电梯全寿命周期故障预测与智能健康管理方法研究

Facing the people's livelihood demands of the safety and reliability of elevator operation, to achieve the life cycle fault prognostics, remaining useful life(RUL) estimation and the intelligent health management for an elevator, this project studies the main fault modes and fault mechanism of an elevator, establishes the fault prognostics models based on the physics of failure(PoF) and validates the availability; studies the fault information perception, data acquisition and condition monitoring methods for the elevator prognostics and health management (PHM)system. Focused on the multi-source, complexity and non-linearity of the monitored data, the fault feature recognition, feature extraction and fusion algorithms are studied to obtain the main feature principal components or health factors. Considering the uncertain conditions, the health assessment and prognostics methodology are studied based on the fusion of the physic failures and data driven. Based on the multi-source data fusion and multiple model combinations within the fault-related conditions, the methodology of remaining useful life(RUL) prediction is studied. The life cycle multi-objective intelligent health management model is established. Based on the fusion models, time series, cost risk and empirical method, the optimal maintenance decision model is established, and then choosing the optimal detection and maintenance period for the availability and economy of the elevator. The fault prognostics and intelligent health management can be realized eventually. This project aims to providing timely warning of faults, reducing the cost within the life cycle, improving the safety, reliability and economy for an elevator.

面向电梯安全可靠运行的重大民生需求，以实现电梯全寿命周期的故障预测、剩余寿命评价和智能健康管理为目标，对电梯进行主要故障模式和机理分析，建立和验证基于物理失效的电梯故障预测模型；研究面向电梯PHM系统的故障信息感知、数据采集和状态监测方法；针对电梯监测数据的多源性、复杂性、非线性等，研究其特征识别、提取及融合算法，获得特征主元或健康因子；研究考虑不确定性条件下基于物理失效和数据驱动融合的电梯健康评估与故障预测方法、考虑故障相关性条件下基于多源数据融合和多种模型组合的电梯剩余寿命评价方法；建立电梯全寿命周期多目标最优预知智能健康管理模型和基于融合模型、时间序列、成本风险、经验方法的最优维修决策模型，制定面向可用性和经济性的最优检测与维修周期，实现对电梯全寿命周期的故障预测与智能健康管理，旨在提供电梯故障及时预警，降低电梯全寿命周期成本，提高电梯运行安全性、可靠性和经济性。

面向电梯安全可靠运行的重大民生需求，本项目以杭州西奥电梯有限公司生产的XO-REZO型高速乘客电梯为研究对象，开展了乘客电梯的故障预测、剩余寿命评价和智能健康管理研究。首先围绕曳引机、制动器等影响电梯运行安全的关键机械部件开展了FMEA分析，通过试验分析并结合工程经验，确定了曳引机、制动器的主要故障模式和故障原因，选用了钢丝绳蠕动距离、制动轮的制动距离作为故障表征参量。随后，确定了数据采集方案，在曳引轮等部件上加装编码器对蠕动距离、制动距离进行测量。然后，建立并得到了蠕动距离和制动距离的数学表达式及其阈值，并开展了仿真分析；在浙江省特科院国家电梯中心、杭州西奥电梯的用户单位以及自行搭建的试验电梯等上开展了蠕动距离、制动距离等的试验，建立了蠕动距离、制动距离的性能退化模型，并开展了剩余寿命研究；提出了基于改进的神经网络、改进迁移算法等的预测方法。然后，建立了一种基于熵值法、层次分析法和灰色关联分析法相结合的电梯运行性能评估模型和方法，解决了传统灰色关联分析指标赋权主观性过强的问题，并开展了电梯性能状态的评估。随后，提出了检测策略决策因子并建立了电梯部件的状态维修决策模型；针对电梯多部件维修的决策问题，建立了电梯机械系统多部件的系统级状态机会维修策略模型；针对维修调度时的路径规划问题，建立了电梯维修多场站调度路径规划的数学模型，并提出了一种基于两阶段求解的求解方法，提高算法的整体求解效率，为多个维修场站的任务调度提供一种新的思路和方法。最后，开发了面向电梯全寿命周期的PHM系统原型机，在杭州西奥电梯的用户单位--杭州临平区西子国际大厦、临和家园、枉山嘉苑等地的电梯上开展了推广与应用。