Code Generation Optimisation for Fire Shadow Image Processing

Andy Sherriff, MBDA

Traditionally, MBDA have spent vast amounts of time and money developing algorithms and specifying these algorithms for implementation within onboard missile computers. On Fire Shadow, MBDA have pioneered a rapid prototyping approach to reduce costs and timescales, and have employed Simulink® as the environment to develop the algorithms and then automatically generate the real-time C code for inclusion on the embedded target processor. This process was successfully piloted on the Fire Shadow Demonstration Phase, where MBDA achieved concept to flight trial within 15 months.

Now MBDA is continuing the rapid prototyping process in the Fire Shadow Assessment Phase, aimed at the first guided firing in 2010. Simulink provides an environment within which models can be easily developed, assessed, and exchanged within an interdisciplinary team. The quality and consistency of the models is controlled by the MBDA Code of Practice which utilises the Simulink Model Advisor to provide automatic checking of models and C code generation settings. Working from a baseline of Release 2007b of MATLAB®, the evolution of the Fire Shadow Image Processing has required a series of model optimisations in order to produce efficient C code for deployment on the target onboard processor. The presentation will highlight the special structuring of the algorithms, which has enabled the design to achieve significant improvements to meet required image frame rates. Many of these features are now standard within later MATLAB versions.

The Use of Model-Based Design in a Formula 1 Team

David Roberts, Renault

Formula 1 is a fast-paced design environment where innovation and optimisation of design solutions within a highly regulated framework are critical to success. Model-Based Design is used increasingly in many aspects of both the design of the car and the analysis of its performance.

This session will survey the areas where Model-Based Design is used in the Renault Formula 1 team, and in particular where MathWorks tools are used to model, optimise, and analyse system performance. Finally, some preliminary results of a study to assess how the suite of MathWorks physical modelling tools could be used to further improve performance will be discussed.

A Collaborative Approach to Three-Phase Motor Controller Development, Targeting DSP and FPGA Architectures

Paul Berry, Emmet Connon, and John Hanna, Thales Belfast

Thales in Belfast has over 50 years' experience in the design, development, and manufacture of lightweight, high-volume missile systems and has been using MathWorks tools to develop missile control systems for over 15 years. However, translating a concept into a workable implementation has never been a simple or short-term task, often requiring several iterations.

This presentation will highlight the advantages of utilising code generation technology for flexible implementation of designs. A three-phase motor controller for the Lightweight Multi-role Missile was developed and refined in Simulink and automatically generated in C for implementation on a DSP to provide rapid proof of concept. As the design matured, HDL autocode technology was harnessed to target the controller for an FPGA. Simulink HDL Coder™ enabled systems and electronics engineers to work together through an integrated design environment. This allowed modifications and bug fixes to the HDL implementation to be introduced and tested efficiently. Accelerated verification and validation of the HDL code was made possible using the fixed-point Simulink algorithm to provide test data to prove the validity of the hardware implementation.

Introducing Model-Based Design and Automatic Code Generation for a New Type of HVDC System

Patrick Mattis and Anthony Totterdell, AREVA

This session describes the introduction of a model-based approach for an HVDC VSC development project at AREVA, in an electrical transmission and distribution company. The central issues discussed are technical and organisational. The technical issues involve creating and maintaining the most suitable model structure for digitising, partitioning, and automatically generating code. The organisational issues revolve around changing work practices and garnering acceptance that Model-Based Design provides enough advantages to blur traditional departmental lines.

The introduction of such an approach is more difficult from an organisational point of view than a technical one and needs to be continuously managed. However, applying Model-Based Design can provide large commercial advantages. Initial testing and prototyping show that significant improvement in productivity with respect to model verification and automatic code generation are possible.

Using Physical Modelling Tools and the Simscape Language to Support System-Level Design

Rick Hyde, The MathWorks

The combined demand for more complex systems and shorter development cycles drives the need for continuous improvement of tools and processes. For example, many companies have moved to using Simulink and Real-Time Workshop® to develop algorithms, thus reducing development times and providing a common environment that both the software and algorithm designers can use to communicate. There is a similar potential for MathWorks physical modelling tools to provide a means for system engineers to work and communicate with control, mechanical, and electronics engineers. Often electronic and mechanical subsystems are designed from the bottom up, the design being an evolution from a previous one. This approach, however, does not support good system-level design where new disruptive technologies are fast becoming available. An example is the recent availability of high-torque electric actuators, which can now be used in place of pneumatic and hydraulic actuation for some applications.

The presentation will be illustrated by two examples. The first will use the Simscape™ language to show how control and analogue electronics engineers can work with a shared specification to de-risk unforeseen implementation issues. The second will explore how MathWorks physical modelling tools can be used to understand, at a system level, the implications of moving from hydraulic to electric actuation.

Development and Use of a Wind Turbine Single-Blade Model

Kelvin Hales, Vestas

This presentation will describe how Vestas has implemented a rigorous model in Simulink of a single turbine blade, identical to that in industry-standard “aeroelastic code,” to improve the visualisation of blade dynamic behaviour, to allow interactive use for “what if” investigations, and to obtain linearised models of blade dynamic response to aerodynamic disturbances.

In the wind energy industry, dynamic simulation of wind-turbine generators is typically carried out using aeroelastic code—that is, computer programs that are written in a high-level computer-programming language and that combine calculation routines for determining the aerodynamic and mechanical forces on the turbine with numerical integration routines for solving the differential equations to obtain the motions and dynamic response of the machine, including the mechanical deformations of the structure.

Using Simulink for modelling enables control engineers to use the simulations interactively, perform run-time investigations and visualisation, and extract linear models.

Design of Complex Electronic Hardware to Meet Safety-Related Requirements Using Model-Based Design

Dr. John Outram, Ultra Electronics PMES

This presentation begins with a brief overview of PMES’s background in producing complex electronic hardware (as defined by DEFSTAN 00-56) for safety-related, naval, nuclear applications. Using rigorous design processes, the nuclear industry is beginning to accommodate the use of programmable hardware, for example FPGAs. Until now, PMES has realised such by hand HDL coding and verification, using a standalone process. However, it is a time-consuming and costly process, which means there is a high penalty for any rework required.

The presentation discusses what motivated PMES to make use of Model-Based Design when updating and improving their process. The process brings the same rigour to system design as was previously applied only to code design, avoids opportunities for misinterpretation of specification documents, and helps avoid rework caused by lack of understanding of systems integration.

The presentation also summarises PMES’s experience, so far, with generation and testing of HDL using MathWorks and third-party tools.

Requirements-Led Model-Based Process to Support Automotive Control System Development

Will Suart, Jaguar Land Rover

This presentation focuses on the use of MathWorks products within Jaguar Land Rover and covers a workflow using control system modelling to verify system requirements. Using the standard V process, the presentation shows the methods and process used to enable control system design, desktop test verification methods, concept code verification, and traceability back to requirements. The presentation also covers the efficiencies of the model-based approach and reuse.

The Application of Parallel Computing to Modelling and Simulation

Jos Martin, The MathWorks

MATLAB, together with Parallel Computing Toolbox™, provides a rich environment for parallel computing that is suitable for beginners and experienced users. This environment encompasses many different areas such as parallel language constructs, parallel numeric algorithms, interaction with resource allocation environments, and low-level MPI-like functionality. This session will introduce some of the features that enable simple parallelization of simulation, modelling, and analysis codes.

Using Stateflow for Model-Based Design and Development of Aircraft Control Systems

Christopher Slack, Airbus

In modern civil aircraft, the fuel system provides more functionality than simply keeping the engines fed. It is also used to provide structural bending relief for the wings, to protect the aircraft flight envelope with centre-of-gravity control, to provide cooling for the engines and hydraulic packs, and to improve aircraft efficiency by trimming the aircraft in flight. These extra functional requirements, together with a plethora of safety requirements, can lead to complexities within the design of the fuel control system. To address them, Airbus has employed Model-Based Design, focusing on the core MathWorks toolsets of MATLAB, Simulink, and Stateflow.

This presentation will focus on the use of Stateflow and associated tools in the development of the Airbus Fuel Control System. The advantages of Model-Based Design to facilitate full validation and verification at system and aircraft level will also be discussed.

Integration of MathWorks Tools into Flight Control Laws Design and Clearance Processes: A BAE Systems Perspective

Matt Lodge, BAE Systems

The rapid development in computer technology over the past 20 years has provided engineers with ever more capable design and analysis tools. With this context in mind, this session will present a historical view of when and why BAE Systems has incorporated MathWorks tools into its flight control laws design and clearance processes.

The main body of the presentation will provide insight into the replacement of in-house developed tools with those from The MathWorks, with particular emphasis on business constraints. Also included is a brief overview of the major BAE Systems products for which flight control laws have been developed using MathWorks tools. The presentation will conclude with a view of future developments.

A Case Study of Application Development and Production Code Generation for a Telematics ECU with Full Unified Diagnostics Services

Ivan Wilson, Embed Limited

Telematics is now moving away from research laboratories and into our vehicles. Volume vehicle manufacturers are looking into fitting telematics units to every vehicle to send data back to a hub for many purposes, including stolen vehicle tracking, fleet management, remote diagnostics, intelligent traffic management, and customer services.

Embed Limited has successfully delivered an advanced telematics ECU that can be line-fitted to volume production vehicles. The application was fully developed within the Simulink environment and the production code fully automatically generated using Real-Time Workshop Embedded Coder. Because the ECU is line-fit, it had to comply with full automotive standards with respect to hardware and software. One example of this is diagnostics services that are required to enable the ECU to be configured and fixed in dealerships and garages.

This study will explain how Embed developed the telematics ECU software. Details such as creating the blocksets for onboard devices like the GPS module and Bluetooth module will be examined as will the extensive use of Simulink libraries to enable unit testing of the features. Particular attention will be given to modelling the diagnostics services as it is unusual for these to be modelled; instead they are usually hand-coded and implemented towards the end of the development cycle. Bringing the diagnostics services into the modelling environment and developing them in conjunction with the rest of the application enabled large time savings.

Embed has shown that using a model-based development process with Simulink and Stateflow has significantly increased productivity. By pushing model-based development into new areas such as unified diagnostic services (UDS), further gains can be made.

Using Simulink in Teams: New Features and Best Practices

Gavin Walker, The MathWorks

MathWorks tools are being used on ever larger and more complex projects, which typically involves larger teams and longer project durations. This brings requirements for the development and adoption of best practices in areas such as:

• Breaking down large designs into appropriately sized components
• Managing changes in large designs and efficiently performing reviews of changes
• Understanding the set of files that make up a project in order to perform configuration management

This session will illustrate some best practices that we have seen to be successful in industry, and will highlight new features and technologies that address common challenges.

A Brief History of the First Full Life-Cycle Model-Based Development at SELEX Galileo Using the MathWorks Toolset

Stuart Brady and Michael Lander, SELEX Galileo

This presentation will explore the full life cycle of a development project that utilised the the MathWorks toolset. We will discuss the required dynamics within the project team in terms of roles, responsibilities, and focus. We will also discuss the integration of the MathWorks toolset into the development environment, providing an optimum solution to a technical problem containing both algorithmic and real-time control challenges. We will provide technical detail of the development and, probably most important, the lessons learnt by the team. We will finish by describing the next steps for the organisation, having now experienced a full life-cycle Model-Based Design project.

The material will be delivered by senior engineers with first-hand experience on the project. We will not present a set of management slides, but a “from the coal face” view of real development on a real project using Model-Based Design techniques.

Fuel Cell Control for Domestic Appliances

Andrew Hazell, Ceres Power

Ceres Power’s goal is to bring solid oxide fuel cell (SOFC) technology into the home in the form of micro-CHP, and in so doing to reduce the consumer’s carbon footprint and energy costs. Turning a novel R&D concept into a fully commercialized product in a short space of time has placed particular demands on the development of the control system. Specifically, rapid concurrent engineering across the business has led to a high degree of uncertainty around software requirements, evolving control platforms, and a need for a high level of responsiveness to change. When coupled with the small team size, these factors made Model-Based Design the only viable approach to software development.

The first part of this talk outlines the background to Ceres Power’s products and how Ceres has used code generation in combination with their own custom-built tools and an Agile development process to deliver embedded software across multiple products. The second part discusses the structure of the control system and introduces some of the challenges specific to control of fuel cell–based domestic CHP.

The Control and Instrumentations Systems for the New Land Speed Record Car BLOODHOUND SSC

Dr. John P. Davis, Dragonfly Technology Ltd. and BLOODHOUND SSC

This presentation will outline the proposed Control and Data System to be fitted to the new Richard Noble–inspired BLOODHOUND SSC land speed record contender with the aim of achieving 1,000 miles per hour.

The control system will be responsible for the various propulsion systems fitted to the car, including a Euro fighter EJ200 jet engine, an 18-inch hybrid rocket, and a V12 internal combustion engine used to pump the oxidizer into the rocket. In addition, the control system will actively control wings mounted at the front and rear of the vehicle to maintain sufficient load on the wheels to ensure the car does not take off.

The control system will use multiple processors operating in real time with redundancy and safety override capability assisting driver Andy Green, current land speed record holder, by displaying data collected from many systems and allowing him to control the vehicle at all times through the regulator measured mile in each direction with a one-hour time frame.

The system is currently under development, with many of the safety subsystems being developed independently. The modular nature of Simulink, Stateflow®, and xPC Target™ software from The MathWorks allows rigorous hardware-in-the-loop testing combining real system and simulated functions prior to the first running of the car in the deserts of South Africa.

Master Class: Verification and Validation

Shane Davies, The MathWorks

Verification and validation is critical for the successful implementation of Model-Based Design on production programs. This master class will introduce MathWorks tools and concepts that are available to address different aspects of verification and validation in Model-Based Design. The class will use MathWorks tools for model verification and validation: SystemTest™, Simulink Verification and Validation™, and Simulink Design Verifier™. The products will be used to demonstrate and explain three different methods for model verification and validation:

• Performing regression testing of Simulink models
• Extending regression tests to achieve model coverage
• Proving critical design properties of a model

The session will provide a brief overview presentation, followed by demonstrations. Attendees are encouraged to engage in the demonstration with their questions and suggestions.

Master Class: Physical Modelling and the Simscape Language

David Sampson, The MathWorks

Following on from the presentation Using Physical Modelling and the Simscape Language to Support System-Level Design, this master class will cover three key topics related to modelling physical systems in Simulink:

• Creating and elaborating physical component blocks by combining basic blocks
• Combining physical domains and Simulink signals to represent control laws and force laws
• Creating custom physical component blocks based on equations using Simscape language

The master class will consist of a discussion of the key issues, illustrated by demonstrations

Master Class: Production Code Generation

Mark Walker, The MathWorks

Code generation, using Real-Time Workshop Embedded Coder™, is a core activity in Model-Based Design for production systems. This master class will demonstrate code generation techniques that users can apply to maximise their development efficiency, including how to:

• Satisfy software requirements as early as possible in the development life cycle
• Apply software detail to a Simulink model
• Get the most out of model optimisations and the code generator's optimisations
• Automate the code generation and software build processes

The session will start with a short presentation and then illustrate each activity through demonstrations.