Radial basis functions methods for solving Fokker-Planck equation

In this paper two numerical meshless methods for solving the Fokker-Planck equation are considered. Two methods based on radial basis functions to approximate the solution of Fokker-Planck equation by using collocation method are applied. The first is based on the Kansa's approach and the other one is based on the Hermite interpolation. In addition, to conquer the ill-conditioning of the problem for big number of collocation nodes, two time domain Discretizing schemes are applied. Numerical examples are included to demonstrate the reliability and efficiency of these methods. Also root mean square and N_e errors are obtained to show the convergence of the methods. The errors show that the proposed Hermite collocation approach results obtained by the new time-Discretizing scheme are more accurate than the Kansa's approach.     

Variable selection for neural networks in multivariate calibration

The problem of variable selection for neural network modeling is discussed in this paper. Two methods that gave the best results in a previous comparative study are presented. One of these methods is a modified version of the Hinton diagrams, the other method is based on saliency estimation and is part of the Optimal Brain Surgeon algorithm for pruning unimportant weights in a neural network. We also propose two new methods, based on the estimation of the contribution of each input variable to the variance of the predicted response. These new methods are designed for situations where input variables are orthogonal, such as the PC scores often used in multivariate calibration. The four methods are tested on synthetic examples, and on real industrial data sets for multivariate calibration. The main characteristics of each method are discussed. In particular, we underline the strong theoretical and experimental limitations of methods like the modified Hinton diagrams, based on weight magnitude estimation. We also demonstrate that although the saliency estimation approach is theoretically more stringent, it gives unstable results on repeated trials. The advantage of the two variance-based approaches is that they are much less dependent on the initial weight randomization than the two other methods, and therefore, the results they produce are more stable and reliable.     

Maximum likelihood multivariate calibration

Two new approaches to multivariate calibration are described that, for the first time, allow information on measurement uncertainties to be included in the calibration process in a statistically meaningful way. The new methods, referred to as maximum likelihood principal components regression (MLPCR) and maximum likelihood latent root regression (MLLRR), are based on principles of maximum likelihood parameter estimation. MLPCR and MLLRR are generalizations of principal components regression (PCR), which has been widely used in chemistry, and latent root regression (LRR), which has been virtually ignored in this field. Both of the new methods are based on decomposition of the calibration data matrix by maximum likelihood principal component analysis (MLPCA), which has been recently described (Wentzell, P. D.; et al. J. Chemom., in press). By using estimates of the measurement error variance, MLPCR and MLLRR are able to extract the optimum amount of information from each measurement and, thereby, exhibit superior performance over conventional multivariate calibration methods such as PCR and partial least-squares regression (PLS) when there is a nonuniform error structure. The new techniques reduce to PCR and LRR when assumptions of uniform noise are valid. Comparisons of MLPCR, MLLRR, PCR, and PLS are carried out using simulated and experimental data sets consisting of three-component mixtures. In all cases of nonuniform errors examined, the predictive ability of the maximum likelihood methods is superior to that of PCR and PLS, with PLS performing somewhat better than PCR. MLLRR generally performed better than MLPCR, but in most cases the improvement was marginal. The differences between PCR and MLPCR are elucidated by examining the multivariate sensitivity of the two methods.     

EMD端点效应的改进型混沌延拓方法及其在机械故障诊断中的应用

针对经验模态分解(Empirical Mode Decomposition,EMD)的端点效应问题,分析了三次样条插值产生端点效应的机理,指出了现有延拓方法的不足,提出采用Lyapunov指数预测模型进行边界延拓,以此减小端点效应在经验模态分解过程中产生的影响。在采用最大Lyapunov指数法进行边界延拓时,针对Lyapunov指数预测模型所表示的平均发散度仅是轨道演化规律的一个近似模型,提出采用局域发散度代替最大Lyapunov指数进行改进,以改进预测模型的预测精度。基于改进型Lyapunov指数边界延拓方法,由于引入时间序列预测模型使端点处的延拓更加合理,所延拓数据更加趋于真实,在希尔伯特-黄变换（Hilbert-Huang Transform,HHT）实现过程中仅需一次延拓,实现了准确的EMD分解。仿真计算和转子系统故障试验分析结果表明,所用方法可以有效解决EMD的端点效应问题。     

Detection of voltage sag sources based on instantaneous voltage and current vectors and orthogonal Clarke?s transformation

Different methods for voltage sag source detection are discussed. They are based either on disturbance energy, voltage-current characteristic or active (real) current component. It is shown that, in the cases of asymmetrical voltage sags, both current-based methods known from literature do not work well. These two methods are therefore generalised using a vector-space approach. Furthermore, three new methods are introduced using orthogonal Clarke's transformation, which can be used to detect the sources of those voltage sags provoked by earth faults. All the discussed methods for voltage sag source detection have been tested by applying extensive simulations, laboratory tests and field testing. The results obtained show the very high effectiveness of the proposed methods, which are superior to the methods known from literature, especially for detecting sources of asymmetrical voltage sags.     

统计圆度公差及其误差评定的研究

 在分析几何圆度公差及其误差评定方法存在问题的基础上,指出现有圆度误差的各种测量评定方法都是基于几何公差的概念。提出应以统计圆度公差作为几何圆度公差的补充与发展。论述了统计圆度误差的意义及其测量评定方法。     

DVB/DOCSIS混合传输技术研究

探讨了有线电视网络DVB和DOCSIS协议在数据传输业务中的融合问题，提出了DVB视频和DOCSIS数据带内、带外两种混合传输技术，并在双向HFC网上实现了基于DOCSIS回传信道的视频点播系统。下行方向利用DVB-C发送视频节目，上行利用DOCSIS回传信道传送用户点播信息，实现交互。同时提出了基于DVB／DOCSIS带内融合技术的机顶盒设计方案。     

Improving the NPD Process by Applying Lean Principles: A Case Study

Abstract:

This article extends the new product development (NPD) literature by presenting a case study of a lean product development (LPD) transformation framework implemented at a U.S. based manufacturing firm. In a departure from typical LPD methods, in this article the design structure matrix and the cause and effect matrix are integrated into the lean transformation framework, allowing analysis of the underlying complexity of a product development (PD) system, and thus facilitating determination of the root causes of wasteful reworks. Several strategies to transform the current PD process into a lean process are discussed. Besides the two-phase improvement plan, a new organizational structure roadmap and a human resources plan are also suggested to support the recommended changes in the NPD process. The results of the first phase show a 32% reduction in PD cycle time due to the proposed NPD process. The article concludes with lessons learned and implications for engineering managers based on the case study.     

Backcalculation of pavement layer elastic modulus and thickness with measurement errors

This paper presents a backcalculation method for pavement layer elastic modulus and thickness. The effect of deflection measurement errors on the backcalculated results is also considered. The falling weight deflectometer (FWD) data are generated by applying a load to the pavement while calculating deflection at various fixed distances from the load centre. The measurement errors in FWD data are simulated by perturbing the theoretical deflections. Using these data, a backcalculation technique based on the improved genetic algorithm is proposed. In order to deal with the measurement errors, besides the common root mean square, a new objective function called area value with correction factor is introduced to the backcalculation algorithm. Numerical examples for two- and four-layer pavement structures are presented, which show the capability of the proposed method in backcalculation of pavement layer modulus and thickness.     

Numerical detection and continuation of saddle-node homoclinic bifurcations of codimension one and two

An extension of an existing truncated boundary-value method for the numerical continuation of connecting orbits is proposed to deal with homoclinic orbits to a saddle-node equilibrium. In contrast to previous numerical work by Schecter and Friedman and Doedel, the method is based on (linear) projection boundary conditions. These boundary conditions, with extra defining conditions for a saddle-node, allow the continuation of codimension-one curves of saddle-node homoclinic orbits. A new test function is motivated for detecting codimension-two points at which loci of saddle-nodes and homoclinic orbits become detached. Two methods for continuing such codim 2 points in three parameters are discussed. The numerical methods are applied to two example systems, modelling a DC Josephson junction and CO oxidation. For the former model, existing numerical results are recovered and extended; for the latter, new dynamical features are uncovered. All computations are performed using AUTO.