基于人工智能算法和CFD仿真的滑动轴承优化设计

发布时间：2018-04-10 22:24

本文选题：滑动轴承 + 雷诺方程　；参考：《哈尔滨工业大学》2014年博士论文

【摘要】：人类最早发明的轴承是滑动轴承，它具有形状简单、生产廉价、寿命长和容易使用等优点。如今，滑动轴承种类丰富，依然广泛应用在中小型卧式水电机组、中高档汽车增压器、内燃机曲轴、舰船螺旋桨、甚至低温火箭发动机涡轮泵中。面向不同的工况和要求，滑动轴承形状变得日益复杂，涉及的材料也延伸到合金和非金属材料等，制造过程通常还需要特殊的工具和技术等。在环境、能源和材料消耗问题日益突出的大背景下，滑动轴承技术发展需要同时兼顾性能可靠、易于维护、减少润滑剂用量、降低能源消耗、最大限度发挥材料和制造技术的潜在能力。滑动轴承的设计分析一直是工业摩擦学研究的重要内容之一，理论上已经基本建立了轴承性能与使用工况之间的数值计算关系，形成了可行的设计分析方法，如有限差分法、有限元法和计算流体力学方法，但是这些方法在工业界应用依然存在收敛问题和准确性判断问题，而且，随着工程设计和应用对滑动轴承的复杂计算精度要求越来越高，这些问题会更加突出。本文主要利用现代数值方法和计算技术来快速设计和优化流体动力润滑滑动轴承，实现设计流程规范快速可靠，特别是在充分发挥润滑剂、制造能力和材料性能潜力的基础上，对现有轴承产品进行系列化优化和改进过程中，，建立高效快速可靠的计算方法和流程，依然具有十分重要的理论意义和实用价值。论文首先用有限差分法来求解流体动压滑动轴承的性能，建立了一种新的简化的对角矩阵方法来求解非线性方程，获得了轴承性能参数的边界可行域；然后利用人工神经网络、遗传算法和人工蜂群算法，建立了轴承润滑剂流量、功率消耗的多目标优化模型，利用此模型，在不更改润滑剂流变参数的前提下，以材料极限性能、制造表面粗糙度和最大温升为控制变量，获得了优化的轴承结构参数，有效降低了轴承流量和功率消耗；接下来利用计算流体力学软件，对经过优化的径向滑动轴承，建立了包含轴承尺寸参数和润滑剂进口几何结构的仿真分析模型，获得了进油口数目、结构形状、尺寸参数与轴承内部温升、润滑剂空化、贫油和入口处润滑剂逆向流动之间的对应关系，为进一步优化轴承结构和参数提供了依据。为验证仿真分析和优化结果，设计制造了滑动轴承性能的试验测试装置，通过试验收集了大量关于轴承工作温度和转矩的数据，理论计算和试验测试结果吻合良好，并用于仿真结果对比和校正仿真分析模型。最后结合真实滑动轴承案例进行了多目标优化分析，再次验证了这些方法和模型能够较大幅度降低轴承功率消耗、润滑剂流量消耗和轴承温升，阐述并给出了清晰的优化流程，将极大地方便滑动轴承工业领域的快速设计、优化和应用。虽然现有独立的轴承性能优化函数模型在理论上是可行的,但是工程应用极为不便；而且因为多个目标函数独立优化，没有建立关联关系，优化结果很难同时满足多变量多目标优化函数，从而导致优化失败。而本文建立的方法成功解决了这一问题，可以实现全局最优，缩短了试验和实际工况应用的距离,使得系列轴承产品的优化不仅高度可行,而且价格低廉。
[Abstract]:The earliest human invention is bearing sliding bearing, it has simple shape, cheap production, the advantages of long service life and easy to use. Now, the sliding bearing variety, is still widely used in medium and small-sized horizontal hydroelectric generating units, high-grade automotive turbocharger, crankshaft, ship propeller, and cryogenic rocket engine turbo pump. For the different conditions and requirements, the sliding bearing shape becomes more complex, involving the material also extends to the alloys and non-metallic materials, manufacturing process usually requires special tools and technology. In the environment, can the big background the increasingly serious problem of source and material consumption, the need of the development of technology while taking into account the reliable performance of sliding bearing, easy maintenance, reduce the amount of lubricant, reducing energy consumption, maximize the potential of materials and manufacturing technology.
Analysis and design of sliding bearing is one of the important content of tribology research industry, the theory has been established between the bearing performance and operation condition of numerical calculation, the formation of a feasible analysis method design, such as finite difference method, finite element method and computational fluid dynamics method, but these methods still exist in the industrial sector the problem, convergence and accuracy of judgment and, with the requirement of complex calculation precision bearings of the engineering design and application is more and more high, these problems will become more prominent. In this paper, using modern numerical methods and computing technology to quickly design and optimization of hydrodynamic lubrication of sliding bearing, fast and reliable design of process specification, especially in full play lubricant, basic manufacturing capacity and material performance potential, a series of optimization and improvement in the process of the existing bearing products, construction It is still of great theoretical significance and practical value to establish a high efficient, fast and reliable calculation method and process.
This paper use finite difference method to solve the hydrodynamic performance of sliding bearing, a new simplified diagonal matrix method to solve nonlinear equations, the boundary of the feasible region bearing performance parameters; then using artificial neural network, genetic algorithm and artificial bee colony algorithm, a multi bearing lubricant flow a multi-objective optimization model of power consumption, using this model, the premise does not change the lubricant rheological parameters, to limit performance, making the surface roughness and the maximum temperature rise as control variables, optimized bearing structure parameters, effectively reducing the bearing flow and power consumption; then using the CFD software, the after optimization of radial sliding bearing, including simulation of bearing size parameters and lubricant inlet geometry analysis model, the oil inlet of the number, structure shape, scale The internal parameters and the bearing temperature rise inch, lubricant cavitation, the relationship between lean and the reverse flow of the lubricant, and provides a basis for the further optimization of bearing structure and parameters.
In order to verify the simulation analysis and optimization, performance test device of sliding bearing is designed and manufactured. Through the experiment collected a large number of bearing working temperature and torque data, theoretical calculation and test results are in good agreement, and for the comparison of the simulation results and calibration simulation analysis model. Finally, the real case of multi-objective sliding bearing optimization analysis proved these methods and models can greatly reduce the power consumption of the bearing, lubricant flow consumption and temperature rise, expounds and gives the optimization process clear, will greatly fast design of sliding bearing industry to facilitate the optimization and application.
Although the existing independent bearing performance optimization function model is feasible in theory, but the application is very inconvenient; and because the multiple objective function optimization of independent, no established relationship, the optimization result is difficult to meet the multi-objective optimization function of many variables, resulting in optimization of failure. And this method is successfully solved this problem, can achieve the global optimum, shorten the test and the actual condition of the application of distance, so the optimization series of products bearing not only highly feasible, and the price is low.

【学位授予单位】：哈尔滨工业大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TH133.31;TP18

【参考文献】