基于改进PSO优化神经网络的水泵全特性预测研究
发布时间：2018-11-13 11:53
【摘要】：水泵广泛使用在国民经济的各个部门，而随着国民经济的发展，对水泵及其输运系统的运行安全性要求越来越高。然而对于水泵的研究还远远不能满足实际需要，尤其是水泵全特性参数的研究。水泵全特性参数能够表示水泵—水轮机的各种工况，对于水泵及其输运系统的水力过渡过程的计算和分析至关重要，而水力过渡过程的计算直接影响水泵输运系统的设计与建成后的运行安全。虽然在水泵全特性数据的获取与预测方面，中外的学者们做了大量工作，但是截止目前实测数据稀少，预测工作的精度仍有待提高，因此寻找更好的模型对水泵全特性参数进行预测具有重要的实际意义。 本文的主要研究内容和成果有： 1、详细介绍了水泵的相关理论，分析了水泵全特性曲线的应用，并对水泵全特性数据的获取及预测方法进行了总结。 2、对神经网络的神经元模型、学习算法、分类及常用模型进行了介绍，分析对比了几种神经网络的特点。对粒子群算法原理及其发展进行了介绍。构建了采用自适应惯性权重的粒子群算法来优化RBF神经网络的预测模型。 3、在MATLAB R2012b平台上，利用MATLAB提供的神经网络工具箱和GUI工具箱，基于提出的预测模型，，开发出水泵全特性曲线参数预测软件。依据现有数据对未知水泵的全特性参数进行预测，采用合理方法对预测结果进行评价分析，并与其它方法进行对比，体现出本文方法的优势。 4、将开发的预测软件应用于实际工程中，优化了水力过渡过程的计算分析，对于水泵输运系统中可能发生的问题，提出合理的防护措施。
[Abstract]:Pumps are widely used in various sectors of the national economy, but with the development of the national economy, the operational safety requirements of pumps and their transport systems are becoming more and more high. However, the research on the pump is far from meeting the actual needs, especially the study of the full characteristic parameters of the pump. The full characteristic parameters of the pump can express the various working conditions of the pump and turbine, which is very important for the calculation and analysis of the hydraulic transition process of the pump and its transportation system. The calculation of hydraulic transition process directly affects the design and operation safety of pump transportation system. Although scholars at home and abroad have done a great deal of work in obtaining and predicting the full characteristic data of pumps, the precision of prediction work needs to be improved because of the scarcity of measured data so far. Therefore, it is of great practical significance to find a better model to predict the full characteristic parameters of the pump. The main contents and achievements of this paper are as follows: 1. The related theory of water pump is introduced in detail, the application of the full characteristic curve of water pump is analyzed, and the method of obtaining and predicting the data of the whole characteristic of water pump is summarized. 2. The neuron model, learning algorithm, classification and common models of neural network are introduced, and the characteristics of several neural networks are analyzed and compared. The principle and development of particle swarm optimization (PSO) are introduced. An adaptive particle swarm optimization (PSO) algorithm is proposed to optimize the prediction model of RBF neural network. 3. On the platform of MATLAB R2012b, using the neural network toolbox and GUI toolbox provided by MATLAB, based on the proposed prediction model, the software for predicting the parameters of the full characteristic curve of water pump is developed. Based on the existing data, the full characteristic parameters of the unknown pump are forecasted. The reasonable method is used to evaluate and analyze the prediction results, and compared with other methods, the advantages of this method are reflected. 4. The developed prediction software is applied to practical engineering, the calculation and analysis of hydraulic transition process are optimized, and reasonable protective measures are put forward for the problems that may occur in the pump transportation system.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TV136.2;TH38
本文编号：2329033
[Abstract]:Pumps are widely used in various sectors of the national economy, but with the development of the national economy, the operational safety requirements of pumps and their transport systems are becoming more and more high. However, the research on the pump is far from meeting the actual needs, especially the study of the full characteristic parameters of the pump. The full characteristic parameters of the pump can express the various working conditions of the pump and turbine, which is very important for the calculation and analysis of the hydraulic transition process of the pump and its transportation system. The calculation of hydraulic transition process directly affects the design and operation safety of pump transportation system. Although scholars at home and abroad have done a great deal of work in obtaining and predicting the full characteristic data of pumps, the precision of prediction work needs to be improved because of the scarcity of measured data so far. Therefore, it is of great practical significance to find a better model to predict the full characteristic parameters of the pump. The main contents and achievements of this paper are as follows: 1. The related theory of water pump is introduced in detail, the application of the full characteristic curve of water pump is analyzed, and the method of obtaining and predicting the data of the whole characteristic of water pump is summarized. 2. The neuron model, learning algorithm, classification and common models of neural network are introduced, and the characteristics of several neural networks are analyzed and compared. The principle and development of particle swarm optimization (PSO) are introduced. An adaptive particle swarm optimization (PSO) algorithm is proposed to optimize the prediction model of RBF neural network. 3. On the platform of MATLAB R2012b, using the neural network toolbox and GUI toolbox provided by MATLAB, based on the proposed prediction model, the software for predicting the parameters of the full characteristic curve of water pump is developed. Based on the existing data, the full characteristic parameters of the unknown pump are forecasted. The reasonable method is used to evaluate and analyze the prediction results, and compared with other methods, the advantages of this method are reflected. 4. The developed prediction software is applied to practical engineering, the calculation and analysis of hydraulic transition process are optimized, and reasonable protective measures are put forward for the problems that may occur in the pump transportation system.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TV136.2;TH38
【参考文献】
相关期刊论文 前4条
1 王文海,冀四梅;水泵性能曲线的数学化处理方法[J];北京建筑工程学院学报;2003年02期
2 张华娟;李春;;基于性能预测的水泵优化设计方法[J];上海理工大学学报;2006年04期
3 郑铭,陈池,袁寿其;水锤数值计算的全特性曲线法[J];农业机械学报;2000年05期
4 聂书彬,关醒凡,刘厚林;利用人工神经网络预测离心泵性能的探索[J];水泵技术;2002年05期
本文编号：2329033
本文链接：https://www.wllwen.com/kejilunwen/jixiegongcheng/2329033.html
