Data-driven Modeling Techniques Research Articles

Data-Driven Degradation Modeling and SOH Prediction of Li-Ion Batteries

Electrified vehicles (EV) and marine vessels represent promising clean transportation solutions to reduce or eliminate petroleum fuel use, greenhouse gas emissions and air pollutants. The presently commonly used electric energy storage system (ESS) is based on lithium-ion batteries. These batteries are the electrified or hybridized powertrain’s most expensive component and show noticeable performance degradations under different use patterns. Therefore, battery life prediction models play a key role in realizing globally optimized EV design and energy control strategies. This research studies the data-driven modelling and prediction methods for Li-ion batteries’ performance degradation behaviour and the state of health (SOH) estimation. The research takes advantage of the increasingly available battery test and data to reduce prediction errors of the widely used semi-empirical modelling methods. Several data-driven modelling techniques have been applied, improved, and compared to identify their advantages and limitations. The data-driven approach and Kalman Filter (KF) algorithm are used to estimate and predict the degradation of the battery during operation. The combined algorithm of Gaussian Process Regression (GPR) and Extended Kalman Filter (EKF) showed higher accuracy than other algorithms.

Data-Driven Modeling of Vibrations in Turbofan Engines Under Different Operating Conditions

Vibration monitoring in aircraft engines is traditionally performed using order tracking methodology. Currently, there are no representative physics-based models for tracked order vibration in aircraft engines. This gap in research is primarily attributed to the limited understanding of mutual interactions and the nonlinear nature of engine vibrations. The objective of the current study is twofold: 1) to present a preliminary investigation of tracked order vibrations in aircraft engines and to statistically analyze them in the context of their operating environment and 2) to leverage emerging data-driven modeling techniques to develop accurate models for the tracked order vibration in a turbofan engine valid over a wide range of operating conditions. Toward the same, we employ multiscale detrended partial cross-correlation analysis to evaluate the interdependencies and cross-correlation structure between various thermomechanical variables and tracked order vibration across multiple time scales. We leverage the understanding of complex correlation structures and subsequently use them to model tracked order vibrations. We develop data-driven models using state-of-the-art time-delay embedded dynamic mode decomposition and compare its performance with Lasso regression.

Shedding light on the state-of-the-art in data-driven modeling of wind farms

Examining data-driven and machine learning modelling techniques to improve wind-farm flow predictions and power harvesting

Eulerian-Lagrangian Modelling of Turbulent Two-Phase Particle-Liquid Flow in a Stirred Vessel: CFD and Experiments Compared

A 3D Eulerian-Lagrangian CFD model is set up to simulate the turbulent flow and mixing of coarse dense particles in a standard batch vessel, mechanically agitated by a down-pumping pitched-blade turbine. Flow is investigated in the just-suspended regime and regimes much above. All simulations are fully and successfully validated by experimental Lagrangian measurements provided by positron emission particle tracking. Predicted distributions of the local 3D phase-velocity components, local particle slip velocities and local particle concentration show very good agreement with experiment. Lagrangian particle trajectories are exploited to infer detailed accurate information on particle circulation time and local residence time. The predicted two-phase flow number is in excellent agreement with experimental values under all flow regimes. Being accurate and robust, the numerical model can be extended to study even more complex particle-liquid suspensions of industrial relevance, and to provide much needed Lagrangian data to other data-driven modelling techniques.

Compressed Machine Learning-Based Inverse Model for Design Optimization of Microwave Components

This article presents a new noniterative inverse modeling technique based on machine learning regression and its applications to microwave design optimization. The proposed inverse model accepts the high-dimensional S-parameters computed at many frequency points as the input and estimates the optimal geometrical/physical parameters of the microwave component as its output. The least-squares support vector machine regression is combined with the principal component analysis to simultaneously overcome both the high-dimensional input space and ill-posed challenges of the inverse modeling. We also propose a new empirical method to find the optimum number of principal components (i.e., the compression level) for each example in an automated way. This makes our proposed model general and easy to use compared with the existing data-driven inverse modeling techniques. The inverse model is trained by a set of scattering parameters computed via a 2-D/3-D solver for few configurations of the geometrical parameters. The feasibility and the accuracy of the proposed optimization scheme are investigated by comparing its predictions with the corresponding optimal configuration estimated via a commercial solver.

Data-driven building energy modelling – An analysis of the potential for generalisation through interpretable machine learning

Data-driven building energy modelling techniques have proven to be effective in multiple applications. However, the debate around the possibility of generalisation is open. Generalisation involves the ability of a machine-learning model to adapt to previously unseen data and perform in a satisfactory way. Besides that, while machine-learning techniques are extremely powerful, interpretability, i.e. the ability for humans to predict how the model output will change in response to a change in input data or algorithmic parameters, is essential to attain a “human-in-the-loop” approach and creating feedback loops aimed at continuous improvement of efficiency measures in buildings.A flexible regression-based approach is developed and tested on a Passive House building in this study. The formulation employs dummy (binary) variables as a piecewise linearization method, and the rules for creating them are explicitly stated to ensure interpretability. Furthermore, the possibility of automating the model selection process using statistical indicators is described, including specific indicators used in Measurement and Verification (M&V) for the acceptance of calibrated energy models.The valuable insights that can be found using data-driven methods are reported and discussed, emphasizing limitations and constraints, as well as the potential for future research focused on systems of (interpretable data-driven) models that can exploit the techniques' spatial and temporal scalability. Finally, the physical interpretation of model coefficients and the analytical formulations for energy model decomposition can be used to supplement the scalability of data-driven techniques and create more sophisticated systems of interconnected models.

Adaptive Positioning Control of Servomotor Based on Model Identification

The new energy vehicle industry needs a large number of vehicle power batteries, and its production depends on high-speed and high-precision battery pole-piece cutting equipment, which is mainly composed of the electrode material roll uncoiling unit, buffering unit, main drive unit, and cutting unit. Due to the nonlinearity and high-speed intermittent motion mode of a servomotor, the existing control technology is difficult to apply to the cutting system. This article presents a multimotor linear velocity planning approach used to ensure synchronous and cooperative operation of the motors in the cutting system. To achieve fast and accurate positioning control of the main drive motor running in fast acceleration and deceleration, periodic and intermittent mode, an adaptive position controller based on data-driven modeling technique is proposed, which makes the parameters of the controller change with the operation state of the main drive motor so that the controller can obtain good positioning accuracy in a wide working range of the main drive motor. Two linear models are established to represent the relationship between the positioning error and the positioning controller parameters and the main drive motor running conditions. The parameters of the two models are estimated offline and the parameters of the adaptive position controller can be obtained online quickly. Thus, the proposed control method for the main drive motor can be easily realized by general PLC in practical application. Simulation study results show that using the proposed method, the positioning error of the main drive motor is less than 0.1 mm when cutting 89-mm-long battery pole-piece and cutting 1 to 3 pieces/s. This result is much better than that of the position controller with fixed parameters.

Machine-learning approach for predicting the occurrence and timing of mid-winter ice breakups on canadian rivers

The increasingly common occurrence of Mid-Winter Breakups (MWBs) in Canadian rivers, consisting of the early breakup of ice cover outside of the typical spring season, is a cause for concern. This study applied various data-driven modelling techniques to predict MWBs occurrence and timing with sufficient lead times on a national scale using a new Canadian River Ice Database (CRID) coupled with National Resources Canada gridded climate data. A two-level machine learning model structure was developed, with the first level model predicting MWB occurrence within a given period and the second level model predicting the timing of an MWB occurrence within that period. Machine learning techniques that can handle class imbalance were employed to address many of the issues inherent in rare event forecasting, including the implementation of data preprocessing, the selection of algorithms and performance metrics suited to rare events. Multiple configurations of both model levels, including variations on time series arrangement and input variables, were tested to select the optimal model structure. The best performing configuration, focussing on a biweekly time period, attained overall accuracies of 80.1% and 77.6% for the first and second level models respectively on the 452 MWBs in the CRID. In addition, probabilistic prediction results were analyzed to evaluate model uncertainty and robustness. This new modelling framework provides the first tool capable of predicting MWBs on a national scale, with easily extendable methodology to locations that have not yet experienced MWBs and can form the basis of vital decision-making support to affected communities.

Accurate discharge coefficient prediction of streamlined weirs by coupling linear regression and deep convolutional gated recurrent unit

ABSTRACT Streamlined weirs, which are a nature-inspired type of weir, have gained tremendous attention among hydraulic engineers, mainly owing to their established performance with high discharge coefficients. Computational fluid dynamics (CFD) is considered as a robust tool to predict the discharge coefficient. To bypass the computational cost of CFD-based assessment, the present study proposes data-driven modeling techniques, as an alternative to CFD simulation, to predict the discharge coefficient based on an experimental dataset. To this end, after splitting the dataset using a k-fold cross-validation technique, the performance assessment of classical and hybrid machine learning–deep learning (ML-DL) algorithms is undertaken. Among ML techniques, linear regression (LR), random forest (RF), support vector machine (SVM), k-nearest neighbor (KNN) and decision tree (DT) algorithms are studied. In the context of DL, long short-term memory (LSTM), convolutional neural network (CNN) and gated recurrent unit (GRU), and their hybrid forms, such as LSTM-GRU, CNN-LSTM and CNN-GRU techniques, are compared using different error metrics. It is found that the proposed three-layer hierarchical DL algorithm, consisting of a convolutional layer coupled with two subsequent GRU levels, which is also hybridized with the LR method (i.e. LR-CGRU), leads to lower error metrics. This paper paves the way for data-driven modeling of streamlined weirs.

Assessing and predicting the severity of mid-winter breakups based on Canada-wide river ice data

Mid-winter breakups (MWBs), consisting of the early breakup of the winter river ice cover before the typical spring breakup season, are becoming increasingly common events in cold region rivers. These events can lead to potentially severe flooding, while also altering the expected spring flow regime, yet data on these events is limited. In this study, a newly released Canadian River Ice Database (CRID), containing river ice data from 196 rivers across Canada obtained from time series analysis, was used to analyse these MWBs on a previously impossible national scale. The CRID data was combined with the Natural Resources Canada (NRCan) gridded daily climate dataset to identify a list of potential hydrologic and climatic drivers for MWB events. Techniques such as correlation analysis, Least Absolute Selection Shrinkage Operator (LASSO) regression, and input omission were combined to select 20 key drivers of the severity of MWB events. A random forest model that was trained with these drivers using data-driven modelling techniques successfully classified the MWBs as either low, medium, or high severity, achieving an overall accuracy of 80%. A new threshold for the prediction of MWB initiation based on climatic conditions was subsequently proposed through the use of optimization via an exhaustive grid search and its accuracy in identifying MWBs exceeded those proposed by previous studies. The new threshold used in conjunction with the random forest model provide valuable tools for both the prediction of MWBs and the assessment of their potential severity.

Adaptive Data Dimensionality Reduction for Chemical Process Modeling Based on the Information Criterion Related to Data Association and Redundancy

Chemical process modeling is the basis for research and applications in related fields. With the development of industrial informatization, data-driven process modeling techniques are increasingly applied in chemical processes, helping to obtain more accurate results with less model development costs. However, due to the high-dimensional nonlinear characteristics of most chemical processes, problems such as the “curse of dimensionality” and information redundancy will render the models more prone to overfitting with reduced accuracies and weaker generalization abilities. Many data dimensionality reduction methods are adopted to mitigate the above problems, but most are limited by inaccurate association measurements and weak redundancy exclusion. In this paper, the extensive existence of data associations and information redundancies is first revealed through the analysis from an information-theoretic perspective. Then, a feature selection method based on conditional refined maximal information coefficient maximization (CRMICM) is proposed to improve the consistency of association measurement and the accuracy of redundancy estimation with limited samples. The final prediction modeling test for an actual fluidized catalytic cracking (FCC) process proves the extensive association between the variables and targets. Only a few variables are essential for the modeling, while the rest are redundant. Compared with other methods, CRMICM achieves the best dimensionality reduction effects on the FCC process data regarding the number of features and model prediction accuracy, showing its good applicability for chemical processes.

Data-driven Crowd Modeling Techniques: A Survey

Data-driven crowd modeling has now become a popular and effective approach for generating realistic crowd simulation and has been applied to a range of applications, such as anomaly detection and game design. In the past decades, a number of data-driven crowd modeling techniques have been proposed, providing many options for people to generate virtual crowd simulation. This article provides a comprehensive survey of these state-of-the-art data-driven modeling techniques. We first describe the commonly used datasets for crowd modeling. Then, we categorize and discuss the state-of-the-art data-driven crowd modeling methods. After that, data-driven crowd model validation techniques are discussed. Finally, six promising future research topics of data-driven crowd modeling are discussed.

Development of stage-discharge rating curve using ANN

Accurate forecasting of river discharge is essential for the efficient operation of water resources systems. Therefore, researchers are consistently developing and improving various techniques to predict river discharge with relative ease and high accuracy, although traditional methods are available. This paper presents mainly three data-driven modelling techniques, namely the stage rating curve (SRC), generalised reduced gradient (GRG) solver, and an artificial neural network (ANN) to accurately model the stage-discharge relationship for local rivers in Trinidad and Tobago using only low flow data. The model that produced the overall superior results was the ANN.

Nonlinear Modeling of a Plate Heat Exchanger of a District Heating System

The integration of renewable and residual energy sources requires a detailed level of knowledge of district heating components such as thermal pipes, pipes, pumps, and heat exchanger of a substation. The thermal energy stored in these components can be utilized to match the heat demand and heat generation in time for next generation district heating networks. The goal of this paper is to provide accurate substation models that can be utilized for controlled energy exchange between primary and secondary networks. The proposed method introduces an efficient, user-friendly, combined parametric and nonparametric nonlinear data-driven modeling technique illustrated on a plate heat exchanger which is installed in the district heating network of the Vrije Universiteit Brussel. A fair comparison between a physics-based white-box and a data-driven black-box modeling techniques is provided. The white-box model is developed on the first principle modeling approach of the heat balance equations, which allows calculating the temperature evolution and heat transfer in the plate heat exchanger. As for the data-driven modeling approach, there are four interrelated consecutive steps. First, the measurement is processed to obtain a frequency response function model. Second, a linear parametric (state-space) model is estimated. Third, a polynomial nonlinear model is built. Last, a decoupled model structure is obtained based on the previously obtained PNLSS model. The proposed black-box method outperforms the white-box techniques in terms of prediction capabilities and computational needs.

Wildland Fire Detection and Monitoring Using a Drone-Collected RGB/IR Image Dataset

Current forest monitoring technologies including satellite remote sensing, manned/piloted aircraft, and observation towers leave uncertainties about a wildfire’s extent, behavior, and conditions in the fire’s near environment, particularly during its early growth. Rapid mapping and real-time fire monitoring can inform in-time intervention or management solutions to maximize beneficial fire outcomes. Drone systems’ unique features of 3D mobility, low flight altitude, and fast and easy deployment make them a valuable tool for early detection and assessment of wildland fires, especially in remote forests that are not easily accessible by ground vehicles. In addition, the lack of abundant, well-annotated aerial datasets – in part due to unmanned aerial vehicles’ (UAVs’) flight restrictions during prescribed burns and wildfires – has limited research advances in reliable data-driven fire detection and modeling techniques. While existing wildland fire datasets often include either color or thermal fire images, here we present (1) a multi-modal UAV-collected dataset of dual-feed side-by-side videos including both RGB and thermal images of a prescribed fire in an open canopy pine forest in Northern Arizona and (2) a deep learning-based methodology for detecting fire and smoke pixels at accuracy much higher than the usual single-channel video feeds. The collected images are labeled to "fire" or "no-fire" frames by two human experts using side-by-side RGB and thermal images to determine the label. To provide context to the main dataset’s aerial imagery, the included supplementary dataset provides a georeferenced pre-burn point cloud, an RGB orthomosaic, weather information, a burn plan, and other burn information. By using and expanding on this guide dataset, research can develop new data-driven fire detection, fire segmentation, and fire modeling techniques.

A Hybrid Surrogate Modeling Approach for Vehicle Crash Simulations

Integrated vehicle safety – a combination of active and passive vehicle safety systems has the potential to increase occupant safety dramatically. To adapt the passive safety components such as belt and airbag to a given collision scenario, the system requires a priori knowledge of the collision consequences, i.e. via a camera, radar or lidar system. Therefore, real-time capable surrogate models are needed, that predict the vehicle decelerations during a crash with high accuracy. To combine the strength of both data-driven and physics-based surrogate modeling techniques, a hybrid surrogate model consisting of a machine learning-enhanced spring-damper-mass model, is developed in this work. Thereby, the spring-damper-mass model describes the deformation behavior of the vehicle structure. A neural network is then trained to predict the stiffness and damping parameters of the simplified physical model based on the impact angle, the velocity and the offset of the vehicles.

A Survey on Design, Actuation, Modeling, and Control of Continuum Robot.

In this paper, we describe the advances in the design, actuation, modeling, and control field of continuum robots. After decades of pioneering research, many innovative structural design and actuation methods have arisen. Untethered magnetic robots are a good example; its external actuation characteristic allows for miniaturization, and they have gotten a lot of interest from academics. Furthermore, continuum robots with proprioceptive abilities are also studied. In modeling, modeling approaches based on continuum mechanics and geometric shaping hypothesis have made significant progress after years of research. Geometric exact continuum mechanics yields apparent computing efficiency via discrete modeling when combined with numerical analytic methods such that many effective model-based control methods have been realized. In the control, closed-loop and hybrid control methods offer great accuracy and resilience of motion control when combined with sensor feedback information. On the other hand, the advancement of machine learning has made modeling and control of continuum robots easier. The data-driven modeling technique simplifies modeling and improves anti-interference and generalization abilities. This paper discusses the current development and challenges of continuum robots in the above fields and provides prospects for the future.

Alzheimer’s disease diagnosis using genetic programming based on higher order spectra features

In Alzheimer’s diagnosis field, Computer-Aided Diagnosis (CADx) technology can improve the work performance of medical researchers and practitioners since it gives early chances to patient’s eligibility for clinical trials. The aim of this study is to develop a novel CADx system for the diagnosis of Alzheimer’s disease (AD) by utilizing genetic programming (GP) as data-driven evolutionary computation based modeling. The proposed method invokes a majority voting based scheme to select a set of most discriminant features which leads to the highest diagnosis accuracy of the final classification. The effectiveness of GP in categorizing patients with Alzheimer’s versus healthy group was revealed by developing models according to their performance in terms of higher-order spectra (HOS) features. The results show that the GP method achieved better performance compared to other the-state-of-the-art approaches. It is also found that the highest accuracy index was yielded by using the proposed data-driven modeling technique. The results of this study emphasize the practicality of GP-based method for developing CADx systems, on the basis of spontaneous speech analysis; can efficiently assist in the diagnosis of Alzheimer’s disease.

OASIS-P: Operable Adaptive Sparse Identification of Systems for fault Prognosis of chemical processes

With the increasing process complexities, data-driven fault prognosis has emerged as a promising fault management tool that predicts and manages abnormal events well in advance. In this paper, we develop a fault prognosis framework named ‘OASIS-P’ by integrating operable adaptive sparse identification of systems (OASIS), which is a data-driven adaptive modeling technique, with a risk-based process monitoring approach and contribution plots. Firstly, OASIS is employed with the risk assessment procedure for the prediction of impending faults. As the OASIS model is adaptive, it copes with the initial fault symptoms and forecasts the future behavior of the process under faulty conditions reasonably well, thereby providing an early fault prediction. Next, the fault isolation step is immediately initiated using contribution plots to identify the faulty variables. Unlike in fault diagnosis, the problem of ambiguity in interpreting contribution results due to fault propagation is not an issue in fault prognosis, if the fault isolation step is implemented at an early stage of the fault before it affects the other variables. Hence, the contribution plots together with OASIS can proactively monitor the process in real-time. As a case study, we demonstrate OASIS-P for fault prognosis of a reactor–separator system.

Ensemble prediction of mean bubble size in a continuous casting mold using data driven modeling techniques

Gas injection into a slab continuous casting mold is a common practice as it enables long casting sequences by reducing SEN clogging. However, too much gas injection can generate lots of bubbles, and bubbles formed inside the mold are also responsible for the occurrence of various quality defects such as deterioration of steel cleanliness, reoxidation of liquid steel, and occurrence of sliver and blister defects, etc. Mean bubble size is a key parameter and controlling it can resolve/reduce these quality issues. Operating parameters such as gas flow rate and liquid flow rate are the major factors affecting the mean bubble size. Two-phase water modeling​ experiments were performed to generate bubbles in the mold, and the mean bubble diameter was captured using various gas and liquid flow rates. Clear images of bubbles were recorded using a high-speed high-resolution camera along with the application of shadowgraphy. Bubble images were processed using an image processing software, ImageJ to obtain bubble characteristics, and Sauter mean diameter was calculated for each operating condition. Advance machine learning techniques such as Multilinear Regression (MLR), Decision Tree (DT), Support Vector Machine (SVM), and Artificial Neural Network (ANN) were used on the experimental data to predict the combined effect of these operating parameters on the mean bubble diameter. All four ML techniques were compared considering the values of cross-validated adjusted R2, and a performance metric is presented to compare the suitability of ML techniques in this case.