Least Squares Support Vector Machine Method Research Articles

Quantitative structure activity relationship study of inhibitory activities of 5-lipoxygenase and design new compounds by different chemometrics methods

A quantitative structure-activity relationship (QSAR) study was conducted for the prediction of inhibitory activity of 1-phenyl(2H)-tetrahydro-triazine-3-one analogues as inhibitors of 5-Lipoxygenase. The inhibitory activities of the 1-phenyl(2H)- tetrahydro-triazine-3-one analogues modeled as a function of molecular structures using chemometrics methods such as multiple linear regression (MLR) and least squares support vector machines (LS-SVM). The obtained models were applied to predict the inhibitory activity of compounds which were not in the modeling procedure. The results of models showed high prediction ability with root mean square error of prediction of 0.167 and 0.061 for MLR and LS-SVM, respectively. The LS-SVM method was used for prediction of inhibitory activity of the new inhibitor derivatives.

Comparison of Statistical Downscaling Methods for Monthly Total Precipitation: Case Study for the Paute River Basin in Southern Ecuador

Downscaling improves considerably the results of General Circulation Models (GCMs). However, little information is available on the performance of downscaling methods in the Andean mountain region. The paper presents the downscaling of monthly precipitation estimates of the NCEP/NCAR reanalysis 1 applying the statistical downscaling model (SDSM), artificial neural networks (ANNs), and the least squares support vector machines (LS-SVM) approach. Downscaled monthly precipitation estimates after bias and variance correction were compared to the median and variance of the 30-year observations of 5 climate stations in the Paute River basin in southern Ecuador, one of Ecuador’s main river basins. A preliminary comparison revealed that both artificial intelligence methods, ANN and LS-SVM, performed equally. Results disclosed that ANN and LS-SVM methods depict, in general, better skills in comparison to SDSM. However, in some months, SDSM estimates matched the median and variance of the observed monthly precipitation depths better. Since synoptic variables do not always present local conditions, particularly in the period going from September to December, it is recommended for future studies to refine estimates of downscaling, for example, by combining dynamic and statistical methods, or to select sets of synoptic predictors for specific months or seasons.

A novel denitration cost optimization system for power unit boilers

Reducing the denitration cost of coal-fired boilers is important to enhance the competitiveness of power generation companies. This study proposes a real operation data-based denitration cost optimization system that guides operators in economically adjusting the operation parameters of boilers. A data-driven least square support vector machine (LSSVM) learning method is utilized to predict the denitration cost of a coal-fired boiler. Back propagation (BP) is used here to select the input variables to simplify the model. With the pre-built BP-LSSVM-based denitration cost model, the genetic algorithm (GA) is then applied to obtain offline optimizations at the typical operating load points, which results in an Offline Optimal Expert Database (OOED). Once a load command is received, fuzzy association rule mining (FARM) is employed to extract the relationship between the operating load point and the optimal adjustable variables (AVs) in the OOED, thereby achieving the online denitration cost optimization of the power plant. For comparison, a single LSSVM method is also employed to build a denitration cost prediction model, and the GA and FARM proposed in this study are compared too. The results show that, compared with the single LSSVM method, the BP-LSSVM method not only predicts more accurately but also lowers the model complexity. In addition, considering the denitration cost, optimization time, and update time, the proposed BP-LSSVM-GA-FARM-based denitration cost optimization system is always better than traditional optimization methods.

Improving Forecasting Accuracy of Streamflow Time Series Using Least Squares Support Vector Machine Coupled with Data-Preprocessing Techniques

Highly reliable forecasting of streamflow is essential in many water resources planning and management activities. Recently, least squares support vector machine (LSSVM) method has gained much attention in streamflow forecasting due to its ability to model complex non-linear relationships. However, LSSVM method belongs to black-box models, that is, this method is primarily based on measured data. In this paper, we attempt to improve the performance of LSSVM method from the aspect of data preprocessing by singular spectrum analysis (SSA) and discrete wavelet analysis (DWA). Kharjeguil and Ponel stations from Northern Iran are investigated with monthly streamflow data. The root mean square error (RMSE), mean absolute error (MAE), correlation coefficient (R) and coefficient of efficiency (CE) statistics are used as comparing criteria. The results indicate that both SSA and DWA can significantly improve the performance of forecasting model. However, DWA seems to be superior to SSA and able to estimate peak streamflow values more accurately. Thus, it can be recommended that LSSVM method coupled with DWA is more promising.

Estimation of the silica solubility in the superheated steam using LSSVM modeling approach

The presence of silica (SiO2) in boiler water causes precipitation and creates hard silicate scale on steam turbine blades. This study assessed the ability of least squares support vector machines (LSSVM) modeling approaches to estimate the solubility of SiO2 in the steam of boilers. A genetic algorithm (GA) and population‐based stochastic search algorithms were employed to identify the optimal LSSVM method variables. Results indicate that the GA‐LSSVM can be used to model the complicated nonlinear relationship between the input and output variables. To predict the solubility of SiO2 in the steam of boilers, the GA‐LSSVM model generated the mean absolute error (MAE) and the coefficient of determination (R2) values of 1.8831 and 0.9997, respectively, for the entire data set. The proposed model provides a distinctly promising approach to estimating the solubility of SiO2 in the steam of boilers. © 2015 American Institute of Chemical Engineers Environ Prog, 35: 596–602, 2016

The application of continuous wavelet transform and least squares support vector machine for the simultaneous quantitative spectrophotometric determination of Myricetin, Kaempferol and Quercetin as flavonoids in pharmaceutical plants

Flavonoids are γ-benzopyrone derivatives, which are highly regarded in these researchers for their antioxidant property. In this study, two new signals processing methods been coupled with UV spectroscopy for spectral resolution and simultaneous quantitative determination of Myricetin, Kaempferol and Quercetin as flavonoids in Laurel, St. John’s Wort and Green Tea without the need for any previous separation procedure. The developed methods are continuous wavelet transform (CWT) and least squares support vector machine (LS-SVM) methods integrated with UV spectroscopy individually. Different wavelet families were tested by CWT method and finally the Daubechies wavelet family (Db4) for Myricetin and the Gaussian wavelet families for Kaempferol (Gaus3) and Quercetin (Gaus7) were selected and applied for simultaneous analysis under the optimal conditions. The LS-SVM was applied to build the flavonoids prediction model based on absorption spectra. The root mean square errors for prediction (RMSEP) of Myricetin, Kaempferol and Quercetin were 0.0552, 0.0275 and 0.0374, respectively. The developed methods were validated by the analysis of the various synthetic mixtures associated with a well- known flavonoid contents. Mean recovery values of Myricetin, Kaempferol and Quercetin, in CWT method were 100.123, 100.253, 100.439 and in LS-SVM method were 99.94, 99.81 and 99.682, respectively. The results achieved by analyzing the real samples from the CWT and LS-SVM methods were compared to the HPLC reference method and the results were very close to the reference method. Meanwhile, the obtained results of the one-way ANOVA (analysis of variance) test revealed that there was no significant difference between the suggested methods.

Modeling urban growth with GIS based cellular automata and least squares SVM rules: a case study in Qingpu–Songjiang area of Shanghai, China

A critical issue in urban cellular automata (CA) modeling concerns the identification of transition rules that generate realistic urban land use patterns. Recent studies have demonstrated that linear methods cannot sufficiently delineate the extraordinary complex boundaries between urban and non-urban areas and as most urban CA models simulate transitions across these boundaries, there is an urgent need for good methods to facilitate such delineations. This paper presents a machine learning CA model (termed MachCA) with nonlinear transition rules based on least squares support vector machines (LS-SVM) to simulate such urban growth. By projecting the input dataset into a high dimensional space using the LS-SVM method, an optimal hyper-plane is constructed to separate the complex boundaries between urban and nonurban land, thus enabling the retrieval of nonlinear CA transition rules. In the MachCA model, the transition rules are yes–no decisions on whether a cell changes its state or not, the rules being dynamically updated for each iteration of the model implementation. The application of the MachCA for simulating urban growth in the Shanghai Qingpu–Songjiang area in China reveals that the spatial configurations of rural–urban patterns can be modeled. A comparison of the MachCA model with a conventional CA model fitted by logarithmic regression (termed LogCA) shows that the MachCA model produces more hits and less misses and false alarms due to its capability for capturing the spatial complexity of urban dynamics. This results in improved simulation accuracies, although with only less than 1 % deviation between the overall errors produced by the MachCA and LogCA models. Nevertheless, the way MachCA model use in retrieving the transition rules provides a new method for simulating the dynamic process of urban growth.

Prediction of stress concentration factor of corrosion pits on buried pipes by least squares support vector machine

Concentrated working stress of a corrosion pit is known to be an important factor inducing the deterioration and consequently the breakage of a buried pipe. In this paper, we investigate the stress concentration factors (SCFs) of isolated elliptical corrosion pits using 3-D finite element analyses. The elliptical pits are fully characterized by their 3-D geometries, i.e., major and minor diameters and pit depth via a series of parametric studies of finite element pipe models. This is to quantify the effect of geometric variations of the pits on SCF. Realizing the fact that the 3-D finite element analysis is computationally intensive, efficient statistically predictive models have been developed based on least squares support vector machine (LS-SVM) realized in a ubiquitous spreadsheet platform. This approach shows very close predictions of SCF to the numerical findings for elliptical corrosion patterns on buried pipes. Particularly, two typical kernel functions have been adopted to cross validate the excellent performance of the LS-SVM method in SCF prediction of corroded pipes.

A preliminary study on time series forecast of fair-weather atmospheric electric field with WT-LSSVM method

Accurate forecasting of fair-weather atmospheric electric field is of great importance to denote the climate change, and it also plays a vital role in lightning forecasting. In this paper, the WT (Wavelet transform)-LSSVM (Least squares support vector machine) method is proposed to develop a model to forecast time series of fair-weather atmospheric electric field. The application experiments with data from Verhnee Dubrovo, Dusheti, and Voeikovo stations show that the predicted data and observed value have the same tendency. The correlation coefficients between them are 0.8149, 0.6395 and 0.7308, while the MSE are 16.9869 V/m, 28.8876 V/m and 32.2163 V/m, respectively. Furthermore, comparing to LSSVM and ANN, WT-LSSVM is a superior method to predict the monthly mean values of fair-weather atmospheric electric field. • A new method is proposed to forecast of fair-weather atmospheric electric field. • The model is a running-extrapolation model based on WT -LSSVM method. • Comparing with LSSVM and ANN, WT-LSSVM method is better for prediction.

Prediction of building energy consumption using an improved real coded genetic algorithm based least squares support vector machine approach

The least-squares support vector machine (LSSVM) strategy has played a crucial role in the forecasting of building energy consumption owing to its remarkable nonlinear mapping capabilities in prediction. In order to build an effective LSSVM method, its two free parameters, the regularization parameter and the kernel parameter, must be selected carefully. However, LSSVM using a conventional real-coded genetic algorithm (RCGA) or differential evolution algorithm (DEA) for determining the aforementioned two parameters consumes excessive amounts of computation time. In this study, a novel LSSVM for effective prediction of daily building energy consumption is designed by utilizing a hybrid of the direct search optimization (DSO) algorithm and RCGA, called the DSORCGA. The proposed DSORCGA differs from the conventional RCGA in terms of the reproduction operator and the crossover operator, and is used to optimize free parameters of LSSVM for faster computation speed and higher predictive accuracy. Finally, in a MATLAB2010a environment, actual building energy consumption data are adopted to run the proposed DSORCGA-LSSVM and conventional RCGA-LSSVM and DEA-LSSVM. Further, the simulation results in the target period are compared with those of actual recorded energy consumption data, and improvement in computation time is revealed via numerical simulation.

Estimation of Failure Rate in Water Distribution Network Using Fuzzy Clustering and LS-SVM Methods

In this study, a novel approach combining fuzzy clustering and Least Squares Support Vector machine (LS-SVM) methods is developed for estimation of failure rate in water distribution networks and for determination of the relationship between failure rate-effective factors. For this aim, failure data observed Malatya water distribution network during 2006–2012 was selected as study area. In first phase, estimation model was developed and tested for the complete data set in estimating the failure rate by LS-SVM method. Then, in order to develop a more sensitive estimation model and to improve the performance of LS-SVM, 9 sub-regions were defined with similar characteristics by using fuzzy clustering method. Then failure rate estimation was carried out for each of the sub-regions using by LS-SVM method. Feed Forward Neural Network (FFNN) and Generalized Regression Neural Network (GRNN) methods were also used for estimation of failure rate and the results were compared with those of LS-SVM. The criteria such as Correlation Coefficient (R), Efficieny (E) and Root Mean Square Error (RMSE) were used to evaluate the performance of models. The results showed that LS-SVM model gives better results in comparison with the FFNN and GRNN models. It was also determined that LSSVM model results for the sub-regions defined by clustering analysis are better and that the clustering analysis increases the estimation model performance in addition to the fact that the estimation results have become better. In conclusion, it can be possible to develop a more sensitive estimation models using fuzzy clustering and LSSVM methods.

The Contrastive Analysis of Predicting Riveting Pieces of Corrosion Fatigue Based on Multiple Kernel Function

In the process of long-term storage, the equipment would happen storage environment contaminated corrosion, mechanical structure stress corrosion damage. Currently,the corrosion fatigue damage prediction accuracy of method was low. Different kernel functions were adopted by this paper to compare based on least squares support vector machine (LSSVM). Besides, comparison was made among the BP neural network method, Standard Support Vector Machines (SVM), Grey System Prediction model Method and the radial basis function kernel least squares support vector machine (RBF_LSSVM) method by the simulation experiment. The optimal results finally were applied to practical engineering. The results showed that high accuracy and performance could be gained by employing the RBF_LSSVM method for predicting the trends of the mechanical structure rivet corrosion.

Application of particle swarm optimization-based least square support vector machine in quantitative analysis of extraction solution of yangxinshi tablet using near infrared spectroscopy

A particle swarm optimization (PSO)-based least square support vector machine (LS-SVM) method was investigated for quantitative analysis of extraction solution of Yangxinshi tablet using near infrared (NIR) spectroscopy. The usable spectral region (5400–6200 cm-1) was identified, then the first derivative spectra smoothed using a Savitzky–Golay filter were employed to establish calibration models. The PSO algorithm was applied to select the LS-SVM hyperparameters (including the regularization and kernel parameters). The calibration models of total flavonoids, puerarin, salvianolic acid B and icariin were established using the optimum hyperparameters of LS-SVM. The performance of LS-SVM models were compared with partial least squares (PLS) regression, feed-forward back-propagation network (BPANN) and support vector machine (SVM). Experimental results showed that both the calibration results and prediction accuracy of the PSO-based LS-SVM method were superior to PLS, BP-ANN and SVM. For PSO-based LS-SVM models, the determination coefficients (R2) for the calibration set were above 0.9881, and the RSEP values were controlled within 5.772%. For the validation set, the RMSEP values were close to RMSEC and less than 0.042, the RSEP values were under 8.778%, which were much lower than the PLS, BP-ANN and SVM models. The PSO-based LS-SVM algorithm employed in this study exhibited excellent calibration performance and prediction accuracy, which has definite practice significance and application value.

Firmness prediction in Prunus persica 'Calrico' peaches by visible/short-wave near infrared spectroscopy and acoustic measurements using optimised linear and non-linear chemometric models.

In this work, near infrared spectroscopy (NIR) and an acoustic measure (AWETA) (two non-destructive methods) were applied in Prunus persica fruit 'Calrico' (n = 260) to predict Magness-Taylor (MT) firmness. Separate and combined use of these measures was evaluated and compared using partial least squares (PLS) and least squares support vector machine (LS-SVM) regression methods. Also, a mutual-information-based variable selection method, seeking to find the most significant variables to produce optimal accuracy of the regression models, was applied to a joint set of variables (NIR wavelengths and AWETA measure). The newly proposed combined NIR-AWETA model gave good values of the determination coefficient (R(2)) for PLS and LS-SVM methods (0.77 and 0.78, respectively), improving the reliability of MT firmness prediction in comparison with separate NIR and AWETA predictions. The three variables selected by the variable selection method (AWETA measure plus NIR wavelengths 675 and 697 nm) achieved R(2) values 0.76 and 0.77, PLS and LS-SVM. These results indicated that the proposed mutual-information-based variable selection algorithm was a powerful tool for the selection of the most relevant variables.

Connectionist approach estimates gas–oil relative permeability in petroleum reservoirs: Application to reservoir simulation

Relative permeability of the petroleum reservoirs is a key parameter for various aspects of the petroleum engineering area like as reservoir simulation, history matching and etc. Due to this fact, various approaches such as experimental, theoretical and numerical approaches have been studied however; such experimental methods are time consuming, complicated and expensive. Based on the addressed disadvantages, robust, rapid, simple and accurate model is needed to represent gas/oil relative permeability through petroleum reservoirs. In this research communication we utilized the concept of various intelligent approaches such as least square support vector machine (LSSVM) which is high attended branches of artificial intelligent approaches. To develop and test the proposed LSSVM approach massive experimental relative permeability data from literature survey was faced to the addressed model. The suggested LSSVM method has low deviation from relevant measured values and statistical factors of the addressed model solutions were calculated. According to the determined statistical factors, the results of the proposed LSSVM approach prove and certify the high performance and low uncertainty of the addressed model in prediction gas/oil relative permeability in petroleum reservoirs. Finally, the suggested LSSVM model could help us to prepare more precise and accurate relative permeability curves without extensive experiment and furthermore, could lead to provide high performance reservoir simulation with low uncertainty.

LS-SVM based substation circuit breakers maintenance scheduling optimization

• Maintenance effect is evaluated based on historical data and statistical data. • LS-SVM method is used to predict defects before and after optimization. • Defects are correlated to power loss in capital means. • Capital means is used as constraints whether the optimization is cost-effective. • Evaluate function and constraint function are proposed. An optimized maintenance schedule for circuit breakers (CBs) can enhance substation reliability and lower maintenance cost. In this paper, a Least Squares Support Vector Machines (LS-SVM) based CB maintenance scheduling optimization approach that considers constraint of cost effect is proposed. Historical operation data are utilized to build a defects tree. A bi-level optimization algorithm is used to choose LS-SVM parameters. The LS-SVM algorithm is used to predict the distribution of defects before and after scheme optimization using aggregated defect data, outage duration, maintenance operation defect detection rate, etc. After the defect loss is quantified based on an expert scoring method and the Gross Domestic Product to power consumption ratio, a cost effect measurement is used to determine the best scheme. The effect of the proposed approach is verified using a numerical simulation of an electric power corporation.

Bearing faults diagnostics based on hybrid LS-SVM and EMD method

In this paper, a novel method that integrates the LS-SVM and Empirical Mode Decomposition (EMD) is proposed to improve the performance of conventional EMD. The analyzed signal is preprocessed with the weighted Least Squares Support Vector Machines (LS-SVM) to suppress the interference of high-frequency intermittent components and other non-Gaussian noises. The denoised signal is extended with LS-SVM rolling forecast modeling. Next, the linear function is used to construct upper and lower envelopes of the extrapolated data in order to determine the temporary mean envelope curve which is then smoothed with the adaptive mapped LS-SVM to obtain the local mean curve. Signal decomposition is self-adaptively performed to achieve IMFs through removal of the smoothed local mean curve. The representative IMF containing fault information is selected for demodulation analysis to identify the fault characteristics. The effectiveness of the proposed method is verified by means of simulations and applications to bearing fault diagnosis.

Neural network and support vector machine predictive control of tert-amyl methyl ether reactive distillation column

An algorithm of model predictive control based on artificial neural network and least-square support vector machine method is presented for a class of industrial process with strong nonlinearity such as tert-amyl methyl ether (TAME). Integral constant is added to improve the performance of the controller. In the present work, two different control methodologies neural network predictive control (NNPC) and support vector machine-based predictive control (SVMPC) are implemented and compared with a conventional proportional-integral-derivative (PID) control methodology to a TAME reactive distillation column. The simulation result shows that both NNPC and SVMPC gives better control performance than PID for set-point change as well as for load change of±10% in methanol feed flow rate and molar ratio of methanol to isoamylene in reactor effluent feed.

Calculation of Hildebrand solubility parameters of some polymers using QSPR methods based on LS-SVM technique and theoretical molecular descriptors

In this work, some chemometrics methods are applied for the modeling and prediction of the Hildebrand solubility parameter of some polymers. A genetic algorithm (GA) method is designed for the selection of variables to construct two models using the multiple linear regression (MLR) and least square-support vector machine (LS-SVM) methods in order to predict the Hildebrand solubility parameter. The MLR method is used to build a linear relationship between the molecular descriptors and the Hildebrand solubility parameter for these compounds. Then the LS-SVM method is utilized to construct the non-linear quantitative structure-activity relationship (QSAR) models. The results obtained using the LS-SVM method are then compared with those obtained for the MLR method; it was revealed that the LS-SVM model was much better than the MLR one. The root-mean-square errors of the training set and the test set for the LS-SVM model were 0.2912 and 0.2427, and the correlation coefficients were 0.9662 and 0.9518, respectively. This paper provides a new and effective method for predicting the Hildebrand solubility parameter for some polymers, and also reveals that the LS-SVM method can be used as a powerful chemometrics tool for the quantitative structure-property relationship (QSPR) studies.

Housing Value Forecasting Based on Machine Learning Methods

In the era of big data, many urgent issues to tackle in all walks of life all can be solved via big data technique. Compared with the Internet, economy, industry, and aerospace fields, the application of big data in the area of architecture is relatively few. In this paper, on the basis of the actual data, the values of Boston suburb houses are forecast by several machine learning methods. According to the predictions, the government and developers can make decisions about whether developing the real estate on corresponding regions or not. In this paper, support vector machine (SVM), least squares support vector machine (LSSVM), and partial least squares (PLS) methods are used to forecast the home values. And these algorithms are compared according to the predicted results. Experiment shows that although the data set exists serious nonlinearity, the experiment result also show SVM and LSSVM methods are superior to PLS on dealing with the problem of nonlinearity. The global optimal solution can be found and best forecasting effect can be achieved by SVM because of solving a quadratic programming problem. In this paper, the different computation efficiencies of the algorithms are compared according to the computing times of relevant algorithms.