沙戈荒区域丰富的风光热资源有利于支撑高能耗数据中心集群快速发展,但会使其面临算力负载强时变性、风光出力间歇性及恶劣天气离网运行可靠性的多重挑战。为此,该文提出一种考虑任务负载需求响应及源荷不确定性的数据中心集群微网电-...沙戈荒区域丰富的风光热资源有利于支撑高能耗数据中心集群快速发展,但会使其面临算力负载强时变性、风光出力间歇性及恶劣天气离网运行可靠性的多重挑战。为此,该文提出一种考虑任务负载需求响应及源荷不确定性的数据中心集群微网电-热设备容量协同优化配置方法。首先,根据计算任务对时延的敏感性,精细化建模可推迟可中断、可推迟不可中断及不可推迟3类任务负载的时间约束,在此基础上综合源荷不确定性建立数据中心集群微网“并网-离网”2阶段分布鲁棒优化模型,采用列与约束生成(column and constraint generation,C&CG)算法求解。以青海某实际数据中心为案例的分析结果表明:所提出的方法可使微网容量配置成本下降约25.8%,弃风率下降约56%,并大幅提高数据中心集群微网离网运行可靠性。该文研究为沙戈荒区域绿色低碳数据中心建设提供了理论支撑。展开更多
If a task is to be included in a reliable test, its features and the level of difficulty should be specified so that we know it is appropriate to the examinees. Beginning from Skehan＇s （1998） cognitive perspective ...If a task is to be included in a reliable test, its features and the level of difficulty should be specified so that we know it is appropriate to the examinees. Beginning from Skehan＇s （1998） cognitive perspective of task features in L2 instruction, the study compares the performance of two kinds of writing tasks from CET （College English Test） in China in order to verify whether the hypothesized task difficulty set by Skehan＇s cognitive framework can predict the actual performance of the examinees in testing context. The study also investigates the examinee＇s own perception of test tasks and the relationship between such perception and their performance. The results of the study show that more cognitively demanding tasks tend to generate greater level of task difficulty and thus elicit poorer examinee performance. However, it does not find enough evidence that there is some systematic competition between accuracy, fluency and complexity in the performance. It also shows the examinees＇ own perception of difficulty is variable and cannot be relied on. There is some reasonable self-awareness of the quality of their performance in terms of accuracy and fluency, but not complexity. It concludes to suggest that the two types of tasks should not be treated as parallel in CET in view of the observed difference on the examinees＇ performance.展开更多
文摘沙戈荒区域丰富的风光热资源有利于支撑高能耗数据中心集群快速发展,但会使其面临算力负载强时变性、风光出力间歇性及恶劣天气离网运行可靠性的多重挑战。为此,该文提出一种考虑任务负载需求响应及源荷不确定性的数据中心集群微网电-热设备容量协同优化配置方法。首先,根据计算任务对时延的敏感性,精细化建模可推迟可中断、可推迟不可中断及不可推迟3类任务负载的时间约束,在此基础上综合源荷不确定性建立数据中心集群微网“并网-离网”2阶段分布鲁棒优化模型,采用列与约束生成(column and constraint generation,C&CG)算法求解。以青海某实际数据中心为案例的分析结果表明:所提出的方法可使微网容量配置成本下降约25.8%,弃风率下降约56%,并大幅提高数据中心集群微网离网运行可靠性。该文研究为沙戈荒区域绿色低碳数据中心建设提供了理论支撑。
文摘If a task is to be included in a reliable test, its features and the level of difficulty should be specified so that we know it is appropriate to the examinees. Beginning from Skehan＇s （1998） cognitive perspective of task features in L2 instruction, the study compares the performance of two kinds of writing tasks from CET （College English Test） in China in order to verify whether the hypothesized task difficulty set by Skehan＇s cognitive framework can predict the actual performance of the examinees in testing context. The study also investigates the examinee＇s own perception of test tasks and the relationship between such perception and their performance. The results of the study show that more cognitively demanding tasks tend to generate greater level of task difficulty and thus elicit poorer examinee performance. However, it does not find enough evidence that there is some systematic competition between accuracy, fluency and complexity in the performance. It also shows the examinees＇ own perception of difficulty is variable and cannot be relied on. There is some reasonable self-awareness of the quality of their performance in terms of accuracy and fluency, but not complexity. It concludes to suggest that the two types of tasks should not be treated as parallel in CET in view of the observed difference on the examinees＇ performance.
