【作者】 王伟；

【导师】 刘延俊； 薛钢；

【作者基本信息】 山东大学 ， 机械制造及其自动化， 2022， 硕士

【摘要】 发展海洋仪器,壮大海洋装备产业,推动海洋能源开发利用工作是建设海洋强国的重要保障。由于海洋的恶劣条件,传统的供电方式已经不能为海洋仪器提供持续稳定的电能,能源供应问题正在制约海洋仪器和装备的大范围推广。我国广阔的海域中蕴藏着丰富的海洋能,作为一种新型可再生能源,它具有能量分布集中、理论能量俘获效率高的特点。推行海洋能源开发,推广海洋波浪能在海洋仪器上的应用,是解决海洋仪器供电问题的主要路径。本文以锚泊浮台波浪能发电装置为基础,研究不同形状浮体的水动力特性,计算圆台形浮体的捕能功率,分析影响装置发电功率的关键因素,建立预测模型对输出功率进行预测,并引入相关参数评价模型的精度。本文首先阐述课题的研究背景和研究意义,介绍几种典型的波浪能发电装置以及它们的应用,探讨发电效率提升策略,回顾国内外关于波浪能发电技术的研究,阐述机器学习算法在该领域的应用,引出本文的主要研究内容。然后介绍本文所采用的波浪理论,包括线性波理论、斯托克斯波理论和不规则波理论,阐明它们的控制方程和边界条件,求解得到速度势函数、波面函数、弥散关系、水质点的速度和加速度。介绍不规则波在固定点和非固定点下波面方程的表达式,阐述描述能量分布的频谱和方向谱。针对不同形状(圆柱、圆锥、半球、圆台)浮体的水动力特性进行研究,建立线性能量输出系统作用下浮体的数学模型,利用数值模拟的方法研究浮体的频域响应特性和P-M谱激励下的时域响应特性,对比分析四种浮体的水动力性能差异,并进一步讨论半顶角对圆台形浮体的水动力特性和运动特性的影响。推导圆台形浮体的功率计算表达式,确定影响装置捕能功率的关键参数。采用拉丁超立方采样的方法建立样本库,计算样本的波浪激励力、附加质量、辐射阻尼和捕能功率,对关键参数进行相关性分析,量化它们对捕能功率的影响程度。预测圆台形浮体的最大捕能功率,设计一个三层的前馈神经网络,划分训练集和验证集,用训练好的模型计算样本在不同波况下的最大捕能功率。最后引入相关系数、均方根误差、误差百分比和P值来分析预测结果和仿真结果之间的一致性,并评价模型的精度。最后,对全文的工作进行总结,列出本文的主要结论和创新点,并进行研究展望。更多还原

【Abstract】 Developing marine instruments,expanding marine equipment industry and promoting the development and utilization of marine energy are important ways to build a marine power country.Due to harsh conditions of ocean,traditional power supply means cannot provide continuous and stable electric energy for marine instruments.The problem of power supply is restricting the wide promotion of marine instruments and equipment.The vast sea area in China is rich in marine energy.As a new type of renewable energy,it has the characteristics of centralized energy distribution and high theoretical energy capture efficiency.Promoting the development and application of marine energy on marine instruments is the main way to solve the power supply problem of instruments.This dissertation studies hydrodynamic characteristics of floating bodies on the wave energy power generation device of anchored floating platform with different shapes.Calculate the capture power of the frustum of a cone shaped floating body,analyze the key factors affecting power generation,establish a prediction model to predict the output power,and introduce relevant parameters to evaluate the accuracy of the model finally.This dissertation firstly describes the research background and significances of this subject,and introduces several typical wave energy converters and their applications.It also discusses the strategies to improve power generation,and reviews wave energy generation technology and application of machine learning in this field at home and abroad.Finally,it illustrates research contents of this dissertation.After that,it introduces wave theory used in this paper,including linear wave theory,Stokes wave theory,and irregular wave theory.After giving governing equations and boundary conditions,velocity potential function,wave surface function,dispersion relationship,velocity and acceleration of water points are calculated.Then,wave surface equation of irregular wave at fixed and non-fixed points,and frequency spectrum and directional spectrum used to describe energy distribution are introduced.It studies hydrodynamic characteristics of floating bodies with different shapes,which are cylinder,cone,hemisphere,and the frustum of a cone shaped floating.A mathematical model with linear power output system is established.The response characteristics of floating body in the time domain under P-M spectrum excitation,as well as in the frequency domain,are studied by numerical simulation.The hydrodynamic differences between four floating bodies are compared and analyzed.Half apex angle’s effects on hydrodynamic and motion characteristics of the frustum of a cone shaped floating body are investigated further.The power calculation formula of the frustum of a cone shaped floating body are deduced,and key parameters that effect the capture power are determined.A sample database is established using the Latin hypercube sampling method.Wave excitation force,additional mass,radiation damping,and capture power of samples are calculated.Finally,a correlation analysis is conducted to quantify parameters’ correlation with capture power.The maximum energy capture power of the dome shaped floating body are predicted.It designs a three-layer feedforward neural network.After dividing the training and verification set,it is used to calculate sample’s maximum capture power under different wave conditions.Introduce the correlation coefficient,root mean square error,error percentage,and P value to analyze the consistency between forecasting results and simulation results and evaluate model’s accuracy.Finally,it summarizes the full text.It lists main conclusions and innovations,and makes research prospects.更多还原