基于模型辨识的在役起重机结构损伤识别理论与方法研究

Focusing on the in-service gantry cranes and pedestal cranes, the theory and methodology for structure damage detection based on the nonlinear vibration model identification is researched systematically. This key idea is to partition the whole structure of the crane into a series of sub-structures firstly and then to establish the nonlinear vibration transmission models between the sub-structures under the moving load exciting, with the analysis of the spatial distribution shape of the feature parameters extracted from the model array, a novel structure damage identification method for in-service cranes is proposed which overcomes the difficult to obtain the baseline from the intact structure that often encountered in most existing structure parameter based damage detection methods. The structure identification method, global optimal parameter estimation algorithm and the model validation criterion for the damage detection oriented nonlinear time series model of crane vibration excited with moving load are developed, and the corresponding high-sensitive, robust and reliable damage detection features from the time-domain and frequency-domain model parameters is constructed. The optimal sub-structure partition method based on the structural mode shape and the dynamic vibration response and its propagation characteristic is put forward and the optimal sensor placement with multi-objective criteria and its optimization algorithm are proposed also. Finally, on the basis of the above research, with the theory of probability statistics, damage mechanics and fracture mechanics, the damage propagation law and the residual strength degradation model for the crane structure is established to assess the structure security and reliability of in-service cranes, the assessment standard is generated for the real operation instruction.

以门式/门座式起重机为对象，将其结构划分为一系列子结构单元，建立移动载荷作用下不同子结构单元间振动信息的非线性传递模型，提出基于所建立的非线性振动传递模型特征参数阵列在子结构空间上分布形态的变化实现在役起重机结构损伤识别的理论与方法，避免传统基于结构参数的损伤识别方法需要建立无损伤结构基准模型的不足。提出移动载荷作用下，面向损伤识别的起重机振动非线性时间序列模型的结构辨识方法、全局最优参数估计算法及模型适用性检验准则，构建具有高检测灵敏度、基于非线性模型时域和频域参数、稳健可靠的损伤识别特征量；提出基于起重机结构振型与动态响应及其传播特性的最优子结构划分方法、面向损伤识别的传感器布置多目标最优化准则及算法，在此基础上，研究基于概率统计、损伤力学和断裂力学的起重机结构损伤演变规律，建立结构剩余强度衰减模型，实现在役起重机结构安全与运行可靠性评估，并编制相应的评估规范。

以在役门式/门座式起重机为对象，将其结构划分为一系列子结构单元，首先，提出了冲击载荷作用下不同子结构单元间振动信息的非线性传递模型——GNARX模型，系统研究了GNARX模型基于渐消记忆最小二乘的参数估计方法、基于参数离差率结构剪枝的结构辨识方法、基于白噪声、Akaike信息准则、残差双谱检验等模型有效性验证方法、基于模型参数的特征提取与模式识别方法，实现了无需健康结构基准模型条件下起重机结构损伤的识别；其次，提出了基于主梁应变模态且无需参考原始振动数据的损伤识别方法，通过应用差分算法对主梁的应变模态值进行处理，结合三次样条插值得到主梁的应变模态差分曲线，选择三次样条插值得到的系数矩阵构建损伤位置敏感系数（IDSI），并计算主梁均布节点的损伤位置敏感系数IDSI来实现损伤定位，进而，使用已得到的损伤区域附近的未损伤节点的数据构建ARMA模型，通过ARMA的预测功能定量分析损伤对应变模态差分曲线的影响，实现了损伤程度的判定；同时，研制了基于时间-数字转换(TDC: Time Digital Converter)技术的起重机无线状态采集系统，采用3G或“ZigBee+WiFi”无线通信技术，实现起重机应变、主梁挠度、运动姿态等的在线检测、远程传输和监测分析，开发完成了面向Internet的起重机远程应变在线监测软件，可定时接收底层检测单元采集的应变及起重机运动姿态等数据，并可再此基础上获取起重机运行载荷谱/应力谱，为起重机安全评价提供数据支撑。最后，结合理论研究与起重机安全评价实践，制定了《港口大型起重机械结构健康监测与安全评价指南》，为规范起重机结构健康监测与安全评价活动，明确起重机结构健康状态评价指标、数据来源、方法、标准及结论，保证工作质量、方便管理，提供了相应的指导。