蓄电池的荷电状态(state of charge,SOC)是表征电池当前剩余电量的重要参数。提出一种基于神经网络和主从式自适应无迹卡尔曼滤波(masterslaveadaptiveunscented Kalmanfilter,MS-UKF)算法的SOC估计方法。首先,建立蓄电池的戴维南(Theve...蓄电池的荷电状态(state of charge,SOC)是表征电池当前剩余电量的重要参数。提出一种基于神经网络和主从式自适应无迹卡尔曼滤波(masterslaveadaptiveunscented Kalmanfilter,MS-UKF)算法的SOC估计方法。首先,建立蓄电池的戴维南(Thevenin)二阶模型,针对开路电压与电池SOC之间的非线性关系,采用神经网络模型代替多项式模型,以提高拟合精度。根据实时测量数据,基于最小二乘法在线确定电池模型的参数。针对传统的扩展卡尔曼滤波(extendedKalmanfilter,EKF)和无迹卡尔曼滤波(unscented Kalman filter,UKF)方法存在噪声方差固定,会产生误差造成估计精度不高的问题,采用MS-AUKF算法。该算法的主滤波器用来估计系统状态,辅助滤波器用来估计噪声方差矩阵。算法每次迭代时更新系统模型的噪声方差,克服了传统卡尔曼滤波算法中,噪声方差初值人为设定可能导致滤波发散的缺点。仿真结果表明,相比于EKF、UKF算法,MSAUKF在估计电池SOC时具有更高的精确度和收敛速度。展开更多
The complexity of fire and smoke in terms of shape, texture, and color presents significant challenges for accurate fire and smoke detection. To address this, a YOLOv8-based detection algorithm integrated with the Con...The complexity of fire and smoke in terms of shape, texture, and color presents significant challenges for accurate fire and smoke detection. To address this, a YOLOv8-based detection algorithm integrated with the Convolutional Block Attention Module (CBAM) has been developed. This algorithm initially employs the latest YOLOv8 for object recognition. Subsequently, the integration of CBAM enhances its feature extraction capabilities. Finally, the WIoU function is used to optimize the network’s bounding box loss, facilitating rapid convergence. Experimental validation using a smoke and fire dataset demonstrated that the proposed algorithm achieved a 2.3% increase in smoke and fire detection accuracy, surpassing other state-of-the-art methods.展开更多
为弥补传统的广义线性模型(generalized linear model,GLM)方法的不足,并探索模式识别在运动脑科学当中的应用价值。使用支持向量机(support vector machine,SVM)模式识别算法,以低频振幅(fractional amplitude of low-frequency fluctu...为弥补传统的广义线性模型(generalized linear model,GLM)方法的不足,并探索模式识别在运动脑科学当中的应用价值。使用支持向量机(support vector machine,SVM)模式识别算法,以低频振幅(fractional amplitude of low-frequency fluctuations,fALFF)、局部一致性(regional homogeneity,ReHo)和度中心度(degree centrality,DC)作为学习特征,对射击运动组和滑冰运动组(分类1)、射击运动组和对照组(分类2)以及速滑运动组和对照组(分类3)之间进行二分类,并计算每一个脑区在分类算法当中的权重。使用留一交叉验证法计算分类正确率,使用总的准确率、接受者操作特性曲线(receiver operating characteristic curve,ROC)、以及预测准确率来衡量机器分类算法的优劣性。结果表明:分类1中SVM算法的正确率较高且分类效果更稳定,总的准确率(total accuracy,tACC)可以维持在96.67%以上,曲线下面积(area under curve,AUC)均为1,说明SVM算法对区分不同项目运动员脑静息态功能特征时更有优势;在分类2和分类3中,SVM算法效果取决于使用的指标。其中,使用fALFF或者综合使用三个静息态指标的分类效果较稳定(tACC均在80%以上,AUC均在0.88以上);小脑在分类1算法中占较多的权重,提示不同运动项目运动员的脑功能活动之间差异最明显的部位主要在小脑上。而分类2和3中,除了小脑,还有一些与运动执行和控制及其他功能活动相关的脑区参与了算法的构成。通过SVM分类算法的应用获得较为理想的结果,展示了模式识别方法在运动科学领域的应用价值。研究成果有助于体育科学研究者从新的角度更加全面地理解运动与脑的关系。展开更多
文摘The complexity of fire and smoke in terms of shape, texture, and color presents significant challenges for accurate fire and smoke detection. To address this, a YOLOv8-based detection algorithm integrated with the Convolutional Block Attention Module (CBAM) has been developed. This algorithm initially employs the latest YOLOv8 for object recognition. Subsequently, the integration of CBAM enhances its feature extraction capabilities. Finally, the WIoU function is used to optimize the network’s bounding box loss, facilitating rapid convergence. Experimental validation using a smoke and fire dataset demonstrated that the proposed algorithm achieved a 2.3% increase in smoke and fire detection accuracy, surpassing other state-of-the-art methods.