The scientific programme of ACES 2020 will be operated by two most renowned research laboratories in the fields of electrical engineering and automation: the Research Centre in Computer Science, Signal and Automatic Control of Lille (CRIStAL, 400 members) and the Laboratory on Electrical Energy and Power Electronics (L2EP, 100 members).
It will take place at the core of the University of Lille campus: in the pioneering-spirited building of Lilliad Learning Centre Innovation, an unprecedented combination for showing, promoting, thinking and making innovation.
The University of Lille has a double Master degree with Harbin Institute of Technology (HIT); the best Chinese students are selected to continue in co-supervised PhD degrees in the CRIStAL or L2EP research laboratories.
Both labs offer PhD theses on various topics including automatic control and electrical systems. Globally ¼ of the PhD are co-supervised with international universities (Canada, China, Europe, North Africa, etc.) and ¼ of the PhD are in collaboration with the industrial world (Alstom, PSA Peugeot Citroën, Renault, Siemens, etc.).
The CRIStAL research laboratory has a joint research laboratory with Nanjing University of Sciences and Technology (NUST), focused on automatic control: the LaFCAS. The L2EP research laboratory has annually held an international Summer School “Energetic Macroscopic Representation, modelling and control of electric vehicles and other applications” since 2006 (even year in Lille, odd year abroad) – the 2008 edition was organised in Harbin (China).
These parnerships and experiences will benefit the ACES 2020 Summer School.
Focusing on the fundamental aspects of control, observation and electrical systems, ACES 2020 will offer top international conferences, lectures, seminars and meetings with leading researchers in the fields of automatic control and electrical systems, simulation session and demonstrations of cutting-edge equipment, scientific and industrial visits, as well as an introduction to research through a supervised project. The quality of the work undertaken and your scientific potential will be assessed.
Particular emphasis will be laid on hybrid systems, embedded and networked control systems, electric vehicles and autonomous vehicles, capable of inventing and developing new breakthroughs and technological developments.
Half of the schedule will be devoted to the 60-hour scientific programme taught in English.
30h / LECTURES & SEMINARS
3h / ASSESSMENT
6h / SIMULATION SESSION
6h / CONFERENCES
12h / SCIENTIFIC RESEARCH PROJECT
3h / SCIENTIFIC VISIT & INDUSTRIAL MEETING
Keywords: Advanced Modelling Methods, Advanced Control Methods, Hybrid and Electric Vehicles, Energetic Macroscopic Representation, Innovative Energy Storage Systems, High-Efficiency Electrical Machines, Networked/Embedded Control, Geo-Localisation.
The teaching programme of ACES 2020 is divided in two parts on the following topics:
1. ELECTRICAL SYSTEMS
C1: Hybrid and Electric Vehicles
State-of-the-art of hybrid and electric vehicles (limits of conventional vehicles, different kinds of electric and hybrid vehicles, challenges of low-carbon vehicles, examples of innovative developments).
C2: Energetic Macroscopic Representation
EMR is the graphical formalism for the analysis of power flow with a system. It leads to define the relevant design, control and energy management of complex energetic systems, such as renewable energy conversion systems or advanced transportation systems.
C3: Simulation of an Electric Vehicle (see below)
2. AUTOMATIC CONTROL
C4: An Introduction to Reinforcement Learning
AI-based systems that are able to learn complex strategies by interacting with their environment have recently gained a lot of attention. One of the most frequently reported such achievement in the media is Google’s AlphaGo which defeated the world Go game champion. These artificially trained decision making systems can also be applied to control theoretic issues. Indeed the feedback loop principle consists in checking if a given input (or command signal) achieves the prescribed output. In the reinforcement learning paradigm, the similarity between the actual output and the prescribed one is regarded as a reward. The learner can thus select inputs that bring as much as cumulated reward over time as possible.
In this course, I will present the general concepts of reinforcement learning which mainly rely on the Markov Decision Process model. I will show that decision functions, also called policies, can learned by maximizing the expected cumulative reward when the set of actions is finite. Extensions to continuous actions spaces will be evoked and examples of learned controllers will be presented.
C5: System Identification
In this course you will be introduced to the basics of the identification of linear systems and, if time permits, we will also discuss its extension to nonlinear and hybrid systems. We will also cover the theoretical basis of system identification, including the realisation theory, minimality, reachability, observability, the existence of a realisation, the notion of Hankel matrix, identifiability and persistence of excitation.
C6: Estimation in Linear and Nonlinear Systems
First, the basic notions of observability will be discussed for linear and nonlinear systems invoking the geometric and algebraic approaches. Second, the problem of the observer design in linear systems will be dealt with (full and reduced order observers). Finally, a nonlinear observer design will be considered, which is converging in a finite time using the notion of homogeneity.
C7: Multi-Sensor Data Fusion for Mobile Robot Localisation
The robot localization is the problem of estimating the pose of the robot relative to a map. In other words, the robot has to answer accurately the question “Where am I?” while using erroneous measurements sensors and inaccurate map.
How can robots determine their state and properties of the surrounding environment from noisy sensor measurements in time? In this module you will learn how to get robots to incorporate uncertainty into estimating state. Specific topics that will be covered include probabilistic models, Bayesian filtering for localization and mapping. The course will cover different topics and techniques in the context of mobile robots localization. We will cover also techniques such as SLAM with the family of Kalman filters, information filters. For this part, the attention will be devoted to the integrity monitoring of the state estimation method.
C8: Analysis and Design of Networked Control Systems with Aperiodic Sampling
Embedded and networked control systems are often required to share a limited amount of computational and transmission resources between different applications. This may lead to fluctuations of the sampling interval, due to the interaction between real-time control algorithms and task / communication scheduling protocols. This class will focus on the stability of networked control systems with time-varying sampling intervals.
You will be presented with basic concepts and recent research directions. An overview of time-delay, hybrid, discrete-time and input-output models for systems with aperiodic sampling will be given and specific analysis tools will be introduced. At last, an emerging topic concerning the design of stabilising state-dependent sampling laws will be discussed.
C9: Integro-Differential Algebra for Parameter Estimation
The parameter estimation problem consists in recovering the values of the parameters of a system (governed for example by a set of nonlinear differential equations) from an incomplete observation of its state. A possible approach consists in computing the so-called input/output equations (using elimination algorithms in the context of differential algebra) which only involve the parameters and the measured signals (i.e. the inputs and outputs). The (numerical) precision of the parameter estimation is usually compromised when the I/O equations involve high order derivations. We will present ongoing research consisting in reducing the order of derivations by integrating the I/O equations.
The Simulation of an Electric Vehicle will be dedicated to the control of the L2EP research laboratory’s electric vehicle. The simulation of the EV of the L2EP research laboratory will be developed using EMR and Matlab-Simulink. You will then have the opportunity to discover experimental facilities and the real vehicle.
Designed to facilitate your admission to Doctoral programmes in France, the Scientific Research Project will require both personal and teamwork. It will include 6 sessions of tutorials or free discussion with researchers and PhD students from the CRIStAL and L2EP laboratories. The project will involve the analysis of articles, bibliographical summaries, a presentation of state-of-the-art trends in the chosen research topic, simulations when appropriate and a final project defence in front of the scientific board.
You will have every opportunity to contact teachers/researchers with a view to supervision of a future PhD programme; assistance will be provided in maintaining contact in order to finalise the project up to enrolment in the doctoral programme.
1. High-Integration Smart Innovative Electric Drive
Prof. Betty LEMAIRE-SEMAIL, Head of the Integrated Smart Energy Converter (CE2I) cluster.
2. Towards Smart Factory for Industry 4.0: Challenges, Design Principles and Technical Approaches
Dr. Jean-Marc VANNOBEL, INCASE European Interregional Project.
3. Power Advanced N-Level Digital Architecture for Simulation of Electrified Vehicles (PANDA)
Prof. Alain BOUSCAYROL, H2020 European project.
4. Campus of University with Mobility Based on Innovation and Carbon Neutral (CUMIN)
Prof. Eric HITTINGER, Interdisciplinary programme of the University of Lille.
5. Scientific Research Clusters and Economic Hubs in Hauts-de-France (Northern France Region)
Transverse conference correlated to the domains of France Excellence Summer Schools students and coordinated by competent clusters and hubs.
You will make the acquaintance of automotive French manufacturers through a meeting of high technological interest with Dr. CHEN Keyu from the innovative company Valeo (automotive supplier and partner to automakers worldwide).
You will visit Xperium, the experimental demonstrations centre of new research development at the University of Lille. You will get to see the most relevant demonstrations in the ACES 2020 topics.
NB: Minor modifications to the scientific programme may occur.