| Along with the rapid development of the country’s automobile industry,the huge consumption of energy and air pollution caused by exhaust emissions are becoming more and more serious,and clean energy vehicles fueled by compressed hydrogen can greatly alleviate the two major problems of energy scarcity and air pollution.In recent years,with the rapid development of advanced composite materials and the improvement and perfection of fiber winding molding process,the development and application of high-pressure hydrogen storage pressure vessels are becoming more and more mature and widely used in various fields.Since there are no design principles and calculation formulas for each structural parameter in the product standards for various types of hydrogen storage pressure vessels,for a long time,the structural design can only rely on the accumulation of practical experience to determine the design scheme.According to the required function of the structure,the geometry of the structure and the selected materials are predetermined according to the experience,and then the structural analysis is carried out and the stresses derived from the analysis are compared to determine whether it is feasible.For the pressure vessel parameter design problem,this paper proposes a parameter prediction model based on an improved Aquila optimization algorithm to optimize the neural network structure.The main work of the paper includes:(1)Aiming at the various problems of insufficient actual prediction accuracy and slow prediction speed of the underlying neural network in practical engineering applications,a prediction model based on the Aquila optimization algorithm to optimize the ENN(Elman Neural Networks,ENN)neural network is constructed to improve the prediction accuracy of this neural network in practical applications.The weights in the neural network as well as the structural parameters are encoded as an individual Aquila in the Aquila optimization algorithm,and the Aquila algorithm can effectively adjust the weight parameters of the neural network to improve the prediction accuracy and generalization ability of the neural network.The simulation experimental results show that the model proposed in this section has better prediction effect on the pressure vessel dataset problem.(2)For the prediction models of Aquila optimization algorithm and ENN,Aquila optimization algorithm has the problems of slow convergence speed,easy to fall into local optimum and possible gradient explosion of ENN,an adaptive Aquila optimization algorithm based on t-distribution is proposed to optimize wavelet ENN model.In the Aquila optimization algorithm,firstly,the tdistribution with thick tails is used to perturb the individual positions to enhance the convergence speed of the algorithm;secondly,the adaptive weight update position strategy is used to improve the position update method of exploring individuals,which enhances the ability of local exploration and global exploitation of the algorithm,and finally,the wavelet transform function is introduced instead of the sigmoid function of the ENN.Among them,the adaptive Aquila optimization algorithm based on t-distribution is tested and validated with three basic algorithms and three improved algorithms based on nine benchmark test function problems,and the experimental results show that the proposed algorithm has better response speed and more stable accuracy.The wavelet function-based ENN model is experimented on the existing pressure vessel data set,and the results show that the improved neural network model has higher accuracy,smaller error and faster speed compared with the traditional ENN.(3)To address the problems of long simulation period and high prediction cost in the application of pressure vessels,the proposed prediction model with optimized neural network structure by improved Aquila optimization algorithm is applied to the type III pressure vessel parameter optimization problem,design software for design of pressure vessel parameters.It is experimentally verified that the optimized neural network prediction model using the improved Aquila optimization algorithm has higher prediction accuracy and generalization capability for pressure vessel parameter prediction,which provides a new idea for optimizing the neural network structure and can be widely applied to various prediction problems. |
