Trained Back Propagation Neural Network Research Articles

Prediction of Salinity Variations in a Tidal Estuary Using Artificial Neural Network and Three-Dimensional Hydrodynamic Models

The simulation of salinity at different locations of a tidal river using physically-based hydrodynamic models is quite cumbersome because it requires many types of data, such as hydrological and hydraulic time series at boundaries, river geometry, and adjusted coefficients. Therefore, an artificial neural network (ANN) technique using a back-propagation neural network (BPNN) and a radial basis function neural network (RBFNN) is adopted as an effective alternative in salinity simulation studies. The present study focuses on comparing the performance of BPNN, RBFNN, and three-dimensional hydrodynamic models as applied to a tidal estuarine system. The observed salinity data sets collected from 18 to 22 May, 16 to 22 October, and 26 to 30 October 2002 (totaling 4320 data points) were used for BPNN and RBFNN model training and for hydrodynamic model calibration. The data sets collected from 30 May to 2 June and 11 to 15 November 2002 (totaling 2592 data points) were adopted for BPNN and RBFNN model verification and for hydrodynamic model verification. The results revealed that the ANN (BPNN and RBFNN) models were capable of predicting the nonlinear time series behavior of salinity to the multiple forcing signals of water stages at different stations and freshwater input at upstream boundaries. The salinity predicted by the ANN models was better than that predicted by the physically based hydrodynamic model. This study suggests that BPNN and RBFNN models are easy-to-use modeling tools for simulating the salinity variation in a tidal estuarine system.

Identifikasi Tahu Berformalin dengan Electronic Nose Menggunakan Jaringan Syaraf Tiruan Backpropagation

During this time to identify pure and formalin tofu based on color and aroma involving human taster. But this tofu tester still has weaknesses such as subjective. Besides that, the standard chemical analytical methods requires a high cost and need expertise to analyzing it. Basically aroma of tofu is determined by volatile compounds such as heksanal, ethanol, and 1-hexanol, while aroma of formalin tofu is determined by volatile compounds such as OH, CO, and hydrocarbon. Electronic nose based on unselected gas sensor array has the ability to analyze samples with complex compositions that can be known characteristics and qualitative analysis of the samples. Stimulus aroma is transformed by electronic nose into fingerprint data then it is used by feature extraction process using the differential method. The results of feature extraction is used to process the back propagation neural network training to obtain optimal parameters. The parameters have been optimized is then tested on a random tofus. Based on test results, ANN-BP can identify samples with 100% accuracy rate so that the identification of a pure tofu and tofu formalin with electronic nose using back propagation neural network analysis has been successfully carried out.

Big Data: A Parallel Particle Swarm Optimization-Back-Propagation Neural Network Algorithm Based on MapReduce.

A back-propagation (BP) neural network can solve complicated random nonlinear mapping problems; therefore, it can be applied to a wide range of problems. However, as the sample size increases, the time required to train BP neural networks becomes lengthy. Moreover, the classification accuracy decreases as well. To improve the classification accuracy and runtime efficiency of the BP neural network algorithm, we proposed a parallel design and realization method for a particle swarm optimization (PSO)-optimized BP neural network based on MapReduce on the Hadoop platform using both the PSO algorithm and a parallel design. The PSO algorithm was used to optimize the BP neural network’s initial weights and thresholds and improve the accuracy of the classification algorithm. The MapReduce parallel programming model was utilized to achieve parallel processing of the BP algorithm, thereby solving the problems of hardware and communication overhead when the BP neural network addresses big data. Datasets on 5 different scales were constructed using the scene image library from the SUN Database. The classification accuracy of the parallel PSO-BP neural network algorithm is approximately 92%, and the system efficiency is approximately 0.85, which presents obvious advantages when processing big data. The algorithm proposed in this study demonstrated both higher classification accuracy and improved time efficiency, which represents a significant improvement obtained from applying parallel processing to an intelligent algorithm on big data.

Fractional vegetation cover estimation algorithm for Chinese GF-1 wide field view data

Abstract Wide field view (WFV) sensor on board the Chinese GF-1, the first satellite of the China High-resolution Earth Observation System, is acquiring multi-spectral data with decametric spatial resolution, high temporal resolution and wide coverage, which are valuable data sources for environment monitoring. The objective of this study is to develop a general and reliable fractional vegetation cover (FVC) estimation algorithm for GF-1 WFV data under various land surface conditions. The algorithm is expected to estimate FVC from GF-1 WFV reflectance data with spatial resolution of 16 m and temporal resolution of four dates. The proposed algorithm is based on training back propagation neural networks (NNs) using PROSPECT + SAIL radiative transfer model simulations for GF-1 WFV canopy reflectance and corresponding FVC values. Green, red and near-infrared bands' reflectances of GF-1 WFV data are the input variables of the NNs, as well as the corresponding FVC is the output variable, and finally 842,400 simulated samples covering various land surface conditions are used for training the NNs. A case study in Weichang County of China, having abundant land cover types, was conducted to validate the performance of the proposed FVC estimation algorithm for GF-1 WFV data. The validation results showed that the proposed algorithm worked effectively and generated reasonable FVC estimates with R 2 = 0.790 and root mean square error of 0.073 based on the field survey data. The proposed algorithm can be operated without prior knowledge on the land cover and has the potential for routine production of high quality FVC products using GF-1 WFV surface reflectance data.

Intraday stock price forecasting based on variational mode decomposition

This paper presents a hybrid predictive model for forecasting intraday stock prices. The proposed model hybridizes the variational mode decomposition (VMD) which is a new multiresolution technique with backpropagation neural network (BPNN). The VMD is used to decompose price series into a sum of variational modes (VM). The extracted VM are used to train BPNN. Besides, particle swarm optimization (PSO) is employed for BPNN initial weights optimization. Experimental results from a set of six stocks show the superiority of the hybrid VMD–PSO–BPNN predictive model over the baseline predictive model which is a PSO–BPNN model trained with past prices.

Fault diagnosis for a solar assisted heat pump system under incomplete data and expert knowledge

Fault diagnosis for a solar assisted heat pump (SAHP) system in the presence of incomplete data and expert knowledge is discussed in this article. A method for parameter learning of Bayesian networks (BNs) from incomplete data based on the back-propagation (BP) neural network and maximum likelihood estimation (MLE), which is called BP-MLE method, is presented. The BP neural network is utilized to impute the missing data and the complete data sets are addressed with MLE to obtain the parameters of BN. A method for parameter estimation under incomplete expert knowledge based on BP neural networks and fuzzy set theory is also presented, which is called BP-FS method. Similarly, the missing information is imputed by the trained BP neural network. Fuzzy set theory is employed to quantify the parameters of BN based on complete qualitative expert knowledge. The presented methods are applied to parameter learning of diagnostic BN for a SAHP system with incomplete simulation data and expert knowledge. The developed BN can perform fault diagnosis with complete or incomplete symptoms.

Forest Fire Smoke Detection Using Back-Propagation Neural Network Based on MODIS Data

Satellite remote sensing provides global observations of the Earth’s surface and provides useful information for monitoring smoke plumes emitted from forest fires. The aim of this study is to automatically separate smoke plumes from the background by analyzing the MODIS data. An identification algorithm was improved based on the spectral analysis among the smoke, cloud and underlying surface. In order to get satisfactory results, a multi-threshold method is used for extracting training sample sets to train back-propagation neural network (BPNN) classification for merging the smoke detection algorithm. The MODIS data from three forest fires were used to develop the algorithm and get parameter values. These fires occurred in (i) China on 16 October 2004, (ii) Northeast Asia on 29 April 2009 and (iii) Russia on 29 July 2010 in different seasons. Then, the data from four other fires were used to validate the algorithm. Results indicated that the algorithm captured both thick smoke and thin dispersed smoke over land, as well as the mixed pixels of smoke over the ocean. These results could provide valuable information concerning forest fire location, fire spreading and so on.

Fusion of Multi-Vision of Industrial Robot in MAS-Based Smart Space

The paper presents a fusion method of muti-vision of industrial robot in a smart space based on multi-agent system(MAS), the robotic multi-vision consists of top-view, side-view, front-view and hand-eye cameras, the moving hand-eye provide vision guidance and give the estimation of robot position, other three cameras are used for target recognition and positioning. Each camera is connected to an agent based on an image-processing computer that aims at analyzing image rapidly and satisfying the real-time requirement of data processing. As a learning strategy of robotic vision, a back-propagation neural network(BPNN) with 3-layer-architecture is first constructed for each agent and is independently trained as a classifier of target recognition using batch gradient descent method based on the region features extracted from the images of target samples(typical mechanical parts), and then the outputs of trained BPNNs in MAS-based smart space are fused with Dempster-Shafer evidence theory to form a final recognition decision, the experimental results of typical mechanical parts show that fusion of multi-vision can improve the robotic vision accuracy and MAS-based smart space will contribute to the parallel processing of immense image data in robotic multi-vision system.

An Optimization System Applied in Optical Design of LED Light Guide Plate

Most of the small and medium-sized backlight modules are applied to a small amount of various products. Since the time to design a product is rigid and the light guide plate (LGP) is the key component that affects the quality of backlight modules, photovoltaic industries can enhance the core competence if they are able to develop high standard light guide plates with fast and efficient methods. The study proposes an optical optimal design system of light guide plate (LGP). The optimization design is conducted in the longitudinal structure of the LGP incidence plane with three LED light sources. Taguchi method is also used in carrying out the design of experiment through the TracePro, optical analysis software, and the experiment data which are employed as the back-propagation neural network (BPNN) training and testing samples and, then, created an optical quality predictor of the longitudinal structure. BPNN can predict the impact of incidence plane luminance versus the different constructed parameters. Finally, the optical quality predictor can effectively generate the optimal parameters settings combined with genetic algorithm (GA). The simulation results show that the proposed system can improve the non-uniformity problem of the incidence plane but also make it easier to design the longitudinal structure of the incidence plane.

A Hybrid of Back Propagation Neural Network and Genetic Algorithm for Modeling and Simulation of Standard Involute Spur Gear

A method is presented for gear modeling and simulation based on a hybrid of back propagation (BP) neural network and genetic algorithm (GA). Generally, the method of mechanical design includes building up 3D models and gets them checked in corresponding software, which is complicated and takes much time. This proposed method offers a quick way for designers, which combines finite element analysis (FEA) with a hybrid of BP neural network and an intelligence global optimization algorithm, i.e. genetic algorithm (GA). ANSYS APDL is applied to build up the FEA database, which offers data for study and training of BP neural network. Then, the GA is applied to optimize the process parameters that would result in optimal solution of the BP neural network goals. The case study demonstrates that the proposed method can get accurate results in a short time.

Research on Short-Term Wind Farm Output Power Prediction Model Based on Meteorological Data Collected by WSN

The prediction of wind farm output power is considered as an effective way to increase the wind power capacity and improve the safety and economy of power system. It is one of the hot research topics on wind power. The wind farm output power is related to many factors such as wind speed, temperature, etc., which is difficult to be described by some mathematical expression. In this paper, Back Propagation (BP) neural network algorithm and genetic algorithm (GA) are combined to establish the prediction model of the short-term wind farm output power based on meteorological data collected by Wireless Sensor Network (WSN). The Meteorological data is used to determine the input variables of the BP neural network. Meanwhile, the GA is used to adjust the value of BP's connection weight and threshold dynamically. Then the trained BP neural network is used to predict the wind power. The experiment results show that our method has better prediction capability compared with that using BP neural network alone or using wind power formulas.

The Defect Classification of TFT-LCD Array Photolithography Process via Using Back-Propagation Neural Network

This research is conducted mainly by using the Auto Optical Inspection (AOI) in the fifth generation TFT-LCD factory. In the development of detect-classification system, we designed the back-propagation neural network which combined with Visual Basic as the interface and MATLAB as an image-processing tool. The system is able to determine and display the detected results. The defect classification mainly designed to detect and classify the following defects: the second layer of the photo resist residue (AS-Residue), the second layer of large-area photo resist residue (AS-BPADJ), and the third layer of photo resist residue (M2-residue) in the Array Photolithography Process. Finally, the result is shown the fact that without the complicated processing procedures, the four defects in the TFT-LCD Array Photo Process can be precisely and quickly classified by imaging processing and back-propagation neural network training. As result, it is feasible to reduce the costs and the risk of human judgments.

Training Back-Propagation Neural Network for Target Localization Using Improved Particle Swarm Optimization

Target localization technology has been intensively studied and broadly applied in many fields. This paper presents one improved particle swarm optimization technique in training a back-propagation neural network for position estimation in target localization. The proposed scheme combines particle swarm optimization (PSO), back-propagation neural network (BP), adaptive inertia weight and hybrid mutation, called IPSO-BP. To verify the proposed IPSO-BP approach, comparisons between the PSO-based BP approach (PSO-BP) and general back-propagation neural network (BP) are made. The computational results show that the proposed IPSO-BP approach exhibits much better performance in the training process and better prediction ability in the validation process than those using the other two base line approaches.

A Study of Various Training Algorithms on Neural Network for Angle based Triangular Problem

This paper examines the study of various feed forward backpropagation neural network training algorithms and performance of different radial basis function neural network for angle based triangular problem. The training algorithms in feed forward back-propagation neural network comprise of Scale Gradient Conjugate Back-Propagation (BP), Conjugate Gradient BP through Polak-Riebre updates, Conjugate Gradient BP through Fletcher-Reeves updates, One Secant BP and Resilent BP. The final result of each training algorithm for angle based triangular problem will also be discussed and compared. General Terms Artificial Neural Network (ANN), Feed Forward Backpropogation (FFB), Training Algorithm.

Concrete Mix Design Using Artificial Neural Network

Concrete mix design is a process based on sound technical principles for proportioning of ingredients in right quantities. This paper demonstrates the applicability of Artificial Neural Networks (ANN) Model for approximate proportioning of concrete mixes. For ANN a trained back propagation neural network is integrated in the model to learn experimental data pertaining to predict 7, 14 and 28-day compressive strength which have been loaded into a model, containing 55 concrete mixtures. The ANN model proposed is based on 5 input parameters such as cement, sand, coarse aggregate, and water and fineness modulus. The proposed concrete mix proportion design is expected to reduce the number of trials in laboratory as well as field, saves cost of material as well as labor and also saves time as it provides higher accuracy. The concrete designed is expected to have higher durability and hence is economical.

A novel particle filter implementation for a multiple-vehicle detection and tracking system using tail light segmentation

This paper proposes a vision-based multiple vehicle automatic detection and tracking system which can be applied in different environments. To detect vehicles, tail light position is utilized for fast vehicle candidate localization. A back propagation neural network (BPNN) trained by a Gabor feature set is used. BPNN verifies vehicle candidates and ensures detection system robustness. In the vehicle tracking step, to overcome multiple vehicle tracking challenges, partial vehicle occlusion and temporarily missing vehicle problems, this paper propose a novel method implementing a particle filter. The color probability distribution function (CPDF) of detected vehicles is used twice in the vehicle tracking sub-system. Firstly, CPDF is adopted to seek potential target vehicle locations; secondly, CPDF is used to measure the similarity of each particle for target vehicle position estimation. Because of various illuminations or target vehicle distances, the same vehicle will generate different CPDFs; the initial CPDF cannot guarantee long-term different scale vehicle tracking. To overcome these problems, an accurate tracking result, which is chosen by a trained BPNN, is used to update target vehicle CPDF. In our experiments, the proposed algorithm showed 84% accuracy in vehicle detection. Videos collected from highways, urban roads and campuses are tested in our system. The system performance makes it appropriate for real applications.

Optimization Design for the Longitudinal Structure of LED Light Guide Plate Using Soft Computing

The study proposes an optical optimal design system of light guide plate (LGP). The optimization design is conducted in the longitudinal structure of the LGP incidence plane with 3-piece LED light source. Taguchi method is also used in carrying out the design of experiment through the TracePro, optical analysis software, and the experiment data were employedd as the back-propagation neural network (BPNN) training and testing samples, and then created an optical quality predictor of the longitudinal structure; BPNN can predict the impact of incidence plane luminance versus the different constructed parameters. Finally, the optical quality predictor can effectively generate the optimal parameters settings combined with genetic algorithm (GA). The simulation results show that the proposed system improves the non-uniformity problem of the incidence plane but also makes it easier to design the longitudinal structure of the incidence plane.

Prediction of Cutting Tool Life Based on ACO-BP Meural Network

For predicting the tool life combine the ant colony optimization(ACO) with the back propagation (BP) neural networks, use the the ACO to train BP neural network, build the prediction model based ACO-BP neural network. Some disadvantages are overcame in the BP algorithm, such as the low convergence speed, easily falling into local minimum point and weak global search capablity in the prediction process. Satisfies the requirement of global search capability and the robustness of the model. The experiment results show the prediction model has high precision in predicting the tool life. By the prediction model can provide a reasonable basis for planing production schedule and cutting tool requirement, calculating the cost, selecting the machining parameters,etc.

An application of back propagation neural network for the steel stress detection based on Barkhausen noise theory

A new method for stress testing based on the theory of Barkhausen noise has been introduced using changing feature values for monitoring stress and temperature. However, changes in temperature not only have an effect on the stress but also the MBN signal itself. In order to get the accurate stress value and eliminate the temperature effect, we proposed a data processing method for stress testing based on MBN. The study found that within the steel elastic range, the Barkhausen noise feature values, including mean value, RMS value, ring numbers, peak value and the ratio of envelope peak and full peak width at half of maximum amplitude decrease with increasing temperature, there is a fixed monotonic relationship which provides a theoretical basis for building the back propagation (BP) neural network model, with stress as the output value and temperature, mean value, RMS value, ring numbers, peak value and the ratio of peak and full width of half maximum as the input values. The MATLAB 7.8.0 neural network toolbox was used to model and simulate the neural network and samples used to validate the trained BP neural network. The results showed that the network had a high degree of accuracy and generalization ability, to get the values of stress.

Automatic Modulation Classifier Using Artificial Neural Network Trained by PSO Algorithm

The back propagation (BP) neural networks have been commonly used for automatic modulation recognition since the late 1990s. However, the back propagation algorithm easily falls into local minimum and the network learning is sensitive to initial weight values which usually determined by experience. The particle swarm optimization (PSO) algorithm is a global heuristic searching technology. By combing these two techniques, this paper presents a novel automatic modulation classifier. By training BP neural network with PSO optimization, this classifier can identify seven digital signals with six input features. The experiment results show that this recognizer achieves about 92% recognition performance at the SNR of 0dB.