This article mainly discusses the application methods and allocation plans of two typesofdrones dealing with wildfires in specific situations. We established a model to coordinateforest fire supporting drones based on computer simulations. In State Victoria, we have collectedandprocessed the relevant data of forest fires with digital image processing technology. We selectedthe most representative area and used the Alpha Shapes algorithm to estimate the boundariesofthe state's forest fire-prone areas. The obtained area can be used to determine a unit range, thenwe used computers to make fire simulations within this range. The model uses securityasanindicator, and through multiple simulations, it provided an estimated value of the number ofdrones needed to ensure a high communication timeliness with the least quantity. Meanwhile, thedrone shifting model is introduced to ensure a smoother transmission: additional Drones will beused when the initial drone returns to the headquarters for charging. The influence of extremeterrain on the signal transmission between Drones is also considered, and a correspondingalgorithm is designed to optimize the Drones ’ positions. By analyzing the historical dataofwildfires in Australia, we divided forest fires into three different levels and calculatedthefrequency of each levels as their weights. Then our model can provide the appropriate quantityofthe drones we need to buy. We prepared a budget request for the CFA based on the model above. Simulation experiments have proved that this model has a strong rob

随着智能制造相关技术的高速发展，传统的制造模式已无法满足订单个性化、多样化、定制化的市场需求，柔性化、智能化和高度集成化的离散生产模式是未来制造业生产模式的发展趋向，多品种、小批量、混合生产是其生产特点。但这种生产模式带来了车间资源组织的复杂性、设备负载的不均性、制造过程的多扰动性、运行状态的不稳定性，并给车间的高效管控带来极大复杂度。 为此，DTWorks 区别于离线仿真系统，并将不局限在现场管理层。通过在线不断收集车间中的扰动事件以及设备的状态变化，测定产品单元级生产工时，并工时为基准进行现场作业改善。以车间设备模型和产品制造模型作为核心模型，以工艺流程为链条串连起制造工序工时。以数据驱动的数字孪生车间为核心载体，以排产计划为管理纲领，工时管理基准作为切入点，寻找引发计划变动的因素，并把变动控制在小范围波动上，从而指导车间作业生产。

问题主题识别是问答系统中的重要步骤，可以定位问题领域，缩小查询范围精化答案。本研究提出引入外部 word2vec 维基百科词嵌入的英文文本分类方法，并针对 MadSci 上三类主题的问题集进行了相关试验。该方法首先训练维基百科语料库获得 word2vec 词向量字典，其次，对问题集进行数据清洗，进行转小写，去标点，词形还原等操作。建立基于词向量的特征提取方法，例如，完全平均法，根据词性加权，根据词长加权等。最后通过 SVM 经典分类器和 KNN 分类器对其进行分类实验。试验结果表明，本研究可以有效的进行英文文本分类，且分类效果显著高于没有引入 word2vec 对照试验的分类结果。