上游泵送机械密封多目标多工况优化研究

为了获得具有较强适应工况变化能力的上游泵送机械密封,以微间隙内流场的CFD计算为基础,建立涉及多工况的多目标优化数学模型,综合分析相关研究成果和密封特性确定了优化变量及其取值范围、工况参数取值范围和优化目标。应用模糊集理论确定不同工况下的液膜刚度和泄漏量的权重因子,并研究人工神经网络模型对优化结果的影响。基于穷尽搜索法和遗传算法对密封进行优化设计,对比分析2种优化方法得到的优化结果,并对优化前后的密封润滑膜流场特性进行BVF诊断分析。优化研究表明:人工神经网络模型很好地拟合了液膜刚度、泄漏量随槽型参数变化的函数关系;2种优化方法都能提高密封性能,但遗传算法优化比穷尽搜索法优化能获得更好的优化结果。

Research on Multi-objective and Multi-condition Optimization of Upstream Pumping Mechanical Seal

To obtain the upstream pumping mechanical seal with strong adaptability to changing working conditions,a mathematical model for multi objective and multi-condition optimization based on CFD flow field calculation was established.The range of working condition,optimization variables (groove depth,spiral angle,groove diameter ratio and groove width ratio) and optimization targets (liquid film stiffness and leakage) were determined.The fuzzy set theory was used to determine the weighting factors of liquid film stiffness and leakage under different working conditions.The influence of artificial neural network model on the optimization results was studied.Based on the full search method and genetic algorithm,the optimal design of the seal was carried out.The optimization results obtained by the two optimization methods were compared and analyzed.The BVF theory was applied to diagnose and analyze the flow field inside the seal before and after optimization.The research results show that the artificial neural network model fits well the function relationship between liquid film stiffness and leakage with the change of groove parameters.Both optimization methods are able to improve the sealing performance.The optimization results of genetic algorithm is better than those of the full search method.