Model-based Techniques Research Articles

POSITION CONTROL OF ELECTRO-HYDRAULIC SERVO SYSTEM WITH FRICTION COMPENSATION

This work presents a model-based friction compensation technique and an integral adaptive backstepping controller to improve the position control of the electrohydraulic servo system. The controller design uses a continuous approximation of the LuGre friction model and the Lyapunov theory of nonlinear systems. The friction compensation enhances system performance, and comparison tests are conducted on a hydraulic servo test bench to validate the control strategy's tracking performance under various conditions. KEYWORDS: Electrohydraulic servo systems; Position control; Integral adaptive backstepping controller; LuGre dynamic friction model; Lyapunov theory; Friction compensation.

Gaussian process regression for forward and inverse kinematics of a soft robotic arm

The use of soft robotics to perform tasks and interact with the environment requires good system identification. Data-driven methods offer a promising alternative where traditional analytical model-based techniques have proven insufficient. However, their use has been limited and under-explored in soft robotics. The novelty of this research lies in the application of Gaussian processes to soft robotics and the exploration of approximate Gaussian processes (AGP) and deep Gaussian processes (DGP) methods. It highlights the advantages of Gaussian processes in modeling uncertainty, incorporating prior knowledge, and handling complex systems. This is achieved through the identification of the forward and inverse kinematics of a two-degree-of-freedom soft robotic arm actuated by three tendons. A comparison is made between different configurations using Gaussian processes and the results are also compared with those obtained from the analytical model of the kinematics and an artificial neural network (ANN). The research contributes to the development of more efficient and accurate techniques for system identification, kinematics modeling, and control in soft robotics.

Learning disentangled representations for controllable human motion prediction

Generative model-based motion prediction techniques have recently realized predicting controlled human motions, such as predicting multiple upper human body motions with similar lower-body motions. However, to achieve this, the state-of-the-art methods require either subsequently learning mapping functions to seek similar motions or training the model repetitively to enable control over the desired portion of body. In this paper, we propose a novel framework to learn disentangled representations for controllable human motion prediction. Our task is to predict multiple future human motions based on the past observed sequence, with the control of partial-body movements. Our network involves a conditional variational auto-encoder (CVAE) architecture to model full-body human motion, and an extra CVAE path to learn only the corresponding partial-body (e.g., lower-body) motion. Specifically, the inductive bias imposed by the extra CVAE path encourages two latent variables in two paths to respectively govern separate representations for each partial-body motion. With a single training, our model is able to provide two types of controls for the generated human motions: (i) strictly controlling one portion of human body and (ii) adaptively controlling the other portion, by sampling from a pair of latent spaces. Additionally, we extend and adapt a sampling strategy to our trained model to diversify the controllable predictions. Our framework also potentially allows new forms of control by flexibly customizing the input for the extra CVAE path. Extensive experimental results and ablation studies demonstrate that our approach is capable of predicting state-of-the-art controllable human motions both qualitatively and quantitatively.

A Robust Model-Based Strategy for Real-Time Fault Detection and Diagnosis in an Electro-Hydraulic Actuator Using Updated Interactive Multiple Model Smooth Variable Structure Filter

Abstract In industries where harsh environments and stringent safety requirements are prevalent, the widespread use of applications has made it essential to focus on fault detection and diagnosis (FDD) in hydraulic actuators. To achieve this, model-based FDD techniques are utilized, which employ estimation tools like observers and filters. However, for many applications, particularly in fluid power systems, observability, and parameter uncertainty pose constraints to extracting information and estimating parameters. To address these issues, an efficient form of interactive multiple model (IMM), called updated IMM (UIMM), is applied to an electro-hydraulic actuator (EHA) to detect and isolate persisting friction and leakage faults. UIMM method progresses through a series of models that correspond to the fault condition's progression instead of considering all models at once (as is done in IMM). This reduces the number of models running simultaneously, providing two significant benefits: enhanced computational efficiency and avoidance of the combinatorial explosion. The smooth variable structure filter with variable boundary layer (SVSF-VBL) is used for state and parameter estimation in conjunction with UIMM. SVSF-VBL is a reliable suboptimal estimation method that performs better than the Kalman filter regarding uncertainties related to the system and modeling. The performance of the UIMM method is validated by the simulation of fault conditions for a typical EHA. A fault tolerable control system (FTCS) has been designed to demonstrate the use of the proposed FDD strategy for fault management in a closed-loop system.

Spatial heterogeneity in intimate partner violence across the 640 districts of India: a secondary analysis of a cross-sectional, population-based survey by use of model-based small-area estimation

Although intimate partner violence (IPV) against women is a substantial challenge in India, response is limited by little evidence on substate prevalence. District-level IPV estimates are essential in targeted response and prevention efforts, but cannot be directly calculated from the National Family Health Surveys (NFHS), which is the main source of nationally representative IPV estimates in India. We aimed to use small-area estimation techniques to derive reliable estimates of physical, emotional, and sexual IPV for the 640 districts of India. For this secondary analysis of a cross-sectional, population-based survey, we used model-based small-area estimation techniques linking data from the 2015-16 NFHS-4 and the 2011 Indian Population and Housing Census (2011 Indian Census) to derive district-level estimates of physical, emotional, and sexual IPV for the 640 districts of India in the previous 12 months. Only women who had ever been married aged 15-49 years, who were interviewed in NFHS-4, and who were included in the domestic violence module were eligible for inclusion in this analysis. Data collection occurred between Jan 20, 2015, and Dec 4, 2016. The 2011 Indian Census was conducted in all 640 districts from Feb 9 to Feb 28, 2011. It collected information on a range of data including sociodemographic data and housing characteristics. The primary outcomes of this analysis were the district-level mean proportions of women who experienced physical IPV, emotional IPV, and sexual IPV in the previous 12 months. This outcome was estimated for all women aged 15-49 years who had ever been married in the 640 districts of India that were included in the 2011 Indian Census. 699 686 women aged 15-49 years were interviewed in NFHS-4. One woman per household in a randomly selected 15% of households was chosen for participation in the domestic violence module, resulting in 83 397 (11·9%) of 699 686 women included. Of these 83 397 women, 14 377 (17·2%) were excluded as they had never been married and 3007 (3·6%) were excluded due to privacy limitations. The mean prevalence of physical IPV in the previous 12 months was 22·5% (95% CI 21·9-23·2), of emotional IPV in the previous 12 months was 11·4% (11·0-11·9), and of sexual IPV in the previous 12 months was 5·2% (4·9-5·5). Model-based estimates revealed intrastate and interstate IPV variations. In Bihar, which had the highest state-level physical IPV prevalence (35·1%, 33·3-37·0), district-level estimates varied from 23·5% (23·0-23·9) in Siwan to 42·7% (42·3-43·1) in Purbi Champaran. In Tamil Nadu, which had the highest state-level emotional IPV prevalence (19·0%, 17·4-20·8), district estimates ranged between 13·7% (13·2-14·1) in Kanniyakumari and 30·2% (29·5-30·8) in Sivaganga. Bihar also had the highest state-level sexual IPV prevalence (11·1%, 9·9-12·4), with estimates ranging between 6·3% (6·1-6·6) in Siwan and 18·1% (17·6-18·6) in Saharsa. Across districts, there was substantial spatial clustering of IPV prevalence. This reliable district-level estimation of IPV prevalence in the 640 districts of India has important policy implications. The ability to track substate levels of IPV over time enables the identification of progress in reducing IPV; recognises the heterogeneity of culture and context in India; and informs the targeting of resources, interventions, and prevention programmes to districts with the greatest need. Bill & Melinda Gates Foundation.

Experimentally Validated Neural Networks for Sensors Redundancy Purposes in Spark Ignition Engines

<div>In the aeronautical field, aircraft reliability is strictly dependent on propulsion systems. Indeed, a reliable propulsion system ensures the safety of pilots and passengers and the possibility of making comfortable flights. Typically, on aircraft equipped with spark ignition (SI) engines, one of the principal requirements to make them reliable is the correct balancing between the intake air mass and fuel flows. Advances in the implementation of sophisticated control and estimation strategies on SI engines allow realizing engines with improved features in terms of performance, reducing pollution emissions, and fuel consumption. Approaches based on sensor redundancy are applied to improve the reliability in measurements of the manifold air pressure (MAP) and flow (MAF) to avoid issues related to possible faults of sensors vital for the correct functioning of SI engines. Model-based estimation techniques, based on the speed–density and alpha-speed methods for determining the MAF in engine control units, are employed to obtain sensor-less redundancy. The prediction of MAP and MAF, for sensors redundancy purposes, can be made through neural networks, allowing the avoidance of effects due to unmodeled dynamical behaviors. A sensor redundancy approach based on feedforward neural networks (FNNs) is proposed in this work for MAP and MAF prediction of a SI engine. The present work focuses on the possibility of estimating the physical quantities related to SI engines, such as the MAP and the MAF, fundamental for their monitoring using neural networks trained by means of a model-based approach avoiding expensive experimental tests for producing training data. A well-known intake manifold dynamical model (IMDM), parametrized based on the CMD 22 aeronautical engine, is employed for generating synthetic training data in steady-state conditions functional for making the chosen FNNs able to predict both MAP and MAF even in transient behavior.</div> <div>The MAP and MAF are predicted through two virtual sensors based on two independent FNNs, having the same inputs, constituted by the engine speed and the throttle angle. An experimental investigation based on an aircraft endurance test of two hours proposed by the European Aviation Safety Agency (EASA) has been made on a controlled and monitored CMD 22 engine for comparing the experimentally measured MAP and MAF with the predicted ones by the FNNs. The results demonstrate the suitability of the proposed approach for sensor redundancy purposes in SI engines to increase their reliability.</div>

Comparative Analysis of Centralized and Federated Learning Techniques for Sensor Diagnosis Applied to Cooperative Localization for Multi-Robot Systems.

Cooperation in multi-vehicle systems has gained great interest, as it has potential and requires proving safety conditions and integration. To localize themselves, vehicles observe the environment using sensors with various technologies, each prone to faults that can degrade the performance and reliability of the system. In this paper, we propose the coupling of model-based and data-driven techniques in diagnosis to produce a fault-tolerant cooperative localization solution. Consequently, prior knowledge can guide a discriminative model that learns from a labeled dataset of appropriately injected sensor faults to effectively identify and flag erroneous readings. Going further in security, we conduct a comparative study on learning techniques: centralized and federated. In centralized learning, fault indicators generated by model-based techniques from all vehicles are collected to train a single model, while federating the learning allows local models to be trained on each vehicle individually without sharing anything but the models to be aggregated. Logistic regression is used for learning where parameters are established prior to learning and contingent upon the input dimensionality. We evaluate the faults detection performance considering diverse fault scenarios, aiming to test the effectiveness of each and assess their performance in the context of sensor faults detection within a multi-vehicle system.

A Multimodal Deep Learning Model Using Text, Image, and Code Data for Improving Issue Classification Tasks

Issue reports are valuable resources for the continuous maintenance and improvement of software. Managing issue reports requires a significant effort from developers. To address this problem, many researchers have proposed automated techniques for classifying issue reports. However, those techniques fall short of yielding reasonable classification accuracy. We notice that those techniques rely on text-based unimodal models. In this paper, we propose a novel multimodal model-based classification technique to use heterogeneous information in issue reports for issue classification. The proposed technique combines information from text, images, and code of issue reports. To evaluate the proposed technique, we conduct experiments with four different projects. The experiments compare the performance of the proposed technique with text-based unimodal models. Our experimental results show that the proposed technique achieves a 5.07% to 14.12% higher F1-score than the text-based unimodal models. Our findings demonstrate that utilizing heterogeneous data of issue reports helps improve the performance of issue classification.

Overview of Position-Sensorless Technology for Permanent Magnet Synchronous Motor Systems

In recent years, permanent magnet synchronous motors (PMSMs) have been widely used in industry. Position-sensorless control has the advantages of reducing costs and improving reliability, and is becoming one of the most promising technologies for permanent magnet synchronous motors. This article reviews the main position-sensorless technologies. The advantages and disadvantages of model-based and saliency-based techniques were summarized and compared. Finally, the developmental trends and research directions of position-sensorless technology were discussed.

Prediction of combustion, performance, and emission parameters of ethanol powered spark ignition engine using ensemble Least Squares boosting machine learning algorithms

This research concentrates on the application of machine learning techniques to predict combustion, performance, and emission parameters in a dual-fuel spark ignition (SI) engine powered by neat gasoline and E20 ethanol dual fuel. The goal is to overcome the limitations posed by repeated engine experiments and nonlinear test results. In order to optimise engine parameters, the research seeks to develop efficient machine learning models with high generalizability and employ an optimisation strategy to determine the optimal engine settings. Input for training and evaluating machine learning algorithms, such as Artificial Neural Networks (ANN), and Ensemble LS Boosting was derived from experimental data from a combustion test engine, which includes Neat gasoline, and Ethanol dual fuel blend E20 at various load conditions. The dataset includes engine combustion, performance, and emission indices such as brake thermal efficiency (BTE), Exhaust Gas Temperature (EGT), Hydrocarbons (HC), Carbon monoxide (CO), Carbon dioxide (CO2), and Nitrogen oxides (NOx), under various operating conditions. Load and brake-specific fuel consumption (BSFC) were training input attributes. Using a comprehensive experimental database of input-output engine parameters, the Artificial Neural Network (ANN) and Ensemble LS Boosting were constructed. The training data points were resampled to generate multiple training datasets for training different models. 50 test samples were used to evaluate the generalisation capability of the machine learning models, while BTE, EGT, CO, CO2, HC and NOx, were the primary parameters subject to prediction. The optimal machine learning method was determined by comparing R-squared (R2) values, root mean square error (RMSE), mean square error (MSE), and mean absolute error (MAE). Using multiple hyperparameter tuning iterations, the agreement between actual and predicted values for diverse Ensemble LS Boost algorithms was evaluated. The Ensemble LS Boost model exhibits the maximum level of agreement between predicted and experimental engine parameters across all datasets when compared to the other ANN models. This finding was corroborated by additional research based on test datasets, specifically the test sample interpolation data, which measures generalisation ability. The study also focuses on developing and applying two unique, interactive Simulink models for the Spark Ignition (SI) engine that are tailored for Neat Gasoline and Ethanol E20 test fuels under all loads. The key component of the model-based development technique in MATLAB and Simulink was the incorporation of sophisticated machine learning algorithms, i.e., Ensemble Least-Squares (LS) Boosting, to the model-based development workflow which produced reliable results. Implementing an Ensemble LS Boost machine learning framework is therefore highly recommended as an efficient method for predicting and optimising the combustion, performance, and emission characteristics of dual-fuel gasoline engines utilising Ethanol-based dual-fuel blends.

Modified extremum seeking controller for hydraulic turbine speed control

In the past, researchers have done various attempts to improve hydraulic turbine precise speed control and use synchronization time between turbine/generator to grid as a test bench. These attempts were either based on improving the instant when phase difference between two systems becomes zero or to control speed precisely using mathematical model-based techniques. Due to highly nonlinear nature of the hydro turbines, the models presented in the past were never perfect. In this research work, a model free approach is adopted to improve the precise speed control of hydraulic turbines. The developed model free system outperformed old model-based techniques in both stable as well as unstable grids. As per results obtained, grid instability had negligible effect on the speed control and synchronization time. The developed system is able control the hydro turbine speed during synchronization up to a 5 mHz level.

Simulation and optimisation of a medium scale reverse osmosis brackish water desalination system under variable feed quality: Energy saving and maintenance opportunity

In this work, we considered model-based simulation and optimisation of a medium scale brackish water desalination process. The mathematical model is validated using actual multistage RO plant data of Al- Hashemite University (Jordan). Using the validated model, the sensitivity of different operating parameters such as pump pressure, brackish water flow rate and seasonal water temperature (covering the whole year) on the performance indicators such as productivity, product salinity and specific energy consumption of the process is conducted. For a given feed flow rate and pump pressure, winter season produces less freshwater that in summer in line with the assumption that winter water demand is less than that in summer.With the soaring energy prices globally, any opportunity for the reduction of energy is not only desirable from the economic point of view but is an absolute necessity to meet the net zero carbon emission pledge by many nations, as globally most desalination plants use fossil fuel as the main source of energy. Therefore, the second part of this paper attempts to minimise the specific energy consumption of the RO system using model-based optimisation technique. The study resulted not only 19 % reduction in specific energy but also 4.46 % increase in productivity in a particular season of the year. For fixed product demand, this opens the opportunity for scheduling cleaning and maintenance of the RO process without having to consider full system shutdown.

Overall competitiveness efficiency: A quantitative approach to the five forces model

This study proposes Porter’s five forces model-based technique that measures a company’s competitiveness through objective indicators and metrics. We introduce a novel model, the overall competitiveness efficiency, which points out that a company’s competitiveness level incorporates capabilities matching five environmental forces. The five forces are: bargaining power of suppliers, bargaining power of buyers, threats of substitutes, threats of new entrants, and rivalry among existing competitors. The overall competitiveness efficiency model also consists of five dimensions: bargaining power against suppliers, bargaining power against buyers, competitive power against substitutes, competitive power against new entrants, and competitive power against competitors. We develop three quantifiable indicators for each dimension and operationalize these indicators with objective metrics. To show the models’ applicability, we conduct a case study in a hydraulics press industry company by collecting quantitative data from document analysis, secondary data analysis, semi-structured interviews, and observations. Our model gives managers insight into improving the company’s competitiveness by focusing on weaknesses and capitalizing on the strengths through measurable competitiveness indicators and metrics. The models’ implications guide managers in making strategic decisions on the competitive position in the market.

Wireless communications empowered by reconfigurable intelligent surfaces: Model-based vs model-free channel estimation

Reconfigurable intelligent surfaces (RISs) are lately being attractive for their great potential in future sixth generation wireless communications (6G), which is attributed to their affordable energy consumption and easy integration. However, the large numbers of low-cost reflecting elements comprising RISs impose challenges for channel acquisition in various RIS-based wireless applications, such as RIS-enhanced orthogonal frequency-division multiplexing and multi-user multiple-input multiple-output systems. In this article, we first overview the state-of-the-art RIS hardware architectures designed to assist channel estimation for RIS-empowered wireless communication systems. We also overview existing channel estimation approaches, which are categorized into model-based and model-free techniques, and discuss their advantages and limitations depending on the RIS deployment. Design challenges with RIS-empowered systems in terms of hardware and other parameter limitations are presented, together with future research directions for channel estimation in RIS-based wireless systems, such as RISs with extremely large numbers of elements, multi-hop communications with RISs, and frequency division duplexing for high mobility systems.

Frequency-Domain Data-Driven Position-Dependent Controller Synthesis for Cartesian Robots

Cartesian robots have position-dependent dynamics that must be taken into account for high-performance applications. Traditional methods design linear time-invariant (LTI) controllers that are robustly stable with respect to position variations but result in reduced performance. Advanced methods require linear parameter varying (LPV) models and LPV controller design methods that are not well-established in the industry. On the other hand, the classical model-based gain-scheduled technique involves parametric identification, high-performance controller design for each position, interpolation of the controller parameters, and real-time controller validation, making it time-consuming and costly. Our approach uses frequency response at different operating points to design an LPV controller using a convex optimization algorithm based on second-order cone programming. The approach is applied to an industrial three-axis Cartesian robot, showing significant improvements over state-of-the-art control design strategies. Data acquisition and controller design can be performed automatically, reducing significantly the engineering costs for controller synthesis.

Optimization-Based Design and Control of Dynamic Systems

Dynamic systems such as robots, autonomous vehicles, and process plants require careful design and control to achieve optimal performance. This paper presents an integrated framework for the simultaneous optimization of system design parameters and control policies. The underlying mathematical formulation utilizes optimization techniques to find the best system configuration and control strategy according to specified objectives. These objectives may include metrics such as tracking error, energy consumption, safety margins, and cost. Both model-based and data-driven techniques are explored for learning the system dynamics and synthesizing optimal controllers. The optimization algorithms leveraged include gradient-based methods, evolutionary algorithms, and reinforcement learning. The benefits of the joint optimization approach are demonstrated through case studies on representative dynamic systems. Results show that the integrated design and control framework outperforms sequential optimization, leading to improved efficiency, responsiveness, and robustness. This underscores the importance of co-optimizing design and control parameters, especially for complex, uncertain systems. The proposed methods provide an effective tool for next-generation automated design of smart, adaptable systems.

HyperJump: Accelerating HyperBand via Risk Modelling

In the literature on hyper-parameter tuning, a number of recent solutions rely on low-fidelity observations (e.g., training with sub-sampled datasets) to identify promising configurations to be tested via high-fidelity observations (e.g., using the full dataset). Among these, HyperBand is arguably one of the most popular solutions, due to its efficiency and theoretically provable robustness. In this work, we introduce HyperJump, a new approach that builds on HyperBand’s robust search strategy and complements it with novel model-based risk analysis techniques that accelerate the search by skipping the evaluation of low risk configurations, i.e., configurations that are likely to be eventually discarded by HyperBand. We evaluate HyperJump on a suite of hyper-parameter optimization problems and show that it provides over one-order of magnitude speed-ups, both in sequential and parallel deployments, on a variety of deep-learning, kernel-based learning and neural architectural search problems when compared to HyperBand and to several state-of-the-art optimizers.

Bayesian model-guided antimicrobial therapy in pediatrics.

Antimicrobials have transformed the practice of medicine, making life-threatening infections treatable, but determining optimal dosing, particularly in pediatric patients, remains a challenge. The lack of pediatric data can largely be traced back to pharmaceutical companies, which, until recently, were not required to perform clinical testing in pediatrics. As a result, most antimicrobial use in pediatrics is off-label. In recent years, a concerted effort (e.g., Pediatric Research Equality Act) has been made to fill these knowledge gaps, but progress is slow and better strategies are needed. Model-based techniques have been used by pharmaceutical companies and regulatory agencies for decades to derive rational individualized dosing guidelines. Historically, these techniques have been unavailable in a clinical setting, but the advent of Bayesian-model-driven, integrated clinical decision support platforms has made model-informed precision dosing more accessible. Unfortunately, the rollout of these systems remains slow despite their increasingly well documented contributions to patient-centered care. The primary goals of this work are to 1) provide a succinct, easy-to-follow description of the challenges associated with designing and implementing dose-optimization strategies; and 2) provide supporting evidence that Bayesian-model informed precision dosing can meet those challenges. There are numerous stakeholders in a hospital setting, and our intention is for this work to serve as a starting point for clinicians who recognize that these techniques are the future of modern pharmacotherapy and wish to become champions of that movement.

Attitude control of 3D soft pneumatic actuators based on BP neural network

Soft pneumatic robotics have attracted considerable attention in recent years due to their deformation capabilities, which far exceed those of conventional robotics. However, precise control of soft pneumatic actuators remains a challenge due to the lack of model-based control techniques. This work aims to employ a high-precision and low-cost backpropagation (BP) neural network-based model method to control a 3D soft pneumatic actuator. Experiments show that this BP neural network-based model control method performs well in terms of precision, in which the errors of bending angle and deflection angle are within 0.8° and 1.2°, respectively, and the end point position error of the soft actuator is less than 2.5 mm, which is significantly better than traditional modeling methods, demonstrating the application potential of soft robots for high-precision operations.

Autonomous Drone Electronics Amplified with Pontryagin-Based Optimization

In the era of electrification and artificial intelligence, direct current motors are widely utilized with numerous innovative adaptive and learning methods. Traditional methods utilize model-based algebraic techniques with system identification, such as recursive least squares, extended least squares, and autoregressive moving averages. The new method known as deterministic artificial intelligence employs physical-based process dynamics to achieve target trajectory tracking. There are two common autonomous trajectory-generation algorithms: sinusoidal function- and Pontryagin-based generation algorithms. The Pontryagin-based optimal trajectory with deterministic artificial intelligence for DC motors is proposed and its performance compared for the first time in this paper. This paper aims to simulate model following and deterministic artificial intelligence methods using the sinusoidal and Pontryagin methods and to compare the differences in their performance when following the challenging step function slew maneuver.