Model-in-the-loop Research Articles

Beat the AI: Investigating Adversarial Human Annotation for Reading Comprehension

Innovations in annotation methodology have been a catalyst for Reading Comprehension (RC) datasets and models. One recent trend to challenge current RC models is to involve a model in the annotation process: Humans create questions adversarially, such that the model fails to answer them correctly. In this work we investigate this annotation methodology and apply it in three different settings, collecting a total of 36,000 samples with progressively stronger models in the annotation loop. This allows us to explore questions such as the reproducibility of the adversarial effect, transfer from data collected with varying model-in-the-loop strengths, and generalization to data collected without a model. We find that training on adversarially collected samples leads to strong generalization to non-adversarially collected datasets, yet with progressive performance deterioration with increasingly stronger models-in-the-loop. Furthermore, we find that stronger models can still learn from datasets collected with substantially weaker models-in-the-loop. When trained on data collected with a BiDAF model in the loop, RoBERTa achieves 39.9F1 on questions that it cannot answer when trained on SQuAD—only marginally lower than when trained on data collected using RoBERTa itself (41.0F1).

Active Damping Control of Multiport DC Power Flow Controller for MMC-MTDC With Unbalanced AC Grid

Multiport dc power flow controller (M-DCPFC) has become one of the most attractive solutions to regulate the power flow in the modular multilevel converter-based multiterminal dc (MMC-MTDC) transmission system. However, the current studies on the M-DCPFC mainly focus on the power flow control under the normal operation in the transmission system. To eliminate the impact of the power oscillation which frequently happens because of the unbalanced ac grid, this article proposes an integration to calibrate the active damping control into the M-DCPFC. It introduces the complementary virtual damping to the transmission lines in order to reduce the dc current fluctuations, which suppresses the dc-side power oscillations. In this article, the influence of the unbalanced ac grid on the MMC-MTDC transmission system is investigated and the active damping control of M-DCPFC is presented in detail. A four-terminal MMC-MTDC transmission system is developed as the test platform in the real-time model-in-the-loop (MIL) simulation. The proposed active damping control is validated and shows its effectiveness under different unbalanced conditions.

Predictive Model Based Low-Speed Adaptive Cruise Control for Autonomous Vehicles

European Union confirmed the “Vision Zero” objective in June 2019, as to achieve zero deaths and serious injuries by 2050. This can only be attained through connected and autonomous vehicles integrated into intelligent transport systems and a sustainable mobility system. This requires a cost-effective, fast,and efficient development process for advanced connected and autonomous vehicle functions. In this article, a methodology to develop low-speed Adaptive Cruise Control (ACC), which is one of the most important functions of an autonomous vehicle, is explained. Vehicle tracking at slow speeds is a problem especially for conventional vehicles with high levels of nonlinearities in the powertrain system. As a part of a university-industry collaboration project “SAE level 3 autonomous bus development”, a flexible and realistic discrete plant model including longitudinal vehicle and powertrain model has been developed and discrete low-speed ACC is designed. The plant model aims to perform detailed and realistic software tests of autonomous features, which interfaces with the vehicle controllers. OKAN_UTAS autocorrectedmulti-parameter longitudinal model is integrated. For engine modeling via shaft dynamometer the 3D map of the engine is reproduced. The transmission characteristics were prepared through the road tests. To increase the reliability of the developed functions, Software in the Loop (SIL) and Model in the Loop (MIL) simulations were conducted before the on-road vehicle tests. Finally, C code with the MISRA C standard of ACC is generated and embedded into a real-time platform. The plant model, ACC design, and Model in the Loop test results are presented.

Experimental Implementation of Power-Split Control Strategies in a Versatile Hardware-in-the-Loop Laboratory Test Bench for Hybrid Electric Vehicles Equipped with Electrical Variable Transmission

The energy management strategy (EMS) or power management strategy (PMS) unit is the core of power sharing control in the hybridization of automotive drivetrains in hybrid electric vehicles (HEVs). Once a new topology and its corresponding EMS are virtually designed, they require undertaking different stages of experimental verifications toward guaranteeing their real-world applicability. The present paper focuses on a new and less-extensively studied topology of such vehicles, HEVs equipped with an electrical variable transmission (EVT) and assessed the controllability validation through hardware-in-the-loop (HiL) implementations versus model-in-the-loop (MiL) simulations. To this end, first, the corresponding modeling of the vehicle components in the presence of optimized control strategies were performed to obtain the MiL simulation results. Subsequently, an innovative versatile HiL test bench including real prototyped components of the topology was introduced and the corresponding experimental implementations were performed. The results obtained from the MiL and HiL examinations were analyzed and statistically compared for a full input driving cycle. The verification results indicate robust and accurate actuation of the components using the applied EMSs under real-time test conditions.

Pre-Deployment Testing of Low Speed, Urban Road Autonomous Driving in a Simulated Environment

<div class="section abstract"><div class="htmlview paragraph">Low speed autonomous shuttles emulating SAE Level L4 automated driving using human driver assisted autonomy have been operating in geo-fenced areas in several cities in the US and the rest of the world. These autonomous vehicles (AV) are operated by small to mid-sized technology companies that do not have the resources of automotive OEMs for carrying out exhaustive, comprehensive testing of their AV technology solutions before public road deployment. Due to the low speed of operation and hence not operating on roads containing highways, the base vehicles of these AV shuttles are not required to go through rigorous certification tests. The way these vehicles’ driver assisted AV technology is tested and allowed for public road deployment is continuously evolving but is not standardized and shows differences between the different states where these vehicles operate. Currently, AVs and AV shuttles deployed on public roads are using these deployments for testing and improving their technology. However, this is not the right approach. Safe and extensive testing in a lab and controlled test environment including Model-in-the-Loop (MiL), Hardware-in-the-Loop (HiL) and Autonomous-Vehicle-in-the-Loop (AViL) testing should be the prerequisite to such public road deployments. This paper presents three dimensional virtual modeling of an AV shuttle deployment site and simulation testing in this virtual environment. We have two deployment sites in Columbus of these AV shuttles through the Department of Transportation funded Smart City Challenge project named Smart Columbus. The Linden residential area AV shuttle deployment site of Smart Columbus is used as the specific example for illustrating the AV testing method proposed in this paper.</div></div>

An Intelligent Type-2 Fuzzy Stabilization of Multi-DC Nano Power Grids

This paper proposes a robust and intelligent droop control method to stabilize a cluster of DC nanogrids (DCNGs) with constant power loads (CPLs). Conventional droop methods suffer from voltage regulation in parallel converters due to the CPLs destructive effects, which cause instability in the DCNGs. To address the shortcomings of the conventional methodologies, this paper proposes a new droop-based single input interval type-2 fuzzy logic controller (SI-IT2-FLC), which enhances the transient response and improve the steady-state performance. The suggested method aims to boost the stability of the paralleled DCNGs in a cluster under the effect of the CPLs, which is inevitable in the practical applications. The efficacy of the suggested scheme is validated in different operating conditions. Moreover, real-time model-in-the-loop (MiL) results show the performance improvements in the transient and steady-state obtained by the novel suggested control approach.

A Novel Deep Reinforcement Learning Controller Based Type-II Fuzzy System: Frequency Regulation in Microgrids

The high-penetration of distributed generation technologies in the form of MicroGrids (MGs), in recent years, has increased the risk of frequency instability since their energy is supplied by renewable energy resources (RESs) with uncertain nature. Under such circumstances, providing an MG model with an efficient load frequency control (LFC) has a fundamental role in restoring the stability of the unstructured power system. In this study, a hybrid power system with the application of the Tidal Power Unit (TPU) and Vehicle-to-Grid (V2G) is effectively planed as an isolated MG. A new fractional gradient descent (FGD) based on a single-input interval type-2 fuzzy logic controller (SIT2-FLC) is suggested as the main LFC controller, where the footprint of uncertainty (FOU) coefficient of the SIT2-FLC is specifically adjusted to enhance the LFC performance. Additionally, a deep deterministic policy gradient (DDPG) with the actor-critic framework is considered to generate the supplementary control action, which is useful for the frequency stabilization by adapting to the randomness of load disturbances and RESs. Lastly, a model-in-the-loop (MiL) simulation is conducted to appraise the feasibility and applicability of the suggested design method from a systemic perspective.

Co-Simulation of Multi-Domain Engine and its Integrated Control for Transient Driving Cycles

Virtualization of powertrain components allows the front-loading of conventional vehicle calibration and validation tasks to Model-in-the-Loop (MiL) and Hardware-in-the-Loop (HiL) simulations. This approach is based on the utilization of highly accurate physics-based powertrain models that enable a seamless system validation using virtual testing methods in order to ensure cost-effective powertrain development by reducing hardware tests. Proper modelling methods target the optimum between parametrization effort, model accuracy and required computing power to grant the real-time (RT) capability of the simulation, which is mandatory for HiL simulation. In this paper two validated modelling approaches and their implementation into a MiL environment are introduced and discussed. The approaches are the MATLAB/Simulink based Mean Value Engine Model (MVEM) and the Fast-Running Modelling (FRM) of GT-Power. After the models integration in a Simulink frame, the responses of a model-based control unit with the two simulation models were evaluated using real experimental data. In transient cycles, the controller showed a different reaction to the feedback signals of the two engine models. The purposes of the conducted investigation are mainly to evaluate strong and weak points of both approaches and to propose the best-practice modelling approaches for virtual calibration and validation. A comparative rating shows the main advantage of the MVEM in the flexibility for HiL-based systems and the model training effort for the FRM.

Scalable Modeling Approach and Robust Hardware-in-the-Loop Testing of an Optimized Interleaved Bidirectional HV DC/DC Converter for Electric Vehicle Drivetrains

Automotive Original Equipment Manufacturers (OEMs) require varying levels of functionalities and model details at different phases of the electric vehicles (EV) development process, with a trade-off between accuracy and execution time. This article proposes a scalable modelling approach depending on the multi-objective targets between model functionalities, accuracy and execution time. In this article, four different fidelity levels of modelling approaches are described based on the model functionalities, accuracy and execution time. The highest error observed between the low fidelity (LoFi) map-based model and the high fidelity (HiFi) physics-based model is 5.04%; while, the simulation time of the LoFi model is ~10 4 times faster than corresponding one of the HiFi model. A detailed comparison of all characteristics between multi-fidelity models is demonstrated in this paper. Furthermore, a dSPACE SCALEXIO Hardware-in-the-Loop (HiL) testbench, equipped with a minimal latency of 18 μ sec, is used for real-time (RT) model implementation of the EV's HV DC/DC converter. The performance of the entire HiL setup is compared with the Model-in-the-Loop (MiL) setup and the highest RMSE is limited to 0.54 among the HiL and MiL results. Moreover, the accuracy (95.7%) of the passive component loss estimation is verified through the Finite Element Method (FEM) software model. Finally, the experimental results of a full-scale 30-kW SiC DC/DC converter prototype are presented to validate the accuracy and correlation between multi-fidelity models. It has been observed that the efficiency deviation between the hardware prototype and multi-fidelity models is less than 1.25% at full load. Furthermore, the SiC Interleaved Bidirectional Converter (IBC) prototype achieves a high efficiency of 98.4% at rated load condition.

Co-simulation Improvement for Uncertain Flexible Robot Arm

Flexible robot arm driven by Brushless DC Motor (BDCM) under uncertainties represents one of the most complex and heterogeneous system. Indeed, the verification phase becomes a great challenge for designers. Avoid and predict risks accurately at earlier stage represents the main purpose of the Computer-Aided Design (CAD) field. This paper treats the case of robotics system for tracking trajectory problem and attempts to improve the verification phase by identifying the most suitable co-simulation technique. For the system analyzed in this paper, the flexible robot arm driven by Brushless DC actuator is verified using the Model In the Loop (MIL) technique, the Software In the Loop (SIL) technique and CODIS + technique. Each one verifies the system according to a particular abstraction level. The performance of each technique is determined by measuring the accuracy and time simulation. Experimental results have revealed that CODIS+ is the most adequate technique for flexible robot arm, outperforming MIL and SIL by 4–5 times.

Measurement Based Validation of an Electro-Pneumatic Gearbox Actuator

The objective of the research is to analyze the behavior of the developed electro-pneumatic actuator model and compare it to the behavior of the real system. The actuator achieves the requested gear changes by moving the two pistons inside the cylinder and it is operated by three-way two-position solenoid valves. Since not all model parameters are exactly known, such as contraction coefficients and friction parameters, they can be estimated based on literature then they can be further tuned to minimize the error of the simulation. The developed nonlinear model is capable of describing the dynamic behavior of the gearbox actuator, thus it can be used to analyze the effects of constructional modifications and it can serve as Model in the Loop (MIL) environment for controller testing.

A New Adaptive Type-II Fuzzy-Based Deep Reinforcement Learning Control: Fuel Cell Air-Feed Sensors Control

This paper proposes a new adaptive controller for air-feed on proton exchange membrane fuel cell (PEMFC) plants. The oxygen excess ratio is first regulated by a single input interval type-2 fuzzy PI (SIT2-FPI) controller during the current variation of the PEMFC system. Then, a deep deterministic policy gradient (DDPG) algorithm is adopted to adaptively adjust the baseline PI coefficients of the SIT2 controller by taking the benefits of the online learning and model-free features of reinforcement learning. In the DDPG structure, an actor-network determines the policy signals, while a critic-network evaluates the quality of the policy provided by the actor. The suggested DDPG algorithm naturally takes into account the baseline PI coefficients into the design objective and offers the SIT2-FPI structure with online coefficient adjusting ability through learning. Based on reward feedback of the oxygen excess ratio error, the weights of the actor and critic networks are updated by the gradient descent technique. Detailed real-time model-in-the loop (MIL) simulation outcomes and comparative analysis are presented to confirm the adaptation capability of the DDPG-based online SIT2-FPI coefficient tuning strategy.

Pulsed power load effect mitigation in DC shipboard microgrids: a constrained modelpredictive approach

This paper proposes a novel model predictive controller to minimise the effect of unknown pulsed loads on the DC microgrid (MG) side of shipboards. It is assumed that the level of the pulsed power loads (PPLs) is not determined in prior for which the authors propose a novel non-linear power observer. By employing the so-called freezing technique, a novel model predictive scheme is utilised to optimally stabilise the overall DC shipboard MG. Furthermore, different constraints on the current of the DC source and the energy storage system (ESS) are considered in the predictive controller to make it more realistic and practical. Compared to the exiting results, the proposed approach can optimally design the injecting current of the ESS so that the practical and physical constraints of the DC MG are also satisfied, which improves the effectiveness and robustness of the proposed controller. To show the merits of the proposed approach, it is tested on a DC MG that feeds one PPL. Real-time model-in-the-loop (MiL) results show the performance improvements in the transient and steady-state obtained by the proposed control method compared to the state-of-the-art methods.

Dynamic Modeling, Control System Design and MIL–HIL Tests of an Unmanned Rotorcraft Using Novel Low-Cost Flight Control System

Unmanned helicopters have gained great importance during recent years due to their special abilities such as hover flight, vertical take-off and landing, maneuverability and superior agility. The advances in electronic devices technologies lead to more powerful and lighter processors to be used in avionic systems which have attracted more attention to these UAVs. The first steps of utilizing an unmanned helicopter are dynamic modeling, control system design and performing model-in-the-loop (MIL) and hardware-in-the-loop (HIL) tests which are presented in this paper. In this research, MIL and HIL tests of an unmanned helicopter are done using novel Linux-based flight control system built on Raspberry Pi board (different from normally used PC-104 and STM- or Arduino-based systems). Dynamic modeling, robust hierarchical control design, flight control system hardware and software architecture and MIL and HIL test results are reported here. By succeeding in these tests, it is shown that the proposed platform can be used in experimental flight tests in next steps.

Model based design applied to ceramic balls grinding

This research uses model based design to evaluate and realize the complex motion of the lower plates that define balls motion during grinding. The Model In the Loop (MIL) was performed through the machine model generated using ADAMS and then exported to Simulink. For a prior evaluation of the machining, the grinding tracks were built through the simulation results performing Software In the Loop (SIL). The balls grinding was performed by an Arduino connected to the machine, finishing Hardware In the Loop (HIL). The technique integrated the simulation and balls grinding processes, increasing the simplicity and reducing the development time.

Dynamic simulation of a combined cycle for power plant flexibility enhancement

System simulation is used in many fields to help design, control or troubleshoot various industrial systems. Within the PUMP-HEAT H2020 project, it is applied to a combined cycles power plant, with innovative layouts that include heat pumps and thermal storage to un-tap combined cycle potential flexibility through low-CAPEX balance of plant innovations. Simcenter Amesim software is used to create dynamic models of all subsystems and their interactions and validate them from real life data for various purpose. Simple models of the Gas Turbine (GT), the Steam loop, the Heat Recovery Steam Generator (HRSG), the Heat Pump and the Thermal Energy storage with Phase Change material are created for Pre-Design and concept validation and then scaled to more precise design. Control software and hardware is validated by interfacing them with detailed models of the virtual plant by Model in the Loop (MiL), Software in the Loop (SiL) and Hardware in the Loop (HiL) technologies. Unforeseen steady state and transient behaviours of the powerplant can be virtually captured, analysed, understood and solved. The purpose of this paper is to introduce the associated methodologies applied in the PUMP-HEAT H2020 project and their respective results.

Experimental Hysteresis Identification and Micro-position Control of a Shape-Memory-Alloy Rod Actuator

In order to exhaustively exploit the high-level capabilities of shape memory alloys (SMAs), they must be applied in control systems applications. However, because of their hysteretic inherent, dilatory response, and nonlinear behavior, scientists are thwarted in their attempt to design controllers for actuators of such kind. The current study aims at developing a micro-position control system for a novel SMA rod actuator. To do so, the hysteretic behavior of the actuator was simulated in the form of a gray-box Wiener model. Based on the experimental training dataset obtained from the actuator, the hysteresis Wiener model was trained using a PSO algorithm. Afterwards that the identified hysteresis Wiener model was validated, the authors formed a model-in-the-loop (MIL) position control system. Next, a PSO algorithm was again set to find the best controller parameters regarding some performance criteria. At the end, implemented on the fabricated prototype (the experimental setup), the designed control system shared such excellent accuracy that makes the fabricated actuator amenable to micro-positioning applications.

Towards increasing nanosatellite subsystem robustness

Short development life cycle and low cost of cubesat-based mission have motivated the growing number of nanosatellite launched in the last decade around the world. Fast and cheaper space project do not guarantee success in orbit. The lack of good practices on design, assembly and tests has been pointed out as one of the major causes to nanosatellite mission failures. Efforts on the use of verification and validation techniques are required. Because the increased use of nanosatellites missions for technology qualification of payloads on orbit, faulty behavior of those payloads can be expected. However, such malfunction shall not represent a risk to the whole mission. Robustness is an important property of reactive critical system not addressed properly in the cubesat standardization. Although significant mitigation of the interface failures has been observed at hardware level in the integration phase of the payloads with the nanosatellite platform, behavior aspects of the communicating subsystems on the use of these interfaces shall be verified. The test systematization of CubeSat-based nanosatellites supported by proper tools is necessary to reduce the mission development cycle in terms of the time consumed by the verification & validation activities. In this paper we present a failure emulator mechanism framework, named FEM, for robustness testing of interoperable software-intensive subsystems onboard nanosatellite. FEM acts in the communication channel being part of the integration test workbench in two phases of nanosatellite design: (i) robustness requirement specification using model in the loop (MIL) and (ii) robustness validation using hardware in the loop (HIL). The architectural aspects of the proposed FEM framework support its instantiation to any communication channel of the CubeSat standard. As an example, FEM prototype was instantiated to I2C communication channel to support NanosatC-BR2 testing. NanosatC-BR2 is a Cubesat based scientific mission, under development and integration at Brazilian Institute for Space Research (INPE), which uses I2C communication channel for its payloads interactions with the On-Board Data Handling computer subsystem (OBC). FEM prototype was used to support OBC integration testing with each payload subsystem at MIL scenario aiming at anticipating the robustness requirement verification on the development lifecycle. Moreover, the FEM prototype was also validated at HIL scenario using Test Cases automatically generated. Results of these two scenarios executions are reported demonstrating in a case study the effectiveness of FEM framework in detecting the lack or noncompliance of robustness requirements by the interoperated subsystems under testing.

Development of a pressure-based technique to control IMEP and MFB50 in a 3.0L diesel engine

A pressure-based technique for the control of IMEP (Indicated Mean Effective Pressure) and MFB50 (crank angle at which 50% of fuel mass fraction has burned) has been developed, assessed and tested by means of MiL (Model-in-the-Loop) on a 4 cylinder 3.0L Euro VI diesel engine.The activity was carried out in the frame of a research project in collaboration with FPT Industrial.The developed controller is of the closed-loop type. It receives, as input, the desired targets of IMEP and MFB50 for each cycle and cylinder and performs a cycle-by-cycle and cylinder-to-cylinder correction of the injected fuel quantity of the main pulse (qmain) and of the start of injection of the main pulse (SOImain), in order to reduce the deviation between the actual and target values of IMEP and MFB50, respectively. The method is referred to as “pressure-based” since it requires the measurement of the in-cylinder pressure trace for each cylinder in order to extract the actual values of IMEP and MFB50. In fact, the actual IMEP value can be estimated by integrating the pressure signal with respect to the in-cylinder volume. At the same time, the actual MFB50 value can be extracted from the heat release curve, which is obtained from the in-cylinder pressure trace by using a single-zone heat release model. The proposed control technique has been developed in Simulink environment, and has been assessed and tested on an engine emulator which is constituted by a GT-power model of the 3.0L diesel engine.The controller has been tested in transient operation over a load ramp profile at different engine speeds and over a WHTC interval, and demonstrated to have a good potential for IMEP and MFB50 control, since it is characterized by a fast response and a limited overshoot behavior.

Design of a Model-Based Fuzzy-PID Controller with Self-Tuning Scaling Factor for Idle Speed Control of Automotive Engine

Automotive engines spend a considerable portion of their life in idle state, where their characteristics are highly nonlinear and time varying. Efficient control of the engine in idle state significantly improves fuel economy, emission level and drivability. In this work, we proposed a combined fuzzy-PID idle speed controller which improves stability, fuel consumption and emission level, during normal operation as well as transient loads. In order to study the performance of control algorithm, XU7-1761cc gasoline port fuel injection engine has been modeled in idle state. This model calculates engine speed as a function of idle air valve position and ignition angle. This model demonstrated maximum of 8% error during parallel model in the loop and engine in the loop operation. After tuning the membership functions, performance of fuzzy-PID controller has been compared with baseline PID controller of engine control unit. Model in the loop simulation demonstrated that fuzzy-PID controller is more accurate than baseline PID controller, unconditionally stable and is able to reduce fuel consumption in the order of 14% compared to baseline controller. Finally, an experimental setup for real-time control of idle air valve based on proposed controller has also been developed.