- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
Hyperspectral signature analysis using neural network for grade estimation of copper ore
摘要： The ever-increasing demand for the different metal and mineral resources from the earth’s subsurface has brought tremendous pressure on the geochemical laboratory for the growing countries. The success of any mining industry relies on the estimated values of ore grade in the mineral deposit. Hence, rapid assessment of ore grade is critical in daily schedule in mines operations. Commonly the assay value is determined by chemical analysis or X-Ray Fluorescence (XRF), which is one of the constrained by real-time grade estimation, duration of sample preparation and processing. Several researches carried out in exploration and revealed that hyperspectral technique is a promising tool for mineral identification and mapping. The goal of the present study is to determine the effectiveness of narrow band spectroscopy in Cu grade estimation. To achieve this, a multilayer feed-forward neural network model has been developed to establish a functional link between hyperspectral signature derived features with the copper grade. Altogether eight different types of features including absorption depth, band depth center, the area under the absorption curve, full width at half maxima were extracted from continuum removed spectra along with derivative reflectance features, e.g. band depth ratio, 1st and 2nd slopes from the hyperspectral profile. The dimensionality was reduced by applying Principal Component Analysis onto the extracted features. The first seven PCAs are then used as input vector of the ANN model. A five-fold cross-validation exercise is carried out for model performance. The high degree of correlation reveals that the PCA generated feature from hyperspectral data coupled with ANN model could be an alternative approach to predict the copper grade for the copper mine.
关键词： copper grade,ore grade estimation,spectral feature,K-Fold cross validation,principal component analysis,artificial neural network
更新于2025-09-23 15:22:29
-
[IEEE IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Valencia, Spain (2018.7.22-2018.7.27)] IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - The Difference of Rrs Product Derived From Modis Meris and Seawifs in South China Sea
摘要： This study presents the difference of normalized remote sensing reflectance (Rrs) between SeaWiFS, MERIS and MODIS with in situ data in the SCS. The results show that all the satellite-derived Rrs are less than the in situ data, and the APD(Absolute Percentage Difference) of the Rrs from the three satellite sensors are mostly less than 20%, except band 412 and 660 nm in global ocean and band 555and 660nm in SCS. The blue band of the three missions have the good performances, whose APD are less than 15%. The performances of MODIS Rrs is better than MERIS (or SeaWiFS). And, compared with each other in the SCS , the results show that the Rrs of the three missions are generally agree well with each other, but they still have big differences in the coastal area. The MODIS Rrs is bigger than MERIS’s and SeaWiFS’s in the 412, 443 and 488 nm band. The differences between MERIS (or SeaWiFS) and MODIS change with band (488 nm is the best one). And the differences among three missions for blue band has seasonal variation in the SCS from 2008 to 2010.
关键词： Validation,SCS,cross validation,MODIS/MERIS/SeaWiFS,Rrs
更新于2025-09-23 15:21:21
-
Machine learning-based mapping of micro-topographic earthquake-induced paleo Pulju moraines and liquefaction spreads from a digital elevation model acquired through laser scanning
摘要： The advent of public open source airborne laser scanning-produced digital elevation models (ALS DEM) has provided new perspectives on glacial geomorphology in the Nordic countries. Seismically-induced micro-topographic paleo-landforms can now be identified and mapped throughout the former Fennoscandian Ice Sheet, allowing spatial safety assessment for nuclear waste disposal. Automated machine learning techniques enable recognition of these fine-scale geomorphological features efficiently and in a consistent way nationwide. The current study focuses on automated recognition of paleo liquefaction spreads and Pulju moraines in northern Finland. Geomorphometric variables in different cell sizes were first derived from the 2 m ALS DEM by Gabor and principal curvature filtering to emphasize the elevational multi-scale texture of these paleo-seismic landforms. The Gabor textural variables were considered as a baseline method and the principal curvature features, including maximum and minimum curvature, were used because they have previously been proven critical in recognition of concave and convex elongated features. Both sets of raster variables were then turned into histogram-based features and input into a non-linear supervised multilayer perceptron early-stop committee which is a neural network classifier. The leave-one-out cross-validation performance results indicated principal curvature features to be highly successful with 94% accuracy. Principal curvatures provided a clear improvement to Gabor based features which provided significantly lower accuracies between 83?85%. The study demonstrates the high success of supervised neural network-based classification of ALS DEM data and derived textural features capturing the multi-scale nature of the micro-topographic liquefaction spreads and Pulju moraines. The approach could be utilized for time-efficient mapping of these paleo-seismic geomorphologies to complete paleo-seismic databases in formerly glaciated regions.
关键词： rotation invariant,histogram-based features,leave-one-out cross-validation,principal curvature,area invariant,multilayer perceptron,landforms,paleo-seismology,geomorphology,Gabor filter,Pulju moraine,liquefaction spreads
更新于2025-09-23 15:19:57
-
Fault diagnosis method of photovoltaic array based on support vector machine
摘要： Photovoltaic (PV) arrays are prone to various faults due to the hostile working environment. This paper presents the fault diagnosis algorithm based on support vector machine (SVM) to detect short circuit, open circuit, and lack of irradiation faults that occurred in PV arrays. By analyzing these faults and I–V characteristic curves of PV arrays, the short-circuit current, open-circuit voltage, maximum-power current, and maximum-power voltage are chosen as input parameters of SVM-based fault diagnosis algorithm. The data pre-processing methods are used to improve the quality of fault data set considering the effects of the quality on the performance of SVM-based fault diagnosis algorithm. The grid search and k-fold cross-validation methods are proposed to optimize the parameters of the SVM-based fault diagnosis algorithm. It gets test accuracy of 97% by testing the trained SVM-based fault diagnosis algorithm with 400 data. The experimental results indicate that the SVM-based fault diagnosis algorithm has higher accuracy and generalization ability than other algorithm for fault diagnosis of PV arrays.
关键词： k-fold cross-validation,PV arrays,data preprocessing,grid search,SVM-based fault diagnosis algorithm
更新于2025-09-12 10:27:22
-
Fault Diagnosis Method of Photovoltaic Array Based on BP Neural Network
摘要： Photovoltaic arrays are prone to various failures due to long-term work. In order to quickly and accurately diagnose the type of failure of the PV array and implement online monitoring of the PV array, this paper proposes the BP neural network for PV array fault diagnosis, and proposes a network search method when training BP neural network. And the K-cross-validation method is used to select the number of hidden layer nodes. The BP neural network fault diagnosis model designed and trained by this method is proved to have high precision.
关键词： BP neural network,hidden layer nodes,fault diagnosis,K-cross-validation,Photovoltaic array
更新于2025-09-11 14:15:04
-
[IEEE IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Valencia, Spain (2018.7.22-2018.7.27)] IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Land Surface Temperature Retrieval from the Infrared Measurements of Advanced Himawari Imager on Himawari-8
摘要： This work addresses Land Surface Temperature (LST) retrieval from the infrared measurements of Advanced Himawari Imager (AHI) on Himawari-8 satellite using the Generalized Split-Window (GSW) algorithm. First, a radiative transfer modeling experiment is conducted using the moderate spectral resolution atmospheric transmittance algorithm and computer model (MODTRAN) 4.0 fed with the SeeBor V5.0 atmospheric profile database to simulate the brightness temperatures in the AHI channels 14 (centered at about 11.2 μm) and 15 (centered at about 12.3 μm) related to Land Surface Emissivities (LSEs) and Total Precipitable Water (TPW). Then, the unknown coefficients of the GSW algorithm are obtained through multi-variable linear regression, in which the simulated data are grouped into several sub-ranges to improve algorithm accuracy. Next, LSTs are derived from the clear-sky AHI measurements in September 2016 over a study area with longitude from 100°E to 145°E and latitude from 15°N to 45°N, where LSEs are deduced from the MOD11C1 V6 product using the baseline fit method, and TPWs are extracted from the European Centre for Medium-range Weather Forecasts (ECMWF) reanalysis data. Finally, the derived LSTs are cross-validated with the MOD11C1 V6 product. The results show that the GSW algorithm developed in this work can accurately retrieve LST from the AHI measurements, and the error is 0.39±1.62 K against the MOD11C1 V6 product.
关键词： Land surface temperature,Advanced Himawari Imager (AHI),the Generalized Split-Window (GSW) algorithm,cross-validation
更新于2025-09-10 09:29:36
-
Fast Microwave Through Wall Imaging Method With Inhomogeneous Background Based on Levenberg-Marquardt Algorithm
摘要： In this paper, a fast solution for microwave through wall imaging (TWI) with nonlinear inversion is proposed to reconstruct the unknown targets embedded in an inhomogeneous background medium. We treat inhomogeneous background, the wall around bounded in a finite domain as a known scatterer, which has the advantage of avoiding the time-consuming calculation of inhomogeneous background Green’s function. Under this scheme, a new approach under the framework of difference integral equation model, i.e., difference Lippmann–Schwinger integral equation, with modified enhanced Levenberg–Marquardt algorithm is proposed. In particular, we used a hybrid regularized technique, i.e., generalized cross-validation and truncated singular value decomposition, to stabilize the inversion. It is shown that the proposed method runs fast and is stable in presence of noise. Also, it is able to alleviate the nonlinearity and reconstruct unknown scatterers of high contrast with respect to the background. Both the numerical and experimental TWI tests validate the efficiency of the proposed inversion method.
关键词： microwave imaging,Generalized cross-validation (GCV) regularization,inhomogeneous background,inverse scattering problems (ISPs),Levenberg–Marquardt (LM) method
更新于2025-09-10 09:29:36
-
[IEEE 2018 37th Chinese Control Conference (CCC) - Wuhan (2018.7.25-2018.7.27)] 2018 37th Chinese Control Conference (CCC) - Near-Infrared Spectrum of Coal Origin Identification Based on SVM Algorithm
摘要： Near infrared spectroscopy is introduced to analyze 243 coal samples of different origins of Australia, Canada, China, Indonesia and Russia, combined with the supportive vector machines (SVM) analysis method. With the pre-processed data from the Principal component analysis (PCA), six supportive vector machines with different kernel functions are employed to discriminate origins of coal samples, namely Linear SVM, Quadratic SVM, Cubic SVM, Fine Gaussian SVM, Medium Gaussian SVM and Coarse Gaussian SVM. Through comparison, Linear SVM has the best performance in prediction accuracy rate while better results are obtained using Medium Gaussian SVM taking accuracy rate and training time into account. It turns out that NIR spectroscopy combined with Medium Gaussian SVM can be used as a good non-destructive method to predict origins of coal, with an accuracy rate of 98.8%, which strengthens the supervision of coal quality.
关键词： Principal component analysis,K- fold cross validation,Supportive vector machines,Near infrared spectroscopy
更新于2025-09-10 09:29:36