Parallel Kalman filter group integrated particle filter method for the train nonlinear operational status...

Area of Science:

• Control Engineering
• Signal Processing
• Transportation Systems

Background:

• Accurate state estimation is crucial for train operation safety and efficiency.
• Existing methods struggle with nonlinear, non-Gaussian, and multi-mode conditions in real-time train dynamics.

Purpose of the Study:

• To develop a high-precision parallel Kalman filter group (NN-HEKFG) integrated with Particle Filter for multi-mode state estimation in train operations.
• To address the challenges of nonlinearity and non-Gaussianity in train running state estimation.

Main Methods:

• Multi-model Gaussian decomposition of probability density functions for state and measurement equations.
• Representation of local state models using multi-dimensional high-order polynomials to establish an expanded dimensional state model.
• Updating local state model parameters and solving particle filtering for global estimation results.
• Establishing a parameter reduction criterion to avoid parameter explosion by re-identifying weights and means.

Main Results:

• The proposed NN-HEKFG demonstrates superior performance compared to standard particle filters and Gaussian sum filters.
• Effective estimation of multi-mode train running states was verified through simulations.
• The parameter reduction criterion successfully mitigated the issue of parameter explosion.

Conclusions:

• The NN-HEKFG offers a robust and high-precision solution for multi-mode state estimation in train operations.
• The method effectively handles nonlinear and non-Gaussian characteristics inherent in train dynamics.
• This approach provides a significant advancement for intelligent train control and safety systems.