Recurrent Neural Network Controller Research Articles

Adaptive control of recurrent neural networks using conceptors.

Recurrent neural networks excel at predicting and generating complex high-dimensional temporal patterns. Due to their inherent nonlinear dynamics and memory, they can learn unbounded temporal dependencies from data. In a machine learning setting, the network's parameters are adapted during a training phase to match the requirements of a given task/problem increasing its computational capabilities. After the training, the network parameters are kept fixed to exploit the learned computations. The static parameters, therefore, render the network unadaptive to changing conditions, such as an external or internal perturbation. In this paper, we demonstrate how keeping parts of the network adaptive even after the training enhances its functionality and robustness. Here, we utilize the conceptor framework and conceptualize an adaptive control loop analyzing the network's behavior continuously and adjusting its time-varying internal representation to follow a desired target. We demonstrate how the added adaptivity of the network supports the computational functionality in three distinct tasks: interpolation of temporal patterns, stabilization against partial network degradation, and robustness against input distortion. Our results highlight the potential of adaptive networks in machine learning beyond training, enabling them to not only learn complex patterns but also dynamically adjust to changing environments, ultimately broadening their applicability.

Optimization of solar energy using recurrent neural network controller with dc-dc boost, Cuk, and single-ended primary inductor converter (SEPIC) Converters

Optimization of solar energy using recurrent neural network controller with dc-dc boost, Cuk, and single-ended primary inductor converter (SEPIC) Converters

Optimization of solar energy using artificial neural network vs recurrent neural network controller with ultra lift Luo converter

Optimization of solar energy using artificial neural network vs recurrent neural network controller with ultra lift Luo converter

Optimization of solar energy using artificial neural network vs recurrent neural network controller with positive output super lift Luo converter

In today’s world, the need for clean energy is crucial. Historically, Renewable energy sources like hydropower, wind, and solar offer sustainable solutions. Photovoltaic (PV) systems convert sunlight into electricity using semiconductor PV cells, which have been efficient for over 30 years. PV cell efficiency depends on irradiance (solar photon intensity) and temperature. Higher irradiance increases efficiency, while higher temperatures decrease it. PV systems, despite low voltage outputs, can be optimized using DC-DC Positive Output Super Lift Luo converters to match load requirements, enhancing system efficiency. Solar irradiance varies throughout the day, affecting PV cell output. Maximum Power Point Trackers (MPPTs) adjust the system's operating point to maintain peak efficiency. This study focuses on designing AI controllers to manage MPPT. We compare the performance of Artificial Neural Networks (ANN) and Recurrent Neural Networks (RNN) using three datasets. The goal is to identify the most efficient AI controller for optimizing solar energy systems.

A Hybrid RBFNN-SPOA Technique for Multi-Source EV Power System with Single-Switch DC-DC Converter

With a single-switched DC-DC converter, a hybrid approach is proposed for a fast-charging (FC) dc to dc converter with the controller for an electric vehicle charging station (EVCS) that combines solar photovoltaic (PV), fuel cells (FC), and a battery energy storage system (BESS). The proposed hybrid technique integrates a radial-basis-function-neural-network (RBFNN) and a student-psychology-optimization-algorithm (SPOA); commonly known as the RBFNN-SPOA technique. The PV-FC hybrid electric vehicle systems, like controllers, electric vehicle, and the vehicle’s power source, are displayed. The proposed converter architectures have less voltage waveform changes, a low voltage stress on semiconductor power sources, and a high-voltage ratio of conversion. The proposed method is to use FC successfully at battery charge states and varied radiation levels. The FC in the controller of SPOA may change the inputs and outputs depending on the battery state of charge (SoC) and PV irradiation. Similarly, the needed input-torque for the motor is forecasted by the recalling recurrent neural network controller using several manually produced torque signals. The performance of the proposed technique is evaluated in MATLAB and is compared to ant colony optimization (ACO), crow optimization algorithm (COA), and fuzzy logic controller (FLC) approaches. According to the results, the efficiency of the proposed technique is 99.17%.

Eliminating Harmonics and Voltage Ripples Using Optimal Constructive Recurrent Neural Network on Boost Rectifiers with Unbalanced Grid Sources

Various new methodologies and techniques have been proposed to improve power quality. One of the parameters that affect power quality is harmonics. Voltage and current unbalance is a frequent problem. Electronic devices such as rectifiers and power distribution networks are negatively affected by this imbalance. The current in the grid will be severely distorted. The DC bus voltage will fluctuate or ripple frequently, perhaps twice as often as the frequency. Direct power control (DPC) and voltage-oriented control (VOC) are two examples of control strategies developed for rectifiers. In classical VOC, the voltage and current in the synchronous frame of reference (dq) are controlled using a proportional-integral (PI) controller. However, the PI controller in controlling the VOC current has the disadvantage that when one variable changes, the other variables will also change. PI cannot control itself independently. it is difficult to achieve an accurate linear mathematical model of the system required for PI controllers. In addition, PI controllers often struggle to work satisfactorily on parameter variations, nonlinearities, and load disturbances. in transient conditions, the Optimal Constructive Recurrent Neural Network(OCRNN) controller takes faster time to reach a stable state. when there is a change in the reference dc voltage, PI has a higher overshot voltage and a lower voltage drop than OCRNN and takes longer than the OCRNN controller.

Parallel Trajectory Training of Recurrent Neural Network Controllers With Levenberg–Marquardt and Forward Accumulation Through Time in Closed-Loop Control Systems

Parallel Trajectory Training of Recurrent Neural Network Controllers With Levenberg–Marquardt and Forward Accumulation Through Time in Closed-Loop Control Systems

An Inexact Sequential Quadratic Programming Method for Learning and Control of Recurrent Neural Networks.

This article considers the two-stage approach to solving a partially observable Markov decision process (POMDP): the identification stage and the (optimal) control stage. We present an inexact sequential quadratic programming framework for recurrent neural network learning (iSQPRL) for solving the identification stage of the POMDP, in which the true system is approximated by a recurrent neural network (RNN) with dynamically consistent overshooting (DCRNN). We formulate the learning problem as a constrained optimization problem and study the quadratic programming (QP) subproblem with a convergence analysis under a restarted Krylov-subspace iterative scheme that implicitly exploits the structure of the associated Karush-Kuhn-Tucker (KKT) subsystem. In the control stage, where a feedforward neural network (FNN) controller is designed on top of the RNN model, we adapt a generalized Gauss-Newton (GGN) algorithm that exploits useful approximations to the curvature terms of the training data and selects its mini-batch step size using a known property of some regularization function. Simulation results are provided to demonstrate the effectiveness of our approach.

Power Quality Gain of Self-Lift Luo Converter used for Hybrid Renewable Power System with Battery Grid Integration

The renewable power sources such as wind as well as photovoltaic (PV) energy are fashionable for residential and commercial applications. For the reason, that eco friendlessness plus expenditure are effective. This paper intends a Self-Lift Luo converter integrated with solar photovoltaic system to the utility grid along with its Adaptive Neuro-Fuzzy Inference System (ANFIS) and Recurrent Neural Network (RNN) control strategies are proposed. By achieve an energy management for the proposed system using Bidirectional Battery converter along with battery system. Power supervision move towards is suggested to control the DC bus voltage by way of Self-Lift Luo converter. The wind power with Doubly Fed Induction Generator (DFIG) based grid integration with a PI controller. Under this work, DSTATCOM has been used to improve the quality of power under different load conditions. The obtained results designate that the proposed approach delivers recovered performance with enhanced efficiency and nominal harmonics. The intact system is validated through a MATLAB simulation.

Salp Swarm Optimized Hybrid Elman Recurrent Neural Network (SSO-ERNN) based MPPT Controller for Solar PV

Renewable energy technologies provide clean and abundant energy that can be self-renewed from natural sources; more support from the public to replace fossil fuels with various renewable energy sources to protect the environment. Although solar energy has less impact on the environment than other renewable sources, the output efficiency is lower due to the different weather conditions. So to overcome that, the MPPT controller is used for tracking peak power and better efficiency. Some conventional methods in MPPT controllers provide less tracking efficiency, and steady-state oscillations occur in maximum power tracking due to the sudden variations in solar irradiance. Thus, in this work salp swarm optimized (SSO) based Elman recurrent neural network (ERNN) controller is proposed to track the maximum power form PV with high efficiency. The weight parameter of ERNN layer is optimized with the help of SSO, which solve the complex problems and give maximum efficiency. The proposed method is performed in MATLAB/Simulink environment, which differs from existing plans and gives a better output efficiency. Using this proposed controller, the system can achieve high tracking efficiency of 99.74% compared to conventional processes.

Modeling and Performance Analysis of a Grid-Connected Photovoltaic System with Advanced Controller considering Varying Environmental Conditions

This paper presents a mathematical model of 255 kW grid-connected solar photovoltaic (SPV) system. To study the performance characteristics of the grid-connected SPV system, a new hybrid adaptive grasshopper optimization algorithm with the recurrent neural network (AGO-RNN) control technique was implemented. Furthermore, the power quality at the point of common coupling (PCC) has been studied using the conventional (PSO) and proposed AGO-RNN controllers. The characteristics of the PV system were analyzed under varying environmental (variable irradiance and temperature) conditions considering 3 different cases such as (i) standard test conditions (STC), (ii) variable radiation with constant temperature, and (iii) variable radiation with variable temperature. For each case, the total harmonic distortion (THD) has been calculated using the proposed AGO-RNN control technique, and the results were compared with particle swarm optimization (PSO) technique. The 255 kW PV model is initially developed and connected to a three-level NPC inverter, an MPPT-based perturbation and observation algorithm. Later, the PV model is controlled by an AGO-RNN pulse width modulation (PWM) controller and is then integrated to the main grid at PCC. The main advantage of this technique is exploiting the separate DC-DC converter between the SPV module and the inverter. Finally, the proposed grid-connected SPV system was simulated on MATLAB for analyzing the performance of the system based on its I-V and P-V characteristics, inverter voltage, grid power, gird voltage, grid current, power factor, and THD under different environmental conditions. The simulation results demonstrate that the current magnitude and THD of the SPVGC system are improved with the cutting-edge AGO-RNN controller compared to PSO in all three different scenarios, and this value is less than 1.6%, which is within the permitted limits of IEC 61727 standards.

Online-Growing Neural Network Control for Dissolved Oxygen Concentration

The activated sludge method is one of the most commonly used methods for wastewater treatment process (WWTP). Among them, the dissolved oxygen (DO) concentration is a key factor affecting microbial metabolism and wastewater treatment effectiveness. However, due to the nonlinearity and dynamicity of WWTP, it is difficult to precisely control the DO concentration by traditional control methods. To solve this problem, the online-growing pipelined recurrent wavelet neural network (OG-PRWNN) control is proposed to improve the DO control accuracy. First, the online growing mechanism is designed to adjust the number of modules of the controller by measuring the control performance. Then, the structure of the controller is automatically determined to meet the different operating conditions of the WWTP. Second, the online algorithm of parameters incorporating adaptive learning rates is designed to train the OG-PRWNN to meet the control requirements. In addition, the stability of the OG-PRWNN controller is analyzed by the Lyapunov stability theorem. Finally, the performance of the controller is verified by the benchmark simulation model of WWTP. Simulation results show that the OG-PRWNN controller can obtain better control accuracy.

Neural-Observer-Based Terminal Sliding Mode Control: Design and Application

In view of the fact that most devices are nonlinear systems, this article proposes a recurrent probabilistic compensation fuzzy neural network (RPCFNN) control scheme based on global fast terminal sliding mode control (GFTSMC) for a class of nonlinear systems with uncertainties. First, GFTSMC is developed to impose a finite-time convergence feature for the considered systems. Second, a novel RPCFNN framework is designed to further enhance the ability to deal with uncertainty. Due to the added probabilistic estimation and dynamic fuzzy operator, the developed RPCFNN controller possesses superior nonlinearity handling capability. Meanwhile, the stability of RPCFNN is proved by the Lyapunov theory. Finally, active power filter is taken as the representative of nonlinear systems to verify the effectiveness and feasibility of the proposed method. Simulation and experimental results show that the novel RPCFNN controller has good steady-state and dynamic performance and has excellent robustness to deal with uncertain disturbances.

Deep-Learning-Based Complex Scene Text Detection Algorithm for Architectural Images

With the advent of smart cities, the text information in an image can be accurately located and recognized, and then applied to the fields of instant translation, image retrieval, card surface information recognition, and license plate recognition. Thus, people’s lives and work will become more convenient and comfortable. Owing to the varied orientations, angles, and shapes of text, identifying textual features from images is challenging. Therefore, we propose an improved EAST detector algorithm for detecting and recognizing slanted text in images. The proposed algorithm uses reinforcement learning to train a recurrent neural network controller. The optimal fully convolutional neural network structure is selected, and multi-scale features of text are extracted. After importing this information into the output module, the Generalized Intersection over Union algorithm is used to enhance the regression effect of the text bounding box. Next, the loss function is adjusted to ensure a balance between positive and negative sample classes before outputting the improved text detection results. Experimental results indicate that the proposed algorithm can address the problem of category homogenization and improve the low recall rate in target detection. When compared with other image detection algorithms, the proposed algorithm can better identify slanted text in natural scene images. Finally, its ability to recognize text in complex environments is also excellent.

RETRACTION NOTICE: Low-voltage ride-through capability enhancement of wind energy conversion system using an ant-lion recurrent neural network controller

RETRACTION NOTICE: Low-voltage ride-through capability enhancement of wind energy conversion system using an ant-lion recurrent neural network controller

Hybrid deep learning diagonal recurrent neural network controller for nonlinear systems

In the present paper, a hybrid deep learning diagonal recurrent neural network controller (HDL-DRNNC) is proposed for nonlinear systems. The proposed HDL-DRNNC structure consists of a diagonal recurrent neural network (DRNN), whose initial values can be obtained through deep learning (DL). The DL algorithm, which is used in this study, is a hybrid algorithm that is based on a self-organizing map of the Kohonen procedure and restricted Boltzmann machine. The updating weights of the DRNN of the proposed algorithm are developed using the Lyapunov stability criterion. In this concern, simulation tasks such as disturbance signals and parameter variations are performed on mathematical and physical systems to improve the performance and the robustness of the proposed controller. It is clear from the results that the performance of the proposed controller is better than other existent controllers.

Recurrent neural network based high-precision position compensation control of magnetic levitation system

For improving the dynamic quality and steady-state performance, the hybrid controller based on recurrent neural network (RNN) is designed to implement the position control of the magnetic levitation ball system in this study. This hybrid controller consists of a baseline controller, an RNN identifier, and an RNN controller. In the hybrid controller, the baseline controller based on the control law of proportional-integral-derivative is firstly employed to provide the online learning sample and maintain the system stability at the early control phase. Then, the RNN identifier is trained online to learn the accurate inverse model of the controlled object. Next, the RNN controller shared the same structures and parameters with the RNN identifier is applied to add the precise compensation control quantity in real-time. Finally, the effectiveness and advancement of the proposed hybrid control strategy are comprehensively validated by the simulation and experimental tests of tracking step, square, sinusoidal, and trapezoidal signals. The results indicate that the RNN-based hybrid controller can obtain higher precision and faster adjustment than the comparison controllers and has strong anti-interference ability and robustness.

Recurrent Neural Network Controllers Synthesis with Stability Guarantees for Partially Observed Systems

Neural network controllers have become popular in control tasks thanks to their flexibility and expressivity. Stability is a crucial property for safety-critical dynamical systems, while stabilization of partially observed systems, in many cases, requires controllers to retain and process long-term memories of the past. We consider the important class of recurrent neural networks (RNN) as dynamic controllers for nonlinear uncertain partially-observed systems, and derive convex stability conditions based on integral quadratic constraints, S-lemma and sequential convexification. To ensure stability during the learning and control process, we propose a projected policy gradient method that iteratively enforces the stability conditions in the reparametrized space taking advantage of mild additional information on system dynamics. Numerical experiments show that our method learns stabilizing controllers with fewer samples and achieves higher final performance compared with policy gradient.

Enhancement neural control scheme performance using PSO adaptive rate: Experimentation on a transesterification reactor

An arbitrary choice of the neural controller adaptive rate can have a negative effect on the performance of the closed-loop system. In this study, we propose a novel methodology for neural controller adaptive rate using Particle Swarm Optimization algorithm. The developed control scheme is composed of a recurrent neural networks emulator and controller with decoupled adaptive rates. Constraints on the adaptive rate are derived from the Lyapunov stability method. Particle Swarm Optimization is proposed as a mechanism to optimize the adaptive rate of the NC to improve the closed-loop performances. The advantages of the proposed new control algorithm are as follows: (1) online optimal choice of adaptive rate, which reduces the effort for searching an adequate neural controller adaptive rate when considering the conventional methods and (2) ensuring stability, faster convergence, disturbance rejection, and good tracking. The efficiency of the proposed PSO adaptive rate is demonstrated with numerical control of SISO nonlinear system. The obtained results prove the efficiency of the proposed NC compared to those obtained with existing methods. An application of the developed approach on a semi-batch reactor is presented to validate simulations results.

Improvement of voltage stability for grid connected solar photovoltaic systems using static synchronous compensator with recurrent neural network

Purpose. This article proposes a new control strategy for static synchronous compensator in utility grid system. The proposed photovoltaic fed static synchronous compensator is utilized along with recurrent neural network based reference voltage generation is presented in grid system network. The novelty of the proposed work consists in presenting a Landsman converter enhanced photovoltaic fed static synchronous compensator with recurrent neural network algorithm, to generate voltage and maintain the voltage-gain ratio. Methods. The proposed algorithm which provides sophisticated and cost-effective solution for utilization of adaptive neuro-fuzzy inference system as maximum power point tracking assures controlled output and supports the extraction of complete power from the photovoltaic panel. Grid is interconnected with solar power, voltage phase angle mismatch, harmonic and voltage instability may occur in the distribution grid. The proposed control technique strategy is validated using MATLAB/Simulink software and hardware model to analysis the working performances. Results. The results obtained show that the power quality issue, the proposed system to overcome through elimination of harmonics, reference current generation is necessary, which is accomplished by recurrent neural network. By recurrent neural network, the reference signal is generated more accurately and accordingly the pulses are generated for controlling the inverter. Originality. Compensation of power quality issues, grid stability and harmonic reduction in distribution network by using photovoltaic fed static synchronous compensator is utilized along with recurrent neural network controller. Practical value. The work concerns the comparative study and the application of static synchronous compensator with recurrent neural network controller to achieve a good performance control system of the distribution network system. This article presents a comparative study between the conventional static synchronous compensator, static synchronous compensator with recurrent neural network and hardware implementation with different load. The strategy based on the use of a static synchronous compensator with recurrent neural network algorithm for the control of the continuous voltage stability and harmonic for the distribution network-linear as well as non-linear loads in efficient manner. The study is validated by the simulation results based on MATLAB/Simulink software and hardware model.