A local neuro-fuzzy network for high-dimensional models and optimization

Relaxed fuzzy Lyapunov approach for dynamic local model networks

“…• It is independent regarding the dimensions of the partition space and the type of the validity function. • It is applicable to arbitrary partitioning strategies, such as: equidistant grid [4]; recursive, axis orthogonal [8], [9], [14]; axis oblique [10], [15]; arbitrary functions [12]; hinging hyperplanes [11]. This paper is organized as follows: The architecture of local model networks is briefly described in section II.…”

“…two dimensional trapezoid validity functions) it is possible to determine the model transition using the phase portrait of the considered local model network or the method presented in [4]. For high dimensional partition spaces and complex partitioning strategies such as [8], [9], [10], [11] and [12] the method of [4] is no longer feasible. In this paper a systematic approach is introduced to determine all occurring model transitions in a dynamic LMN.…”

Nonlinear System Identification through Local Model Approaches: Partitioning Strategies and Parameter Estimation

“…For the integration of the above mentioned methodologies in a local model network weighted TLS and GTLS parameter estimation algorithms are presented in this work that allow for individual weighting of data records. Recent local model network approaches utilise the prediction error for partitioning, Abonyi et al (2002); Hametner & Jakubek (2007); Jakubek & Keuth (2006). If some or all signals involved in the parameter estimation process are corrupted by noise this approach is no longer feasible.…”

“…Johansen et al (2000). Other methods make use of the input/output data of the system to identify suitable subdomains, Jakubek & Keuth (2006); Nelles (2002). The identification and partitioning algorithm presented in this contribution is based on an iterative decomposition that works in the partition space rather than in the input-or product space.…”

Genetic-fuzzy model of diesel engine working cycle

“…Most publications about engine modeling with the use of artificial intelligence methods concern the control of engine work [6] and exhaust gas emission [7,8]. Spray penetration in the diesel engine, depending on fuel pressure and density, was also modeled, by means of the neuro-fuzzy system [9].…”