Search published articles
Showing 4 results for Neuro-Fuzzy Network
Mahdi Shahab, Majid Moavenian,
Volume 17, Issue 4 (6-2017)
Abstract
Design of fault detection and diagnosis systems (FDDS), although extending the control strategies, they are challenged by controller interferences in fault diagnosis. In this study, in order to improve performance and accuracy of FDDS in the fault detection process, considering influential parameters and the level of corresponding interferences is investigated. To achieve this enterprise, a powerful method in fault pattern recognition of industrial plants based on dynamic behavior and dynamic model by using soft computing is designed and tested on simulated suspension system of a vehicle. The suspension system is one the parts, most affecting reliability and safety of the vehicle. For investigating the level of interference caused by the control unite, the simulations of both passive and active (equipped with hydraulic actuator) suspension systems are utilized in association with the control unite. The results of tests under variable circumstances (using random values) demonstrate that the presence of control unite, strict the FDDS process and reduces the robustness of the system against disturbances and noise. Considering the way in which the control unite affects the process, application of suggested solutions in this research, have a considerable impact on amendment of the adverse effects.
Fault detection program which is provided by Matlab software benefits special possibilities to investigate and define the effect of controlling unite and can be considered as a useful device to facilitate and precipitate conduction of tests in different stages of the research.
H. Pourhashem , A. Jamali, N. Narimanzade ,
Volume 19, Issue 2 (2-2019)
Abstract
Because of the widespread application in complex modeling based on experimental data, neuro-fuzzy networks have attracted the attention of researchers. In the neuro-fuzzy inference system, the objective is to reduce the system's prediction error relative to the actual data. The regulation of parameters of neuro-fuzzy network is very important and affects its performance. So, a new optimization algorithm based on Particle Swarm Optimization (PSO) and Differential Evolution (DE) has been proposed. In this algorithm, the coefficients of the operator speed are calculated dynamically, using fuzzy logic. These coefficients are set according to the generation number and variance of the particles. Proposed operator leads the particles to explore and exploit the search domain more precisely. Next, the performance of the proposed algorithm is checked by optimizing three benchmarks and comparing it with the results, which are obtained by conventional PSO and DE. The results show that the proposed algorithm obtained better solution in comparison with DE and PSO and proved its performance and efficiency. Finally, a neuro-fuzzy system has been employed to forecast the time series of Mackey-Glass. The parameters of this neuro-fuzzy network are optimized by the new algorithm and the PSO and DE method multi-objectively and the Pareto charts obtained by each method of optimization are compared with each other, indicating the better performance of the new algorithm.
M. Mohammadi Soleymani , S. Mirzadeh,
Volume 20, Issue 9 (9-2020)
Abstract
Due to the importance of tumbling mills in processing industries and factories and the lack of an acceptable model for identifying and predicting their performance, it is necessary to optimize these complexes, non-linear, and large systems. This paper aimed to study multi-objective optimization of operating parameters in a tumbling mill. To evaluate the effects of the mill working parameters such as mill speed, ball filling, slurry concentration, and slurry filling on grinding process, power draw, wear of lifters and size distribution of the mill product, it was tried to manufacture a pilot model with a smaller size than the actual mill. For this aim, a mill with 1×0.5m was implemented. The feed of the mill is copper ore with a size smaller than 1 inch. The experiments were done at 65 to 85% of the critical speed. In addition, the combination of the balls was used as grinding media with 10 to 30% of the total volume of the mill. Slurry concentration is 40 to 80% (the weight fraction of solid in slurry) and the slurry filling is between 0.5 and 2.5. In this paper, Adaptive Neuro-Fuzzy Inference System (ANFIS) based multi-objective optimization (NSGA-II) of tumbling mill is done. Level diagrams are used to select the best solution from the Pareto front. The results showed that the best grinding occurs at 70-80% of the critical speed and ball filling of 15-20%. Optimized grinding was observed when the slurry volume is 1-1.5 times of the ball bed voidage volume and the slurry concentration is between 60 and 70%.
Yousef Shahsbi, Erfan Mirshekari,
Volume 25, Issue 1 (12-2024)
Abstract
This research examines the optimization of expansion loops in steam pipeline systems using a neuro-fuzzy network. Stress analysis was conducted based on the ASME B31.3 design code using CAESAR II software. Additionally, a neuro-fuzzy network was developed and optimized in MATLAB. The results indicate that the neuro-fuzzy network outperforms traditional methods and the MLP neural network. Combining this network with the Bee Colony Optimization algorithm led to the identification of an optimal loop that minimizes pipeline length and reduces static and thermal stresses. The optimized loop obtained from the Perceptron network increased the loop length by 20 cm (1.14%) and reduced the total sum of standard stresses by 14.6%. In contrast, the optimized loop from the neuro-fuzzy network reduced the loop length by 120 cm (6.78%) and decreased the total sum of standard stresses by 9.5%. These findings demonstrate that the application of artificial intelligence techniques in expansion loop design significantly reduces thermal stresses and enhances design efficiency
