State Estimation Using Dependent Evidence Fusion: Application to Acoustic Resonance-Based Liquid Level Measurement

Area of Science:

• Engineering
• Information Theory
• Signal Processing

Background:

• State estimation in dynamic systems commonly relies on known statistical properties of noise.
• Practical applications often involve bounded noises where precise distributions are unknown.
• Existing methods struggle with uncertainty quantification under bounded noise conditions.

Purpose of the Study:

• To develop a novel state estimation method for dynamic systems with bounded noises.
• To leverage Dempster-Shafer (DS) evidence theory for fusing dependent evidence under uncertainty.
• To improve the accuracy of state estimation in practical engineering applications.

Main Methods:

• A novel state estimation method based on Dempster-Shafer (DS) evidence theory is proposed.
• Dependent evidence is generated from state equations, observation equations, and actual system observations.
• The method is iteratively implemented, fusing evidence at each time step considering bounded noises.

Main Results:

• The proposed method successfully fuses dependent evidence from multiple sources under bounded noise constraints.
• Iterative implementation provides reliable state estimation values at every time step.
• Application to a low-frequency acoustic resonance level gauge yielded high-accuracy measurement results.

Conclusions:

• The Dempster-Shafer based state estimation method effectively handles bounded noises.
• The approach offers a robust solution for state estimation problems where noise distributions are not precisely known.
• The method demonstrates practical utility and high accuracy in real-world applications like acoustic resonance level gauging.