Outline. 1. Introduction. 2. Model Based Control Design. 3. Software Development for Flight Control Algorithms. 4. Lessons Learned / Impacts

Transcription

1 Assisted and Auto Mode RSw second lane active [CSw healthy] CSw first input active Rx active Autopilot Control [Rx healthy] [CSw fail] Loss of servo [RSw healthy] [CSw healthy] Manual Control [Rx fail] CSw first input active Rx active RSw first lane active [CSw fail] Loss of servo [Rx healthy] [RSw fail] [CSw healthy] CSw second input active Rx active Autopilot Control [Rx healthy] [CSw fail] Loss of servo [Rx fail] Loss of aircraft [CSw healthy] CSw first input active Rx active Manual Control [Rx healthy] RSw second lane active [Rx fail] [CSw fail] Loss of servo [RSw healthy] [Rx fail] Loss of aircraft [CSw healthy] CSw second input active Rx active Autopilot Control [RSw fail] [Rx healthy] [CSw fail] Loss of servo [Rx fail] Loss of aircraft Lehrstuhl für Flugsystemdynamik Technische Universität München.4 Loss of Aircraft PSw fail CSw -N fail Bat fail Bat fail RSw fail PSw fail Rx fail Rx fail Markus Hornauer, Falko Schuck und Florian Holzapfel

2 Outline. Introduction. Model Based Control Design 3. Software Development for Flight Control Algorithms 4. Lessons Learned / Impacts 5. Summary

3 Introduction Motivation 3 Continuity of development process: From requirements capturing to verification on target system High level of reuse to increase efficiency Better requirements through pre-simulation Challenge: Software engineers are usually no control experts, flight mechanics / flight control engineers are usually no software engineers!

4 Introduction TUM FSD - DA4M-NG Flying test bed for the Free State of Bavaria 4

5 Introduction Project Example FLY SMART 5 Development of an certifiable autopilot for CS-3 Aircraft using modular avionic components Optimization of Model-Based Software Development Methods and Techniques Formal and Integral Process Definition for Model-Based Software Development System and Software Synthesis, Capture and Validation Safety Assessment Functional Hazard Analysis Safety Derivation (Preliminary) System Safety Assessment (FTA, FMEA, Simulation) System and Software Development Hardware and Software Architecture Interfaces and Redundancy Management Monitoring and Voting Autopilot / Flight Control Algorithms Development System and Software Integration, Test and Verification Requirement Based Testing Hardware in the Loop (HIL) System and Aircraft Integration Flight Testing Analysis and Verification based on Flight Test Data

6 Outline. Introduction. Model Based Control Design 3. Software Development for Flight Control Algorithms 4. Lessons Learned / Impacts 5. Summary

7 Model Based Control Design Introduction 7.4. Flight Control Design is not a single point issue but spans a large operation envelope / domain.5.5 Plant analysis Controller Design Gain Design Controller assessment needs to be automated Non-compliance needs to be detected automatically Reports and evaluations need to be organized automatically

8 Software Development for Flight Control Algorithms Control Algorithm Development Life Cycle System Process (Flight Control System) System Allocated to Flight Control Algorithm Safety Assessment 8 Verification of Plant Model Verification of Reduced Order System Plant Modelling Process (Failure, Disturbance and Uncertainty Models) (Systems: Sensors and Actuation) LOS Model Extraction Process (Trimming, Linearization, Reduction of Order) System Analysis Process High Fidelity Plant Model Reduced Order Plant Model Analysis Results Design Validation Control Structure Design Process Gain Design Process Gain Parameters (and Ranges) Controller Layout Transfer Function / Matrix (Inner Loops) Design Model Loop Back from Software Process (System Engineering Process) Automated Assessment on reduced Model (classical: Linear) Automated Assessment on High Fidelity Model Assessment Results Assessment Results (Proof of Stability, etc.) Artifacts to Software Process

9 Model Based Control Design Trim and Linearize 9 Trim applications: Computation of performance data: - Flight envelope parameter configurable (Altitude, Velocity, Center of gravity, flap configuration, ). - Automated trim point calculation for each flight condition. - Output of all important trim result data (trim success, used trim strategy, number of iterations, ) for evaluation of the trim results. x S a b x h z d T p V h g Y& a b Haddad Constraint p, q, r, F, Q [ ] f, a, b, g HD V K V g Y& K, K Y& V a b p q r F Q Y l m h x h z d T x = p() = x S () = x S () = Haddad = Haddad = Haddad = Haddad = Haddad u egal egal egal = p() = x S (3) = x S (4) = x S (5) = x S (6) Nonlinear Equation solver x p r f, S f ( x, u) x& h( x, u) y! x& V& = r() a& = r() b& = r(3) p& = r(4) q& = r(5) r& = r(6) F& = (auto) Q& = (auto) Y& = p(4)(auto) l& egal m& egal h& V sin g F Y B r V& a& b& p& q& r& r f x, p S Numerical Linearization: - Free configuration of the desired model parameters (States, Inputs, outputs), which shall be extracted from the nonlinear model. - For each trim point, a linearized model of the nonlinear equation system is generated x u dx du x& y linsys linsys f ( x dx, u du) dx& h( x dx, u Ax Cx linsys linsys u Du du) y linsys linsys dy Numerical Linearization d& x dy

10 Model Based Control Design Example: DA4 Longitudinal Controller Design with MATLAB Linearization Phase (deg) - - Config Flaps = Gear = Config Flaps = 7 Gear = -3 - Phase (deg) Magnitude (db) Config 3 Flaps = 4 Gear = Config 4 Flaps = Gear = -3 - Frequency (rad/s) Frequency (rad/s) Phase (deg) Magnitude (db) Frequency (rad/s) - Frequency (rad/s) Config 5 Flaps = 7 Gear = Extended Plant Dynamics FP Frequency (rad/s) Bode - All borders combined ds - - Config Config Config Config Config Config Flaps = Flaps = 7 3 Flaps = 4 4 Flaps = 5 Flaps = 7 6 Flaps = Gear = Gear = Gear = Gear = Gear = Gear = Controller Design qcom-csas Config 6 Flaps = 4 Gear = -3 - Frequency (rad/s) Magnitude (db) Automatic trim and linearization for the whole flight envelope (h Ma) as well as all configurations (gear, flaps) Plant analysis Magnitude (db) Trim qcmd,s Eq F Prefilter V/g cos g sinf tanf Frequency (rad/s) qcmd Kds,q Stick-toCommand Shape - G (s) hcmd Controller hcmd, el Delay after FCC etc. qmeas H. Peak response Amplitude Rise time HQ Sensor.4 HQ requirements BW Aircraft qturn,c Delay to FCC C* hmec Plant Turn compensation CAP GA(s) Actuator Delay assumption Time (seconds) Actuator assumption Automatic HQ analysis (Dropback, CAP, C*, Neal Smith, Bandwidth,...) based on w and T (z =.7) for desired short period dynamics q

11 Model Based Control Design Example: DA4 Longitudinal Controller Design with MATLAB 4 L LEVE L LEVE L LEVE L LEVE Level L LEVE 3 - n/ a (g/rad) Category A & C flight phases (CAP - damping ratio boundaries) LEVEL LEVEL Level Level / CAP 4 qmax/qss CAP. 3.6 Closed Loop Resonance [db] Clas. I,II-C,IV (rad/sec) Clas. II-L,III SP Category C flight phases CAP boundaries Note: The boundaries for values n/a outside the range show n are defined by straight-line extensions LEVEL. Clas. II-L,III wn LEVEL. Clas. I,II-C,IV 6 Level Lag - Level LEVEL 3 - Lead Pilot Phase Compensation [deg] at Omega = 3.5 rad/s 8 zsp Automatic controller stability and Handling Qualities analysis featuring LOES (low order system generation) for all trim conditions and configurations (~ points) Automatic Gain design for the controller to comply the requirements for all points..5 Level / db/qss Controller implementation 3 Extended Plant Dynamics FP Closed loop pull & release Open loop pull & release Closed loop doublet Open loop doublet q [ /s] qcom-csas ds qcmd,s qcmd Eq Kds,q F Stick-toCommand Shape Prefilter - G (s) hcmd Controller hcmd, el Delay after FCC GA(s) hmec Aircraft q Actuator Plant qturn,c q [ /s] V/g cos g sinf tanf qmeas H Turn compensation Delay to FCC - Sensor Time [s] 6 8 Time [s] Gibson s dropback criterion /q db db - db.5 Exc Acc q peak/qss - db db db db db db db 6 db Mismatch - Phase - db 3 db 5 ess iv e (P IO p os si ble) epta ble.5 db Phase [ ] peak ss 5 Frequenzgangbetrag des ORK [db] Magnitude [db] Mismatch - Amplitude Frequency [rad/s] Phase des ORK [ ] - -5

12 g-tracking / g ( ) Model Based Control Design Example: DA4 Longitudinal Controller Design with MATLAB Tracking - Josef-Niederl-Run Con-act--K-stick-.74-MC-grad-.mat Error-RMS:.54 Error-rate-RMS: Finetracking-RMS: Magnitude (db) Time (s) - - Y C / Y CL / w co,ac =.89 rad/s Y C Y C L Magnitude (db) Y P x Y C w CO =.6993 w CO-A = Comp. Pursuit Magnitude (db) 3 Y P - Pilotgain@co:.79 Y P Y P -pursuit Microsoft Excel -3 - Frequency (rad/s) - - Frequency (rad/s) - - Frequency (rad/s) Kohärenz Kohärenz Kohärenz Kohärenz.6 Kohärenz.6.4 Kohärenz Automatic simulation result analysis Deflection [-] Frequency (rad/s) Frequency (rad/s) Frequency (rad/s) Stick - : RMS:.9797 Mittelw.: e-5 Force-RMS: Speed-RMS: d S.4 d Kraft 5. -Stick Time (s) Force [N] Export: Analysis results. Automatic report generator. Automatic compliance matrix. Simulator flight testing with pilots Automatic data analysis of the simulation results with Matlab Export of the analysis results to Excel

13 Outline. Introduction. Model Based Control Design 3. Software Development for Flight Control Algorithms 4. Lessons Learned / Impacts 5. Summary

14 Software Development for Flight Control Algorithms Architecture of a Flight Control System 4 Flight Control Computer & Flight Control Algorithms Controller Algorithms (Design Model) (DO-33 Example 5) Input- / Output-Conditioning, Robustness and Interface Framework (Design Model) (DO-33 Example ) Real Time Operating System (DO-78C COTS) BSP, Drivers, Runtime- and Interface Framework ANSI C (DO-78C) Electronic Hardware COTS and Special Built (no FPGA) Sensors Actuators

15 Software Development for Flight Control Algorithms DO-33 MB..6.3 Examples of Model Usage Process that generates the life-cycle data MB Example MB Example MB Example 3 MB Example 4 (See Note ) MB Example 5 System Requirement and System Design Process allocated to software from which the Model is developed from which the Model is developed from which the Model is developed from which the Model is developed Design Model Software Requirement and Software Design Process from which the Model is developed Specification Model (See Note ) Specification Model Design Model Design Model Formalized Design Textual description (See Note 3) Software Coding Process Source Code Source Code Source Code Source Code Source Code

16 Software Development for Flight Control Algorithms Software Module Life Cycles 6 DO-78C System allocated to Software DO-33 SW High Level, SW Pre-Design and SW Module Interfaces SW Design (SW Low Level and Architecture) Design Model Design Model COTS Manual Coding Automatic Code Generation Automatic Code Generation Software Integration Software Hardware Integration At least four SW modules: one PSAC several SDPs and SVPs RTOS BSP, Drivers and Data Handling Framework Signal Conditioning and Data I/O Framework (Example ) Flight Control Application (Example 5)

17 Software Development for Flight Control Algorithms Equipment Development Life Cycle 7 System Allocated to Software Software Components: Component W.. R D C - I Component X.. R I Component Y.. R C I Component Z.. R C I C I R D C - I Legend: R C Coding D Design I Integration Software Product Source: DO-78 B/C Figure 3- Example of a Software Project Using Four Different Development Sequences

18 Software Development for Flight Control Algorithms Equipment Development Life Cycle Pre-Development Step 8 System Engineering Process (Flight Control System) Customer Specification Safety Safety Assessment SW / HW Planning Process Equipment Specification Design Model Derived High Level PDR (Preliminary Design Review) SW- and HW- Validation Preliminary SW/HW Design / SW/HW-Architecture / Coding / Preliminary Testing (prelim. Test Cases) SW / HW Process Incremental Development Process Demonstrator (Prototype) Software Document High Level Derived Low Level Hardware Document High Level Preliminary DO-6F Environmental Testing System RIG Testing Flight Testing Preliminary SW Documents Preliminary HW Documents Next Slide Development Workflow Generation of Artifacts Subject to Safety Assessment

19 Software Development for Flight Control Algorithms Equipment Development Life Cycle Core Development Step 9 Previous Slide CDR (Critical Design Review) Software / (System) Review SW Design Review Design Model Review Model Coverage Analysis Code Review SW / (System) Update SW Design Process SW Coding, Integration and Pre-Testing SW / (System) Document SW / Model Design Description Model Coverage Analysis Data Source Code Simulation Results CDR (Critical Design Review) HW Review HW Design Review HW Update HW Design Process HW Implementation HW Testing HW Document HW Conceptual Design HW Detailed Design Production Data HW / SW Integration Testing (Req. Based Functional Testing Structural Coverage Analysis) Test Results Additional Artifacts: Qualification Testing (Functional and Environmental Testing) Structural Coverage Analysis Data Documentation of Performed Reviews Series Production Final Product Development Workflow Generation of Artifacts

20 Software Development for Flight Control Algorithms Tool Chain Structure and Workflow for Power PC Generation and deployment of source and object code to target environment Strategy on qualifiable verification tools All tools in use in aerospace applications Model Development and Verification Embedded MATLAB, Simulink, Stateflow Simulink Model Advisor, Report Generator and Model Coverage Source Code Development and Verification Simulink Embedded Coder Simulink Code Inspector PolySpace VerOCode and VeroSource (structural code coverage) Startup, Runtime, Robustness and Interface Framework ANSI C <hand coded> Debugging and Target Testing Lauterbach TRACE 3 Debugger (debugging and PIL) Bernecker + Rainer (B&R) (hardware in the loop test bed) + Structural Coverage on Target Simulink Code Inspector Integrated Build In-Circuit Debugging Embedded Coder Source Code Object Code Target Deployment Model Link Code Reference Integrated Debugging Hardware & Processor In the Loop (HIL / PIL)

21 Software Development for Flight Control Algorithms Flight Control Computer Hardware (FCC) General Description Development of a certifiable avionics platform for research and demonstration of algorithms in a closed to product environment together with industry partners. The focus is on a scalable and modular system which can be used in different applications. High Performance Power PC Architecture: 67 MHz performance (available with separate graphic controller) Double precision floating point unit High speed interface to I/O module Cortex M I/O Interface Module Featuring two I/O processors Providing up to UART interfaces, 4 ARINC 85 Interfaces, ARINC 49 Interfaces, 4 bit analog inputs and multiple discrete I/Os Extendable Platform Configuration Single string for simple applications Redundant system with high speed interface between two lanes Fail operational triplex system with internal redundancy Certification Aspects All used processors and complex elements already in use in aerospace systems (COTS) System was design to ensure easy testability

22 Software Development for Flight Control Algorithms Testing in Final Target Environment Processor In The Loop Test Bench DA4 Total System Simulation HIL Diamond DSIM Flight Training Device Actuator Test Bench High Fidelity Aircraft Model on Automation PC B&R Automation Debugger Integration in Simulink Traceability between Simulink and Trace3 Aircraft Flight Test Installation Flight Control Computers (FCCs) Target testing is required for generated and handwritten code (make it efficient!) PIL is a proven method for testing equivalence to Simulink (reuse of test cases) HIL testing is important to develop good and robust control algorithms (e.g. latencies) based testing of the final product is a fundamental idea of DO-78 B/C

23 Outline. Introduction. Model Based Control Design 3. Software Development for Flight Control Algorithms 4. Lessons Learned / Impacts 5. Summary

24 Lessons Learned / Impacts 4 Estimation of latencies Software Modeling Guidelines Definition of Software Parameter Data Items Availability / Correctness of Parameters (especially plant parameters) Testing according to software methods Test Cases for inner loops

25 Outline. Introduction. Model Based Control Design 3. Software Development for Flight Control Algorithms 4. Lessons Learned / Impacts 5. Summary

26 Summary 6 MBD helps engineers to keep focus on technical task Simulation / analysis based derivation of requirements: basis for consistent requirements and powerful method for early validation through simulation Clean workflows are required! Continuity and automation opens opportunities for SMEs to enter avionics market

27 Thank you for your attention Contact: Technische Universität München