人工免疫和证据理论集成应用于旋转机械并发故障诊断中的研究

发布时间：2018-03-06 20:38

本文选题：人工免疫　切入点：证据理论　出处：《太原理工大学》2011年硕士论文　论文类型：学位论文

【摘要】：旋转机械广泛应用于工业领域,其主要由转子、支撑转子的轴承系统、定子或机器壳体、联轴器等部件构成,通过旋转运动来完成工作。旋转机械作为一些工业部门的关键核心设备,企业的生产将直接受到这些设备运行状况优劣的影响,如果发生故障停机,将带来巨大的经济损失和严重的甚至灾难性的后果。故障诊断技术,即了解机械的运行状态,预测其可靠性,识别机械故障的部位、原因、危险程度等,并预报其发展趋势,最后针对诊断情况指导实施维护的技术,其目的是为了减少故障发生带来的损失。随着科技的发展,机械设备的复杂程度越来越高,并发故障已成为故障诊断的常见问题,而且并发故障具有的复杂性、层次性、相关性、不确定性等特点,给正确诊断并发故障带来了很大困难。本文即针对这一问题,将人工免疫与证据理论集成,构建了并发故障诊断模型,通过实验验证,该模型可以有效的实现对并发故障的诊断。受生物免疫系统的启发发展而来的人工免疫系统,为故障诊断带来了新的思路。其中由生物免疫系统“自己”-“非己”识别机理衍生得到的阴性选择算法,使得人工免疫系统应用于故障诊断中可以有效的识别机械设备的工作状态。结合无量纲指标不受载荷、工况、转速等工作条件影响的优势,构建多无量纲免疫检测器,利用这些无量纲免疫检测器可以使从传感器获取的机械设备的振动信号分析处理为各无量纲指标范围,作为各种故障的特征信息进行后续的分析判断。证据理论在量测和组合、不确定性的表示方面的优势使得许多学者将它引入并发故障诊断中来。用证据理论对人工免疫系统的无量纲免疫检测器得到的故障特征信息进行融合来实现对故障的最终判定,这就是本文提出的人工免疫与证据理论集成应用于旋转机械故障诊断的思路。本文在实验过程中应用了优化的无量纲指标,这些新的无量纲指标是利用遗传算法优化得到的,较好的解决了原有无量纲指标只对部分故障敏感的缺陷。并对证据理论进行了扩展,充分考虑了不同指标对不同故障的诊断能力、敏感程度不同的特性,采用加权证据理论的方式对故障特征信息进行融合,从而提高了诊断的可靠性和灵敏度。实验验证,人工免疫和证据理论集成的方法对于旋转机械并发故障的诊断切实可行。
[Abstract]:Rotating machinery is widely used in industry, which is mainly composed of a rotor bearing system supported rotor, stator or machine casing, couplings and other components, through rotation of rotating machinery to complete the work. As the key equipment in some industries, the production of the enterprise will have a direct impact on the health of the equipment operation, if a fault occurs stop, will bring huge economic losses and even disastrous consequences. The fault diagnosis technology, to understand the mechanical running condition, forecast the reliability, identify causes of mechanical failure of the parts, degree of risk, and forecast its development trend, according to the diagnosis of the situation to guide the implementation of maintenance technology, its purpose is to to reduce the loss caused by the fault. With the development of science and technology, mechanical equipment becomes more and more complex, concurrent fault has become a common problem of fault diagnosis, But with the complexity of concurrent fault hierarchy, correlation, uncertainty, brought great difficulties to the correct diagnosis of concurrent fault. This article is aimed at this problem, the integration of artificial immunity and evidence theory, constructs the concurrent fault diagnosis model, through experimental verification, the model can be the diagnosis of concurrent fault effectively.
Artificial immune system inspired by the biological immune system development, brings a new approach for fault diagnosis. The biological immune system "-" non self recognition mechanism derived from negative selection algorithm, the application of artificial immune system in fault diagnosis of mechanical equipment to identify effective working condition with dimensionless index is not affected by the load conditions, the working conditions of the influence of the speed advantage, build multi dimensionless immune detectors, using these dimensionless immune detectors can make the vibration signals of mechanical equipment obtained from the sensor analysis for the dimensionless index range, as the characteristic information of fault analysis and judgment in the future. Evidence theory test and combination in the amount of uncertainty is expressed in terms of advantage makes many scholars put it into the concurrent fault diagnosis. The use of evidence theory The fault feature information obtained from the non dimensional immune detector of immune system is fused to achieve the final judgement of faults. This is the idea of artificial immune and evidence theory integrated in rotating machinery fault diagnosis.
The dimensionless index in this paper are applied in the optimization, the new dimensionless index is obtained by using genetic algorithm optimization, to solve the defects of original dimensionless index only part of the fault sensitive. And the evidence theory is extended, considering the different indicators of the different fault diagnosis ability. The characteristics of different sensitive degree, using the weighted evidence theory to fault information fusion, so as to improve the reliability of diagnosis and sensitivity. Experimental verification, artificial immune system and evidence theory integrated method for rotary machinery fault diagnosis to concurrency is feasible.

【学位授予单位】：太原理工大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH165.3;TP18

【参考文献】