Digital Transformation#1: MATLAB e Simulink per supportare Modellazione & Simulazione dei sistemi in Industry 4.0 - Aldo Caraceto Application ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Digital Transformation#1: MATLAB e Simulink per supportare Modellazione & Simulazione dei sistemi in Industry 4.0 Aldo Caraceto Application Engineering Group © 2020 The MathWorks, Inc. 1
Modellazione & Simulazione dei sistemi in Industry 4.0 - Agenda Orario Titolo della Sessione 09:00 Digital Transformation Industriale: opportunità, sfide e soluzioni per lo sviluppo prodotto 09:10 MATLAB & Simulink per Modellazione & Simulazione di sistemi virtuali in Industry 4.0 09:20 Esempi di sviluppo di sistemi di controllo per apparati meccatronici • Modellazione di plant a partire da equazioni/multifisici • Progettazione di controlli continui e a logiche ad eventi discreti 10:05 Q&A 2
Digital Transformation: Learnings from studies and programs ▪ Customers want increasingly individualized products. “sample-size 1” ▪ Autonomous machines which do not require costly programming to meet new requirements. “Smart products” ▪ Intelligent products that collect data to optimize processes and develop new products ▪ Competitive threats from big players offering internet-related and IT products and services. ▪ Opportunities for innovative business models and services. Particularly for SME’s. “Servitization” 5
Digital Transformation of the Industry: is everywhere – Higher flexibility given by small batches production with the economies of scale – Higher speed from prototyping to mass production using innovative technologies – Increased productivity thanks to lower set-up time and reduced downtimes – Improved quality and scrap reduction thanks to real time production monitoring through sensors – Higher competitiveness of products thanks to additional functionalities enabled by Internet Of Things 6
The birth of New Challenges designing multi-domain, smart, connected systems ▪ Too slow because process is serial and fragmented, many iterations are needed ▪ Components over- under dimensioned ▪ System Performance issues detected too late in integration phase ▪ Need risky/expensive physical machine testing ▪ Tuning and commissioning is lengthy ▪ Need to design more intelligent and connected systems ▪ Need customizable systems without extensive re-design and programming 7
The birth of New Challenges designing multi-domain, smart, connected systems ▪ Too slow because process is serial and fragmented, many iterations are needed ▪ Components over- under dimensioned ▪ System Performance issues detected too late in integration phase ▪ Need risky/expensive physical machine testing ▪ Tuning and commissioning is lengthy ▪ Need to design more intelligent and connected systems ▪ Need customizable systems without extensive re-design and programming 8
Key Enabler: Mechatronics Combination of mechanical-, computer-, telecommunications-, systems- and control engineering with electronics 10
Key Enabler: Cyber Physical System ▪ A mechanism controlled by computer-based algorithms, tightly integrated with the Internet ▪ Process control based on embedded systems ▪ Examples: smart grid, autonomous automobile, medical monitoring, robotics 11
Key Enabler: Digital Twin ▪ A digital replica of physical assets, that can be used for various purposes. ▪ Integrate machine learning and analytics to create living digital simulation models that continuously learn and update themselves 12
MATLAB e Simulink per Modellazione & Simulazione di sistemi virtuali in Industry 4.0 13
Modellazione e Simulazione in Industry 4.0 Industry 4.0 Levers ▪ «Rapid Experimentation and Value Drivers Simulation» è una leva fondamentale per ridurre Time-to-Market ▪ La simulazione è solo un elemento; altri devono essere resi disponibili per il massimo risultato possibile. ▪ Molteplici driver determinano il successo di un progetto: es. «Time-to-Market», senza «Quality»? “Industry 4.0. How to navigate digitalization of the manufacturing sector” – McKinsey Digital, 2016 14
Modellazione e Simulazione in Industry 4.0 15
Modellazione e Simulazione in Industry 4.0 “Product Life Cycle Risk Management”, Jan Machac, Frantisek Steiner and Jiri Tupa, 2017 16
Model-Based Design RESEARCH REQUIREMENTS What if you were able to verify your system’s behavior through the entire design process? DESIGN Environment Models TEST & VERIFICATION Physical Plant Models Control / Supervisory Logic Models IMPLEMENTATION C, C++ IEC HDL MCU DSP PLC FPGA ASIC INTEGRATION / COMMISSIONING 17
Model-Based Design RESEARCH REQUIREMENTS Step 1: Desktop Simulation DESIGN ▪ Prototype new functionality and Environment Models combine with existing code TEST & VERIFICATION Physical Plant Models ▪ Perform automated system tests Control / Supervisory Logic Models that would not be feasible outside of simulation ▪ Optimize parameters (software, IMPLEMENTATION mechanics, hydraulics, etc.) C, C++ IEC HDL MCU DSP PLC FPGA ASIC INTEGRATION / COMMISSIONING 18
Model-Based Design RESEARCH REQUIREMENTS Step 2: Hardware in the Loop DESIGN ▪ Emulate the behavior of the physical Environment Models system in real-time TEST & VERIFICATION Physical Plant Models ▪ Connect the virtual plant to your Control / Supervisory Logic Models PLC or industrial PC IMPLEMENTATION C, C++ IEC HDL MCU DSP PLC FPGA ASIC INTEGRATION / COMMISSIONING 19
Model-Based Design RESEARCH REQUIREMENTS Step 3: Production Use DESIGN ▪ Design and test hardware Environment Models independent functionality TEST & VERIFICATION Physical Plant Models Control / Supervisory Logic Models IMPLEMENTATION C, C++ IEC HDL MCU DSP PLC FPGA ASIC INTEGRATION / COMMISSIONING 20
Simulink to support Model-Based Design 21
Esempi di sviluppo di Sistemi di Controllo per Apparati Meccatronici 22
Modeling Physical Systems with MathWorks Products Modeling Approaches First Principles Modeling Data-Driven Modeling Physical Networks Statistical Methods Programming (Statistics & Machine System (MATLAB, C) (Simscape Products) Learning Toolbx) Identification (System Identification Block Diagram Toolbox) (Simulink) Modeling Language Neural Networks (Simscape language) (Deep Learning Toolbox) Symbolic Methods Parameter Tuning (Symbolic Math (Simulink Design Optimization) Toolbox) 23
Modellazione di Sistemi a partire da Misurazioni sul Campo 24
Modeling Approaches First Principles Data-Driven Physical Networks Statistical Methods Programming System Identification Measured Block Diagram Model Modeling Language Neural Networks Symbolic Methods ▪ Purpose: Model an existing design (real or virtual) ▪ Requirements: – Relevant set of measured data is available – Design and physical parameters will not be changed 25
Modeling Approaches: System Identification System Measured input + error - Model Minimize 26
Estimation and Validation Go Together ▪ A large enough model can reproduce a measured output arbitrarily well. We must verify that model is relevant for other data – data that was not used for estimation, but was collected for the same system. Error Validation data Estimation data Number of parameters 27
Example: Indentification of a Linear System 28
Using System Identification Toolbox ▪ Use two data sets for estimation and validation ▪ Estimate a variety of models: ▪ Linear models – Transfer functions, state space, etc. ▪ Nonlinear models – ARX- type and Hammerstein- Wiener ▪ Nonparametric – Impulse and frequency response ▪ Grey-Box models – Models with known structure but unknown parameters 30
Using an Estimated Model in Simulink ▪ Use models estimated in System Identification Toolbox directly in a Simulink model ▪ Blocks available for source, sink and models 31
Modellazione di Sistemi Multifisici 32
Physical Modeling First Principles Data-Driven Physical Networks Programming Block Diagram Modeling Language Symbolic Methods ▪ Purpose: Explore design or physical parameters ▪ Requirements: – Physics of system are well-known – Component-level models exist or can be created 33
Motivation Controller Plant Controller Plant Controller Plant 34
Optimize System-Level Performance Actuators Sensors u + s1 s2 System y s3 Controller Plant Simulating plant and controller in one environment allows you to optimize system-level performance – Automate tuning using optimization algorithms – Accelerate process using parallel computing 35
Detect Integration Issues Earlier Actuators Sensors + uSystem s1 s2 System y s3 System Specification Model Controller Plant Controls engineers and domain specialists can work together to detect integration issues in simulation – Convert models to C code for HIL tests – Share with internal users with fewer licenses – Share with external users while protecting IP 36
Build Accurate Models Quickly ▪ Simply connect the FSpring = k Spring *(xMass ) components you need dxMass FDamper = bDamper *( ) dt d2 xMass −FSpring − FDamper ▪ The more complex the = dt 2 mMass system, the more value you get from Simscape ▪ Resulting model is intuitive, easy to modify, and easy for others Input/Output Block Diagram Simscape to understand 37
Build Accurate Models Quickly Fortran, C++ Domain Expertise Coding Effort System MATLAB, Simulink System Specification Domain Expertise Coding Effort Model Simscape Domain Exp. Coding Eff. Get from specification to model even faster Spend more time designing, less time modeling 38
Example: Robot Arm and Conveyer Belts 39
Example: Modeling Contact Force Between Two Solids 40
Example: Modeling a Three-Phase Inverter 41
Simscape Products ▪ Simscape platform – Foundation libraries in many domains – Language for defining custom blocks Isothermal ▪ Extension of MATLAB Liquid – Simulation engine and custom diagnostics ▪ Simscape add-on libraries – Extend foundation domains with components, effects, parameterizations – Multibody simulation – Editing Mode permits use of add-ons with Simscape license only – Models can be converted to C code 42
Optimize Your Entire Engineering System Multidomain Simscape Domain Exp. Coding Eff. Plant Model Multibody Domain Exp. Coding Eff. Mechanical Driveline Domain Exp. Coding Eff. Hydraulic Fluids Domain Exp. Coding Eff. Electronic Power Systems Electrical Domain Exp. Coding Eff. Simulate the entire system in a single environment – Does not require learning multiple tools or co-simulation 43
Simscape Add-on Libraries ▪ Simscape Electrical – Electronics, mechatronics, and power systems ▪ Simscape Driveline – Gears, leadscrew, clutches, tires, engines ▪ Simscape Multibody – Multibody systems: joints, bodies, frames ▪ Simscape Fluids – Pumps, actuators, pipelines, valves, tanks 44
Modellazione di Sistemi a partire da Equazioni 45
Equation–based Modeling First Principles Data-Driven Programming Block Diagram Modeling Language Symbolic Methods ▪ Purpose: Explore design or physical parameters ▪ Requirements: – Physics of system are well-known – System-level equations can be derived and implemented 46
Customize and Extend Simscape Libraries for a Custom DC Motor 47
Dal Disegno Meccanico alla Regolazione dell’unità di Motion Control per un sistema Meccatronico 48
Optimizing Time-Domain Responses of a Simulink Model ▪ Specify desired behavior by either graphically shaping the desired response or typing in numeric values ▪ Add design requirements without adding blocks to the model ▪ Use multiple objectives and constraints simultaneously ▪ Monitor all plots in one window ▪ Perform optimization faster with Parallel Computing Toolbox and Fast Restart 49
Progettazione del sistema di regolazione del tiro per film plastici Closed-loop model Outputs Control logic 50
What is Stateflow? Extend Simulink with state charts and flow graphs Design supervisory control, scheduling, and mode logic Model state discontinuities and instantaneous events 51
How Does Stateflow Work with Simulink? Simulink excels at Stateflow excels at continuous changes in instantaneous changes in dynamic systems. dynamic systems. Real-world systems have to respond to both continuous and instantaneous changes. suspension dynamics gear changes manufacturing robot propulsion system operation modes liftoff stages Use both Simulink and Stateflow so that you can use the right tool for the right job. 52
Key Takeaways ▪ MATLAB e Simulink forniscono un ambiente integrato per sviluppare progetti innovativi all’interno del paradigma di Industry 4.0 ▪ MATLAB e Simulink supportano efficacemente la modellazione & simulazione di sistemi complessi, fornendo: 1. strumenti per intercettare eventuali errori nelle fasi preliminari 2. funzionalità per limitarne l’introduzione accidentale. ▪ MATLAB e Simulink garantiscono un supporto completo e un flusso di lavoro ininterrotto all’interno del Model-Based Design 53
MATLAB e Simulink per supportare Modellazione & Simulazione dei sistemi in Industry 4.0 54
You can also read