1 January 2026 in Power Electronics and Drives
Drum-type flying shears represent a technological system used in material processing lines (MPL). During operation, they must withstand an impact shear torque of very short duration with an amplitude comparable to the rated motor torque. The position of the shears is derived from the strip speed, meaning that any change in strip speed is directly reflected in the shears’ speed. Therefore, the primary control task is to ensure precise synchronisation between the strip speed and the shears at the moment cutting begins, as well as to minimise the speed drop of the shears during the cutting process. This prevents deformation of the cut strip and deterioration of the dimensional accuracy of the cut sheets. Previous research has shown that conventional control methods supplemented with shear torque compensation can significantly improve speed stability during cutting. Building on these findings, this paper analyses the application of artificial intelligence (AI) elements to optimise the control of flying drum shears. The study compares traditional control approaches with methods based on fuzzy logic and neural networks (NNs), aiming to further improve system dynamics and minimise speed deviations during cutting. The proposed control structures and compensations are verified through simulation, and the results are presented.
1 January 2026 in Power Electronics and Drives
Modern energy systems require efficient power management, making multi-active bridge (MAB) converters ideal for electric vehicle (EV) charging and DC microgrids. However, higher port counts lead to magnetic coupling and non-linear dynamics, complicating independent control. This paper proposes a generalised power flow decoupling strategy for a five-port modular multi-active bridge (MMAB) converter using the Newton–Raphson (NR) method. Unlike traditional reduced-order models, a current-based model is employed to maintain modularity. A key contribution is the integration of the Moore–Penrose pseudoinverse into the NR solver to mitigate Jacobian singularities and ensure numerical stability. The algorithm is implemented on a dual-core central processing unit (CPU), achieving a deterministic 40 μs execution time. Furthermore, this study identifies performance degradation caused by parasitic inductances in the modular AC bus and proposes an optimised design using magnetic flux cancellation. Modelling precision is further refined through offline fast Fourier transform (FFT)-based parameter identification. Experimental validation on a five-port MMAB prototype confirms that the proposed method provides superior steady-state accuracy and robust dynamic decoupling during rapid transient load transitions.