New approaches for determining a common set of weights for a voting system

Data Envelopment Analysis (DEA) is one of the methods that have been proposed to determine the weights in a ranked voting system. DEA solves one model corresponding to each candidate and evaluates candidates with their own weights. However, sometimes solving only one model, and evaluating the candidates based on a common set of weights, is preferred. In this paper, some drawbacks of existing models are explained and new approaches for determining a common set of weights are proposed. Numerical examples are utilized to illustrate the content of the paper.     

塞浦路斯的水资源及地表观光的鉴别和管理

随着塞浦路斯人口和观光事业的较大的增长，未来的几十年将面临土地和水资源的缺乏．本文在遥感和GIS的基础上，应用三个不同等级的方法确定土地的利用，并用两个决策支持工具确定土地未来的适宜性使用，以及一个水文学的模型用于水资源的分析．     

用遗传算法优化神经网络初始权重的方法

基于遗传算法的全局搜索和BP网络局部精确搜索的特性,通过采用遗传算法优化网络初始权重的方法,将遗传算法和BP算法有机结合,做到了优势互补,并在提高洪水智能预报精度,特别是提高洪峰预报精度的研究中得到了很好的应用。在珠江流域西江洪水预报系统中,以历史水文资料进行检验的结果表明,洪峰时段的预报精度明显高于平水期的预报精度,证明了这种方法的有效性和可靠性。     

一种新型双权值人工神经元网络的数据拟合研究

在本文中提出了一种针对新型双权值神经元网络的数据拟合算法.采用这种新型网络结构和算法,可以克服传统的通用前馈网络中BP算法易陷入局部极小的问题.通过实验比较证明在相同的网络规模下,采用这种新型网络结构和算法可以取得比径向基(RBF)网络更高的拟合精度和更少的迭代次数.     

层次分析法在评价某企业报酬要素权重中应用

利用层次分析(AHP)法，对该企业中层干部报酬要素按层次进行分解细化，得到4个基本要素和26个子要素，通过对各判断矩阵求权重向量和一致性检验，最终确定各报酬要素的权重，定性与定量相结合，较合理的解决了企业报酬要素权重如何确定的问题，实践证明，该方法所得的结果在该企业运行中是可行的，具有指导意义的。     

A new method to assess the power performance during a lifting task in young adults

This study designs and verifies a new test for evaluating performance during a lifting task. 32 young men performed deadlift to high pull with increasing weights, and deadlift and upright row with a previously established weight of 70% 6RM, either using a Smith machine or free weights. Peak and mean power were significantly higher during deadlift to high pull with free weights than on the Smith machine, however only at ⩾50 kg. Their maximal values were achieved at about 80% and 70% 1RM. The deadlift to high pull with free weights is a reliable test with ICC > 0.80 and SEM < 10% for mean power. It is also a sensitive test for distinguishing lifting performance among healthy young individuals. It may be applied in functional performance testing of college graduate students and office workers with a prevalently sedentary lifestyle as well as construction workers with job demands based on lifting tasks.     

模糊综合评判在西南矿区沿空留巷适应性评价中的应用

沿空留巷技术在实际应用中表现出很大的优越性,为了使该技术在中小型煤矿中得到推广,综合运用层次分析法、模糊综合评判法建立了沿空留巷适应性评价模型。根据专家现场经验和试验研究结果,确定了影响适应性等级的6个主要地质因素,分别为煤层倾角、煤层厚度、巷道埋深、直接顶厚度影响系数、直接顶岩性和顶板完整性。采用层次分析法,通过构造判断矩阵,确定各个因素的权重。以西南某煤矿回采巷道为例,确定其留巷适应性等级为Ⅴ级,与实际留巷条件基本一致,表明该模型是合理可行的,能够实现对沿空留巷适应性等级进行评价。     

Deriving the priority weights from hesitant multiplicative preference relations in group decision making

In this paper, we investigate the deviation of the priority weights from hesitant multiplicative preference relations (HMPRs) in group decision-making environments. As basic elements of HMPRs, hesitant multiplicative elements (HMEs) usually have different numbers of possible values. To correctly compute or compare HMEs, there are two principles to normalize them, i.e., the α-normalization and the β-normalization. Based on the α-normalization, we develop a new goal programming model to derive the priority weights from HMPRs in group decision-making environments. Based on the β-normalization, a consistent HMPR and an acceptably consistent HMPR are defined, and their desired properties are studied. A convex combination method is then developed to obtain interval weights from an acceptably consistent HMPR. This approach is further extended to group decision-making situations in which the experts evaluate their preferences as several HMPRs. Finally, some numerical examples are provided to illustrate the validity and applicability of the proposed models.     

融合注意力机制和多任务学习的机器人抓取检测算法

抓取主要分为抓取检测、轨迹规划和执行环节，准确的抓取检测是完成抓取任务的关键。为进行更准确的抓取检测，提高机器人抓取性能表现，本研究以关键点检测算法为基础，提出了一种融合注意力和多任务学习的抓取检测算法。首先，针对任务特点，在特征提取环节引入CA(coordinate attention)注意力模块，显式的学习通道和空间特征，充分利用特征信息。其次，在损失函数环节加入多任务权重学习算法，学习抓取中心坐标、抓手开合宽度及旋转角度信息的最优权重。最后，在Cornell数据集以及更大规模的Jacquard数据集上进行试验。研究结果表明，所提方法相比滑动窗口和锚框类型等经典方法在检测速率上有明显提升，且与单纯的关键点检测方法相比有更高的准确率，所提模型在两个数据集上分别取得98.8%和95.7%的准确率。检测示例体现出所提模型对于非常规物体也有良好的抓取结果，不同Jaccard系数条件下的抓取结果显示模型在精准抓取方面有优秀性能，而对于权重学习算法的不同初始值试验则表明所提模型具有良好的鲁棒性。此外，通过消融实验分析了不同模块对于模型性能表现的影响程度。     

基于模糊综合评判和综合赋权的抗灾型配电网骨干网架规划

根据抗灾型配电网的特点和差异化规划的理念,建立了计及电网安全性、灾况供电可靠性、特级负荷覆盖率、差异化成本和差异化效益的抗灾型中压配电网骨干网架评判模型,提出了一种基于模糊理论、综合赋权和成本效益分析法的抗灾型配电网骨干网架改造决策方法。该方法在选取影响电网抗灾性能的主要因素的基础上,通过高斯隶属函数建立了模糊评判矩阵。同时,利用改进层次分析法(IAHP)和熵权法确定了主客观统一的综合权重,并由模糊综合评判法进行了各配电线路的重要性排序,然后由成本效益分析法确定了最优骨干网架。最后,将该方法应用于南方某地区配电网抗灾骨干网架改造方案决策分析中,案例分析结果验证了所提方法的有效性和实用性。