Abstract:In this paper an adaptive state observer and parameter identification algorithm for a linear time-varying system are developed under condition that the state matrix of the system contains unknown time-varying parameters of a known form. The state vector is observed using only output and input measurements without identification of the unknown parameters. When the state vector estimate is obtained, the identification algorithm is applied to find unknown parameters of the system.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is how to design an adaptive state observer and a parameter identification algorithm in a linear time - varying system (LTV) when the state matrix of the system contains unknown time - varying parameters. Specifically, the paper focuses on estimating the state vector only through input and output measurements when the system state vector cannot be directly measured, and further using these estimates to identify the unknown time - varying parameters in the system.
### Problem Background
In the control of dynamic systems, the observation of the system state is an important issue. In some cases, the state vector can be directly measured using sensors, but in many cases, the state vector cannot be directly measured. Therefore, a state - observation algorithm needs to be designed to estimate the state vector. However, when the system contains unknown parameters, especially when these parameters are time - varying, traditional state - observation methods (such as the Kalman filter, Luenberger observer, etc.) are no longer applicable.
### Main Contributions of the Paper
1. **Adaptive State Observer**: The paper proposes a new adaptive state observer that can estimate the state vector based on input and output measurements when the system state matrix contains unknown time - varying parameters.
2. **Parameter Identification Algorithm**: After obtaining the state vector estimate, the paper further proposes a parameter identification algorithm for estimating the unknown time - varying parameters in the system.
### Specific Methods
- **State Observer Design**:
- Assume that the system state matrix \(A(t)\) can be expressed as the sum of a known part \(A_0(t)\) and an unknown part \(D(t)\), that is, \(A(t)=A_0(t)+D(t)\).
- Design an observer in the form of:
\[
\dot{\hat{x}} = M\hat{x}+Nu + G\dot{y}+LC\tilde{x}
\]
- By choosing appropriate matrices \(M, N, G, L\), make the estimation error \(\tilde{x}=x - \hat{x}\) converge to zero.
- **Parameter Identification**:
- Using the estimated value of the state vector, through a series of transformation and filtering operations, transform the parameter identification problem into the form of a linear regression model.
- Use the gradient algorithm or other linear regression parameter estimation methods to estimate the unknown parameters.
### Numerical Simulation
The paper verifies the effectiveness of the proposed algorithm through a specific numerical example. The simulation results show that both the estimation error of the state vector and the estimation error of the parameters can converge to zero, proving the feasibility and effectiveness of the algorithm.
### Conclusion
The paper successfully solves the state - observation and parameter - identification problems in a linear time - varying system containing unknown time - varying parameters, and provides a new adaptive observer and parameter - identification algorithm. These methods are not only innovative in theory but also have high practical value in practical applications.