Optimal power system operation using parallel processing system and PSO algorithm

In recent studies, PSO algorithm is applied to solve OPF problem. However, population based optimization method requires higher computing time to find optimal point. This shortcoming is overcome by a straightforward parallelization of PSO algorithm. The developed parallel PSO algorithm is implemented on a PC-cluster system with 8 Intel Pentium IV 2 GHz processors. The proposed approach has been tested on the test systems. The results showed that computing time of parallelized PSO algorithm can be reduced by parallel processing without losing the quality of solution.     

Efficient three‐dimensional parallel simulations of PHEMTs

An efficient 3D semiconductor device simulator is presented for a memory distributed multiprocessor environment using the drift–diffusion (D–D) approach for carrier transport. The current continuity equation and the Poisson equation, required to be solved iteratively in the D–D approach, are discretized using a finite element method (FEM) on an unstructured tetrahedral mesh. Parallel algorithms are employed to speed up the solution. The simulator has been applied to study a pseudomorphic high electron mobility transistor (PHEMT). We have carried out a careful calibration against experimental I–V characteristics of the 120 nm PHEMT achieving an excellent agreement. A simplification of the device buffer, which effectively reduces the mesh size, is investigated in order to speed up the simulations. The 3D device FEM simulator has achieved almost a linear parallel scalability for up to eight processors. Copyright © 2005 John Wiley & Sons, Ltd.     

A parallel genetic algorithm approach to solving the unitcommitment problem: implementation on the transputer networks

Through a constraint handling technique, this paper proposes a parallel genetic algorithm (GA) approach to solving the thermal unit commitment (UC) problem. The developed algorithm is implemented on an eight-processor transputer network, processors of which are arranged in master-slave and dual-direction ring structures, respectively. The proposed approach has been tested on a 38-unit thermal power system over a 24-hour period. Speed-up and efficiency for each topology with different number of processor are compared to those of the sequential GA approach. The proposed topology of dual-direction ring is shown to be well amenable to parallel implementation of the GA for the UC problem     

不对称三相P—Q分解法潮流的并行计算

本文利用解耦－补偿法模型，通过对三序间弱耦合的补偿，实现了三序间的解耦求解；构筑了不对称三相潮流的同步并行算法，实现了并行计算，并对计算结果进行了分析。     

一种基于边界元离散的导热问题几何边界识别算法

采用基于边界元法和共轭梯度法的反演算法求解二维导热边界识别问题,即几何反问题。正问题采用边界元法求解,反问题的解则在正问题的基础上通过共轭梯度法优化目标函数得到。考虑了未知边界为偏心圆、正弦曲线、椭圆等形状时的识别情况,讨论了初值、测量误差和测量点数等因素对反演解精度的影响。结果表明,该方法能将各种不规则边界识别出来,对初值和测量误差不敏感,测量点数适当减少对反演解精度影响不大,测量点离未知边界越近反演解精度越高。     

基于Spark的并行模拟退火算法求解TSP

模拟退火算法是求解无约束优化问题的有效方法,但求解旅行商问题时存在精度较差、容易陷入局部最优且收敛速度慢等缺点。为了改进上述问题,本文提出了一种基于Spark平台的并行模拟退火算法。修改模拟退火算法的降温函数,构造旅行商问题的解空间,采用大邻域搜索技术和2-opt算子增强局部搜索能力,引入OX交叉思想增强全局搜索能力,提出交叉协同试验并行策略与Spark平台并行实现。选取若干TSPLIB数据集进行仿真实验,对求解质量和运行时间两个方面进行测试,与其它Spark框架的并行算法进行对比实验。仿真结果表明,该算法求解精度有较大的提高,求解速度上对比其他算法提升3-10倍,能够有效求解旅行商问题。     

Parallel computation of 3-D electromagnetic scattering using finite elements

The finite element method (FEM) with local absorbing boundary conditions has been recently applied to compute electromagnetic scattering from large 3-D geometries. In this paper, we present details pertaining to code implementation and optimization. Various types of sparse matrix storage schemes are discussed and their performance is examined in terms of vectorization and net storage requirements. The system of linear equations is solved using a preconditioned biconjugate gradient (BCG) algorithm and a fairly detailed study of existing point and block preconditioners (diagonal and incomplete LU) is carried out. A modified ILU preconditioning scheme is also introducted which works better than the traditional version for our matrix systems. The parallelization of the iterative sparse solver and the matrix generation/assembly as implemented on the KSR1 multiprocessor is described and the interprocessor communication patterns are analysed in detail. Near-linear speed-up is obtained for both the iterative solver and the matrix generation/assembly phases. Results are presented for a problem having 224,476 unknowns and validated by comparison with measured data.     

基于Spark的并行k均值聚类模拟退火算法求解MMTSP

多起点闭回路多旅行商问题是旅行商问题的扩展。针对这个问题文中提出了一种基于Spark框架的并行k均值聚类模拟退火算法。该算法首先采用k均值聚类算法将所有城市分类,然后对应每个类建立一个旅行商问题,并通过一种改进的模拟退火算法对旅行商问题求解,MMTSP的解由这些类的最短路径之和计算得出。所提算法采用先聚类再执行模拟退火算法的求解策略可以极大的缩减模拟退火的搜索空间,并且由于Spark框架可以将聚类算法分好的若干类并行求解,从而更快的得到MMTSP问题的最优解。选取TSPLIB数据库中若干测试实例进行仿真实验,对求解精度和运行时间两个方面进行测试,与其他几种相关算法进行对比实验。实验结果表明,与目前FCMPGA、IPGA、IWO等算法相比,求解精度提高了5%~40%,求解效率上对比其他算法提升1~5倍,尤其在K值较大时表现更优。     

TLM modelling using an SIMD computer

A major limitation of the transmission-line matrix (TLM) method used to solve Maxwell's equations is the long computation time required. The TLM scattering calculations involved can, however, be viewed as parallel in nature. This paper describes an effort to reduce computational time by using an SIMD, DAP multiprocessor computer employed to solve a two-dimensional TLM electromagnetic field formulation. A parallel algorithm based on the TLM scattering algorithm is designed and implemented using FORTRAN- PLUS Enhanced on an AMT DAP 510 machine. Here the connectivity of the DAP is exploited to simulate the intrinsic scattering behaviour on which the TLM algorithm relies. The results show that parallel processing on an SIMD machine such as the DAP is advantageous, especially for higher-order mesh sizes.     

Application of parallel genetic algorithms to generation expansion planning using parallel processors

This paper presents an application of parallel genetic algorithms (PGA) to the optimal long-range generation expansion planning. The problem can be formulated as a combinatorial optimization problem that determines the order of introduced generation units at each interval of the year. The proposed method considers introduced power limits of each technology, maximum loads at each interval, and load duration curves at each interval. Appropriate string representation for the problem is presented. Binary and decimal coding and three selection methods are compared. The method is developed on a transputer that is one of the parallel processors. The feasibility of the proposed method is demonstrated using a typical expansion problem with four technologies and five intervals. The method is then compared with conventional dynamic programming and a simple genetic algorithm with promising results.     

Parallel processing for analogue fault diagnosis

The computer-aided testing (CAT) problem is inherently a large scale systems problem, it is essential to exploit whatever computational power is available to reduce the computational requirements for both on-line and off-line test processes. In particular, digital CAT algorithms often use some degree of parallel processing in their implementation. Therefore, the degree to which an analogue CAT algorithm can be implemented in parallel becomes a significant factor in determining its viability. In this paper, a parallel test algorithm and the associated architecture for the test system for analogue circuits and systems are proposed. the proposed test algorithm is applicable to testing multiple faults. Based on a combinatorial covering set approach, the conditions for the non-sequential execution of the test algorithm is analysed and the parallel processing to generate the test results is also addressed. Although this paper is motivated on the discussions of analogue fault diagnosis, the proposed parallel algorithm and the combinatorial covering approach can also be applied to digital testing.     

Creation of sparse boundary element matrices for 2-D andaxi-symmetric electrostatics problems using the bi-orthogonal Haarwavelet

In our previous paper we described the creation of sparse boundary element matrices that arise from Laplace's equation with mixed boundary conditions using an orthogonal wavelet basis. In this paper we examine the properties of wavelet expansions, and propose a way to construct even sparser matrices than the ones we obtained using the conventional Haar wavelet. We observe that the number of `vanishing moments' of a wavelet qualitatively determines the density of BEM matrices. Then we introduce a way of constructing a wavelet that has more vanishing moments than Haar's wavelet, but still retains a form that is very simple to implement. The time to solution employing the new wavelet is comparable to the Haar wavelet; both are faster than algorithms that rely on a conventional piecewise constant basis. However the quality of the solution is much better, for a given sparsity, the L₂ error of the source distribution is as much as an order of magnitude lower with the new bi-orthogonal wavelet     

一种暂态稳定并行仿真的改进算法及其加速比分析

提出了一种基于因子路径树网络划分的暂态稳定空间并行仿真的改进算法。采用新的网络划分性能评价指标、任务划分后续调整策略进行任务划分,使各处理器间计算负荷的分配更为合理;同时,还提出了该算法在理想状况下加速比的计算公式,分析了影响并行仿真计算效率的各个因素,并以某3872节点系统为例介绍了一种估计实现最大加速比所需处理器数目的方法。在Cluster1350集群系统上的算例表明,所提改进算法提高了仿真效率。     

Concurrent composite reliability evaluation using the state sampling approach

This paper presents a parallel methodology for composite reliability evaluation using Monte Carlo simulation by the state sampling approach. The methodology is based on coarse-grain parallelism, in which the adequacy analyses of the sampled system states are performed by different processors and the parallel process convergence is controlled by an asynchronous algorithm. The methodology was implemented on parallel and distributed processing environments: a ten-node IBM RS/6000 SP distributed memory parallel computer and a network of eight IBM RS/6000 43P workstations. The results obtained in tests with actual power system models show high speed-up and efficiency on both platforms and very good scalability.     

OPTIMIZATION ISSUES IN FINITE ELEMENT CODES FOR SOLVING OPEN DOMAIN 3D ELECTROMAGNETIC SCATTERING AND CONFORMAL ANTENNA PROBLEMS

The first part of the paper presents the implementation and performance of a new absorbing boundary condition (ABC) for truncating finite element meshes. This ABC can be applied conformally to the surface of the structure for scattering and antenna radiation calculations. Consequently, the computational domain is reduced dramatically, thus allowing the simulation of much larger structures, and results are presented for three-dimensional bodies. The latter part of the paper discusses optimization issues relating to the solver's CPU speed on parallel and vector processors. It is shown that a jagged diagonal storage scheme leads to a four-fold increase in the FLOP rate of the code, and a standard matrix profile reduction algorithm substantially reduces the inter-processor communication.     

3D Parallel Simulations of Fluctuation Effects in pHEMTs

The use of 3D simulations is essential in order to study the effects of fluctuations when devices are scaled to deep submicron dimensions. A 3D drift-diffusion device simulator has been developed to effectively simulate pseudomorphic high electron mobility transistors (pHEMTs) on a distributed memory multiprocessor computer. The drift-diffusion equations are discretized using a finite element method on an unstructured tetrahedral mesh. The obtained set of equations is solved in parallel on an arbitrary number of processors using the message-passing interface library. We have applied our simulator to a 120 nm pHEMT based on the Al_0.3Ga_0.7As/In_0.2Ga_0.8As interface and carried out a calibration to real experimental data.     

Parallel-in-time implementation of transient stability simulationson a transputer network

The most time consuming computer simulation in power system studies is the transient stability analysis. Parallel processing has been applied for time domain simulations of power system transient behavior. In this paper, a parallel implementation of an algorithm based on Shifted-Picard dynamic iterations is presented. The main idea is that a set of nonlinear Differential Algebraic Equations (DAEs), which describes the system, can be solved by the iterative solution of a linear set of DAEs. The time behavior of the linear set of differential equations can be obtained by the evaluation of the convolution integral. In the parallel-in-time implementation of the proposed algorithm, each processor is devoted to the evaluation of the complete set of variables relative to each time step. The quadrature formula, adopted for the integral evaluation, can be easily parallelized by using a number of processors equal to the number of time steps. The algorithm, implemented on a transputer network with 32 Inmos T800/20 adopting a uni-directional ring topology, has been tested on standard power systems     

Parallel FETI‐DP algorithm for efficient simulation of large‐scale EM problems

An efficient parallelization of the dual‐primal finite‐element tearing and interconnecting (FETI‐DP) algorithm is presented for large‐scale electromagnetic simulations. As a nonoverlapping domain decomposition method, the FETI‐DP algorithm formulates a global interface problem, whose iterative solution is accelerated with a solution of a global corner problem. To achieve a good load balance for parallel computation, the original computational domain is decomposed into subdomains with similar sizes and shapes. The subdomains are then distributed to processors based on their close proximity to minimize inter‐processor communication. The parallel generalized minimal residual method, enhanced with the iterative classical Gram‐Schmidt orthogonalization scheme to reduce global communication, is adopted to solve the global interface problem with a fast convergence rate. The global corner‐related coarse problem is solved iteratively with a parallel communication‐avoiding biconjugate gradient stabilized method to minimize global communication, and its convergence is accelerated by a diagonal preconditioner constructed from the coarse system matrix. To alleviate neighboring communication overhead, the non‐blocking communication approach is employed in both generalized minimal residual and communication‐avoiding biconjugate gradient stabilized iterative solutions. Three numerical examples are presented to demonstrate the accuracy, scalability, and capability of the proposed parallel FETI‐DP algorithm for electromagnetic modeling of general objects and antenna arrays. Copyright © 2016 John Wiley & Sons, Ltd.     

Fault Location Observability Using Phasor Measurement Units in a Power Network Through Deterministic and Stochastic Algorithms

This article studies deterministic and stochastic algorithms for placing minimum number of phasor measurement units (PMUs) in a power system in order to locate any fault in the power system. The optimization problem is initially formulated in a mixed integer linear programing framework with binary-valued variables as well as in a binary integer linear programing model. Then, the optimization problem is formulated as an equivalent non-linear programing model, minimizing a quadratic objective function subject to equality non-linear constraints defined over a bounded and closed set. The problem is solved by using a Sequential Quadratic Programming algorithm. The non-linear program is illustrated with a 7-bus test system. Also, stochastic algorithms such as binary-coded genetic algorithm and particle swarm optimization have been implemented in solving the optimal PMU placement under fault condition. The accuracy of suggested algorithms is independent from the fault type and its resistance. The optimization models are applied to the IEEE systems. The numerical results indicate that the proposed algorithms locate minimizers at the optimal objective function value in complete agreement with those obtained by branch-and-bound algorithms.