A new study published in Engineering introduces a neural-network-based switching output regulation controller (NN-SORC) for high-speed nano-positioning stages, aiming to suppress hysteresis nonlinearity in piezoelectric actuation and improve tracking performance under switching reference signals. The research, conducted by a team from Huazhong University of Science and Technology and the University of Victoria, presents a mechatronic platform and control framework that supports reliable micro- and nano-scale manipulation in precision detection and manufacturing applications.

The team built a symmetrically driven nano-positioning stage using multiple parallel-bonded thin piezoelectric ceramic layers, paired with capacitive displacement sensors and voltage amplifiers. To overcome slow floating-point computation and low compilation efficiency in real-time control, they developed an field-programmable gate array–central processing unit (FPGA–CPU) dual-layer data-processing architecture, where the FPGA handles high-speed signal conversion and algorithm execution, and the CPU manages parameter tuning and status monitoring. This structure supports a sampling computation frequency up to 10 MHz for the inner loop and 100 kHz for the outer loop, enabling efficient high-precision control signal calculation.

The researchers employed feedback linearization to transform the hysteresis nonlinearity of the piezoelectric actuator system into a switched tracking error model. The NN-SORC uses a neural network to adaptively adjust critical control parameters, forming a closed-loop regulation mechanism for switching reference signals. Using Lyapunov stability theory and the average dwell-time technique, the team derived sufficient conditions to ensure the asymptotic stability of the closed-loop system under valid switching signals. They further extended the results to handle piecewise continuous references that are not twice differentiable, providing a stability criterion based on the minimum dwell time of the reference switching sequence.

Experimental verification was carried out on the established test bench with a 10 μm stroke and 140 Hz bandwidth, comparing the NN-SORC with a finely tuned proportional–integral–derivative (PID) controller and a Prandtl–Ishlinskii inverse compensation scheme. Test results indicate that the proposed method reduces tracking errors for both frequency-switching cosinusoidal and triangular reference signals across different operating frequencies within the system bandwidth. The control system maintains stable tracking during reference switching when the switching constraints are satisfied, confirming the effectiveness of the theoretical conditions.

The work integrates a high-speed hardware platform, adaptive intelligent control, and switched-system stability analysis, offering a practical solution for precision motion control in nano-positioning systems. Future work will extend the method to address dual-axis coupling effects and further enhance the performance of multi-axis nanopositioning systems used in advanced micro- and nano-fabrication processes.

The paper “Neural Network-Based Switching Output Regulation Control for High-Speed Nano-Positioning Stages,” is authored by Hongwei Sun, Ning Xing, Jiayu Zou, Yuqi Rong, Yang Shi, Han Ding, Hai-Tao Zhang. Full text of the open access paper: https://doi.org/10.1016/j.eng.2025.07.023. For more information about Engineering, visit the website at https://www.sciencedirect.com/journal/engineering.