基于可拓神经网络的火灾探测算法

为解决传统单一传感器式的火灾探测器容易造成火灾报警的漏报和误报的问题,采用多传感器信息融合技术,将温度、烟雾浓度和CO浓度等多个参数相结合,进行综合分析,对火灾进行早期预测。采用可拓神经网络作为数据融合算法,以温度、烟雾浓度、CO气体浓度三个物理参量作为输入,以三种火灾预警等级作为输出。通过仿真分析结果表明：火灾正确识别率很高,达到93.9%以上。同时通过与传统BP神经网络的对比,表明可拓神经网络在数据融合的速度和可靠性上有突出的优势,从而使可拓神经网络实际应用于火灾早期预测成为可能。     

Straight ladder inclined angle in a field environment: the relationship among actual angle,method of set-up and knowledge

Ladder inclined angle is a critical factor that could lead to a slip at the base of portable straight ladders, a major cause of falls from heights. Despite several methods established to help workers achieve the recommended 75.5° angle for ladder set-up, it remains unclear if these methods are used in practice. This study explored ladder set-up behaviours in a field environment. Professional installers of a company in the cable and other pay TV industry were observed for ladder set-up at their worksites. The results showed that the actual angles of 265 ladder set-ups by 67 participants averaged 67.3° with a standard deviation of 3.22°. Although all the participants had training on recommended ladder set-up methods, only 3 out of 67 participants applied these methods in their daily work and even they failed to achieve the desired 75.5° angle. Therefore, ladder set-up remains problematic in real-world situations.

Practitioner Summary: Professional installers of a cable company were observed for portable straight ladder set-up at their worksites. The ladder inclined angle averaged 67.3° with a standard deviation of 3.22°, while the recommended angle is 75.5°. Only a few participants used the methods that they learned during training in their daily work.     

基于自然语言检索的综合相似度计算算法

随着人工智能的发展,自然语言检索已逐步成为信息检索领域中的研究热点,而文本相似度计算算法直接决定着检索效果.在分析已有研究工作的基础之上,分别在统计信息层面和语义信息层面上提出了对应的相似度计算算法,最后提出了一种综合相似度计算算法,并通过相关实验验证了综合相似度计算算法对自然语言智能检索的有效性.     

基于改进光线投射算法的体数据显示

光线投射算法属于直接体绘制（DVR）中应用比较广泛的算法,其优点是绘制质量高,但是存在采样点计算量大,绘制速度慢的问题.针对这一问题,本文利用投射光线在物空间的传递性质,提出了一种改进的计算采样点位置的算法,加快采样点的获取速度,提高图像三维重建的效率.该算法在PC机平台上得到了实现,不仅在图像质量上得到保证而且绘制速度又有很大提高,为图像的三维重建提供了有效的手段.     

MDA建模的AOP扩展策略及其比较

对于MDA横切于核心业务逻辑的关注点对封装的破坏的问题,本文给出把AOP引入到MDA的扩展策略和主要方法,并对不同的扩展策略进行了比较。     

基于神经网络非线性模型的多级工作点阶跃响应扩展DMC预测控制

利用前馈神经网络建立对象的非线性预测模型,在不同工作点做阶跃响应,建立其局部线性模型,用隶属函数将局部线性模型加权得到全局线性模型,全局线性模型用于滚动优化,非线性模型用于预测系统输出和校正线性模型,实现非线性预测控制,仿真结果表明该方法控制效果良好,可满足实时要求。     

动态可扩展操作系统的研究

动态扩展技术的研究能够解决操作系统设计中许多长期存在并难以解决的问题。这些问题包括：如何在提高应用程序的性能同时保证系统的安全性;如何提高操作系统使用的灵活性;如何包容和管理随着操作系统发展而不断增加的复杂性。文章对当前各种操作系统扩展类型进行详细分类,具体介绍了各种动态扩展技术,评估了不同的扩展技术对操作系统性能和安全性的影响,最后提出了需要研究的内容以及其中的关键问题。     

Isokinetic strength characteristics in wrist flexion and extension

A laboratory experiment was conducted to measure strength characteristics in dynamic (isokinetic) wrist flexion and extension. Twenty four college-age males exerted their maximum torque in both concentric flexion and extension at 60, 120, and 180°/s of angular velocity through a ±60° range of deviation from wrist neutral. Results show that velocity and motion direction significantly effected both peak torque as well as the postural displacement of peak torque. The value of peak torque decreased with an increase in velocity and the wrist angle at peak torque generally moved to a more deviated, flexed posture (from neutral) with increasing velocity as well. Peak torque for all velocity and motion-type conditions tested occurred in a flexed posture relative to neutral. It is anticipated that these results may be of use as biomechanically based considerations in the evaluation and design of upper extremity tasks involving wrist flexion/extension as well as to perhaps give insight into functional characteristics of the wrist. Finally, regression equations were developed to aid in the prediction of peak torque based upon task, individual and/or population parameters.

Relevance to industry

Results from this study should enhance the overall understanding of wrist functioning. Specifically, motion type, velocity of movement and wrist posture are important ergonomic design considerations. These results can also be used to modify existing biomechanical models that do not consider wrist variables.     

子句型缺省逻辑中的分情形推理

引进一种树型方法以研究缺省逻辑中分情形推理下的Roos扩张,深入讨论了Roos扩张的计算,并分析了Roos扩张与Reiter扩张的关系.为计算Roos扩张,引入了从子句集分解最小文字集的算法.方法对于在缺省逻辑中计算Roos扩张以及分析分情形推理的计算复杂性是有用的.