Modelling of the Relaxation Least Squares-Based Neural Networks and Its Application

A relaxation least squares-based learning algorithm for neual networks is proposed. Not only does it have a fast convergence rate, but it involves less computation quantity. Therefore, it is suitable to deal with the case when a network has a large scale but the number of training data is very limited. It has been used in converting furnace process modelling, and impressive result has been obtained.

Robust On-Line Fault Diagnosis for Nonlinear Difference-Algebraic Systems Using Least Squares Estimate

A robust on-line fault diagnosis methor based on least squares estimate for nonlinear difference-algebraic systems (DAS) with uncertainties is proposed. Based on the known nominal model of the DAS, this method firstly constructs an auxiliary system consisting of a difference equation and an algebraic equation, then, based on the relationship between the state deviation and the faults in the difference equation and the relationship between the algebraic variable deviation and the faults in algebraic equation, it identifies the faults on-line through least squares estimate. This method can not only detect, isolate and identify faults for DAS, but also give the upper bound of the error of fault identification. The simulation results indicate that it can give satisfactory diagnostic results for both abrupt and incipient faults.

Application of nonlinear partial least square in catalyst modeling

The neural network partial least square (NNPLS) method was used to establish a robust reaction model for a multi-component catalyst of methane oxidative coupling. The details, including the learning algorithm, the number of hidden units of the inner network, activation function, initialization of the network weights and the principal components, are discussed. The results show that the structural organizations of inner neural network are 1-10-5-1, 1-8-4-1, 1-8-5-1, 1-7-4-1, 1-8-4-1, 1-8-6-1, respectively. The Levenberg-Marquardt method was used in the learning algorithm, and the central sigmoidal function is the activation function. Calculation results show that four principal components are convenient in the use of the multi-component catalyst modeling of methane oxidative coupling. Therefore a robust reaction model expressed by NNPLS succeeds in correlating the relations between elements in catalyst and catalytic reaction results. Compared with the direct network modeling, NNPLS model can be adjusted by experimental data and the calculation of the model is simpler and faster than that of the direct network model.

Neural Network inverse Adaptive Controller Based on Davidon Least Square

General neural network inverse adaptive controller has two flaws: the first is the slow convergence speed; the second is the invalidation to the non-minimum phase system. These defects limit the scope in which the neural network inverse adaptive controller is used. We employ Davidon least squares in training the multi-layer feedforward neural network used in approximating the inverse model of plant to expedite the convergence, and then through constructing the pseudo-plant, a neural network inverse adaptive controller is put forward which is still effective to the nonlinear non-minimum phase system. The simulation results show the validity of this scheme.

Weight-based shape modification of NURBS curves by constrained optimization

A new method for shape modification of non-uniform rational B-splines (NURBS) curves was presented, which was based on constrained optimization by means of altering the corresponding weights of their control points. Using this method, the original NURBS curve was modified to satisfy the specified geometric constraints, including single point and multi-point constraints. With the introduction of free parameters, the shapes of modified NURBS curves could be further controlled by users without destroying geometric constraints and seem more naturally. Since explicit formulae were derived to compute new weights of the modified curve, the method was simple and easy to program. Practical examples showed that the method was applicable for computer aided design (CAD) system.

Study on the tradeoff between interpretability and precision in fuzzy modeling

An approach to identifying fuzzy models considering both interpretability and precision was proposed. Firstly, interpretability issues about fuzzy models were analyzed. Then, a heuristic strategy was used to select input variables by increasing the number of input variables, and the Gustafson-Kessel fuzzy clustering algorithm, combined with the least square method, was used to identify the fuzzy model. Subsequently, an interpretability measure was described by the product of the number of input variables and the number of rules, while precision was weighted by root mean square error, and the selection objective function concerning interpretability and precision was defined. Given the maximum and minimum number of input variables and rules, a set of fuzzy models was constructed. Finally, the optimal fuzzy model was selected by the objective function, and was optimized by a genetic algorithm to achieve a good tradeoff between interpretability and precision. The performance of the proposed method was illustrated by the well-known Box-Jenkins gas furnace benchmark; the results demonstrate its validity.

Quaternion based joint DOA and polarization parameters estimation with stretched three-component electromagnetic vector sensor array

The three-component electromagnetic vector sensor (EMVS) consisting of co-centered orthogonally oriented x-dipole, z-dipole and z-loop is considered. In order to make full use of the spatial aperture of each component, the original uniform linear three-component EMVS array (ULTEA) is stretched into one half-wavelength spaced uniform linear loop subarray (ULLSA) along the z axis, and one sparse uniform linear co-centered orthogonally oriented dual-dipole (CODD) subarray (SULCSA) along the x axis. Then, a generalized rotation invariance based quaternion multiple signal classification (GRIQ-MUSIC) algorithm is presented for direction of arrival (DOA) and polarization parameters estimation. According to the proposed algorithm, the elevation angles are firstly estimated based on the half-wavelength spaced ULLSA. Then the polarization phase differences and azimuth angles are obtained based on the coupling relationship between the angle domain and polarization domain, but the azimuth angles are in coarse-resolution since the array aperture is not utilized. Next, the SULCSA is used to re-estimate the azimuth angles in fine-resolution, and the ambiguity problem can be resolved by the least square method. Finally, based on the estimated elevation angles, azimuth angles and polarization phase differences, the corresponding auxiliary polarization angles can be estimated by N times one-dimensional parameter search, where N is the sources number, and the parameters are matched automatically. Based on the GRIQ-MUSIC algorithm, the high dimensional parameters search problem of the conventional Q-MUSIC algorithm is simplified to a one-dimensional parameter search problem, thus the proposed algorithm not only reduces the computation complexity considerably, but also avoids the performance degradation caused by the failure in parameters pairing. The simulation examples demonstrate the effectiveness and feasibility of the proposed algorithm. © 1990-2011 Beijing Institute of Aerospace Information.

Design and Experiments of Ultrasound Image-Guided Multi-DOF Robot System for Brachytherapy

To implant radioactive seeds through a needle precisely and safely, a novel multi-DOF surgical robotic system is presented in this paper for percutaneous prostate intervention through the patient’s perineum under real-time ultrasound image guidance. The proposed robot, which is designed with 9-DOF, consists of a 3-DOF automatic location platform for position adjustment, 2-DOF for automatic ultrasonic probe adjustment mounted with electromagnetic trackers, and 4-DOF for manually adjusting the guided template. Meanwhile, a new registration method based on the quaternion algorithm and least square method is developed, and the needle insertion is performed under the real-time guidance of a navigation system. Furthermore, the robot system has undergone some preliminary experiments with a laser tracker to evaluate the repeatability and accuracy of the robot system. The location error of the puncture needle tip can be controlled under 0.7 mm in air for the whole robotic system. The acquired results endorse the precision of the robot system for prostate seed implantation brachytherapy. © 2017, Tianjin University and Springer-Verlag GmbH Germany.

An Iterative Learning Approach to Identify Fractional Order KiBaM Model

This paper discusses the parameter and differentiation order identification of continuous fractional order KiBaM models in ARX (autoregressive model with exogenous inputs) and OE (output error model) forms. The least squares method is applied to the identification of nonlinear and linear parameters, in which the Grünwald-Letnikov definition and short memory principle are applied to compute the fractional order derivatives. An adaptive P-type order learning law is proposed to estimate the differentiation order iteratively and accurately. Particularly, a unique estimation result and a fast convergence speed can be arrived by using the small gain strategy, which is unidirectional and has certain advantages than some state-of-art methods. The proposed strategy can be successfully applied to the nonlinear systems with quasi-linear characteristics. The numerical simulations are shown to validate the concepts. © 2017 Chinese Association of Automation.

Stable Estimation of Horizontal Velocity for Planetary Lander with Motion Constraints

The planetary lander usually selects image feature points and tracks them from frame to frame in order to determine its own position and velocity during landing. Aiming to keep features tracking in consecutive frames, this paper proposes an approach of calculating the field of view (FOV) overlapping area in a 2D plane. Then the rotational and translational motion constraints of the lander can be found. If the FOVs intersects each other, the horizontal velocity of the lander is quickly estimated based on the least square method after the ill-conditioned matrices are eliminated previously. The Monte Carlo simulation results show that the proposed approach is not only able to recover the ego-motion of planetary lander, but also improves the stabilization performance. The relationship of the estimation error, running time and number of points is shown in the simulation results as well. © 2014 Chinese Association of Automation.

Crystallite Orientation Analysis for a Multidirectionally Cold Rolled α-Ti Alloy (TA7) Thin Sheet

A microcomputer software determining the crystallite orientation distribution functions (ODFs) and inverse pole figures of hexagonal materials with or without orthorhombic physical symmetry has been worked out the first time. The texture measurements and the ODF calculations were performed for a multidirectionally cold rolled α-Ti alloy (TA7) sheet by the application of this software. It is shown that the rolling planes of most grains in the sheet tend to be parallel to (0001) with a deviation to the extent of 40° and is shown a predominance of the orientation zone containing (1016), while the rolling directions are, as a whole, uniformly distributed along all the directions over the rolling planes. Of all the texture components, (2¯117) [01¯10] is slightly stronger than the other.

Time -Sharing Power Supply Optimal Decision forElectrolytic Deposition Process of Zinc Based onSimulated Annealing Algorithm

According to time-sharing valuation principle (TSVP) of power supply, the relationships of current density and current efficiency at different acidities are obtained based on the processed data of electrolytic deposition process of zinc (EDPZ) with the least square method (LSM). Thus an optimal model of time-sharing power supply system for EDPZ is established, which has been optimized by use of an improved efficient simulated annealing algorithm (SAA). Practical results show that industrial and mining enterprises can obtain enormous economic benefits every year.

Determination of Crystal Structure of Britholite-Y

Britholite-Y is milkwhite, rosy in colour. The measured specific gravity is 4.35, with Ng′ = 1.791, Np′ = 1.784. As determined by electron microprobe, its crystal formula is (Y2.82Ca1.58Ce0.27Dy0.21Er0.11)5 [(Si2.95P0.05)3O12](OH,F), space group = C22-P21, a = (0.9504±0.0005) nm, b = (0.9414±0.0004) nm, c = (0.6922±0.0002) nm, r = (119.71±0.04)°, V = (53.79±0.04) nm3, Z = 2. Least-squares refinements with 2272 independent reflection (F0>3σ|F0|) yielded R = 0.111. The change of symmetry group P63/m of apatite to P21 of britholite-Y results from the shifts of Y, Ca, Si, O and (OH) atoms (anions) from the pseudohexagonal P63/m equivalent positions (6h), (4f), (12i) and (2a) as produced by distortion of the polyhedra with 7-, 9-, and 4-corners. The substitution of Ca in apatite for Y and the order distribution of atoms Y and Ca on the equivalent positions (6h) and (4f) have contributed to the distortion of 7- and 9- cornered polyhedra. The substitutions of Si4+ for P5+ and OH- for F- may also be effected. The crystal chemistry of apatite-group minerals is also discussed.

STUDY OF C_D, C_M, C_L AND WAVE FORCES ON PIPE LINES NEAR THE OCEAN BOTTOM

The wave forces on pipes near the ocean bottom are studied with model test in this paper. Comparisons of four different methods show that the uncommonly used method - Least Square Mass (LSM) is the most stable one. The in-line and transverse forces on the pipe, which is placed at various distances from the bottom of the wave tank, are measured. The force transfer coefficients (CD, CM, CL) are much larger when the pipe is near the bottom than when it is free. But the coefficients do not increase unlimitedly: they reach their maxima at ζ (gap ratio = e / D) = 0.1 (gap ratio ζ = e / D). An interesting result is that the ratio of lift force frequency (f′L) to wave frequency (fw) is 2.0. In the paper, the curves of coefficients are given.

Inverse Design of Electron Lens

The inverse design of electron lens is realized by two different methods in this paper. One is damped least square method and the other is the artificial neural network method. Their merits and defects are discussed according to our calculation results in the paper. In the condition of selecting the learning samples properly, the artificial neural network has obvious advantages in the inverse design of electron lens. It is an effective method to solve the inverse design problem in the electron optic system.

A 3D shell-like approach using element-free Galerkin method for analysis of thin and thick plate structures

A new efficient meshless method based on the element-free Galerkin method is proposed to analyze the static deformation of thin and thick plate structures in this paper. Using the new 3D shell-like kinematics in analogy to the solid-shell concept of the finite element method, discretization is carried out by the nodes located on the upper and lower surfaces of the structures. The approximation of all unknown field variables is carried out by using the moving least squares （MLS） approximation scheme in the in-plane directions, while the linear interpolation is applied through the thickness direction. Thus, different boundary conditions are defined only using displacements and penalty method is used to enforce the essential boundary conditions. The constrained Galerkin weak form, which incorporates only dis- placement degrees of freedom （d.o.f.s）, is derived. A modified 3D constitutive relationship is adopted in order to avoid or eliminate some self-locking effects. The numeric efficiency of the proposed meshless formulation is illustrated by the numeric examples.

ANALYSIS OF 2-D THIN STRUCTURES BY THE MESHLESS REGULAR HYBRID BOUNDARY NODE METHOD

Thin structures are generally solved by the Finite Element Method(FEM), using plate or shell finite elements which have manylimitations in applications, such as numerical locking, edge effects,length scaling and the cnvergence problem. Recently, by proposing anew approach to tranting the nearly- singular integrals, Liu et al.developed a BEM to successfully solve thin structures with thethickness-to- length ratios in the micro-or nano-scales. On the otherhand, the meshless Regular Hybrid Boundary Node Method (RHBNM), whichis proposed by the current authors and based on a modified functionaland the Moving Least-Square (MLS) approximation, has very promisingapplications for engineering problems owing To its meshless natureand dimension-reduction advantage, and not involving any singular ornearly-singular Integrals. Test examples show that the RHBNM can alsobe applied readily to thin structures with high accu- Racy withoutany modification.

Meshless analysis of an improved element-free Galerkin method for linear and nonlinear elliptic problems

We first give a stabilized improved moving least squares （IMLS） approximation, which has better computational stability and precision than the IMLS approximation. Then, analysis of the improved element-free Galerkin method is provided theoretically for both linear and nonlinear elliptic boundary value problems. Finally, numerical examples are given to verify the theoretical analysis.

The improved element-free Galerkin method forthree-dimensional wave equation

The paper presents the improved element-free Galerkin （IEFG） method for three-dimensional wave propa- gation. The improved moving least-squares （IMLS） approx- imation is employed to construct the shape function, which uses an orthogonal function system with a weight function as the basis function. Compared with the conventional moving least-squares （MLS） approximation, the algebraic equation system in the IMLS approximation is not ill-conditioned, and can be solved directly without deriving the inverse matrix. Because there are fewer coefficients in the IMLS than in the MLS approximation, fewer nodes are selected in the IEFG method than in the element-free Galerkin method. Thus, the IEFG method has a higher computing speed. In the IEFG method, the Galerkin weak form is employed to obtain a dis- cretized system equation, and the penalty method is applied to impose the essential boundary condition. The traditional difference method for two-point boundary value problems is selected for the time discretization. As the wave equations and the boundary-initial conditions depend on time, the scal- ing parameter, number of nodes and the time step length are considered for the convergence study.

Oblique Wave Trapping by a Surface-Piercing Flexible Porous Barrier in the Presence of Step-Type Bottoms

The present study deals with the oblique wave trapping by a surface-piercing flexible porous barrier near a rigid wall in the presence of step-type bottoms under the assumptions of small amplitude water waves and the structural response theory in finite water depth.The modified mild-slope equation along with suitable jump conditions and the least squares approximation method are used to handle the mathematical boundary value problem.Four types of edge conditions,i.e.,clamped-moored,clamped-free,moored-free,and moored-moored,are considered to keep the barrier at a desired position of interest.The role of the flexible porous barrier is studied by analyzing the reflection coefficient,surface elevation,and wave forces on the barrier and the rigid wall.The effects of step-type bottoms,incidence angle,barrier length,structural rigidity,porosity,and mooring angle are discussed.The study reveals that in the presence of a step bottom,full reflection can be found periodically with an increase in(i)wave number and(ii)distance between the barrier and the rigid wall.Moreover,nearly zero reflection can be found with a suitable combination of wave and structural parameters,which is desirable for creating a calm region near a rigid wall in the presence of a step bottom.