High-Precision Control of Aviation Photoelectric-Stabilized Platform Using Extended State Observer-Based Kalman Filter

Lu Wang, Xiantao Li, Yuzhang Liu, Dapeng Mao, Bao ZhangSensors 2023 · DOI

TL;DR

This paper tackles the challenge of maintaining precise line-of-sight (LOS) stability in airborne photoelectric platforms, which is degraded by both external disturbances and sensor noise. They propose a novel control strategy that combines an Extended State Observer (ESO) for disturbance rejection with a Kalman Filter (KF) for sensor noise reduction, achieving significantly improved stability accuracy.

What Problem Does This Paper Address?

Airborne photoelectric stabilization platforms are crucial for tasks like target acquisition and tracking on moving vehicles. Their effectiveness hinges on maintaining a stable line-of-sight (LOS) despite the inherent challenges of their operating environment. Two primary culprits degrade this stability: external disturbances (like aircraft vibrations, wind, and friction) and sensor noise, particularly from gyroscopes. Traditional methods like Extended State Observers (ESOs) are good at compensating for disturbances but struggle when sensor noise is high, as they can amplify this noise, leading to controller saturation and reduced performance. Conversely, noise filtering techniques can sometimes attenuate useful signals or introduce lag, impacting real-time performance. This paper aims to bridge this gap by developing a control approach that simultaneously addresses both disturbance and noise issues to achieve higher LOS stability accuracy.

Methods

The core of this paper is the proposed ESO + KF algorithm. Here's how it works:

  • Extended State Observer (ESO): The ESO is used to estimate and compensate for the "lumped disturbance" acting on the system. This disturbance encompasses both external environmental factors and internal system uncertainties. By treating these as a single observable state, the ESO can actively counteract them.
  • Kalman Filter (KF): The KF is then applied to the system's output after the ESO has compensated for the disturbance. The KF's role is to filter out the high-frequency noise from the gyroscope and other sensors.
  • Combined Approach (ESO + KF): The key innovation is the iterative interaction between the ESO and KF. The KF filters the sensor noise, providing a cleaner signal to the ESO. This cleaner signal allows the ESO to observe the disturbance more accurately. In turn, a more accurate disturbance observation by the ESO leads to a better-filtered state for the KF, creating a synergistic loop that enhances both disturbance rejection and noise reduction. The paper derives the mathematical framework for this combined algorithm.

Datasets

Not discussed in the paper. The paper focuses on experimental verification using a laboratory setup.

Key Insights and Ideas

  • Synergistic Noise and Disturbance Handling: The most significant insight is that effectively improving LOS stability requires tackling both sensor noise and external disturbances simultaneously. Treating them independently or sequentially can lead to suboptimal performance.
  • ESO Amplifies Noise: The paper highlights a critical limitation of traditional ESOs: their tendency to amplify sensor noise, which can degrade performance, especially in noisy environments.
  • KF for Real-Time Noise Reduction: The Kalman filter is identified as a suitable tool for real-time noise reduction in this context, offering better performance and less lag compared to traditional low-pass filters.
  • Iterative Improvement: The ESO + KF approach creates a feedback loop where each component enhances the other's performance, leading to a more robust and accurate control system.

Key Results

The paper presents experimental results demonstrating the effectiveness of the ESO + KF algorithm:

  • Static Noise Reduction: In static conditions, the ESO + KF reduced gyroscope noise from ±0.102°/s to ±0.038°/s, outperforming a traditional 150 Hz low-pass filter (±0.042°/s).
  • Dynamic Disturbance Rejection: Under dynamic disturbance conditions (1°/1 Hz and 3°/1 Hz swing), the ESO + KF significantly improved LOS stabilization accuracy:
  • At 1°/1 Hz, the standard deviation of the off-target quantity was reduced from 6.072°/s (traditional ESO) to 2.721°/s (ESO + KF), a 44.81% improvement.
  • At 3°/1 Hz, the standard deviation decreased from 15.60°/s (traditional ESO) to 6.15°/s (ESO + KF), a 39.42% improvement.
  • Reduced Controller Burden: The ESO + KF algorithm resulted in less ripple in the controller's driving amount, reducing the burden on the controller and mitigating potential system resonance.
  • Improved Real-Time Performance: The KF component provided better real-time performance with less lag (0.12 cycles) compared to a traditional low-pass filter (0.27 cycles).

Limitations and Future Work

  • Limitations:
    • The paper acknowledges that while the ESO + KF reduces noise and disturbance, there are still residual errors.
  • The performance of the algorithm is dependent on the accurate modeling of the system and the noise characteristics.
  • The paper mentions that traditional low-pass filters have limitations in terms of bandwidth and lag, which the KF aims to overcome, but further optimization of KF parameters might be beneficial.
  • Future Work:
    • The authors suggest that increasing the gain of the ESO controller within the ESO + KF framework can further improve disturbance suppression capabilities.
  • The paper implies that further research could explore adaptive tuning of the ESO and KF parameters for even better performance across a wider range of conditions.
  • The conclusion mentions a flight verification on a helicopter, suggesting that real-world flight testing is a crucial next step to validate the algorithm's performance in actual operational environments.

Summary updated

BibTeX
@article{wang2023high,
  author = {Lu Wang and Xiantao Li and Yuzhang Liu and Dapeng Mao and Bao Zhang},
  title = {{High-Precision Control of Aviation Photoelectric-Stabilized Platform Using Extended State Observer-Based Kalman Filter}},
  journal = {Sensors},
  year = {2023},
  doi = {10.3390/s23229204},
}
Figure