当前位置:首页 >> 机械/仪表 >>

混合智能技术及其在故障诊断中的应用研究中英文摘要


论文中英文摘要
作者姓名:雷亚国 论文题目:混合智能技术及其在故障诊断中的应用研究









机械、运载、能源、冶金、石化等国民经济和国防建设中的大型复杂机械设备,长期在 重载、疲劳、腐蚀、高温等复杂恶劣的工况下运行,设备中的关键零部件会不可避免地发生 不

同模式和不同程度的故障。例如,汽轮发电机组、航空发动机、燃汽轮机等大型旋转机械 出现动静碰摩、密封摩擦、轴瓦破碎、转子裂纹等故障现象;高速机车、连铸连轧机组、数 控机床等装备中的主轴、齿轮、轴承和丝杠经常出现磨损、腐蚀、剥落、胶合、擦伤、裂纹、 断齿等故障现象;大型内燃机、往复式压缩机等装备在变工况作用下,曲轴、轴承、连杆、 活塞等产生磨损、疲劳断裂以及穴蚀、烧蚀等故障现象。对这些大型复杂设备进行状态监测 与故障诊断密切关系到生产系统的正常运行、生产效率的提高、产品质量的保证、生态环境 的保护以及维修管理的科学化与现代化等一系列重要问题,因此倍受世界各国的广泛关注。 由于大型复杂机械设备的故障具有复杂性、不确定性、响应微弱性、多故障耦合性、相关性 等特点,因此,其状态监测与故障诊断要比常规设备的监测诊断要求高、难度大,对于其早 期故障和复合故障的诊断更是难上加难。 传统的故障诊断大多是由诊断专家手工进行, 因此使用者的经验和专业知识就尤为重要; 同时对于结构复杂、监测单元多、自动化程度高的机械设备,需要分析的数据量十分巨大。 如果这些大量的数据全部依靠诊断专家手工来分析显然是不可行的,因此必须提高设备故障 诊断的自动化、智能化程度。近几年,国内外学者将专家系统、模糊逻辑、神经网络、遗传 算法、聚类分析、支持向量机、粒子群算法、粗糙集理论等人工智能技术应用于设备的故障 诊断中,从而进行设备的智能故障诊断,在工程实际中取得了显著成效。随着研究和应用的 深入,发现这些技术各具特色、各有不足,在一定的条件和场合下有效。面临的诸如诊断信 息不完整、模糊隶属函数的人为确定、专家系统的知识获取困难、神经网络缺乏故障训练样 本等问题,限制了这些单一智能技术的应用。尤其对于大型复杂机械设备早期故障和复合故 障,单一智能技术存在诊断精度低、泛化能力弱和通用性不强等问题,严重制约了大型复杂 设备早期故障和复合故障的有效检测与诊断。因此急需新的技术和方法来解决这一工程实际 中面临的棘手问题。 借鉴“优势互补”和“分而治之”思想、利用人工智能技术与算法、结合机械故障诊断 理论的混合智能故障诊断技术便应运而生。混合智能故障诊断技术辉煌的发展远景是可预见 的,但目前纸面上的研究十分活跃,这种纸上谈兵的讨论,并没有在实际应用中充分地发挥 作用。迄今为止,实用化的混合智能诊断系统为数不多。无容置疑,研究混合智能故障诊断 技术以解决大型复杂设备早期故障和复合故障的诊断问题是十分困难而又十分有意义的课 题。论文正是针对这一极具挑战而又诱人的课题,对混合智能故障诊断技术进行了深入的研 究,旨在提出新的混合智能诊断算法与模型,并应用于工程实际,有效地提高高速机车、汽 轮发电机组、往复式压缩机等大型复杂机械设备的服役性能和预防重大事故,取得显著经济 效益和社会效益。因此,研究混合智能故障诊断技术具有重要的科学理论意义和工程应用价 值。 论文首先构造了混合智能故障诊断框架,给予混合智能故障诊断新的涵义:利用专家系 统、模糊逻辑、神经网络、遗传算法、支持向量机等单一智能技术之间的差异性和互补性, 扬长避短、优势互补,并结合现代信号处理技术和特征提取方法,将它们以某种方式综合、 集成或融合,以弥补单一智能故障诊断技术存在的缺陷,有效提高大型复杂机械设备的早期 故障和复合故障诊断的准确性和鲁棒性,确定故障发生的位置和模式,并估计其严重程度。

论文阐述了模糊逻辑、神经网络、聚类算法和遗传算法等单一智能技术的基本概念和原 理,针对每种技术举例说明,论证其优势并指出其不足。论证了两种适合于处理非平稳、非 线性信号的现代信号处理技术:小波包分析和经验模式分解。小波包分析是小波变换的延伸, 它以不同的尺度将动态信号正交地分解到相互独立的频带中,提供无冗余、不疏漏的独立频 带分解信号的特征信息;经验模式分解方法基于信号的局部特征时间尺度,把动态信号分解 为若干个本征模式分量,正交地给出分解信号的本征信息。所以二者分别从不同角度来分析 信号,各具特色,为混合智能故障诊断提取全面的故障特征信息。 为了提高大型复杂机械设备早期故障诊断的准确性,提出了一种基于现代信号处理、特 征评估和神经网络的混合智能故障诊断模型。该模型用小波包和经验模式分解对振动信号进 行分解,分别提取原始信号和各分解信号的统计特征组成联合特征,然后对联合特征进行评 估,并计算评估因子。根据评估因子的大小选取敏感特征作为径向基函数神经网络的输入, 从而实现设备不同故障的自动识别。通过对滚动轴承早期损伤和烟气轮机转子微弱碰摩故障 的诊断研究,应用结果表明:利用小波包分析和经验模式分解方法能从动态信号中获得丰富 的故障特征信息;利用特征评估方法能够从原始特征集中精确地评选出敏感特征,从而明显 地提高了机械设备早期故障诊断的准确率。 针对大型复杂机械设备中早期故障和复合故障的诊断难题,利用统计分析、经验模式分 解、补偿距离评估技术、自适应神经模糊推理系统和遗传算法等技术,提出了综合多征兆域 特征集和多分类器集成的混合智能诊断模型。该模型运用多种信号预处理方法挖掘潜藏在动 态信号中的故障信息,并综合利用从不同层面表征机械设备运行状态的统计特征,构成多征 兆域特征集来全面揭示故障特性;利用基于自适应神经模糊推理系统的多分类器既相互独立 又互为补充的优势来弥补单一分类器诊断性能的不足。综合运用多元征兆域特征集的多态混 合智能模型能够将不同工况、不同损伤程度的实验台滚动轴承故障进行准确的识别,而且有 效地诊断了高速电力机车滚动轴承早期故障和复合故障。 针对故障诊断中应用最多的无监督聚类算法—模糊 C 均值聚类存在的 3 个问题,提出了 基于模糊聚类和神经网络的混合智能聚类新算法。 该算法设计神经网络自适应学习特征权值; 运用点密度函数计算样本权值;使用聚类评价指标自动确定聚类数。赋予特征和样本以权重, 强调敏感特征和典型样本对聚类的贡献,削弱无关特征和模棱两可样本对聚类的干扰,以提 高聚类的性能。 采用国际比较聚类算法性能的典型数据 IRIS 验证了混合智能聚类算法的有效 性。在高速电力机车滚动轴承微弱故障和复合故障耦合的诊断问题中,混合智能聚类算法取 得了满意的诊断结果,进而验证了该算法的实用性和推广性能。 论文最后设计和开发了两种不同模式的远程监测与混合智能故障诊断系统:C/S 模式下 的“潜艇模型振动监测与分析系统”和 B/S 模式下的“皮带输送机轴承状态监测与故障诊断 系统” ;阐述了两个系统的总体框架和诊断功能;提出了基于潜艇壳体时、频域特征、专家系 统、K 近邻分类以及集成算法等的混合智能振动源辨识方法,实现了引起潜艇壳体声功率增 大的振动源的准确定位;提出了皮带输送机轴承状态监测与故障诊断的混合智能方法,该方 法综合运用了滚筒轴承在运行过程中的多物理场信息、集成多规则推理机进行故障诊断,已 成功地应用于工程实际输送机状态监测与故障诊断。

关键词: 能诊断

小波包分析

经验模式分解

人工神经网络

模糊聚类

混合智

Research on Hybrid Intelligent Technique and Its Applications in Fault Diagnosis
Lei Yaguo

ABSTRACT

Large-scale and complex mechanical equipments widely used in national economics and national defense constructions such as machinery, vehicles, energy sources, metallurgy, petro-chemistry, etc., usually operate under complicated and terrible conditions such as heavy duty, fatigue, erosion, high temperature etc. Therefore, it is inevitable for the key components of these equipments to suffer faults with various modes and different severity degrees. For example, rotor/stator rub-impact, seal friction, bearing bush fragmentation, rotor crack, etc. occur in large-scale rotating machinery including turbine generators, aeroengines, gas turbines, etc. Wear, corrosion, flaking, scuffing, rub, crack, broken teeth, etc. frequently happen in main spindles, gears, bearings and lead screws of equipments like high-speed locomotives, continuous casting and rolling lines, NC machine tools, etc. Fatigue fracture, cavitation, ablation, etc. appear in crank shafts, bearings, connecting rods, pistons etc. of machinery operating under variable conditions such as large-scale internal combustion engines, reciprocating compressors, etc. The condition monitoring and fault diagnosis for these large-scale and complex mechanical equipments is closely related to a series of issues, such as the normal operation of production systems, the improvement of production efficiency, the assurance of production quality, the protection of entironment, and the scientific modernization of maintenance and management, etc.. Thus, it has received intensive attention all over the world. However, faults of large-scale and complex mechanical equipments are characterized by complexity, uncertainty, response weakness, multi-fault coupling, correlation, etc., and therefore the condition monitoring and fault diagnosis for these equipments is more difficult than that of conventional equipments. Especially, it is much harder to detect and diagnose incipient faults and multiple faults occurring in the large-scale and complex mechanical equipments. Traditional fault diagnosis techniques are performed by diagnosticians manually and therefore expertise and special knowledge are extremely important to apply them successfully. Moreover, for mechanical equipments having complex structures, many monitoring cells and high degrees of automation, there is lots of data to be analyzed in fault diagnosis. Obviously, it is impossible for diagnosticians to manually process so many data. Thus, the degree of automation and intelligence of mechanical fault diagnosis should be enhanced. Recently, researchers have applied artificial

intelligent techniques to fault diagnosis of mechanical equipments, such as expert systems, fuzzy logic, neural networks, genetic algorithms, clustering analysis, support vector machines, particle swarm algorithm, rough set theory, etc. Correspondingly, prominent achievements have been obtained in the field of intelligent fault diagnosis. With the advancement of studies and applications, however, researchers find that individual intelligent techniques have their advantages and shortcomings as well, and an individual technique just performs well for specific cases. Such as the incompleteness of diagnosis information, the subjective determination of fuzzy memberships, the knowledge acquirement of expert systems, the lack of training samples of neural networks, etc. have greatly limited the further applications of individual intelligent techniques. Especially, for incipient faults and multiple faults of large-scale and complex mechanical equipments, the diagnosis accuracy using an individual intelligent technique is quite low and generalization ability is very weak. Thus, it is urgent and necessary to develop novel techniques and methods to solve these hard practical engineering problems. Hybrid intelligent fault diagnosis technique is proposed by using the idea of “complementary advantages” and “divide-and-conquer”, and combining artificial intelligent techniques and algorithms with mechanical fault diagnosis theory. The application prospects of hybrid intelligent fault diagnosis technique are quite wide. However, lots of investigations have been carried out just on the paper and the studies based on an armchair strategist are not effective in solving practical engineering problems. So far, practical hybrid intelligent diagnosis systems are quite few. Undoubtedly, it is extremely difficult but highly significant to apply hybrid intelligent fault diagnosis technique to diagnosing incipient faults and multiple faults of large-scale and complex mechanical equipments. This thesis just approaches this challenging and attractive subject, and intensively explores the hybrid intelligent fault diagnosis technique. The study objective of the thesis is to present novel hybrid intelligent diagnosis algorithms and models to solve the practical engineering problems, to effectively improve the service performance of large-scale and complex mechanical equipments and to obtain evident economic and social benefits. Thus, research on hybrid intelligent fault diagnosis technique has important scientific and theoretical significance and engineering values. The thesis constructs the framework of hybrid intelligent fault diagnosis technique and gives it new meaning first. According to the diversity and the complementarity among individual intelligent techniques, i.e. expert systems, fuzzy logic, artificial neural networks, genetic algorithms, support vector machines etc., we utilize their own merits and overcome their own shortcomings, and reinforce their advantages. By synthesizing, integrating or fusing these individual intelligent techniques, and different modern signal processing techniques and feature extraction methods, we can overcome the shortcomings of individual intelligent fault diagnosis techniques. Therefore,

hybrid intelligent fault diagnosis technique can effectively improve the diagnosis accuracy and robustness of incipient faults and multiple faults occurring in large-scale and complex mechanical equipments, ascertain the fault locations, and evaluate the fault severities. The thesis introduces basic conceptions and principles of the individual intelligent techniques, i.e. fuzzy logic, neural networks, clustering algorithms, genetic algorithms, etc., and provides an illustration for each technique to show its advantages and disadvantages. Two modern signal processing techniques suitable to nonstationary and nonlinear signals, wavelet packet analysis (WPA) and empirical mode decomposition (EMD), have been advanced. WPA is an extended result of wavelet transform (WT). It orthogonally decomposes a dynamic signal into several independent frequency bands that link up mutually without redundant or omitted information. EMD, which is based on the local characteristic time scales of a signal, adaptively decomposes the dynamic signal into a series of intrinsic mode functions (IMFs) and orthogonally presents intrinsic information of the signal. Thus, WPA and EMD have their own characteristics and process the dynamic signal from different aspects, respectively. They can extract rich fault characteristic information for hybrid intelligent fault diagnosis. In order to enhance the diagnosis accuracy of incipient faults occurring in large-scale and complex mechanical equipments, a hybrid intelligent fault diagnosis model is constructed based on modern signal processing techniques, feature evaluation and neural networks. In this model, WPT and EMD are respectively used to preprocess vibration signals to mine fault characteristic information accurately. Then, statistical parameters are extracted from each of the original vibration signals and preprocessed signals to form a combined feature set. Moreover, the distance evaluation technique is utilized to calculate evaluation factors of the combined feature set. Finally, according to the evaluation factors, the corresponding sensitive features are selected and input into the network to automatically identify different machine operating conditions. This model is applied to identifying the incipient defect of rolling element bearings and the slight rub of a gas turbine rotor. The results demonstrate that rich fault characteristic information can be precisely extracted by WPA and EMD, the sensitive features can be easily selected from a large number of features with the feature evaluation method, and therefore the diagnosis accuracy of the incipient faults has been evidently improved. Aiming at the difficulties in diagnosing incipient faults and compound faults of large-scale and complex equipments, the thesis proposes a novel hybrid intelligent diagnosis model based on feature sets from multiple symptom domains and multiple classifier combination. The model combines statistics analysis, EMD, the compensation distance evaluation technique, adaptive neuro-fuzzy inference systems (ANFISs) and genetic algorithms. It employs several signal preprocessing methods to extract the fault information embedded in dynamic signals. Time- and

frequency-domain statistical features that reflect the equipment operating conditions from various aspects are synthesised to construct multiple feature sets, which are able to fully reveal fault characteristics. Based on the independency and the complementarity of multiple ANFISs with the different input feature sets, we integrate them and develop the hybrid intelligent diagnosis model to overcome the shortcomings of the diagnosis performance of individual intelligent classifiers. The practical application results show the proposed hybrid model is able to reliably recognize not only incipient faults but also compound faults in locomotive rolling element bearings. Aiming at the three disadvantages of fuzzy C-means (FCM), the most popular unsupervised clustering algorithm used in the fault diagnosis field, a hybrid intelligent clustering algorithm is developed. In this algorithm, a new neural network is designed to adaptively learn feature weights; the density function of data point is used to compute sample weights; the cluster validity index is adopted to automatically determine the cluster number. The feature weights and the sample weights are assigned to the corresponding features and samples to emphasize the leading effect of sensitive features and typical samples, and weaken the interference of unrelated features and vague samples to improve the clustering performance. The test result of the benchmark data IRIS demonstrates the validity of the proposed algorithm. The algorithm is also employed to the incipient fault and compound fault diagnosis of locomotive rolling element bearings. The results show that the hybrid intelligent clustering algorithm enables to automatically and accurately identify the faults and have a strong practicability and generalization. Two modes of remote condition monitoring and hybrid intelligent fault diagnosis systems are designed and developed at the end of the thesis. One is based on C/S mode entitled “monitoring and analysis system of vibration for the submarine model”. The other is based on B/S mode entitled “bearing condition monitoring and fault diagnosis system of strap transportation machines”. The frameworks and diagnosis functions of the two systems are introduced. In system “monitoring and analysis system of vibration for the submarine model”, a hybrid intelligent vibration source identification method for the submarine model is presented by combining time- and frequency-domain features, expert systems and K nearest neighbor algorithms. The vibration source causing the increase of the submarine sound power can be exactly located by the hybrid intelligent identification method. In system “bearing condition monitoring and fault diagnosis system of strap transportation machines”, a hybrid intelligent diagnosis method by utilizing comprehensive information from multiple physical fields and integrating multiple rule-based inference engines is developed and successfully applied to condition monitoring and fault diagnosis of real-world transportation machines.

Key words:

Wavelet packet analysis; Empirical mode decomposition; Artificial

neural network; Fuzzy clustering; Hybrid intelligent diagnosis


相关文章:
基于多种方法综合的智能电网故障诊断研究
智能电网故障诊断研究 摘要 随着信息技术的发展、电力系统运行的日益复杂以及...由2种以上智能方法构成的混合系统在输配电网络故障诊断中的应用越来越 广泛, 将...
浅析人工智能技术在设备故障诊断中的应用
浅析人工智能技术在设备故障诊断中的应用 浅析人工智能技术在设备故障诊断中的应用...随着人工智能技术的 发展以及相关领域学科研究的深入,设备故障诊断也朝着智能化...
人工智能在故障诊断领域的应用
人工智能在故障诊断领域的应用摘要:本文在简要介绍故障诊断和人工智能的基础上,详细分析人工智能在故障诊断领域中的应 用, 包括智能故障诊断技术和故障诊断专家系统, ...
人工智能技术在电力系统故障诊断中应用
人工智能技术在电力系统故障诊断中应用_能源/化工_工程科技_专业资料。人工智能技术在电力系统故障诊断中应用摘要】 及时发现故障对于提高电力系统稳定性有着至关重...
输电网络故障诊断中人工智能技术的应用
输电网络故障诊断中人工智能技术的应用摘要】人们的生活离不开电能的正常供给,但是输电网络发生故障又是 不可避免的,为了快速的对故障进行诊断,这就需要应用...
哈工大 智能故障诊断技术在轴承中的应用
创新研修课论文 智能故障诊断技术 在轴承中的应用 班号: 学号: 姓名: 日期: 2012.01.04 摘要: 滚动轴承作为机械设备中重要的旋转零件和机械设备的重要故障源之...
人工智能在电气工程自动化中的应用——论文
人工智能电气工程自动化中的应用摘要:随着社会不断...人工智能 (Artificial Intelligence) , 英文缩写为 ...在电气设备故障诊断中, 人工智能技术中的模糊理论、...
设备故障诊断原理技术及应用
设备故障诊断原理技术及应用_机械/仪表_工程科技_专业...2. 开展故障诊断技术研究的意义应用故障诊断技术对...机械故障诊断技术的发展趋势 (1)混合智能故障诊断...
电气自动化技术中人工智能的应用研究
电气自动化技术中人工智能的应用研究摘要:随着科技的...智能理论分析人工智能(Artificial Intelligence),英文...(4)电气故障诊断中人工智能的应用。在电气设备的...
波形分析在发动机故障诊断中的应用研究
波形分析在发动机故障诊断中的应用研究_能源/化工_工程...当前,由于电子技术智能传感 2 器技术、集成电路...英文、有实践经验的机电一体化人才加入进来,充实汽车...
更多相关标签:
中英文摘要 | 论文中英文摘要格式 | 论文中英文摘要翻译 | 中英文摘要格式 | 论文中英文摘要 | latex中英文摘要 | 毕业论文中英文摘要 | 论文中英文摘要的格式 |