Machine-learning based prediction of hydrogen/methane mixture solubility in brine

With regard to underground hydrogen storage projects, presuming that the hydrogen storage site has served as a repository for methane, the coexistence of a blend of methane and hydrogen is anticipated during the incipient stage of hydrogen storage. Therefore, the solubility of hydrogen/methane mixtures in brine becomes imperative. On the contrary, laboratory tasks of such measurements are hard because of its extreme corrosion ability and flammability, hence modeling methodologies are highly preferred. Therefore, in this study, we seek to create accurate data-driven intelligent models based upon laboratory data using hybrid models of adaptive neuro-fuzzy inference system (ANFIS) and least squares support vector machine (LSSVM) optimized with either particle swarm optimization (PSO), genetic algorithm (GA) and coupled simulated annealing (CSA) to predict hydrogen/methane mixture solubility in brine as a function of pressure, temperature, hydrogen mole fraction in hydrogen/methane mixture and brine salt concentration. The results indicate that almost all the gathered experimental data are technically suitable for the model development. The sensitivity study shows that pressure and hydrogen mole fraction in the mixture are strongly related with the solubility data with direct and indirect effects, respectively. The analyses of evaluation indexes and graphical methods indicates that the developed LSSVM-GA and LSSVM-CSA models are the most accurate as they exhibit the lowest AARE% and MSE values and the highest R-squared values. These findings show that machine learning methods could be a useful tool for predicting hydrogen solubility in brine encountered in underground hydrogen storage projects, aiding in the advancement of intelligent, affordable, and secure hydrogen storage technologies.

Displacement prediction of tunnel entrance slope based on LSSVM and bacterial foraging optimization algorithm

In order to realize the effective prediction of landslide risk in the tunnel entrance area, an multivariate time series model is established on the basis of the traditional model, taking temperature and rainfall factors as additional input indicators. Bacterial foraging optimization algorithm (BFOA) is used to search the global optimal solution of the key parameters γ and [Formula: see text] of least squares support vector machine (LSSVM) to improve its regression accuracy, and the evolved LSSVM is used to describe the aforementioned multivariate time series model. At the same time, a remote real-time internet of things (IoT) monitoring system for the tunnel entrance section, including monitoring indicators such as surface subsidence, temperature, and rainfall, has also been designed and implemented, providing a stable and accurate data source for the realization of this prediction model. Based on the engineering measurement data, the accuracy of the established model is checked and analyzed, the optimal value of historical data amount is determined to be 5 days, and the optimal value of prediction step is 1 day. The research results are applied in the construction of Wendong tunnel of Molin expressway, Yunnan, China. Practice shows that the prediction results of the multivariate time series model established in this study is accurate. This method can realize the prediction and early warning of slope risk, which provides a effective technical means for risk control of tunnel portal section.

Non-smooth optimization algorithm to solve the LINEX soft support vector machine

The Support Vector Machine (SVM) is a cornerstone of machine learning algorithms. This paper proposes a novel cost-sensitive model to address the challenges of class-imbalanced datasets inherent to SVMs. Integrating soft-margin SVM with the asymmetric LINEX loss function, this approach effectively tackles issues in scenarios with noisy data or overlapping classes. The LINEX loss function, which resembles the hinge and square loss functions, facilitates efficient model training with reduced sample penalties. Despite the resulting model's nonsmooth nature due to a constraint inequality, optimization is achieved using a Primal-Dual method, capitalizing on the convexity of the optimization function. This method enhances the model's noise robustness while preserving its original form. Extensive experiments validate the model's effectiveness, showcasing its superiority over traditional methods. Statistical tests further corroborate these findings.

Evaluation of Tidal Effect in Long-Strip DInSAR Measurements Based on GPS Network and Tidal Models

A long-strip differential interferometric synthetic aperture radar (DInSAR) measurement based on multi-frame image mosaicking is currently the realizable approach to measure large-scale ground deformation. As the spatial range of the mosaicked images increases, the nonlinear variation of ground ocean tidal loading (OTL) displacements is more significant, and using plane fitting to remove the large-scale errors will produce large tidal displacement residuals in a region with a complex coastline. To conveniently evaluate the ground tidal effect on mosaic DInSAR interferograms along the west coast of the U.S., a three-dimensional ground OTL displacements grid is generated by integrating tidal constituents’ estimation of the GPS reference station network and global/regional ocean tidal models. Meanwhile, a solid earth tide (SET) model based on IERS conventions is used to estimate the high-precision SET displacements. Experimental results show that the OTL and SET in a long-strip interferogram can reach 77.5 mm, which corresponds to a 19.3% displacement component. Furthermore, the traditional bilinear ramp fitting methods will cause 7.2~20.3 mm residual tidal displacement in the mosaicked interferograms, and the integrated tidal constituents displacements calculation method can accurately eliminate the tendency of tidal displacement in the long-strip interferograms.

Generative Adversarial Network to evaluate quantity of information in financial markets

Nowadays, the information obtainable from the markets are potentially limitless. Economic theory has always supported the possible advantage obtainable from having more information than competitors, however quantifying the advantage that these can give has always been a problem. In particular, in this paper we study the amount of information obtainable from the markets taking into account only the time series of the prices, through the use of a specific Generative Adversarial Network. We consider two types of financial instruments traded on the market, stocks and cryptocurrencies: the first are traded in a market subject to opening and closing hours, whereas cryptocurrencies are traded in a 24/7 market. Our goal is to use this GAN to be able to “convert” the amount of information that the different instruments can have in discriminative and predictive power, useful to improve forecast. Finally, we demonstrate that by using the initial dataset with the 5 most important feature useds by traders, the prices of cryptocurrencies present higher discriminatory and predictive power than stocks, while by adding a feature the situation can be completely reversed.

Displacement Estimation of Six-Pole Hybrid Magnetic Bearing Using Modified Particle Swarm Optimization Support Vector Machine

In order to solve the problems of the large volume and high cost of a six-pole hybrid magnetic bearing (SHMB) with displacement sensors, a displacement estimation method using a modified particle swarm optimization (MPSO) least-squares support vector machine (LS-SVM) is proposed. Firstly, the inertial weight of the MPSO is changed to achieve faster iterations, and the prediction model of an LS-SVM-based MPSO is built. Secondly, the prediction model is simulated and verified according to the parameters optimized by the MPSO, and the predicted values of MPSO and PSO are compared. Finally, static and dynamic suspension experiments and a disturbance experiment are carried out, which verify the robustness and stability of the displacement estimation method.

Predicting methane solubility in water and seawater by machine learning algorithms: Application to methane transport modeling

The upward migration of methane from natural gas wells associated with fracking operations may lead to contamination of groundwater resources and surface leakage. Numerical simulations of methane transport in the subsurface environment require knowledge of methane solubility in the aqueous phase. This study employs machine learning (ML) algorithms to predict methane solubility in aquatic systems for temperatures ranging from 273.15 to 518.3 K and pressures ranging from 1 to 1570 bar. Four regression algorithms including regression tree (RT), boosted regression tree (BRT), least square support vector machine (LSSVM) and Gaussian process regression (GPR) were utilized for predicting methane solubility in pure water and mixed aquatic systems containing Na⁺, K⁺, Ca²⁺, Mg²⁺, Cl^- and SO₄^-2. The experimental data collected from the literature were used to implement the models. We used Grid search (GS), Random search (RS) and Bayesian optimization (BO) for tuning hyper-parameters of the ML models. Moreover, the predicted values of methane solubility were compared against Spivey et al. (2004) and Duan and Mao (2006) equations of state. The results show that the BRT-BO model is the most rigorous model for the prediction task. The coefficient of determination (R²) between experimental and predicted values is 0.99 and the mean squared error (MSE) is 1.19 × 10^-7. The performance of the BRT-BO model is satisfactory, showing an acceptable agreement with experimental data. The comparison results demonstrated the superior performance of the BRT-BO model for predicting methane solubility in aquatic systems over a span of temperature, pressure and ionic strength that occurs in deep marine environments.

Non-Contact Speech Recovery Technology Using a 24 GHz Portable Auditory Radar and Webcam

Language has been one of the most effective ways of human communication and information exchange. To solve the problem of non-contact robust speech recognition, recovery, and surveillance, this paper presents a speech recovery technology based on a 24 GHz portable auditory radar and webcam. The continuous-wave auditory radar is utilized to extract the vocal vibration signal, and the webcam is used to obtain the fitted formant frequency. The traditional formant speech synthesizer is selected to synthesize and recover speech, using the vocal vibration signal as the sound source excitation and the fitted formant frequency as the vocal tract resonance characteristics. Experiments on reading single English characters and words are carried out. Using microphone records as a reference, the effectiveness of the proposed speech recovery technology is verified. Mean opinion scores show a relatively high consistency between the synthesized speech and original acoustic speech.

A novel optimal-hybrid model for daily air quality index prediction considering air pollutant factors

Accurate and reliable air quality index (AQI) forecasting is extremely crucial for ecological environment and public health. A novel optimal-hybrid model, which fuses the advantage of secondary decomposition (SD), AI method and optimization algorithm, is developed for AQI forecasting in this paper. In the proposed SD method, wavelet decomposition (WD) is chosen as the primary decomposition technique to generate a high frequency detail sequence WD(D) and a low frequency approximation sequence WD(A). Variational mode decomposition (VMD) improved by sample entropy (SE) is adopted to smooth the WD(D), then long short-term memory (LSTM) neural network with good ability of learning and time series memory is applied to make it easy to be predicted. Least squares support vector machine (LSSVM) with the parameters optimized by the Bat algorithm (BA) considers air pollutant factors including PM_2.5, PM₁₀, SO₂, CO, NO₂ and O₃, which is suitable for forecasting WD(A) that retains original information of AQI series. The ultimate forecast result of AQI can be obtained by accumulating the prediction values of each subseries. Notably, the proposed idea not only gives full play to the advantages of conventional SD, but solve the problem that the traditional time series prediction model based on decomposition technology can not consider the influential factors. Additionally, two daily AQI series from December 1, 2016 to December 31, 2018 respectively collected from Beijing and Guilin located in China are utilized as the case studies to verify the proposed model. Comprehensive comparisons with a set of evaluation indices indicate that the proposed optimal-hybrid model comprehensively captures the characteristics of the original AQI series and has high correct rate of forecasting AQI classes.

Short-Term Wind Speed Forecasting Based on Hybrid Variational Mode Decomposition and Least Squares Support Vector Machine Optimized by Bat Algorithm Model

With the integration of wind energy into electricity grids, wind speed forecasting plays an important role in energy generation planning, power grid integration and turbine maintenance scheduling. This study proposes a hybrid wind speed forecasting model to enhance prediction performance. Variational mode decomposition (VMD) was applied to decompose the original wind speed series into different sub-series with various frequencies. A least squares support vector machine (LSSVM) model with the pertinent parameters being optimized by a bat algorithm (BA) was established to forecast those sub-series extracted from VMD. The ultimate forecast of wind speed can be obtained by accumulating the prediction values of each sub-series. The results show that: (a) VMD-BA-LSSVM displays better capacity for the prediction of ultra short-term (15 min) and short-term (1 h) wind speed forecasting; (b) the proposed forecasting model was compared with wavelet decomposition (WD) and ensemble empirical mode decomposition (EEMD), and the results indicate that VMD has stronger decomposition ability than WD and EEMD, thus, significant improvements in forecasting accuracy were obtained with the proposed forecasting models compared with other forecasting methods.

In silico prediction of pesticide aquatic toxicity with chemical category approaches

Aquatic toxicity is an important issue in pesticide development. In this study, using nine molecular fingerprints to describe pesticides, binary and ternary classification models were constructed to predict aquatic toxicity of pesticides via six machine learning methods: Naïve Bayes (NB), Artificial Neural Network (ANN), k-Nearest Neighbor (kNN), Classification Tree (CT), Random Forest (RF) and Support Vector Machine (SVM). For the binary models, local models were obtained with 829 pesticides on rainbow trout (RT) and 151 pesticides on lepomis (LP), and global models were constructed on the basis of 1258 diverse pesticides on RT and LP and 278 on other fish species. After analyzing the local binary models, we found that fish species caused influence in terms of accuracy. Considering the data size and predictive range, the 1258 pesticides were also used to build global ternary models. The best local binary models were Maccs_ANN for RT and Maccs_SVM for LP, which exhibited accuracies of 0.90 and 0.90, respectively. For global binary models, the best model was Graph_SVM with an accuracy of 0.89. Accuracy of the best global ternary model Graph_SVM was 0.81, which was a little lower than that of the best global binary model. In addition, several substructural alerts were identified including nitrobenzene, chloroalkene and nitrile, which could significantly correlate with pesticide aquatic toxicity. This study provides a useful tool for an early evaluation of pesticide aquatic toxicity in environmental risk assessment.

Robust Recurrent Kernel Online Learning

We propose a robust recurrent kernel online learning (RRKOL) algorithm based on the celebrated real-time recurrent learning approach that exploits the kernel trick in a recurrent online training manner. The novel RRKOL algorithm guarantees weight convergence with regularized risk management through the use of adaptive recurrent hyperparameters for superior generalization performance. Based on a new concept of the structure update error with a variable parameter length, we are the first one to propose the detailed structure update error, such that the weight convergence and robust stability proof can be integrated with a kernel sparsification scheme based on a solid theoretical ground. The RRKOL algorithm automatically weighs the regularized term in the recurrent loss function, such that we not only minimize the estimation error but also improve the generalization performance through sparsification with simulation support.

Design and Implementation of Foot-Mounted Inertial Sensor Based Wearable Electronic Device for Game Play Application

Wearable electronic devices have experienced increasing development with the advances in the semiconductor industry and have received more attention during the last decades. This paper presents the development and implementation of a novel inertial sensor-based foot-mounted wearable electronic device for a brand new application: game playing. The main objective of the introduced system is to monitor and identify the human foot stepping direction in real time, and coordinate these motions to control the player operation in games. This proposed system extends the utilized field of currently available wearable devices and introduces a convenient and portable medium to perform exercise in a more compelling way in the near future. This paper provides an overview of the previously-developed system platforms, introduces the main idea behind this novel application, and describes the implemented human foot moving direction identification algorithm. Practical experiment results demonstrate that the proposed system is capable of recognizing five foot motions, jump, step left, step right, step forward, and step backward, and has achieved an over 97% accuracy performance for different users. The functionality of the system for real-time application has also been verified through the practical experiments.

Multiscale asymmetric orthogonal wavelet kernel for linear programming support vector learning and nonlinear dynamic systems identification

Support vector regression for approximating nonlinear dynamic systems is more delicate than the approximation of indicator functions in support vector classification, particularly for systems that involve multitudes of time scales in their sampled data. The kernel used for support vector learning determines the class of functions from which a support vector machine can draw its solution, and the choice of kernel significantly influences the performance of a support vector machine. In this paper, to bridge the gap between wavelet multiresolution analysis and kernel learning, the closed-form orthogonal wavelet is exploited to construct new multiscale asymmetric orthogonal wavelet kernels for linear programming support vector learning. The closed-form multiscale orthogonal wavelet kernel provides a systematic framework to implement multiscale kernel learning via dyadic dilations and also enables us to represent complex nonlinear dynamics effectively. To demonstrate the superiority of the proposed multiscale wavelet kernel in identifying complex nonlinear dynamic systems, two case studies are presented that aim at building parallel models on benchmark datasets. The development of parallel models that address the long-term/mid-term prediction issue is more intricate and challenging than the identification of series-parallel models where only one-step ahead prediction is required. Simulation results illustrate the effectiveness of the proposed multiscale kernel learning.

A linear recurrent kernel online learning algorithm with sparse updates

In this paper, we propose a recurrent kernel algorithm with selectively sparse updates for online learning. The algorithm introduces a linear recurrent term in the estimation of the current output. This makes the past information reusable for updating of the algorithm in the form of a recurrent gradient term. To ensure that the reuse of this recurrent gradient indeed accelerates the convergence speed, a novel hybrid recurrent training is proposed to switch on or off learning the recurrent information according to the magnitude of the current training error. Furthermore, the algorithm includes a data-dependent adaptive learning rate which can provide guaranteed system weight convergence at each training iteration. The learning rate is set as zero when the training violates the derived convergence conditions, which makes the algorithm updating process sparse. Theoretical analyses of the weight convergence are presented and experimental results show the good performance of the proposed algorithm in terms of convergence speed and estimation accuracy.

Non-linear multivariate modeling of cerebral hemodynamics with autoregressive Support Vector Machines

Cerebral blood flow (CBF) is normally controlled by myogenic and metabolic mechanisms that can be impaired in different cerebrovascular conditions. Modeling the influences of arterial blood pressure (ABP) and arterial CO(2) (PaCO(2)) on CBF is an essential step to shed light on regulatory mechanisms and extract clinically relevant parameters. Support Vector Machines (SVM) were used to model the influences of ABP and PaCO(2) on CBFV in two different conditions: baseline and during breathing of 5% CO(2) in air, in a group of 16 healthy subjects. Different model structures were considered, including innovative non-linear multivariate autoregressive (AR) models. Results showed that AR models are significantly superior to finite impulse response models and that non-linear models provide better performance for both structures. Correlation coefficients for multivariate AR non-linear models were 0.71 ± 0.11 at baseline, reaching 0.91 ± 0.06 during 5% CO(2). These results warrant further work to investigate the performance of autoregressive SVM in patients with cerebrovascular conditions.

Financial time series prediction using least squares support vector machines within the evidence framework

The Bayesian evidence framework is applied in this paper to least squares support vector machine (LS-SVM) regression in order to infer nonlinear models for predicting a financial time series and the related volatility. On the first level of inference, a statistical framework is related to the LS-SVM formulation which allows one to include the time-varying volatility of the market by an appropriate choice of several hyper-parameters. The hyper-parameters of the model are inferred on the second level of inference. The inferred hyper-parameters, related to the volatility, are used to construct a volatility model within the evidence framework. Model comparison is performed on the third level of inference in order to automatically tune the parameters of the kernel function and to select the relevant inputs. The LS-SVM formulation allows one to derive analytic expressions in the feature space and practical expressions are obtained in the dual space replacing the inner product by the related kernel function using Mercer's theorem. The one step ahead prediction performances obtained on the prediction of the weekly 90-day T-bill rate and the daily DAX30 closing prices show that significant out of sample sign predictions can be made with respect to the Pesaran-Timmerman test statistic.

Support vector regression and least squares support vector regression for hormetic dose-response curves fitting

Accurate description of hormetic dose-response curves (DRC) is a key step for the determination of the efficacy and hazards of the pollutants with the hormetic phenomenon. This study tries to use support vector regression (SVR) and least squares support vector regression (LS-SVR) to address the problem of curve fitting existing in hormesis. The SVR and LS-SVR, which are entirely different from the non-linear fitting methods used to describe hormetic effects based on large sample, are at present only optimum methods based on small sample often encountered in the experimental toxicology. The tuning parameters (C and p1 for SVR, gam and sig2 for LS-SVR) determining SVR and LS-SVR models were obtained by both the internal and external validation of the models. The internal validation was performed by using leave-one-out (LOO) cross-validation and the external validation was performed by splitting the whole data set (12 data points) into the same size (six data points) of training set and test set. The results show that SVR and LS-SVR can accurately describe not only for the hermetic J-shaped DRC of seven water-soluble organic solvents consisting of acetonitrile, methanol, ethanol, acetone, ether, tetrahydrofuran, and isopropanol, but also for the classical sigmoid DRC of six pesticides including simetryn, prometon, bromacil, velpar, diquat-dibromide monohydrate, and dichlorvos.

Training Recurrent Networks by Evolino

In recent years, gradient-based LSTM recurrent neural networks (RNNs) solved many previously RNN-unlearnable tasks. Sometimes, however, gradient information is of little use for training RNNs, due to numerous local minima. For such cases, we present a novel method: EVOlution of systems with LINear Outputs (Evolino). Evolino evolves weights to the nonlinear, hidden nodes of RNNs while computing optimal linear mappings from hidden state to output, using methods such as pseudo-inverse-based linear regression. If we instead use quadratic programming to maximize the margin, we obtain the first evolutionary recurrent support vector machines. We show that Evolino-based LSTM can solve tasks that Echo State nets (Jaeger, 2004a) cannot and achieves higher accuracy in certain continuous function generation tasks than conventional gradient descent RNNs, including gradient-based LSTM.

Support Vector Echo-State Machine for Chaotic Time-Series Prediction

A novel chaotic time-series prediction method based on support vector machines (SVMs) and echo-state mechanisms is proposed. The basic idea is replacing "kernel trick" with "reservoir trick" in dealing with nonlinearity, that is, performing linear support vector regression (SVR) in the high-dimension "reservoir" state space, and the solution benefits from the advantages from structural risk minimization principle, and we call it support vector echo-state machines (SVESMs). SVESMs belong to a special kind of recurrent neural networks (RNNs) with convex objective function, and their solution is global, optimal, and unique. SVESMs are especially efficient in dealing with real life nonlinear time series, and its generalization ability and robustness are obtained by regularization operator and robust loss function. The method is tested on the benchmark prediction problem of Mackey-Glass time series and applied to some real life time series such as monthly sunspots time series and runoff time series of the Yellow River, and the prediction results are promising.