基于GCN-Transformer的车辆换道行为建模与轨迹预测方法

对车辆换道行为建模并准确预测未来行驶轨迹对交通流的稳定与安全至关重要,为了解决目前大多数轨迹预测模型在同时捕捉车辆之间的空间相关性和时间依赖性上能力不足的问题,结合车辆轨迹的时空特点,本研究提出了一种基于长短期记忆网络、图卷积网络和Transformer编码器的改进建模策略。首先利用长短期记忆网络,对目标车辆和周围车辆在换道临界点前三秒内的状态信息分别进行轨迹编码,接着通过图卷积网络提取空间交互特征,然后通过Transformer编码器提取时间交互特征,最后将前三个模块处理后的特征向量合并后,输入至长短期记忆网络进行解码,得到目标车辆未来五秒的行驶路径预测输出。在NGSIM数据集和HighD数据集上进行实验,并与多种基准模型对比,结果表明:在2秒内的预测时域下,本文模型与PiP模型和DLM模型不差上下,但优于其他LSTM改进模型;在3~5秒内的预测时域下,本文模型优于各基准模型。本文还通过消融实验,证明了设计的时空特征提取模型对模型准确预测的有效贡献。

An integrated flexible self-healing Zn-ion battery using dendrite-suppressible hydrogel electrolyte and free-standing electrodes for wearable electronics

Among many aqueous batteries,flexible zinc-ion(Zn-ion)battery becomes the focus owing to the merits of low cost,non-toxicity,and safety.Here,a Zn dendrite-suppressible hydrogel electrolyte with both flexible and self-healing properties is developed via photoinitiated polymerization.The cross-linked structure of the polyacrylamide-N,N'-methylenebisacrylamide(PAM-MBA)-Zn/Mn hydrogel endows an enlarged chemical stable window,high ionic conductivity,and low polarization potential.After cycling at the current density of 0.5 mA·cm^(−2)for 250 h,Zn‖Zn symmetrical cell based on PAM-MBA-Zn/Mn electrolyte delivers a low polarization of 40 mV.The suppressed dendrite growth is ascribed to the uniform Zn deposition/stripping on anode.The galvanostatic intermittent titration technique curves display that the Zn-ion battery constructed by the PAM-MBA-Zn/Mn hydrogel electrolyte,free-standing FeVO_(4)/carbon cloth cathode,and Zn nanosheets/carbon cloth anode presents low reaction resistance and fast diffusion coefficient,indicating good endurance of cycling at high current densities.The battery with PAM-MBA-Zn/Mn hydrogel electrolyte presents a good flexible and self-healing performance.After bending 0°,60°,90°,and 180°for 30 times,batteries deliver stable capacities.Even cutting into ten pieces,the battery could self-heal and display a potential retention of 93.7%compared to the fresh cell.A good rate-performance is also achieved.After cutting/healing three times during cycling,capacity recovers well compared to the first-time cutting/healing.Moreover,the battery exhibits good flexibility using in an electric watch,indicating a promising potential for wearable electronics.

Predicting adsorption ability of adsorbents at arbitrary sites for pollutants using deep transfer learning

Accurately evaluating the adsorption ability of adsorbents for heavy metal ions(HMIs)and organic pollutants in water is critical for the design and preparation of emerging highly efficient adsorbents.However,predicting adsorption capabilities of adsorbents at arbitrary sites is challenging,with currently unavailable measuring technology for active sites and the corresponding activities.Here,we present an efficient artificial intelligence(AI)approach to predict the adsorption ability of adsorbents at arbitrary sites,as a case study of three HMIs(Pb(Ⅱ),Hg(Ⅱ),and Cd(Ⅱ))adsorbed on the surface of a representative two-dimensional graphitic-C_(3)N_(4).We apply the deep neural network and transfer learning to predict the adsorption capabilities of three HMIs at arbitrary sites,with the predicted results of Cd(Ⅱ)>Hg(Ⅱ)>Pb(Ⅱ)and the root-mean-squared errors less than 0.1 eV.The proposed AI method has the same prediction accuracy as the ab initio DFT calculation,but is millions of times faster than the DFT to predict adsorption abilities at arbitrary sites and only requires one-tenth of datasets compared to training from scratch.We further verify the adsorption capacity of g-C_(3)N_(4) towards HMIs experimentally and obtain results consistent with the AI prediction.It indicates that the presented approach is capable of evaluating the adsorption ability of adsorbents efficiently,and can be further extended to other interdisciplines and industries for the adsorption of harmful elements in aqueous solution.