基于PSO-BP神经网络的平流层风场短期快速预测

平流层风场环境对临近空间低速飞行器驻空飞行性能有重要影响。研究了基于PSO-BP神经网络的平流层区域风场建模与快速预测方法,根据历史风场数据,采用主成分分析法对数据进行降维处理,通过BP神经网络对风场进行预测建模,利用粒子群优化（PSO）算法对其进行优化,采用Biharmonic样条曲面插值方法构建区域预测风场。以南海地区5年历史风场为对象,对比分析了基于BP神经网络和基于PSO-BP神经网络的风场预测模型,结果表明:使用具有全局寻优特性的PSO算法改进BP神经网络,能够有效避免传统BP神经网络易陷入局部最优的缺点,提高预测精度;通过结合PSO-BP神经网络预测与Biharmonic样条曲面插值,可实现区域风场的预测。研究结果可为临近空间低速飞行器的轨迹规划与区域驻留等任务的高精度区域快速预报风场提供解决途径。      

多目标组合运输物流量预测建模算法

提出了一种新的多目标组合运输物流量预测建模算法.以时间、领域、影响以及组合运输为基准,运用系统工程理论思想设计出一种四维的物流量影响因素模型,并运用结构方程模型对所建模型做了优化,提取出组合运输物流量的核心影响因素.在改进的神经网络算法的基础上结合遗传算法,提出了一种结合遗传算法的改进的神经网络新算法,弥补了改进的神经网络算法上的缺陷,在多目标组合运输物流量预测的实例应用中,该算法不仅有很高的预测精度,而且具有收敛速度快、运行稳定的特点.      

一种非线性系统集员辨识算法

针对带有未知有界噪声的非线性动态系统的鲁棒辨识问题,提出了一种新的非线性动态系统的集员辨识算法.利用径向基函数神经网络的逼近能力,根据系统的输入输出数据,选用径向基函数神经网络对未知非线性系统建模.径向基函数神经网络的中心被确定之后,考虑到建模误差与系统噪声有界,利用径向基函数神经网络为参数线性模型的特点,使用参数线性集员辨识算法辨识径向基函数神经网络的输出权值.由于集员辨识算法所得到的是网络输出权值的集合估计,在系统运行过程中,可以很方便地利用所建模型预测实际系统的输出范围.仿真表明,集员辨识算法辨识网络的输出权值比最小二乘法较少的受未知动态系统噪声分布的影响.     

基于改进极限学习机的光谱定量建模方法

依据近红外光谱(NIR)产生原理,提出了粒子群优化极限学习机(PSO-ELM)算法,运用于小样本氨水浓度定量分析。通过优化极限学习机(ELM)隐藏节点参数,解决了极限学习机由于输入权值和隐含层偏差随机产生的建模结果具有随机性的问题,提高了预测模型的稳定性、精确度和泛化性能。经实验验证,优化后的PSO-ELM相比ELM,模型预测集均方根误差由0.01166减小至0.00322,预测集相关系数由0.9951提高至0.9979。将优化后的模型预测结果与支持向量机(SVM)、BP神经网络算法等传统方法的建模结果进行对比,优化后的PSO-ELM算法具有较高的精确度和良好的泛化性能,模型预测效果优于传统的定量回归分析算法。     

基于时序建模的卫星故障检测方法

为解决当前卫星故障检测面临的依赖规则库、多元特征融合不足以及数据正负样本分布不均衡等问题,从卫星数据的时序特性出发,提出基于时序建模的卫星故障检测方法与半监督模型,实现卫星数据规律的有效挖掘与数据驱动的故障检测。考虑卫星数据间的时序关联,提出基于长短期记忆神经网络的卫星故障检测方法,并引入滑动窗口机制实现卫星数据的有效预测与故障检测。考虑卫星数据多元特征参数间的关联关系,引入时间卷积和自编码器神经网络,同时建模不同时刻、多元特征参数间的依赖关系,实现融合多元特征参数进行卫星故障的有效检测。以某型号卫星电源分系统为实验对象,仿真结果表明,所提算法和模型在关键指标方面优于BP神经网络等传统故障检测方法和模型。     

基于神经网络的磁平静期赤道电集流预测

利用BP神经网络技术分别对2008年后磁平静期印度扇区、秘鲁扇区以及CHAMP卫星的赤道电集流（EEJ）变化进行预测，其中神经网络训练数据为对应的2000—2007年磁平静期EEJ观测数据，输入参量为天数、地方时、太阳天顶角、太阳活动指数（F_10.7）、太阴时以及卫星地理经度，输出参量为EEJ.对EEJ预测结果进行了统计学分析，并且与实际观测结果进行对比.结果表明:BP神经网络对事件中EEJ的变化具有很好的预测能力，预测结果能够反映EEJ的重要分布特征；EEJ预测值与观测值之间具有很好的相关性，其中地磁台站观测值与预测值相关性系数可达85%以上.此外，将BP神经网络模型的预测结果与Yamazaki提出的经验模型结果进行对比，结果显示BP神经网络与其经验模型性能相当.研究结果表明，BP神经网络技术在平静期EEJ变化预测方面性能优异，具有良好的应用前景.      

基于平流层风场预测的浮空器轨迹控制

平流层风场环境对浮空器设计和轨迹控制具有重要影响。针对平流层风场建模,以长沙地区2005—2010年的风场数据为例,首先采用本征正交分解（POD）方法对风场数据进行降阶处理;然后分别采用Fourier级数与BP神经网络算法对平流层风场进行预测,并对2种模型的预测精度进行比较分析;最后通过建立临近空间浮空器的动力学模型和高度调控模型,分析2种风场预测模型对浮空器轨迹控制的影响。研究结果表明,相对于Fourier预测模型,基于BP神经网络预测模型的预测精度更高,可信度更强,能够更好地为浮空器飞行轨迹控制提供参考价值。      

基于BP神经网络的纳卫星轨道温度预测

为了实现纳卫星在轨温度的预测,在对纳卫星热系统动态特性模型分析的基础上,建立BP神经网络预测模型实现纳卫星在轨温度的预测.通过分析纳卫星热系统动态特性模型,得到用于BP神经网络预测模型的输入、输出变量以及训练神经网络所需的数据样本.BP神经网络预测模型分别以纳卫星外壳、辐射器、舱内仪器的热流及温度值为神经网络输入、输出,预测纳卫星10s后的轨道温度.经验证,神经网络预测模型预测结果与纳卫星实际轨道温度吻合较好,表明神经网络预测模型是快捷有效的.      

利用人工神经网络快速计算木星系磁坐标

在木星辐射带研究中，从地理坐标向磁坐标的准确转换是建模基础.以往的建模中，磁壳参数L值的计算基于磁偶极场假设，该方法精确度较差.结合最新的高精度木星磁场模型JRM09，本文提出基于磁力线追踪法的木星磁坐标计算方法，并分析其合理性和必要性.要求精确度较高时，磁力线追踪法计算耗时很长.本文在磁力线追踪法的基础上进行改进，提出基于人工神经网络的磁坐标快速计算方法.该方法包括分类器和拟合器.分类器基于Adaboost算法的BP神经网络，用于预测某地理坐标是否在内磁层，如果在内磁层，则用拟合器计算L值.拟合器采用遗传算法优化BP神经网络.结果表明，分类器的分类错误率在3%以内，而拟合器的预测误差在7%以内.以Juno号一圈探测轨道为例，利用神经网络的磁坐标计算法比磁力线追踪计算法速度快3个数量级以上.基于人工神经网络的磁坐标快速计算方法可用于未来木星辐射带的研究.      

一种高精度GPS卫星钟差预报方法

为了获得实时高精度GPS钟差,提出了采用快速星历建模进行短期预报。文章先对钟差数据提取趋势项,再利用傅里叶分析研究其周期特征以确定建模与预报时间段长度,最后利用径向基函数(Radial Basis Function,RBF)神经网络建模实时预报钟差。由于RBF神经网络用于非线性数据建模效果良好,在提取线性趋势项并合理确定建模周期后,该方法能够得到较好的预报结果。实际预报结果表明,文中方法得到的预报钟差精度高于超快速星历,能够满足分米级实时精密定位的要求。     

Comparison of ANN and MLR models for estimating solar radiation in Turkey using NOAA/AVHRR data

In this paper, the estimation capacities of MLR and ANN are investigated to estimate monthly-average daily SR over Turkey. The satellite data are used for 73 different locations over Turkey. Land surface temperature, altitude, latitude, longitude and month are offered as the input variables for modeling ANN and MLR to get SR. Estimations of SR are evaluated with the meteorological values by using the statistical bases. The obtained results indicated that the ANN model could achieve a satisfactory performance when compared to the MLR model. Moreover, it is understood that more accurate results in estimation of SR are obtained in the use of satellite data, rather than the use of meteorological station data. Finally, the built ANN model is used to estimate the yearly average of daily SR over Turkey. As a result, satellite-based SR map for Turkey is generated.     

Regional modeling of the ionosphere using adaptive neuro-fuzzy inference system in Iran

In recent years, new techniques and algorithms such as Artificial Neural Networks (ANNs), Fuzzy Inference Systems (FIS) and Genetic Algorithm (GA) have been used as alternative statistical tools in modeling and forecasting issues. These methods have been extensively used in the field of geosciences and atmospheric physics. The main purpose of this paper is to combine FIS and ANNs for local modeling of the ionosphere Total Electron Content (TEC) in Iran. An Adaptive Neuro-Fuzzy Inference System (ANFIS) is developed for TEC modeling. Also, Multi-Layer Perceptron ANN (MLP-ANN) and ANN based on Radial Base Functions (RBF) have been designed for analyzing ANFIS results. Observations of 29 Global Positioning System (GPS) stations from the Iranian Permanent GPS Network (IPGN) have been used in 3 different seasons in 2015 and 2016. These stations are located at geomagnetic low latitudes region. Out of these 29 stations, 24 stations for training and 5 stations for testing and validating were selected. The relative and absolute errors have been used to evaluate the accuracy of the proposed model. Also, the results of this paper are compared with the International Reference Ionosphere model (IRI2016). The maximum values of the average relative error for RBF, MLP-ANN, ANFIS and IRI2016 methods are 13.88%, 11.79%, 10.06%, and 18.34%, respectively. Also, the maximum values of the average absolute error for these methods are 2.38, 2.21, 1.5 and 3.36 TECU, respectively. Comparison of diurnal predicted TEC from the ANFIS, RBF, MLP-ANN and IRI2016 models with GPS-TEC revealed that the ANFIS provides more accurate predictions than the other methods in the test area.     

Prediction of landslides using ASTER imagery and data mining models

The aim of this study was to identify landslide-related factors using only remotely sensed data and to present landslide susceptibility maps using a geographic information system, data-mining models, an artificial neural network (ANN), and an adaptive neuro-fuzzy interface system (ANFIS). Landslide-related factors were identified in Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) satellite imagery. The slope, aspect, and curvature of topographic features were calculated from a digital elevation model that was made using the ASTER imagery. Lineaments, land-cover, and normalized difference vegetative index layers were also extracted from the imagery. Landslide-susceptible areas were analyzed and mapped based on occurrence factors using the ANN and ANFIS. The generalized bell-shaped built-in membership function of the ANFIS was applied to landslide susceptibility mapping. Analytical results were validated using landslide test location data. In the validation results, the ANN model showed 80.42% prediction accuracy and the ANFIS model showed 86.55% prediction accuracy. These results suggest that the ANFIS model has a better performance than does the ANN in predicting landslide susceptibility.     

Selection of classification techniques for land use/land cover change investigation

The concerns over land use/land cover (LULC) change have emerged on the global stage due to the realisation that changes occurring on the land surface also influence climate, ecosystem and its services. As a result, the importance of accurate mapping of LULC and its changes over time is on the increase. Landsat satellite is a major data source for regional to global LULC analysis. The main objective of this study focuses on the comparison of three classification tools for Landsat images, which are maximum likelihood classification (MLC), support vector machine and artificial neural network (ANN), in order to select the best method among them. The classifiers algorithms are well optimized for the gamma, penalty, degree of polynomial in case of SVM, while for ANN minimum output activation threshold and RMSE are taken into account. The overall analysis shows that the ANN is superior to the kernel based SVM (linear, radial based, sigmoid and polynomial) and MLC. The best tool (ANN) is then applied on detecting the LULC change over part of Walnut Creek, Iowa. The change analysis of the multi temporal images indicates an increase in urban areas and a major shift in the agricultural practices.     

Thermal and TEC anomalies detection using an intelligent hybrid system around the time of the Saravan,Iran, (Mw = 7.7) earthquake of 16 April 2013

A powerful earthquake of M_w = 7.7 struck the Saravan region (28.107° N, 62.053° E) in Iran on 16 April 2013. Up to now nomination of an automated anomaly detection method in a non linear time series of earthquake precursor has been an attractive and challenging task. Artificial Neural Network (ANN) and Particle Swarm Optimization (PSO) have revealed strong potentials in accurate time series prediction. This paper presents the first study of an integration of ANN and PSO method in the research of earthquake precursors to detect the unusual variations of the thermal and total electron content (TEC) seismo-ionospheric anomalies induced by the strong earthquake of Saravan. In this study, to overcome the stagnation in local minimum during the ANN training, PSO as an optimization method is used instead of traditional algorithms for training the ANN method. The proposed hybrid method detected a considerable number of anomalies 4 and 8 days preceding the earthquake. Since, in this case study, ionospheric TEC anomalies induced by seismic activity is confused with background fluctuations due to solar activity, a multi-resolution time series processing technique based on wavelet transform has been applied on TEC signal variations. In view of the fact that the accordance in the final results deduced from some robust methods is a convincing indication for the efficiency of the method, therefore the detected thermal and TEC anomalies using the ANN + PSO method were compared to the results with regard to the observed anomalies by implementing the mean, median, Wavelet, Kalman filter, Auto-Regressive Integrated Moving Average (ARIMA), Support Vector Machine (SVM) and Genetic Algorithm (GA) methods. The results indicate that the ANN + PSO method is quite promising and deserves serious attention as a new tool for thermal and TEC seismo anomalies detection.     

Modelling of air temperature using remote sensing and artificial neural network in Turkey

The aim of this research was to forecast monthly mean air temperature based on remote sensing and artificial neural network (ANN) data by using twenty cities over Turkey. ANN contained an input layer, hidden layer and an output layer. While city, month, altitude, latitude, longitude, monthly mean land surface temperatures were chosen as inputs, and monthly mean air temperature was chosen as output for network. Levenberg–Marquardt (LM) learning algorithms and tansig, logsig and linear transfer functions were used in the network. The data of Turkish State Meteorological Service (TSMS) and Technological Research Council of Turkey–Bilten for the period from 1995 to 2004 were chosen as training when the data of 2005 year were being used as test. Result of research was evaluated according to statistical rules. The best linear correlation coefficient (R), and root mean squared error (RMSE) between the estimated and measured values for monthly mean air temperature with ANN and remote sensing method were found to be 0.991–1.254 K, respectively.     

太阳质子事件警报

采用人工神经网络预报方法,利用太阳耀斑的日面位置、X射线辐射的峰值流量及其上升时间、2695MHz和8800MHz微波辐射的半积分流量等5个物理参量,提出了一个新的太阳质子事件警报方案,预报太阳质子事件的发生及其流量和时间．该方案在本文检验中达到93．75％的预报准确率．     

风云二号星上转发器功率放大器的非线性分析与模拟

针对风云二号气象卫星星上转发器放大器的非线性特性及其带来的交调问题进行了分析。对实测给出的放大器单载波功率转移曲线的非线性数据,采用多项式分析和人工神经网络(ANN)模拟的方法进行了拟合和分析,并对所得结果作了比较。实验表明,文中提出的用ANN 概念和方法进行非线性分析与模拟,对解决非线性问题具有一定的理论意义和实际应用价值。     

基于神经网络及深层充电的电子通量反演模型

为了实现对空间高能电子通量的估计及航天器深层充放电的风险评估,基于深层充电和空间电子环境的关联性,利用人工神经网络（ANN）建立了一种由深层充电反演空间高能电子环境的模型。以深层充电探测电流密度及电子能量作为模型输入,电子通量作为模型输出,使用AE9模型对神经网络进行训练,将神经网络的MSE降低到了0.04122,并使用Giove A卫星的深层充电探测数据及GOES卫星的电子通量探测数据验证了模型反演电子环境的准确性。同时对由探测电流计算航天器典型介质材料最大内电场的神经网络模型进行了研究,以实现对航天器内充电风险实时评估。     

Models for estimating daily photosynthetically active radiation in oceanic and mediterranean climates and their improvement by site adaptation techniques

In this work Photosynthetically Active Radiation (PAR) in oceanic and mediterranean climates is modeled. Twenty-two different models have been developed and tested: eleven Multilinear Regression (MR) models and eleven Artificial Neuron Network (ANN) models, using combinations of variables such as Global Horizontal Irradiance (GHI), Global Extraterrestrial Irradiance (G₀), Temperature (T) and Relative Humidity (RH). Data provided by Satellite Application Facility on Climate Monitoring (CM SAF) are used to develop and train the models, while the models have been validated using field data from four stations located in Spain, covering the different study climates.According to the results, zones with different climate conditions need different models, both for the case of MR and ANN. The results show the need of including the GHI in all models in order to obtain accurate estimates; in fact, the presence of more variables only improves slightly the results in mediterranean climate, while in oceanic climate no improvement is observed.On the other hand, comparing MR and ANN models, ANN models did not show better results than those of MR models in no one of the cases studied. Regarding the climate, both types of models are clearly better for the mediterranean case than for the oceanic one. In order to improve the performance of the model for oceanic climate a correction based on the site adaptation technique was carried out. The good results obtained by this technique fully justify its use.The best proposed models provide better performance than other models which are restricted to certain locations. Besides, the clustering technique based on the PAR variable, used in this work, allows obtaining useful models for a whole region. Finally, another advantage of this methodology is that there is no need of ground measurements for its development, except for the site adaptation technique.