Accurate predictive maintenance based on four parameters

本项目旨在验证基于四个参数的预测性维护假设。魁北克北部地区的经济高度依赖铁和铝的生产和转型。相关的工业活动保证了大部分人口的就业。传统的维护已不足以保证这些公司的竞争力。优化的资产管理和维护对于降低生产成本和机器停机时间至关重要。该项目将导致高精度维修方法的发展，改进3D建模和仿真培训，并在预测性维修方面建立一极专业知识。开发的模型将在两个试验台进行测试和完善；一个在c懊悔工厂Îles，另一个在Metal7使用的典型传送带上。对于不同的轴承和轴，将对两个台架进行应力、剪切和扭转以及疲劳和数值模拟。模拟将在不同的扭矩、转速和质量偏移情况下运行。同时，对这些机器部件的故障频率进行统计数据分析，并根据机器工作场景推断出故障趋势的信息。同时，使用制造商的规范，在可用的情况下，将建立机器操作场景的类似故障趋势。最后，试验台将配备加速度计、热传感器和对准传感器，以便实时跟踪机器元件在时间和机器工作情况下的故障。其思想是根据每个测量参数（数值模拟，统计数据，制造商的规格和仪器仪表数据）和不同的机器建立

This project is aimed at validating the assumption of predictive maintenance based on four parameters. The economy of the Northern Quebec region is highly dependent on iron and aluminium production and transformation. Related industrial activities ensure employment for a large proportion of the population. Traditional maintenance is no longer enough to ensure competitiveness of these companies. Optimised asset management and maintenance is essential to reduce production cost and machine downtime. This project will lead to the development of a high precision maintenance method, improve training in 3D modelling and simulation, and establish a pole of expertise in predictive maintenance. The developed model will be tested and refined on two testbeds; one at Cégep de Sept Îles industrial and on a typical conveyor used at Metal7. For different bearings and shafts, the stress, shear and torsion as well as fatigue and numerical simulation will be performed for both benches. The simulations will be run for different torques, rotational speeds and mass offset scenarios. In parallel, statistical data related to frequency of failure of these machine elements will be analysed and information inferred as to failure tendencies according to machine working scenarios. At the same time, using manufacturer's specifications, when available, similar failure trends with respect to machine operating scenario will be established. Finally, the testbeds will be instrumented with accelerometers, heat sensors, and alignment sensors in order to follow in real time, failure of the machine elements with respect to time and machine working scenario. The idea is to establish according to each measurement parameter (numerical simulation, statistical data, manufacturers' specifications and instrumentation data) and for different machine