Nowadays, radio broadcasting plays an important role in people's daily life. However, unauthorized broadcasting stations may seriously interfere with normal broadcastings and further disrupt the management of civilian spectrum resources. Since they are easily hidden in the spectrum and are essentially the same as normal signals, it still remains challenging to automatically and effectively identify unauthorized broadcastings in complicated electromagnetic environments. In this paper, we introduce the manifold regularization‐based deep convolutional autoencoder (MR‐DCAE) model for unauthorized broadcasting identification. The specifically designed autoencoder (AE) is optimized by entropy‐stochastic gradient descent, then the reconstruction errors in the testing phase can be adopted to determine whether the received signals are authorized. To make this indicator more discriminative, we design a similarity estimator for manifolds spanning various dimensions as the penalty term to ensure their invariance during the back‐propagation of gradients. In theory, the consistency degree between discrete approximations in the manifold regularization (MR) and the continuous objects that motivate them can be guaranteed under an upper bound. To the best of our knowledge, this is the first time that MR has been successfully applied in AE to promote cross‐layer manifold invariance. Finally, MR‐DCAE is evaluated on the benchmark data set AUBI2020, and comparative experiments show that it achieves state‐of‐the‐art performance. To help understand the principle behind MR‐DCAE, convolution kernels and activation maps of test signals are both visualized. It can be observed that the expert knowledge hidden in normal signals can be extracted and emphasized, rather than simple overfitting.

MR-DCAE: Manifold regularization-based deep convolutional autoencoder for unauthorized broadcasting identification

Electroencephalogram (EEG) signals, generated during the neuron firing, are an effective way of predicting such seizure and it is used widely in recent days for classifying and predicting seizure activity. But EEG signals generated during an epileptic seizure are highly nonstationary and dynamic in nature and contain very crucial information about the state of the brain. Due to this randomness, the accuracy of analysis of EEG data by conventional and visual methods is reduced drastically. This paper aims at enhancing epilepsy seizure detection using deep learning models with an FPGA implementation of the short-time Fourier transform block. Detection of seizure has been achieved in the following stages: (1) time–frequency analysis of EEG segments using STFT; (2) extraction of frequency bands and features of interest; and (3) seizure detection using convolutional neural network (CNN) and bidirectional long short-term memory (Bi-LSTM). For this work, the Bonn EEG dataset has been used. The maximum error of ~ 0.13% was encountered while the comparison of STFT output generated via proposed hardware architecture vs the output generated via simulation was done. The average classification accuracy of 93.9% and 97.2% was achieved by CNN and Bi-LSTM models, respectively, considering all frequency bands for epileptic and non-epileptic patients.

Accuracy Enhancement of Epileptic Seizure Detection: A Deep Learning Approach with Hardware Realization of STFT

This paper gives a class of novel predefined-time sliding mode surfaces based on the time-regulator function, forcing the states on them to approach the origin in a predefined time, which can be employed to replace the linear sliding mode surface that is used in the existing literatures focused on the predefined-time set stabilization problems such that the zero-error predefined-time stability can be achieved. In this way, the settling time is independent of the initial conditions and can be explicitly predefined as a specific control parameter. Later, such sliding mode surfaces are utilized to construct the nonsingular sliding mode control (SMC) for both uncertain second-order systems and disturbed Euler–Lagrange systems. The sufficient conditions for the proposed control schemes to guarantee the predefined-time stability, satisfactory capability of disturbance rejection, and nonsingularity of the control input are derived through systematic stability analysis. Finally, several numerical simulations are performed on generalized uncertain second-order systems and 2-DOF (degree of freedom) robot manipulator to show the effectiveness and the performance of the presented control schemes.

A novel sliding surface design for predefined-time stabilization of Euler–Lagrange systems

Iterative learning control (ILC) is a high performance control scenario that is widely applied to systems that repeat a given task or operation defined over a finite duration, and has been introduced to point‐to‐point motion tasks in existing work. However, its design degree of freedom has not been fully utilized to optimize performance beyond tracking accuracy in constrained conditions. The framework of point‐to‐point ILC in this article is extended within discrete linear time‐invariant (LTI) system, so as to take the tracking time instants of desired positions as changing variables. Therefore, it is possible to achieve the objective of minimizing energy while maintaining the required tracking accuracy. The multiobjective optimization problem is divided into two sub‐problems, which are solved with an iterative algorithm composed of norm‐optimal ILC approach as well as the coordinate descend method. Furthermore, the impact of model uncertainty on algorithm performance is also considered, and the iterative algorithm is further extended to capture constrained systems. The algorithm is robust to the model uncertainty and has a certain robustness to output disturbances. Finally, the validity of the proposed algorithm is verified by a twin rotor aerodynamic system (TRAS) model.

Robust point‐to‐point iterative learning control for constrained systems: A minimum energy approach

Emergency fault affected wide-area automatic generation control via large-scale deep reinforcement learning

Iterative learning control (ILC) is a high-performance technique for repeated control tasks with design postulates on a fixed reference profile and identical initial conditions. However, the tracking performance is only critical at few points in point-to-point tasks, and their initial conditions are usually trial-varying within a certain range in practice, which essentially degrades the performance of conventional ILC algorithms. Therefore, this study reformulates the ILC problem setup for point-to-point tasks and considers the effort of trial-varying initial conditions in algorithm design. To reduce the tracking error, it proposes a worst-case norm-optimal problem and reformulates it into a convex optimisation problem using the Lagrange dual approach. In this sense, a robust ILC algorithm is derived based on iteratively solving this problem. The study also shows that the proposed robust ILC is equivalent to conventional norm-optimal ILC with trial-varying parameters. A numerical simulation case study is conducted to compare the performance of this algorithm with that of other control algorithms while performing a given point-to-point tracking task. The results reveal its efficiency for the specific task and robustness against trial-varying initial conditions.

https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/iet-cta.2020.0557

Jian Li

Papers

Robust point-to-point iterative learning control with trial-varying initial conditions