基于卷积神经网络的航天复合材料缺陷智能检测

针对传统的缺陷图像识别处理方式存在着准确度与辨识度不足,且处理缺陷种类单一的问题,提出了一种基于Cascade R-CNN和Mask R-CNN的神经网络模型。首先,为了提高缺陷检测的可视化效果和检测准确度,在实例分割卷积网络Mask R-CNN的基础上,结合级联神经网络Cascade R-CNN结构,组合成了新的级联实例分割Cascade Mask R-CNN网络;其次,对组合而成的级联卷积神经网络进行了训练,将训练好的模型对复合材料缺陷图像进行了检测。实验结果表明:检测的平均准确度达到了91.5%,平均置信度达到了97.3%,达到了检测精度的要求。该研究成果可运用于航天复合材料缺陷识别。     

标准化学习管理系统的研究

针对现阶段低水平的学习管理系统重复开发、不同系统之间不能互操作,提出了可互操作的标准化学习管理系统的模型,并对各种不同类型数据信息进行XML绑定,实现不同系统之间的互操作。     

A method of predicting visual detectability of low-velocity impact damage in composite structures based on logistic regression model

This paper proposed a new method for quantitative assessment of visual detectability of damage based on logistic regression, using the Probability of Detection (POD) as a criterion. Experiments were performed to establish the massive hit/miss data of visual inspection. Authoritative investigations verified the reliability of the data. The prediction function concluded comprises more than one flaw size parameters, including the depth and diameter of the dents. The results show that the depth and diameter of the dents are pivotal for the evaluation of detectability; the type of detection, the detection distance, and the qualifications of personnel are critical external factors to be considered. This function, with an accuracy rate of nearly 85%, is capable of predicting the visual detection probability of impact damage under various detection environments, which will provide a reference for the damage tolerance design of composite materials and field maintenance in the Non-Destructive Testing (NDT) field.     

人工智能在气动设计中的应用与展望

在人工智能技术高速发展的浪潮下,智能化技术为空气动力学的研究提供了新的思路和手段。各国学者在人工智能与空气动力学设计的综合应用方面开展了诸多有益的探索与尝试。目前人工智能方法已被用于设计对象描述、数值求解、非线性映射等气动设计的关键环节中。实现了自适应设计参数探索、高效气动特征求解、快速数据降维与映射、智能优化等,提高了气动设计的速度、准确性、鲁棒性与全局性。概述了气动设计的发展现状、人工智能技术的研究现状以及机器学习在气动设计中的应用现状。展望了深度学习在气动设计上的应用前景。提出了以机器为核心根据优化阶段实时调整优化方案及走向的高度智能化气动设计概念——"机器设计"。强调了开展智能可诠释设计研究的重要性。     

Regression model for civil aero-engine gas path parameter deviation based on deep domain-adaptation with Res-BP neural network

The variations in gas path parameter deviations can fully reflect the healthy state of aero-engine gas path components and units; therefore, airlines usually take them as key parameters for monitoring the aero-engine gas path performance state and conducting fault diagnosis. In the past, the airlines could not obtain deviations autonomously. At present, a data-driven method based on an aero-engine dataset with a large sample size can be utilized to obtain the deviations. However, it is still difficult to utilize aero-engine datasets with small sample sizes to establish regression models for deviations based on deep neural networks. To obtain monitoring autonomy of each aero-engine model, it is crucial to transfer and reuse the relevant knowledge of deviation modelling learned from different aero-engine models. This paper adopts the Residual-Back Propagation Neural Network (Res-BPNN) to deeply extract high-level features and stacks multi-layer Multi-Kernel Maximum Mean Discrepancy (MK-MMD) adaptation layers to map the extracted high-level features to the Reproduce Kernel Hilbert Space (RKHS) for discrepancy measurement. To further reduce the distribution discrepancy of each aero-engine model, the method of maximizing domain-confusion loss based on an adversarial mechanism is introduced to make the features learned from different domains as close as possible, and then the learned features can be confused. Through the above methods, domain-invariant features can be extracted, and the optimal adaptation effect can be achieved. Finally, the effectiveness of the proposed method is verified by using cruise data from different civil aero-engine models and compared with other transfer learning algorithms.     

专科学生英语学习动机研究

学习动机与学生的学习态度、学习策略及学习成绩是紧密相关的.它直接制约着学习的积极性,极大地影响着学生的外语学习效果.本文介绍了对大学生英语学习动机研究的过程及其教学启示.     

人工智能气动特性预测技术在火箭子级落区控制项目的应用

发展了一种基于人工智能算法的气动特性预测技术,在开展部分工况风洞试验基础上,结合少量数值仿真结果,通过机器学习模型预测全部工况气动特性.该方法能够降低研制成本,缩短周期.先后解决了相关函数选择、模型超参数训练、数据检验和"人在回路"应用等关键算法与技术问题,应用于运载火箭子级栅格舵落区控制项目气动研制,获得了设计所需完...     

基于深度学习算法的极化合成孔径雷达通用分类器设计

地物分类是PolSAR(极化合成孔径雷达)的重要应用方向。传统算法需要基于特定数据人工选取特征和设计分类器,而深度学习算法能够自行从海量数据中提取层次化特征。在深度学习算法总结的基础上,结合深度学习和PolSAR大数据,提出了一种高效率、高精度的通用分类器设计方法。使用人工标记的数据训练CNN(深度卷积网络),自动化地进行特征学习和提取,并实现高精度的地物自动分类。在具有不同分辨率的机载和星载PolSAR数据上对通用分类器进行测试,都能快速、准确地分类。研究成果可快速将PolSAR数据转译为更直观的地物分类结果,对海量数据,特别是GF-3卫星PolSAR图像的利用有一定的辅助价值。     

基于深度卷积神经网络的遥感影像目标检测

随着遥感影像数据规模的快速扩张,如何高效准确地识别遥感影像中的典型目标成为当前的研究热点。为解决传统遥感影像目标检测方法准确率低的问题,用基于深度卷积神经网络进行遥感影像目标检测,在遥感影像数据集上用基于Faster-RCNN的神经网络模型对VGG16卷积网络进行训练,对输入的遥感影像通过区域推荐网络标注出待检目标的包围框和置信度,实现对遥感影像的目标检测。以飞机和油罐为例,在TensorFlow深度学习框架下实现了数据预处理、网络训练、目标检测等功能,并在当前测试数据集上取得了较高的检测准确率和置信度。该研究成果可应用于遥感影像解译和处理等相关领域。     

基于Sarsa(λ)强化学习的空间机械臂路径规划研究

针对目标特性未知的在轨操作环境,研究了典型空间操作机械臂的路径规划策略。采用Sarsa(λ)强化学习方法实现目标跟踪及避障的自主路径规划与智能决策,该方法将机械臂系统的每节臂视为一个决策智能体,通过感知由目标偏差和障碍距离程度组成的二维状态,设计符合人工经验的拟合奖赏函数,进行各臂转动动作的强化训练,最终形成各智能体的状态-动作值函数表,即可作为机械臂在线路径规划的决策依据。将本方法应用于多自由度空间机械臂路径规划任务,仿真结果表明新算法能在有限训练次数内实现对移动目标的稳定跟踪与避障,同时各智能体通过学习所得的状态-动作值函数表,具备较强的后期在线自主调整能力,从而验证了算法较强的鲁棒性和智能性。