Papers by Keyword: Adaptive Neural Network

Abstract: Most of metallurgical performance testing devices use small high-temperature furnace to simulate physical environment for the sample testing. Since the controlled object has the dynamic characteristics of nonlinear, time-varying, large delay and large inertia during heating process, it is difficult to establish an accurate models to control thermal processes and optimize. This paper presents an adaptive neural fuzzy modeling approach based on T-S model for the heating process. Using the fuzzy system structure identification and parameter identification, the more accurate nonlinear model can be obtained. Duo to the fuzzy neural network has the capability of autonomous, quickly and effectively converging to the required relations of the input and output, the modeling accuracy has been improved. The simulation results demonstrate the effectiveness of the proposed algorithm, and the method can provide a reference for obtaining accurate nonlinear model.

Abstract: It is unchangeable fact that there is great stiffness spring flexible load in electro-hydraulic force servo control system, and there is low oscillation frequency second order differential link in the numerator of the transfer function. On the other hand, the frequency response ability of the system is influenced badly from the link and the system may oscillate easily even is instability. Aiming at this special performance of the electro-hydraulic force servo control system and its small open loop gain characteristic, the adaptive neural network control strategy is adopted to the controller of the electro-hydraulic force servo control system this special performance system in order to improve the dynamic performance of the system. The model of the system with the adaptive neural network control strategy is built and the simulation and experiment study is done. Comparing the control result of the controller with the adaptive neural network control strategy to the control result of the controller with the traditional PID, and from the results of simulation and experiment, we can know that the mathematic model of the electro-hydraulic force control system controlled by the adaptive neural network control strategy is correct, and we can find that the controller with the new designed control strategy can not only increase the frequency response ability of system, but also improve the precision of system, and the controller can quicken the response ability of the system obviously.

Abstract: Mine ventilation system is a repairable system, with the characters of time-varied and randomness. It made us encounter some difficulties when we discussed the system failure evolution process and its operational reliability, as well as when we sought the system reliability parameters developing trends following with ventilation system operation time variation and air adjustment. Firstly, according to the reliability theory, the failure process characteristic values, which affect the ventilation system, were defined. And then, by mean of the analysis of the failure process deterioration of mine ventilation system, the deterioration and its amelioration discriminants are given basing on two parameters’ Weibull process. Finally, on the basis of adaptive neural network technology, the ventilation system failure processes were simulated, the failure process characteristic parameters were determined, and the failure curves were drawn accordingly. The results show that the failure process of the experiment mine ventilation system is a Non-homogeneous Poisson Process; the mean time between failures of the ventilation system is between 160h and 170h; the failure ratio curves present the trends of concave or convex variation when the system failure process deterioration or amelioration.

Abstract: This paper presents a neural network adaptive image edge detection method, and from neural network theory, this paper gives the formula of adaptive neural network algorithm; quantitative given the momentum factor and error, momentum factor and error on the weight vector of norm of the gradient of the quantitative relationship; and gives the algorithm flow diagram. Through experiment we get the conclusion: by using this adaptive neural network for image edge detection is feasible, and it has good generalization ability.

Abstract: Rolling temperature is an important factor affecting mechanical properties of hot rolled strip significantly. Traditional techniques cannot meet higher precision control imperatives. In the present work, a novel knowledge-based system has been developed for the temperature prediction in hot strip mills. Neural network has been used for this purpose, which is an intelligent technique that can solve nonlinear problem of temperature control by learning from the samples. Furthermore, an annealing robust learning algorithm was presented to adjust the hidden node parameters as well as the weights of the adaptive neural networks. Simulations in a multi-object mode have been carried out to verify the effectivity of new neural optimization system. Calculation results confirm the feasibility of this approach and show a good agreement with experimental values obtained from a steel plant.