Asynchronous n-step Q-learning adaptive traffic signal control

Ensuring transportation systems are efficient is a priority for modern society. Intersection traffic signal control can be modeled as a sequential decision-making problem. To learn how to make the best decisions, we apply reinforcement learning techniques with function approximation to train an adaptive traffic signal controller. We use the asynchronous n-step Q-learning algorithm with a two hidden layer artificial neural network as our reinforcement learning agent. A dynamic, stochastic rush hour simulation is developed to test the agent’s performance. Compared against traditional loop detector actuated and linear Q-learning traffic signal control methods, our reinforcement learning model develops a superior control policy, reducing mean total delay by up 40% without compromising throughput. However, we find our proposed model slightly increases delay for left turning vehicles compared to the actuated controller, as a consequence of the reward function, highlighting the need for an appropriate reward function which truly develops the desired policy.     

Social networks and joint/solo activity–travel behavior

The social dimension of activity–travel behavior has recently received much research attention. This paper aims to make a contribution to this growing literature by investigating individuals’ engagements in joint activities and activity companion choices. Using activity–travel diary data collected in Hong Kong in 2010, this study examines the impact of social network attributes on the decisions between solo and joint activities, and for joint activities, the choices of companions. Chi-square difference tests are used to assess the importance of social network variables in explaining joint activity behavior. We find that the inclusion of social network attributes significantly improves the goodness-of-fit of the model with only socioeconomic variables. Specifically, individuals receiving emotional support and social companionship from family members/relatives are found to more likely undertake joint activities with their family members/relatives; the size of personal social networks is found to be a significant determinant of companion choices for joint activities; and activity companions are found to be significant determinants of travel companions. The findings of this study improve the understanding about activity–travel, especially joint activity–travel decisions.     

Short-term speed predictions exploiting big data on large urban road networks

Big data from floating cars supply a frequent, ubiquitous sampling of traffic conditions on the road network and provide great opportunities for enhanced short-term traffic predictions based on real-time information on the whole network. Two network-based machine learning models, a Bayesian network and a neural network, are formulated with a double star framework that reflects time and space correlation among traffic variables and because of its modular structure is suitable for an automatic implementation on large road networks. Among different mono-dimensional time-series models, a seasonal autoregressive moving average model (SARMA) is selected for comparison. The time-series model is also used in a hybrid modeling framework to provide the Bayesian network with an a priori estimation of the predicted speed, which is then corrected exploiting the information collected on other links. A large floating car data set on a sub-area of the road network of Rome is used for validation. To account for the variable accuracy of the speed estimated from floating car data, a new error indicator is introduced that relates accuracy of prediction to accuracy of measure. Validation results highlighted that the spatial architecture of the Bayesian network is advantageous in standard conditions, where a priori knowledge is more significant, while mono-dimensional time series revealed to be more valuable in the few cases of non-recurrent congestion conditions observed in the data set. The results obtained suggested introducing a supervisor framework that selects the most suitable prediction depending on the detected traffic regimes.     

基于加权Kendall相关系数和序贯代理模型的有限元模型修正

为了提高桥梁结构有限元模型修正的效率和效果，提出基于加权Kendall相关系数和序贯代理模型的有限元模型修正方法。首先，建立基于目标响应误差和待修正设计参数灵敏度的加权Kendall相关系数指标，并采用凝聚层次聚类算法，合理确定待修正设计参数的数量和位置，保证待修正设计参数对目标响应具有合适的解耦能力；其次，为解决传统一次性代理模型法在构造代理模型过程中产生的欠采样或者过采样的问题，采用基于FLOLA-Voronoi通用序贯设计策略的序贯代理模型法，提高构造代理模型过程中试验设计样本的利用率和模型修正效率；最后，根据一座斜拉桥的静动力荷载试验数据进行有限元模型修正。结果表明：基于FLOLA-Voronoi算法的序贯代理模型能够合理地确定构造代理模型所需的样本数量和位置，使得在相同样本数量的情况下，序贯代理模型比一次性代理模型具有稍好的精度指标；采用加权Kendall相关系数指标的聚类方法可以得到具有合适解耦能力的待修正设计参数，使得修正前大误差目标响应的误差显著下降，小误差目标响应的误差基本都处于合理范围内，有效地降低了目标响应的整体误差；同时，有限元理论静动力响应与实测静动力响应的平均误差分别由5.56%、8.43%降低至1.87%、3.41%，并且相较于常规方法，修正精度更好，所提方法可以用于桥梁结构有限元模型修正，可得到准确的桥梁结构数值模型。     

融合D-S理论的锅炉人因贝叶斯网络安全性定量分析

针对船用锅炉人因安全性分析中存在的知识不确定性,采用D-S证据理论对多专家信息进行融合并建立考虑人因的贝叶斯网络,得到节点条件概率的区间表示形式.经加权平均后代入贝叶斯网络计算,与面向对象贝叶斯网络和FTA等方法的对比显示,该方法能够更加有效地融合不同专家信息,也更为符合工程实际.     

Real time detection of driver attention: Emerging solutions based on robust iconic classifiers and dictionary of poses

Real time monitoring of driver attention by computer vision techniques is a key issue in the development of advanced driver assistance systems. While past work mostly focused on structured feature-based approaches, characterized by high computational requirements, emerging technologies based on iconic classifiers recently proved to be good candidates for the implementation of accurate and real-time solutions, characterized by simplicity and automatic fast training stages.In this work the combined use of binary classifiers and iconic data reduction, based on Sanger neural networks, is proposed, detailing critical aspects related to the application of this approach to the specific problem of driving assistance. In particular it is investigated the possibility of a simplified learning stage, based on a small dictionary of poses, that makes the system almost independent from the actual user.On-board experiments demonstrate the effectiveness of the approach, even in case of noise and adverse light conditions. Moreover the system proved unexpected robustness to various categories of users, including people with beard and eyeglasses. Temporal integration of classification results, together with a partial distinction among visual distraction and fatigue effects, make the proposed technology an excellent candidate for the exploration of adaptive and user-centered applications in the automotive field.     

轮机管系单元设计和制造安装要点

基于单元组装的概念及基本形式,分析机舱中单元组装的方案及优点,讨论单元设计要点及注意事项,为轮机管系单元设计和制造安装提供借鉴.     

Elman递归神经网络在结构分析中的应用

给出了Elman动态递归神经网络的网络结构和基本原理。基于Elman递归神经网络能够逼近任意非线性函数的特点,提出了一种基于Elman递归神经网络建立结构分析模型的方法。利用Elman递归神经网络对桁架进行建模,真实地反映了桁架结构的动态特性。     

Energy-aware load-balaning fault-tolerant routing scheme for ad hoc sensor networks

Most routing protocols for sensor networks try to extend network lifetime by minimizing the energy consumption, but have not taken the network reliability into account. An energy-aware, load-balancing and fault-tolerant routing scheme, termed as ELFR was propsed to adapt to the harsh environment. First a network robustness model was presented. Based on this model, the route discovery phase was designed to make the sensors to construct into a hop-leveled network which is mesh structure. A cross-layer design was adopted to measure the transmission delay so as to detect the failed nodes. The routing scheme works with acknowledge （ACK） feedback mechanism to transfer control messages to avoid producing extra control overhead messages. When nodes fail, the new healthy paths will be selected locally without rerouting. Simulation results show that our scheme is much robust, and it achieves better energy efficiency, load balancing and maintains good end-to-end delay.     

����GA¬¬-BP�ĸ��ٹ�·����ѵ�ģ������������о�

鉴于模糊神经网络具有良好的非线性特性、学习能力、自适应能力和抗干扰能力,本文将模糊神经网络技术引入到高速公路入口匝道控制中。提出一种基于GA和BP算法的模糊神经网络控制器,并对控制器进行了详细设计。设计过程主要分为三部分：输入输出参数的选择、模糊神经网络的结构设计以及基于GA-BP的学习算法设计。最后,使用MATLAB软件对其进行了仿真。仿真结果表明,本文提出的方法是有效的,较之基于BP的模糊神经网络控制和ALINEA控制,能更好地稳定主线交通流密度。