Back Propagation Neutral Network Research Articles

Improved Prediction of Michaelis Constants in CYP450-Mediated Reactions by Resilient Back Propagation Algorithm

In drug metabolism reactions, it has become increasingly important to measure Michaelis constants (Km), which are used for a variety of purposes, including identification of enzymes involved in drug metabolism, prediction of drug-drug interactions, etc. Cytochrome P450s (CYPs) comprise a super family of major human enzymes responsible for drug metabolism. Hence, computational prediction of Km in CYP-mediated reactions facilitates drug development in an efficient and economical way. In this study, we firstly constructed a large dataset of ten CYP isoforms associated with 169 binding substrates, and 210 experimental Km values in CYP-mediated reactions. To predict Km of substrates metabolized by various CYP isoforms, we developed a general prediction model by using resilient back-propagation neutral network algorithm, based on the structural and physicochemical properties of the substrates and the metabolic specificity of the enzymes. The predictive Km values achieve a squared cross-validation correlation coefficients (Q2) of 0.73 with the experimental values, which is better than that of the existing models. Moreover, our model can predict Kmvalues of the compounds metabolized by a wide range of CYP isoforms. This tool will be useful in large-scale drug screening studies for CYP enzymes and helpful in the drug design and development.

New pattern recognition system in the e-nose for Chinese spirit identification**Project supported by the National High Technology Research and Development Program of China (Grant No. 2013AA030901) and the Fundamental Research Funds for the Central Universities, China (Grant No. FRF-TP-14-120A2).

This paper presents a new pattern recognition system for Chinese spirit identification by using the polymer quartz piezoelectric crystal sensor based e-nose. The sensors are designed based on quartz crystal microbalance (QCM) principle, and they could capture different vibration frequency signal values for Chinese spirit identification. For each sensor in an 8-channel sensor array, seven characteristic values of the original vibration frequency signal values, i.e., average value (A), root-mean-square value (RMS), shape factor value (Sf), crest factor value (Cf), impulse factor value (If), clearance factor value (CLf), kurtosis factor value (Kv) are first extracted. Then the dimension of the characteristic values is reduced by the principle components analysis (PCA) method. Finally the back propagation (BP) neutral network algorithm is used to recognize Chinese spirits. The experimental results show that the recognition rate of six kinds of Chinese spirits is 93.33% and our proposed new pattern recognition system can identify Chinese spirits effectively.

Remaining Useful Life Prediction of Rolling Bearings Using an Enhanced Particle Filter

This paper presents an enhanced particle filter (PF) approach for predicting remaining useful life (RUL) of rolling bearings. In the presented approach, particles in each recursive step are used to determine an alterable importance density function and the backpropagation neutral network is utilized to improve the particle diversity before resampling. Based on the enhanced PF, the framework of online rolling bearing RUL prediction is designed and a multiorder autoregressive model is used to construct the dynamic model for PF. Case studies performed on a simulation example and two test-to-failure experiments indicate that the presented approach can accurately predict the RUL of rolling bearings and it can achieve better performance than the traditional PF-based approach and commonly used support vector regression approach.

Mapping dustfall distribution in urban areas using remote sensing and ground spectral data

The aim of this study was to utilize remote sensing and ground-based spectral data to assess dustfall distribution in urban areas. The ground-based spectral data denoted that dust has a significant impact on spectral features. Dusty leaves have an obviously lower reflectance than clean leaves in the near-infrared bands (780–1,300nm). The correlation analysis between dustfall weight and spectral reflectance showed that spectroscopy in the 350–2,500-nm region produced useful dust information and could assist in dust weight estimation. A back propagation (BP) neutral network model was generated using spectral response functions and integrated remote sensing data to assess dustfall weight in the city of Beijing. Compared with actual dustfall weight, validation of the results showed a satisfactory accuracy with a lower root mean square error (RMSE) of 3.6g/m2. The derived dustfall distribution in Beijing indicated that dustfall was easily accumulated and increased in the south of the city. In addition, our results showed that construction sites and low-rise buildings with inappropriate land use were two main sources of dust pollution. This study offers a low-cost and effective method for investigating detailed dustfall in an urban environment. Environmental authorities may use this method for deriving dustfall distribution maps and pinpointing the sources of pollutants in urban areas.

Diesel Engine Inlet and Exhaust System Fault Detection Based on PSO-BP NN

There is high failure rate in diesel engine inlet and exhaust system, so the inlet and exhaust system condition monitoring and fault diagnosis is of great significance.To overcome drawbacks of pure BP algorithm, a heuristic algorithm is adopted to give a transition from particle swarm search to gradient descending search to set up a Particle Swarm Optimized Back Propagation (PSO-BP) Neutral Network (NN). Then vibration signals have been measured from a diesel cylinder head by simulating two faults:the gas leak and abnormal lash; The diesel engine vibration frequency energy signals after wavelet packet decomposition are taken as the input feature vectors of NN. Finally, we carry out fault detection experiment by the PSO-BP NN model to validate the method. Experiment results show that the PSO-BP NN has more fast convergence speed and higher diagnosis accuracy than BP NN, and it provides a new fault detection method for engine inlet and exhaust system.

Estimation of Protein Content in Plant Leaves using Spectral Reflectance: A Case Study in Euonymus japonica

Reflectance spectroscopy has been widely applied in the field of environmental studies. In this study, a low-cost, rapid, and nondestructive method using spectral reflectance was explored to evaluate protein concentrations in plant leaves of Euonymus japonica. Proteins in leaf samples were extracted and separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis, and five specific protein bands of interest were identified and quantified. Correlation analysis indicated that spectral reflectance had significant relationships with ribulose bisphosphate carboxylase a (r = −0.43) and chlorophyll a-b binding protein (r = 0.53). A linear regression model and a quadratic regression model were formulated to directly and rapidly estimate the concentration of these two proteins with R 2 = 0.61 and 0.7, respectively. To more accurately estimate the concentration of proteins, a precise inversion was established by a back propagation neutral network model using plant spectral absorption and position parameters, and the R 2 values for proteins ribulose bisphosphate carboxylase, chlorophyll a-b binding protein, oxygen evolving enhancer protein, and ATP synthase subunit beta were 0.90, 0.91, 0.91, and 0.93, respectively. The models established in this study were shown to be useful tools for studies of plant biochemical components and health under different environmental conditions.

Predication Method for China's Civil Aviation Fuel Consumption

With the China’s civil aviation industry gradual market-oriented and the rapid development of China’s economy, China’s civil aviation transportation fuel consumption has grown significantly in nearly past three decades. Therefore, it’s a very important strategic significance of the prediction of China’s civil aviation transportation fuel consumption. In this paper, gray system and neural network approach, combined with China’s civil aviation industry 1980-2010 total traffic volume of the data, we establish gray system GM (1,1) model and BP neural network model for civil aviation transport volume. Training and simulation of the back propagation neutral network model and the gray system GM(1,1) used MATLAB. BP neural network modeling takes into account in three factors: the number of aircraft aviation industry, flight hours and total turnover. The fitting precision of the gray system GM(1,1) model is 64.2% while the fitting precision of the back propagation neutral network model is 90.16%. Thus, the back propagation neutral network model is better for estimating Civil aviation fuel consumption.

Design of Tiltrotor Flight Control System in Conversion Mode Using Improved Neutral Network PID

Considering the high-order, time-varying and nonlinear characteristics of the tiltrotor aircraft in conversion mode, the control requirements and the specific conversion scheme are presented first. An improved back propagation (BP) neural network PID control algorithm is then proposed, utilizing the strong nonlinear mapping characteristics of neural network. In the proposed algorithm, an inertia term is added to the gradient-descent-based weight adjustment formula of the conventional BP neutral network in order to increase the convergence rate and obtain satisfactory control effects. Finally simulation results indicate the effectiveness of the proposed algorihm.

Adaptive Chaotic Prediction Algorithm of RBF Neural Network Filtering Model Based on Phase Space Reconstruction

With the analysis of the technology of phase space reconstruction, a modeling and forecasting technique based on the Radial Basis Function (RBF) neural network for chaotic time series is presented in this paper. The predictive model of chaotic time series is established with the adaptive RBF neural networks and the steps of the chaotic learning algorithm with adaptive RBF neural networks are expressed. The network system can enhance the stabilization and associative memory of chaotic dynamics and generalization ability of predictive model even by imperfect and variation inputs during the learning and prediction process by selecting the suitable nonlinear feedback term. The dynamics of network become chaotic one in the weight space. Simulation experiments of chaotic time series produced by Lorenz equation are proceeded by a RBF neural network .The experimental and simulating results indicated that the forecast method of the adaptive RBF neutral network compared with the forecast method of back propagation (BP) neutral network based on the chaotic learning algorithm has faster learning capacity and higher accuracy of forecast．The method provides a new way for the chaotic time series prediction.

Control Model of Vehicle Turning Stability Based on Neural Network

In order to research the turning stability of a semi-active suspension vehicle, the suspension damping BP (Back Propagation) neutral network PID (Proportion–Integral–Differential) control technique was proposed. The ideal and actual yaw rate error is taken as the control targets to carry on the turning stability closed-loop control. The vehicle yaw rate is controlled to analyze the impacts of the suspension damping change on the wheel vertical load and roll. The MATLAB software was applied to carry out simulation according to the single lane steering and step turning working conditions. The results show that the suspension damping can be controlled to control the vehicle yaw and cornering motions in the process of the vehicle turning on the high adhesion road. When the front suspension damping is increased and the rear suspension damping is decreased synchronously, the load transfer of the left and right wheels of the front and rear axles is obviously decreased, thus being able to effectively restrain the over-steering behavior of the vehicle.

Aero-engine blade fatigue analysis based on nonlinear continuum damage model using neural networks

Fatigue life and reliability of aero-engine blade are always of important significance to flight safety. The establishment of damage model is one of the key factors in blade fatigue research. Conventional linear Miner’s sum method is not suitable for aero-engine because of its low accuracy. A back propagation neutral network (BPNN) based on the combination of Levenberg-Marquardt (LM) and finite element method (FEM) is used to describe process of nonlinear damage accumulation behavior in material and predict fatigue life of the blade. Fatigue tests of standard specimen made from TC4 are carried out to obtain material fatigue parameters and S-N curve. A nonlinear continuum damage model (CDM), based on the BPNN with one hidden layer and ten neurons, is built to investigate the nonlinear damage accumulation behavior, in which the results from the tests are used as training set. Comparing with linear models and previous nonlinear models, BPNN has the lowest calculation error in full load range. It has significant accuracy when the load is below 500 MPa. Especially, when the load is 350 MPa, the calculation error of the BPNN is only 0.4%. The accurate model of the blade is built by using 3D coordinate measurement technology. The loading cycle in fatigue analysis is defined from takeoff to cruise in 10 min, and the load history is obtained from finite element analysis (FEA). Then the fatigue life of the compressor blade is predicted by using the BPNN model. The final fatigue life of the aero-engine blade is 6.55×104 cycles (10 916 h) based on the BPNN model, which is effective for the virtual design of aero-engine blade.

Comparison of three models of forest biomass estimation

Aims Quantitative estimation of forest biomass is significant to studies of global carbon storage and carbon cy-cle.Our objective is to develop models to estimate forest biomass accurately.Methods Multi-stepwise regression model,traditional back propagation (BP) neutral network model and BP neutral network model based on Gaussian error function (Erf-BP) were developed to estimate forest biomass in Changbai Mountain of Heilongjiang,China according to TM imagery and 133 plots of forest inventory data.There were 71 dependent variables of geoscience and remote sensing.Important findings The precisions and root mean square errors of multi-stepwise regression model,traditional BP neutral network model and Erf-BP were 75%,26.87 t?m-2;80.92%,21.44 t?m-2 and 82.22%,20.83 t?m-2,re-spectively.Therefore,the relations between forest biomass and various factors can be better modeled and de-scribed by the improved Erf-BP.

The Study of Combination Model Forecasting of Dynamic Traffic Volumes Based on MATLAB

The most important part of ITS (Intelligent Technology System) is the forecasting of dynamic traffic information reporting in a timely manner. Based on the results of past empirical studies, this article is presenting and analyzing two models based on Grey Theory and the BP (Back propagation) Neural Network. In accordance with the advantages and disadvantages of these two models, we established a new combination model. This article also presents the calculation of dynamic traffic stream volumes based on the application of two models (Grey Theory and BP Neutral Network).The results predicted an increase in the overall prediction accuracy of road traffic volume.

Analysis of a Residential Building Energy Consumption Demand Model

In order to estimate the energy consumption demand of residential buildings, this paper first discusses the status and shortcomings of current domestic energy consumption models. Then it proposes and develops a residential building energy consumption demand model based on a back propagation (BP) neural network model. After that, taking residential buildings in Chongqing (P.R. China) as an example, 16 energy consumption indicators are introduced as characteristics of the residential buildings in Chongqing. The index system of the BP neutral network prediction model is established and the multi-factorial BP neural network prediction model of Chongqing residential building energy consumption is developed using the Cshap language, based on the SQL server 2005 platform. The results obtained by applying the model in Chongqing are in good agreement with actual ones. In addition, the model provides corresponding approximate data by taking into account the potential energy structure adjustments and relevant energy policy regulations.

Cloud security defence to protect cloud computing against HTTP-DoS and XML-DoS attacks

Cloud computing is still in its infancy in regards to its software as services (SAS), web services, utility computing and platform as services (PAS). All of these have remained individualized systems that you still need to plug into, even though these systems are heading towards full integration. One of the most serious threats to cloud computing itself comes from HTTP Denial of Service or XML-Based Denial of Service attacks. These types of attacks are simple and easy to implement by the attacker, but to security experts they are twice as difficult to stop. In this paper, we recreate some of the current attacks that attackers may initiate as HTTP and XML. We also offer a solution to traceback through our Cloud TraceBack (CTB) to find the source of these attacks, and introduce the use of a back propagation neutral network, called Cloud Protector, which was trained to detect and filter such attack traffic. Our results show that we were able to detect and filter most of the attack messages and were able to identify the source of the attack within a short period of time.

Bonding Interface Imaging and Shear Strength Prediction by Ultrasound

The propagation of a longitudinal ultrasonic wave normally incident upon an adhesively bonded structure is studied. The structure consists of adherend and adhesive layers with finite thickness. Interfaces between adherend and adhesive are regarded as distributed springs. Theoretical and experimental results show that resonant frequencies of the bonded structure vary sensitively with the interface stiffness constants and adhesive thickness, and these interface characteristics are inversed by the simulation annealing (SA) method. Furthermore, the distribution image of interface stiffnesses is compared with the state of fracture interface, and quantitative prediction of shear strength is achieved based on the distribution of interface stiffnesses and adhesive thicknesses by using a back-propagation neutral network. The average relative error of the shear strength from prediction to real value is 10.7%.

Neurofuzzy Genetic System for Selection of Construction Project Managers

Choosing a project manager for a construction project—particularly, large projects—is a critical project decision. The selection process involves different criteria and should be in accordance with company policies and project specifications. Traditionally, potential candidates are interviewed and the most qualified are selected in compliance with company priorities and project conditions. Precise computing models that could take various candidates' information into consideration and then pinpoint the most qualified person with a high degree of accuracy would be beneficial. On the basis of the opinions of experienced construction company managers, this paper, through presenting a fuzzy system, identifies the important criteria in selecting a project manager. The proposed fuzzy system is based on IF-THEN rules; a genetic algorithm improves the overall accuracy as well as the functions used by the fuzzy system to make initial estimates of the cluster centers for fuzzy c-means clustering. Moreover, a back-propagation neutral network method was used to train the system. The optimal measures of the inference parameters were identified by calculating the system's output error and propagating this error within the system. After specifying the system parameters, the membership function parameters—which by means of clustering and projection were approximated—were tuned with the genetic algorithm. Results from this system in selecting project managers show its high capability in making high-quality personnel predictions.

Calculation of Heat of Formation for Tensile Cyclic Molecules in Condensed Phase Using Electrostatic Potential and QSPR Schemes

AbstractPDDG/PM3 method was evaluated as a suitable semiempirical calculation for heat of formation of tensile cyclic organic molecules in gas phase and used in subsequent researches. In order to compute the heat of formation of these compounds in condensed phase, electrostatic potential scheme was performed to obtain the area of the Bader surface at B3LYP/6‐311G∗︁∗︁ level. The root mean square (rms) of the calculated condensed phase heat of formation for 58 molecules from corresponding experimental values is 4.81 kcal/mol. Six QSPR schemes for predicting solid phase heat of formation of 38 model molecules with physically meaningful descriptors from genetic algorithm selection were established. These selected descriptors were recalculated by DFT B3LYP/6‐311G∗︁∗︁ and MP2/6‐311G∗︁∗︁ methods, which could then give similar and better results. The QSPR schemes adopted back‐propagation neutral network (BPNN) and support vector machine (SVM) as analysis tools. It was suggested that selection of descriptors with strong physical significance in QSPR calculation should be important.

Optimization of the Dyeing Process and Prediction of Quality Characteristics on Elastic Fiber Blending Fabrics

This study aims to find the optimal conditions for dyeing polyester (PET) and Lycra®-blended fabric and predict the quality characteristics, where PET and Lycra®-blended fabric were taken as raw material with dispersed dyes using a one-bath two-section dyeing method, characterizing the color strength of gray fabric. Adopting the Taguchi method for parameter design, machine working temperature, dyeing time, dye concentration, and bath ratio, which have an influence on dyeing, were chosen as control factors to conduct experiments using orthogonal arrays, and analysis of variance was incorporated to determine optimal processing conditions, significant factors, and percent contribution. With the smaller-the-better characteristic for color strength of gray fabric as the target characteristic, calculations were conducted to confirm the reproducibility of the experiment. We found from experimental results that color strength for gray fabric dyed under optimal conditions was closer to the target value. In addition, we constructed a prediction system based on the factors significantly influencing dyeing performance by a integrating genetic algorithm (GA) with a back propagation neutral network (BPNN), so as to find optimal connection weight values. Hence, learning algorithm efficiency was enhanced with decreased dependency on initial conditions, together with a more robust learning algorithm, so as to predict color strength for gray fabric. Moreover, improvements of BPNN, which is most widely used at present, will be influenced by choice of initial learning parameters more often than not, hence there is a disadvantage in that it is probable that only local optimal solutions are found instead of global optimal solutions.

Drift counteraction for an electronic nose

Abstract Most gas sensors do not give stable responses over a long period of time. This drift causes electronic noses to classify gases wrongly after some period of time. In this paper this problem has been studied, and two methods for counteracting drift are proposed. One is based on reference patterns that are updated when new measurements are made, and the other uses models known from system identification theory. The two methods are suited for different experimental situations, and both give a better result than a back-propagation neutral network classifier when they are used under the right conditions.