一种深度回声状态网络的输入尺度自适应算法

深度回声状态网络是回声状态网络与深度学习思想的结合,合理选取不同谱半径的内部状态矩阵和弱积分参数能有效增强深度回声状态网络的多尺度时域特性。利用数据可视化分析输出矩阵在不同网络层中的分布关系,发现高层网络中部分神经元处于饱和工作状态且该状态抑制了网络动态预测能力。提出一种深度回声状态网络的输入矩阵自适应算法,在对网络内部状态的均值和方差进行递推估计的基础上判断神经元饱和状态,通过自适应调整各层输入权重的值来增强神经元动态性。数值计算结果表明,基于输入尺度自适应算法的深度回声状态网络相对同等规模的单层回声状态网络对于动态系统的预测精度有成倍提升。     

基于深度强化学习的深圳市急救车调度算法

在院前急救领域中,急救反应时间是指患者拨打急救电话后,急救车到达现场的时间。传统急救车调度算法未全面考虑急救环境的动态性和复杂性因素,导致模型优化的急救反应时间与实际情况存在偏差。将急救车调度问题建模成马尔科夫决策过程,构建基于深度强化学习的急救车调度算法。以多层感知机作为评分网络结构,通过将急救站的动态信息映射为各个急救站的得分,确定急救车被调往各急救站的概率。同时,结合急救车调度的动态决策特点,利用强化学习中演员-评论家框架下的近端策略优化算法改进评分网络参数。在深圳市急救中心真实急救数据集上的实验结果表明,相比Fixed、DSM、MEXCLP等算法,该算法在每个急救事件中的急救反应时间平均缩短约80 s,并且在10 min内急救车的平均到达比例为36.5%,能够实时地将急救车调度到合适的急救站。     

滚筒式采煤机煤岩截割力学特性及测试系统研究

为实现采煤机在煤岩截割过程中滚筒载荷受力的测试与分析,现场建立等比例大型采煤机力学特性分析研究实验平台,采用等效结构惰轮轴传感器测试方法,针对滚筒式采煤机在煤岩截割过程中采煤机滚筒载荷变化进行实时动态在线监测,采用无线信号发射装置实现测试数据的实时传输。现场截割实验结果表明,在采煤机截割煤岩过程中,滚筒x,y,z三向上的载荷峰值差分别为29.941,17.459和7.371kN,载荷变化显著,测试结果符合现场实际工况,为实现采煤机煤岩动态识别以及自动调高控制提供了重要的理论和数值依据。     

Reprogrammable Assembly of Molecular Motor on Solid Surfaces via Dynamic Bonds

Controllable assembly of molecular motors on solid surfaces is a fundamental issue for providing them to perform physical tasks. However, it can hardly be achieved by most previous methods due to their inherent limitations. Here, a general strategy is designed for the reprogrammable assembly of molecular motors on solid surfaces based on dynamic bonds. In this method, molecular motors with disulfide bonds can be remotely, reversibly, and precisely attached to solid surfaces with disulfide bonds, regardless of their chemical composition and microstructure. More importantly, it not only allows encoding geometric information referring to a pattern of molecular motors, but also enables erasing and re‐encoding of geometric information via hemolytic photocleavage and recombination of disulfide bonds. Thus, solid surfaces can be regarded as “computer hardware”, where molecular motors can be reformatted and reprogramed as geometric information.     

Optimal markdown policy of perishable food under the consumer price fairness perception

Markdown policies are widely used by retailers to sell perishable food. Consumers purchase food at different prices during different sales periods. Some consumers may compare their experience with others who purchase the same items. Price unfairness or inequity is perceived when different prices are quoted without reasonable explanations. This study develops an optimal markdown model for perishable food pricing to optimise the food retailer revenue and enable a maximum aggregated consumer utility considering individuals’ price fairness perception. The model serves as the first step in evaluating trade-offs between food retailer revenue and consumer utility. In addition, it enables consumer utility to be depicted through perceived price fairness by including the effects of food perishability and scarcity. Another innovative feature is the equalisation of the consumer average aggregated utility during different sales periods as a condition of intertemporal price fairness perception. The proposed model is compared with two benchmark models to justify the effectiveness and advantages in the example. Finally, a sensitive analysis based on the food deterioration rate, consumer food desire rate and consumer average reservation price is conducted to justify the manner in which these factors influence the optimal pricing policy, and managerial insights are suggested for food retailers.     

A mathematical model for vehicle routing problem under endogenous uncertainty

In this study, a multistage stochastic programming (SP) model is presented for a variant of single-vehicle routing problem with stochastic demands from a dynamic viewpoint. It is assumed that the actual demand of a customer becomes known only when the customer is visited. This problem falls into the category of SP with endogenous uncertainty and hence, the scenario tree is decision-dependent. Therefore, nonanticipativity of decisions is ensured by conditional constraints making up a large portion of total constraints. Thus, a novel approach is proposed that considerably reduces the problem size without any effect on the solution space. Computational results on some test problems are reported.     

Strain Rate Effects on Dynamic Fractures in Fine‐grained Granitic Rock

The effect of dynamic strain rates on failure responses of a fine‐grained granitic rock is studied experimentally and theoretically. Theoretical investigation employs a model incorporating dynamic fracture criterion with damage mechanics theory. Experimental investigation is conducted using split Hopkinson pressure bar device. In order to investigate the effects of microstructure on dynamic fracture failure under different loading rates, fragment debris of each tested specimen is collected and analyzed. It is found through the debris analysis that the granitic rock breaks down into the fragment debris in grain size scales and the effect of strain rates on the formation of fragment debris appears to be related to the microstructure of the rock. It is also found that dynamic inertia induced by the dynamic loading can reduce the effect of friction confinement generated by the contact between the cylindrical specimen and two split Hopkinson pressure bars on the dynamic responses of the specimen. Theoretical evaluations agree with the corresponding experimental observations.     

Dynamic flow‐based particle splitting in smoothed particle hydrodynamics

In this work, a dynamic procedure for local particle refinement to be used in smoothed particle hydrodynamics (SPH) is presented. The algorithm is able to consistently produce successive levels of particle splitting in accordance to a flow‐based criterion. It has been applied together with accurate and robust formulations for variable spatial resolution in the framework of a semi‐implicit, truly incompressible scheme for SPH. Different test cases have been considered to assess the capabilities and advantages of the proposed procedure, namely, the laminar flow around circular and square obstacles in a plane channel for various regimes. Such flow cases entail the simulation of attached and separated shear layers, recirculating flow, vortex shedding and surface discontinuities. The results obtained for two levels of particle splitting have demonstrated that significant improvements may be obtained with respect to uniform particle spacing solutions in a variety of situations, thus presenting an excellent trade‐off between accuracy and computational cost. Copyright © 2015 John Wiley & Sons, Ltd.     

Drug self-assembly: A phenomenon at the nanometer scale with major impact in the structure–biological properties relationship and the treatment of disease

Under water-rich conditions, small amphiphilic and hydrophobic drug molecules self-assemble into supramolecular nanostructures. Thus, substantial modifications in their interaction with cellular structures and the ability to reach intracellular targets could happen. Additionally, drug aggregates could be more toxic than the non-aggregated counterparts, or vice versa. Moreover, since self-aggregation reduces the number of effective “monomeric” molecules that interact with the target, the drug potency could be underestimated. In other cases, the activity could be ascribed to the non-aggregated molecule while it stems from its aggregates. Thus, drug self-assembly could mislead from drug throughput screening assays to advanced preclinical and clinical trials. Finally, aggregates could serve as crystallization nuclei. The impact that this phenomenon has on the biological performance of active compounds, the inconsistent and often controversial nature of the published data and the need for recommendations/guidelines as preamble of more harmonized research protocols to characterize drug self-aggregation were main motivations for this review. First, the key molecular and environmental parameters governing drug self-aggregation, the main drug families for which this phenomenon and the methods used for its characterization are described. Then, promising nanotechnology platforms investigated to prevent/control it towards a more efficient drug development process are briefly discussed.     

Identification of Pivotal Causes and Spreaders in the Time‐Varying Fault Propagation Model to Improve the Decision Making under Abnormal Situation

Under abnormal conditions, timely and effective decisions of system recovery and protective measures are of great significance for safety‐critical systems. The knowledge of the roles that network nodes play in the spreading process is crucial for developing efficient maintenance decisions; for singling out and preferential control, the ‘pivotal spreaders’ may be a way to maximize the chances to timely hinder the fault pervasion. Inspired by the inhomogeneous topological nature of a complex fault propagation network, this study is devoted to exploring the spreading capabilities of nodes regarding both structural connectivity and causal influence strength, so as to provide decisions of preferential recovery actions under specific fault scenarios. Specifically, the dynamic betweenness centrality and nonsymmetrical entropy are incorporated to adaptively measure the system‐wide fault diffusion risk of a set of controllable fault events. In order to model the dynamics and uncertainties involved in the complex fault spreading process, we introduce the model of a dynamic uncertain causality graph, based on which solutions of time‐varying structure decomposition and causality reduction are adopted to improve the reasoning efficiency. Verification experiments consisting of simulated calculation cases and generator faults of a nuclear power plant show empirically the effectiveness and applicability of this method in large‐scale engineering practice. Copyright © 2014 John Wiley & Sons, Ltd.