Dislocation parameters of Gonghe earthquake jointly inferred by using genetic algorithms and least squares method

The Second Crustal Deformation Monitoring Center, China Seismological Bureau, has detected a marked uplift associated with the Gonghe Ms=7.0 earthquake on April 26, 1990, Qinghai Province. From the observed vertical deformations and using a rectangular uniform slip model in a homogeneous elastic half space, we first employ genetic algorithms (GA) to infer the approximate global optimal solution, and further use least squares method to get more accurate global optimal solution by taking the approximate solution of GA as the initial parameters of least squares. The inversion results show that the causative fault of Gonghe Ms=7.0 earthquake is a right-lateral reverse fault with strike NW60°, dip SW and dip angle 37°, the coseismic fracture length, width and slip are 37 km, 6 km and 2.7 m respectively. Combination of GA and least squares algorithms is an effective joint inversion method, which could not only escape from local optimum of least squares, but also solve the slow convergence problem of GA after reaching adjacency of global optimal solution.

Improved scheme to accelerate sparse least squares support vector regression

The pruning algorithms for sparse least squares support vector regression machine are common methods, and easily com- prehensible, but the computational burden in the training phase is heavy due to the retraining in performing the pruning process, which is not favorable for their applications. To this end, an im- proved scheme is proposed to accelerate sparse least squares support vector regression machine. A major advantage of this new scheme is based on the iterative methodology, which uses the previous training results instead of retraining, and its feasibility is strictly verified theoretically. Finally, experiments on bench- mark data sets corroborate a significant saving of the training time with the same number of support vectors and predictive accuracy compared with the original pruning algorithms, and this speedup scheme is also extended to classification problem.

The posterior selection method for hyperparameters in regularized least squares method

The selection of hyperparameters in regularized least squares plays an important role in large-scale system identification. The traditional methods for selecting hyperparameters are based on experience or marginal likelihood maximization method, which are inaccurate or computationally expensive. In this paper, two posterior methods are proposed to select hyperparameters based on different prior knowledge (constraints), which can obtain the optimal hyperparameters using the optimization theory. Moreover, we also give the theoretical optimal constraints, and verify its effectiveness. Numerical simulation shows that the hyperparameters and parameter vector estimate obtained by the proposed methods are the optimal ones.

Prediction of Time Series Empowered with a Novel SREKRLS Algorithm

For the unforced dynamical non-linear state–space model,a new Q1 and efficient square root extended kernel recursive least square estimation algorithm is developed in this article.The proposed algorithm lends itself towards the parallel implementation as in the FPGA systems.With the help of an ortho-normal triangularization method,which relies on numerically stable givens rotation,matrix inversion causes a computational burden,is reduced.Matrix computation possesses many excellent numerical properties such as singularity,symmetry,skew symmetry,and triangularity is achieved by using this algorithm.The proposed method is validated for the prediction of stationary and non-stationary Mackey–Glass Time Series,along with that a component in the x-direction of the Lorenz Times Series is also predicted to illustrate its usefulness.By the learning curves regarding mean square error(MSE)are witnessed for demonstration with prediction performance of the proposed algorithm from where it’s concluded that the proposed algorithm performs better than EKRLS.This new SREKRLS based design positively offers an innovative era towards non-linear systolic arrays,which is efficient in developing very-large-scale integration(VLSI)applications with non-linear input data.Multiple experiments are carried out to validate the reliability,effectiveness,and applicability of the proposed algorithm and with different noise levels compared to the Extended kernel recursive least-squares(EKRLS)algorithm.

Deep Learning to Find Key Brain Electrode Positions in Emotional EEG

At present, the channels used for EEG acquisition are more than 16, which makes it difficult to wear EEG caps and has poor contact. Therefore, it brings difficulties to the collection of brain waves is not conducive to converting research into applications. It is a well worth studying work in researching how to find the key brain electrode in the existing brain wave, which will greatly reduce the number of EEG acquisition points during application, making it easier to translate the research into practical application. This paper takes emotional EEG as an example to study how to find the key brain electrode points of emotional EEG with deep learning method. Firstly, using the least square regression algorithm to calculate the characteristic coefficients of each electrode point;secondly, according to the law of the characteristic coefficient value, grouping the key EEG poles for experiment. In the grouping experiment, the Conv1d-GRU model used to train and verify the EEG data of the corresponding electrode points. Finally, from the results of various grouping experiments, it concluded that the selection method of the key EEG level points should be the electrode points with positive characteristic coefficient, and the accuracy of verification is 97.6%. With experiments, it confirmed that there are key electrode points in the detection of emotional EEG by 16-channel OpenBCI. There are only six key electrode points of emotional EEG;that is to say, the EEG data collected by only six key electrode points can identify seven kinds of emotional EEG. .

Correlation-weighted least squares residual algorithm for RAIM

The Least Squares Residual(LSR)algorithm,one of the classical Receiver Autonomous Integrity Monitoring(RAIM)algorithms for Global Navigation Satellite System(GNSS),presents a high Missed Detection Risk(MDR)for a large-slope faulty satellite and a high False Alarm Risk(FAR)for a small-slope faulty satellite.From the theoretical analysis of the high MDR and FAR cause,the optimal slope is determined,and thereby the optimal test statistic for fault detection is conceived,which can minimize the FAR with the MDR not exceeding its allowable value.To construct a test statistic approximate to the optimal one,the CorrelationWeighted LSR(CW-LSR)algorithm is proposed.The CW-LSR test statistic remains the sum of pseudorange residual squares,but the square for the most potentially faulty satellite,judged by correlation analysis between the pseudorange residual and observation error,is weighted with an optimal-slope-based factor.It does not obey the same distribution but has the same noncentral parameter with the optimal test statistic.The superior performance of the CW-LSR algorithm is verified via simulation,both reducing the FAR for a small-slope faulty satellite with the MDR not exceeding its allowable value and reducing the MDR for a large-slope faulty satellite at the expense of FAR addition.

An enhanced least squares residual RAIM algorithm based on optimal decentralized factor

The Least Squares Residual(LSR)algorithm is commonly used in the Receiver Autonomous Integrity Monitoring(RAIM).However,LSR algorithm presents high Missed Detection Risk(MDR)caused by a large-slope faulty satellite and high False Alert Risk(FAR)caused by a small-slope faulty satellite.In this paper,the LSR algorithm is improved to reduce the MDR for a large-slope faulty satellite and the FAR for a small-slope faulty satellite.Based on the analysis of the vertical critical slope,the optimal decentralized factor is defined and the optimal test statistic is conceived,which can minimize the FAR with the premise that the MDR does not exceed its allowable value of all three directions.To construct a new test statistic approximating to the optimal test statistic,the Optimal Decentralized Factor weighted LSR(ODF-LSR)algorithm is proposed.The new test statistic maintains the sum of pseudo-range residual squares,but the specific pseudo-range residual is weighted with a parameter related to the optimal decentralized factor.The new test statistic has the same decentralized parameter with the optimal test statistic when single faulty satellite exists,and the difference between the expectation of the new test statistic and the optimal test statistic is the minimum when no faulty satellite exists.The performance of the ODFLSR algorithm is demonstrated by simulation experiments.

A novel procedure for identifying a hybrid QTL-allele system for hybrid-vigor improvement, with a case study in soybean(Glycine max)yield

“Breeding by design” for pure lines may be achieved by construction of an additive QTL-allele matrix in a germplasm panel or breeding population, but this option is not available for hybrids, where both additive and dominance QTL-allele matrices must be constructed. In this study, a hybrid-QTL identification approach, designated PLSRGA, using partial least squares regression(PLSR) for model fitting integrated with a genetic algorithm(GA) for variable selection based on a multi-locus, multi-allele model is described for additive and dominance QTL-allele detection in a diallel hybrid population(DHP). The PLSRGA was shown by simulation experiments to be superior to single-marker analysis and was then used for QTL-allele identification in a soybean DPH yield experiment with eight parents. Twenty-eight main-effect QTL with 138 alleles and nine QTL × environment QTL with 46 alleles were identified, with respective contributions of 61.8% and 23.5% of phenotypic variation. Main-effect additive and dominance QTL-allele matrices were established as a compact form of the DHP genetic structure. The mechanism of heterosis superior-to-parents(or superior-to-parents heterosis, SPH) was explored and might be explained by a complementary locus-set composed of OD+(showing positive over-dominance, most often), PD+(showing positive partial-to-complete dominance, less often) and HA+(showing positive homozygous additivity, occasionally) loci, depending on the parental materials. Any locus-type, whether OD+, PD + and HA+, could be the best genotype of a locus. All hybrids showed various numbers of better or best genotypes at many but not necessarily all loci, indicating further SPH improvement. Based on the additive/dominance QTL-allele matrices, the best hybrid genotype was predicted, and a hybrid improvement approach is suggested. PLSRGA is powerful for hybrid QTL-allele detection and cross-SPH improvement.

A COLUMN RECURRENCE ALGORITHM FOR SOLVING LINEAR LEAST SQUARES PROBLEM

A new column recurrence algorithm based on the classical Greville method and modified Huang update is proposed for computing generalized inverse matrix and least squares solution. The numerical results have shown the high efficiency and stability of the algorithm.

Modeling the dynamic behavior of a lithium-ion battery for electric vehicles using numerical optimization

In order to simulate the dynamical behavior of a lithium ion traction battery used in elec tric vehicles, an equivalent circuit based battery model was established. The methodology in the guide document of the ADVISOR software was used to determine the initial parameters of the model as a function of state of charge （ SoC ） over an experimental data set of the battery. A numerically nonlinear least squares algorithm in SIMULINK design optimization toolbox was applied to further op timize the model parameters. Validation results showed that the battery model could well describe the dynamic behavior of the lithinm ion battery in two different battery loading situations.

Randomized Algorithms for Orthogonal Nonnegative Matrix Factorization

Orthogonal nonnegative matrix factorization(ONMF)is widely used in blind image separation problem,document classification,and human face recognition.The model of ONMF can be efficiently solved by the alternating direction method of multipliers and hierarchical alternating least squares method.When the given matrix is huge,the cost of computation and communication is too high.Therefore,ONMF becomes challenging in the large-scale setting.The random projection is an efficient method of dimensionality reduction.In this paper,we apply the random projection to ONMF and propose two randomized algorithms.Numerical experiments show that our proposed algorithms perform well on both simulated and real data.

Novel approach of crater detection by crater candidate region selection and matrix-pattern-oriented least squares support vector machine

Impacted craters are commonly found on the surface of planets, satellites, asteroids and other solar system bodies. In order to speed up the rate of constructing the database of craters, it is important to develop crater detection algorithms. This paper presents a novel approach to automatically detect craters on planetary surfaces. The approach contains two parts： crater candidate region selection and crater detection. In the first part, crater candidate region selection is achieved by Kanade-Lucas-Tomasi （KLT） detector. Matrix-pattern-oriented least squares support vector machine （MatLSSVM）, as the matrixization version of least square support vector machine （SVM）, inherits the advantages of least squares support vector machine （LSSVM）, reduces storage space greatly and reserves spatial redundancies within each image matrix compared with general LSSVM. The second part of the approach employs MatLSSVM to design classifier for crater detection. Experimental results on the dataset which comprises 160 preprocessed image patches from Google Mars demonstrate that the accuracy rate of crater detection can be up to 88%. In addition, the outstanding feature of the approach introduced in this paper is that it takes resized crater candidate region as input pattern directly to finish crater detection. The results of the last experiment demonstrate that MatLSSVM-based classifier can detect crater regions effectively on the basis of KLT-based crater candidate region selection.

Virtual Inertia Estimation Method of DFIG-based Wind Farm with Additional Frequency Control

With the increasing penetration of wind power,using wind turbines to participate in the frequency regulation to support power system has become a clear consensus.To accurately quantify the inertia provided by the doubly-fed induction generator(DFIG)based wind farm,the frequency response model of DFIG with additional frequency control is established,and then by using Routh approximation,the explicit expression of the virtual moment of inertia is derived for the DFIG gridconnected system.To further enhance the availability of the expression,an estimation method is proposed based on the matrix pencil method and the least squares algorithm for estimating the virtual moment of inertia provided by the wind farm.Finally,numerical results tested by a DFIG grid-connected system and a modified IEEE 30-bus system verify the derived expression of the virtual moment of inertia and the proposed estimation method.

Forward/backward prediction solution for adaptive noisy FIR filtering

An important and hard problem in signal processing is the estimation of parameters in the presence of observation noise.In this paper, adaptive finite impulse response （FIR） filtering with noisy input-output data is considered and two developed bias compensation least squares （BCLS） methods are proposed.By introducing two auxiliary estimators, the forward output predictor and the backward output predictor are constructed respectively.By exploiting the statistical properties of the cross-correlation function between the least squares （LS） error and the forward/backward prediction error, the estimate of the input noise variance is obtained; the effect of the bias can thereafter be removed.Simulation results are presented to illustrate the good performances of the proposed algorithms.

Design, Implementation and Control of an Amphibious Spherical Robot

We proposed and implemented a leg-vector water-jet actuated spherical robot and an underwater adaptive motion control system so that the proposed robot could perform exploration tasks in complex environments.Our aim was to improve the kinematic performance of spherical robots.We developed mechanical and dynamic models so that we could analyze the motions of the robot on land and in water.The robot was equipped with an Inertial Measurement Unit(IMU)that provided inclination and motion information.We designed three types of walking gait for the robot,with different stabilities and speeds.Furthermore,we proposed an online adjustment mechanism to adjust the gaits so that the robot could climb up slopes in a stable manner.As the system function changed continuously as the robot moved underwater,we implemented an online motion recognition system with a forgetting factor least squares algorithm.We proposed a generalized prediction control algorithm to achieve robust underwater motion control.To ensure real-time performance and reduce power consumption,the robot motion control system was implemented on a Zynq-7000 System-on-Chip(SoC).Our experimental results show that the robot’s motion remains stable at different speeds in a variety of amphibious environments,which meets the requirements for applications in a range of terrains.

Fast adaptive principal component extraction based on a generalized energy function

By introducing an arbitrary diagonal matrix, a generalized energy function (GEF) is proposed for searching for the optimum weights of a two layer linear neural network. From the GEF, we derive a recur- sive least squares (RLS) algorithm to extract in parallel multiple principal components of the input covari- ance matrix without designing an asymmetrical circuit. The local stability of the GEF algorithm at the equilibrium is analytically verified. Simulation results show that the GEF algorithm for parallel multiple principal components extraction exhibits the fast convergence and has the improved robustness resis- tance to the eigenvalue spread of the input covariance matrix as compared to the well-known lateral inhi- bition model (APEX) and least mean square error reconstruction (LMSER) algorithms.