Fuzzy neural network learning based on hierarchical agglomerative T-S fuzzy inference

Abstract

It is well-known that the accuracy of classification prediction is relatively high, but the prediction result is obscure in concept since result is given in two-value form (0 or 1) which says that red tide exists or does not exist. On the other hand, the accuracy of numerical prediction is relatively low, but it can offer density value of plankton which influences red tide. In order to combine characteristics of the above mentioned two methods, a prediction method for red tide which is mixed with integration model of hierarchical agglomerative T-S fuzzy inference is proposed. In the thesis, through using the proposed prediction method mixed with integration model of hierarchical agglomerative T-S fuzzy inference, taking respective advantages of classification prediction and numerical prediction in prediction process for reference, and through experiment and comparison, it is proved that this algorithm is better than LMBP algorithm in prediction accuracy which shows the validity of the proposed algorithm. In the next step, it is mainly to further study the practical application of the algorithm, and to apply this prediction model to red tide warning system, and also to conduct experimental verification for a certain period by using actual marine environment.