Ensemble Classifier Design Based on Perturbation Binary Salp Swarm Algorithm for Classification

Multiple classifier system exhibits strong classification capacity compared with single classifiers,but they require significant computational resources.Selective ensemble system aims to attain equivalent or better classification accuracy with fewer classifiers.However,current methods fail to identify precise solutions for constructing an ensemble classifier.In this study,we propose an ensemble classifier design technique based on the perturbation binary salp swarm algorithm(ECDPB).Considering that extreme learning machines(ELMs)have rapid learning rates and good generalization ability,they can serve as the basic classifier for creating multiple candidates while using fewer computational resources.Meanwhile,we introduce a combined diversity measure by taking the complementarity and accuracy of ELMs into account;it is used to identify the ELMs that have good diversity and low error.In addition,we propose an ECDPB with powerful optimizing ability;it is employed to find the optimal subset of ELMs.The selected ELMs can then be used to forman ensemble classifier.Experiments on 10 benchmark datasets have been conducted,and the results demonstrate that the proposed ECDPB delivers superior classification capacity when compared with alternative methods.

Multi-Strategy-Driven Salp Swarm Algorithm for Global Optimization

In response to the shortcomings of the Salp Swarm Algorithm (SSA) such as low convergence accuracy and slow convergence speed, a Multi-Strategy-Driven Salp Swarm Algorithm (MSD-SSA) was proposed. First, food sources or random leaders were associated with the current bottle sea squirt at the beginning of the iteration, to which Levy flight random walk and crossover operators with small probability were added to improve the global search and ability to jump out of local optimum. Secondly, the position mean of the leader was used to establish a link with the followers, which effectively avoided the blind following of the followers and greatly improved the convergence speed of the algorithm. Finally, Brownian motion stochastic steps were introduced to improve the convergence accuracy of populations near food sources. The improved method switched under changes in the adaptive parameters, balancing the exploration and development of SSA. In the simulation experiments, the performance of the algorithm was examined using SSA and MSD-SSA on the commonly used CEC benchmark test functions and CEC2017-constrained optimization problems, and the effectiveness of MSD-SSA was verified by solving three real engineering problems. The results showed that MSD-SSA improved the convergence speed and convergence accuracy of the algorithm, and achieved good results in practical engineering problems.

sALP、sNTX与下颌骨骨密度改变的相关性分析

目的:探讨下颌牙列缺损人群血清碱性磷酸酶(sALP)和I型胶原氨基末端肽(sNTX)随年龄变化及其下颌骨PIM指数之间的关系。方法:选用下颌牙列缺损健康患者60例,年龄20～79岁,分成6个年龄段,10岁为一年龄段,其中男性40名、女性20名。全景曲面断层采用PMI指数测量受试者,同时抽取血清测试骨生化指标sALP和sNTX。根据测试数据进行分析。结果:PMI和sALP、sNTX的性别间比较:4个指标均未见具有统计学意义的性别差别。下颌骨密度指标和骨生化指标与年龄关系:下颌骨密度指标(sPMI、iPMI)与年龄呈高度负相关(r值为-0.845,-0.917,P<0.001),骨生化指标(sALP、sNTX)与年龄呈正相关(r值为0.946,0.423,P<0.001)。sALP和sNTX在20～39岁年龄段较低,自40～49岁年龄段开始明显升高(P<0.001),sPMI和iPMI在50～59岁年龄段较30～39岁之前年龄段显著下降(P<0.001)。结论:年龄是下颌骨密度指数和骨生化指标变化的重要因素;sALP和sNTX反映下颌骨高骨转换状态,是下颌骨量丢失的重要因素之一;sPMI和iPMI是反映下颌骨密度改变的敏感指标。

基于Salp群算法的多堆燃料电池系统效率优化控制方法

为提高多堆燃料电池系统(multi-stack fuel cell system,MFCS)整体效率和维持母线电压的稳定,该文提出一种基于Salp群算法(Salp swarm algorithm,SSA)的MFCS效率优化控制方法。利用SSA算法的快速搜索能力实时优化系统整体效率,实现多个燃料电池间功率的合理分配,并通过下垂控制策略维持母线电压长期稳定。最后,在RT-LAB上搭建硬件在环(hardware-in-the-loop,HIL)仿真平台,与平均功率分配方法和Daisy链式功率分配方法进行对比分析,从功率、效率、容错能力三方面做实验测试。结果表明,所提控制方法既可以保证MFCS整体效率实时优化,稳定母线电压,也可以增强系统容错性,减少MFCS运行成本且能提高燃料电池耐久性和抗扰动能力。

Pilot Allocation Optimization Using Enhanced Salp Swarm Algorithm for Sparse Channel Estimation

Pilot pattern has a significant effect on the performance of channel estimation based on compressed sensing.However,because of the influence of the number of subcarriers and pilots,the complexity of the enumeration method is computationally impractical.The meta-heuristic algorithm of the salp swarm algorithm(SSA)is employed to address this issue.Like most meta-heuristic algorithms,the SSA algorithm is prone to problems such as local optimal values and slow convergence.In this paper,we proposed the CWSSA to enhance the optimization efficiency and robustness by chaotic opposition-based learning strategy,adaptive weight factor,and increasing local search.Experiments show that the test results of the CWSSA on most benchmark functions are better than those of other meta-heuristic algorithms.Besides,the CWSSA algorithm is applied to pilot pattern optimization,and its results are better than other methods in terms of BER and MSE.

Hybrid Chaotic Salp Swarm with Crossover Algorithm for Underground Wireless Sensor Networks

Resource management in Underground Wireless Sensor Networks(UWSNs)is one of the pillars to extend the network lifetime.An intriguing design goal for such networks is to achieve balanced energy and spectral resource utilization.This paper focuses on optimizing the resource efficiency in UWSNs where underground relay nodes amplify and forward sensed data,received from the buried source nodes through a lossy soil medium,to the aboveground base station.A new algorithm called the Hybrid Chaotic Salp Swarm and Crossover(HCSSC)algorithm is proposed to obtain the optimal source and relay transmission powers to maximize the network resource efficiency.The proposed algorithm improves the standard Salp Swarm Algorithm(SSA)by considering a chaotic map to initialize the population along with performing the crossover technique in the position updates of salps.Through experimental results,the HCSSC algorithm proves its outstanding superiority to the standard SSA for resource efficiency optimization.Hence,the network’s lifetime is prolonged.Indeed,the proposed algorithm achieves an improvement performance of 23.6%and 20.4%for the resource efficiency and average remaining relay battery per transmission,respectively.Furthermore,simulation results demonstrate that the HCSSC algorithm proves its efficacy in the case of both equal and different node battery capacities.

Optimization of Cognitive Radio System Using Self-Learning Salp Swarm Algorithm

CognitiveRadio(CR)has been developed as an enabling technology that allows the unused or underused spectrum to be used dynamically to increase spectral efficiency.To improve the overall performance of the CR systemit is extremely important to adapt or reconfigure the systemparameters.The Decision Engine is a major module in the CR-based system that not only includes radio monitoring and cognition functions but also responsible for parameter adaptation.As meta-heuristic algorithms offer numerous advantages compared to traditional mathematical approaches,the performance of these algorithms is investigated in order to design an efficient CR system that is able to adapt the transmitting parameters to effectively reduce power consumption,bit error rate and adjacent interference of the channel,while maximized secondary user throughput.Self-Learning Salp Swarm Algorithm(SLSSA)is a recent meta-heuristic algorithm that is the enhanced version of SSA inspired by the swarming behavior of salps.In this work,the parametric adaption of CR system is performed by SLSSA and the simulation results show that SLSSA has high accuracy,stability and outperforms other competitive algorithms formaximizing the throughput of secondary users.The results obtained with SLSSA are also shown to be extremely satisfactory and need fewer iterations to converge compared to the competitive methods.

Distribution and abundance of euphausiid larvae and salps during austral summers in Prydz Bay,Antarctica

The distribution and abundance of euphausiid larvae and salps was studied from samples collected in 2002 and 2006 from Prydz Bay.Antarctica. Larvae of Thysanoessa macrura and Euphausia superba were mainly distributed in the north of the continental shelf.T.macrura was more abundant and had a relatively wider distribution.In 2006,with ice having retreated and higher seawater temperatures and chlorophyll a levels,E.superba and T.macrura occurred in higher abundances and at more mature developmental stages.Euphausia crystallorophias was mainly distributed in the neritic region.In 2002,with severe ice conditions in the neritic region,abundance of E.crystallorophias was only 95.6 ind·（1000 m）^（-3）.In 2006 when a polynya existed,the abundance of E.crystallorophias reached 43966.6 ind·（1000 m）^（-3）.The population mainly consisted of metanauplius（MN） and calyptopis I（CD.Salps,mostly Salpa thompsoni,had a low abundance in Prydz Bay.In 2002,S.thompsoni was only found at one station in the north of the bay with an abundance of 10 ind·（1000 m）^（-3）.In 2006,S.thompsoni was found at three stations located near the continental slope and average abundance reached 146.7 ind·（1000 m）^（-3）.Environmental factors,such as the timing of ice melt,polynya formation and food concentration appear to have a marked effect on the distribution and abundance of euphausiid larvae and salps.

Improved Reptile Search Algorithm by Salp Swarm Algorithm for Medical Image Segmentation

This study proposes a novel nature-inspired meta-heuristic optimizer based on the Reptile Search Algorithm combed with Salp Swarm Algorithm for image segmentation using gray-scale multi-level thresholding,called RSA-SSA.The proposed method introduces a better search space to find the optimal solution at each iteration.However,we proposed RSA-SSA to avoid the searching problem in the same area and determine the optimal multi-level thresholds.The obtained solutions by the proposed method are represented using the image histogram.The proposed RSA-SSA employed Otsu’s variance class function to get the best threshold values at each level.The performance measure for the proposed method is valid by detecting fitness function,structural similarity index,peak signal-to-noise ratio,and Friedman ranking test.Several benchmark images of COVID-19 validate the performance of the proposed RSA-SSA.The results showed that the proposed RSA-SSA outperformed other metaheuristics optimization algorithms published in the literature.

A Boosted Communicational Salp Swarm Algorithm: Performance Optimization and Comprehensive Analysis

The Salp Swarm Algorithm (SSA) is a recently proposed swarm intelligence algorithm inspired by salps, a marine creature similar to jellyfish. Despite its simple structure and solid exploratory ability, SSA suffers from low convergence accuracy and slow convergence speed when dealing with some complex problems. Therefore, this paper proposes an improved algorithm based on SSA and adds three improvements. First, the Real-time Update Mechanism (RUM) underwrites the role of ensuring that excellent individual information will not be lost and information exchange will not lag in the iterative process. Second, the Communication Strategy (CMS), on the other hand, uses the multiplicative relationship of multiple individuals to regulate the exploration and exploitation process dynamically. Third, the Selective Replacement Strategy (SRS) is designed to adaptively adjust the variance ratio of individuals to enhance the accuracy and depth of convergence. The new proposal presented in this study is named RCSSSA. The global optimization capability of the algorithm was tested against various high-performance and novel algorithms at IEEE CEC 2014, and its constrained optimization capability was tested at IEEE CEC 2011. The experimental results demonstrate that the proposed algorithm can converge faster while obtaining better optimization results than traditional swarm intelligence and other improved algorithms. The statistical data in the table support its optimization capabilities, and multiple graphs deepen the understanding and analysis of the proposed algorithm.

Double Mutational Salp Swarm Algorithm:From Optimal Performance Design to Analysis

The Salp Swarm Algorithm(SSA)is a population-based Meta-heuristic Algorithm(MA)that simulates the behavior of a group of salps foraging in the ocean.Although the basic SSA has stable exploration capability and convergence speed,it still can fall into local optimum when solving complex optimization problems,which may be due to low utilization of population information and unbalanced exploration-to-exploitation ratio.Therefore,this study proposes a Double Mutation Salp Swarm Algorithm(DMSSA).In this study,a Cuckoo Mutation Strategy(CMS)and an Adaptive DE Mutation Strategy(ADMS)are introduced into the structure of the original SSA.The former mutation strategy is summarized as three basic operations:judgment,shuffling,and mutation.The purpose is to fully consider the information among search agents and use the differences between different search agents to participate in the update of positions,making the optimization process both diverse in exploration and minor in randomness.The latter strategy employs three basic operations:selection,mutation,and adaptation.As the follower part,some individuals do not blindly adopt the original follow method.Instead,the global optimal position and differences are considered,and the variation factor is adjusted adaptively,allowing the new algorithm to balance exploration,exploitation,and convergence efficiency.To evaluate the performance of DMSSA,comparisons are made with numerous algorithms on 30 IEEE CEC2014 benchmark functions.The statistical results confirm the better performance and significant difference of DMSSA in solving benchmark function tests.Finally,the applicability and scalability of DMSSA to optimization problems with constraints are further confirmed in three experiments on classical engineering design optimization problems.The source code of the proposed algorithm will be available at:https://github.com/ncjsq/Double-Mutational-Salp-Swarm-Algorithm.

基于双层优化VMD-LSTM的农村超短期电力负荷预测

稳定的供电是农村发展建设的有力保障,而电力负荷水平是建设效果的重要衡量标准,因此建立精确的负荷预测模型可以更准确直观显现电力负荷情况,为供电公司制定决策提供有力支撑。由于LSTM负荷预测模型在数据预测方面存在收敛性差、预测精度不高等问题,为提高模型的预测精度,提出一种基于双层优化VMD-LSTM的超短期电力负荷预测方法。首先提出麻雀算法优化变分模态分解(sparrow variational mode decomposition,SVMD),通过SVMD将原始数据转化为模态分量(intrinsic mode functions,IMF);其次采用改进樽海鞘群算法(association salp swarm algorithm,ASSSA)优化LSTM模型。通过引入4种策略增强标准樽海鞘算法优化能力;最后将各模态分量分别代入到新模型并进行叠加预测。选取辽宁省某市某乡村10kV变压器真实历史负荷数据,以均方根误差(RMSE)、平均绝对误差(MAE)、平均绝对百分比误差(MAPE)、拟合度(R^(2))作为评价指标,并与其他基础预测模型进行对比,结果表明,改进后的算法在计算精度、稳定性方面均优于其他基础预测模型。

路径规划问题的多策略改进樽海鞘群算法研究

针对移动机器人寻找最优路径问题,提出了一种融合无标度网络、自适应权重和黄金正弦算法变异策略的樽海鞘群算法BAGSSA(Adaptive Salp Swarm Algorithm with Scale-free of BA Network and Golden Sine)。首先,生成一个无标度网络来映射跟随者的关系,增强算法全局寻优的能力,在追随者进化过程中集成自适应权重ω,以实现算法探索和开发的平衡;同时选用黄金正弦算法变异进一步提高解的精度。其次,对12个基准函数进行仿真求解,实验数据表明平均值、标准差、Wilcoxon检验和收敛曲线均优于基本樽海鞘群和其他群体智能算法,证明了所提算法具有较高的寻优精度和收敛速度。最后,将BAGSSA应用于移动机器人路径规划问题中,并在两种测试环境中进行仿真实验,仿真结果表明,改进樽海鞘群算法较其他算法所寻路径更优,并具有一定理论与实际应用价值。

基于改进樽海鞘群算法的含瓦斯煤破裂过程信号特征识别

针对标准樽海鞘群算法存在的计算精度不足、易陷入局部停滞等缺陷,提出一种多策略融合的樽海鞘群算法。在初始化阶段,引入线性同余法随机发生器;利用野马算法优化樽海鞘领导者位置;采用金豺算法改进樽海鞘种群追随机制。通过测试函数寻优对比实验,证明多策略融合的樽海鞘群算法相比于其他智能算法在鲁棒性与稳定性方面均有显著提升。将多策略融合的樽海鞘群算法应用到含瓦斯煤破裂过程信号特征识别,实验结果表明:提出的含瓦斯煤破裂过程信号特征识别模型具有更好的表现,准确率可达93.33%,相比其他识别模型,识别率更高。

井下电力电缆故障定位研究

针对传统井下电力电缆故障定位方法依赖主观参数选择和抗噪性能较差,无法满足强噪声背景下井下电力电缆故障精确定位要求的问题,提出了一种基于樽海鞘群算法(SSA)优化变分模态分解(VMD)并结合改进型Teager能量算子(NTEO)的井下电力电缆故障定位方法。针对VMD在信号分解上存在的模态混叠、过分解和欠分解问题,采用SSA以模糊熵为适应度函数对VMD模态数K和惩罚因子α2个参数进行优化,得到更能反映故障特征信息的本征模态函数;采用NTEO对本征模态函数进行首波波头标定,得到首末两端的波头到达时刻,根据双端测距法得出故障位置。采用PSCAD/EMTDC进行井下电力电缆故障仿真,模拟具有强背景噪声的井下故障信号,结果表明:①在理想电流信号中加入9,12 dB噪声后,SSA-VMD的信噪比最低,皮尔逊相关系数最大,说明SSA-VMD在最大程度降噪的同时,能很好地保留信号的特征信息。②在不同过渡电阻下,SSA-VMD-NTEO的定位精度较高。③在不同故障相角下,SSA-VMD-NTEO在采样点上出现不同,但定位位置没有改变,依旧保持较高的定位精度。④在不同故障距离下,SSA-VMD-NTEO均能保证较高的定位精度。⑤在井下较大噪声和10 MHz采样频率下,SSA-VMD-NTEO较小波模极大值和VMD+NTEO 2种方法的定位精度具有明显优势。

基于超参数优化的电力负荷预测模型研究

电力负荷数据的多样性与复杂性,会导致负荷预测过程中出现超参数难以确定、拟合效果较差和预测精度不高等问题。针对以上问题,提出一种基于樽海鞘群算法的融入注意力机制的双向长短期记忆神经网络模型——SSA-AM-BiLSTM模型。该模型使用BiLSTM学习特征的内部变化规律,引入注意力机制为特征进行权重分配,并且利用樽海鞘群算法优化网络超参数。基于具体数据集进行的负荷预测仿真实验表明,相较于GRU、LSTM、AM-BiLSTM和PSO-AM-BiLSTM模型,所提出的SSA-AM-BiLSTM模型的MAPE分别减少了2.15%、1.93%、1.42%和0.45%,并且优化了拟合效果,显著提高了预测精度。

改进樽海鞘算法求解带时间窗的应急选址路径问题

目的为使应急物资及时高效地送到灾区,针对多目标应急选址-路径问题,在考虑灾区的时间窗及物资运输过程中道路安全的情况下,以最小化经济成本、最小化时间惩罚成本及最大化道路安全性为目标,构建多目标优化模型。同时,设计改进的樽海鞘算法求解问题,以验证模型的可行性和算法的有效性。方法根据模型的特征对樽海鞘算法进行改进,运用随机生成和贪心算法相结合的方式生成初始解,利用交叉算子和邻域搜索算子改进原始算法的位置更新操作,引入非支配排序遗传算法(NSGA-Ⅱ)的精英保留策略,以提高算法的性能。结果经过多个算例测试,该算法能快速获得一簇Pareto解,与基本樽海鞘算法进行对比后可知,改进后的算法性能更优越。结论对于灾后及时响应的应急选址路径问题,采用改进的樽海鞘算法具有一定优越性,并在多个目标权衡的情况下,可供决策者根据目标的偏好找到较满意的解,对于研究应急选址路径问题具有一定的参考价值。

考虑电堆性能一致性的燃料电池混合动力系统多目标优化能量管理方法

为优化燃料电池混合动力系统(fuel cell hybrid power system,FCHPS)并延长其使用寿命,该文提出一种考虑电堆性能一致性的多目标优化能量管理方法。该方法的目的是降低系统等效氢耗、提高燃料电池系统内电堆组运行效率的同时限制锂电池荷电状态(state of charge,SOC)波动。由于电堆组的性能会在实际运行过程中发生退化,因此该方法还考虑了电堆组的性能状态差异,通过限制性能较差电堆的运行压力,以延长系统寿命。为实现这一目的采用樽海鞘群算法(salpswarmalgorithm,SSA)对目标函数进行优化求解,得到系统最优功率分配。最后,基于RT-LAB半实物仿真平台,将所提方法与有限状态机控制方法进行对比,实验结果表明所提出的方法能够有效降低系统氢耗,提高电堆组效率的同时减缓性能较差电堆的功率波动,维持系统一致性,有利于系统长期稳定运行。

趋优变异反向学习的樽海鞘群与蝴蝶混合优化算法

针对蝴蝶优化算法(butterfly optimization algorithm,BOA)易陷入局部最优,且收敛速度慢和寻优精度低等问题,提出了一种趋优变异反向学习的樽海鞘群与蝴蝶混合优化算法(hybrid optimization algorithm for salp swarm and butterfly with reverse mutation towards optimization learning,OMSSBOA)。引入柯西变异对最优蝴蝶个体进行扰动,避免算法陷入局部最优;将改进的樽海鞘群优化算法(salp swarm algorithm,SSA)嵌入到BOA,平衡算法全局勘探和局部开采的比重,进而提高算法收敛速度;利用趋优变异反向学习策略扩大算法搜索范围并提升解的质量,进而提高算法的寻优精度。将改进算法在10种基准测试函数上进行仿真实验,结果表明,改进算法具有较好的寻优性能和鲁棒性。

CEEMD-VMD与参数优化SVM结合的托辊轴承故障诊断

针对托辊轴承工作环境复杂、提取故障特征困难等问题,提出一种基于互补集合经验模态分解(Complementary ensemble empirical mode decomposition, CEEMD)和变分模态分解(Variational modal decomposition, VMD)相结合的降噪方法。首先,利用CEEMD将采集到的信号进行分解,依据相关系数和峭度筛选分量并进行重构,生成新的信号;然后,利用VMD将新的信号进行再分解,并基于包络熵和包络谱峭度组合的复合指标优选本征模态分量(Intrinsic mode functions, IMF);最后,提取相应的特征输入樽海鞘群优化支持向量机(Salp swarm optimization support vector machine, SSO-SVM)模型完成故障诊断。实验结果表明:对于正常轴承、轴承内圈故障、轴承外圈故障三种情况,诊断准确率达97.78%。与单一降噪方法相比,该方法可以有效提高故障信号的信噪比,降噪效果明显。