7.1 Sensors and state estimation for airborne systems

Sensors and state estimation are crucial for airborne wind energy systems. They provide vital data on position, orientation, wind conditions, and environmental factors. This information enables accurate control and optimal energy harvesting in dynamic atmospheric conditions.

State estimation techniques fuse data from multiple sensors to determine the system's current state. Advanced methods like Kalman filters and machine learning approaches handle the complexities of airborne systems, improving accuracy and reliability in challenging environments.

Essential Sensors for Airborne Wind Energy

Inertial and Positioning Sensors

Top images from around the web for Inertial and Positioning Sensors

• 

  WES - Clustering wind profile shapes to estimate airborne wind energy production View original

  Is this image relevant?

• 

  Apollo Command Module Inertial Measurement Unit (IMU) | Flickr View original

  Is this image relevant?

• 

  BEE Board and Daisy-7 IMU-GPS module View original

  Is this image relevant?

• 

  WES - Clustering wind profile shapes to estimate airborne wind energy production View original

  Is this image relevant?

• 

  Apollo Command Module Inertial Measurement Unit (IMU) | Flickr View original

  Is this image relevant?

1 of 3

Top images from around the web for Inertial and Positioning Sensors

• 

  WES - Clustering wind profile shapes to estimate airborne wind energy production View original

  Is this image relevant?

• 

  Apollo Command Module Inertial Measurement Unit (IMU) | Flickr View original

  Is this image relevant?

• 

  BEE Board and Daisy-7 IMU-GPS module View original

  Is this image relevant?

• 

  WES - Clustering wind profile shapes to estimate airborne wind energy production View original

  Is this image relevant?

• 

  Apollo Command Module Inertial Measurement Unit (IMU) | Flickr View original

  Is this image relevant?

1 of 3

• Inertial Measurement Units (IMUs) measure acceleration, angular velocity, and orientation of the airborne system
  • Typically contain accelerometers, gyroscopes, and sometimes magnetometers
  • Provide high-frequency data for short-term motion tracking
• Global Positioning System (GPS) receivers provide accurate position and velocity information
  • Offer global coverage and long-term stability
  • Update rate typically lower than IMUs (1-10 Hz)
• Barometric pressure sensors measure altitude and assist in vertical position estimation
  • Utilize atmospheric pressure changes to determine height above sea level
  • Complement GPS for improved vertical accuracy

Wind and Force Measurement Sensors

• Wind sensors measure wind speed and direction crucial for optimal energy harvesting
  • Anemometers measure wind speed (cup, propeller, or sonic types)
  • Wind vanes determine wind direction
  • Pitot tubes measure airspeed for fast-moving airborne systems
• Tension sensors monitor tether forces and stress in airborne wind energy systems
  • Strain gauges or load cells measure mechanical forces
  • Critical for preventing tether overload and optimizing power generation

Environmental Awareness Sensors

• Optical sensors aid in obstacle detection and environmental awareness
  • Cameras provide visual information for navigation and obstacle avoidance
  • LiDAR (Light Detection and Ranging) offers precise 3D mapping of surroundings
• Magnetometers measure the Earth's magnetic field to determine device heading
  • Assist in orientation estimation when combined with IMU data
  • Susceptible to magnetic interference from nearby structures or electronics

State Estimation in Airborne Systems

Fundamental State Estimation Techniques

• State estimation determines system's current state based on sensor measurements and models
  • State typically includes position, velocity, and orientation
  • Combines noisy sensor data with system dynamics for optimal estimates
• Kalman filtering provides optimal state estimation in linear systems with Gaussian noise
  • Recursive algorithm that minimizes mean squared error
  • Consists of prediction and update steps
• Extended Kalman Filter (EKF) and Unscented Kalman Filter (UKF) handle nonlinear systems
  • EKF linearizes system around current estimate
  • UKF uses deterministic sampling to propagate probability distributions

Advanced State Estimation Methods

• Particle filters, or Sequential Monte Carlo methods, used for non-Gaussian and highly nonlinear problems
  • Represent probability distributions using a set of weighted particles
  • Particularly useful for complex environments or multi-modal distributions
• Complementary filters combine high-frequency and low-frequency sensor data
  • Example: Fusing high-frequency IMU data with low-frequency GPS updates
  • Simple yet effective for attitude estimation in small aerial vehicles
• Model-based estimation techniques incorporate dynamic models of the airborne system
  • Improve state predictions by leveraging known system behavior
  • Can account for external forces (wind, tether dynamics) in state estimates
• Machine learning approaches employed for state estimation in complex environments
  • Neural networks can learn nonlinear system dynamics from data
  • Recurrent neural networks (RNNs) or Long Short-Term Memory (LSTM) networks suitable for sequential data

Sensor Fusion for State Estimation Accuracy

Sensor Fusion Architectures and Techniques

• Sensor fusion combines data from multiple sensors for improved accuracy and reliability
  • Exploits complementary strengths of different sensor types
  • Mitigates weaknesses or failures of individual sensors
• Centralized fusion architectures process all sensor data in a single estimator
  • Optimal when computational resources are available
  • Can become computationally intensive for large sensor networks
• Decentralized approaches use multiple local estimators
  • Distribute computational load across multiple nodes
  • More robust to single-point failures
• Bayesian inference provides a probabilistic framework for combining sensor measurements
  • Incorporates prior knowledge with new observations
  • Handles uncertainty in both measurements and prior beliefs

Advanced Fusion Methods and Reliability Enhancements

• Covariance Intersection fuses estimates without knowing their degree of independence
  • Useful when correlation between different estimates is unknown
  • Provides conservative but consistent fusion results
• Multi-rate fusion techniques handle sensors with different sampling rates
  • Interpolation or extrapolation used to align measurements in time
  • Ensures consistent state estimates across varying sensor update frequencies
• Fault detection and isolation methods improve system reliability
  • Identify sensor failures or anomalies in real-time
  • Exclude or down-weight faulty sensor data in fusion process
• Adaptive fusion algorithms adjust sensor weighting based on estimated reliability
  • Dynamically adapt to changing environmental conditions
  • Example: Reducing GPS weight in urban canyons with poor satellite visibility

Challenges of Sensor Integration in Airborne Wind Energy

Environmental and Physical Constraints

• Environmental factors affect sensor performance and reliability
  • Temperature variations can cause sensor drift or bias
  • Humidity may impact certain sensor types (optical sensors)
  • Electromagnetic interference can disrupt GPS or magnetometer readings
• Weight and power constraints limit sensor integration in airborne systems
  • Lightweight sensors preferred to maximize payload capacity
  • Low-power sensors extend operational time of battery-powered systems
• Sensor calibration and alignment errors propagate through estimation process
  • Misaligned IMU axes lead to orientation estimation errors
  • GPS antenna offset from center of gravity causes position discrepancies

Data Processing and Communication Challenges

• Communication bandwidth limitations restrict real-time data transmission
  • High-bandwidth sensors (cameras, LiDAR) may require on-board processing
  • Data compression techniques can help mitigate bandwidth constraints
• Computational constraints on embedded systems limit algorithm complexity
  • Simplified estimation algorithms may be necessary for real-time operation
  • Hardware acceleration (GPUs, FPGAs) can enable more complex algorithms
• Sensor drift and bias over time lead to accumulating errors
  • Periodic recalibration or in-flight bias estimation required
  • Sensor fusion can help detect and compensate for individual sensor drift
• Integration of heterogeneous sensors presents synchronization challenges
  • Different update rates and latencies must be accounted for
  • Time stamping and buffering techniques ensure consistent state estimates