Rita Lovassy

Parameter optimisation in fuzzy flip-flop-based neural networks

This paper presents a method for optimising the parameters of fuzzy flip-flop-based neural networks (FNN) consisting of fuzzy J-K and D flip-flop neurons based on various popular fuzzy operations using bacterial memetic algorithm with the modified operator execution order (BMAM). In early works, the authors proposed the Levenberg-Marquardt algorithm (LM) a widely used second order gradient type training algorithm for fuzzy neural networks variables optimisation. The BMAM local and global search evolutionary approach is a bacterial type memetic algorithm which executes several LM cycles during the bacterial mutation after each mutational step, using the LM method more efficiently. Numerical experiments were performed to show the function approximation capability of various quasi optimised FNN types based on fuzzy J-K and D flip-flop neurons using algebraic, Lukasiewicz, Yager, Dombi, Hamacher and Frank norms, trained with LM method and BMAM algorithm.

Learning the optimal parameter of the Hamacher t-norm applied for fuzzy-rule-based model extraction

Mamdani-type inference systems with trapezoidal-shaped fuzzy membership functions play a crucial role in a wide variety of engineering systems, including real-time control, transportation and logistics, network management, etc. The automatic identification or construction of such fuzzy systems input output data is one of the key problems in modeling. In the past years, the authors have investigated several different fuzzy t-norms, among others, algebraic and trigonometric ones, and the Hamacher product by substituting the standard “min” t-norm operation, in order to achieve better model fitting. In the present paper, the focus is on examining the general parametric Hamacher t-norm, where the free parameter quite essentially influences the quality of modeling and the learning capability of the model identification system. Based on a wide scope of simulation experiments, a quasi-optimal interval for the value of the Hamacher operator is proposed.

A possible control structure for production lines optimization

This paper proposes a new control structure for automatic production lines. To handle the production lines flexibility and to solve the optimal control problem a new methodology is suggested. In our approach the production line workstations are connected sequential, coordinated and controlled with individual (industrial) Personal Computers (PCs) interconnected in grid. These computers are able to supply the whole system control task and to optimize the production. The suggested method advantages and disadvantages are enumerated. Possible control system architecture is presented.

Fuzzy Flip-Flops Revisited

J-K flip-flops are elementary digital units providing sequential features/memory functions. Their definitive equation is used both in the minimal disjunctive and conjunctive forms. Fuzzy connectives do not satisfy all Boolean axioms, thus the fuzzy equivalents of these equations result in two non-equivalent definitions, “reset and set type” fuzzy flip-flops (F3) by Hirota & al. when introducing the concept of F3. There are many alternatives for “fuzzifying” digital flip-flops, using standard, algebraic or other connectives. The paper gives an overview of some of the most famous F3-s by presenting their definitions and presenting graphs of the inner state for a typical state value situation. Then a pair of non-associative operators is introduced, and the properties of the respective F3 are discussed. The investigation of possible fuzzy flip-flops is continued by examining Turksen’s IVFS, its midpoint values, and by introducing “minimized IVFS” (MIVFS), along with the MIVFS midpoints.

Fuzzy Flip-Flop based Neural Networks as a novel implementation possibility of multilayer perceptrons

Fuzzy Flip-Flop based Neural Networks (FNN) constructed from fuzzy D flip-flops are studied as a novel technique to implement multilayer perceptrons. The starting point of this approach is the concept of fuzzy flip-flop (F3), as the extension of the binary counterpart. Fuzzy D flip-flop based neurons are viewed, as sigmoid function generators. Their characteristic equations contain simple fuzzy operations, thus enabling easy implementability. FNNs have an interconnected fuzzy neuron structure composed from a large number of neurons acting in parallel which are capable of learning, and are suitable for function approximation. In this paper we propose the FPGA implementation of Łukasiewicz operations, furthermore of fuzzy D flip-flop neurons based on Łukasiewicz norms.

Intelligent reconfigurable universal fuzzy flip-flop

In this paper a universal fuzzy flip-flop is proposed that can be reconfigured as a fuzzy SR, D, JK, or T flip-flop. When integrated with a multi layer neural network, the resulting reconfigurable fuzzy-neural structure showed excellent learning ability. The sigmoid activation function of neurons in the hidden layers of the multilayer neural network was replaced by the quasi-sigmoidal transfer characteristics of the universal fuzzy flip-flop in the reconfigurable fuzzy-neural structure. Experimental results showed that the reconfigurable fuzzy-neural structure can be effectively trained using either a large or sparse set of data points to closely approximate nonlinear input functions.

Progressive bacterial algorithm

The purpose of this paper is to present a new version of the Bacterial Algorithms used for fuzzy rule base extraction called Progressive Bacterial Algorithm. In order to explore high quality models with very good speed of convergence towards the optimal rule base, we develop an improved version of the Bacterial Evolutionary and former Bacterial Memetic Algorithms. It is shown, in case of multidimensional reference problems, by comparing with existing methods, that an efficient and fast convergent tool is obtained.

Spatial load forecast for Electric Vehicles

In this paper the Electric Vehicles (EVs) spatial load forecast is studied, highlighting their current situation in Hungary. The impact of electric vehicles charging on electricity consumption, which generates a concentrated power demand at specific points of the network, is modeled. The extra power requires the development of network capacity, the extra energy requires more power generation. It is a new task to forecast and model the new demand in time and space for the network - and generation planning.

Fuzzy neural networks stability in terms of the number of hidden layers

This paper introduces an approach for studying the stability, and generalization capability of one and two hidden layer Fuzzy Flip-Flop based Neural Networks (FNNs) with various fuzzy operators. By employing fuzzy flip-flop neurons as sigmoid function generators, novel function approximators are established that also avoid overfitting in the case of test data containing noisy items in the form of outliers. It is shown, by comparing with existing standard tansig function based approaches that reducing the network complexity networks with comparable stability are obtained. Finally, examples are given to illustrate the effect of the hidden layer number of neural networks.

Robustness of Fuzzy Flip-Flop based Neural Networks

In this paper the robustness of three different types of Fuzzy Flip-Flop based Neural Network (FNN) and the standard tansig based neural networks is compared from the various test function approximation goodness points of view. It is tested how well the fuzzy flip-flop based and the simulated neural networks handle the test data sets outlier points. The robust design of the FNN is presented, and the best suitable fuzzy neuron type is emphasized. Furthermore, the sensitivity of fuzzy neural networks to the fuzzy neuron type and hidden layers neuron number is evaluated.