求解约束优化问题的ε-DE 算法

差分进化(differential evolution,简称DE)算法解决约束优化问题(constrained optimization problems,简称COPs)时通常采用可行解优先的比较规则,但是该方法不能利用种群中不可行解的信息.设计了可以利用不可行解信息的ε-DE算法.该算法通过构造一种比较准则,使得进化过程可以充分利用种群中优秀不可行解的信息.该准则通过引入种群约束允许放松程度的概念,在进化初始阶段使可行域边界上且拥有较优目标函数的不可行解进入种群;随着进化代数增加,种群约束允许放松程度不断减小,使得种群中不可行解数量减少,直到种群约束允许放松程度为0,种群完全由可行解组成.此外,还选择了一种改进的DE算法作为搜索算法,使得进化过程具有较快的收敛性.13个标准Benchmark函数实验仿真的结果表明:ε-DE算法是目前利用DE算法解决COPs问题中效果最好的.     

粉尘浓度测量装置利用涡流发生效应的优化研究

对现阶段电荷感应法测量粉尘浓度不够精确的问题进行了研究。依据涡流发生效应和文丘里效应,对现有环状粉尘传感器的测量管道进行改造。通过在文丘里管道的喉道中间段不同对称位置加入三片三角形涡流发生器,使气流通过管道时不易产生气体分离效应;在发生器尾部产生空气旋涡,使得环式静电传感器所感应的电荷量增大。利用PROE 5.0软件建立所设计管道3D模型,并利用Fluent软件对改进的管道进行气固两相流仿真,获得改进管道内粉尘颗粒速度分布云图及喉道段内粉尘的速度值。使用环形静电传感器的数学模型,分别计算了文丘里管道和改进管道在不同粒径比下的感应电荷量。结果表明:改进管道的结构在实验中使得感应电荷量大幅提高,在各阶段测得感应电荷量提高20%以上。通过试验可知,在各个浓度情况下测量精度得到了较大的提升,大气监测和燃煤电厂粉尘浓度在线测量精度都有所提高。     

改进自适应微分进化算法求解全局优化问题

针对微分进化(DE: differential evolution)算法在进化后期收敛速度慢,收敛精度低,易陷入局部最优解等缺点。本文通过改进DE的变异方程,并引入一种新的控制参数自适应策略,提出了一种改进自适应微分进化(IADE: improved adaptive differential evolution)算法。进化过程中IADE将根据个体适应值与父代平均适应值之间的关系动态地调整控制参数。同时,采用10个常用于优化算法比较的标准函数对IADE和其它改进DE算法进行对比试验,实验结果表明IADE算法不仅能够显著地提高收敛速度和收敛精度,而且具有非常好的鲁棒性,从而使得该算法能够满足过程优化的实时性、准确性以及稳定性要求。     

基于改进PSO和DE的混合算法

研究粒子群优化(PSO)算法和差分进化(DE)算法的优缺点,通过改进PSO算法并与DE算法混合,得到一种双种群的新型混合全局优化算法。经过对5个标准测试函数的大量实验计算表明,该算法能有效克服PSO算法和DE算法的缺陷,使寻优精度有较大改进,在高维情况下表现更加突出。     

采用改进DE算法优化高刺激率AEP的刺激序列

刺激序列的性质直接影响高刺激率听觉诱发电位(AEP)去卷积的性能,自动生成满足需求的刺激序列可以为高刺激率AEP的基础和应用研究带来极大便利。本文以刺激序列的抖动量为优化变量利用差分进化(DE)算法定义了约束条件下的目标函数。我们根据抖动量的变化范围,改进了DE搜索的变异算子实现搜索空间动态缩减。该方法可以方便地生成各种参数(包括刺激率、频带范围、扫程长度和采样频率)的低抖动率刺激序列。通过实测脑电信号合成的数据检验,本方法得到的各种刺激序列都取得了较好的效果。     

一种求解随机有限缓冲区流水线调度的混合差分进化算法

针对随机有限缓冲区流水线调度问题(Flow shop scheduling problem, FSSP), 提出混合差分进化(Differential evolution, DE)算法OHTDE, 用来最小化提前/拖后指标和最小化总体完成时间指标. OHTDE将DE和最优计算量分配(Optimal computing budget allocation, OCBA)技术以及假设检验(Hypothesis test, HT)有效结合. DE用于执行全局搜索和局部搜索; OCBA用于对有限计算量进行合理分配, 从而保证优质解得到较多仿真计算量, 提高了在噪声环境下获得优质解的置信度; HT用于在统计意义上比较解的性能, 从而一定程度上避免在解空间相近区域进行重复搜索. 进而, 对由OCBA和HT确定的优质解执行一种特殊的交叉操作, 加强DE的局部开发能力. 同时也采用有限马氏链理论对OHTDE的随机收敛性进行了分析. 仿真实验和算法比较验证了算法的有效性和鲁棒性.     

RBF网络的微分进化正交最小二乘算法

研究用于径向基函数(RBF)网络训练的一种微分进化正交最小二乘(DEOLS)算法。把微分进化(DE)算法的种群作为正交最小二乘(OLS)算法的候选径向基函数集合,利用OLS对DE的种群个体进行评断,以确定RBF网络的隐结点的数目、中心和宽度。该算法融合了DE的强大搜索能力和OLS的高效评断能力,隐结点的选择比OLS要合理,同时避免DE的复杂性。最后使用实验验证了该算法的优越性。     

基于模拟退火的混合差分进化算法及其在联合补货-配送集成优化中的应用

针对差分进化 (Differential evolution, DE)算法搜索效率较低和容易陷入局部最优的缺点,设计了基于SA的混合差分进化算法(SA-based Hybrid DE, SAHDE),以提高DE算法的全局寻优能力。该算法采用自适应变异算子和交叉算子,并结合模拟退火(Simulated Annealing, SA)算法的Metropolis 准则。首先通过标准测试函数对改进的SAHDE进行性能测试,证明了该算法比DE、自适应混合DE (Adaptive Hybrid DE, AHDE)和遗传算法(Genetic Algorithm, GA)更有效。进而将该算法运用到联合补货-配送集成优化(典型NP-hard)问题的求解中,通过大规模的算例分析,证实SAHDE在解决联合补货-配送优化问题比DE、AHDE和GA更有效。     

基于Memetic框架和改进DFP局部搜索的改进差分进化算法

为提高差分进化(DE)算法对性连续优化问题的求解能力、增强算法的适应性,提出了一种基于局部快速收敛算法的Memetic进化算法。改进了Davidon-Fletcher-Powell方法,得到了具有强搜索能力的局部搜索算法——NDFP。当进化过程中出现具有优秀特质的个体时,NDFP可以使该个体沿着局部最优解的方向快速进化。为综合NDFP和DE的优势,提出局部搜索的执行策略来平衡全局搜索和局部搜索的关系,使得NDFP对DE的优化具有更为广泛的适应性。在CEC2005和CEC2013 Benchmark的53个测试函数上的实验结果表明,同DE/current-to-best/1、SaDE和EPSDE算法相比,NDFP-DE进化算法具有更高的求解精度和稳定性。     

改进DE算法优化高刺激率AEP的刺激序列

刺激序列的性质直接影响高刺激率听觉诱发电位(Auditory evoked potentials,AEP)去卷积的性能,自动生成满足需求的刺激序列可以为高刺激率AEP的基础和应用研究带来极大便利.以刺激序列的抖动量为优化变量利用差分进化(Differential evolution,DE)算法定义了约束条件下的目标函数.根据抖动量的变化范围,改进了DE搜索的变异算子实现搜索空间动态缩减.该方法可以方便地生成各种参数(包括刺激率、频带范围、扫程长度和采样频率)的低抖动率刺激序列.通过实测脑电信号合成的数据检验,本方法得到的各种刺激序列都取得了较好的效果.     

多通道连续卷积神经网络脑电信号情绪识别

针对不同个体的脑电信号差异大且易受到环境因素影响的问题, 结合去基线干扰及脑电通道选择方法, 提出一种基于连续卷积神经网络的情绪分类识别算法. 首先进行基线信号的微分熵(differential entropy, DE)特征的选取研究, 将数据处理为多通道输入后使用连续卷积神经网络进行分类实验, 然后选择最佳电极个数. 实验结果表明, 将实验脑电信号微分熵与被试者实验脑电前一秒的基线信号微分熵的差值映射为二维矩阵后, 在频率维度组合为多通道的形式作为连续卷积神经网络的输入, 在22通道上唤醒度和效价的分类平均准确率为95.63%和95.13%, 接近32通道的平均准确率.     

Using differential evolution for fine tuning naïve Bayesian classifiers and its application for text classification

The Naive Bayes (NB) learning algorithm is simple and effective in many domains including text classification. However, its performance depends on the accuracy of the estimated conditional probability terms. Sometimes these terms are hard to be accurately estimated especially when the training data is scarce. This work transforms the probability estimation problem into an optimization problem, and exploits three metaheuristic approaches to solve it. These approaches are Genetic Algorithms (GA), Simulated Annealing (SA), and Differential Evolution (DE). We also propose a novel DE algorithm that uses multi-parent mutation and crossover operations (MPDE) and three different methods to select the final solution. We create an initial population by manipulating the solution generated by a method used for fine tuning the NB. We evaluate the proposed methods by using their resulted solutions to build NB classifiers and compare their results with the results of obtained from classical NB and Fine-Tuning Naïve Bayesian (FTNB) algorithm, using 53 UCI benchmark data sets. We name these obtained classifiers NBGA, NBSA, NBDE, and NB-MPDE respectively. We also evaluate the performance NB-MPDE for text-classification using 18 text-classification data sets, and compare its results with the results of obtained from FTNB, BNB, and MNB. The experimental results show that using DE in general and the proposed MPDE algorithm in particular are more convenient for fine-tuning NB than all other methods, including the other two metaheuristic methods (GA, and SA). They also indicate that NB-MPDE achieves superiority over classical NB, FTNB, NBDE, NBGA, NBSA, MNB, and BNB.     

Banerjee-GCD与Banerjee-Bound联合数组相关性测试

以Banerjee-GCD方法和Banerjee-Bound方法为基础，充分考虑了两者的测试结果之间的相互影响以及程序并行化对相关性测试的要求，从而提出了一个在统一的框架下利用Banerjee-GCD方法与Banerjee-Bound方法对不同的相关向量进行测试的联合数组相关性测试方法，该方法在保持执行时间效率的前提下提高了测试的精确性和结果的有效性，并且能够处理一部分非线性下标表达式的情况。     

基于FPGA和无线通信的密立根油滴测试系统设计

为了提高密立根油滴实验中仪器测量的精确性,改善传统实验教学指导模式,提出了一种基于现场可编程门阵列（FPGA）和无线通信的密立根油滴实验测试系统。利用 FPGA 高速地采集和存储油滴运动的控制电压和下落时间,并构建了基于 ZigBee 的分布式网络对数据进行无线传输。开发了基于 C/ S 架构的实验数据管理系统,实现了上位机实时获取、处理、存储并显示实验数据的功能,并利用 SQL Server 数据库实现数据的本地保存和远程服务器的同步。实践证明：该系统具有灵活高效的实时教学指导模式,且测量结果的平均相对误差小于0.06%,相较于传统实验仪器有效地提高了测量精确度。     

基于k-ary消减的快速最大公约数算法

最大公约数(GCD)算法中,对于输入B和C,利用Sorenson的右移k-ary消减思想提出一个算法用于寻找整数x和y,使得x和y满足Bx-Cy在二进制表示下低比特位部分为0,即Bx-Cy=0(mod 2^e),其中e是常数正整数。利用该算法能够右移较多比特并大规模降低循环次数。再结合模算法,提出了快速GCD算法,其输入规模为n比特时最差复杂度仍然是O(n²),但最好的情况下复杂度能达到O(nlog²n log logn)。实验数据表明,对于20万以上比特规模的输入,快速GCD算法比Binary GCD算法速度快;对100万比特规模的输入,快速GCD算法速度是Binary GCD算法的两倍。     

Method for finding multiple roots of polynomials

Conventional numerical methods for finding multiple roots of polynomials are inaccurate. The accuracy is unsatisfactory because the derivatives of the polynomial in the intermediate steps of the associated root-finding procedures are eliminated. Engineering applications require that this problem be solved. This work presents an easy-to-implement method that theoretically completely resolves the multiple-root issue. The proposed method adopts the Euclidean algorithm to obtain the greatest common divisor (GCD) of a polynomial and its fast derivative. The GCD may be approximate because of computational inaccuracy. The multiple roots are then deflated into simple ones and then determined by conventional root-finding methods. The multiplicities of the roots are accordingly calculated. A detailed derivation and test examples are provided to demonstrate the efficiency of this method.     

Calorimetric measurement of the standard enthalpy of formation of ZnSb at 298 K

After a critical review of the literature data on the standard enthalpy of formation of ZnSb, discrepancies between various experimental studies are highlighted. Moreover, experimental values disagree with values calculated by ab initio methods. As many of the experimental methods used suffer from some serious drawbacks, a new determination of the standard enthalpy of formation of ZnSb by an alternative calorimetric method, drop solution calorimetry in liquid tin, has been performed. Two different synthesis methods have been used to obtain a pure ZnSb phase and drop solution experiments were performed at 665 and at 974 K. The standard enthalpy of formation values derived from these experiments are − 6.1±2.5 kJ mol of atoms⁻¹ for the ZnSb sample prepared by ball milling and − 7.9±0.4 kJ mol of atoms⁻¹ for the ZnSb sample prepared by melting. These results are discussed and compared to literature data.     

Analysis of mutation vectors selection mechanism in differential evolution

Differential evolution (DE), which is one of the most popular evolution algorithms, has received much attention from researchers and engineers. In DE, mutation operation has a great impact on the performance of the algorithm. It generates mutant vectors by adding difference vectors to the base vector. Obviously, the chosen vectors in the mutation operation should not be equal to each other or to the target vector. This paper designs four experiments to analyze this session of DE and tries to determine whether this constraint is necessary. The theoretical analysis and experimental results show that without this constraint, the DE algorithm may perform better or at least not worse. Moreover, based on the experimental results, we can also summarize some rules for when and how to use this constraint to enhance the performance of the DE algorithm. This can help researchers to improve or apply the DE algorithm well.     

基于差分演化算法的粗糙集属性约简

属性约简是粗糙集(RS)理论的核心内容之一。应用差分演化(DE)算法求解最小属性约简是一个新的方向。对差分演化算法进行了改进,给出了一种新的适应值函数的定义形式;并在此基础上提出了基于差分演化算法的属性约简算法。最后利用多组数据对该算法进行了仿真实验,并与现有算法进行了比较分析。实验结果表明该算法是有效的,能快速地进行属性约简。     

B-spline neural network design using improved differential evolution for identification of an experimental nonlinear process

B-Spline Neural Network (BSNN), a type of basis function neural network, is trained by gradient-based methods which may fall into local minima during the learning procedure. To overcome the limitations encountered by gradient-based optimization methods, we propose differential evolution (DE) – an evolutionary computation methodology – which can provide a stochastic search to adjust the control points of a BSNN. In this paper, we propose six DE approaches using chaotic sequences based on logistic mapping to train a BSNN. Chaos describes the complex behavior of a nonlinear deterministic system. The application of chaotic sequences instead of random sequences in DE is a powerful strategy to diversify the DE population and improve the DE's performance in preventing premature convergence to local minima. The numerical results presented here indicate that chaotic DE was effective for building a good BSNN model for the nonlinear identification of an experimental nonlinear yo–yo motion control system.