An evolutionary recurrent network which automates the design of recurrent neural/fuzzy networks using a new evolutionary learning algorithm is proposed in this paper. This new evolutionary learning algorithm is based on a hybrid of genetic algorithm (GA) and particle swarm optimization (PSO), and is thus called HGAPSO. In HGAPSO, individuals in a new generation are created, not only by crossover and mutation operation as in GA, but also by PSO. The concept of elite strategy is adopted in HGAPSO, where the upper-half of the best-performing individuals in a population are regarded as elites. However, instead of being reproduced directly to the next generation, these elites are first enhanced. The group constituted by the elites is regarded as a swarm, and each elite corresponds to a particle within it. In this regard, the elites are enhanced by PSO, an operation which mimics the maturing phenomenon in nature. These enhanced elites constitute half of the population in the new generation, whereas the other half is generated by performing crossover and mutation operation on these enhanced elites. HGAPSO is applied to recurrent neural/fuzzy network design as follows. For recurrent neural network, a fully connected recurrent neural network is designed and applied to a temporal sequence production problem. For recurrent fuzzy network design, a Takagi-Sugeno-Kang-type recurrent fuzzy network is designed and applied to dynamic plant control. The performance of HGAPSO is compared to both GA and PSO in these recurrent networks design problems, demonstrating its superiority.

In this paper, a TSK-type recurrent fuzzy network (TRFN) structure is proposed. The proposal calls for the design of TRFN by either neural network or genetic algorithms depending on the learning environment. A recurrent fuzzy network is described which develops from a series of recurrent fuzzy if-then rules with TSK-type consequent parts. The recurrent property comes from feeding the internal variables, derived from fuzzy firing strengths, back to both the network input and output layers. In this configuration, each internal variable is responsible for memorizing the temporal history of its corresponding fuzzy rule. The internal variable is also combined with external input variables in each rule's consequence, which shows an increase in network learning ability. TRFN design under different learning environments is next advanced. For problems where supervised training data is directly available, TRFN with supervised learning (TRFN-S) is proposed, and a neural network (NN) learning approach is adopted for TRFN-S design. An online learning algorithm with concurrent structure and parameter learning is proposed. With flexibility of partition in the precondition part, and outcome of TSK-type, the TRFN-S displays both small network size and high learning accuracy. For problems where gradient information for NN learning is costly to obtain or unavailable, like reinforcement learning, TRFN with Genetic learning (TRFN-G) is put forward. The precondition parts of TRFN-G are also partitioned in a flexible way, and all free parameters are designed concurrently by genetic algorithm. Owing to the well-designed network structure of TRFN, TRFN-G, like TRFN-S, is characterized by high learning accuracy. To demonstrate the superior properties of TRFN, TRFN-S is applied to dynamic system identification and TRFN-G to dynamic system control. By comparing the results to other types of recurrent networks and design configurations, the efficiency of TRFN is verified.

This paper proposes a self-evolving interval type-2 fuzzy neural network (SEIT2FNN) with online structure and parameter learning. The antecedent parts in each fuzzy rule of the SEIT2FNN are interval type-2 fuzzy sets and the fuzzy rules are of the Takagi-Sugeno-Kang (TSK) type. The initial rule base in the SEIT2FNN is empty, and the online clustering method is proposed to generate fuzzy rules that flexibly partition the input space. To avoid generating highly overlapping fuzzy sets in each input variable, an efficient fuzzy set reduction method is also proposed. This method independently determines whether a corresponding fuzzy set should be generated in each input variable when a new fuzzy rule is generated. For parameter learning, the consequent part parameters are tuned by the rule-ordered Kalman filter algorithm for high-accuracy learning performance. Detailed learning equations on applying the rule-ordered Kalman filter algorithm to the SEIT2FNN consequent part learning, with rules being generated online, are derived. The antecedent part parameters are learned by gradient descent algorithms. The SEIT2FNN is applied to simulations on nonlinear plant modeling, adaptive noise cancellation, and chaotic signal prediction. Comparisons with other type-1 and type-2 fuzzy systems in these examples verify the performance of the SEIT2FNN.

A recurrent self-organizing neural fuzzy inference network (RSONFIN) is proposed in this paper. The RSONFIN is inherently a recurrent multilayered connectionist network for realizing the basic elements and functions of dynamic fuzzy inference, and may be considered to be constructed from a series of dynamic fuzzy rules. The temporal relations embedded in the network are built by adding some feedback connections representing the memory elements to a feedforward neural fuzzy network. Each weight as well as node in the RSONFIN has its own meaning and represents a special element in a fuzzy rule. There are no hidden nodes (i.e., no membership functions and fuzzy rules) initially in the RSONFIN. They are created on-line via concurrent structure identification (the construction of dynamic fuzzy if-then rules) and parameter identification (the tuning of the free parameters of membership functions). The structure learning together with the parameter learning forms a fast learning algorithm for building a small, yet powerful, dynamic neural fuzzy network. Two major characteristics of the RSONFIN can thus be seen: 1) the recurrent property of the RSONFIN makes it suitable for dealing with temporal problems and 2) no predetermination, like the number of hidden nodes, must be given, since the RSONFIN can find its optimal structure and parameters automatically and quickly. Moreover, to reduce the number of fuzzy rules generated, a flexible input partition method, the aligned clustering-based algorithm, is proposed. Various simulations on temporal problems are done and performance comparisons with some existing recurrent networks are also made. Efficiency of the RSONFIN is verified from these results.

This paper proposes a reinforcement ant optimized fuzzy controller (FC) design method, called RAOFC, and applies it to wheeled-mobile-robot wall-following control under reinforcement learning environments. The inputs to the designed FC are range-finding sonar sensors, and the controller output is a robot steering angle. The antecedent part in each fuzzy rule uses interval type-2 fuzzy sets in order to increase FC robustness. No a priori assignment of fuzzy rules is necessary in RAOFC. An online aligned interval type-2 fuzzy clustering (AIT2FC) method is proposed to generate rules automatically. The AIT2FC not only flexibly partitions the input space but also reduces the number of fuzzy sets in each input dimension, which improves controller interpretability. The consequent part of each fuzzy rule is designed using Q-value aided ant colony optimization (QACO). The QACO approach selects the consequent part from a set of candidate actions according to ant pheromone trails and Q-values, both of whose values are updated using reinforcement signals. Simulations and experiments on mobile-robot wall-following control show the effectiveness and efficiency of the proposed RAOFC.

This paper proposes an evolutionary-group-based particle-swarm-optimization (EGPSO) algorithm for fuzzy-controller (FC) design. The EGPSO uses a group-based framework to incorporate crossover and mutation operations into particle-swarm optimization. The EGPSO dynamically forms different groups to select parents in crossover operations, particle updates, and replacements. An adaptive velocity-mutated operation (AVMO) is incorporated to improve search ability. The EGPSO is applied to design all of the free parameters in a zero-order Takagi-Sugeno-Kang (TSK)-type FC. The objective of EGPSO is to improve fuzzy-control accuracy and design efficiency. Comparisons with different population-based optimizations of fuzzy-control problems demonstrate the superiority of EGPSO performance. In particular, the EGPSO-designed FC is applied to mobile-robot navigation in unknown environments. In this application, the robot learns to follow object boundaries through an EGPSO-designed FC. A simple learning environment is created to build this behavior without an exhaustive collection of input-output training pairs in advance. A behavior supervisor is proposed to combine the boundary-following behavior and the target-seeking behavior for navigation, and the problem of dead cycles is considered. Successful mobile-robot navigation in simulation and real environments verifies the EGPSO-designed FC-navigation approach.