基于Unity3D的工作面液压支架群组空间支护姿态数字孪生重构方法

煤炭是我国的关键能源,保障煤炭资源安全高效开采对我国具有重要战略意义。我国煤炭资源以井工开采为主,生产环境较恶劣且工人劳动强度大,智能化开采因其减人提效理念内涵成为我国煤炭行业高质量发展的必由之路。液压支架是井工开采关键支护装备,其支护性能同时受本体支护姿态及群组关联姿态影响。受井下复杂开采条件影响,液压支架支护姿态信息获取困难,因此如何智能精准地感知液压支架群组空间支护姿态是制约智能化建设的重大难题之一。为精准描述工作面液压支架群组空间支护姿态,首先分析了液压支架群组空间连接关系,建立了基于多杆双驱的液压支架自姿态感知模型和基于空间多线协同定姿的架群姿态感知模型,分析了不同求解策略对自姿态模型解算性能影响。随后借助三维互动平台Unity3D搭建了工作面液压支架群组数字孪生环境,开发了基于工作面真实场景数据驱动的可视化系统。最后,依托山东某矿对所述方法开展了工程验证,结果表明提出的液压支架群组空间支护姿态孪生重构方法能精确映射实际液压支架群组真实空间支护位态,解决了液压支架支护本体支护姿态信息易受现场振动干扰和支架群组关联姿态信息缺失难题。论文所提方法为工作面多智能体运行行为全场景虚拟再现及工作面智能化建设提供思路借鉴。

基于改进Cascade Mask R-CNN与协同注意力机制的群猪姿态识别

猪体姿态识别有助于实现猪只健康状况预警、预防猪病爆发,是当前研究热点。针对复杂场景下群猪容易相互遮挡、粘连,姿态识别困难的问题,该研究提出一种实例分割与协同注意力机制相结合的两阶段群猪姿态识别方法。首先,以Cascade Mask R-CNN作为基准网络,结合HrNetV2和FPN模块构建猪体检测与分割模型,解决猪体相互遮挡、粘连等问题,实现复杂环境下群猪图像的高精度检测与分割;在上述提取单只猪基础上,构建了基于协同注意力机制(coordinate attention,CA)的轻量级猪体姿态识别模型(CA-MobileNetV3),实现猪体姿态的精准快速识别。最后,在自标注数据集上的试验结果表明,在猪体分割与检测环节,该研究所提模型与MaskR-CNN、MSR-CNN模型相比,在AP_(0.50)、AP_(0.75)、AP_(0.50:0.95)和AP_(0.5:0.95-large)指标上最多提升了1.3、1.5、6.9和8.8个百分点,表现出最优的分割与检测性能。而在猪体姿态识别环节,所提CA-MobileNetV3模型在跪立、站立、躺卧、坐立4种姿态类上的准确率分别为96.9%、99.1%、99.5%和98.6%,其性能优于主流的MobileNetV3、ResNet50、DenseNet121和VGG16模型,由此可知,该研究模型在复杂环境下群猪姿态识别具有良好的准确性和有效性,为实现猪体姿态的精准快速识别提供方法支撑。

IMPROVED PARTICLE SWARM OPTIMIZATION ALGORITHM FOR INTELLIGENTLY SETTING UAV ATTITUDE CONTROLLER PARAMETERS

An improved particle swarm optimization （PSO） algorithm is investigated in the optimization of the attitude controller parameters of unmanned aerial vehicle （UAV）. Considering the stagnation phenomenon in the later phase of the basic PSO algorithm caused by the diversity scarcity of particles, a modified PSO algorithm is presented. For the basic PSO algorithm, the velocity of each particle is adjusted according to the inertia motion, the swarm previous best position and its own previous best position. However, in the improved PSO algorithm, each particle only learns from another randomly selected particle with higher performance, besides keeping the inertia motion. The inertia weight of the improved PSO algorithm is a random number. The modification decreases the uncertain parameters of the algorithm, simplifies the learning mechanism of the particle, and enhances the diversity of the swarm. Furthermore, a UAV attitude control system is built, and the improved PSO algorithm is applied in the optimized tuning of four controller parameters. Simulation results show that the improved PSO algorithm has stronger global searching ability than the common PSO algorithms, and obtains better UAV attitude control parameters.