桥式起重设备载荷预测研究

论文总字数：33990字

摘 要

桥式起重机具有作业稳定、起重量大、承载能力强，跨度大、整体稳定性好、可在特定范围内吊重行走、使用寿命长等特点，是目前使用最广泛、拥有量最大的一种轨道运行式起重机。桥式起重机的状态评估和预测对于安全生产具有重要意义。桥式起重机的健康状况和其所受的载荷息息相关，载荷的大小直接影响着桥式起重机的工作状态和健康状况。为了合理评估和预测起重机的剩余寿命，制定最优的维修养护计划，需要预测起重机未来所受载荷的大小。

本文以桥式起重机作为工程背景，基于某桥式起重机2018年的起吊载荷数据，建立基于Matlab的BP神经网络预测模型，对起重机的载荷进行预测。首先详细介绍了BP神经网络的基本原理，包括常用算法、激活函数等概念，编制了BP神经网络算法程序，将其应用于桥式起重机载荷预测研究。然后在BP神经网络算法框架基础上进行改进，改进方法一为降低网络灵敏度，改进方法二为运用遗传算法对网络的初始权值和阀值进行优化，改进方法三为运用粒子群优化算法进行初始权值阀值优化，将三种改进后的BP神经网络和原网络的计算结果进行了比较和分析。结果表明，降低网络灵敏度的BP神经网络训练速度最快，遗传算法优化初始权值阀值的改进算法训练精度最高，粒子群算法优化初始权值和阀值的改进算法训练速度和训练精度居中，性能最为均衡。运用三种优化算法后BP神经网络的预测性能均在原有的基础上有大幅提升，具备可靠性，可以实际用于解决桥式起重机的载荷预测问题。

关键词：载荷预测；BP神经网络；网络灵敏度；遗传算法；粒子群算法

Abstract

With the characteristics of stable operation, large lifting capacity, strong carrying capacity, large span, good overall stability, hoisting and walking in a certain range and long service life, the bridge crane is the most widely used and the largest orbital cranes. Due to the large size and long-term high-load operation of the bridge crane, it is difficult to carry out high-frequency manual maintenance and maintenance. Reasonable evaluation and prediction of its state are of great significance for safe production. The health of the bridge crane is closely related to the load it receives. The size of the load directly affects the working condition and health of the bridge crane. In order to properly estimate and predict the remaining life of the crane, to develop an optimal maintenance and maintenance plan, it is necessary to predict the load of the crane in the future.

In this paper, the bridge crane is used as the engineering background. Based on the hoisting load data of a bridge crane in 2018, a BP neural network prediction model based on Matlab is established to predict the load of the crane. Firstly, the basic principles of BP neural network are introduced in detail, including the commonly used algorithms and activation functions. The BP neural network algorithm program is compiled and applied to the research of bridge crane load prediction. Then, based on the BP neural network algorithm framework, the improvement method is to reduce the network sensitivity. The second method is to use the genetic algorithm to optimize the initial weight and threshold of the network. The third method is to use the particle swarm optimization algorithm to initialize. The weight threshold optimization is used to compare and analyze the results of the three improved BP neural networks and the original network. The results show that the BP neural network with reduced network sensitivity has the fastest training speed. The improved algorithm of genetic algorithm optimization initial weight threshold has the highest training precision. The particle swarm optimization algorithm optimizes the initial weight and threshold training speed and training accuracy. The most balanced. After using three optimization algorithms, the prediction performance of BP neural network is greatly improved on the basis of the original, and it has reliability, which can be used to solve the load prediction problem of bridge cranes.

Key words: Structural Health Monitoring; BP neural network; network sensitivity; genetic algorithm; particle swarm optimization

目 录

摘 要 Ⅰ

Abstract Ⅱ

第一章 绪论 1

1.1 引言 1

1.2 桥式起重机介绍 1

1.2.1 简介 1

1.2.2 主要问题 1

1.3 研究现状 2

第二章 BP神经网络在桥式起重机载荷预测中的应用 4

2.1 概述 4

2.1.1 基本概念 4

2.1.2 主要特点 5

2.1.3 缺点及不足 6

2.2 算法实现 6

2.2.1 常用的反向传播算法 6

2.2.2 激活函数 8

2.3 在桥式起重机载荷预测中的应用 11

2.3.1 数据来源及分析 11

2.3.1 网络构建与训练 14

第三章 降低网络灵敏度的改进BP神经网络及其应用 20

3.1 概述 20

3.2 算法实现 20

3.3 在桥式起重机荷载预测中的应用 21

第四章 基于遗传算法优化初始权值阀值的改进BP神经网络及其应用 26

4.1 概述 26

4.1.1 基本概念 26

4.1.2 遗传算法的一般步骤 26

4.2 算法实现 28

4.2.1 改进的适应度函数 28

4.2.2 改进遗传算法的主要操作步骤 29

4.3 在桥式起重机载荷预测中的应用 30

第五章 基于粒子群算法优化初始权值阀值的改进BP神经网络及其应用 37

5.1 概述 37

5.1.1 基本概念 37

5.1.2 类比过程 37

5.2 算法实现 37

5.2.1 粒子群优化算法的操作步骤 37

5.3 在桥式起重机载荷预测中的应用 39

第六章 结论与展望 46

6.1 主要结论 46

6.2 研究展望 49

参考文献 50

附录 遗传算法功能代码 52

致谢 56

剩余内容已隐藏，请支付后下载全文，论文总字数：33990字