Rotorcraft ASIAS - HAI Heli-Expo - Helicopter Association International
Transcription
Rotorcraft ASIAS - HAI Heli-Expo - Helicopter Association International
Rotorcraft ASIAS (Aviation Safety Information Analysis and Sharing) Brian Haggerty - HAI Cliff Johnson - FAA Kipp Lau - HAI Keith Cianfrani - FIT Kyle Collins - GT Presentation Outline • • • • • • • Program Overview and Background Overview of R-ASIAS system Overview of FDM (Flight Data Monitoring) Review of Operator Interface and Architecture Overview Research Efforts How to Participate Questions 2 Federal Aviation Administration CLIFF JOHNSON 3 PEGASAS Description • PEGASAS = The Partnership to Enhance General Aviation Safety, Accessibility and Sustainability (PEGASAS) is a Federal Aviation Administration (FAA) Center of Excellence for General Aviation. • Established: Dec. 2012; constitutes a 10-year partnership with the FAA. • Mission: The mission of PEGASAS is to enhance general aviation safety, accessibility, and sustainability by partnering the FAA with a national network of world-class researchers, educators and industry leaders. • Linkage to Rotorcraft ASIAS: PEGASAS supports Rotorcraft ASIAS research efforts by providing world-class researchers well-versed in causal modeling, data mining, analysis, and prototype software application development. PEGASAS is examining new ways to analyze and implement FDM techniques within the overall helicopter community (i.e. safety tools, policies, metrics, etc.). 4 What is Rotorcraft ASIAS? • A system for secure, confidential, and protected safety analysis of flight data records • Supported by the FAA in its mission to promote and advance flight safety • Participating Operators are the key stakeholders, users and beneficiaries of the system • Developed and maintained by an independent 3rd party with strong ties to operators and industry (HAI) 5 Why Rotorcraft ASIAS 6 U.S. Helicopter Fatal Accident & Fatality Rates *CY 14, Jan-Jul Only 7 Rotorcraft ASIAS Overview • How can we encourage and cultivate participation in this vital R&D effort among helicopter operators, manufacturers, and other stakeholders for safety purposes (i.e. How can you help us help you to increase safety in your daily operations?) – Solicit Helicopter Flight Data Monitoring (HFDM) data for ASIAS across representative mission types – Encourage Helicopter Subject Matter Expert Advice/Input (i.e. Pilots, Safety Analysts, etc.) towards R&D activities and working groups – Promote benefits of contributing data to ASIAS Knowledge of “what you don’t know” (i.e. hidden risks/dangers) only visible via sharing of information among parties Ability to promote increased situational awareness and safety within helicopter operations Incorporate new safety analysis tools into helicopter operations 8 Rotorcraft ASIAS Research Team Members (Lead PEGASAS school) Cliff Johnson Rotorcraft ASIAS Technical Monitor FAA WJHTC Atlantic City, NJ Prof. Dimitri Mavris – PI Hernando Jimenez, Ph.D. – Co-PI Kyle Collins, Ph.D. Simon Briceno, Ph.D. Alek Gravilovski Prof. Karen Marais – PI Prof. Inseok Hwang – Co-PI Arjun Rao Sanghyun Shin Ed DiCampli, Chief Operating Officer– PI Harold Summers, Director of Flight Operations Brian Haggerty, Deputy Director of Flight Operations Jay Clark, Manager, Information Systems Kipp Lau, RASIAS FDM Consultant Scott Collins, RASIAS IT Consultant Prof. Steve Cusick – PI Keith Cianfrani Gabrielle Landry 9 Concept of Rotorcraft ASIAS Rotorcraft ASIAS All Missions Flight Training Charter EMS Oil/Gas Logging Police News Cargo Lift … Participating Operators Own-Access User FAA, Sponsor Secure System Database Analysis Toolkit Interface & Display Aggregate Results De-identified Data Access User System Developer & Administrator Analysis Tools Visualizations, Algorithms, Event Definitions PEGASAS Research Team for Rotorcraft ASIAS 10 Helicopter FDM Data Gathering & Analysis for ASIAS (Rotorcraft ASIAS) Research Question Research Requirement Outcomes Outputs Implementation Plan • Rotorcraft ASIAS methodology & integration plan • Develop advisory materials. • “The U.S. Helicopter Safety Team has identified a goal of an 80% reduction in the helicopter accident rate (2006-2016), with a long-term vision of zero tolerance = zero accidents. The #1 proposed solution in order to reach this goal is the adoption of Rotorcraft FDM across the industry.” • Research question: How do we develop tools, techniques, policies, and procedures that enable the sharing and analysis of rotorcraft FDM data within ASIAS? • Sponsor: Mark Liptak/AVP-200 • Performer: Cliff Johnson/ANG-E272. • Collection of FDM Data for Rotorcraft ASIAS • Mission-Specific FDM parameters & exceedances • Quantification of undesired rotorcraft events • Helicopter FDM parameter portfolios • Rotorcraft safety data mining techniques • Contribution to the reduction of the helicopter & general aviation fatal accident rates (FAA Destination 2025 Plan). • Expansion of ASIAS safety initiatives to the rotorcraft communities. • Outreach efforts to contribute to the reduction of the GA/rotorcraft fatal accident rate. • Development of new Tools & Techniques for the Helicopter Community to conduct safety analyses in accordance with proper ASIAS Executive Board & USHST governance structures. • Expand ASIAS to additional participants. • Disseminate best practices/guidance material for Rotorcraft FDM • Revise Rotorcraft training materials. • Support U.S. Helicopter Safety Team (USHST) initiatives. • Rotorcraft ASIAS prototype system 11 Helicopter Association International KIPP LAU 12 What is Flight Data Monitoring (FDM) • “Snapshots” of various airborne parameters recorded at least 1X/second • Data is recorded by dedicated hardware that monitors parameters and records them to media for later access • Parameters are recorded and can be evaluated to indicate exceedances • Multiple parameters can be combined to identify events 13 HFDM programs: Pre 2009 (individual “silos”) Aircraft ERA Sector: CHC Bristo w Cougar PHI Others Oil and Gas S-76, S-92 SA-365/EC155, AS332/EC225, EC135 BH412 BH212 BH206 ACH HEMS Aircraft S-76C+ Others HFDM programs: 2009-2014 (IHST HeliShare) Aircraft S-76, S-92 ERA Bristo w PHI Sector: CHC CHI SA-365/EC155, AS332/EC225, EC145, EC135 BH412, BH212, BH407 BH206 Cougar Shell Metro Chevro n Oil and Gas Etc… PHI HEMS Air Method s ACH Aircraft EC145, EC135 S-76C+ BH407 Others HFDM programs: 2015 and beyond (R-ASIAS conceptual) ERA CHC 7Bar CHI CalStar Bristo w PHI Sector: Cougar Chevro n Oil and Gas Shell MedFlt AMG H Metro PHI HEMS Boston MedFlt Air Method s ACH Others … Las Vegas PD Fairfax Co. Others Parameters, Rates & Exceedances What events might they help diagnose? • Loss of Control • Loss of tail rotor effectiveness • Unstabilized Approach • Autorotation • Tailstrike • Abnormal Runway Contact • Ground Collision Proximity Warning/Controlled Flight Into Terrain • Ground Resonance • Flight Control System Failures • Instrument Failures • Engine Failures • Rotor Shaft Failures • Excess Loading (G-Forces) • Weather (Turbulence, Winds hear, Microburst, Thunderstorm, etc.) • Outside of the Envelope Flight 17 How is the Data Utilized? • Algorithms based on various combinations of even basic recorded parameters can re-create a flight or event in the flight • Aggregation can be accomplished by operator, mission segment or aircraft type to well define “normal operations” • Large batches of flight data can be scanned quickly to identify operations outside of normal parameters 18 Big data is about challenges and opportunities Characteristics • Growing quantity of data • Quickening speed of data generation • Increase in types of data • Veracity of the data Opportunities • Making better informed decisions • Discovering hidden insights, Anomalies, forensics, patterns and trends • Automation 19 Operator Process Operator Operator Automated 20 IT Architecture of Rotorcraft ASIAS Presentation / Web Tier ... Web server presenting pages to end users and dispatching/scheduling workload to the application layer Application Layer ... Query data, provide analysis and return results to presentation layer Data Tier RDBMS Cluster HADOOP MapR Cluster File Server cluster External Data Sources 21 Guiding Principles • Robust computer, network, systems, data and IT security are paramount to a successful R-ASIAS implementation • Security best practices and capabilities have been designed into the systems architecture from the very start to assure a comprehensive approach to this critical capability Data Access • Voluntary sharing of information (only within the research team via signed non-disclosure agreements) for research purposes • Operators sign agreements with HAI, HAI has signed agreements with PEGASAS (university community) • HFDM data secured and protected from unauthorized disclosure including de-identification of data Computer Systems Security • FISMA – Federal Information Security Management Act RASIAS will follow guidelines, standards and best practices as outlined by FISMA • Access – Understand how data access is granted, who it is granted to and what they have access to • Encryption – End to end encryption of all data during communication as well as encryption of sensitive “data at rest” • Integrity – Measure and validate data integrity Systems & Data Security is a Process Security Toolkit • Firewalls • Software Updates • User Access Rules • Active Logging • Log Auditing • Encryption • Authentication Continuous Process Build Innovate Audit Secure Test Benefits to Participation • Gain insight into how your flight practices compare in the industry • Potential to increase safety by examining flight profiles • Examine flight profiles to facilitate noise abatement • Provide anonymous data to increase industry wide understanding of flight profiles • Potential to decrease insurance costs 26 Florida Institute of Technology LTC (RET) KEITH M CIANFRANI, MAS, RSP Georgia Institute of Technology KYLE COLLINS, Ph.D. 27 Overview of Rotorcraft ASIAS RE&D Efforts • • • • • Examine HFDM process Develop rotorcraft ASIAS architecture Devise HFDM data mining techniques Develop novel tools for safety analysis Study HFDM events – Minimum list SME vetted events (parameters, rates, exceedances) – Mission specific events Analysis Flight tests • Develop data enhanced helicopter simulation models – New event discovery – Performance envelope margin analysis – Missing parameter estimates • Secure operators 28 Benchmarking the State of the Art of Rotorcraft FDM Literature Review 1. 2. 3. 4. 5. 6. Rotorcraft Safety Key Stakeholders Lessons Learned and Success Stories FDM Systems Regulations, Policy, Technical Standards Reported Benefits and Costs • ~130 documents/ reports reviewed • ~55 included in review • Two SME reviews, 3 incremental revisions Benchmarking of the FDM Process 1. 2. 3. 4. 5. 6. Background Typical FDM Process Planning and Preparation Steering Committee FDM Reporting Challenges and Solutions for Smaller Operators 7. HFDM Best Practices R-ASIAS Architecting and Development Architecting of Rotorcraft ASIAS System 1. Development and refinement of concept of operations 2. Requirements analysis 3. Tool mock-ups and use cases 4. Analysis prototyping and algorithm definition 5. Visualization Secure Operator Participation 1. 2. 3. 4. Establish relationship with R-ASIAS users Articulate value proposition Facilitate execution of agreements with HAI Develop IRB protocols and support infrastructure for ethical research with “human data” 5. Capture the voice of participating operators in the development of system capabilities Although the FAA has sponsored this project, it neither endorses nor rejects the findings of this research. The presentation of this information is in the interest of invoking technical community comment on the results and conclusions of the research." Review and Advancement of FDM Events, Parameters, Rates and Exceedances Identify and Review Additional Events, Parameters, and Exceedances Identify FDM Data Processing Techniques 1. 2. 3. 4. Examine and document data processing Sparseness and aberrant entries Correction measures Recommendations for data processing and storage 1. SME approach – engagement with operators and experts 2. Taxonomical approach – mapping of safety space and FDM events to discover gaps 3. Data-based approach – Data mining and new trend-discovery algorithms Review Current Events, Parameters, Rates and Exceedances 1. Compiled repository of FDM event, parameter, and threshold definitions 2. Compiled repository of FDR equipment and recording rates 3. Referenced from published HFDM studies and operator FDM program information Generic Event Name Excessive Roll Angle Excessive Roll Angle Excessive Roll Rate Excessive Roll Rate Excessive Roll Rate Defined Event Name (explicitly defined thresholds) Excessive Roll Attitude Below 500 ft AGL Radio Height, Absolute (Roll Angle) Flight Phase OG, HV,TR,APP CICTT Phase of Flight Low Altitude/Hover Excessive Bank Roll Air Maneuvering High Roll Rate Above 500 ft AGL High Roll Rate Below 500 ft AGL Radio Height, Absolute (Roll Angle) Radio Height, Absolute (Roll Angle) Air Maneuvering OG Low Altitude/Hover Excessive Roll Rate Roll rate Air Maneuvering Parameters Category Crew action - Control Crew action - Control Crew action - Control Crew action - Control Crew action - Control Analysis of FDM Voice and Video Recordings 1. Benchmark CVR state of the art 2. Develop video pattern recognition capabilities 3. Video-based attitude indicator as complement to flight data records Although the FAA has sponsored this project, it neither endorses nor rejects the findings of this research. The presentation of 30 this information is in the interest of invoking technical community comment on the results and conclusions of the research." Safety Tools Survey and Opportunities for Flight Data Utilization Safety Software 1. Survey and review of FDM, Risk Management Systems (RMS) and Safety Management Systems (SMA) software 2. Identification and prioritization of analysis features for enhanced FDM capabilities 3. Prototyping of enhanced FDM capabilities Data Fusion 1. Survey and review of existing aviation databases 2. Prototyping of safety enhancements through data fusion Safety Metrics 1. Exhaustive survey and review of aviation safety metrics 2. Prototyping of identified safety metrics based on FDM records Data-enhanced Performance Models 1. Performance models can greatly enhance FDM programs: • Review event definitions, identify/define new events • Examine performance envelope, back off safety margins • Infer parameter values when data is missing or corrupt 2. Develop rotorcraft models and capability to calibrate with operator’s flight data records Mapping Accident Cause to Exceedances 1. Mapping of accidents to exceedances benefits safety and risk studies, directs attention to critical FDM events 2. Analysis of accident data, assess severity and frequency 3. Risk-mapping of accidents to occurrences with statistical analysis 4. Mapping of FDM events to occurrences with SME input Although the FAA has sponsored this project, it neither endorses nor rejects the findings of this research. The presentation of this information is in the interest of invoking technical community comment on the results and conclusions of the research." 31 Design and Analysis of FDM Equipment Flight Experiments Support Operators with Small or Partial FDM Programs 1. Support FAA in the design of flight tests to investigate variations in equipment and installation 2. Conduct analysis of effects and from flight test data to determine the settings with optimal results and least variability Flight tests Analysis 1. Produce recommendations of resources and tools for smaller operators with small or partial FDM programs 2. Whenever possible, procure some of these resources 3. Emphasis on maximum impact and benefit • What does a full/complete FDM program look like? • What do FDM programs of small operators look like in reality? What are they missing? What makes them “partial”? • How can operators with a small or partial FDM program be supported? What support options offer the most benefit at the least expense? Recorder and installation design of experiments Study of Data Mining Techniques for FDM Applications 1. Exhaustive survey and review of data mining and current applications to FDM 2. Prototyping of “proven” data mining approaches 3. Development and prototyping of novel data mining algorithms 4. All capabilities made available to participating operators “New” safety conditions identified with unsupervised machine learning algorithm Conditions identified with event definitions Although the FAA has sponsored this project, it neither endorses nor rejects the findings of this research. The presentation of 32 this information is in the interest of invoking technical community comment on the results and conclusions of the research." Benefits to Participation • Research and Development Phase – During the formative years, industry SMEs have the opportunity to provide feedback on program structure and governance (ex. Qualifications of Helicopter Issue Analysis Team members) – Opportunity through industry working group (PEGASAS|HAI outreach) to provide input that will ultimately lead to HFDM system improvements: Common recorded parameter sets Increased number and fidelity of recorded parameters Advanced algorithms used to analyze rotorcraft ops Improvements in analysis tools Benefits to Participation • ASIAS Implementation Phase – Opportunity to participate in data collection activities that may lead to future improved operational safety and efficiencies. – Opportunity to participate in industry info sharing activities (InfoShare) that leads to additional safety enhancements. – Opportunity to use ASIAS info to identify threats (beyond internal reporting) – Similar to the CAST model, a collaboration of industry and government experts provide input on directed studies to solve issues in the NAS. Rotorcraft specific issues could include: Operations around oil rigs (gas discharges) Electronic news gathering ops Helicopter Air Ambulance ops Tour operators (traffic conflicts) Benefits to Participation • ASIAS Mature Phase – Opportunity to compare operations (metrics) to other like or dissimilar operators by (through ASIAS portal): Region, Aircraft Type, Mission Profile, Etc. – Opportunity to learn from advanced studies which fuse data from multiple sources: ATC, Weather, Other operators of similar and dissimilar aircraft (GA to airliners), TAWS alerts, TCAS alerts, Terrain maps, Etc. – Operators benefit from airspace improvements (examples) NORCAL (min vectoring altitudes over Mt. Diablo) SOCAL (improved VFR corridors near Burbank) – Operators benefit from infrastructure and procedural improvements (examples) Newark (rough runways) Orlando (unstable approaches due to underlying GA airport) Albuquerque (VFR approach procedures to avoid TAWS alerts) Rotorcraft ASIAS Timeline Research Prototype || Full Integration with ASIAS Develop Rotorcraft ASIAS System Requirements analysis, system architecting and design, implementation, standards for data formatting/processing prototyping, testing, incremental delivery of tools and capabilities, integration with existing ASIAS communities Secure Operator Participation Develop governance, establish agreements, insure data protection and confidentiality, test and monitor data transfer, elicit operator feedback, ensure and monitor value added to operators, enhance system to meet emerging operator needs Support Rotorcraft ASIAS Research Establish generic event set, identify event set gaps, video/audio processing, safety metrics, software capabilities, data fusion, accident mapping, performance models, FDM flight testing, data mining and knowledge discovery with FDM data Conduct Outreach and Community Engagement Establish outreach efforts within the Helicopter Community, present research topics & results at Heli-Expo, industry forums/events, HFDM Working Groups, mitigation partner 2013 2015 2017 2019 36 Summary / Next Steps • Interact and engage with operators • Site visits to discuss details, learn about your needs and see how we can add value to your operations • Establish data agreements • Participate 37 How To Participate This project will combine data analysis and engagement with rotorcraft stakeholders and subject matter experts to review the state of the art of Rotorcraft Flight Data Management (FDM) Contact Us General Information or to participate Email: [email protected] Website: https://www.pegasas.aero/projects.php?p=2 FAA Sponsoring Office: Cliff Johnson Email: [email protected] Phone: (609) 485-6181 System Safety Section, ANG-E272 Aviation Research Division William J. Hughes Technical Center 38 Questions? Presenters • Brian Haggerty – HAI – Phone: (703) 683-4646 – Email: [email protected] • Cliff Johnson - FAA • Kipp Lau – HAI • LTC (ret) Keith M Cianfrani - FIT • Kyle Collins, Ph.D. - GT – Phone: (609) 485-6181 – Email: [email protected] – Phone: (502) 649-3211 – Email: [email protected] – Phone: (267) 377-5364 – Email: [email protected] – Phone: (404) 385-2786 – Email: [email protected] 39