YOLSO: You Only Look Small Object

Small object detection is challenging and far from satisfactory. Most general object detectors suffer from two critical issues with small objects: (1) Feature extractor based on classification network cannot express the characteristics of small objects reasonably due to insufficient appearance information of targets and a large amount of background interference around them. (2) The detector requires a much higher location accuracy for small objects than for general objects. This paper proposes an effective and efficient small object detector YOLSO to address the above problems. For feature representation, we analyze the drawbacks in previous backbones and present a Half-Space Shortcut(HSSC) module to build a background-aware backbone. Furthermore, a coarse-to-fine Feature Pyramid Enhancement(FPE) module is introduced for layer-wise aggregation at a granular level to enhance the semantic discriminability. For loss function, we propose an exponential L1 loss to promote the convergence of regression, and a focal IOU loss to focus on prime samples with high classification confidence and high IOU. Both of them significantly improves the location accuracy of small objects. The proposed YOLSO sets state-of-the-art results on two typical small object datasets, MOCOD and VeDAI, at a speed of over 200 FPS. In the meantime, it also outperforms the baseline YOLOv3 by a wide margin on the common COCO dataset.     

基于VMD和BSA-KELM的高陡边坡位移预测模型研究

边坡位移的时间序列曲线存在复杂的非线性特性,传统的预测模型精度不足以满足预测要求。为此提出了基于变分模态分解的鸟群优化-核极限学习机的预测模型,并用于河北省某水泥厂的边坡位移预测。该方法首先采用VMD把边坡位移序列分解为一系列的有限带宽的子序列,再对各子序列分别采用相空间重构并用核极限学习机预测,采用鸟群算法优化相空间重构的嵌入维度和KELM中惩罚系数和核参数三个数值,以取得最优预测模型。最后将各个子序列预测值叠加,得到边坡位移的最终预测值。结果表明：和KELM、BSA-KELM、EEMD-BSA-KELM模型相比,基于VMD的BSA-KELM预测精度更高,为边坡位移的预测提供一种有效的方法。     

Designing optical glasses by machine learning coupled with a genetic algorithm

Engineering new glass compositions have experienced a sturdy tendency to move forward from (educated) trial-and-error to data- and simulation-driven strategies. In this work, we developed a computer program that combines data-driven predictive models (in this case, neural networks) with a genetic algorithm to design glass compositions with desired combinations of properties. First, we induced predictive models for the glass transition temperature (T_g) using a dataset of 45,302 compositions with 39 different chemical elements, and for the refractive index (n_d) using a dataset of 41,225 compositions with 38 different chemical elements. Then, we searched for relevant glass compositions using a genetic algorithm informed by a design trend of glasses having high n_d (1.7 or more) and low T_g (500 °C or less). Two candidate compositions suggested by the combined algorithms were selected and produced in the laboratory. These compositions are significantly different from those in the datasets used to induce the predictive models, showing that the used method is indeed capable of exploration. Both glasses met the constraints of the work, which supports the proposed framework. Therefore, this new tool can be immediately used for accelerating the design of new glasses. These results are a stepping stone in the pathway of machine learning-guided design of novel glasses.     

A novel strategy based on salp swarm algorithm for extracting the maximum power of proton exchange membrane fuel cell

Cell temperature and water content of the membrane have a significant effect on the performance of fuel cells. The current-power curve of the fuel cell has a maximum power point (MPP) that is needed to be tracked. This study presents a novel strategy based on a salp swarm algorithm (SSA) for extracting the maximum power of proton-exchange membrane fuel cell (PEMFC). At first, a new formula is derived to estimate the optimal voltage of PEMFC corresponding to MPP. Then the error between the estimated voltage at MPP and the actual terminal voltage of the fuel cell is fed to a proportional-integral-derivative controller (PID). The output of the PID controller tunes the duty cycle of a boost converter to maximize the harvested power from the PEMFC. SSA determines the optimal gains of PID. Sensitivity analysis is performed with the operating fuel cell at different cell temperature and water content of the membrane. The obtained results through the proposed strategy are compared with other programmed approaches of incremental resistance method, Fuzzy-Logic, grey antlion optimizer, wolf optimizer, and mine-blast algorithm. The obtained results demonstrated high reliability and efficiency of the proposed strategy in extracting the maximum power of the PEMFC.     

基于改进SOA-VMD的磁声发射信号去噪算法

磁声发射（MAE）是铁磁性材料磁化过程中产生的声发射信号，在构件应力检测和微观损伤检测中有着广泛的应用。针对MAE信号非稳态、复杂性、衰减性等特点，提出海鸥算法结合变分模态分解（SOA-VMD）的去噪方法，为克服海鸥算法求解过程中易陷入局部最优解问题，利用柯西变异算子产生随机迭代过程，使改进算法即柯西变异海欧算法（CVSOA）跳出早熟收敛。采用以幅值谱熵为适应度函数，优化VMD算法中分解模态个数K和二次惩戒因子α两个参数，将含噪声的MAE信号进行VMD分解重构。经仿真信号和实际检测信号分析表明，改进后的CVSOA-VMD算法全局寻优能力和去噪性能优于传统的SOA-VMD算法，降噪后的MAE信号特征值对于不同应力下均方根、偏斜度特征值的重复性更好，可靠性更高。     

基于PBFT区块链技术的电网企业仓储系统

提出了一种基于实用拜占庭容错(PBFT)算法的区块链技术,首先对传统的实用拜占庭容错算法原理进行了阐述,该传统算法包含前期、需求、预准备、准备、确认、答复6个阶段,但传统算法具有实时性差、缺乏惩罚机制、带宽高的缺点。针对出现的这些问题,又对传统算法进行了改进,具体涉及记账节点、共识过程以及视图切换过程。通过测试进一步证明了该改进算法的实用性,并将该算法应用于电网企业中,构建的虚拟仓库实现了联储联备,降低了库存资金的耗费,并且提高了电网企业库存管理的效率。     

基于改进D?Lite遗传算法路径规划研究

针对遗传算法在移动机器人路径规划中易产生早熟现象和收敛速度慢的问题,提出了改进的D~* Lite遗传算法。该算法将D~* Lite算法和遗传算法相结合,通过引入碰撞系数和可视检测技术以提高路径安全性,寻找最短路径。在遗传算法设计中加入动态调整交叉与变异概率,以解决算法在路径规划中因陷入局部最优值而不能到达目标点的问题。最后,通过实验仿真可知:与蚁群算法和免疫遗传算法相比,改进的D~* Lite遗传算法执行效率高,可以快速规划出全局最优路径。     

基于神经网络和遗传算法的锭子弹性管性能优化

为得到减振弹性管对下锭胆的支承弹性和锭子高速运动下的稳定性等性能的最优匹配效率,依据减振弹性管的等效抗弯刚度及底部等效刚度系数公式,利用MatLab数值分析软件构建弹性管抗弯刚度和底部挠度数学模型。首先,结合Isight优化软件基于径向基神经网络构建其近似模型,且使精度达到可接受水平,并以模型的关键结构参数弹性模量、螺距、槽宽、壁厚为设计变量,结合遗传算法对弹性管抗弯刚度和底部挠度进行多目标优化设计,得到Pareto最优解集和Pareto前沿图,确定出减振弹性管结构工艺参数的优化方案。通过对优化数据进行分析发现,该方案在保证减振弹性管弹性的同时,其底部振幅明显减弱。     

喷油方式对斜齿轮齿面温度的影响

为探讨喷油方式对齿面温度的影响，以斜齿轮为研究对象，通过CFD仿真和试验研究不同喷油方式对斜齿轮齿面温度的影响。针对斜齿轮喷油润滑过程的特点，分析其齿面摩擦生热量，简化轮齿喷油润滑计算模型，并基于CFX进行齿面对流换热仿真研究，得到了不同喷油方式下的齿面温度；基于功率开放式斜齿轮试验平台，对不同喷油方式下的齿面温度进行了试验测量，与仿真计算结果进行对比分析。结果表明：在高速重载斜齿轮传动系统中，啮出侧喷油润滑的方式比啮入侧喷油润滑方式的冷却降温效果要好。     

Optimal design of nonlinear model predictive controller based on new modified multitracker optimization algorithm

The controller design for the robotic manipulator faces different challenges such as the system's nonlinearities and the uncertainties of the parameters. Furthermore, the tracking of different linear and nonlinear trajectories represents a vital role by the manipulator. This paper suggests an optimal design for the nonlinear model predictive control (NLMPC) based on a new improved intelligent technique and it is named modified multitracker optimization algorithm (MMTOA). The proposed modification of the MTOA is carried out based on opposition-based learning (OBL) and quasi OBL approaches. This modification improves the exploration behavior of the MTOA to prevent it from becoming trapped in a local optimum. The proposed method is applied on the robotic manipulator to track different linear and nonlinear trajectories. The NLMPC parameters are tuned by the MMTOA rather than the trial and error method of the designer. The proposed NLMPC based on MMTOA is compared with the original MTOA, genetic algorithm, and cuckoo search algorithm in literature. The superiority and effectiveness of the proposed controller are confirmed to track different linear and nonlinear trajectories. Furthermore, the robustness of the proposed method is emphasized against the uncertainties of the parameters.