Feature selection(FS)is essential in machine learning(ML)and data mapping by its ability to preprocess high-dimensional data.By selecting a subset of relevant features,feature selection cuts down on the dimension of t...Feature selection(FS)is essential in machine learning(ML)and data mapping by its ability to preprocess high-dimensional data.By selecting a subset of relevant features,feature selection cuts down on the dimension of the data.It excludes irrelevant or surplus features,thus boosting the performance and efficiency of the model.Particle Swarm Optimization(PSO)boasts a streamlined algorithmic framework and exhibits rapid convergence traits.Compared with other algorithms,it incurs reduced computational expenses when tackling high-dimensional datasets.However,PSO faces challenges like inadequate convergence precision.Therefore,regarding FS problems,this paper presents a binary version enhanced PSO based on the Support Vector Machines(SVM)classifier.First,the Sand Cat Swarm Optimization(SCSO)is added to enhance the global search capability of PSO and improve the accuracy of the solution.Secondly,the Latin hypercube sampling strategy initializes populations more uniformly and helps to increase population diversity.The last is the roundup search strategy introducing the grey wolf hierarchy idea to help improve convergence speed.To verify the capability of Self-adaptive Cooperative Particle Swarm Optimization(SCPSO),the CEC2020 test suite and CEC2022 test suite are selected for experiments and applied to three engineering problems.Compared with the standard PSO algorithm,SCPSO converges faster,and the convergence accuracy is significantly improved.Moreover,SCPSO’s comprehensive performance far exceeds that of other algorithms.Six datasets from the University of California,Irvine(UCI)database were selected to evaluate SCPSO’s effectiveness in solving feature selection problems.The results indicate that SCPSO has significant potential for addressing these problems.展开更多
With the large-scale integration of renewable energy sources into the grid,distribution networks are increasingly challenged by issues related to renewable energy accommodation and the mainte-nance of power quality st...With the large-scale integration of renewable energy sources into the grid,distribution networks are increasingly challenged by issues related to renewable energy accommodation and the mainte-nance of power quality stability.To address the challenge that existing partitioning methods are inad-equate for the planning and operation needs of active distribution networks under frequently changing power flow conditions,a three-stage dynamic partitioning approach is proposed based on an im-proved sand cat swarm optimization(ISCSO)algorithm.Firstly,a comprehensive dynamic partitio-ning index is developed by integrating both structural and functional metrics,including modularity,voltage regulation capability,and regional renewable energy accommodation capacity.Secondly,to overcome the limitations of the conventional sand cat swarm optimization,namely its weak global ex-ploration ability and tendency to fall into local optima in the later optimization stages,chaotic map-ping is employed to initialize a uniformly distributed population.A nonlinear sensitivity mechanism is introduced to balance global exploration and local exploitation,alongside the design of a particle encoding and position updating scheme tailored for dynamic partitioning.Furthermore,a‘state re-tention-local adjustment-global reconstruction’partitioning structure is developed.To avoid unnec-essary partition changes under minor source-load fluctuations,the concept of overlapping nodes is introduced,enabling fine-tuned adjustments under such conditions.Finally,two experimental sce-narios are designed to validate the proposed method.Simulation results demonstrate strong electrical coupling performance and show that the method enhances voltage regulation and renewable energy integration capabilities across regions.展开更多
Skin cancer is one of the most dangerous cancer.Because of the high melanoma death rate,skin cancer is divided into non-melanoma and melanoma.The dermatologist finds it difficult to identify skin cancer from dermoscop...Skin cancer is one of the most dangerous cancer.Because of the high melanoma death rate,skin cancer is divided into non-melanoma and melanoma.The dermatologist finds it difficult to identify skin cancer from dermoscopy images of skin lesions.Sometimes,pathology and biopsy examinations are required for cancer diagnosis.Earlier studies have formulated computer-based systems for detecting skin cancer from skin lesion images.With recent advancements in hardware and software technologies,deep learning(DL)has developed as a potential technique for feature learning.Therefore,this study develops a new sand cat swarm optimization with a deep transfer learning method for skin cancer detection and classification(SCSODTL-SCC)technique.The major intention of the SCSODTL-SCC model lies in the recognition and classification of different types of skin cancer on dermoscopic images.Primarily,Dull razor approach-related hair removal and median filtering-based noise elimination are performed.Moreover,the U2Net segmentation approach is employed for detecting infected lesion regions in dermoscopic images.Furthermore,the NASNetLarge-based feature extractor with a hybrid deep belief network(DBN)model is used for classification.Finally,the classification performance can be improved by the SCSO algorithm for the hyperparameter tuning process,showing the novelty of the work.The simulation values of the SCSODTL-SCC model are scrutinized on the benchmark skin lesion dataset.The comparative results assured that the SCSODTL-SCC model had shown maximum skin cancer classification performance in different measures.展开更多
光伏阵列通常被安装在恶劣的室外环境中,因此在运行过程中易发生故障。为了准确识别光伏阵列的故障类型,提出沙猫群优化支持向量机(sand cat swarm optimization support vector machine,SCSO-SVM)用于光伏组件故障识别,且对比支持向量...光伏阵列通常被安装在恶劣的室外环境中,因此在运行过程中易发生故障。为了准确识别光伏阵列的故障类型,提出沙猫群优化支持向量机(sand cat swarm optimization support vector machine,SCSO-SVM)用于光伏组件故障识别,且对比支持向量机(support vector machine,SVM)、粒子群优化支持向量机(particle swarm optimized support vector machine,PSO-SVM)、遗传优化支持向量机(genetic optimized support vector machine,GA-SVM)、麻雀优化支持向量机(sparrow optimized support vector machine,SSA-SVM)、灰狼优化支持向量机(gray wolf optimized support vector machine,GWO-SVM)和鲸鱼优化支持向量机(whale optimized support vector machine,WOA-SVM)算法。首先,六种SVM混合算法都克服了SVM诊断结果易受参数初始值影响的缺点,识别精度相较传统SVM算法都有所提升,但是识别时间都增加。其次,7种算法中SCSO-SVM识别效果最好,克服了SVM易受参数初始值的影响,相较SVM识别精度提高了约9.4594%;是因为更能有效找到SVM惩罚因子和核函数参数。然后,对于同一种算法而言,算法的识别精度是随输入特征减少而降低的,是因为输入特征越少,越不能有效表征光伏组件在不同故障类型下的输出属性。但算法的识别时间却不是随输入特征减少而减短。所以选取合适的输入特征才能兼顾算法的故障识别准确率和效率。最后,发现七种算法的识别效果依赖于数据集的影响。原因可能是各个算法参数选择过多导致泛化性有差异,且依赖参数初始值选择。展开更多
基金supported by the Fundamental Research Funds for the Central Universities of China(No.300102122105)the Natural Science Basic Research Plan in Shaanxi Province of China(2023-JC-YB-023).
文摘Feature selection(FS)is essential in machine learning(ML)and data mapping by its ability to preprocess high-dimensional data.By selecting a subset of relevant features,feature selection cuts down on the dimension of the data.It excludes irrelevant or surplus features,thus boosting the performance and efficiency of the model.Particle Swarm Optimization(PSO)boasts a streamlined algorithmic framework and exhibits rapid convergence traits.Compared with other algorithms,it incurs reduced computational expenses when tackling high-dimensional datasets.However,PSO faces challenges like inadequate convergence precision.Therefore,regarding FS problems,this paper presents a binary version enhanced PSO based on the Support Vector Machines(SVM)classifier.First,the Sand Cat Swarm Optimization(SCSO)is added to enhance the global search capability of PSO and improve the accuracy of the solution.Secondly,the Latin hypercube sampling strategy initializes populations more uniformly and helps to increase population diversity.The last is the roundup search strategy introducing the grey wolf hierarchy idea to help improve convergence speed.To verify the capability of Self-adaptive Cooperative Particle Swarm Optimization(SCPSO),the CEC2020 test suite and CEC2022 test suite are selected for experiments and applied to three engineering problems.Compared with the standard PSO algorithm,SCPSO converges faster,and the convergence accuracy is significantly improved.Moreover,SCPSO’s comprehensive performance far exceeds that of other algorithms.Six datasets from the University of California,Irvine(UCI)database were selected to evaluate SCPSO’s effectiveness in solving feature selection problems.The results indicate that SCPSO has significant potential for addressing these problems.
基金Supported by the Technology Project of State Grid Corporation Headquarters(No.5100-202322029A-1-1-ZN)the 2024 Youth Science Foun-dation Project(No.62303006).
文摘With the large-scale integration of renewable energy sources into the grid,distribution networks are increasingly challenged by issues related to renewable energy accommodation and the mainte-nance of power quality stability.To address the challenge that existing partitioning methods are inad-equate for the planning and operation needs of active distribution networks under frequently changing power flow conditions,a three-stage dynamic partitioning approach is proposed based on an im-proved sand cat swarm optimization(ISCSO)algorithm.Firstly,a comprehensive dynamic partitio-ning index is developed by integrating both structural and functional metrics,including modularity,voltage regulation capability,and regional renewable energy accommodation capacity.Secondly,to overcome the limitations of the conventional sand cat swarm optimization,namely its weak global ex-ploration ability and tendency to fall into local optima in the later optimization stages,chaotic map-ping is employed to initialize a uniformly distributed population.A nonlinear sensitivity mechanism is introduced to balance global exploration and local exploitation,alongside the design of a particle encoding and position updating scheme tailored for dynamic partitioning.Furthermore,a‘state re-tention-local adjustment-global reconstruction’partitioning structure is developed.To avoid unnec-essary partition changes under minor source-load fluctuations,the concept of overlapping nodes is introduced,enabling fine-tuned adjustments under such conditions.Finally,two experimental sce-narios are designed to validate the proposed method.Simulation results demonstrate strong electrical coupling performance and show that the method enhances voltage regulation and renewable energy integration capabilities across regions.
基金supported by the Technology Development Program of MSS [No.S3033853]by the National University Development Project by the Ministry of Education in 2022.
文摘Skin cancer is one of the most dangerous cancer.Because of the high melanoma death rate,skin cancer is divided into non-melanoma and melanoma.The dermatologist finds it difficult to identify skin cancer from dermoscopy images of skin lesions.Sometimes,pathology and biopsy examinations are required for cancer diagnosis.Earlier studies have formulated computer-based systems for detecting skin cancer from skin lesion images.With recent advancements in hardware and software technologies,deep learning(DL)has developed as a potential technique for feature learning.Therefore,this study develops a new sand cat swarm optimization with a deep transfer learning method for skin cancer detection and classification(SCSODTL-SCC)technique.The major intention of the SCSODTL-SCC model lies in the recognition and classification of different types of skin cancer on dermoscopic images.Primarily,Dull razor approach-related hair removal and median filtering-based noise elimination are performed.Moreover,the U2Net segmentation approach is employed for detecting infected lesion regions in dermoscopic images.Furthermore,the NASNetLarge-based feature extractor with a hybrid deep belief network(DBN)model is used for classification.Finally,the classification performance can be improved by the SCSO algorithm for the hyperparameter tuning process,showing the novelty of the work.The simulation values of the SCSODTL-SCC model are scrutinized on the benchmark skin lesion dataset.The comparative results assured that the SCSODTL-SCC model had shown maximum skin cancer classification performance in different measures.
文摘光伏阵列通常被安装在恶劣的室外环境中,因此在运行过程中易发生故障。为了准确识别光伏阵列的故障类型,提出沙猫群优化支持向量机(sand cat swarm optimization support vector machine,SCSO-SVM)用于光伏组件故障识别,且对比支持向量机(support vector machine,SVM)、粒子群优化支持向量机(particle swarm optimized support vector machine,PSO-SVM)、遗传优化支持向量机(genetic optimized support vector machine,GA-SVM)、麻雀优化支持向量机(sparrow optimized support vector machine,SSA-SVM)、灰狼优化支持向量机(gray wolf optimized support vector machine,GWO-SVM)和鲸鱼优化支持向量机(whale optimized support vector machine,WOA-SVM)算法。首先,六种SVM混合算法都克服了SVM诊断结果易受参数初始值影响的缺点,识别精度相较传统SVM算法都有所提升,但是识别时间都增加。其次,7种算法中SCSO-SVM识别效果最好,克服了SVM易受参数初始值的影响,相较SVM识别精度提高了约9.4594%;是因为更能有效找到SVM惩罚因子和核函数参数。然后,对于同一种算法而言,算法的识别精度是随输入特征减少而降低的,是因为输入特征越少,越不能有效表征光伏组件在不同故障类型下的输出属性。但算法的识别时间却不是随输入特征减少而减短。所以选取合适的输入特征才能兼顾算法的故障识别准确率和效率。最后,发现七种算法的识别效果依赖于数据集的影响。原因可能是各个算法参数选择过多导致泛化性有差异,且依赖参数初始值选择。
