advertisement

PENNSTATE 1 8 5 5 Project PS 5.2 Simulation and Control of Shipboard Launch and Recovery Operations PI: Asst. Prof. Joseph F. Horn Tel: (814) 865 6434 Email: joehorn@psu.edu Graduate Students: Dooyong Lee, PhD Candidate Derek Bridges, PhD Candidate 2005 RCOE Program Review May 3, 2005 PENNSTATE 1 8 5 5 Background / Problem Statement • The shipboard launch and recovery task is one of the most challenging, training intensive, and dangerous of all rotorcraft operations • The helicopter / ship dynamic interface (DI) is difficult to accurately model • Industry and government could use better tools for analyzing shipboard operations to reduce the flight test time and cost to establish safe operating envelopes • Workload requirements could be reduced using tasktailored control systems for shipboard operations Technical Barriers • Accurate models require the integration of the time-varying ship airwake and the flight dynamics of the helicopter • Currently pilot workload requirements and HQ analysis must be assessed using expensive flight tests and piloted simulation. Better engineering tools needed to reduce costs for analyzing current and future ships / aircraft. • A practical, fully autonomous or piloted assisted landing AFCS has not yet been developed, need to assess requirements and potential benefits PENNSTATE 1 8 5 5 Task Objectives: • Develop advanced simulation model of the shipboard dynamic interface • Validate the model using available test data • Use the model to develop advanced flight control systems to address workload issues in the DI Approaches: • Develop a MATLAB/SIMULINK based simulation of the H-60 based on GenHel (will facilitate model improvements and control law development) • Develop a maneuver controller to simulate pilot control during launch and recovery operations • Integrate simulation with ship airwake models, investigate relative effects of steady and timeaccurate CFD wakes, and stochastic wake models based on CFD and flight test data • Simulate UH-60 operating off LHA and validate model with JSHIP flight test data • Develop new concepts in AFCS design for shipboard operations • Develop a real-time simulation facility of shipboard operations (using DURIP funds) Expected Results: • A simulation tool for analyzing handling qualities in the DI and predicting safe landing envelopes • A methodology for designing a task-tailored AFCS for shipboard operations • A conceptual design of an autonomous landing systems and assessment of the system requirements for such a system (possible UAV applications) PENNSTATE PSU DI Simulation Program 1 8 5 5 • Developed “tunable” pilot model for different levels of tracking tolerance • Integrated CFD solutions of ship airwake with non-linear flight dynamics model • Demonstrated using UH-60A / LHA combination, same as JSHIP flight test program • Validated model with flight test data from JSHIP program • Evaluate task tailored control laws Human pilot model (Optimal control model) Time-accurate ship airwake from CFD Real-time simulation Matlab based DI simulation program (based on GENHEL) Stochastic airwake model Task-tailored control law design (using CONDUIT) Validation with flight test data (from JSHIP program) PENNSTATE Stochastic Ship Airwake Modeling 1 8 5 5 • A method for extracting equivalent airwake disturbances from flight test data (or high order simulation model) has been developed Method is similar to that used for turbulence models developed at NASA Ames (Ref. Labows and Tischler et al, MacFarland - SORBET Model) Filters are derived to simulate the spectral properties of the airwake, can compare to traditional turbulence models (e.g. von Karman, Dryden) Spectral filters are based on von Karman model, and modified to fit the desired forms of spectral characteristics • Stochastic airwake model can be readily used for flight control optimization Designed to fit the spectral properties of the airwake Stochastic airwake model White noise Linear filter + Pilot stick inputs + + Helicopter Dynamics Optimized to reject disturbances SAS PENNSTATE Stochastic Ship Airwake Modeling PSD of vertical gust component, (ft/sec)2/(rad/sec) Sample Results for Vertical Component, 0° WOD 1 8 5 5 “Best Fit” Spectral Filter Lw =10.7156 ft, sw = 4.81 ft/sec Extracted from simulation with full time-varying airwake 2 Lw L L 1 1.5913 w s 0.9173 w s V V V H s 2 3 L L L 1 2.5601 w s 3.3169 w s 1.8985 w s V V V von Karman Turbulence Model Lw = 37.8667 ft, sw = 2.8067 ft/sec 2 Lw Lw Lw 4s w 1 2.7478 s 0.3398 s V V V H s 2 3 L L L 1 2.9958 w s 1.9754 w s 0.1539 w s V V V Frequency (rad/sec) 4s w PENNSTATE Stochastic Ship Airwake Modeling 1 8 5 5 • Comparison of response with stochastic airwake model, equivalent disturbances and full time-varying airwake (spectral data averaged over five runs) Stochastic Airwake Autospectra identified Equivalent Airwake Disturbances Full Time-Varying Airwake - Input autospectrum(30 deg WOD), dB PED PED COL COL LON LON LAT LAT - Input autospectrum(0 deg WOD), dB by CIFER Frequency (rad/sec) Frequency (rad/sec) PENNSTATE Task-Tailored Control Design 1 8 5 5 • Using CONDUIT to optimize SAS gains • Include ADS-33 HQ specs as constraints in optimization • Include longitudinal acceleration feedback and pitch attitude feedback Airwake Spectral Filters Longitudinal acceleration feedback to improve gust response Pitch attitude feedback to provide closed-loop stability Optimize for minimal gust response PENNSTATE Stability Augmentation System • Optimize gains using CONDUIT Based on phase-lag compensator Design parameters include the prefix “dpp_” Roll SAS Pitch SAS Yaw SAS 1 8 5 5 PENNSTATE HQ specs 1 8 5 5 • Selected design specs from CONDUIT as constraints Closed-loop eigenvalues(EigLcG1), Gain/Phase margin(StbMgG1), Crossover frequency(CrsLnG1), Bandwidth for roll/pitch(BnwAtH1) • New spec for disturbance rejection(DisRnL1) Based on psd of angular rate response to corresponding gust input White noise Transfer function q( s) H ( s) qg ( s) PSD Magnitude [dB] (Example) – Pitch rate Level III Level II Level I Frequency [rad/sec] PENNSTATE HQ Specification Window 1 8 5 5 • Original SAS configurations - 30 degree WOD condition (1) roll pitch (2) yaw (3) CrsLnG1 (1) CrsLnG1 (2) CrsLnG1 (3) EigLcG1 (1) EigLcG1 (2) EigLcG1 (3) StbMgG1 (1) StbMgG1 (2) StbMgG1 (3) BnwAtH1 (1) BnwAtH1 (2) DisRnL1 (2) DisRnL1 (3) DisRnL2 (1) H H S J H J PENNSTATE HQ Specification Window 1 8 5 5 • Modified SAS configurations - 30 degree WOD condition (1) roll pitch (2) yaw (3) CrsLnG1 (1) CrsLnG1 (2) CrsLnG1 (3) EigLcG1 (1) EigLcG1 (2) EigLcG1 (3) StbMgG1 (1) StbMgG1 (2) StbMgG1 (3) BnwAtH1 (1) BnwAtH1 (2) DisRnL1 (2) DisRnL1 (3) DisRnL2 (1) H H S J H J PENNSTATE Simulation Results - Hovering Flight 1 8 5 5 • Angular rate responses (deg/sec) Result with Original SAS configurations Result with Optimized SAS configurations - 30 degree WOD R, deg/sec R, deg/sec Q, deg/sec Q, deg/sec P, deg/sec P, deg/sec - 0 degree WOD Time [sec] Time [sec] PENNSTATE Simulation Results - Hovering Flight 1 8 5 5 • SAS outputs (%) Result with Original SAS configurations Result with Optimized SAS configurations - 30 degree WOD YSAS, % YSAS, % PSAS, % PSAS, % RSAS, % RSAS, % - 0 degree WOD Time [sec] Time [sec] PENNSTATE Simulation Results - Hovering Flight • Pilot stick inputs (%) 1 8 5 5 Lateral cyclic input : Left 0%, Right 100% Longitudinal cyclic input : Forward 0% , Aft 100% Collective input : Down 0%, Up 100% Pedal input : Left 0%, Right 100% Result with Original SAS configurations Result with Optimized SAS configurations - 30 degree WOD PED, % PED, % COL, % COL, % LON, % LON, % LAT, % LAT, % - 0 degree WOD Time [sec] Time [sec] PENNSTATE Simulation Results - Hovering Flight • Angular rate autospectrum (dB) 1 8 5 5 Autospectra identified by CIFER Result with Original SAS configurations Result with Optimized SAS configurations - 30 degree WOD R, dB R, dB Q, dB Q, dB P, dB P, dB - 0 degree WOD Frequency [rad/sec] Frequency [rad/sec] PENNSTATE Simulation Results - Hovering Flight • Pilot stick input autospectrum (dB) 1 8 5 5 Autospectra identified by CIFER Result with Original SAS configurations Result with Optimized SAS configurations - 30 degree WOD PED, dB PED, dB COL, dB COL, dB LON, dB LON, dB LAT, dB LAT, dB - 0 degree WOD Frequency [rad/sec] Frequency [rad/sec] PENNSTATE H infinity Controller for SAS 1 8 5 5 • Include frequency-dependent weight functions for control inputs and outputs • Produce a controller K∞ to reduce the tracking deviations to reject disturbances • We is a high-gain low-pass filter for good tracking and disturbance rejection • Wu is a low-gain high-pass filter to improve the robustness and to limit the control activity Gust Filter (Wg) dw ref d + + + - dt Aircraft (UH-60) + +y Weighting (We) ee eu Weighting (Wu) u H∞ controller (K∞) PENNSTATE H infinity Controller Design 1 8 5 5 • Obtain a controller solving a classical 4-block problem based on 8-rigid-state linearized aircraft model 3 diagonal components of weighting functions iterate to find the optimal weighting parameters x Ax B1w B2u z C1 x D11w D12u y C2 x D21w D22u x A z C1 y C 2 , x= Aircraft We Wu B1 B2 x D11 D12 w D21 D22 u , w = dw dt , z= ee eu A G C1 C2 14-state H∞ controller ,u = rsas psas ysas B2 D11 D12 D21 D22 B1 , y= p q r PENNSTATE Simulation Results - Hovering Flight 1 8 5 5 • Angular rate responses (deg/sec) Result with Original SAS configurations Result with Optimized SAS configurations Result with H infinity controller - 30 degree WOD R, deg/sec R, deg/sec Q, deg/sec Q, deg/sec P, deg/sec P, deg/sec - 0 degree WOD Time [sec] Time [sec] PENNSTATE Simulation Results - Hovering Flight 1 8 5 5 • SAS outputs (%) Result with Original SAS configurations Result with Optimized SAS configurations Result with H infinity controller - 30 degree WOD YSAS, % YSAS, % PSAS, % PSAS, % RSAS, % RSAS, % - 0 degree WOD Time [sec] Time [sec] PENNSTATE Simulation Results - Hovering Flight • Pilot stick inputs (%) 1 8 5 5 Result with Original SAS configurations Result with Optimized SAS configurations Result with H infinity controller - 30 degree WOD PED, % PED, % COL, % COL, % LON, % LON, % LAT, % LAT, % - 0 degree WOD Time [sec] Time [sec] PENNSTATE Simulation Results - Hovering Flight • Angular rate autospectrum (dB) Result with Original SAS configurations Result with Optimized SAS configurations Result with H infinity controller Autospectra identified by CIFER - 30 degree WOD R, dB R, dB Q, dB Q, dB P, dB P, dB - 0 degree WOD Frequency [rad/sec] 1 8 5 5 Frequency [rad/sec] PENNSTATE Simulation Results - Hovering Flight • Pilot stick input autospectrum (dB) Autospectra identified by CIFER Result with Original SAS configurations Result with Optimized SAS configurations Result with H infinity controller - 30 degree WOD PED, dB PED, dB COL, dB COL, dB LON, dB LON, dB LAT, dB LAT, dB - 0 degree WOD Frequency [rad/sec] 1 8 5 5 Frequency [rad/sec] PENNSTATE Rotorcraft Flight Simulator 1 8 5 5 • Flight dynamics model is based on Genhel • Use FlightGear environment for visualization • Integrated with time-varying airwake data from CFD • Integrated with CHARM freewake model PENNSTATE Schedule and Milestones Tasks • Update GenHel Simulation for shipboard simulation • Develop simplified MATLAB Sim for control design • Interface GenHel with ship air wake solutions and ship motion • Develop maneuver controller • Validation of DI simulation (using JSHIP data) • Develop stochastic airwakes disturbance model and develop physical understanding • Develop real-time simulation capability at PSU • Incorporate CHARM free wake into the model • Task tailored control law design, support with real-time simulator at PSU • Lee PhD Degree • Derek Bridges PhD Degree 2001 2002 1 8 5 5 2003 2004 2005 Completed Short Term Long Term 2006 PENNSTATE 1 8 5 5 2004 Accomplishments • Developed stochastic airwake disturbance model for 0° and 30° WOD, use for off-line analysis, real-time simulation and flight control design • Real-time simulation facility is ready, integrated with time-varying airwake model and CHARM freewake model • Developed task-tailored control laws using CONDUIT and H infinity control method • Presented results at 2004 AIAA AFM conference, paper published in AIAA Journal of Aircraft, paper submitted to Journal of Aerospace Engineering (special issue on shipboard aviation) Planned Accomplishments for 2005 • Will present results at 2005 AHS Forum and submit as journal article • Continue to update and improve model, include the deck ground effects • Further study in task tailored control laws to improve disturbance rejection • Expand flight control design efforts, autonomous landing flight control system, position hold over ship deck • Investigate use of equivalent airwake disturbances as tool for validating ship CFD airwake models. PENNSTATE 1 8 5 5 Technology Transfer Activities: • Presented results at 2004 AIAA AFM Conference • Briefing to Navy Flight Dynamics Group at in Summer 2004, planning further interaction. Leveraging or Attracting Other Resources or Programs: • Obtained JSHIP flight test data for validation, Cdr. Kevin Delemar at NRTC is contact • Integrating with CHARM free wake model • Integrated model and controllers with simulation facility developed under DURIP funds Recommendations at the 2004 Review: Get with Navy to focus the project and also to interface with CFD activities (flow field). Actions Taken: Met with Navy. Received recommendations and we are planning more interaction. Proposed use of equivalent airwake disturbance model as tool for validation of CFD airwakes. PENNSTATE Overview of Accomplishments 2001-2005 1 8 5 5 Stochastic Airwake Disturbance Model • Method for extracting equivalent disturbances from simulation with full CFD airwake (can also be applied to flight test data) • Derived spectral filters to represent airwake disturbances PSD of vertical gust component, (ft/sec)2/(rad/sec) Advanced Simulation Model for Shipboard Operations • Interface with time accurate CFD solutions of ship airwake • High order Peters-He inflow • Tunable OCM pilot model • MATLAB / Simulink version of model for rapid development and control design • Validation against JSHIP flight test data • Implemented in real-time simulation facility at PSU “Best Fit” Spectral Filter Lw =10.7156 ft, sw = 4.81 ft/sec Extracted from simulation with full time-varying airwake H s 4s w 2 Lw L L 1 1.5913 w s 0.9173 w s V V V 2 3 Lw L L s 3.3169 w s 1.8985 w s V V V 1 2.5601 von Karman Turbulence Model Lw = 37.8667 ft, sw = 2.8067 ft/sec H s 4s w 2 Lw L L 1 2.7478 w s 0.3398 w s V V V 2 3 Lw L L s 1.9754 w s 0.1539 w s V V V 1 2.9958 Frequency (rad/sec) Task-Tailored Control Design for Shipboard Operations • Optimized SAS for operation in airwake using CONDUIT® • Use spectral filters in control synthesis • Optimized SAS using H∞ synthesis Airwake Spectral Filters Longitudinal acceleration feedback to improve gust response Pitch attitude feedback to provide closed-loop stability Publications • 5 conference papers, 1 journal paper published, 1 journal paper under review Optimize for minimal gust response Future Path PENNSTATE 1 8 5 5 Additional Basic Research • Should pursue similar analyses to study effects of building airwakes on UAVs operating in urban areas, proposed as follow on for next RCOE • Potential to investigate impacts on shipboard handling qualities requirements – Maritime ADS-33. • Could make further efforts to pursue the fully coupled problem, model effect of rotor wake on ship airwake, would need more CFD expertise Transition to Applications / Applied Research • Apply equivalent airwake disturbance method to validate ship airwake CFD analysis. Airwake disturbance can be extracted from flight test and compared to simulation with CFD wake • Use stochastic airwake model as a simplified and more compact model for use in trainers • Apply maneuver controller and simulation for analysis of new aircraft and new ship designs