Integrated classification method of tight sandstone reservoir based on principal component analysise simulated annealing genetic algorithmefuzzy cluster means

In this research,an integrated classification method based on principal component analysis-simulated annealing genetic algorithm-fuzzy cluster means(PCA-SAGA-FCM)was proposed for the unsupervised classification of tight sandstone reservoirs which lack the prior information and core experiments.A variety of evaluation parameters were selected,including lithology characteristic parameters,poro-permeability quality characteristic parameters,engineering quality characteristic parameters,and pore structure characteristic parameters.The PCA was used to reduce the dimension of the evaluation pa-rameters,and the low-dimensional data was used as input.The unsupervised reservoir classification of tight sandstone reservoir was carried out by the SAGA-FCM,the characteristics of reservoir at different categories were analyzed and compared with the lithological profiles.The analysis results of numerical simulation and actual logging data show that:1)compared with FCM algorithm,SAGA-FCM has stronger stability and higher accuracy;2)the proposed method can cluster the reservoir flexibly and effectively according to the degree of membership;3)the results of reservoir integrated classification match well with the lithologic profle,which demonstrates the reliability of the classification method.

Polarimetric Meteorological Satellite Data Processing Software Classification Based on Principal Component Analysis and Improved K-Means Algorithm

With the increasing variety of application software of meteorological satellite ground system, how to provide reasonable hardware resources and improve the efficiency of software is paid more and more attention. In this paper, a set of software classification method based on software operating characteristics is proposed. The method uses software run-time resource consumption to describe the software running characteristics. Firstly, principal component analysis (PCA) is used to reduce the dimension of software running feature data and to interpret software characteristic information. Then the modified K-means algorithm was used to classify the meteorological data processing software. Finally, it combined with the results of principal component analysis to explain the significance of various types of integrated software operating characteristics. And it is used as the basis for optimizing the allocation of software hardware resources and improving the efficiency of software operation.

NONLINEAR DATA RECONCILIATION METHOD BASED ON KERNEL PRINCIPAL COMPONENT ANALYSIS

In the industrial process situation, principal component analysis (PCA) is ageneral method in data reconciliation. However, PCA sometime is unfeasible to nonlinear featureanalysis and limited in application to nonlinear industrial process. Kernel PCA (KPCA) is extensionof PCA and can be used for nonlinear feature analysis. A nonlinear data reconciliation method basedon KPCA is proposed. The basic idea of this method is that firstly original data are mapped to highdimensional feature space by nonlinear function, and PCA is implemented in the feature space. Thennonlinear feature analysis is implemented and data are reconstructed by using the kernel. The datareconciliation method based on KPCA is applied to ternary distillation column. Simulation resultsshow that this method can filter the noise in measurements of nonlinear process and reconciliateddata can represent the true information of nonlinear process.

Research and Application of a Multi-Field Co-Simulation Data Extraction Method Based on Adaptive Infinitesimal Element

Regarding the spatial profile extraction method of a multi-field co-simulation dataset,different extraction directions,locations,and numbers of profileswill greatly affect the representativeness and integrity of data.In this study,a multi-field co-simulation data extractionmethod based on adaptive infinitesimal elements is proposed.Themultifield co-simulation dataset based on related infinitesimal elements is constructed,and the candidate directions of data profile extraction undergo dimension reduction by principal component analysis to determine the direction of data extraction.Based on the fireworks algorithm,the data profile with optimal representativeness is searched adaptively in different data extraction intervals to realize the adaptive calculation of data extraction micro-step length.The multi-field co-simulation data extraction process based on adaptive microelement is established and applied to the data extraction process of the multi-field co-simulation dataset of the sintering furnace.Compared with traditional data extraction methods for multi-field co-simulation,the approximate model constructed by the data extracted from the proposed method has higher construction efficiency.Meanwhile,the relative maximum absolute error,root mean square error,and coefficient of determination of the approximationmodel are better than those of the approximation model constructed by the data extracted from traditional methods,indicating higher accuracy,it is verified that the proposed method demonstrates sound adaptability and extraction efficiency.

Application of nonlinear partial least square in catalyst modeling

The neural network partial least square (NNPLS) method was used to establish a robust reaction model for a multi-component catalyst of methane oxidative coupling. The details, including the learning algorithm, the number of hidden units of the inner network, activation function, initialization of the network weights and the principal components, are discussed. The results show that the structural organizations of inner neural network are 1-10-5-1, 1-8-4-1, 1-8-5-1, 1-7-4-1, 1-8-4-1, 1-8-6-1, respectively. The Levenberg-Marquardt method was used in the learning algorithm, and the central sigmoidal function is the activation function. Calculation results show that four principal components are convenient in the use of the multi-component catalyst modeling of methane oxidative coupling. Therefore a robust reaction model expressed by NNPLS succeeds in correlating the relations between elements in catalyst and catalytic reaction results. Compared with the direct network modeling, NNPLS model can be adjusted by experimental data and the calculation of the model is simpler and faster than that of the direct network model.

Low-dimensional multi-spectral space for color reproduction based on nonnegative constrained principal component analysis

In order to overcome the shortcomings that the reconstructed spectral reflectance may be negative when using the classic principal component analysis （PCA）to reduce the dimensions of the multi-spectral data, a nonnegative constrained principal component analysis method is proposed to construct a low-dimensional multi-spectral space and accomplish the conversion between the new constructed space and the multispectral space. First, the reason behind the negative data is analyzed and a nonnegative constraint is imposed on the classic PCA. Then a set of nonnegative linear independence weight vectors of principal components is obtained, by which a lowdimensional space is constructed. Finally, a nonlinear optimization technique is used to determine the projection vectors of the high-dimensional multi-spectral data in the constructed space. Experimental results show that the proposed method can keep the reconstructed spectral data in [ 0, 1 ]. The precision of the space created by the proposed method is equivalent to or even higher than that by the PCA.

Comparison of Algorithms for an Electronic Nose in Identifying Liquors

When the electronic nose is used to identify different varieties of distilled liquors, the pattern recognition algorithm is chosen on the basis of the experience, which lacks the guiding principle. In this research, the different brands of distilled spirits were identified using the pattern recognition algorithms （principal component analysis and the artificial neural network）. The recognition rates of different algorithms were compared. The recognition rate of the Back Propagation Neural Network （BPNN） is the highest. Owing to the slow convergence speed of the BPNN, it tends easily to get into a local minimum. A chaotic BPNN was tried in order to overcome the disadvantage of the BPNN. The convergence speed of the chaotic BPNN is 75.5 times faster than that of the BPNN.

Application of SVM and PCA-CS algorithms for prediction of strip crown in hot strip rolling

To make up the poor quality defects of traditional control methods and meet the growing requirements of accuracy for strip crown,an optimized model based on support vector machine(SVM)is put forward firstly to enhance the quality of product in hot strip rolling.Meanwhile,for enriching data information and ensuring data quality,experimental data were collected from a hot-rolled plant to set up prediction models,as well as the prediction performance of models was evaluated by calculating multiple indicators.Furthermore,the traditional SVM model and the combined prediction models with particle swarm optimization(PSO)algorithm and the principal component analysis combined with cuckoo search(PCA-CS)optimization strategies are presented to make a comparison.Besides,the prediction performance comparisons of the three models are discussed.Finally,the experimental results revealed that the PCA-CS-SVM model has the highest prediction accuracy and the fastest convergence speed.Furthermore,the root mean squared error(RMSE)of PCA-CS-SVM model is 2.04μm,and 98.15%of prediction data have an absolute error of less than 4.5μm.Especially,the results also proved that PCA-CS-SVM model not only satisfies precision requirement but also has certain guiding significance for the actual production of hot strip rolling.

Improved Face Recognition Method Using Genetic Principal Component Analysis

An improved face recognition method is proposed based on principal component analysis （PCA） compounded with genetic algorithm （GA）, named as genetic based principal component analysis （GPCA）. Initially the eigenspace is created with eigenvalues and eigenvectors. From this space, the eigenfaces are constructed, and the most relevant eigenfaees have been selected using GPCA. With these eigenfaees, the input images are classified based on Euclidian distance. The proposed method was tested on ORL （Olivetti Research Labs） face database. Experimental results on this database demonstrate that the effectiveness of the proposed method for face recognition has less misclassification in comparison with previous methods.

Spatiotemporal Modeling of Monthly Precipitation in the Upper Shiyang River Watershed in West Central Gansu,Northwest China

This paper outlines a methodology to estimate monthly precipitation surfaces at 1-kin resolution for the Upper Shiyang River watershed （USRW） in northwest China. Generation of precipitation maps is based on the application of a four-variable genetic algorithm （GA） trained on 10 years of weather and ancillary data, i.e., surface air temperature, relative humidity, Digital Elevation Model-derived estimates of elevation, and time of year collected at 29 weather stations in west-central Gansu and northern Qinghai province. An observed-to-GA predicted data comparison of 10 years of precipitation collected at the 29 weather stations showed that about 84% of the variability in observed values could be explained by the trained GA, including variability in two independent datasets. Point-comparisons of observed and modeled precipitation along an elevation-rainfall gradient demonstrated near-similar spatiotemporal patterns. A precipitation surface for USRW for July, 2005, was developed with the trained GA and input surfaces of surface air temperature and relative humidity generated from Moderate Resolution Imaging Spectroradiometer sensor （MODIS） products of land surface temperature. Spatial tendencies in predicted maximum and minimum values of surface air temperature, relative humidity, and precipitation within a 2-kin radius circle around selected weather stations were in close agreement with the values measured at the weather stations.

Support vector classifier based on principal component analysis

Support vector classifier （SVC） has the superior advantages for small sample learning problems with high dimensions, with especially better generalization ability. However there is some redundancy among the high dimensions of the original samples and the main features of the samples may be picked up first to improve the performance of SVC. A principal component analysis （PCA） is employed to reduce the feature dimensions of the original samples and the pre-selected main features efficiently, and an SVC is constructed in the selected feature space to improve the learning speed and identification rate of SVC. Furthermore, a heuristic genetic algorithm-based automatic model selection is proposed to determine the hyperparameters of SVC to evaluate the performance of the learning machines. Experiments performed on the Heart and Adult benchmark data sets demonstrate that the proposed PCA-based SVC not only reduces the test time drastically, but also improves the identify rates effectively.

A nonlinear PCA algorithm based on RBF neural networks

Traditional PCA is a linear method, but most engineering problems are nonlinear. Using the linear PCA in nonlinear problems may bring distorted and misleading results. Therefore, an approach of nonlinear principal component analysis (NLPCA) using radial basis function (RBF) neural network is developed in this paper. The orthogonal least squares (OLS) algorithm is used to train the RBF neural network. This method improves the training speed and prevents it from being trapped in local optimization. Results of two experiments show that this NLPCA method can effectively capture nonlinear correlation of nonlinear complex data, and improve the precision of the classification and the prediction.

Nonlinear Dynamic Analysis of MPEG-4 Video Traffic

The main research motive is to analysis and to veiny the inherent nonlinear character of MPEG-4 video. The power spectral density estimation of the video trafiic describes its 1/f^β and periodic characteristics.The priraeipal compohems analysis of the reconstructed space dimension shows only several principal components can be the representation of all dimensions. The correlation dimension analysis proves its fractal characteristic. To accurately compute the largest Lyapunov exponent, the video traffic is divided into many parts.So the largest Lyapunov exponent spectrum is separately calculated using the small data sets method. The largest Lyapunov exponent spectrum shows there exists abundant nonlinear chaos in MPEG-4 video traffic. The conclusion can be made that MPEG-4 video traffic have complex nonlinear be havior and can be characterized by its power spectral density,principal components, correlation dimension and the largest Lyapunov exponent besides its common statistics.

Hyper-Parameter Optimization of Semi-Supervised GANs Based-Sine Cosine Algorithm for Multimedia Datasets

Generative Adversarial Networks(GANs)are neural networks that allow models to learn deep representations without requiring a large amount of training data.Semi-Supervised GAN Classifiers are a recent innovation in GANs,where GANs are used to classify generated images into real and fake and multiple classes,similar to a general multi-class classifier.However,GANs have a sophisticated design that can be challenging to train.This is because obtaining the proper set of parameters for all models-generator,discriminator,and classifier is complex.As a result,training a single GAN model for different datasets may not produce satisfactory results.Therefore,this study proposes an SGAN model(Semi-Supervised GAN Classifier).First,a baseline model was constructed.The model was then enhanced by leveraging the Sine-Cosine Algorithm and Synthetic Minority Oversampling Technique(SMOTE).SMOTE was used to address class imbalances in the dataset,while Sine Cosine Algorithm(SCA)was used to optimize the weights of the classifier models.The optimal set of hyperparameters(learning rate and batch size)were obtained using grid manual search.Four well-known benchmark datasets and a set of evaluation measures were used to validate the proposed model.The proposed method was then compared against existing models,and the results on each dataset were recorded and demonstrated the effectiveness of the proposed model.The proposed model successfully showed improved test accuracy scores of 1%,2%,15%,and 5%on benchmarking multimedia datasets;Modified National Institute of Standards and Technology(MNIST)digits,Fashion MNIST,Pneumonia Chest X-ray,and Facial Emotion Detection Dataset,respectively.

Device Anomaly Detection Algorithm Based on Enhanced Long Short-Term Memory Network

The problems in equipment fault detection include data dimension explosion,computational complexity,low detection accuracy,etc.To solve these problems,a device anomaly detection algorithm based on enhanced long short-term memory(LSTM)is proposed.The algorithm first reduces the dimensionality of the device sensor data by principal component analysis(PCA),extracts the strongly correlated variable data among the multidimensional sensor data with the lowest possible information loss,and then uses the enhanced stacked LSTM to predict the extracted temporal data,thus improving the accuracy of anomaly detection.To improve the efficiency of the anomaly detection,a genetic algorithm(GA)is used to adjust the magnitude of the enhancements made by the LSTM model.The validation of the actual data from the pumps shows that the algorithm has significantly improved the recall rate and the detection speed of device anomaly detection,with the recall rate of 97.07%,which indicates that the algorithm is effective and efficient for device anomaly detection in the actual production environment.

局部线性下的函数型主成分聚类算法

函数型聚类分析在统计学领域被广泛关注,其分析过程通常在降维目标实现后进行。为了有效解决函数型主成分聚类问题,文章结合局部线性嵌入算法(Locally Linear Embedding,LLE)在非线性空间下的适用性,提出了一种局部线性下的函数型主成分分析模型(LLE Function Principle Component Analysis,LFPCA)。首先,采用函数型主成分分析法作为降维目标方法,改进了FPCA的算法模型,通过将LLE算法的权重系数矩阵与函数型主成分定义相结合,构建出一个适用于非线性空间下的聚类算法;其次,在求解算法的过程中定义了函数型主成分得分,并结合EM算法构建出GMM模型来近似函数型算法的概率密度函数,使模型更高效且适用性更强;最后,通过随机模拟实验及应用分析验证了LFPCA算法模型在真实数据集上具有良好的聚类效能。

多策略改进黏菌算法阶段优化HSVM变压器故障辨识

为解决变压器故障诊断精度较低的问题,提出了一种多策略改进黏菌算法(ISMA)阶段优化混合核支持向量机(HSVM)的变压器故障诊断新方法。首先,利用主成分分析(PCA)来消除变量之间的信息冗余并降低数据集维度。其次,引入黏菌算法(SMA),并结合Logistic混沌映射、二次插值、自适应权重多策略改进SMA,以提高SMA算法收敛速度和局部搜索能力;然后,与原始SMA、WHO和GWO算法进行寻优测试,对比验证改进后SMA算法的优越性;最后,使用改进SMA算法分阶段对混合核支持向量机参数寻优,构建ISMA-HSVM变压器故障诊断模型。将降维后的特征数据输入HSVM模型与BPPN、ELM和SVM进行比较,HSVM模型的诊断准确性分别提高了5.55%、8.89%、5.55%。使用ISMA优化HSVM模型参数,与WHO、GWO、SMA算法优化效果比较,结果准确性提高了13.33%、12.22%、5.55%。其中,ISMA-HSVM模型的诊断精度为93.33%。实验结果表明,所提模型有效提升故障诊断分类性能,且具有较高的故障诊断精度。

基于机器学习的茶树DNA聚类算法

为了研究茶树基因序列的聚类问题,设计一种基于累计方差贡献率进行改进的核主成分分析(KPCA)与k均值(k-means)++聚类算法相结合的降维聚类算法(KPCA-k-means++)。将基因库数据集筛选分组后,利用k-mers算法提取基因数据的数据特征,根据累计方差贡献率的占比大于85%的标准确定降维主元个数对KPCA进行降维改进并采用k-means++算法对降维后数据聚类,通过CH(Calinski-Harabaze Index)指标和响应时间分析聚类结果。结果表明:在单独聚类、KPCA聚类、改进PCA聚类、改进KPCA聚类4种处理方式中,改进KPCA-k-means++算法在不同处理方式和不同样本数的对比下,CH指标均为最高,与未改进时相比平均高出33%。在响应时间方面,改进KPCA-k-means++算法与同样改进PCA-k-means++算法在不同聚类数和样本数的对比下响应时间均较短。改进KPCA-k-means++算法能够保证对于茶树的基因序列的聚类准确率和聚类速度,表现出极好的聚类稳定性。

基于IWOA-ELM的模拟电路故障诊断方法

针对模拟电路故障诊断中非线性和高维度输出信号带来的诊断困难问题,提出一种基于改进鲸鱼算法(IWOA)优化极限学习机(ELM)的模拟电路故障诊断方法。首先,采用主成分分析(PCA)法对初始故障电路特征进行降维;其次,在鲸鱼算法的基础上引入Tent映射来初始化种群,并且加入了非线性时变因子、自适应权重以及随机差分变异策略;再利用改进后的鲸鱼算法对ELM进行优化;最后将降维后的故障特征向量输入ELM中得到故障诊断结果。通过Sallen-Key带通滤波器电路以及CSTV滤波器电路仿真测试实例表明:IWOA优化ELM的故障诊断方法具有更优的故障诊断性能,故障诊断准确率高达99.41%。