由于水质数据特征复杂、关联度参差不齐而导致溶解氧浓度预测难度较大,为提高水质溶解氧浓度预测的准确性,提出了一种基于特征工程和北方苍鹰优化算法的长短期记忆网络(Feature Engineering-Northern Goshawk Optimization-Long Short T...由于水质数据特征复杂、关联度参差不齐而导致溶解氧浓度预测难度较大,为提高水质溶解氧浓度预测的准确性,提出了一种基于特征工程和北方苍鹰优化算法的长短期记忆网络(Feature Engineering-Northern Goshawk Optimization-Long Short Term Memory,FE-NGO-LSTM)混合模型。首先对水质数据集进行缺失值补齐、特征筛选与特征多项式构造,然后基于NGO-LSTM模型优化模型参数,提升预测性能;对不同多项式阶数下的特征预测效果进行分析之后,将该模型与基于灰狼优化算法、鲸鱼优化算法及粒子群优化算法的LSTM模型进行对比;最后,在太湖流域东苕溪城南监测断面对该模型进行了验证,计算FE-NGO-LSTM模型预见期为4,8,12,16,20,24 h的预测结果。试验结果显示:当多项式阶数为2阶时,模型预测效果最好,FE-NGO-LSTM模型相比基于其他优化算法的LSTM模型,平均绝对误差、均方误差、均方根误差分别至少降低9.0%,12.9%及6.3%,且随着预见期的增加,预测误差仍在可接受范围内,说明FE-NGO-LSTM模型在预测溶解氧浓度时具有一定优势与泛化性。展开更多
We developed a habitat suitability model for predicting nest locations of breeding Northern Goshawks (Accipiter gentilis) in the high-elevation mixed forest and shrub-steppe habitat of south-central Idaho, USA. We use...We developed a habitat suitability model for predicting nest locations of breeding Northern Goshawks (Accipiter gentilis) in the high-elevation mixed forest and shrub-steppe habitat of south-central Idaho, USA. We used elevation, slope, aspect, ruggedness, distance-to-water, canopy cover, and individual bands of Landsat imagery as predictors for known nest locations with logistic regression. We found goshawks prefer to nest in gently-sloping, east-facing, non-rugged areas of dense aspen and lodgepole pine forests with low reflectance in green (0.53 - 0.61 μm) wavelengths during the breeding season. We used the model results to classify our 43,169 hectare study area into nesting suitability categories: well suited (8.8%), marginally suited (5.1%), and poorly suited (86.1%). We evaluated our model’s performance by comparing the modeled results to a set of GPS locations of known nests (n = 15) that were not used to develop the model. Observed nest locations matched model results 93.3% of the time for well suited habitat and fell within poorly suited areas only 6.7% of the time. Our method improves on goshawk nesting models developed previously by others and may be applicable for surveying goshawks in adjacent mountain ranges across the northern Great Basin.展开更多
文摘由于水质数据特征复杂、关联度参差不齐而导致溶解氧浓度预测难度较大,为提高水质溶解氧浓度预测的准确性,提出了一种基于特征工程和北方苍鹰优化算法的长短期记忆网络(Feature Engineering-Northern Goshawk Optimization-Long Short Term Memory,FE-NGO-LSTM)混合模型。首先对水质数据集进行缺失值补齐、特征筛选与特征多项式构造,然后基于NGO-LSTM模型优化模型参数,提升预测性能;对不同多项式阶数下的特征预测效果进行分析之后,将该模型与基于灰狼优化算法、鲸鱼优化算法及粒子群优化算法的LSTM模型进行对比;最后,在太湖流域东苕溪城南监测断面对该模型进行了验证,计算FE-NGO-LSTM模型预见期为4,8,12,16,20,24 h的预测结果。试验结果显示:当多项式阶数为2阶时,模型预测效果最好,FE-NGO-LSTM模型相比基于其他优化算法的LSTM模型,平均绝对误差、均方误差、均方根误差分别至少降低9.0%,12.9%及6.3%,且随着预见期的增加,预测误差仍在可接受范围内,说明FE-NGO-LSTM模型在预测溶解氧浓度时具有一定优势与泛化性。
文摘We developed a habitat suitability model for predicting nest locations of breeding Northern Goshawks (Accipiter gentilis) in the high-elevation mixed forest and shrub-steppe habitat of south-central Idaho, USA. We used elevation, slope, aspect, ruggedness, distance-to-water, canopy cover, and individual bands of Landsat imagery as predictors for known nest locations with logistic regression. We found goshawks prefer to nest in gently-sloping, east-facing, non-rugged areas of dense aspen and lodgepole pine forests with low reflectance in green (0.53 - 0.61 μm) wavelengths during the breeding season. We used the model results to classify our 43,169 hectare study area into nesting suitability categories: well suited (8.8%), marginally suited (5.1%), and poorly suited (86.1%). We evaluated our model’s performance by comparing the modeled results to a set of GPS locations of known nests (n = 15) that were not used to develop the model. Observed nest locations matched model results 93.3% of the time for well suited habitat and fell within poorly suited areas only 6.7% of the time. Our method improves on goshawk nesting models developed previously by others and may be applicable for surveying goshawks in adjacent mountain ranges across the northern Great Basin.
文摘为解决风电场并网时的功率波动影响电网稳定性的问题,提出一种基于北方苍鹰(northern goshawk optimization,NGO)算法优化变分模态分解(variational mode decomposition,VMD)参数的混合储能功率分配策略。首先,按照风电场并网技术规范,采用自适应平均滤波法对风力发电功率进行滤波,并由滤波后的并网功率计算出波动功率。然后,采用NGO优化VMD算法中分解模态数K值和二次惩罚因子α值的最优值组合,将波动功率信号经VMD分解后实现在锂电池和超级电容器的功率分配,最后,采用双重模糊控制对混合储能系统(hybrid energy storage system,HESS)的荷电状态(state of charge,SOC)进行优化,完成HESS功率的二次分配。仿真结果表明,该控制策略不仅能够满足风电并网最大功率波动要求,还可以保持SOC维持在合理范围,实现HESS长期安全运行。
