We discuss a variety of adaptive critic designs (ACDs) for neurocontrol. These are suitable learning in noisy, nonlinear, and nonstationary environments. They have common roots as generalizations dynamic programming neural reinforcement approaches. Our discussion these origins leads to an explanation three design families: heuristic programming, dual globalized (GDHP). The main emphasis is on DHP GDHP advanced ACDs. suggest two new modifications the original that currently only working...

Three networks are compared for low false alarm stock trend predictions. Short-term trends, particularly attractive neural network analysis, can be used profitably in scenarios such as option trading, but only with significant risk. Therefore, we focus on limiting alarms, which improves the risk/reward ratio by preventing losses. To predict exploit time delay, recurrent, and probabilistic (TDNN, RNN, PNN, respectively), utilizing conjugate gradient multistream extended Kalman filter training...

Theoretical predictions of circuit failure in an Electromagnetic Pulse (EMP) environment require a knowledge levels for each component the due to surge voltages or currents. For most circuits, semiconductor devices are weakest elements with respect such failure. This paper presents results extensive experimental program determine pulse power junctions. Approximately 80 different types silicon diodes and transistors were studied variations junction areas from 10-4to 10-1 cm2 widely varying...

This article presents a novel efficient experience-replay-based adaptive dynamic programming (ADP) for the optimal control problem of class nonlinear dynamical systems within Hamiltonian-driven framework. The quasi-Hamiltonian is presented policy evaluation with an admissible policy. With quasi-Hamiltonian, composite critic learning mechanism developed to combine instantaneous data historical data. In addition, pseudo-Hamiltonian defined deal performance optimization problem. Based on...

This paper uses data collected at Central and South West Services Fort Davis wind farm (USA) to develop a neural network based prediction of power produced by each turbine. The generated electric turbines changes rapidly because the continuous fluctuation speed direction. It is important for industry have capability perform this diagnostic purposes-lower-than-expected may be an early indicator need maintenance. In paper, characteristics generation are first evaluated in order establish...

This paper presents the design of an optimal neurocontroller that replaces conventional automatic voltage regulator (AVR) and turbine governor for a turbogenerator connected to power grid. The uses novel technique based on adaptive critic designs (ACDs), specifically heuristic dynamic programming (HDP) dual (DHP). Results show both neurocontrollers are robust, but DHP outperforms HDP or controllers, especially when system conditions configuration change. also shows how nonlinear systems,...

Genetic regulatory network inference is critically important for revealing fundamental cellular processes, investigating gene functions, and understanding their relations. The availability of time series expression data makes it possible to investigate the activities whole genomes, rather than those only a pair genes or among several genes. However, current computational methods do not sufficiently consider temporal behavior this type lack capability capture complex nonlinear system...

Three-phase grid-connected converters are widely used in renewable and electric power system applications. Traditionally, controlled with standard decoupled d-q vector control mechanisms. However, recent studies indicate that such mechanisms show limitations their applicability to dynamic systems. This paper investigates how mitigate restrictions using a neural network rectifier/inverter. The implements programming algorithm is trained by backpropagation through time. To enhance performance...

This paper develops optimal control protocols for the distributed output synchronization problem of leader-follower multiagent systems with an active leader. Agents are assumed to be heterogeneous different dynamics and dimensions. The desired trajectory is preplanned generated by Other follower agents autonomously synchronize leader interacting each other using a communication network. in sense that it has nonzero input so can act independently update its keep followers away from possible...

Swarm intelligence has emerged as a worthwhile class of clustering methods due to its convenient implementation, parallel capability, ability avoid local minima, and other advantages. In such applications, validity indices usually operate fitness functions evaluate the qualities obtained clusters. However, are data dependent designed address certain types data, selection different may critically affect cluster quality. Here, we compare performances eight well-known widely used indices,...

This paper investigates how to train a recurrent neural network (RNN) using the Levenberg-Marquardt (LM) algorithm as well implement optimal control of grid-connected converter (GCC) an RNN. To successfully and efficiently RNN LM algorithm, new forward accumulation through time (FATT) is proposed calculate Jacobian matrix required by algorithm. explores incorporate FATT into The results show that combination algorithms trains RNNs better than conventional backpropagation presents analytical...

This brief presents a partially model-free solution to the distributed containment control of multiagent systems using off-policy reinforcement learning (RL). The followers are assumed be heterogeneous with different dynamics, and leaders active in sense that their inputs can nonzero. Optimality is explicitly imposed solving problem not only drive agents' states into convex hull leaders' but also minimize transient responses. Inhomogeneous algebraic Riccati equations (AREs) derived solve...

This paper focuses on current control in a permanent-magnet synchronous motor (PMSM). has two main objectives: the first objective is to develop neural-network (NN) vector controller overcome decoupling inaccuracy problem associated with conventional proportional-integral-based vector-control methods. The NN developed using full dynamic equation of PMSM, and trained implement optimal based approximate programming. second evaluate robust adaptive performance against that standard under...

In this article, a data-driven distributed control method is proposed to solve the cooperative optimal output regulation problem of leader-follower multiagent systems. Different from traditional studies on regulation, adaptive internal model originally developed, which includes and observer estimate leader's dynamics. Without relying dynamics systems, we have two reinforcement learning algorithms, policy iteration value iteration, learn controller through online input state data, estimated...

In this article, we present an intermittent framework for safe reinforcement learning (RL) algorithms. First, develop a barrier function-based system transformation to impose state constraints while converting the original problem unconstrained optimization problem. Second, based on optimal derived policies, two types of feedback RL algorithms are presented, namely, static and dynamic one. We finally leverage actor/critic structure solve online guaranteeing optimality, stability, safety....

This article presents a model-based hybrid adaptive dynamic programming (ADP) framework consisting of continuous feedback-based policy evaluation and improvement steps as well an intermittent implementation procedure. results in ADP with quantifiable performance guaranteed closed-loop stability the equilibrium point. To investigate effect aperiodic sampling on communication bandwidth control algorithms, we use Hamiltonian-driven unified framework. With such framework, it is shown that there...

In this article, we consider an iterative adaptive dynamic programming (ADP) algorithm within the Hamiltonian-driven framework to solve Hamilton-Jacobi-Bellman (HJB) equation for infinite-horizon optimal control problem in continuous time nonlinear systems. First, a novel function, "min-Hamiltonian," is defined capture fundamental properties of classical Hamiltonian. It shown that both HJB and policy iteration (PI) can be formulated terms min-Hamiltonian framework. Moreover, develop ADP...