Research on the controller of an arc welding process based on a PID neural network

A controller based on a PID neural network （PIDNN） is proposed for an arc welding power source whose output characteristic in responding to a given value is quickly and intelligently controlled in the welding process. The new method syncretizes the PID control strategy and neural network to control the welding process intelligently, so it has the merit of PID control rules and the trait of better information disposal ability of the neural network. The results of simulation show that the controller has the properties of quick response, low overshoot, quick convergence and good stable accuracy, which meet the requirements for control of the welding process.     

Direct heuristic dynamic programming based on an improved PID neural network

In this paper,an improved PID-neural network(IPIDNN) structure is proposed and applied to the critic and action networks of direct heuristic dynamic programming(DHDP).As one of online learning algorithm of approximate dynamic programming(ADP),DHDP has demonstrated its applicability to large state and control problems.Theoretically, the DHDP algorithm requires access to full state feedback in order to obtain solutions to the Bellman optimality equation. Unfortunately,it is not always possible to access all the states in a real system.This paper proposes a solution by suggesting an IPIDNN configuration to construct the critic and action networks to achieve an output feedback control.Since this structure can estimate the integrals and derivatives of measurable outputs,more system states are utilized and thus better control performance are expected.Compared with traditional PIDNN,this configuration is flexible and easy to expand. Based on this structure,a gradient decent algorithm for this IPIDNN-based DHDP is presented.Convergence issues are addressed within a single learning time step and for the entire learning process.Some important insights are provided to guide the implementation of the algorithm.The proposed learning controller has been applied to a cart-pole system to validate the effectiveness of the structure and the algorithm.     

Sailboat navigation control system based on spiking neural networks

In this paper, we presented the development of a navigation control system for a sailboat based on spiking neural networks (SNN). Our inspiration for this choice of network lies in their potential to achieve fast and low-energy computing on specialized hardware. To train our system, we use the modulated spike time-dependent plasticity reinforcement learning rule and a simulation environment based on the BindsNET library and USVSim simulator. Our objective was to develop a spiking neural network-based control systems that can learn policies allowing sailboats to navigate between two points by following a straight line or performing tacking and gybing strategies, depending on the sailing scenario conditions. We presented the mathematical definition of the problem, the operation scheme of the simulation environment, the spiking neural network controllers, and the control strategy used. As a result, we obtained 425 SNN-based controllers that completed the proposed navigation task, indicating that the simulation environment and the implemented control strategy work effectively. Finally, we compare the behavior of our best controller with other algorithms and present some possible strategies to improve its performance.     

Adaptive backstepping output feedback control for SISO nonlinear system using fuzzy neural networks

In this paper, a new fuzzy-neural adaptive control approach is developed for a class of single-input and single-output （SISO） nonlinear systems with unmeasured states. Using fuzzy neural networks to approximate the unknown nonlinear functions, a fuzzy- neural adaptive observer is introduced for state estimation as well as system identification. Under the framework of the backstepping design, fuzzy-neural adaptive output feedback control is constructed recursively. It is proven that the proposed fuzzy adaptive control approach guarantees the global boundedness property for all the signals, driving the tracking error to a small neighbordhood of the origin. Simulation example is included to illustrate the effectiveness of the proposed approach.     

输出概率密度函数形状的多步预测控制

A predictive control strategy is proposed for the shaping of the output probability density function (PDF) of linear stochastic systems. The B-spline neural network is used to set up the output PDF model and therefore converts the PDF-shaping into the control of B-spline weights vector. The Diophantine equation is then introduced to formulate the predictive PDF model, based on which a moving-horizon control algorithm is developed so as to realize the predictive PDF tracking performance.     

Multi-step Predictive Control of a PDF-shaping Problem

A predictive control strategy is proposed for the shaping of the output probability density function (PDF) of linear stochastic systems. The B-spline neural network is used to set up the output PDF model and therefore converts the PDF-shaping into the control of B-spline weights vector. The Diophantine equation is then introduced to formulate the predictive PDF model, based on which a moving-horizon control algorithm is developed so as to realize the predictive PDF tracking performance.     

Heuristic dynamic programming-based learning control for discrete-time disturbed multi-agent systems

Owing to extensive applications in many fields, the synchronization problem has been widely investigated in multi-agent systems. The synchronization for multi-agent systems is a pivotal issue, which means that under the designed control policy, the output of systems or the state of each agent can be consistent with the leader. The purpose of this paper is to investigate a heuristic dynamic programming (HDP)-based learning tracking control for discrete-time multi-agent systems to achieve synchronization while considering disturbances in systems. Besides, due to the difficulty of solving the coupled Hamilton– Jacobi–Bellman equation analytically, an improved HDP learning control algorithm is proposed to realize the synchronization between the leader and all following agents, which is executed by an action-critic neural network. The action and critic neural network are utilized to learn the optimal control policy and cost function, respectively, by means of introducing an auxiliary action network. Finally, two numerical examples and a practical application of mobile robots are presented to demonstrate the control performance of the HDP-based learning control algorithm.     

Bipartite consensus for nonlinear time-delay multiagent systems via time-varying gain control method

The bipartite consensus problem is addressed for a class of nonlinear time-delay multiagent systems in this paper. Therein, the uncertain nonlinear dynamics of all agents satisfy a Lipschitz growth condition with unknown constants, and part of the state information cannot be measured. In this case, a time-varying gain compensator is constructed, which only utilizes the output information of the follower and its neighbors. Subsequently, a distributed output feedback control protocol is proposed on the basis of the compensator. According to Lyapunov stability theory, it is proved that the bipartite consensus can be guaranteed by means of the designed control protocol. Different from the existing literature, this paper studies the leader–follower consensus problem under a weaker connectivity condition, i.e., the signed directed graph is structurally balanced and contains a directed spanning tree. Two simulation examples are carried out to show the feasibility of the proposed control strategy     

PI control based on fuzzy set-point weighting tracking for hydraulic crane boom system

A PI control strategy based on fuzzy set-point weighting following was proposed for the active damping control of a hydraulic crane boom system （HCBS）. Two valve-controlled PI controllers, which include a proportional feedforward controller based on fuzzy set-point weighting following and a limited semi-integrator（LSI）, are designed respectively. LSI is used to limit output signal and to prevent wind up at the low frequency of the spectrum. By using a range camera and an electronic feedback control, the tip damping on the HCBS can be adjusted artificially. A collaborative control simulation technique of HOPSAN and MATLAB/SIMULINK is applied to the controller design. Simulation results show that the proposed PI control system has less overshoot as well as faster response. The tip damping on the HCBS during operation is improved.     

Potential-based obstacle avoidance in formation control

Based on the double integrator mathematic model, a new kind of potential function is presented in this paper by referring to the concepts of the electric field; then a new formation control method is proposed, in which the potential functions are used between agent-agent and between agent-obstacle, while state feedback control is applied for the agent and its goal. This strategy makes the whole potential field simpler and helps avoid some local minima. The stability of this combination of potential functions and state feedback control is proven. Some simulations are presented to show the rationality of this control method.     

The adaptive control using BP networks for a nonlinear servo-motor

The servo-motor possesses a strongly nonlinear property due to the effect of the stimulating input voltageload-torque and environmental operating conditions.So it is rather diffcult to derive a traditional mathematical model which is capable of expressing both its dynamics and steady-state characteristics.A neural network-based adaptive control strategy is proposed in this paper.In this method,two neural networks have been adopted for system identification(NNI)and control(NNC),respectively.Then,the commonly-used specialized learning has been modified,by taking the NNI output as the approximation output of the servo-motor during the weights training to get sensitivity information.Moreover,the rule for choosing the learning rate is given on the basis of the analysis of Lyapunov stability.Finally,an example of applying the proposed control strategy on a servo-motor is presented to show its effectiveness.     

非线性伺服电动机的神经网络逆控制

伺服电动机由于存在接触过程的非线性、温漂等非线性因素的影响,很难建立其精确的数学模型,使得基于数学模型的控制困难.针对伺服电动机存在的非线性问题,提出了一种新颖的基于BP神经网络直接逆控制方法.首先,利用BP神经网络建立系统的正向模型(NNI),然后,设计基于神经网络的直接逆控制器(NNC),实现了对伺服电动机的自适应控制.在Lyapunov稳定性分析的基础上,给出了BP算法学习算子的选择方案,保证神经网络权值训练的快速收敛,同时,对训练BP神经网络控制器的专用算法(specialized learning)进行改进,利用NNI的输出求取权值调整的灵敏度函数.数字仿真结果表明提出的控制算法是简单有效的.     

局部递归神经网络控制器及其应用

基于人工神经网络提出了一种局部递归神经网络控制器。在描述了带有输出反馈和激活反馈的网络控制器的结构组成并定义了作为设计目标的误差函数后，采用带有弹性的梯度下降法，获得适用于实时在线调整权值的修正公式，给出了所提的网络控制器的设计步骤及其控制策略。将所提出的网络控制器应用到典型的单级倒立摆的实验系统中，将实验所获得的结果与LQY方法的实验结果进行了对比。     

Design and implementation of a neuro-adaptive backstepping controller for buck converter fed PMDC-motor

A neuro-adaptive backstepping control (NABSC) method using single-layer Chebyshev polynomial based neural network is proposed for the angular velocity tracking in buck converter fed permanent magnet dc (PMDC)-motor. Owing to their universal approximation property, neural networks have been utilized for approximating the unknown nonlinear profile of instantaneous load torque. The inherent computational complexity of the neural network based adaptive scheme has been circumvented through the use of orthogonal Chebyshev polynomials as basis functions. A detailed stability and transient performance analysis has been conducted using Lyapunov stability criteria. The proposed control scheme is shown to yield a superior output performance with enhanced robustness for wide variations in load torque and set-point changes, compared to existing conventional approaches based on adaptive backstepping. The theoretical propositions are verified on an experimental prototype using dSPACE, Control Desk DS1103 setup with an embedded TM320F240 Digital Signal Processor proving its applicability to real-time electrical systems. The efficiency of the proposed strategy is quantified using performance measures and are evaluated against the conventional adaptive backstepping control (ABSC) methodology. Ultimately, this investigation confirms the effectiveness of the proposed control scheme in achieving an enhanced output transient performance while faithfully realizing its control objective in the event of abrupt and uncertain load variations.     

Learning from NN output feedback control of nonlinear systems in Brunovsky canonical form

In this paper, we investigate the learning issue in the adaptive neural network (NN) output feedback control of nonlinear systems in Brunovsky canonical form with unknown affine term. With only output measurements, a high-gain observer (HGO) is employed to estimate the derivatives of the system output which may be associated with the generation of peaking phenomenon. The adverse effect of peaking on learning and its elimination strategies are analyzed. When the gain of HGO is chosen too high, it may cause the failure of learning from the unknown closed-loop system dynamics. Hence, the gain of HGO is not chosen too high to relieve peaking and guarantee the accuracy of the estimated system states. Then, learning from the unknown closed-loop system dynamics can be achieved. When repeating the same or similar control tasks, a neural learning controller is presented which can effectively recall and reuse the learned knowledge to guarantee the output tracking performance. Finally, simulation results demonstrate the effectiveness of the proposed scheme.     

Highly constrained neural networks for industrial quality control

In this work we investigate techniques for embedding domain-specific spatial invariances into highly-constrained neural networks. This information is used to drastically reduce the number of weights which have to be determined during the learning phase, thus allowing us to apply artificial neural networks to problems characterized by a relatively small number of available examples. As an application of the proposed methodology, we study the problem of optical inspection of machined parts. More specifically, we have characterized the performance of a network created according to this strategy, which accepts images of parts under inspection at its input and issues a flag at its output which states whether the part is defective. The results obtained so far show that the proposed methodology provides a potentially relevant approach for the quality control of industrial parts, as it offers both accuracy and short software development time, when compared with a classifier implemented using a standard approach.     

动态自适应模块化神经网络结构设计

针对全连接前馈神经网络不能有效应对时变系统的问题, 提出一种动态自适应模块化神经网络结构. 该网络采用减法聚类算法在线辨识工况数据的空间分布, 利用RBF 神经元实现对数据样本空间的划分, 并结合模糊策略将不同子样本空间的数据动态分配给不同的子网络, 最后对各子网络的输出进行集成. 该模块化网络中子网络数量和子网络规模都能根据所学时变任务动态自适应调整. 通过对不同时变系统的预测表明了该网络能够有效跟踪时变系统.     

Adaptive neural output regulation control of a solar power plant

This work proposes an indirect adaptive nonlinear control scheme based on a recurrent neural network and the output regulation theory.The neural model is first trained off-line, being further improved by means of an on-line learning strategy using the Lyapunov and nonlinear observation theories.The regulation problem is solved by an iterative strategy, formulated as an eigenvalue assignment problem, ensuring the convergence of the regulation equations.The strategy was tested on a distributed collector field of a solar power plant (Plataforma Solar de Almería, Spain). Experimental results, collected on the solar power plant, show the effectiveness of the proposed approach.     

Direct adaptive neural control of nonlinear systems with extreme learning machine

A direct adaptive neural control scheme for a class of nonlinear systems is presented in the paper. The proposed control scheme incorporates a neural controller and a sliding mode controller. The neural controller is constructed based on the approximation capability of the single-hidden layer feedforward network (SLFN). The sliding mode controller is built to compensate for the modeling error of SLFN and system uncertainties. In the designed neural controller, its hidden node parameters are modified using the recently proposed neural algorithm named extreme learning machine (ELM), where they are assigned random values. However, different from the original ELM algorithm, the output weight is updated based on the Lyapunov synthesis approach to guarantee the stability of the overall control system. The proposed adaptive neural controller is finally applied to control the inverted pendulum system with two different reference trajectories. The simulation results demonstrate good tracking performance of the proposed control scheme.     

ILC-Based Fixed-Structure Controller Design for Output PDF Shaping in Stochastic Systems Using LMI Techniques

In this paper, a generalized state-space controller design for the shaping of the output probability density function (PDF) is presented for non-Gaussian dynamical stochastic systems. A radial basis function (RBF) neural network is used to approximate the output PDF of the system. Such a neural network consists of a number of weights and corresponding basis functions. Using such an approximation, the dynamics of the original stochastic system can be expressed as the dynamics between the control input and the weights of the RBF neural network. The task of output PDF control can therefore be reduced to a RBF weight control together with an adaptive tuning of the basis function parameters (i.e., the centers and widths of the basis functions). To achieve this aim, the control horizon is divided into certain intervals hereinafter called batches. Using these definitions, the whole control strategy consists of three stages, namely (a) sub-space parameter identification of the dynamic nonlinear model (that relates the control signal to the weights of the RBF neural network); (b) Weight tracking controller design using an LMI-based convex optimization technique; and (c) RBF basis functions shape tuning in terms of their centers and widths using an iterative learning control (ILC) framework. Among the above stages, the first two are performed within each batch, while stage (c) is carried out between any two adjacent batches. Such an algorithm has the advantage of the batch-by-batch improvement of the closed-loop output PDF tracking performance. Moreover, the controller mentioned in stage (b) is a general controller in a state-space form. Stability analysis has been performed and simulation results are included to show the effectiveness of the proposed method, where encouraging results have been made.