Researchers Develop Algorithm for Parameter Optimization in Nonlinear Wiener Systems with Backlash

Researchers have proposed a new algorithm for optimizing parameters in extended Wiener systems affected by backlash nonlinearity, particularly when dealing with binary-valued data. The algorithm uses repetitive learning and fractional order techniques to enhance estimation accuracy. Accurate parameter estimation is essential for high-performance modeling and adaptive control in nonlinear systems.

The study focuses on Wiener systems, which consist of a linear dynamic block followed by a static nonlinear block. Backlash nonlinearity introduces challenges such as hysteresis, complicating parameter identification. Binary-valued data, often encountered in practical applications like control systems with limited sensors, further limits the information available for estimation.

The proposed algorithm employs repetitive learning to iteratively refine parameter estimates based on repeated system operations. Fractional order calculus is integrated to handle the non-integer dynamics inherent in such systems. com reported that this method achieves improved convergence and robustness compared to traditional approaches.

The algorithm optimizes parameters by minimizing an error function derived from observed binary outputs.

It addresses the identifiability issues in systems with backlash by incorporating a hysteresis model. Simulations and experimental validations demonstrate its effectiveness in scenarios with noisy or sparse data. Background on Wiener systems dates back to their use in chemical engineering and signal processing since the mid-20th century.

Recent advancements have extended these models to include nonlinearities like backlash, common in mechanical systems such as gears and actuators. The stakes involve improving control in industries like robotics and manufacturing, where imprecise models can lead to inefficiencies or failures.