基于虚拟材料模型的螺栓松动预测研究

Bolt loosening prediction method based on virtual material model

螺栓连接在风力发电机等复杂装备中得到了广泛应用。设备的工作载荷及机械振动会引起螺栓松动,导致螺栓连接失效,严重影响设备的正常运行。为此,以海上风力发电机塔筒连接螺栓为研究对象,建立了螺栓结合面连接刚度模型,进行了塔筒法兰连接螺栓的松动状态预测研究。首先,建立了螺栓结合面刚度,通过虚拟材料模型对其进行了表征,探究了预紧力对虚拟材料层模型的影响特性,为有限元模型提供了参数;然后,通过力锤实验,并结合实测数据建立了高精度的有限元模型;最后,通过法兰连接件的前10阶扭转、弯曲固有频率,以任意两阶模态的频率变化平方比、预紧力变化前后频率相对变化比分别作为螺栓定位、定量指标,进行了螺栓松动预测。研究结果表明:随着预紧力的增加,固有频率呈现逐渐增大的趋势,但灵敏度稍有下降;通过对螺栓松动测试结果的统计可得,螺栓松动定位识别精度达到95.24%;单、多螺栓预紧力下降程度定量精度分别为93.4%、90.2%;基于虚拟材料模型的螺栓松动预测方法具备精度高、通用性强的特点,可为螺栓连接的数字孪生实时监测提供重要参考。

Bolting is widely used in complex equipment such as wind turbines.The working load and mechanical vibration of the equipment will cause bolt loosening,and lead to the failure of bolt connection,which will seriously affect the normal operation of the equipment.Therefore,the connection stiffness model of bolt joint surface was established by taking the connection bolts of offshore wind turbine tower barrel as the research object,and the loosening state prediction of the connection bolts of tower barrel flange was carried out.Firstly,a virtual material model was used to characterize the connection stiffness of the bolt joint surface,the influence of preload on the virtual material layer model was explored,and the parameters for the finite element simulation were provided.Then,a high precision finite element model was established by combining the force hammer experiment with the measured data.Finally,the first 10 torsional and bending natural frequencies of flange connectors were used,the square ratio of frequency change of any two modes and the relative change ratio of frequency before and after the preload change were respectively used as bolt quantitative indicators for bolt positioning and loosening prediction.The results show that with the increase of preload,the natural frequency of tower connection increases gradually,but the sensitivity of natural frequency decreases slightly.According to the statistics of the test results of bolt loosening,the identification accuracy of bolt loose positioning reaches 95.24%.The prediction accuracy of preload reduction degree of single bolt and multiple bolt is 93.4%and 90.2%respectively.The research of bolt loosening prediction based on virtual material model has the characteristics of high precision and strong universality,which can provide an important basis for real-time monitoring of digital twin.