This paper investigates active damping of LCL filter resonance in grid-connected inverters with only inverter current feedback control, since it only needs to sample one current to realize both current control and inverter protection. The traditional single-loop inverter current control (SLICC) can damp the LCL filter resonance actively. However, if the control delay is considered in digital control, the system stability will depend on the ratio of the LCL resonance frequency to the sampling frequency , and the valid damping region is only up to . Considering that the design region of the LCL resonance frequency is up to , the system can easily become unstable due to the LCL resonance frequency shifting. Thus, this paper proposes an improved active damping method based on SLICC, including the asymmetric regular sampling method and delay compensation method. The improved sampling method minimizes the control delay without introducing a switching ripple, and the delay compensation method further compensates for the delay effect. With a proper parameter design, the upper limit of the valid damping region is extended up to , which can cover all the possible resonance frequencies, and it has inherent robustness against grid-impedance variation. Finally, a few simulations in MATLAB/SIMULINK and experiments based on a 6 kW prototype are performed to verify the theoretical analysis.

After eliminating the speed sensor in the linear induction motor (LIM) high-performance closed-loop control system, the speed feedback information is missing in the speed closed loop. The accuracy of speed observation results is affected by changes in magnetizing inductance and primary resistance. This effect can cause significant oscillations in the results of the speed sensorless control system, preventing them from converging. An enhanced model reference adaptive system (MRAS) multi-parameter parallel identification methodology based on the interconnected second-order super-twisting algorithm (SOSTA) is proposed. To enhance the system’s dynamic performance, we designed an improvement to the MRAS observer based on the SOSTA, with a focus on the LIM state-space equation that considers dynamic edge-end effects. The impact of parameter alterations on the LIM system is examined. To improve speed observation accuracy and system stability, a two-parameter MRAS identification model was created. The Popov hyperstability principle was used to formulate control laws for these two parameters, ultimately enabling the identification of these two parameters. The identified values were fed back to the speed observation and control system, which reduces the coupling of these two parameters and speed. Simulation and hardware-in-the-loop experiments demonstrate that the observation system estimates speed accurately when these two parameters undergo abrupt changes within the rated speed range, enhancing the precision and robustness of the speed sensorless control system.

The fault detection of the suspension system in a maglev train is of great importance for its operational safety and reliability. However, in random matrix theory (RMT), the size of the random matrix direct impacts the result of the mean spectral radius (MSR). In this paper, a state-of-the-art fault detection method for suspension systems is proposed using optimized RMT. The random matrix with the largest number of eigenvalues is obtained by reshaping the original data, with the help of the auto-correlation length from the correlation analysis. Finally, the optimized MSR is applied to detect the fault. The results of the experiment illustrate that the proposed method is applicable and effective.

With the mature development of artificial intelligence technology, the application of intelligent control algorithms in control systems has become a trend to meet the high-performance requirements of modern society. This paper proposes a deep deterministic policy gradient (DDPG) controller design method based on deep reinforcement learning to improve system control performance. Firstly, the optimal control policy of the DDPG algorithm is derived from the Markov decision process and the Actor–Critic algorithm. Secondly, in order to avoid local optima in traditional control systems, the capacity and the settlement method of the DDPG experience pool are adjusted to absorb positive experience to accelerate convergence and to complete efficient training. In response, and to solve the overestimation of the Q value in DDPG, the overall structure of the Critic network is changed to shorten the convergence period of DDPG at low learning rates. Finally, a first-order inverted pendulum control system was constructed in a simulation environment to verify the control effectiveness of PID, DDPG, and improved DDPG. The simulation results reveal that the improved DDPG controller has a faster response to disturbances, smaller displacement, and angular displacement of the first-order inverted pendulum. The simulation further proves that the improved DDPG algorithm has better stability and convergence and stronger anti-interference ability and stability recovery. This control method provides a certain reference for the application of reinforcement learning in traditional control systems.