models corresponding to the FAO-56 Penman Monteith (FAO-56 PM), Turc, and Hargreaves (HG) methods for estimating daily reference evapotranspiration (ETo). The daily pooled climate data from 15 different locations under 4 different agro-ecological regions (AERs: semi-arid, arid, sub-humid, and humid) in India are used as an input to GWNN models. The inputs include three combinations of climate data (minimum and maximum air temperatures, minimum and maximum relative humidity, wind speed, and solar radiation) and the target consists of the FAO-56 PM estimated ETo. Further, the GWNN models were applied to 15 individual model development locations and 10 different model testing locations to test the generalizing capability. Comparison of developed GWNN models was made with the classic generalized artificial neural network (GANN), generalized linear regression (GLR), generalized wavelet regression (GWR), and corresponding conventional methods to test the superiority of one model over the other. Results reveal that the GWNN models followed by GANN models performed better than GWR and GLR models for four AERs. The testing results suggest that the GWNN and GANN models have better generalizing capabilities than the GWR and GLR for almost all region locations.

Discriminating Biomass and Nitrogen Status in Wheat Crop by Spectral Reflectance Using Artificial Neural Networks

Precision Agriculture has the goal of reducing cost which is difficult when it is related to fertilizers application. Nitrogen (N) is the nutrient absorbed in greater amounts by crops and the N fertilizers application present significant costs. The use of spectral reflectance sensors has been studied to identify the nutritional status of crops and prescribe varying N rates. This study aimed to contribute to the determination of a model to discriminating biomass and nitrogen status in wheat through two sensors, GreenSeeker and Crop Circle, using the Resilient Propagation and Backpropagation Artificial Neural Networks algorithms. As a result was detected a strong correlation to the sensor readings with the aboveground biomass production and N extraction by plants. For both algorithms it was established a satisfactory model for estimating wheat dry biomass production. The best Backpropagation and Resilient Propagation models defined showed better performance for the GreenSeeker and Crop Circle sensors, respectively.

An Improved Method for Sizing Standalone Photovoltaic Systems Using Generalized Regression Neural Network

In this research an improved approach for sizing standalone PV system (SAPV) is presented. This work is an improved work developed previously by the authors. The previous work is based on the analytical method which faced some concerns regarding the difficulty of finding the model’s coefficients. Therefore, the proposed approach in this research is based on a combination of an analytical method and a machine learning approach for a generalized artificial neural network (GRNN). The GRNN assists to predict the optimal size of a PV system using the geographical coordinates of the targeted site instead of using mathematical formulas. Employing the GRNN facilitates the use of a previously developed method by the authors and avoids some of its drawbacks. The approach has been tested using data from five Malaysian sites. According to the results, the proposed method can be efficiently used for SAPV sizing whereas the proposed GRNN based model predicts the sizing curves of the PV system accurately with a prediction error of 0.6%. Moreover, hourly meteorological and load demand data are used in this research in order to consider the uncertainty of the solar energy and the load demand.

A novel user classification method for femtocell network by using affinity propagation algorithm and artificial neural network.

An artificial neural network (ANN) and affinity propagation (AP) algorithm based user categorization technique is presented. The proposed algorithm is designed for closed access femtocell network. ANN is used for user classification process and AP algorithm is used to optimize the ANN training process. AP selects the best possible training samples for faster ANN training cycle. The users are distinguished by using the difference of received signal strength in a multielement femtocell device. A previously developed directive microstrip antenna is used to configure the femtocell device. Simulation results show that, for a particular house pattern, the categorization technique without AP algorithm takes 5 indoor users and 10 outdoor users to attain an error-free operation. While integrating AP algorithm with ANN, the system takes 60% less training samples reducing the training time up to 50%. This procedure makes the femtocell more effective for closed access operation.

GANN models for reference evapotranspiration estimation developed with weather data from different climatic regions

Accurate estimation of reference evapotranspiration (ET0) becomes imperative for better managing the more and more limited agricultural water resources. This study examined the feasibility of developing generalized artificial neural network (GANN) models for ET0 estimation using weather data from four locations representing different climatic patterns. Four GANN models with different combinations of meteorological variables as inputs were examined. The developed models were directly tested with climatic data from other four distinct stations. The results showed that the GANN model with five inputs, maximum temperature, minimum temperature, relative humidity, solar radiation, and wind speed, performed the best, while that considering only maximum temperature and minimum temperature resulted in the lowest accuracy. All the GANN models exhibited high accuracy under both arid and humid conditions. The GANN models were also compared with multivariate linear regression (MLR) models and three conventional methods: Hargreaves, Priestley–Taylor, and Penman equations. All the GANN models showed better performance than the corresponding MLR models. Although Hargreaves and Priestley–Taylor equations performed slightly better than the GANN models considering the same inputs at arid and semiarid stations, they showed worse performance at humid and subhumid stations, and GANN models performed better on average. The results of this study demonstrated the great generalization potential of artificial neural techniques in ET0 modeling.

Process Parameters Optimization for Isothermal Forging of Ti-6Al-4V Alloy using Taguchi Method and Artificial Neural Network

Enabling and developing the required „Process‟ efficiency in an Input-Process-Output System, is bound to provide effective solutions for today‟s need for maximized production especially in dwindling Input conditions. Process optimization is a significant and contributing step towards such process efficiency as it paves way for improved overall productivity. The present work involves, an approach using soft computing paradigm for the process parameter optimization of multiple input single output isothermal forging of Ti-6Al-4V alloy. In this paper a combination of Taguchi‟s L27 Orthogonal Array (OA-L27) along with back- propagation artificial neural network and engineering optimization concepts to determine the optimal process parameter settings of forging temperature, strain rate and strain for the isothermal forging. The optimum solution is valid in the ranges of forging process parameters that were used for training the artificial neural network.

High-Degree-of-Freedom Engine Modelling for Control Design Using a Crank-Angle-Resolved Flame Propagation Simulation and Artificial Neural Network Surrogate Models

Non-linearity of the engine system creates a challenge in building a reliable control-oriented model (COM). The main source of non-linearity is the complex nature of the combustion process. Modern engine system configurations are increasingly complex and predicting their transient response poses additional difficulty. In the present paper, a COM is developed to address the challenges and capture the behaviour of a high-degree-of-freedom engine system. Engine combustion models are created by utilizing the high-fidelity engine cycle simulation to characterize the effects of main parameters, such as turbulence, air—fuel ratio, and residual fraction, and subsequently capturing the interrelationships with artificial neural networks. Then, system dynamics are accounted for by adding manifold and actuator dynamics models. The capabilities of the proposed COM are demonstrated using a spark-ignition engine with a dual-independent cam phasing as a test case. The results indicate the model's ability to accurately predict engine responses to an arbitrary schedule of engine control inputs over the feasible operating range.

Preliminary assessment of a model to predict mold contamination based on microbial volatile organic compound profiles

Identification of mold growth based on microbial volatile organic compounds (MVOCs) may be a viable alternative to current bioaerosol assessment methodologies. A feed-forward back propagation (FFBP) artificial neural network (ANN) was developed to correlate MVOCs with bioaerosol levels in built environments. A cross-validation MATLAB script was developed to train the ANN and produce model results. Entech Bottle-Vacs were used to collect chemical grab samples at 10 locations in northern NY during 17 sampling periods from July 2006 to August 2007. Bioaerosol samples were collected concurrently with chemical samples. An Anderson N6 impactor was used in conjunction with malt extract agar and dichloran glycerol 18 to collect viable mold samples. Non-viable samples were collected with Air-O-Cell cassettes. Chemical samples and bioaerosol samples were used as model inputs and model targets, respectively. Previous researchers have suggested the use of MVOCs as indicators of mold growth without the use of a pattern recognition program limiting their success. The current proposed strategy implements a pattern recognition program making it instrumental for field applications. This paper demonstrates that FFBP ANN may be used in conjunction with chemical sampling in built environments to predict the presence of mold growth.

The ANN Method of Obtaining Fatigue Load Spectrum of Overhead Traveling Crane

For the realization of remaining fatigue life estimation, prevention of catastrophic accidents and safety of the cranes in service, it is necessary to obtain the actual load spectrum of overhead traveling crane. Due to the limitation of condition in field experiment, it is difficult to implement a large number of actual Load Spectrum experiments. So, the present paper will focus on overhead traveling cranes for general purpose and conduct data investigation. In a certain period，the numbers of work cycles corresponding to different lifting loads for different rated lifting capacity are collected. Based on Levenberg-Marquardt Back Propagation(LMBP) Artificial Neural Network(ANN), the equivalent load spectrum, which is equivalent to the actual load spectrum, is acquired. The example demonstrates: It is feasible to establish the equivalent load spectrum for some types of crane using ANN. Meanwhile, the equivalent load spectrums of the typical cranes in this type could be obtained and predicted through the trained network. It has provided the reliable data for the reliability analysis and fatigue life estimation of the crane. In the approach to establish the crane load spectrum, an exploring and initiating study work have been done.

Development and Validation of GANN Model for Evapotranspiration Estimation

The present study was carried out to develop generalized artificial neural network (GANN) based reference crop evapotranspiration models corresponding to FAO-56 PM, FAO-24 Radiation, Turc, and FAO-24 Blaney–Criddle methods. The generalized ANN models were developed using the data from four California Irrigation Management Information System (CIMIS) stations, namely, Davis, Castroville, Mulberry, and West Side Field Station. The average weighted standard error of estimate (WSEE) for the developed models, namely, GANN (4-5-1), GANN (3-4-1), GANN (5-6-1), and GANN (6-7-1) corresponding to the FAO-24 Blaney–Criddle, FAO-24 Radiation, Turc, and FAO-56PM was 0.72, 0.85, 0.63, and 0.48mmday−1, respectively. The developed ANN models were applied at 2 CIMIS stations namely, Lodhi and Fresno, without any local training. The average WSEE for models GANN (4-5-1), GANN (3-4-1), GANN (5-6-1), and GANN (6-7-1) was 0.68, 0.71, 0.65, and 0.46mmday−1, respectively In addition, the GANN (4-5-1) model corresponding to FAO-24 Blaney–Criddle was directly applied to four Indian locations, namely, Hoshangabad, Gwalior, Jabalapur, and Pendra. The model gave the average WSEE of 0.57mmday−1. Based on the results it was concluded that the GANN models can be used directly to predict evapotranspiration (ETo) under the arid conditions, since they performed better than the conventional evapotranspiration (ETo) estimation method.

EEG power spectrum and neural network based sleep-hypnogram analysis for a model of heat stress

An effective application of back- propagation artificial neural network (ANN) in preparation of sleep-hypnogram based on electroencephalogram (EEG) power spectra under acute as well as chronic heat stress has been presented. Rats were divided in three groups (i) acute heat stress-subjected to a single exposure for four hours at 38 degrees C; (ii) chronic heat stress-exposed for 21 days daily for one hour at 38 degrees C, and (iii) handling control groups. The preprocessed EEG signals were fragmented in two-second artifact free epochs for calculation of power spectra, training and testing of ANN. The power spectrum analyses of EEG show that changes in higher frequency components (beta(2)) were significant in all sleep-wake states following both acute and chronic heat stress conditions. The power of beta(2) activity after acute heat exposure was significantly decreased during SWS (slow wave sleep) (P < 0.05) and REM (rapid eye movement) sleep (P < 0.05), while reverse was observed in AWA (awake state) (P < 0.05). Following chronic heat exposure, beta(2) activity was found increased in all three sleep-wake stages (P < 0.05). The ANN used for sleep-hypnogram preparation contains 64 nodes in input layer, weighted from power spectrum data from 0 to 32 Hz, 14 nodes in hidden layer and 3 output nodes. The results obtained from the study, suggest increased sleep efficiency following acute exposure to heat stress while fragmented sleep with decreased sleep efficiency following chronic heat stress. The ANN can be used for the analysis of stressful events by calculating the sleep-EEG alterations.

Artificial neural network analysis of preparation of nano α-Al2O3 powders by thermal decomposition of ammonium aluminium carbonate hydroxide

Nano α-Al2O3 powders were prepared by thermal decomposition of ammonium aluminium carbonate hydroxide (AACH). The effects of α-Al2O3 crystal seeds, reactant concentration, content of active agent (PEG), annealing temperature and ZnF2 addition on the preparation of nano α-Al2O3 powder were studied. The results showed that nano α-Al2O3 powder with apparent crystallite size ∼40 nm could be prepared, and the complete transformation temperature of α-Al2O3 could be reduced from 1250 to 1050°C as α-Al2O3 seeds and ZnF2 were added. The prepared nano α-Al2O3 powders were characterised by X-ray diffraction (XRD), differential thermal analysis and thermogravimetric (TG-DTA) and scanning electron microscopy (SEM). A back propagation (BP) artificial neural network (ANN) has been used to establish models between the relative content of the nano α-Al2O3 powders and the preparation parameters. The results of checking experiments were in accordance with the calculated values.

An integrated method for multiobjectives cell formation in cellular manufacturing systems

In this research, an integrated method was developed to solve a multiobjective cell formation problem. The method consists of an integer programming model, a heuristic algorithm for generating alternative cell formations, a computer simulation model for investigating system performances and a Back Propagation (BP) artificial neural network model for selecting the best system configuration. The criteria considered in solving the overall cell formation problem include system related costs, part completion times, queue lengths and machine utilisation. The BP neural network is designed to incorporate expert experience to evaluate different cell formations considering system costs and performances. Numerical examples are provided to test and illustrate the model and the developed method.

GSM Mobile Station Location Using Reference Stations and Artificial Neural Networks

In this paper we present a novel approach to the automatic GSM mobile station location. The approach is based on measurement of radio signal strengths from a number of the neighboring base stations (antennas) and estimation of the mobile station position using trained artificial neural network (ANN) models. First, we present an improved version of our previous positioning back propagation (BP) ANN multi-level perceptron (MLP) model that further improves positioning accuracy. Then, we extend the MLP primary ANN model by introducing correctional factors obtained from a number of reference stations with known positions. Two new models with the improved location accuracy, both aimed at real-time application, are presented. The first model is using differential range to improve the estimated location of the mobile station. The second is using small-scale secondary neural networks trained with data obtained from reference stations, in addition to the primary ANN, to correct location accuracy.

A trial of artificial neural networks for automatically estimating the age of fish

Artificial neural networks (ANNs) have the potential to automate routine ageing of fish with the benefit of increased speed in processing, greater objectivity and repeatability of estimates, and a mechanism for quantifying uncertainty of age estimates. ANN models were tested as a means of objectively replicating the age estimates of an experienced human reader. Feed-forward back- propagation ANNs, with three layers of neurons (input, hidden and output), were trained to classify the age of previously aged samples of three temperate species. Three ANN structures, where the number of neurons in the hidden layer was varied, were tested for each species. Inputs to each ANN were pixel brightness values along transects across images of sectioned otoliths. The ANN predicted age-class membership by the position of the neuron in the output layer with the highest value. After training, at least one of the three ANN structures correctly classified the age of fish from unseen transects for two members of the Sparidae family Acanthopagrus butcheri and Pagrus auratus at an accuracy level approaching that of an expert reader. For a member of the Merlucciidae family, Macruronus novaezelandiae, however, which is a species with more complex otolith structure, error rates were high for all three ANN structures tested.