ID 原文 译文
22125 其次，为提高稀疏阵列下矩阵填充方法重构接收信号矩阵性能和以此为基础的 2D-DOA 估计精度，提出基于遗传算法(GA)的稀疏采样阵列优化方法。 Then, a sparse sampling array optimization method based on Genetic Algorithm (GA) is studied to enhance the performance of Matrix Completion (MC) and DOA.
22126 最后，将 APG 和 MUSIC算法相结合，在重构完整平面阵列接收信号矩阵的基础上完成 2 维波达方向估计。 Finally, APG and MUSIC are employed to reconstruct the received signal matrix and estimate the direction of wave arrived, respectively.
22127 计算机仿真结果表明，该方法在保证 2 维波达方向估计精度前提下，大幅提高阵元利用率，有效降低空间谱平均旁瓣，与常规 2D-DOA 估计方法相比具有优势。 Computer simulation results show that the proposed method improves the utilization rate of array and reduces the average side lobe of spatial spectrum effectively, compared with the conventional 2D-DOA methods.
22128 针对目前人员定位方法普遍存在易受环境影响，累计误差大等问题，该文提出一种利用地图先验知识与井下人员行进方向识别相结合的位置校正方法。 Focusing on the problem that the personnel positioning methods are seriously influenced by the indoor environment, big cumulative error and other issues, a method is proposed to correct the position, which combines the prior knowledge of the map and the heading recognition.
22129 该方法首先通过线性判别分析(LDA)降低传感器特征集的维度，之后利用随机森林(RF)与设置阈值的方法对井下人员的行进方向分类并标记特殊点，将特殊点与巷道结构的先验知识进行匹配，修正并更新通过步行者航位推算算法(PDR)得到的井下人员的初步运动轨迹。 Firstly, the dimension of the feature set is reduced by Linear Discriminant Analysis (LDA). Then, the heading of the underground personnel is classified and the special points are marked through combining Random Forest (RF) and the method of setting a threshold value. Finally, the movement trajectory of the underground personnel, which is obtained by the Pedestrian Dead Reckoning (PDR) algorithm, is corrected and updated by matching the special point with the prior knowledge of the roadway structure.
22130 实验结果表明：LDA 的预处理方法能够有效提高后续分类器的精度高达 6%以上。 The experimental results show that the pre-processing method of LDA can effectively improve the precision of the classifier by more than 6%.
22131 该文提出的位置估算方法能够有效减小累积误差，具有较高的准确性和鲁棒性，活动识别精度能够达到 98%，可以实现可靠的实时定位。 The proposed method can effectively reduce the cumulative error, with high accuracy and robustness. The activity recognition accuracy can reach 98%, which can achieve reliable real-time location.
22132 顺序回归是机器学习领域中介于分类和回归之间的有监督问题。 Ordinal regression is one of the supervised learning issues, which resides between classification and regression in machine learning fields.
22133 在实际中，许多带有序关系标签的问题都可以被建模成顺序回归问题，因此顺序回归受到众多学者的关注。 There exist many real problems in practice, which can be modeled as ordinal regression problems due to the ordering information between labels. Therefore ordinal regression has received increasing interest by many researchers recently.
22134 基于极限学习机(ELM)的算法能有效避免因迭代过程陷入的局部最优解，减少训练时间，但基于极限学习机的算法在顺序回归问题上的研究较少。 The Extreme Learning Machine (ELM)-based algorithms are easy to train without iterative algorithm and they can avoid the local optimal solution; meanwhile they reduce the training time compared with other learning algorithms. However, the ELM-based algorithms which are applied to ordinal regression have not been exploited much.