Local Minimum Value Research Articles

Study on Water Production Function of Rice Based on RAGA-BP Neural Network Model

The water production function model of rice based on RAGA-BP neural network model is established by taking the advantages of RAGA-BP neural network model (high ability in solving nonlinear problem, high rate of convergence, etc.). Compared with traditional Jensen model, this method is featured by high flexibility and intelligence, avoiding the inconvenience caused by pre-establishing mathematical expression. Compared with classical BP neural network model, this method can overcome the problem in local minimum value and low rate of convergence. The deficient irrigation experiment of rice in 2015 shows that RAGA-BP neural network model has high adaptability during the process of fitting for solution of water production function of rice; the accuracy and speed are high, which provides new effective method for calculation of water production function of rice.

Generalized incremental frequency method for topological designof continuum structures for minimum dynamic compliance subject to forced vibration at a prescribed low or high value of the excitation frequency

This paper deals with topological design optimization of elastic, continuum structures without damping that are subjected to time-harmonic, design-independent external dynamic loading with prescribed excitation frequency, amplitude and spatial distribution. The admissible design domain, the boundary conditions, and the available amount(s) of material(s) for single- or bi-material structures are given. An important objective of such a design problem is often to drive the resonance frequencies of the structure as far away as possible from the given excitation frequency in order to avoid resonance and to reduce the vibration level of the structure. In the present paper, the excitation frequency is defined to be ‘low’ for positive values up to and including the fundamental resonance frequency of the structure, and to be ‘high’ for values beyond that. Our paper shows that it may be very important to consider different design paths in problems of minimum dynamic compliance in order to obtain desirable solutions for prescribed excitation frequencies, and a so-called ‘incremental frequency technique’ (IF technique) is applied for this. Subsequently, the IF technique is integrated into an extended, systematic method named as the ‘generalized incremental frequency method’ (GIF method) that is developed for gradient based dynamic compliance minimization for not only ‘low’, but also ‘high’ excitation frequencies of the structure. The GIF method performs search for and determination of the solution to the minimum dynamic compliance design problem, but this is subject to the complexity that problems with prescribed high excitation frequencies exhibit disjointed design sub-spaces. Each of these sub-spaces is associated with a local minimum value of the dynamic compliance, so in general the ‘global optimum solution’ will have to be selected as the ‘best’ solution from among a number of local candidate solutions. Illustrative examples of application of the IF technique and the GIF method are presented in the paper.

Penerapan Fitur Warna Untuk Identifikasi Plasmodium Falciparum pada Sediaan Apus Darah Menggunakan MK-Means dan Jaringan Backpropagation

<p><strong>Abstract - This research proposed a system to identify Plasmodium falciparum on blood smear using the neural network backpropagation. Modified K-Means (MK-Means) is used to separate between the object with the background image because that method was able to equalize the value of fitness at all Center cluster so there is no dead center and can also cope with the local minimum value. The extraction of the features used in this study consists of color features i.e. calculation of the mean, standard deviation, skewness, curtosis and entropy of co-occurent matrix with the purpose to get the values of all the trait value image, obtained are then used to train a neural network with the backpropagation training algorithm. Method of backpropagation networks capable of acquiring knowledge even though there is no certainty, able to perform a generalization and extraction of a specific data pattern.</strong></p><p><strong> </strong><strong>The image of the preparations blood smear are classified using the method of neural network Backpropagation. The test results obtained from Tropozoit with the accuracy 100%, scizon 80% and gametocytes 80%. Identification is then obtained outcomes the introduction with an average accuracy of 86,66%.</strong></p>

An improved particle swarm optimization algorithm used for BP neural network and multimedia course-ware evaluation

The original BP neural network has some disadvantages, such as slow convergence speed, low precision, which is easy to fall into local minimum value. So this paper proposes an improved particle swarm optimization (PSO) algorithm to optimize BP neural network. In this new algorithm, PSO uses improved adaptive acceleration factor and improved adaptive inertia weight to improve the initial weight value and threshold value of BP neural network. And we give the detailed improved process. At the end, simulation results show that the new algorithm can improve convergence rate and precision of prediction of BP neural network, which reduces the error of prediction. At the end, we use multimedia evaluation model to verify the new method’s performance.

Effect of vortex shedding in unsteady aerodynamic forces for a low Reynolds number stationary wing at low angle of attack

This study elucidates the relation between wake vortex shedding and aerodynamic force fluctuations for a low Reynolds number wing from time resolved particle image velocimetry (TR-PIV) experimental measurements. The results reveal a periodic lift and drag variation within the shedding cycle and resolve the frequencies of those fluctuations from a proper orthogonal decomposition (POD) and power spectral density (PSD) analysis. To show the effect of vortex shedding on the body force fluctuations, the evolution of instantaneous aerodynamic forces is compared to the pressure field of the fluid flow and to the vortical structures in the wake of the airfoil. A six step model describing the vortex-force relation is proposed. It shows that changes in lift such as maximum lift and minimum lift are associated with the detachment of a vortex. It also shows that the minimum or local minimum drag value is obtained at the onset formation of a vortex on the airfoil wake. Similarly, the maximum or local maximum drag is obtained at the onset formation of the saddle on the airfoil wake. The model further explains the asymmetry observed in the unsteady drag force evolution. The model can be used to optimize flow control and fluid-structure interaction applications.

Simple automatic strategy for background drift correction in chromatographic data analysis

Chromatographic background drift correction, which influences peak detection and time shift alignment results, is a critical stage in chromatographic data analysis. In this study, an automatic background drift correction methodology was developed. Local minimum values in a chromatogram were initially detected and organized as a new baseline vector. Iterative optimization was then employed to recognize outliers, which belong to the chromatographic peaks, in this vector, and update the outliers in the baseline until convergence. The optimized baseline vector was finally expanded into the original chromatogram, and linear interpolation was employed to estimate background drift in the chromatogram. The principle underlying the proposed method was confirmed using a complex gas chromatographic dataset. Finally, the proposed approach was applied to eliminate background drift in liquid chromatography quadrupole time-of-flight samples used in the metabolic study of Escherichia coli samples. The proposed method was comparable with three classical techniques: morphological weighted penalized least squares, moving window minimum value strategy and background drift correction by orthogonal subspace projection. The proposed method allows almost automatic implementation of background drift correction, which is convenient for practical use.

Simulation Study on Optimizing Neural Network in Short-Term Electric Load Prediction

It researches the short-term electric load prediction and short-term electric load has the characteristics of time-varying, uncertainty, nonlinearity, etc., so the traditional linear prediction method cannot correctly describe the changing rule of the short-term electric load prediction, and neural network has the deficiencies including local minimum value of neural network, over-fitting and weak generalization ability, and the prediction accuracy is lower. In order to improve the accuracy of the short-term electric load prediction, this paper proposes a short-term electric load prediction model (IQPSOBPNN) based on optimizing BP neural network. Firstly, it improves Quantum Particle Swarm Optimization to optimize the BP neural network parameters, and then adopts the optimized BP neural network to conduct modeling for the nonlinear change law of the short-term electric load prediction. Finally, it takes simulation test for the model performance. The simulation result shows that IPQPSO solves the problems of the BP neural network, and improve the prediction accuracy of the short-term electric load and reduce the prediction error.

Dynamics of the entropic insertion of a large sphere into a cylindrical vessel.

Insertion of a solute into a vessel comprising biopolymers is a fundamental function in a biological system. The entropy originating from the translational displacement of solvent particles plays an essential role in the insertion. Here we study the dynamics of entropic insertion of a large spherical solute into a cylindrical vessel. The solute and the vessel are immersed in small spheres forming the solvent. We develop a theoretical method formulated using the Fokker-Planck equation. The spatial distribution of solute-vessel entropic potential, which is calculated by the three-dimensional integral equation theory combined with rigid-body models, serves as input data. The key quantity analyzed is the density of the probability of finding the solute at any position at any time. It is found that the solute is inserted along the central axis of the vessel cavity and trapped at a position where the entropic potential takes a local minimum value. The solute keeps being trapped without touching the vessel inner surface. In a significantly long time τ, the solute transfers to the position in contact with the vessel bottom possessing the global potential minimum along the central axis. As the solute size increases, τ becomes remarkably longer. We also discuss the relevance of our result to the functional expression of a chaperonin/cochaperonin in the assistance of protein folding.

The role of extreme orbits in the global organization of periodic regions in parameter space for one dimensional maps

We show that extreme orbits, trajectories that connect local maximum and minimum values of one dimensional maps, play a major role in the parameter space of dissipative systems dictating the organization for the windows of periodicity, hence producing sets of shrimp-like structures. Here we solve three fundamental problems regarding the distribution of these sets and give: (i) their precise localization in the parameter space, even for sets of very high periods; (ii) their local and global distributions along cascades; and (iii) the association of these cascades to complicate sets of periodicity. The extreme orbits are proved to be a powerful indicator to investigate the organization of windows of periodicity in parameter planes. As applications of the theory, we obtain some results for the circle map and perturbed logistic map. The formalism presented here can be extended to many other different nonlinear and dissipative systems.

Low-Loss and Compact Silicon Rib Waveguide Bends

Waveguide bends support intrinsically leaky propagation modes due to unavoidable radiation losses. It is known that the losses of deep-etched/strip waveguide bends increase inevitably for decreasing radius. Here, we theoretically and experimentally demonstrate that this result is not directly applicable to shallow-etched/rib waveguide bends. Indeed, we show that the total losses caused by the bends reach a local minimum value for a certain range of compact radii and rib waveguide dimensions. Specifically, we predicted the minimum intrinsic losses $\mu \text{m}$ bend radii in a 220 nm-thick and 400 nm-wide silicon rib waveguide with 70 nm etching depth. This unexpected outcome, confirmed by experimental evidence, is due to the opposite evolution of radiation (bending) losses and losses caused by the coupling to lateral slab modes (slab leakage) as a function of the bend radius, hence creating an optimum loss region. This result may have important implications for the design of compact and low-loss silicon nanophotonic devices.

A Cuckoo Search-Support Vector Machine Model for Predicting Dynamic Measurement Errors of Sensors

Sensors play a very important role in the Internet of Things. Error correction is of great significance to achieve sensor precision. Currently, accurately predicting the future dynamic measurement error is an effective way to improve sensor precision. Aiming to solve the problem of low model accuracy in traditional dynamic measurement error prediction, this paper employs the support vector machine (SVM) to predict the dynamic measurement error of sensors. However, the performance of the SVM depends on setting the appropriate parameters. Hence, the cuckoo search (CS) algorithm is adopted to optimize the key parameters to avoid the local minimum value which can occurs when using the traditional method of parameter optimization. To validate the predictive performance of the proposed CS-SVM model, the dynamic measurement error data for two sensors are applied to establish a predictive model. The root mean squared error and the mean absolute percentage error are employed to evaluate the models' performances. These results are also compared with those obtained from the SVM optimized by a grid search and the particle swarm optimization method. The experiments show that the SVM model based on the CS algorithm achieves more accurate prediction and is more effective in predicting dynamic measurement errors for sensors than the previous models.

Monocular scene flow estimation via variational method

Scene flow provides the 3D motion field of point clouds, which correspond to image pixels. Current algorithms usually need complex stereo calibration before estimating flow, which has strong restrictions on the position of the camera. This paper proposes a monocular camera scene flow estimation algorithm. Firstly, an energy functional is constructed, where three important assumptions are turned into data terms derivation: a brightness constancy assumption, a gradient constancy assumption, and a short time object velocity constancy assumption. Two smooth operators are used as regularization terms. Then, an occluded map computation algorithm is used to ensure estimating scene flow only on un-occluded points. After that, the energy functional is solved with a coarse-to-fine variational equation on Gaussian pyramid, which can prevent the iteration from converging to a local minimum value. The experiment results show that the algorithm can use three sequential frames at least to get scene flow in world coordinate, without optical flow or disparity inputting.

Hysteresis Model of Magnetically Controlled Shape Memory Alloy Based on a PID Neural Network

Magnetically controlled shape memory alloys(MSMA) are a new kind of smart material that can be used in micro-displacement and micro-positioning applications. However, the hysteresis nonlinearity of this material is an obstacle to achieve high precision accuracy [1]. The hysteresis nonlinearity of MSMA can be described by the relationship between the input and output signal which is called hysteresis loop [2]. As shown in figure 1(a), hysteresis loops can be divided into major and minor hysteresis loops. The input and output signals have the same phase at the extrema and the phase of the output signal lags behind the input signal at all other times. A major hysteresis loop takes shape when the input signal rises from an initial value to a maximum value before decreasing to a minimum value (which is same as the initial value). Minor hysteresis loops occur when the input signal rises from a local minimum value (which is different from the overall minimum value) to a local maximum value (which is different from the overall maximum value) before decreasing to a local minimum value.

Study on Urban Remote Sensing Classification Based on Improved RBF Network and Normalized Difference Indexes

Aiming at the complexity of ground objects in urban area, and the difficulty in distinguishing ground objects using spectral characteristics, we extracted normalized different indexes, namely Modified Normalized Difference Water Index (MNDWI), Soil Adjusted Vegetation Index (SAVI ) and Normalized Difference Building Index (NDBI) , as the key auxiliary information for land use classification of urban area. To solve problems of RBF neural network, such as local minimum values and discrete output value in output layer, we used max-min distance means to initialize RBF center, and introduced equilibrium factor into Gauss function to improve RBF neural network learning algorithm. On this basis, a new urban area classification model was proposed based on improved RBF network and normalized difference indexes. At last, NanChong city in SiChuan province of China was taken as the study area, and TM images was used as experiment data to test the model proposed in this paper. The results showed that, based on the improved RBF network, with the help of spectral band information, the classification overall accuracy was 89.97%, Kappa coefficient was 0.88; using both spectral band information and normalized difference indexes, the classification overall accuracy was 95.02%, Kappa coefficient is 0.94, the classification overall accuracy was improved by 5.05%. Also, the experiment results showed that, with the help of spectral band information and normalized difference indexes, the classification overall accuracy of MLC, BP and improved RBF network was 90.12%, 93.63%, 95.02%, respectively, which means RBF has an advantage of fusing geological parameters in classification.

Research on Hybrid Algorithm Based on BP Neural Network and Genetic Algorithm

BP neural network can approximate nonlinear functions in any degree of accuracy. However, it exhibits the shortcoming of slow convergence, and it is easy to fall into local minimum value. The genetic algorithm has academic ad- vantages both at home and abroad. On the basis of analysis on features of genetic algorithm and BP neural network, the explanations on the necessity of combining genetic algorithm and neural network are provided. In order to improve the structure and setting parameters of neural network, a hybrid algorithm which can optimize BP neural network based on genetic algorithm is proposed. Hybrid algorithm optimizes the weights of BP neural network by an improved genetic al- gorithm, and makes up the neural network with global random search, which has the capability of genetic algorithm and likely falls into the local optimal solution. Meanwhile, the changes of traditional mechanisms on the same generation crossover in genetic algorithm, and the use of parent and offspring cross to avoid premature loss of evolution ability of genetic algorithm are also discussed in this paper. The algorithm is validated by using standard data sets wine and letter- recognition of UCI database. The results show that the hybrid algorithm proposed in this paper can improve the conver- gence speed of BP neural network and speed up the training process of this network.

Modeling of transition boiling under an impinging water jet

Liquid jet impingement is known for its high capability of extracting heat from the surface it impinges on. Steady state subcooled boiling experiments were carried out in the stagnation region of a water planar jet of 0.6 and 0.75m/s jet velocity and 15°C of subcooling. The transition boiling regime is found to have a region of constant heat flux. After the critical heat flux (CHF) is reached, the heat flux decreases till a local minimum value is reached; called the first minimum. The heat flux then increases till it reaches a relatively constant value, even with the increase of the surface degree of superheat. The constant heat flux region is called the shoulder heat flux. With the aid of high speed imaging, Rayleigh–Taylor instability has been observed at the liquid–vapor interface which is accelerated by gravity and jet dynamic pressure. A modified expression for the interface acceleration is proposed. The vapor layer formed in the transition boiling is not stable; it follows periodic cycles of break up and formation. Two surface wetting mechanisms are proposed based on observations: (i) direct liquid touching the surface at moderate degrees of superheat and (ii) liquid columns intruding the vapor layer and touching the surface at discrete locations at high degrees of superheat. A new wall heat flux partitioning model is proposed. The proposed model divides the wall heat flux, based on the two observed wetting mechanisms, into two components: (i) quenching heat flux and (ii) intrusion heat flux. The current model estimates the wall heat flux during transition boiling covering the surface superheat range from the CHF till the Leidenfrost point with a normalized root mean square error (NRMSE) of 33%.

Permeability Prediction of Tight Sandstone Reservoirs Using Improved BP Neural Network

By analyzing the permeability controlling factors of tight sandstone reservoir in Wuhaozhuang Oil Field, the permeability is considered to be mainly controlled by porosity, clay content, irreducible water saturation and diagenetic coefficient. Because the conventional BP algorithm has its drawbacks such as slow convergence speed and easy falling into the local minimum value, an improved three-layer feed-forward BP neural network model is built by MATLAB neural network toolbox to predict permeability according to the four permeability controlling factors, while studying samples of model are selected based on the representative core analysis data. The simulation based on improved neural network model shows that the improved model has a faster convergence speed and better accuracy. The consistency between model prediction value and lab test value is good and the mean squared error is less. Therefore, the new model can meet the needs of the development geology research of oil field better in the future.

AF-DHNN: Fuzzy Clustering and Inference-Based Node Fault Diagnosis Method for Fire Detection.

Wireless Sensor Networks (WSNs) have been utilized for node fault diagnosis in the fire detection field since the 1990s. However, the traditional methods have some problems, including complicated system structures, intensive computation needs, unsteady data detection and local minimum values. In this paper, a new diagnosis mechanism for WSN nodes is proposed, which is based on fuzzy theory and an Adaptive Fuzzy Discrete Hopfield Neural Network (AF-DHNN). First, the original status of each sensor over time is obtained with two features. One is the root mean square of the filtered signal (FRMS), the other is the normalized summation of the positive amplitudes of the difference spectrum between the measured signal and the healthy one (NSDS). Secondly, distributed fuzzy inference is introduced. The evident abnormal nodes’ status is pre-alarmed to save time. Thirdly, according to the dimensions of the diagnostic data, an adaptive diagnostic status system is established with a Fuzzy C-Means Algorithm (FCMA) and Sorting and Classification Algorithm to reducing the complexity of the fault determination. Fourthly, a Discrete Hopfield Neural Network (DHNN) with iterations is improved with the optimization of the sensors’ detected status information and standard diagnostic levels, with which the associative memory is achieved, and the search efficiency is improved. The experimental results show that the AF-DHNN method can diagnose abnormal WSN node faults promptly and effectively, which improves the WSN reliability.

Dynamic viscoelastic properties of Chinese fir under cyclical relative humidity variation

The dynamic viscoelastic properties of Chinese fir (Cunninghamia lanceolata) with a moisture content of 10.65 % (30 °C, 58 % RH) under cyclical relative humidity (RH) variation (90–10 % RH) were examined. After seven-cycle RH changes, moisture content of specimens decreased. The storage modulus and loss factor increased and decreased, respectively. The occurrences of local maximum or minimum values of the storage modulus and loss factor lagged behind the occurrences of the maximum or minimum values of RH. Both lagged time and changing rate of storage modulus and loss factor decreased with the increasing cyclical times. When RH decreased from 90 to 10 % RH, the increasing rate of the storage modulus decreased with the decreasing of RH, while the loss factor decreased almost linearly. The unsteady effect and its partial recovery were observed during the desorption and the adsorption process, respectively. With the increasing cyclical times, the unsteady effect decreased and wood returned back to the dynamic equilibrium state gradually.

Boron removal from metallurgical-grade silicon by CaO–SiO2 slag refining

Boron removal from metallurgical-grade silicon (MG-Si) using CaO–SiO2 slag was studied by employing a medium-frequency electromagnetic induction furnace. The relationship between the optical basicity (Λ) of the CaO–SiO2 slag and the distribution coefficient of boron (LB) was investigated. Consequently, the local minimum and maximum LB values of 0.72 and 1.58 are obtained when Λ = 0.56 and Λ = 0.71, respectively. The boron content in MG-Si decreases gradually with refinement time increasing, down to a minimum value of 4.73 × 10−6. The controlling step in the removal of boron from MG-Si is not the chemical reaction at the interface of the slag and silicon. Instead, the controlling step is a diffusion mass transfer, in which boron impurities diffuse from molten silicon to the interface of the slag and silicon, or B2O3 formed by the chemical reaction diffuses from the slag–silicon interface to molten slag.