Strength of Stacking Technique of Ensemble Learning in Rockburst Prediction with Imbalanced Data: Comparison of Eight Single and Ensemble Models

Rockburst is a common dynamic geological hazard, severely restricting the development and utilization of underground space and resources. As the depth of excavation and mining increases, rockburst tends to occur frequently. Hence, it is necessary to carry out a study on rockburst prediction. Due to the nonlinear relationship between rockburst and its influencing factors, artificial intelligence was introduced. However, the collected data were typically imbalanced. Single algorithms trained by such data have low recognition for minority classes. In order to handle the problem, this paper employed stacking technique of ensemble learning to establish rockburst prediction models. In total, 246 sets of data were collected. In the preprocessing stage, three data mining techniques including principal component analysis, local outlier factor and expectation maximization algorithm were used for dimension reduction, outlier detection and outlier substitution, respectively. Then, the pre-processed data were split into a training set (75%) and a test set (25%) with stratified sampling. Based on the four classical single intelligent algorithms, namely k-nearest neighbors (KNN), support vector machine (SVM), deep neural network (DNN) and recurrent neural network (RNN), four ensemble models (KNN–RNN, SVM–RNN, DNN–RNN and KNN–SVM–DNN–RNN) were built by stacking technique of ensemble learning. The prediction performance of eight models was evaluated, and the differences between single models and ensemble models were analyzed. Additionally, a sensitivity analysis was conducted, revealing the importance of input variables on the models. Finally, the impact of class imbalance on the prediction accuracy and fitting effect of models was quantitatively discussed. The results showed that stacking technique of ensemble learning provides a new and promising way for rockburst prediction, which exhibits unique advantages especially when using imbalanced data.

     

Geochemical characteristics of hadal sediment in the northern Yap Trench

Journal of Oceanology and Limnology - Two sediment cores were retrieved in the hadal zone of the Yap Trench, and their concentrations of six major elements Mg, Al, Ca, Ti, Mn and Fe and nine trace...     

The Characteristics of Storm Wave Behavior and Its Effect on Cage Culture Using the ADCIRC+SWAN Model in Houshui Bay,China

The current storm wave hazard assessment tends to rely on a statistical method using wave models and fewer historical data which do not consider the effects of tidal and storm surge.In this paper,the wave-current coupled model ADCIRC+SWAN was used to hindcast storm events in the last 30 years.We simulated storm wave on the basis of a large set of historical storms in the North-West Pacific Basin between 1985 and 2015 in Houshui Bay using the wave-current coupled model ADCIRC+SWAN to obtain the storm wave level maps.The results were used for the statistical analysis of the maximum significant wave heights in Houshui Bay and the behavior of wave associated with storm track.Comparisons made between observations and simulated results during typhoon Rammasun(2014)indicate agreement.In addition,results demonstrate that significant wave height in Houshui Bay is dominated by the storm wind velocity and the storm track.Two groups of synthetic storm tracks were designed to further investigate the worst case of typhoon scenarios.The storm wave analysis method developed for the Houshui Bay is significant in assisting government's decision-making in rational planning of deep sea net-cage culture.The method can be applied to other bays in the Hainan Island as well.     

Laboratory Investigation into the Evaporation of Natural-Gas Condensate Oils: Hints for the Sanchi Oil Spill

The Sanchi oil tanker collision in the East China Sea on January 6th, 2018 has caused worldwide attention due to its uniqueness. A considerable amount of h     

气候态调整对华北冬、夏季气候监测的影响研究

本研究对比分析了不同气候态下,华北冬、夏季降水及气温的差异,分析了气候平均值的改变对历史极端事件监测的可能影响。研究结果发现,1991～2020年（简称气候Ⅱ态）的冬季和夏季的平均降水量均略多于1981～2010年（简称气候Ⅰ态）,但接近或略少于1961～2020年的平均降水量,平均降水量逐年变化幅度冬季Ⅱ态小于Ⅰ态,夏季反之。气候Ⅱ态冬季降水空间分布不均,夏季较Ⅰ态呈“中部减少,东西增加”的分布型。冬季和夏季极端降水阈值Ⅱ态（0.86 mm和22.0 mm）较Ⅰ态（0.83 mm和21.6 mm）均略有提高,造成近60年华北大部基于Ⅱ态阈值的冬、夏季极端降水日数较Ⅰ态略减少。此外,气候Ⅱ态的华北冬、夏季平均气温均明显高于Ⅰ态,也高于1961～2020年平均值。Ⅱ态气温较Ⅰ态基本呈全区增加特征,但空间分布不均匀。冬季极端低温和夏季极端高温阈值Ⅱ态（-9.8°C和27.9°C）较Ⅰ态（-10.2°C和27.5°C）均有所有所提高,造成华北大部分地区基于Ⅱ态阈值的近60年冬季极端低温日数较Ⅰ态有所增加,夏季极端高温日数较Ⅰ态存在不同程度的减少。因此,新气候态下华北气温和降水均值,华北大部极...     

黑河流域降水统计——动力降尺度问题研究

依据区域气候模式RIEMS2.0输出的3 km高分辨率数据和站点降水记录分析了中国西北黑河流域降水的动力降尺度和统计—动力降尺度问题,检验了多种因子组合下多元线性回归（MLR）和贝叶斯模式平均（BMA）降尺度模型,评估了降尺度降水的均方根误差、相关系数、方差百分率及“负降水”偏差率等方面的统计特征。结果表明,动力降尺度降水相关系数最高,误差也最大,降水方差达到观测值的1.5～2倍;除相关系数外,统计—动力降尺度模型的几个统计特征均最优,纯统计模型次之。检验表明,仅用700 hPa位势高度场、经向风和比湿等构建的统计降尺度模型估计的站点降水相关系数较低,均方根误差也较大。当在统计降尺度模型中引入模式降水因子后站点降水的估计得到明显改善,其中MLR类模型的降水相关系数和方差百分率均明显高于BMA类模型,均方根误差二者相当,但前者“负降水”出现频次明显大于后者,“负降水”偏差主要出现在降水稀少的冬半年及黑河中、下游干旱或极端干旱区,上游出现频率较低,其中MLR类模型“负降水”出现频次明显高于BMA类模型,后者仅出现在黑河中、下游地区。包含模式降水因子的统计—动力降尺度模型能减少“负降水”出现...     

关于发展人工影响天气数值模式的一些问题

人工影响天气的学科基础是中小尺度天气动力学与云降水物理学,需要将天气-动力-云降水物理耦合为一体。考虑到目前将天气动力学性质的基础数值模式用于人工影响天气中的问题,从数值模式动力方程、模式分辨率、云物理过程、数值求解方案、初边值条件等方面系统地探索了发展人工影响天气数值模式中一些需要重点解决、且不可忽视的特色问题,并举例对相关问题提出了解决思路和方法。期望提出的问题有助于构思更适合于人工影响天气数值模式,使数值模式功能真正向满足人工影响天气的要求靠近一步。     

Optimizing GPS-guidance transit route for cable crane collision avoidance using artificial immune algorithm

GPS Solutions - A cable crane collision accident usually causes serious damage to workers, equipment and materials. Owing to the emergence of global position system (GPS), some new collision...     

地质环境信息系统上线安全检测工作实践

本文针对地质环境领域业务系统的特点,提出了一种信息系统上线安全检测的方法,深入探讨了上线安全检测实践中的基本流程、测评方法与测评成果,为提升地质环境信息系统安全运行和服务能力提供了技术支撑.