MATLAB Neural Network Toolbox Research Articles

Energy saving effect prediction and post evaluation of air-conditioning system in public buildings

Accurate energy saving effect evaluation analysis of building energy efficiency retrofit is of benefit to obtain technology optimization and fast return of investment. According to the implement sequence, evaluation methods can be divided into post evaluation and prediction evaluation. The energy saving effect of an air-conditioning system retrofit project was analyzed by these two models respectively. The post evaluation model was built based on the spot test data and a parameter called as Refrigeration Operation Energy saving Effect Ratio (ROEER). The prediction evaluation model was built based on Back-Propagation Artificial Neural Network by the use of MATLAB Neural Network Toolbox. The comparison result between these two kinds of evaluation models match well with each other. These two models can be used to predict and evaluate energy saving effect of air-conditioning system retrofit to further improve the real energy saving effect of building energy efficiency retrofit.

Identification and control using MLP, Elman, NARXSP and radial basis function networks: a comparative analysis

This paper describes four neural networks multilayer perceptron (MLP) network, Elman network, NARXSP network and radial basis function (RBF) network. Neural networks are applied for identification and control of DC servo motor and benchmark nonlinear system. Number of epochs required and time taken to train the controller are shown in the form of bar plots for four neural networks. Levenberg-Marquardt algorithm is used for training the controller using neural network toolbox in MATLAB. Each neural network controller is run ten times. Their performances are compared for each run in terms of number of epochs required and time taken to train each controller for tracking a reference trajectory.

Forecast and Analysis of Transformation Point of TC6 Titanium Alloy

Transformation point of titanium alloy is an important material parameter in hot forming and heat treatment for titanium alloy. Used the transformation points of TC6 titanium alloy determined by metallographic method as desired outputs, a model for forecasting the transformation point of TC6 titanium alloy was established using MATLAB neural network toolbox under the guidance of back propagation neural network. Used the established model to forecast the transformation point of TC6 titanium alloy and analyze the effects of the contents of nine elements on the transformation point of TC6 titanium alloy, and the results show that the transformation point of TC6 titanium alloy increases with enhancement of N, O, C and Al contents, while decreases with the enhancement of Cr, Si, Mo, Fe and H contents.

BP Neural Network in Prediction of the Constant-Current Hydrostatic Bearing Static Stiffness

SKZT3500 NC rotary table adopts constant-current hydrostatic bearing and unloading guide two sets of hydraulic system. Aiming at the characteristics of two sets of hydraulic system, this paper deduces the constant-current hydrostatic bearing static stiffness formula. Then, the theory and algorithm of BP neural network were applied to predict the constant-current hydrostatic bearing static stiffness, based on experimental measurements in a physical prototype and neural network toolbox of MATLAB. Testing results show that BP neural network can accurately forecast the constant-current hydrostatic bearing of the static stiffness.

A Spatial Econometric Analysis of China’s Manufacturing Agglomeration based on Geoda and Matlab

Industrial agglomeration has gradually become an economic focus in recent years. Scholars has done a lot of research about the formation mechanism of industry agglomeration and its influencing factors, but the spatial correlation of variables has still been neglected. Firstly this paper gives a brief introduction about Geoda software and Matlab neural network toolbox, then use spatial statistical methods to describe the 1999-2008 China's manufacturing industry agglomeration. Secondly this paper uses spatial econometric methods to analyze the influencing factors of China’s provincial manufacturing Agglomeration. The results show that the spatial econometric model is superior to the traditional econometric models and the analysis based on spatial econometric model are more accurate. Finally, the paper also gives a brief forecast of the manufacturing Agglomeration.

Research on the Prediction Model of Laser Surface Hardening Index on Cylinder Liner Based on RBF

Based on the data obtained from the perpendicular experiment of laser surface hardening on HT150 cylinder liner of a certain motor engine, a Radial Basis Function (RBF) neural network prediction model of laser surface hardening index about HT150 cylinder liner is established by Matlab neural network toolbox in this paper. The parameters of input layer are affirmed by analyzing influential factors of the hardened index ,and the best form of the network is affirmed by selecting suitable spread in function-newrb( ),and as a result, the prediction accuracy and the adaptability of the network are improved. The result of testing the model indicates that the model based on RBF neural network has a good generalizing capability. Compared with traditional Back-Propagation Network (BP network), the result indicates that RBF has better accuracy and adaptability.

Study on Simulation of EDM Milling Technology Based on MATLAB

Due to the complexity of electrical discharge machining (EDM) milling technology, an artificial nerve network simulation and prediction model based on MATLAB for EDM milling technology is proposed. The three -layer BP neural network model is built by applying MATLAB technology and combining with the database toolbox and artificial neural network toolbox in MATLAB, it has the ability of getting experimental data from EDM milling technology database and calculating by neural network and then outputting the prediction result. Finally, the model is used to simulate the relations between material removal rate (MRR), surface roughness and process parameters, and analyzed the technological rule. It is proved that the prediction error is controlled within 10% and the simulation result can successfully map the technological rule of EDM milling technology.

Technology Route Planning of Large Composite Parts

In the current paper optimal design of the manufacturing technology route has been studied. For finding out the optimal technology route of the large composite plastic products the structure of the technology process has to be cut down into different process segments: vacuum forming technology, technology for post-forming operations, strengthening and assembling. The vacuum forming process has been analyzed with different materials like ABS/PMMA, Polycarbonate and Acrylic Plexiglas. In the glass-fiber reinforcement (polyester resin with randomly oriented short glass fibers) process the reinforcement layer thickness has been analyzed and optimized. The Neural Network meta-modeling technique has been used. The optimization of the plastic sheet and its reinforcement layer thickness has been performed using the surrogate design model. The Finite Element Analysis simulation has performed with optimal thickness values to verify the prediction accuracy of the surrogate model. For optimal selection of technology route the corresponding optimization model has been proposed. Model has based on maximization of total profit, where constraints are the workstation time capacities, material availability and optimal technology structure. A neural network approach and MatLab Neural Network Toolbox has been used.

Coal Mine Safety Evaluation with V-Fold Cross-Validation and BP Neural Network

A suit of coal mine safety evaluation indexes have been propounded under the study of current situation about coal mine safety assement. Coal mine safety evaluation system based on BP neural network has been established by employing MATLAB neural network toolbox, and the actual index data will be utilized to validate marking effect through using V-fold Cross-validation technique, and then this will be applied to coal mine safety evaluation system to provide grounds for safety production and management in coal mine.

Back propagation neural network based product cost estimation at an early design stage of passenger vehicles

The lack of effective cost estimation method is the bottleneck for product cost control at the early design stage. In this article, the product costs of sedan, sports utility vehicle and multi-purpose vehicle – the three major categories of passenger vehicles in the Chinese market are studied. A cost estimation architecture is proposed and a back propagation neural network based cost estimation method is developed for the early design stage of passenger vehicles. Users can change parameters to check corresponding influences on product cost or to compare the costs of different manufacturers. As the confidential cost information is inaccessible, the product cost is calculated backward from the price with the product cost ratio which is estimated by a panel of experts with Delphi method. Rather than introducing pilot data, real world data is adopted in this study. The prices and specifications of passenger vehicles are retrieved through the internet while the back propagation neural network is trained with the neural network toolbox of Matlab(TM) 7.1. Two neural network models are evaluated and the test reveals that the model selection has strong relation with the training data set. A case study on how this method is applied in a Chinese automobile company is also given.

Internal validation of an artificial neural network for prostate biopsy outcome

To carry out an internal validation of the retrospectively trained artificial neural network (ANN) 'ProstataClass'. A prospectively collected database of 393 patients undergoing 8-12 core prostate biopsy was analyzed. Data of these patients were applied to the online available ANN 'ProstataClass' using the Elecsys total prostate-specific antigen (tPSA) and free PSA (fPSA) assays. Beside the internal validation of the ANN 'ProstataClass' an additional ANN (named as ANN internal validation: ANNiv) only using the 393 prospective patient data was evaluated. The new ANN model was constructed with the MATLAB Neural Network Toolbox. Diagnostic accuracy was evaluated by receiver operator characteristic (ROC) curves comparing the areas under the ROC curves (AUC) and specificities at 90% and 95% sensitivity. Within a tPSA range of 1.0-22.8 ng/mL, 229 men (58.3%) had prostate cancer (PCa). tPSA, %fPSA and the number of positive digital rectal examinations (DRE) differed significantly from the cohort of patients of the ANN 'ProstataClass', whereas age and prostate volume were comparable. AUCs for tPSA, %fPSA and the ANN 'ProstataClass' were 0.527, 0.726 and 0.747 (P = 0.085 between %fPSA and ANN). The AUC of the ANNiv (0.754) was significantly better compared with %fPSA (P = 0.021), whereas the AUC of two ANN models built on external cohorts (0.726 and 0.729) showed no differences to %fPSA and the other ANN models. Significant differences of DRE status and %fPSA medians decrease the power of the 'ProstataClass' ANN in the internal validation cohort. The effect of retrospective data evaluation the 'ProstataClass' cohort and prospective fPSA measurement may be responsible for %fPSA differences. All ANN models built with different PSA and fPSA assays performed equally if applied to the two cohorts.

Artificial neural networks and neuro-fuzzy inference systems as virtual sensors for hydrogen safety prediction

Hydrogen is increasingly investigated as an alternative fuel to petroleum products in running internal combustion engines and as powering remote area power systems using generators. The safety issues related to hydrogen gas are further exasperated by expensive instrumentation required to measure the percentage of explosive limits, flow rates and production pressure. This paper investigates the use of model based virtual sensors (rather than expensive physical sensors) in connection with hydrogen production with a Hogen ®20 electrolyzer system. The virtual sensors are used to predict relevant hydrogen safety parameters, such as the percentage of lower explosive limit, hydrogen pressure and hydrogen flow rate as a function of different input conditions of power supplied (voltage and current), the feed of de-ionized water and Hogen ®20 electrolyzer system parameters. The virtual sensors are developed by means of the application of various Artificial Intelligent techniques. To train and appraise the neural network models as virtual sensors, the Hogen ®20 electrolyzer is instrumented with necessary sensors to gather experimental data which together with MATLAB neural networks toolbox and tailor made adaptive neuro-fuzzy inference systems (ANFIS) were used as predictive tools to estimate hydrogen safety parameters. It was shown that using the neural networks hydrogen safety parameters were predicted to less than 3% of percentage average root mean square error. The most accurate prediction was achieved by using ANFIS.

Training Radial Basis Function Networks with Differential Evolution

In this paper, Differential Evolution (DE) algorithm, a new promising evolutionary algorithm, is proposed to train Radial Basis Function (RBF) network related to automatic configuration of network architecture. Classification tasks on data sets: Iris, Wine, New-thyroid and Glass are conducted to measure the performance of neural networks. Compared with a standard RBF training algorithm in Matlab neural network toolbox, DE achieves more rational architecture for RBF networks. The resulting networks hence obtain strong generalization abilities.

Correlation of Vapour Liquid Equilibria of Binary Mixtures Using Artificial Neural Networks

In this paper, a back propagation artificial neural network (BP-ANN) model is presented for the simultaneous estimation of vapour liquid equilibria (VLE) of four binary systems viz chlorodifluoromethan-carbondioxide, trifluoromethan-carbondioxide, carbondisulfied-trifluoromethan and carbondisulfied-chlorodifluoromethan. VLE data of the systems were taken from the literature for wide ranges of temperature (222.04–343.23K) and pressure (0.105 to 7.46MPa). BP-ANN trained by the Levenberg-Marquardt algorithm in the MATLAB neural network toolbox was used for building and optimizing the model. It is shown that the established model could estimate the VLE with satisfactory precision and accuracy for the four systems with the root mean square error in the range of 0.054–0.119. Predictions using BP-ANN were compared with the conventional Redlich-Kwang-Soave (RKS) equation of state, suggesting that BP-ANN has better ability in estimation as compared with the RKS equation (the root mean square error in the range of 0.115–0.1546).

DGPS Correction Prediction Using Artificial Neural Networks

This paper focuses on modelling and predicting differential GPS corrections transmitted by marine radio-beacon systems using artificial neural networks. Various neural network structures with various training algorithms were examined, including Linear, Radial Biases, and Feedforward. Matlab Neural Network toolbox is used for this purpose. Data sets used in building the model are the transmitted pseudorange corrections and broadcast navigation message. Model design is passed through several stages, namely data collection, preprocessing, model building, and finally model validation. It is found that feedforward neural network with automated regularization is the most suitable for our data. In training the neural network, different approaches are used to take advantage of the pseudorange corrections history while taking into account the required time for prediction and storage limitations. Three data structures are considered in training the neural network, namely all round, compound, and average. Of the various data structures examined, it is found that the average data structure is the most suitable. It is shown that the developed model is capable of predicting the differential correction with an accuracy level comparable to that of beacon-transmitted real-time DGPS correction.

Radial basis function neural network-based model predictive control for freeway traffic systems

A method of Radial Basis Function neural network-based Model Predictive Control (RBF-MPC) for freeway traffic systems is proposed in this paper. Because of nonlinearity and uncertainty of freeway traffic flow, an accurate mathematics model cannot be obtained. Therefore, RBF neural networks employed to predict the future behaviours of freeway traffic flow are designed based on the MATLAB neural network toolbox. Then, to handle nonlinearity, time delay, uncertainty and strong disturbance, RBF-MPC for ramp metering is proposed. A Genetic Algorithm (GA) is used in the receding horizon optimisation. Compared with the no-control case and optimal-control case, the simulation results demonstrate that the proposed approach can alleviate traffic jams and increase main road capacity; thus the efficiency of freeway traffic is improved tremendously.

Design of microwave structures with MEFISTO‐3D NOVA and MATLAB optimization and neural network toolboxes

AbstractThis paper introduces two time‐domain field‐based optimization procedures for microwave engineering. The methods are built on the foundations of MATLAB's optimization and neural network toolboxes. The first procedure makes use of a direct connection linking MATLAB's optimization toolbox with MEFISTO‐3D NOVA. In this approach the field simulator acts as an objective function server for the optimization toolbox; these two programs work cooperatively with each other to tune the structure parameters to obtain a target response. The second procedure is an indirect optimization approach that makes use of MATLAB's neural network toolbox in conjunction with MEFISTO‐3D NOVA to create a neural network model to emulate the structure of interest; the resulting neural network model is then used an objective function server in a normal MATLAB optimization process. Copyright © 2006 John Wiley & Sons, Ltd.

Integrating Matlab Neural Networks Toolbox functionality in a fully reusable software component library

In this work, we present a reusable software component library, called IMO.Net Artificial Neural Networks library, which encapsulates the functionality of the Matlab Neural Networks Toolbox (MNNT). The MNNT is a powerful tool to work with neural networks. However, MNNT has not been conceived as a reusable and integrable software piece, and its results are clearly inadequate to be used for development of applications. The component library presented in this paper is fully reusable, allowing the integration of the neural networks toolbox objects and functions in software applications, independently of the platform and tools used to build it. Furthermore, this library provides two different sets of classes to the programmers, one of them presenting an application program interface (API) similar to the Matlab toolbox, and the other a fully object-oriented designed API, which is easier to use and more adequate for object-oriented and rapid application development. This library combines the advantages of being fully reusable and internally employing the MNNT to perform the algorithmic task, consequently inheriting all its power, robustness and comprehensiveness, but without its reusability limitations.

Neural network approach to contingency screening and ranking in power systems

An artificial neural network (ANN) approach to power system contingency analysis is proposed. Using fast voltage and line-flow contingency screening. Full AC load flow is performed for each contingency case. The off-line results of full AC load flow calculations are used to construct two kinds of performance indices, namely the real power performance index ( PI MW) and voltage performance index ( PI V), which reflect the degree of severity of contingencies. The results from off-line load flow calculations are used to train the “ screening module”, which is a multi-layered perceptron (MLP) network, for estimating the performance indices ( PI MW, PI V PI V). The MLP is trained to classify the contingencies either as critical or non-critical cases using back-propagation (BP) algorithm. The screened critical contingencies are passed on to the “ ranking module” for ranking of the contingencies. The effectiveness of the proposed method is demonstrated by contingency screening and ranking on a standard 6-bus and IEEE 14-bus systems. The performance of the proposed method is compared with a traditional Newton Raphson (NR) method and the results discussed. The proposed methodology was implemented using the MATLAB Neural Network Toolbox. The generalization capability of the trained neural network was able to identify unknown contingencies with large range of operating conditions and changes in network topology. The proposed approach to contingency analysis was found to be suitable for fast voltage and line-flow contingency screening and ranking.

A speed and accuracy test of backpropagation and RBF neural networks for small-signal models of active devices

Backpropagation networks are compared to radial basis function (RBF) networks when it comes to small signal modeling RF/microwave active devices. The modeled device is a 4×50 μm gate width, 0.25 μm gate length gallium arsenide (GaAs) Metal semiconductor field-effect transistor (MESFET). It is the authors’ intent to prove that RBF networks provide much better performance than backpropagation when it comes to this type of modeling. First, two separate backpropagation networks are created to determine the best training algorithm in terms of convergence speed. Then, the backpropagation network, using its best training algorithm, is compared to the RBF network in terms of speed and accuracy. Simulation results are presented in tables and figures for better understanding. All tests and simulations for the backpropagation and RBF networks are done using Matlab's Neural Network Toolbox.