Learning and evolution are two fundamental forms of adaptation. There has been a great interest in combining learning and evolution with artificial neural networks (ANNs) in recent years. This paper: 1) reviews different combinations between ANNs and evolutionary algorithms (EAs), including using EAs to evolve ANN connection weights, architectures, learning rules, and input features; 2) discusses different search operators which have been used in various EAs; and 3) points out possible future research directions. It is shown, through a considerably large literature review, that combinations between ANNs and EAs can lead to significantly better intelligent systems than relying on ANNs or EAs alone.

/pdf/evolving-artificial-neural-networks-4kzrbu7pkn.pdf

Evolving artificial neural networks

Inverted pendulum system is a typical rapid, multivariable, nonlinear, absolute instability and non-minimum phase system, and it is a favorite problem in the field of control theory and application. In its control, the current main control method includes in fuzzy control, variable structure control and robust control etc. For fuzzy control of a double inverted pendulum, the research is focused on how to solve the “rule explosion” problem. The model and characteristics of the system are detailed analyzed; a status fusion function is designed using information fusion. By using it, the output variables of the system with six dimensions is synthesized as two variables: error and variation of error. From the fuzzy control theory, we also design the fuzzy controller of the double inverted pendulum system in MATLAB, and carried out the system simulation in Simulink, results show that the method is feasible.

Fuzzy control of a double inverted pendulum based on information fusion

Resolved acceleration control method or computed torque method is used for nonlinear control of industrial manipulators, which is composed of a model base portion and a servo portion. The servo portion is a close loop with respect to the position and velocity. On the other hand, the model base portion has the inertia term, gravity term and centrifugal/Coriolis term, which work for canceling the nonlinearity of manipulator. In order to realize high control stability, the position and velocity gains used in the servo portion should be selected suitably. In this paper, a simple but effective fine tuning method after manual tuning is introduced for the position and velocity feedback gains in the servo portion. At the first step, base values of the gains are roughly selected by a controller designer, e.g., considering the critically damped condition. After that, the base values are finely tuned by genetic algorithms. Genetic algorithms search for the better combination of the position and velocity gains. Simulations are conducted using a dynamic model of PUMA560 manipulator to validate the effectiveness of the proposed method.

Simulation of Fine Gain Tuning Using Genetic Algorithms for Model-Based Robotic Servo Controllers

An Indexed Bibliography of Genetic Algorithms in Control

The paper proposes a fast online learning method for neural network structures by using genetic algorithm (GA) and dynamic back propagation algorithm (BP) jointly. GA is used in the coarse tuning process which adjusts interconnection weights of the neural network. The dynamic back propagation algorithm is subsequently applied to achieve fine adjusting of the network weights. The fitness function, based on the squared error between the teaching signal and the network output value, is redefined at every time step and the proposed GA based algorithm solves a nonstationary function optimization task. At every time step the solution with the best fitness function is used for current representation of the neural network weights and biases. It is shown through the simulations and real time temperature control of drying oven that this learning algorithm has faster convergence ability and better performance on reducing mapping error in the online learning neural network structures. This leads to an improvement of the transient response of neuro adaptive systems. The proposed method has the potential to be applied to many practical areas such as system modeling and control, signal processing and pattern recognition.

Fast genetic on-line learning algorithm for neural network and its application to temperature control

The authors consider stabilization of an inverted pendulum which can be controlled by moving a cart in an intelligent way. Here, the authors adopt a PID(proportional plus integral plus derivative) control method to stabilize the pendulum since the PID controller has been extensively used in the industrial world. This controller requires the determination of PID control gains, but it is difficult to select the best gains theoretically. Thus, there have been many approaches to determine them empirically. Most of them are based on experience of operators and knowledge. Here, the authors propose a method using neural networks to tune the PID gains such that human operators tune the gains adaptively according to the environmental condition and systems specification. The tuning method is based on the error backpropagation method (BP method) and hence, it may be trapped in a local minimum. In order to avoid the local minimum problem, the authors use the genetic algorithm to find the initial values of the connection weights of the neural network and initial values of PID gains. The experimental results show the effectiveness of the present approach.

K. Kitagawa

Papers

Stabilization of inverted pendulum by the genetic algorithm