Intelligent Control Architecture Research Articles

Adaptive active inceptor design under shared control architecture for nonlinear pilot-induced oscillations

This paper presents a Shared Control Architecture (SCA) between a human pilot and a smart inceptor for nonlinear Pilot Induced Oscillations (PIOs), e.g., category II or III PIOs. One innovation of this paper is that an intelligent shared control architecture is developed based on the intelligent active inceptor technique, i.e., Smart Adaptive Flight Effective Cue (SAFE-Cue). A deep reinforcement learning approach namely Deep Deterministic Policy Gradient (DDPG) method is chosen to design a gain adaptation mechanism for the SAFE-Cue module. By doing this, the gains of the SAFE-Cue will be intelligently tuned once nonlinear PIOs triggered; meanwhile, the human pilot will receive a force cue from the SAFE-Cue, and will consequently adapting his/her control policy. The second innovation of this paper is that the reward function of the DDPG based gain adaptation approach is constructed according to flying qualities. Under the premise of considering failure situation, task completion qualities and pilot workload are also taken into account. Finally, the proposed approach is validated using numerical simulation experiments with two types of scenarios: lower actuator rate limits and airframe damages. The Inceptor Peak Power-Phase (IPPP) metric is adopted to analyze the human-vehicle system simulation results. Results and analysis show that the DDPG based sharing control approach can well address nonlinear PIO problems consisting of Categories II and III PIO events.

Intelligent direct thrust control for multivariable turbofan engine based on reinforcement and deep learning methods

This paper proposes a novel intelligent direct thrust control (IDTC) architecture for a two-spool turbofan engine, in which the thrust is directly controlled by the IDTC architecture with two control variables instead of transforming thrust command to other commands. To realize thrust control, the unmeasurable thrust should be estimated first. A dynamic thrust model (DTM) is proposed to characterize the strong non-linearity of the engine. With the dynamic characteristic data set generation method, the date diversity increases and the thrust under random engine dynamics can be estimated more accurately. For the multi-variable optimal thrust controller, reinforcement learning (RL) method proximal policy optimization (PPO) algorithm is applied in the intelligent controller (IC) design process. Considering the engine characteristics, the reward system and the controller structure are redesigned to realize thrust optimal control and limit protection. The network used in both the DTM and the IC designing are the long short-term memory (LSTM) recurrent neural network (RNN), which can better process the engine time series and get better performance and accuracy. As the traditional evaluation criteria are too conservative to evaluate the overall ability of the proposed IDTC architecture in the full envelope, new criteria are proposed. Numerical simulations shows the satisfactory thrust estimation and control performance of the proposed IDTC architecture in the full envelope above the idle, and verify the superiority against other methods. Also, the strategy selection of the IC under extreme control tasks also shows the intelligence of the controller and gives new ideas for the engine control law design.

Key Technology and Application of Production Scheduling for Heating Forming of Forgings: A Review

Forgings represented by rocket body rings, engine casings, vehicle drive shafts, etc., are key components of important equipment in the fields of aerospace, automobiles and high-speed rail. In recent years, with the rapid development of the manufacturing industry, it is also facing an increasingly fierce international competition environment. In order to adapt to the new production mode and quickly respond to the changing market demand, forging enterprises urgently need a reasonable and efficient forging production scheduling method, and based on the excellent production organization, in order to further build an intelligent production line, develop a forging production intelligent management and control integration architecture. This paper analyzes the production scheduling problem of forging thermoforming from two aspects: forging production line scheduling and forging production workshop scheduling. The research progress of optimization objectives and optimization algorithms of production scheduling is systematically reviewed. The subsystems serving the production and processing process and the intelligent management and control architecture based on system integration are summarized. It is of great significance to effectively reduce production costs, improve product quality, realize energy saving and emission reduction in the production process and promote further intelligent upgrading of the forging industry through production scheduling and intelligent management and control of the heating and forming process of forgings.

Multiagent Reinforcement-Learning-Aided Service Function Chain Deployment for Internet of Things

Nowadays, the compelling applications of the Internet of Things (IoT) bring unexpected economic benefits to our daily lives. But at the same time, it also poses huge challenges to service providers. Diverse proprietary hardware (i.e., firewall and code conversion) have to be deployed in networks for meeting different applications’ requirements. Recently, network functions virtualization (NFV) is considered a promising technique. In the NFV-enabled architecture, network services can be implemented via a set of orderly virtual network functions (VNFs) on standardized compute nodes, which is termed service function chains (SFCs). However, with the explosion of IoT applications, embedding multiple SFCs in a shared NFV-enabled infrastructure becomes a challenging problem. Centralized schemes suffer from the scalability and private issue, while distributed schemes suffer from the nonconvergence problem. In this article, we propose a hybrid intelligent control architecture, which adopts the centralized training and distributed execution paradigm. A centralized critic is introduced to ease the training process of the distributed network nodes. Besides, considering the competitive behavior of users, we formulate the resource allocation problem as a multiuser competition game model. Based on this, we proposed a multiagent reinforcement learning-based SFCs deployment algorithm.

Deep Reinforcement Learning for Integrated Non-Linear Control of Autonomous UAVs

In this research, an intelligent control architecture for an experimental Unmanned Aerial Vehicle (UAV) bearing unconventional inverted V-tail design, is presented. To handle UAV’s inherent control complexities, while keeping them computationally acceptable, a variant of distinct Deep Reinforcement Learning (DRL) algorithm, namely Deep Deterministic Policy Gradient (DDPG) is proposed. Conventional DDPG algorithm after being modified in its learning architecture becomes capable of intelligently handling the continuous state and control space domains besides controlling the platform in its entire flight regime. Nonlinear simulations were then performed to analyze UAV performance under different environmental and launch conditions. The effectiveness of the proposed strategy is further demonstrated by comparing the results with the linear controller for the same UAV whose feedback loop gains are optimized by employing technique of optimal control theory. Results indicate the significance of the proposed control architecture and its inherent capability to adapt dynamically to the changing environment, thereby making it of significant utility to airborne UAV applications.

OPTIMIZATION OF OPERATIONAL CONTROL OF AUTONOMOUS PHOTO-DIESEL POWER SUPPLY SYSTEM WITH DC BUS

Link for citation: Lukutin B.V., Muravyev D.I. Optimization of operational control of autonomous photo-diesel power supply system with DC bus. Bulletin of the Tomsk Polytechnic University. Geo Аssets Engineering, 2022, vol. 333, no. 4, рр. 224-235. In Rus.
 The relevance. An increase in the contribution of renewable energy sources to the generation of the autonomous hybrid energy industry is important in order to reduce the environmental impact of diesel generator plants and decrease operating costs in the production of electrical energy in smart microgrids with distributed generation, including so-called «green» generation. This issue has the highest relevance for cell towers of mobile communication, rotational residential camps, meteorological stations and other remote consumers, the rated capacity of which is tens to hundreds of kW. Objective: the development of computer models for optimizing the management of the operational staff and modes of an autonomous photo-diesel power supply system with a DC bus, so as to make it possible to minimize the operating time of a diesel generator plant by increasing the contribution of the photovoltaic plant at optimal technical and economic indicators. The use of a DC bus for the electricity integration from distributed energy sources is considered promising in reducing the number of hours of inefficient operation of a diesel generator plant, which will make it possible to save diesel fuel and extend the life expectancy of the diesel generator. Novelty. This work expands the known options for modeling and optimizing the modes of a standalone hybrid DC power plant in terms of using an objective function with appropriate restrictions, adapting models of power equipment for the intended purposes of modeling, presenting the structure of a feasibility study based on commercially available characteristics of the elements of the power supply system. Methods. The study entails the development of computer models of the intelligent control architecture of a standalone hybrid power plant including a photovoltaic plant, a diesel generator plant, an electric energy storage system, a remote consumer, as well as auxiliary converter devices. Modeling of control processes is implemented in the Stateflow library, as well as by using the language syntax of the Matlab high-level package, the capabilities of which are acceptable for the intended purposes. Results. The analysis of the results shows that DC PV-diesel power supply systems can be cost-effective in all scenarios that include different types of batteries while for systems without power storage, DC distribution is often not cost-effective. The results will be of interest to specialists developing or operating standalone power supply systems and organizations planning to upgrade existing diesel power plants.

Network Intelligent Control and Traffic Optimization Based on SDN and Artificial Intelligence

For telecom operators, it is of great significance to employ artificial intelligence (AI) and big data technology in a software-defined network (SDN) in order to achieve intelligent network control, traffic management and optimization. This paper proposes a solution for intelligent work control and traffic optimization. This paper is mainly focused on SDN-based network traffic algorithm optimization and experimental verification. In this paper, we design a network control mechanism for network intelligent control as well as solutions for traffic optimization based on SDN and artificial intelligence. We analyze operators’ network requirements (e.g., the carrying of the 5th generation mobile network (5G) service, multi-protocol label switching virtual private networks optimization, cloudification of services and the IP backbone network). Then, we propose an intelligent network control architecture based on SDN and artificial intelligence. The proposed architecture consists of three modules, including a network status collection/perception module, an AI intelligent analysis module and an SDN controller module. Moreover, this paper also analyzes the objects of traffic optimization as well as routing calculation algorithms (e.g., the greedy algorithm, the top-k-shortest paths (KSP) algorithm) and routing optimization algorithms (e.g., particle swarm optimization, simulated annealing and genetic algorithms). In addition, we also put forward three optimization algorithms for the operator’s network, namely, network congestion control and prevention algorithms, resource preemption algorithms and balance of the entire network traffic algorithms. Then, we propose optimization algorithms for the above three objectives of operator network optimization, respectively. Finally, we conduct large-scale experiments to verify the effectiveness of the control mechanism and algorithms. The experimental results demonstrate that the use of SDN and artificial intelligence in operator networks can realize network intelligent control and traffic optimization more intelligently.

In-Network Intelligence Control: Toward a Self-Driving Networking Architecture

The past few years have witnessed the compelling applications of the Internet of Things (IoT) in our daily life. Meanwhile, with the explosion of IoT devices and various applications, the expectations for the performance, reliability, and security of networks are greater than ever. The current end-host-based or centralized control framework incurs too much communication and computation overhead, therefore exhibiting tardiness and clumsiness in responding to network dynamics. Recently, with the advancement of programmable network hardware, it is possible to implement network functions inside the network. However, current in-network schemes are largely dependent on the manual process, which presents poor scalability and robustness. Therefore, in this article, we present a new intelligent network control architecture, in-network intelligence control. We design intelligent in-network devices that can automatically adapt to network dynamics by leveraging powerful machine learning adaptive abilities. In addition, to enhance the collaboration among distributed in-network devices, a centralized management plane is introduced to ease the training process of distributed switches. To demonstrate the technical feasibility and performance advantage of our architecture, we present three use cases: in-network load balance, in-network congestion control, and in-network DDoS detection.

Toward intelligent continuous assistance

Technology supported assistance is a research area dedicated to support both older adults and, at some level, their caregivers in a variety of situations and contexts. A number of projects doing detailed evaluation both with robots and/or ICT-based intelligent devices have identified as open challenges the need to guarantee both continuity and variability of service according to context interpretation. This paper starts from the willingness to study how both continuity and variability can be pursued by leveraging and integrating results from research areas like artificial intelligence (AI), cognitive systems, psychology and sensor networks. Some of these technological skills are needed for example by an assistive robot and still represent open challenges in AI. This paper presents a medium term research initiative aiming at synthesizing an enhanced (intelligent) control architecture for assistive robots that take advantage from the continuous flow of information provided by a sensor network. The paper presents two main results: (a) starting from the analysis of requirements coming from the real world, it envisages a conceptual cognitive architecture highlighting the functional requirements and the key capabilities characterizing an “ideal” intelligent assistive robot; (b) it presents a prototype of a testbed architecture called KOaLa (Knowledge-based cOntinuous Loop) which integrates sensor data representation, knowledge reasoning and decision making capabilities showing its novelty in a realistic scenario.

An Intuitive End-to-End Human-UAV Interaction System for Field Exploration.

This paper presents an intuitive end-to-end interaction system between a human and a hexacopter Unmanned Aerial Vehicle (UAV) for field exploration in which the UAV can be commanded by natural human poses. Moreover, LEDs installed on the UAV are used to communicate the state and intents of the UAV to the human as feedback throughout the interaction. A real time multi-human pose estimation system is built that can perform with low latency while maintaining competitive performance. The UAV is equipped with a robotic arm, kinematic and dynamic attitude models for which are provided by introducing the center of gravity (COG) of the vehicle. In addition, a super-twisting extended state observer (STESO)-based back-stepping controller (BSC) is constructed to estimate and attenuate complex disturbances in the attitude control system of the UAV, such as wind gusts, model uncertainties, etc. A stability analysis for the entire control system is also presented based on the Lyapunov stability theory. The pose estimation system is integrated with the proposed intelligent control architecture to command the UAV to execute an exploration task stably. Additionally, all the components of this interaction system are described. Several simulations and experiments have been conducted to demonstrate the effectiveness of the whole system and its individual components.

Intelligent control method for traffic flow at urban intersection based on vehicle networking

In order to overcome the problems of large number of queued vehicles and high delay in current traffic control methods for urban intersection, this paper proposes an intelligent traffic control method based on vehicle network for urban intersection. This method uses the vehicle network to design the intelligent traffic control architecture of urban intersection, which is divided into data acquisition layer, data processing layer and intelligent traffic control layer. Data processing layer is divided into vehicle data processing unit (on-board unit, OBU) and data transmission unit (road side unit, RSU). Two-way data transmission between vehicle and wide area network is realised. Green duration optimisation mechanism is introduced to realise intelligent control of traffic flow at intersections. The experimental results show that the proposed method has fewer queued vehicles, better control performance and reliability as well as lower vehicle delay.

Adaptive nonlinear active suspension control based on a robust road classifier with a modified super-twisting algorithm

For the suspension system equipped with nonlinear hydraulic actuators and excited by external road conditions, a road adaptive intelligent suspension control strategy is developed. In this work, (1) a multi-phase intelligent road adaptive control architecture is developed to enhance the ride comfort in the presence of varying road excitations; (2) a modified algorithm is proposed to improve the system performance. Initially based on the nonlinear system dynamics, a sliding mode controller based on an improved super-twisting algorithm is proposed. In the Off-line phase, the optimized control parameters based on particle swarm optimization (PSO) approach for each road level are determined and supplied to a probabilistic neural network (PNN)-based classifier for training. In the On-line phase, the PNN classifier employs the measured unsprung mass acceleration to determine the road level and supplies the information to the controller database. Based on the classified road level, corresponding control parameters as determined by PSO are then selected. These control parameters are then supplied to the nonlinear controller which provides the active control. The closed-loop stability of the proposed approach is proved, and the simulation results for different road levels are presented to show the effectiveness of the proposed approach.

TIDE: Time-relevant deep reinforcement learning for routing optimization

Abstract Routing optimization has been researched in network design for a long time, and plenty of optimization schemes have been proposed from both academia and industry. However, such schemes are either too complicated in applications or far from the optimal performance. In recent years, with the development of Software-defined Networking (SDN) and Artificial Intelligence (AI), AI-based methods of routing strategy are being considered. In this paper, we propose TIDE, an intelligent network control architecture based on deep reinforcement learning that can dynamically optimize routing strategies in an SDN network without human experience. TIDE is implemented and validated on a real network environment. Experiment result shows that TIDE can adjust the routing strategy dynamically according to the network condition and can improve the overall network transmitting delay by about 9% compared with traditional algorithms.

Software-defined converged access network with cross-layer intelligent control architecture

Abstract The optical and wireless convergence technology is one of the principal means for the problem of last mile access. The software-defined network (SDN) supported by OpenFlow protocol is a promising centralized control architecture. Combining these two aspects, we propose a software-defined access network control architecture for the optical and wireless convergence technology, and design its functional modules and the overall interaction process. The proposed architecture adopts the idea of hierarchical control, introduces the OpenFlow agent, dynamically decentralizes some of the controller's authorities to the optical line terminal (OLT), and enhances its overall disaster response capability by extending the control depth of the controller to the optical network unit (ONU). The controller obtains relevant information in real time from a global perspective to achieve more comprehensive dynamic and flexible control, which effectively improves the network resource utilization and meets different service requirements of different users.

A MPTCP-Based Network Architecture for Intelligent Train Control and Traffic Management Operations

This paper investigates a novel communication system architecture for traffic management and data control in railway scenarios. The proposed solution should be integrated in the existing European Railway Traffic Management System/European Train Control System, the most advanced and successful standard even outside the European countries. Our system integrates public land mobile networks that usually offer best-effort packet services, with a quality-of-service (QoS) guaranteed satellite network. It represents a cost-effective substitution for 4G and 5G networks, aiming to replace current well-known GSM-R standard. The coexistence of heterogeneous networks is guaranteed by the multi-path transmission control protocol (TCP) protocol that addresses specific “add and drop” subflow policies and priority handling logics, in order to realize efficient seamless handovers. Providing priority to TCP subflows corresponds to select QoS-guaranteed, and the best effort networks able to fulfill railway requirements. The logic of adding and dropping subflows exploits both a real-time check of the current status of the serving network and “a priori” network performance information. Experimental results have been carried out in both city and harsh environments, where the main performance metrics ( i.e. , average delay, jitter, and bit rate) have been assessed.

A Highly Accurate Model-Free Motion Control System with a Mamdani Fuzzy Feedback Controller combined with a TSK Fuzzy Feed-forward Controller

In this paper, a new intelligent robot motion control architecture --- a highly accurate model-free fuzzy motion control- is proposed in order to achieve improved robot motion accuracy and dynamic performance. Its architecture combines a Mamdani fuzzy proportional (P) and a conventional integral (I) plus derivative (D) controller for the feedback part of the system, and a Takagi-Sugeno-Kang fuzzy controller for the feed-forward, nonlinear part. The fuzzy P + ID controller improves the performance of the nonlinear system, and the TSK fuzzy controller uses a TSK fuzzy inference system based on extended subtractive- clustering method which integrates information on joint angular displacement, velocity and acceleration for torque identification. The advantage of this kind of model-free control is that it uses the information directly from the input/output of the nonlinear system, without any complex robot model computation, in order to decrease the control system's sensitivity to any dynamical uncertainty. Furthermore, parametric search for clustering parameters in extended subtractive clustering secures the high accuracy of the system identification. Consequently, this proposed model-free fuzzy motion control benefits from the advantages of two kinds of fuzzy system. It not only incorporates flexible design, good performance and simple conception but also ensures precise motion control and great robustness. Comparisons with other intelligent models and results from numerical studies on a 4-bar planar parallel mechanism show the effectiveness and competitiveness of the proposed control.

Development of a prototype smart home intelligent lighting control architecture using sensors onboard a mobile computing system

As smartphones become increasingly powerful and ubiquitous, integrating them into intelligent lighting systems can boost both convenience and energy efficiency. This paper presents an intelligent lighting system prototype with enhanced security features for smart homes. The custom-built Android mobile application made use of the onboard ambient light sensor to run a novel closed-loop feedback algorithm to implement daylight harvesting. A cost analysis shows that the whole system setup is slightly cheaper than commercial products and due to its daylight harvesting capabilities, has potential for monetary savings in the long run, outperforming current commercial products.

Intelligent Supervisory Control of Discrete Event Systems - A Holon based Approach

Supervisory Control of Discrete Event Systems (DEVS) requires potential capabilities for constant adaptation and fast responses to the frequent un-expected disturbances occur in process and changes in production orders. Both hierarchical and heterarchical architecture control systems being developed so far have had limited success in achieving such potential capabilities. Hierarchical architecture systems due to their rigid structures inhibit inadequate responses to the process disturbances. Heterarchical architecture systems are quite good in dealing with the disturbances and are highly adaptable, but such structures cannot guarantee high performance due to unpredictable behavior. The shortcomings of the said systems develops the need for designing Holon based structure in which entities are able to communicate, cooperate, and have certain level of intelligence that enable them to be highly autonomous. In this paper we present a methodology that encapsulates individual needs of the physical layer, in general discrete event objects, in order to promote a goal orientated infrastructure with negotiation and cooperation capabilities at the computational layer of Discrete Event Systems. This methodology propagates chains of cooperation between holons seeking satisfactory goal realization. Architecture for a Holonic-based intelligent Supervisory Control of DEVS has been proposed along with a guide-line for its product ionisation. The newly proposed taxonomy of individual holons occurring in a discrete event system as a supervisory controller will allow to understand interaction mechanisms in a complex Holon’s society, and to obtain design methodology for a Holonic based control of DEVS.

Intelligent Control Architecture for Assistive Mobile Robots

The current technology has provided the necessary means for creating better assistive tools. Among these tools, the interest on assistive robotics has been growing, since its cost is being reduced and new equipment is introduced on the market. Yet, developing robots for aiding therapy is not an easy task because robotics is intrinsically multidisciplinary. Among the several research fields contributing to robotics research, two are of particular interest: control architectures and human–robot interaction. This paper covers ongoing research projects on assistive mobile robots for child rehabilitation developed in collaboration with Brazilian and Colombian Universities and proposes an intelligent control architecture based on human–robot interaction for assistive applications. Furthermore, it presents a case study on assistive robots for special education.

The improvement of centralized intelligent control architecture and data collection algorithm

The commonly used structure of centralized temperature control systems is the bi-layer structure with upper computer directly controlling thermostats. But this kind of structure has obvious drawbacks, such as high cost, inability to scale up, low degree of intelligence, etc.. This paper presents a design scheme of an intelligent three-tier structure, namely thermostats, intelligent logging devices, and upper computer. The upper computer can read data directly from both thermostats and logging devices. This scheme can solve the problems in bi-layer structure well, and it is also suitable for other types of large-scale control system. However, there still exists data loss problem in this scheme when systems reach a great scale. A method of using historical data in the database and linear feature of temperature curves is also proposed in this paper to improve data collection algorithm, which preferably solve this data loss problem. Appropriate products have been developed based on the techniques of this scheme, and have been applied in practical industrial production and experimental teaching.