Network Controller Research Articles

Q-learning based model-free input-output feedback linearization control method⋆

Input-output feedback linearization is a nonlinear control method that relies on a precise dynamical model. Combining Q-learning techniques, an input-output feedback linearization correction framework is presented to accomplish model-free feedback linearization of affine nonlinear systems in order to tackle the problem caused by the unknown dynamics model. This framework formulates a model reference tracking control problem that guides the input-output relationship of the nonlinear system into a linear relationship. Due to the two Lie derivative terms present in the feedback linearized controller, the controller is designed as a dual network structure. To overcome the issue of coupling in the dual-network controller, a model-free Q-learning method is presented to solve the unknown controller network weights. The proposed method is experimentally validated on a single-link flexible joint manipulator system, and the resultant linearized system exhibits dynamics similar to the desired linear system in a new tracking task, proving the effectiveness of the proposed method.

Role of glycosphingolipid biosynthesis coregulators in malignant progression of thymoma.

As the most common malignancy from mediastinum, the metabolic reprogramming of thymoma is important in its development. Nevertheless, the connection between the metabolic map and thymoma development is yet to be discovered. Thymoma was categorized into three subcategories by unsupervised clustering of molecular markers for metabolic pathway presentation in the TCGA dataset. Different genes and functions enriched were demonstrated through the utilization of metabolic Gene Ontology (GO) analysis. To identify the main contributors in the development of thymic malignancy, we utilized Gene Set Enrichment Analysis (GSEA), Gene Set Variation Analysis (GSVA), and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. The prognosis of thymoma was evaluated by screening the essential pathways and genes using GSVA scores and machine learning classifiers. Furthermore, we integrated the transcriptomics findings with spectrum metabolomics investigation, detected through LC-MS/MS, in order to establish the essential controller network of metabolic reprogramming during thymoma progression. The thymoma prognosis is related to glycosphingolipid biosynthesis-lacto and neolacto series pathway, of what high B3GNT5 indicate poor survival. The investigation revealed that glycosphingolipid charts have a significant impact on metabolic dysfunction and could potentially serve as crucial targets in the clinical advancement of metabolic therapy.

Gaussian PI Controller Network Classifier for Grid-Connected Renewable Energy System

Multi-port converters are considered as exceeding earlier period decade owing to function in a combination of different energy sources in a single processing unit. Renewable energy sources are playing a significant role in the modern energy system with rapid development. In renewable sources like fuel combustion and solar energy, the generated voltages change due to their environmental changes. To develop energy resources, electric power generation involved huge awareness. The power and output voltages are plays important role in our work but it not considered in the existing system. For considering the power and voltage, Gaussian PI Controller-Maxpooling Deep Convolutional Neural Network Classifier (GPIC-MDCNNC) Model is introduced for the grid-connected renewable energy system. The input information is collected from two input sources. After that, input layer transfer information to hidden layer 1 fuzzy PI is employed for controlling voltage in GPIC-MDCNNC Model. Hidden layer 1 is transferred to hidden layer 2. Gaussian activation is employed for determining the output voltage with help of the controller. At last, the output layer offers the last value in GPIC-MDCNNC Model. The designed method was confirmed using one and multiple sources by stable and unpredictable input voltages. GPIC-MDCNNC Model increases the performance of grid-connected renewable energy systems by enhanced voltage value compared with state-of-the-art works. The control technique using GPIC-MDCNNC Model increases the dynamics of hybrid energy systems connected to the grid.

An Advanced Artificial Neural Network Energy Management in Standalone PV Systems

With the ever-increasing prevalent power crisis and pollution of the environment, solar power, has attracted greater attention as a new and clean energy source. It provides an alternative solution for isolated sites with an unavailable grid connection. However, it is not without any drawbacks, mainly its intermittent nature, related primarily owing to its reliance on meteorological variables such as the temperature outside and the amount of sunlight. In effect, the PV systems that produced electrical energy could well display an electricity excess or deficit at the loads level, likely to result in system service discontinuity. In this respect, the present paper is designed to provide an intelligent management strategy to PV station owners with a dump load. It can involve serving two customers simultaneously according to the following scenarios: the PV production installation of the customer1 is greater than their required load; however, the customer2's neighboring station does not have enough power to cover its electrical load. This case brings electrical energy from the initial station to make up for the shortfall, and vice versa. Lithium-ion batteries step in the case when the essential electrical power cannot be delivered either by the local station or the neighboring one or to keep the accumulated power excess. If one of the stations (1 or 2) detects a power surplus and the batteries are completely charged, the generated power excess must be redirected to a secondary load, commonly known as the dump load. Relying on the artificial neural network controller, the suggested exchange control is used for two independent PV-battery stations with dump load. The MATLAB/Simulink attained simulation turns out to demonstrate the advanced controller’s noticeable performance and effectiveness in managing the standalone PV system’s operability in terms of continuous electrical energy delivery flow to the resistive load while reducing power waste and increasing the lithium-ion battery lifespan.

Adaptive tracking control for a class of nonlinear systems with input dead‐zone and actuator failure

In this paper, the adaptive tracking problem for a class of strictly feedback nonlinear time‐delay systems with output constraints, actuator failures, and input dead zones is studied. First, in the design of controller, the absolute value function and integral term are introduced into the candidate functions to deal with nonlinear factors and distributed time‐varying delays in the system. In addition, an adaptive neural network controller is designed to ensure the stability of the controlled system, and the system tracking error can converge to the origin without violating the constraints. Finally, simulation results verify the feasibility of the proposed control strategy.

Cascade feedforward neural network and deep neural network controller on photovoltaic system with cascaded multilevel inverters: Comparison on standalone and grid integrated system

The introduction of a micro-grid-based power generation network will help to meet the demands of consumers while reducing environmental impact. Several industrialized and emerging countries allocate considerable resources to renewable energy-based power generation and invest significant sums of money in this area. This study examines the challenges involved with electricity generation through photovoltaic (PV) systems and the integration of the same with the grid to mitigate power quality issues and improve the power factor for various loading conditions. An innovative multilayer inverter for grid-connected PV systems has been developed to enhance the voltage profile and resulted in a drop in total harmonic distortion (THD). A cascade multilevel inverter (associated with a grid-integrated PV system and managed using multiple innovative artificial intelligence controllers) was developed in this research project. Various advanced intelligent controllers, such as cascade feedforward neural networks (CFFNN) and deep neural networks (DNN), have been analyzed under various operating situations and observed that the THD of voltage, current at the grid, and the load is less than 3 % as per the IEEE 519 standards along with this power factor is maintained nearly unity for the grid-connected system. The quality of power in terms of voltage, frequency, total harmonics distortion, and power factor are improved by using a novel deep neural network algorithm in a cascaded multilevel inverter and verified according to IEEE 1547 and IEEE 519 standards to determine the efficacy of the proposed system.

Do Deep Reinforcement Learning Agents Model Intentions?

Inferring other agents’ mental states, such as their knowledge, beliefs and intentions, is thought to be essential for effective interactions with other agents. Recently, multi-agent systems trained via deep reinforcement learning have been shown to succeed in solving various tasks. Still, how each agent models or represents other agents in their environment remains unclear. In this work, we test whether deep reinforcement learning agents trained with the multi-agent deep deterministic policy gradient (MADDPG) algorithm explicitly represent other agents’ intentions (their specific aims or plans) during a task in which the agents have to coordinate the covering of different spots in a 2D environment. In particular, we tracked over time the performance of a linear decoder trained to predict the final targets of all agents from the hidden-layer activations of each agent’s neural network controller. We observed that the hidden layers of agents represented explicit information about other agents’ intentions, i.e., the target landmark the other agent ended up covering. We also performed a series of experiments in which some agents were replaced by others with fixed targets to test the levels of generalization of the trained agents. We noticed that during the training phase, the agents developed a preference for each landmark, which hindered generalization. To alleviate the above problem, we evaluated simple changes to the MADDPG training algorithm which lead to better generalization against unseen agents. Our method for confirming intention modeling in deep learning agents is simple to implement and can be used to improve the generalization of multi-agent systems in fields such as robotics, autonomous vehicles and smart cities.

DC BOOST CONVERTER USING A FUZZY CONTROL SYSTEM, SIMULATED IN SIMULINK

Decisive for the operation of voltage, current and power converters is the choice of control method. Control of complex dynamic systems in conditions of insufficient or fuzzy information requires the involvement of non-standard approaches during the construction of the control system. There are two possible ways out of the traditional framework of linear control. The first, is to develop more accurate nonlinear models on which high-performance controller design can be based. However, when choosing this way, one has to develop very complex control algorithms, and, consequently, the controller will perform complex mathematical calculations, which can have a negative impact on the control speed and response. Another way is to use a neural network controller based on fuzzy logic. Fuzzy logic is a generalization of mathematical logic and set theory, which is a function of the object belonging to a set and can take any value in the range [0; 1], not only 0 or 1. In the case of fuzzy logic we don't need simulation, because the whole work on designing of the controller is reduced to the transformation of the rules included in the fuzzy logic into the algorithm of automatic control.

О выборе архитектуры нейрорегулятора

In practice, the choice of the type of neural network is carried out empirically based on an experience of an investigator and many training attempts. At the same time, the excessive complexity of the neural network leads to an increase in its training time, and in some cases, to the impossibility of learning at all. Thus, the justification of the choice of an artificial neural network structure and/or its preliminary calculation based on other models is an urgent task. An equally important task is the choice of an initial weighting coefficients of an neural network, the choice of which determines the speed of convergence of search algorithms. This paper demonstrates several approaches to solving the problem of choosing an architecture and initializing a weighting coefficients of a neural network. One of them is carried out on the basis of a previously calculated function using Petri nets. This approach is demonstrated for solving various tasks, which include the implementation of functions using previously defined neural network models of the simplest logical operations "and", "or", etc. An approach is given that allows optimizing an architecture of a neural network that solves the problem of approximating functions of one and several variables. The principle of determining an architecture and initial weight coefficients is also used in the tasks of training neural networks with reinforcement. A separate section is devoted to the formation of a methodology for determining an architecture and initialization of a weighting coefficients of a neural network of the controller based on information about the controller obtained by a modal method using a polynomial matrix decomposition of a system. The problem of synthesis of a neural network controller for an object model containing nonlinearities and nonparametric uncertainties in the control channel is solved.

Design and Simulation of a PID Neural Network Controller for PMDC Motor Speed and Position Control

Direct current (DC) motors have many difficulties when controlling angular velocity in a variety of applications. The perfect controller cannot be carried out by traditional control alone due to the nonlinear properties of DC motors, design constraints, and mechanical variations caused by the operation conditions. This study proposes a design for an artificial neural network based PID controller (ANNPID) to control the speed of a permanent magnet DC motor (PMDC) in two methods. A detailed analysis is performed based on the simulation results of both methods. The proposed controllers are numerically simulated for various test conditions including; set-point changes, step changes in the load torque, and parameter variations, then the suggested techniques were compared in a comparative study with a traditional PID controller based on the transient response specifications and the performance indices to validate the performance of the controllers. The simulation results demonstrated that the controllers have improved dynamics, static performance, and less overshoot. The methods described here achieve control more effectively than the conventional control approaches under both nominal and disturbed test conditions over different operating ranges.

LSTM‐based adaptive robust nonlinear controller design of a single‐axis hydraulic shaking table

AbstractThe shaking table has been used extensively in the structure test field to verify the structure's performance against various vibrations, for example earthquakes. In order to replicate the vibrations, which are measured by the acceleration signal, the model of the shaking table should be thoroughly constructed to design the controller. However, parametric uncertainty and strong nonlinearity, such as the nonlinear friction, make it an obstacle to obtaining an accurate model. A neural network‐based controller, specifically a long short‐term memory neural network‐based neural network controller, is designed in this paper to address this issue. The nonlinear systems are estimated by the neural network's universal approximation characteristics, and a long short‐term memory neural network is utilized to optimize the time‐series‐related errors. Furthermore, a robust sliding mode controller is utilized to compensate for the residual error of the neural network and other uncertainties. The semi‐global asymptotic stability of the controller is proved by Lyapunov analysis. Comparative experimental results indicate the superiority of the proposed controller.

Path Tracking for Car-like Robots Based on Neural Networks with NMPC as Learning Samples

In the field of path tracking for car-like robots, although nonlinear model predictive control (NMPC) can handle the system constraints well, its real-time performance is poor. To solve this problem, a neural network control method with NMPC as the learning sample is proposed. The design process of this control method includes establishing the NMPC controller based on the time-varying local model, generating learning samples based on this NMPC controller, and training to obtain the neural network controller. The proposed controller is tested by a joint simulation of MATLAB and Carsim and compared with other controllers. According to the simulation results, the accuracy of the NN controller is close to that of the NMPC controller and far better than that of the Stanley controller. In all simulations, the absolute value of displacement error of the NN controller does not exceed 0.2854 m, and the absolute value of heading error does not exceed 0.2279 rad. In addition, the real-time performance of the NN controller is better than that of the NMPC controller. The maximum time cost and average time cost of the NN controller are, respectively, 40.91% and 22.37% smaller than those of the NMPC controller under the same conditions.

Linear Active Disturbance Rejection Control-Based Diagonal Recurrent Neural Network for Radar Position Servo Systems with Dead Zone and Friction

This paper proposes a control scheme for the radar position servo system facing dead zone and friction nonlinearities. The controller consists of the linear active disturbance rejection controller (LADRC) and diagonal recurrent neural network (DRNN). The LADRC is designed to estimate in real time and compensate for the disturbance with vast matched and mismatched uncertainties, including the internal dead zone and friction nonlinearities and external noise disturbance. The DRNN is introduced to optimize the parameters in the linear state error feedback (LSEF) of the LADRC in real time and estimate the model information, namely Jacobian information, of the plant on-line. In addition, considering the Cauchy distribution, an adaptive tracking differentiator (ATD) is designed in order to manage the contradiction between filtering performance and tracking speed, which is introduced to the LADRC. Another novel idea is that the back propagation neuron network (BPNN) is also introduced to tune the parameters of the LADRC, just as in the DRNN, and the comparison results show that the DRNN is more suitable for high precision control due to its feedback structure compared with the static BPNN. Moreover, the regular controller performances and robust performance of the proposed control approach are verified based on the radar position servo system by MATLAB simulations.

Design and analysis of PV fed high-voltage gain DC-DC converter using PI and NN controllers

The entire globe is now focused on generating power using solar photovoltaic (SPV) cells. The SPV is generating the DC power, by using power converters to satisfy the load requirement. In this article, a modified High-Voltage DC-DC converter was explained. The main intention of the modified converter is to decrease the reverse recovery currents and voltage stress across the switch, similarly, it generates high voltage at the converter output. The modified converter contains mainly-one diode, two capacitors, and Coupled Inductor (CI). When the converter switch comes under active state two capacitors are discharged in series, similarly when the converter switch under inactive state those two capacitors are charged in parallel through the coupled inductor (CI) energy so, to get a high voltage at the converter output side. To reuse the leakage-inductor (LI) energy of the CI with the help of a passive clamp circuit for reducing the voltage stress across the switch, similarly switch has less resistance so conduction losses also decreased. In this manner, the converter efficiency was improved and the diode recovery problem also solved. For the converter duty cycle controlling, Proportional-Integral (PI) and Neural Network (NN) controllers were used. The performance and analysis of the modified converter were described in detail with the help of both controllers. Here, 48 V is doubled to 400 V. By modifying the converter's parameters, its behavior is evaluated. In this converter, voltage spikes and reverse recovery currents should be minimal. The converter's switching pulse can be adjusted by PI or NN. The entire system was designed in MATLAB/ Simulink tool.

A neuro‐evolutionary approach for software defined wireless network traffic classification

AbstractAccurate network traffic classification is an essential and challenging issue for wireless network management and survivability. Existing network traffic classification algorithms, on the other hand, cannot meet the required specifications of real networks' in terms of user privacy control overhead, latency, and above all, classification speed. For wireless network traffic classification, machine learning‐based and hybrid optimization techniques have been deployed. This paper takes a software‐defined wireless network (SDWN) architecture for network traffic classification into account. Because the proposed scheme is perfectly contained within the network controller,the SDWN controller's higher processing capability, global visibility, and programmability can be used to achieve real‐time, adaptive, and precise traffic classification. In this paper, a neuro‐evolutionary approach is proposed in which the feed forward neural network (FFNN) is the base classifier and particle swarm optimization (PSO) is used to train the FFNN to accurately classify traffic while minimizing communication overhead between the controller and the SDWN switches. Simulation experiments were conducted by acquiring real‐world internet datasets to test the efficacy of the proposed scheme. The results and the state‐of‐the‐art comparisons show that the proposed approach has outperformed in terms of accuracy in wireless traffic classification.

Intent-Driven Security Policy Management for Software-Defined Systems

Different network controllers are utilized in a multi-domain software-defined systems (SDx) to manage the networking resources. However, these controllers operate using a different high-level language (intent). Thus, the admin needs to perform cross-layer translation from the user requirements to the underlying network controller format, increasing human-in-the-loop overhead. There are two primary security and management challenges involved in managing multi-domain controllers. The first challenge is how to design an SDN controller language that can effectively convert human-specified networking policies at the control plane into the network flow rules level at the data plane. The second challenge is how to reduce the complexity of network flow rules conflict checking at the data plane. To address these challenges, we present a new intent-based security policy enforcement solution called INTPOL. First, INTPOL provides a unified intent rules that abstracts the network admin from the underlying network controller’s format. Second, INTPOL develops a networking service solution to use a bounded formal model for network service compliance checking that significantly reduces the complexity of flow rules conflicts checking at the data plane level. Finally, INTPOL is expendable from a single SDN domain to multiple SDN domains and hybrid networks by applying network service function chaining (SFC) for inter-domain policy management.

Guest Editorial: Network Applications Software for Vertical IoT Industry

Network Applications is defined as a set of services that provide certain functionalities to the verticals and their associated use cases. In practice, this software piece consumes the exposed Application Programmable Interfaces (APIs) from the network, e.g., northbound APIs of 5G core network and RAN Intelligent Controller (RIC), and edge computing APIs. Considering the levels of interaction and trust, Network Applications can either be integrated with (a part of) vertical application or exposes its APIs (e.g., service APIs) to be further consumed by vertical applications. In short, Network Applications could be viewed as a separate middleware layer, which is contributed by both network operators and third parties, to simplify the vertical system deployments on a large scale and extend to vertical IoT industries.

Development of neural network controller for control of debutanization process in rectification column of gas-fractionating plant

In process systems, stabilizing parameters and bringing the process to technological mode is a rather difficult task. Control and control of parameters in manual mode does not produce effective results, and for this reason the introduction of an improved control system is required to eliminate human factor errors. Currently, the best efficiency is shown by neural networks, which are used for an advanced control and prediction system. The purpose of the study: in this project, a neural network controller is being developed to control the debutanization process of the gas fractionation plant in the Matlab Simulink environment. Cascade neural networks consisting of a certain number of layers and neurons are used to develop the project. During the work, a cascade neural network was selected, in which the parameters of the technological system were trained.

Deep Reinforcement Learning-Based QoS Optimization for Software-Defined Factory Heterogeneous Networks

The routing algorithm based on a single routing metric parameter is difficult to meet the Quality of Services (QoS) requirements of multi-source data flows in the software-defined factory heterogeneous network, resulting in link congestion and waste of network resources. To solve the problem, this paper proposes a QoS optimization method for software-defined factory heterogeneous networks based on the double deep Q network (DDQN). First, a QoS optimization architecture is proposed for software-defined factory heterogeneous networks, and the optimization function for network latency and load balancing is established. Then, the DDQN algorithm is used to obtain the optimal paths of multi-source data flows, and the optimal path is uniformly issued by the software-defined network controller. Experiments showed that the proposed algorithm has good convergence and generalizability. Compared with the existing algorithms, it outperforms in network latency, network jitter, and network throughput, improving network load balancing and routing efficiency. This research is able to provide a solution to the QoS optimization and dynamic traffic scheduling in smart heterogeneous networks and provide a technical reference point for realizing customized smart factories.

Power oscillation damping control using PI-neuron network controller for distributed generator grid connection with MVDC

Distributed generator (DG) connection to the system with the DC grid called medium voltage direct current (MVDC) grid connection has received attention and gradually integrated into the distribution grid. The linear controller, such as the PI controller, usually uses the MVDC grid control for power oscillation damping. The PI controller is limited and does not show satisfactory results when the load and parameters of the system itself are changed. This paper proposes the PI with neuron network (NN) as a feed-forward controller (PINNF) to improve the control of the DG grid connection with the MVDC. The proposed PINNF controller is applied to control the power oscillation damping. Since the NN can estimate the proper feed-forward control signals in each situation to the control system, the proposed PINNF controller performs better than the conventional PI controller. The effectiveness of the proposed PINNF controller is validated using nonlinear dynamic simulations on the MATLAB/Simulink program. Four case tests are presented and discussed in this paper. Results indicate the improvement of power oscillation damping stability and performance in the MVDC grid connection with the proposed PINNF controller.