In view of the shortcomings of the survivability analysis method for networking information-centric system of systems architecture in integrity and fidelity, an analysis method of survivability based on multi-layer de...
详细信息
ISBN:
(纸本)9798331540845;9789887581598
In view of the shortcomings of the survivability analysis method for networking information-centric system of systems architecture in integrity and fidelity, an analysis method of survivability based on multi-layer dependent network is proposed. By constructing the multi-layer network structure model of networking information-centric system of systems, the essential characteristics of the architecture under attack are described and the architecture survivability index is constructed, considering the complex coupling relationship among elements and between layers. And then case analysis is carried out, showing that this method has a good performance, which lays the foundation for realizing architecture optimization and improving the survivability of the architecture.
The study uses a recurrent meta-cognitive fuzzy neural network (RMCFNN) to present an adaptive fractional order (FO) terminal sliding mode control (TSMC) method for the robust current management of active power filter...
详细信息
ISBN:
(纸本)9798331517939;9788993215380
The study uses a recurrent meta-cognitive fuzzy neural network (RMCFNN) to present an adaptive fractional order (FO) terminal sliding mode control (TSMC) method for the robust current management of active power filter (APF). By taking into account the fact that the external disturbances and parametric perturbations of the APF are bounded, a fractional order terminal sliding mode control is created. Due to an additional degree of freedom, the suggested scheme with a FO sliding surface can provide improved finite-time high-precision tracking performance as compared to the traditional TSMC approach. Next, in order to obtain an absorbing model-free feature resulting from RMCFNN, a novel observer-based FOTSMC is constructed. The construction of specialized online updating systems for the parameters and structure of RMCFNN aims to enhance the capacity to manage uncertainties. Meanwhile, Lyapunov theory can be used to obtain finite-time convergence characteristic and closed-loop stability. Ultimately, the findings of modeling and experimentation show that the suggested observer-based FOTSMC has better controlperformance than other current schemes and is simple to build using a microcontroller.
The PIM (Processing in Memory) architecture, performing MAC operations inside memory, garners attention as a next-gen deep learning processor by eliminating memory movement between memory and computation units, unlike...
详细信息
ISBN:
(纸本)9798331517939;9788993215380
The PIM (Processing in Memory) architecture, performing MAC operations inside memory, garners attention as a next-gen deep learning processor by eliminating memory movement between memory and computation units, unlike NPUs and GPUs. However, applying conventional network pruning for the same purpose faces challenges due to small memory and analog-based MAC operations in PIM. This paper proposes techniques for effective network pruning, demonstrating how weight pruning based on temperature/humidity modeling can mitigate inference noise in PIM. Additionally, it introduces a grouping-based importance metric for channel pruning applicable to any hardware. Both approaches quantitatively enhance performance in simulation, proving their efficacy.
Multifunctional hardware technologies for neuromorphic computing are essential for replicating the complexity of biological neural systems, thereby improving the performance of artificial synapses and neurons. Integra...
详细信息
Multifunctional hardware technologies for neuromorphic computing are essential for replicating the complexity of biological neural systems, thereby improving the performance of artificial synapses and neurons. Integrating ionic and spintronic technologies offers new degrees of freedom to modulate synaptic potentiation and depression, introducing novel magnetic functionalities alongside the established ionic analogue behavior. We demonstrate that magneto-ionic devices can perform as synaptic elements with dynamically tunable depression linearity controlled by an external magnetic field, a functionality reminiscent of neuromodulation in biological systems. By applying magnetic fields we significantly reduce the nonlinearity of synaptic depression, transitioning from an exponential dependence to a linear response at higher fields. Neural network simulations reveal that this magnetically induced linearity enhancement improves learning accuracy across a wide range of learning rates, which is retained after the magnetic field is removed. These findings highlight the versatility and promise of magneto-ionic devices for developing tunable synaptic elements for neuromorphic hardware.
The proposed model predicts a 1.0 second time from now based on the accelerator, brake, and steering angle. Since the proposed model predictive control requires future speed information, a Neural network, a type of ma...
详细信息
ISBN:
(纸本)9798331517939;9788993215380
The proposed model predicts a 1.0 second time from now based on the accelerator, brake, and steering angle. Since the proposed model predictive control requires future speed information, a Neural network, a type of machine learning, is used to develop a vehicle speed prediction model. Time series data from human drivers will be used as training data. The developed model will be used for model predictive control, and the performance of the model will be evaluated. In the model predictive control, the fuel consumption of the HEV vehicle before and after using the vehicle speed prediction model will be assessed and studied before the model predictive control is incorporated, and the SOC of the HEV vehicles will also be evaluated and investigated by HEV simulation of MATLAB/Simulink.
Knowledge transfer in multi-agent reinforcement learning (MARL) is crucial for improving learning efficiency and performance in various cooperative tasks. However, the large exploration space in MARL often leads to ch...
详细信息
ISBN:
(纸本)9798331517939;9788993215380
Knowledge transfer in multi-agent reinforcement learning (MARL) is crucial for improving learning efficiency and performance in various cooperative tasks. However, the large exploration space in MARL often leads to challenges in achieving efficient knowledge transfer. In this study, we introduce a multi-agent network randomization (MANR) method to enhance the generalization ability. By incorporating randomness into the agents' training data, the MANR method diversifies the state space and leads to more robust learning results, especially when applying to transfer learning. We apply this method in the StarCraft multi-agent challenge (SMAC) environment and demonstrate significant improvements in performance compared to traditional transfer learning approaches. Our results show an 7.29% increase in the test win rate with the MANR method, along with increased robustness, highlighting its effectiveness in stabilizing learning performance. These results suggest that MANR is a promising approach for robust knowledge transfer in MARL, potentially applicable to more complex and dynamic real-world scenarios.
Industrial control System cybersecurity has become an important study area after the occurrence of several mediatic events in the 2010's (Stuxnet, BlackEnergy, Industroyer). Two common characteristics of these att...
详细信息
ISBN:
(纸本)9798350369588;9798350369595
Industrial control System cybersecurity has become an important study area after the occurrence of several mediatic events in the 2010's (Stuxnet, BlackEnergy, Industroyer). Two common characteristics of these attacks are the fact that they were not violating the communication protocols being "stealth" for classical pattern-based detection methods and that they explicitly target the physical process. In this paper we study the performance and explainability of an artificial intelligence based detection system for the detection of such sophisticated attacks.
In this work, a recursive algorithm has been developed for heterogeneous network distributed systems (NDS) communicating over an undirected network to solve H-infinity optimal distributed tracking control problem of c...
详细信息
ISBN:
(纸本)9798350310375
In this work, a recursive algorithm has been developed for heterogeneous network distributed systems (NDS) communicating over an undirected network to solve H-infinity optimal distributed tracking control problem of continuoustime systems as a convex problem. Recent studies on NDS have studied the tracking control problem with decentralized performance functions defined for each subsystem in the network, on the contrary, a global performance function has been defined in this work for the whole NDS. An optimal distributed control problem has been defined as a sequential convex optimization problem benefiting off-policy reinforcement learning with sparsity constraints introduced on the symmetric positive definite matrix. Finally, the efficacy of the proposed algorithm is shown on a group of heterogeneous unmanned aerial vehicles (UAV) communicating over an undirected network.
networked controlsystems (NCSs) are an example of task-oriented communication systems, where the purpose of communication is real-time control of processes over a network. In the context of NCSs, with the processes s...
详细信息
ISBN:
(纸本)9781538674628
networked controlsystems (NCSs) are an example of task-oriented communication systems, where the purpose of communication is real-time control of processes over a network. In the context of NCSs, with the processes sending their state measurements to the remote controllers, the deterioration of controlperformance due to the network congestion can be partly mitigated by shaping the traffic injected into the network at the transport layer (TL). In this work, we conduct an extensive performance evaluation of selected TL protocols and show that existing approaches from communication and control theories fail to deliver sufficient controlperformance in realistic network scenarios. Moreover, we propose a new semantic-aware TL policy, which uses the process state information to filter the most relevant updates and the network state information to prevent delays due to network congestion. The proposed mechanism is shown to outperform all the considered TL protocols with respect to controlperformance.
The need for renewable energy in power systems is growing exponentially. Several algorithms may be used to track the Maximum Power Point (MPP) quickly and precisely. This research provides a comparison and analysis of...
详细信息
ISBN:
(纸本)9798350372113;9798350372106
The need for renewable energy in power systems is growing exponentially. Several algorithms may be used to track the Maximum Power Point (MPP) quickly and precisely. This research provides a comparison and analysis of different control techniques for the maximum power point tracking (MPPT) of a photovoltaic system subject to varying irradiance and temperature by using three algorithms which are Perturb and Observe (PO), Artificial Neural network (ANN), and Hybrid NN-PO. The three MPPT algorithms were created in a standalone photovoltaic system with a boost converter to maintain the maximum power point of the solar panel. Using MATLAB/SIMULINK software, the performance of these controllers is evaluated under varying irradiance and temperature conditions. Under the 100 (W/m2s) slope, PO's efficiency is the lowest, at 96.443% and the hybrid efficiency is nearly identical to the ANN algorithm at 99,996% and 99,997%, respectively. Based on the simulation that has been demonstrated, the Perturb and Observe (PO) algorithm exhibits the lowest performance in the simulation with time response. The Hybrid Neural network and Neural network algorithm performs better than PO. At the same time, hybrid efficiency is similar to the ANN algorithm.
暂无评论