机械制动器摩擦故障快速融合诊断与智能预报方法研究

This project focuses on the friction faults of mechanical brakes (FFMB) which are defined as any braking performance deteriorations connected with abnormal tribological variations of the brake pairs. By taking the disc brake as research object, this project will investigate on the fast fusion diagnosis and intelligent forecast methods of the FFMB for avoiding any connected braking accidents. Firstly, based on some simulating braking experiments, the dynamic friction features of the brakes will be extracted and the objective method characterizing the dynamic friction states will be established. Then the fault features of the FFMB will be extracted and the fault mode will be identified based on the objective tribological characterizing method. Secondly, based on some FFMB diagnosis experiments, the quick signal collecting method, the high reliability signal transmission technology and the fast fusion analysis method based on multi-sensor data will be bulit up. Then the fast fusion diagnosis method based on multi-source informations for the FFMB will be established by the evidence theory and artifical intelligent (AI) technology. Finally, an intelligent predicting method for the friction state of brakes will be founded based on the AI technology and the simulating braking experiments. The intelligent forecast method for the FFMB will be established based on the fuzzy expert system and the diagnosis experiments. It is believed that the results may be valuable and practical for improving the working reliability of the mechanical brakes and the braking safety of the mechanical systems.

本项目将机械制动器由于摩擦状态异常变化而导致其制动性能下降的现象称为"摩擦故障"，以盘式制动器作为研究对象，以避免摩擦故障引发的制动事故为研究目标，开展制动器摩擦故障快速融合诊断与智能预报方法研究。首先，通过模拟制动试验，研究制动过程中的动态摩擦特征，构建制动器动态摩擦状态的客观表征方法，并基于此研究制动器摩擦故障的特征提取和模式识别方法。其次，通过制动器摩擦故障诊断试验，研究制动器摩擦状态信号的快速采集方法、高可靠性传输技术和多传感器信息快速融合分析机制；基于证据理论和人工智能技术，研究制动器摩擦故障的异类多源信息快速融合诊断方法。最后，基于人工智能技术和制动器摩擦试验结果，研究制动器摩擦状态的智能预测方法；基于模糊专家系统和制动器故障诊断试验结果，研究制动器摩擦故障的智能预报方法。本项目的研究结果对于提高机械制动器的工作可靠性和保障机械系统的制动安全，都将具有重要的理论价值和现实意义。

本项目针对机械制动器因摩擦状态异常变化而导致制动性能下降的摩擦故障现象，建立了制动器摩擦状态的客观表征及融合监测方法、摩擦故障的模式识别及智能诊断与预报方法，研究结果可为提高机械制动器工作可靠性、避免摩擦故障引发的制动事故提供重要的理论基础和技术指导。首先，通过开展制动摩擦学试验，研究了制动过程中摩擦因数、摩擦温升等动态摩擦特征状态量与制动工况等外部影响因素间的内在联系，构建了制动器动态摩擦状态客观表征方法；其次，通过开展制动器摩擦状态监测试验，针对制动过程时间较短的特点，分析了PLC两种通讯方式在短时制动过程中的应用方式，建立了制动器摩擦状态信号快速采集及高可靠性传输方法；再次，从传热学、摩擦学等基本原理出发，分析了预置、预埋及红外三种摩擦温升监测方法的特征与缺陷，并分别对其进行了修正，进而建立了基于支持向量机的摩擦温升融合分析与监测模型，并通过了仿真试验进行了验证；最后，通过开展制动器摩擦故障诊断试验，基于人工智能技术构建了制动器摩擦故障模式识别与智能诊断模型；进一步，基于摩擦故障智能诊断模型和盘式制动器模拟制动试验台，构建了制动器摩擦故障智能诊断与预报系统，实现了对制动器摩擦因数过小、不稳定、突变以及摩擦温升过高等摩擦故障的诊断与预报。在项目执行期内，已发表（或录用）学术论文6篇，其中：SCI源刊3篇；申请并公开发明专利2项；培养硕士研究生2人。