Studies on Frequency Response Optimized Integrators Considering Second Order Derivative

Sheng Lei,Alexander Flueck

DOI: https://doi.org/10.48550/arXiv.2101.03266

2021-01-09

Abstract:This paper presents comprehensive studies on frequency response optimized integrators considering second order derivative regarding their numerical error, numerical stability and transient performance. Frequency domain error analysis is conducted on these numerical integrators to reveal their accuracy. Numerical stability of the numerical integrators is investigated. Interesting new types of numerical stability are recognized. Transient performance of the numerical integrators is defined to qualitatively characterize their ability to track fast decaying transients. This property is related to unsatisfactory phenomena such as numerical oscillation which frequently appear in time domain simulation of circuits and systems. Transient performance analysis of the numerical integrators is provided. Theoretical observations from the analysis of the numerical integrators are verified via time domain case studies.

Numerical Analysis,Systems and Control

What problem does this paper attempt to address?

This paper attempts to solve the problems of accuracy, numerical stability and transient performance of numerical integrators when dealing with high - frequency components and fast transient signals in time - domain simulation. Specifically:

1. **Improve the accuracy of numerical integrators**:
- By optimizing the frequency response, the error introduced by the numerical integrator when processing signals with specific dominant frequency components is minimized.
- Consider the second - order derivative to improve the accuracy of ordinary differential equations (ODEs) in the discretization process.

2. **Analyze numerical stability**:
- Research the numerical stability of the numerical integrator, especially identify some new types of numerical stability, and ensure that the numerical solution can correctly follow the trend of the analytical solution under different step sizes.

3. **Define and analyze transient performance**:
- Define transient performance to qualitatively describe the ability of the numerical integrator to track fast - decaying transients and avoid undesirable phenomena such as numerical oscillation.

### Main contributions

1. **Frequency response analysis**:
- Conduct error analysis in the frequency domain to reveal the accuracy performance of the numerical integrator under non - zero frequency components, rather than just the traditional truncation error analysis based on Taylor expansion.

2. **Numerical stability analysis**:
- Analyze the numerical stability of the numerical integrator, discover new types of numerical stability, and discuss the conditions and impacts of these new types.

3. **Transient performance definition and analysis**:
- Define transient performance to qualitatively describe the performance of the numerical integrator when dealing with fast - decaying transient signals, and analyze the transient performance of several numerical integrators.

4. **Verification and case study**:
- Verify the results of error analysis and transient performance analysis through time - domain case studies to ensure the correctness and practicality of theoretical analysis.

### Examples of mathematical formulas

- **General form of ODE**:
\[
\dot{x}=f(t, x, u)
\]
where \(t\) represents time, \(x\) represents state variables, \(u\) represents input, and \(f\) is a function depending on \(t\), \(x\) and \(u\).

- **Discretization formula of numerical integrator considering second - order derivative**:
\[
hx_{t + 1}=a_{-1}x_{t - 1}+b_0x_t + b_{-1}x_{t - 1}+c_0\ddot{x}_t + c_{-1}\ddot{x}_{t - 1}
\]

- **Frequency - domain error expression**:
\[
X(s)-a_{-1}X(s)e^{-sh}-b_0sX(s)-b_{-1}sX(s)e^{-sh}-c_0s^2X(s)-c_{-1}s^2X(s)e^{-sh}
\]

Through these analyses, the paper aims to provide more efficient and accurate numerical integration methods for time - domain simulation of electronic circuits and power systems.