In this paper the Hamming distance is used to contr ol individual difference in the process of creating an original population, and a peak-depot is established to preserve information of different peak-points. So me n...In this paper the Hamming distance is used to contr ol individual difference in the process of creating an original population, and a peak-depot is established to preserve information of different peak-points. So me new methods are also put forward to improve optimization performance of genet ic algorithm, such as point-cast method and neighborhood search strategy around peak-points. The methods are used to deal with genetic operation besides of cr ossover and mutation, in order to obtain a global optimum solution and avoid GA ’s premature convergence. By means of many control rules and a peak-depot, the new algorithm carries out optimum search surrounding several peak-points. Alon g with evolution of individuals of population, the fitness of peak-points of pe ak-depot increases continually, and a global optimum solution can be obtained. The new algorithm searches around several peak-points, which increases the prob ability to obtain the global optimum solution to the best. By using some example s to test the modified genetic algorithm, the results indicate what we have done makes the modified genetic algorithm effectively to solve both of linear optimi zation problems and nonlinear optimization problems with restrictive functions.展开更多
为解决因排查效率低、数据更新不及时等因素导致低压配电网户变关系连接形式与实际不符的问题,提出一种基于角度分段线性近似(anglepiecewiselinearrepresentation,APLR)和改进密度峰值聚类(improved clustering by fast search find of...为解决因排查效率低、数据更新不及时等因素导致低压配电网户变关系连接形式与实际不符的问题,提出一种基于角度分段线性近似(anglepiecewiselinearrepresentation,APLR)和改进密度峰值聚类(improved clustering by fast search find of density peaks,ICFSFDP)相结合的户变关系识别方法。首先,根据电压曲线中相邻线段的角度变化量提取曲线的转折点,利用APLR对曲线进行自适应降维重构;随后,使用ICFSFDP算法对降维数据组展开聚类分析,在决策图中由拟合函数与坐标轴围成面积的最小值得到最优类簇数目,进而得到聚类和非聚类中心用户;最后,使用动态时间弯曲(dynamic time warping,DTW)距离计算聚类和非聚类中心用户之间的距离相似度,进而得到户变关系。将所提方法应用于模拟和真实数据中,均可证实所提方法的有效性。算例分析结果表明:该方法能够对时间间隔不同、不等维的序列进行分析,且不需要人为设定聚类算法的参数,户变关系识别准确率高。展开更多
Diamond search （DS） is an excellent fast block matching motion estimation （BMME） algorithm. In this paper, we propose an improved diamond search （IDS） algorithm, which revises the two search patterns of DS. The ...Diamond search （DS） is an excellent fast block matching motion estimation （BMME） algorithm. In this paper, we propose an improved diamond search （IDS） algorithm, which revises the two search patterns of DS. The proposed algorithm is compared with several mainstream algorithms. The simulation results show that the proposed algorithm over DS can be up to 20% gain on speedup on average, while maintain the similar or even better quality, both objectively and subjectively. The proposed algorithm is also competitive with other fast algorithms.展开更多
实际工程中,光伏阵列在随机变化的环境中会出现局部遮光的情况,从而导致光伏阵列的功率-电压特性曲线会呈现多峰值状态,传统的最大功率点跟踪(maximum power point tracking, MPPT)算法易陷入局部最优解,追踪速度和精准度无法得到满足...实际工程中,光伏阵列在随机变化的环境中会出现局部遮光的情况,从而导致光伏阵列的功率-电压特性曲线会呈现多峰值状态,传统的最大功率点跟踪(maximum power point tracking, MPPT)算法易陷入局部最优解,追踪速度和精准度无法得到满足。针对这一问题,提出一种基于布谷鸟搜索算法(cuckoo search algorithm, CS)和电导增量法(conductivity increment method, CI)结合的光伏MPPT算法,在算法前期利用布谷鸟搜索算法将大步长和小步长交替使用使得全局搜索能力增强,找到全局最大功率点所处区域附近;在后期,采用步长小、控制精度高的CI进行局部寻优,快速准确地锁定到最大功率点。在MATLAB/Simulink中搭建仿真模型,并与原始布谷鸟搜索算法和粒子群优化(particle swam optimization, PSO)算法进行比较。仿真结果表明,将CS与CI结合的算法使得收敛速度更快,精度更高,稳定状态时功率曲线的波动更小。展开更多
目的 针对旋转机械故障诊断过程中存在故障信号特征提取困难、故障诊断过程有标签数据较少、故障诊断准确率低等问题,提出自适应变分模态分解算法(Adaptive Variational Mode Decomposition,AVMD)与密度峰值算法优化的模糊C均值算法(Clu...目的 针对旋转机械故障诊断过程中存在故障信号特征提取困难、故障诊断过程有标签数据较少、故障诊断准确率低等问题,提出自适应变分模态分解算法(Adaptive Variational Mode Decomposition,AVMD)与密度峰值算法优化的模糊C均值算法(Clustering by Fast Search and Find of Density Peaks Optimizing Fuzzy C-Means,DPC-FCM)结合的无监督诊断方法。方法 首先,将多尺度排列熵与峭度相结合的综合系数作为适应度函数,对VMD算法的惩罚因子alpha和模态个数K进行参数寻优,提取分解后本征模态函数(Intrinsic Mode Function,IMF)的平均样本熵与平均模糊熵,并输入至聚类算法中。其次,提出利用密度峰值聚类算法确定FCM的初始聚类中心,降低聚类结果的随机性。结果 将提出的无监督故障诊断模型应用到滚动轴承试验信号中,实现了准确的故障诊断。结论 AVMD在故障提取方面具有优越性,同时DPC算法可以有效提高FCM算法无监督聚类的准确性,二者结合可以有效实现旋转机械故障的智能分类。展开更多
文摘In this paper the Hamming distance is used to contr ol individual difference in the process of creating an original population, and a peak-depot is established to preserve information of different peak-points. So me new methods are also put forward to improve optimization performance of genet ic algorithm, such as point-cast method and neighborhood search strategy around peak-points. The methods are used to deal with genetic operation besides of cr ossover and mutation, in order to obtain a global optimum solution and avoid GA ’s premature convergence. By means of many control rules and a peak-depot, the new algorithm carries out optimum search surrounding several peak-points. Alon g with evolution of individuals of population, the fitness of peak-points of pe ak-depot increases continually, and a global optimum solution can be obtained. The new algorithm searches around several peak-points, which increases the prob ability to obtain the global optimum solution to the best. By using some example s to test the modified genetic algorithm, the results indicate what we have done makes the modified genetic algorithm effectively to solve both of linear optimi zation problems and nonlinear optimization problems with restrictive functions.
文摘为解决因排查效率低、数据更新不及时等因素导致低压配电网户变关系连接形式与实际不符的问题,提出一种基于角度分段线性近似(anglepiecewiselinearrepresentation,APLR)和改进密度峰值聚类(improved clustering by fast search find of density peaks,ICFSFDP)相结合的户变关系识别方法。首先,根据电压曲线中相邻线段的角度变化量提取曲线的转折点,利用APLR对曲线进行自适应降维重构;随后,使用ICFSFDP算法对降维数据组展开聚类分析,在决策图中由拟合函数与坐标轴围成面积的最小值得到最优类簇数目,进而得到聚类和非聚类中心用户;最后,使用动态时间弯曲(dynamic time warping,DTW)距离计算聚类和非聚类中心用户之间的距离相似度,进而得到户变关系。将所提方法应用于模拟和真实数据中,均可证实所提方法的有效性。算例分析结果表明:该方法能够对时间间隔不同、不等维的序列进行分析,且不需要人为设定聚类算法的参数,户变关系识别准确率高。
基金Supported by the National High Technology Research and Development Program of China (2001AA132050-03)the Key Foundation of Ministry of Education of China (211CERS-10)
文摘Diamond search （DS） is an excellent fast block matching motion estimation （BMME） algorithm. In this paper, we propose an improved diamond search （IDS） algorithm, which revises the two search patterns of DS. The proposed algorithm is compared with several mainstream algorithms. The simulation results show that the proposed algorithm over DS can be up to 20% gain on speedup on average, while maintain the similar or even better quality, both objectively and subjectively. The proposed algorithm is also competitive with other fast algorithms.
文摘目的 针对旋转机械故障诊断过程中存在故障信号特征提取困难、故障诊断过程有标签数据较少、故障诊断准确率低等问题,提出自适应变分模态分解算法(Adaptive Variational Mode Decomposition,AVMD)与密度峰值算法优化的模糊C均值算法(Clustering by Fast Search and Find of Density Peaks Optimizing Fuzzy C-Means,DPC-FCM)结合的无监督诊断方法。方法 首先,将多尺度排列熵与峭度相结合的综合系数作为适应度函数,对VMD算法的惩罚因子alpha和模态个数K进行参数寻优,提取分解后本征模态函数(Intrinsic Mode Function,IMF)的平均样本熵与平均模糊熵,并输入至聚类算法中。其次,提出利用密度峰值聚类算法确定FCM的初始聚类中心,降低聚类结果的随机性。结果 将提出的无监督故障诊断模型应用到滚动轴承试验信号中,实现了准确的故障诊断。结论 AVMD在故障提取方面具有优越性,同时DPC算法可以有效提高FCM算法无监督聚类的准确性,二者结合可以有效实现旋转机械故障的智能分类。
