Particle Swarm for Training Cascade Neural Network Based Models/Estimators

详细信息    查看全文

• 作者：V. Saranya^a ; ^{soni90kumar@yahoo.com} ; R. Raghuraman^a ; K. Arulselvam^a ; K. Thirunavukarasu^a ; A. Venkadesan^b
• 关键词：Particle Swarm Optimization ; Cascade Architecture ; Neural Network ; Nonlinear Models ; Estimators
• 刊名：Procedia Engineering
• 出版年：2012
• 出版时间：2012
• 年：2012
• 卷：38
• 期：Complete
• 页码：2433-2452
• 全文大小：3610 K

文摘

Artificial Neural Networks (ANNs) are powerful tools for modelling the non-linear functions. The performance of NN model to a large extent depends on the learning algorithm. The learning algorithm is used to find the optimum parameters (weights and biases) of the given neural architecture. The learning algorithm used in this paper is a population based optimization technique namely Particle Swarm Optimization (PSO). The PSO algorithm can be used for both the continuous/discontinuous excitation function. The PSO algorithm is used to train the Single Neuron Cascaded (SNC) Neural Architecture. The efficacy of the PSO trained SNC-NN is illustrated for modelling the nonlinear functions namely sine function and one bench mark function. To further illustrate the effectiveness of the designed PSO trained SNC-NN model, the practical example taken in this paper is the NN based flux estimator used in Drives applications.