Neural Network Control System Research Articles

Informative modeling of subjective reality for intellectual anthropomorphic robots

This research is aimed at solving the problem of safeguarding robot operation by modeling the subjective reality of an intellectual anthropomorphic robot. Analysis has been carried out on trends in designing advanced robot control systems and methods of improving the safety of robot usage. The following conclusions are drawn: the risk of causing damage to human life, health or property increases when people interact with robots; it is necessary to combine neural network control systems with expert control systems in order to enhance the intellect of a robot and raise the level of trust held by humans towards robots; developing norms and regulations, designing collaborative robots, drawing visualization diagrams for safety zones cannot really provide the depth of robot socialization sufficient for the human society; an informative model of the subjective reality could be integrated into an intellectual anthropomorphic robot to provide a safer ‘human-robot’ interaction. Such an approach should result in achieving the greatest trust of humans and their safety due to anthropomorphic conversion of both the external design of a robot and its internal control structure. By its nature, the research is interdisciplinary – it is run in the fields of artificial intelligence and philosophy of subjective reality.

Research on H∞ Controller of Cyber Attacks and Eventriggering Mechanism

Aiming at the limited network bandwidth and various cyber attacks, the H∞ neural network controller triggered by discrete events is studied. In order to further improve the utilization of network bandwidth, the event triggering mechanism is improved by taking advantage of the advantages of event triggering in network transmission. At the same time, due to various uncertain and deliberate network attacks, data tampering and packet loss problems are caused. A new neural network control system model is established based on the attack mode obeying Bernoulli probability and dynamic integration. By using Lyapunov function, sufficient conditions are obtained to ensure the asymptotic stability of the closed-loop system, which meets the requirements of H∞. Finally, a numerical example is given to verify the validity of the results.

Reachable Set Estimation for Neural Network Control Systems: A Simulation-Guided Approach

The vulnerability of artificial intelligence (AI) and machine learning (ML) against adversarial disturbances and attacks significantly restricts their applicability in safety-critical systems including cyber-physical systems (CPS) equipped with neural network components at various stages of sensing and control. This article addresses the reachable set estimation and safety verification problems for dynamical systems embedded with neural network components serving as feedback controllers. The closed-loop system can be abstracted in the form of a continuous-time sampled-data system under the control of a neural network controller. First, a novel reachable set computation method in adaptation to simulations generated out of neural networks is developed. The reachability analysis of a class of feedforward neural networks called multilayer perceptrons (MLPs) with general activation functions is performed in the framework of interval arithmetic. Then, in combination with reachability methods developed for various dynamical system classes modeled by ordinary differential equations, a recursive algorithm is developed for over-approximating the reachable set of the closed-loop system. The safety verification for neural network control systems can be performed by examining the emptiness of the intersection between the over-approximation of reachable sets and unsafe sets. The effectiveness of the proposed approach has been validated with evaluations on a robotic arm model and an adaptive cruise control system.

Stabilisation of event‐triggered‐based neural network control system and its application to wind power generation systems

This study addresses the event-triggered (ET)-based stabilisation problem of neural-network-based control system (NNBCS) and illustrates the direct application to wind power generation system. In this regard, the novel ET-based controller algorithm is designed for NNBCS instead of sampled data controller (sampling will be initiated at a fixed rate regardless whether it is required or not) which reduces the computation complexity by avoiding the unnecessary details over the transmission. The novel stability and stabilisation conditions are expressed in terms of linear matrix inequalities that are derived through constructing the time-dependent Lyapunov functional candidate. For deepening the knowledge in the outcomes of the proposed conditions, the study numerically evaluates the dynamic models such as variable-speed wind turbine drive system, permanent magnet synchronous motors model and traditional inverted pendulum model and validates the effectiveness of the proposed controller scheme. Finally, a comparison result shows the superiority of the derived conditions.

Neural network control system of cooperative robot based on genetic algorithms

Attitude detection of cooperative robots can help robots recognize environment, understand tasks, and improve the safety, accuracy and efficiency of robot work. For robots with unknown configuration parameters, their configuration can be estimated; it is convenient for further kinematics and dynamics analysis. For an economical cooperative robot, its kinematic parameters can be corrected by attitude detection. An independent robot attitude detection system can be used as a secondary auxiliary system, and the fault diagnosis system of the robot is composed of the high-precision encoder system of the robot body. This paper studies the neural network control system of cooperative robot based on genetic algorithm. In this paper, the robot is taken as the research object. Aiming at its strong conjunction, non-linearity and multivariable characteristics, the problem of robot motion control based on neural network is mainly discussed. On this basis, the basic genetic algorithm and an improved genetic algorithm called messy are used to optimize the structure of the neural network, so as to better realize the motion control problem of the robot. The results of this study show that: After optimization with messy genetic algorithm, not only the tracking effect is better, but also the number of hidden layer nodes of the network is reduced from 12 to 7. This greatly simplifies the structure of the network and makes the design and training of the network relatively simple.

Research on parallel nonlinear control system of PD and RBF neural network based on U model

ABSTRACTThe modelling problem of nonlinear control system is studied, and a higher generality nonlinear U model is established. Based on the nonlinear U model, RBF neural network and PD parallel control algorithm are proposed. The difference between the control input value and the output value of the neural network is taken as the learning target by using the online learning ability of the neural network. The gradient descent method is used to adjust the PD output value, and ultimately track the ideal output. The Newton iterative algorithm is used to complete the transformation of the nonlinear model, and the nonlinear characteristic of the plant is reduced without loss of modelling precision, consequently, the control performance of the system is improved. The simulation results show that RBF neural network and PD parallel control system can control the nonlinear system. Moreover, the control system with Newton iteration can improve the control effect and anti-interference performance of the system.

Formation of information signs of unmanned aerial vehicles use efficiency

The subject matter of the article is an unmanned aerial vehicle and methods for determining the effectiveness of its use. The goal of the work is formation of information signs of unmanned aerial vehicles use effectiveness. The following tasks were solved in the article: formation of information signs of the effectiveness of unmanned aerial vehicles using the mathematical apparatus of probability theory and mathematical statistics, the analysis of the survivability of unmanned aerial vehicles by developing a two-level unmanned aerial vehicle control system using a neural network control system. The following methods used are: methods of probability theory and mathematical statistics, neural network management methods. The following results were obtained: The generated informational signs of unmanned aerial vehicles use effectiveness with application of the mathematical apparatus of probability theory and mathematical statistics, which made it possible to obtain an analytical expression for determining the time of information transmission by an unmanned aerial vehicle, being the main criterion for the effectiveness of its use in a given situation. An analysis of the survivability of unmanned aerial vehicles by developing a two-level control system for an unmanned aerial vehicle using a neural network control system that allows you to increase the survivability and thereby the likelihood of completing a task by an unmanned aerial vehicle. It was determined that an essential feature of the UAV functioning processes is their randomness, which is caused by the incomplete certainty of the conditions in which these processes occur, as well as various random deviations and errors that occur during the collection of information, the generation of control signals and their execution. Thus, the result of UAV functioning is randomly and quantitatively characterized by the laws of distribution of parameters expressing this result.

A Novel Reference Model-Based Neural Network Approach to Temperature Control System

With the continuous development of deep learning, neural network with its excellent self-learning performance obtains a series of a major breakthrough in target detection, image recognition and so on. In this paper, the temperature control system based on a neural network combined with I-PD compensation is proposed. To improve the neural network self-learning efficiency, the reference model is introduced for providing the teaching signal of neural network. The system simulations are carried out in MATLAB/SIMULINK environment to verify the control efficiency of the proposed reference model based neural network control system. The experiments are carried out on the DSP based temperature control system platform, the results are compared to the conventional I-PD control system to verify the control efficiency.

Design and Speed-Adaptive Control of a Powered Geared Five-Bar Prosthetic Knee Using BP Neural Network Gait Recognition.

To improve the multi-speed adaptability of the powered prosthetic knee, this paper presented a speed-adaptive neural network control based on a powered geared five-bar (GFB) prosthetic knee. The GFB prosthetic knee is actuated via a cylindrical cam-based nonlinear series elastic actuator that can provide the desired actuation for level-ground walking, and its attitude measurement is realized by two inertial sensors and one load cell on the prosthetic knee. To improve the performance of the control system, the motor control and the attitude measurement of the GFB prosthetic knee are run in parallel. The BP neural network uses input data from only the GFB prosthetic knee, and is trained by natural and artificially modified various gait patterns of different able-bodied subjects. To realize the speed-adaptive control, the prosthetic knee speed and gait cycle percentage are identified by the Gaussian mixture model-based gait classifier. Specific knee motion control instructions are generated by matching the neural network predicted gait percentage with the ideal walking gait. Habitual and variable speed level-ground walking experiments are conducted via an able-bodied subject, and the experimental results show that the neural network control system can handle both self-selected walking and variable speed walking with high adaptability.

Safety Verification of Cyber-Physical Systems with Reinforcement Learning Control

This paper proposes a new forward reachability analysis approach to verify safety of cyber-physical systems (CPS) with reinforcement learning controllers. The foundation of our approach lies on two efficient, exact and over-approximate reachability algorithms for neural network control systems using star sets, which is an efficient representation of polyhedra. Using these algorithms, we determine the initial conditions for which a safety-critical system with a neural network controller is safe by incrementally searching a critical initial condition where the safety of the system cannot be established. Our approach produces tight over-approximation error and it is computationally efficient, which allows the application to practical CPS with learning enable components (LECs). We implement our approach in NNV, a recent verification tool for neural networks and neural network control systems, and evaluate its advantages and applicability by verifying safety of a practical Advanced Emergency Braking System (AEBS) with a reinforcement learning (RL) controller trained using the deep deterministic policy gradient (DDPG) method. The experimental results show that our new reachability algorithms are much less conservative than existing polyhedra-based approaches. We successfully determine the entire region of the initial conditions of the AEBS with the RL controller such that the safety of the system is guaranteed, while a polyhedra-based approach cannot prove the safety properties of the system.

Adaptive neural network control for a nonlinear Euler‐Bernoulli beam in three‐dimensional space with unknown control direction

SummaryIn this paper, an adaptive neural network control system is developed for a nonlinear three‐dimensional Euler‐Bernoulli beam with unknown control direction. The Euler‐Bernoulli beam is modeled as a combination of partial differential equations (PDEs) and ordinary differential equations (ODEs). Adaptive radial basis function–based neural network control laws are designed to determine approximation of disturbances. A projection mapping operator is adopted to realize bounded approximation of disturbances. A Nussbaum function is introduced to compensate for the unknown control direction. The goal of this study is to suppress the vibrations of the Euler‐Bernoulli beam in three‐dimensional space. In addition, unknown control direction problem and bounded disturbances are considered to guarantee that the signals of the system are uniformly bounded. Numerical simulations demonstrate the effectiveness of the proposed method.

DSP Implementation of a Neural Network Vector Controller for IPM Motor Drives

This paper develops a neural network (NN) vector controller for an interior mounted permanent magnet (IPM) motor by using a Texas Instrument TMS320F28335 digital signal processor (DSP). The NN controller is developed based on the complete state-space equation of an IPM motor and it is trained to achieve optimal control according to approximate dynamic programming (ADP). A DSP-based NN control system is built for an IPM motor drives system, and a high efficient DSP program is developed to implement the NN control algorithm while considering the limited memory and computing capability of the TMS320F28335 DSP. The DSP-based NN controller is able to manage IPM motor control in linear, over, and six-step modulation regions to improve the efficiency of IPM drives and to allow for the full utilization of DC bus voltage with space-vector pulse-width modulation (SVPWM). The experiment results show that the proposed NN controller is able to operate with a sampling period of 0.1ms, even with limited DSP resources of up to 150 MHz cycle time, which is applicable in practical motor industrial implementations. The NN controller has demonstrated a better current and speed tracking performance than the conventional standard vector controller for IPM operation in both the linear and over-modulation regions.

A Dynamic Neural Network Approach for Efficient Control of Manipulators

Redundancy resolution is a critical problem in the control of manipulators. The dual neural network, as a special type of recurrent neural networks that are inherently parallel processing models, is widely investigated in past decades to control manipulators. However, to the best of our knowledge, existing dual neural networks require a full knowledge about manipulator parameters for efficient control. We make progress along this direction in this paper by proposing a novel model-free dual neural network, which is able to address the learning and control of manipulators simultaneously in a unified framework. Different from pure learning problems, the interplay of the control part and the learning part allows us to inject an additive noise into the control channel to increase the richness of signals for the purpose of efficient learning. Due to a deliberate design, the learning error is guaranteed for convergence to zero despite the existence of additive noise for stimulation. Theoretical analysis reveals the global stability of the proposed neural network control system. Simulation results verify the effectiveness of the proposed control scheme for redundancy resolution of a PUMA 560 manipulator.

The neural network-based control system of direct current motor driver

<span lang="EN-US">This article aims to propose an adaptive control system for the direct current motor driver based on the neural network. The control system consists of two neural networks: the first neural network is used to estimate the speed of the direct current motor and the second neural network is used as a controller. The plant in this research includes motor and the driver circuit so it is a complex model. It is difficult to determine the exact parameters of the plant so it is difficult to build the controller. To solve the above difficulties, the author proposes an adaptive control system based on the neural network to control the plant reach the high quality in the case of unknowing the parameters of the plant. The results are that the control quality of the system is very good, the response speed always follows the desired speed and the transition time is small. The simulation results of the neural network control system are shown and compared with that of a PID controller to demonstrate the advantages of the proposed method.</span>

高超声速飞行器的反步滑模神经网络控制系统

高超声速飞行器的反步滑模神经网络控制系统

Improving the energy efficiency of oil production using identification and prediction of operating modes of production wells based on data analysis methods, machine learning and neural networks

Currently, there is a need to improve the systems and control of pumping equipment in the oil and gas production and oil and gas transport industries. Therefore, an adaptive neural network control system for an electric drive of a production well was developed. The task of expanding the functional capabilities of asynchronous electric motors control of the oil and gas production system using the methods of neural networks is solved. We have developed software modules of the well drive control system based on the neural network, an identification system, and a scheme to adapt the control processes to changing load parameters, that is, to dynamic load, to implement the entire system for real-time control of the highspeed process. In this paper, based on a model of an identification block that includes a multilayered neural network of direct propagation, the control of the well system was implemented. The neural network of the proposed system was trained on the basis of the error back-propagation algorithm, and the identification unit works as a forecaster of system operation modes based on the error prediction. In the initial stage of the model adaptation, some fluctuations of the torque are observed at the output of the neural network, which is associated with new operating conditions and underestimated level of learning. However, the identification object and control system is able to maintain an error at minimum values and adapt the control system to a new conditions, which confirms the reliability of the proposed scheme.

Ship roll damping via direct inverse neural network control system

Undesired roll motions of surface vessels under sea way have negative influence on passengers, crew members, equipment and cargo. Damping roll motions is a complex problem due to unpredictable and irregular nature of marine environmental disturbances that induce roll motions. There are several stabilizing systems regarding roll motions such as gyrostabilizers, anti-roll tanks, active fins and rudder roll stabilizers. Among all types of roll stabilizers or roll damping systems active fins practically represents the most effective roll damping technique and yet the most commonly used roll stabilizer. Assuming that the longitudinal location of fins corresponds to the vessel center of gravity a one degree of freedom nonlinear model can be used to identify ship dynamics. In this paper the performance and robustness of a proposed control system based on neural network approach applied to active fins that takes into account roll response spectral characteristics transitions due to not only varying sea state but also changing sailing conditions is assessed. The suggested control system achieves robust performance with adequate roll damping effect that ranges from 50 to 65% roll reduction according to severity of weather conditions and sailing status.

Development of a procedure for the synthesis of a micro gas turbine engine neural controller with account for fuel consumption constraints

Neural networks are often used to model dynamic processes in objects and to synthesize their control systems. However, their application in real systems is now rather limited due to insufficient research into the process of creating control systems. The issues of creating neural network control systems for gas turbine engines, taking into account both their nonlinear dynamics and the fuel consumption constraints depending on the engine operating mode, remain practically unexplored. To take into account the fuel consumption constraints, a method was developed for modifying the misalignment between the actual and target RPM values during the training of the neural controller. The resulting neural controller is characterized by implicit fuel consumption constraints and non-linear dynamics of the engine itself. The developed method for modifying the neural network training error allows one to synthesize a nonlinear control system, taking into account the requirements for limitations in the automatic mode.

Adaptive recurrent neural network motion control for observation class remotely operated vehicle manipulator system with modeling uncertainty

Precise motion control of remotely operated vehicles plays an important role in a great number of submarine missions. However, the high-performance operations are difficult to realize due to the uncertainty in system modeling with self-disturbance. On the basis of the multibody system dynamics, self-disturbances from the tether and manipulator have been systematically analyzed in order to transform them into observed forces. A novel S surface–based adaptive recurrent wavelet neural network control system has been proposed on the nonlinear control of underwater vehicles, with its recurrent wavelet neural network structure designed for the approximation of the uncertain dynamics. Moreover, a robust function has been proposed to improve system robustness and convergence. The comparison shows that the remotely operated vehicle operation performance including the three-dimensional path following and vehicle-manipulator coordinate control has been greatly improved.

Self-Organizing Recurrent Wavelet Fuzzy Neural Network-Based Control System Design for MIMO Uncertain Nonlinear Systems Using TOPSIS Method

The major objective of this study is to design an effective control algorithm for dealing with multiple-input–multiple-output uncertain nonlinear systems. Novelty advantages of the proposed method include: (1) The network has the maximum initial rules; it helps to increase the responsiveness of the system; (2) the network has two dynamic thresholds: One dynamic threshold is utilized to consider whether to retain or to delete the existing rules and the other is used for generating a new rule; (3) the fuzzy neural network-based system can automatically construct the network structure and adjust the parameters of the system; (4) the network uses multiple combination techniques, such as sliding mode control, adaptive control, recurrent unit, wavelet function, fuzzy logic, neural network, and technique for order of preference by similarity to ideal solution multi-criteria decision analysis method. Based on the advantages of the above techniques, a self-organizing recurrent wavelet fuzzy neural network control system is designed comprising a main controller and a robust compensator. The gradient descent method is used to online tune the parameters for the main controller, and a Lyapunov stability theorem is applied to guarantee the system’s stability. Finally, the proposed control system is applied to a nonlinear chaotic system, an inverted double-pendulum system, and an unmanned aerial vehicle motion control to verify the effectiveness of the proposed control scheme. The simulation results show that the proposed control scheme can achieve favorable control performance.