【作者】 赵凤强；

【导师】 马孝江；

【作者基本信息】 大连理工大学 ， 机械电子工程， 2006， 博士

【摘要】 随着全球城市化进程的加快,城市规模的急剧膨胀和城市建设的不断发展,城市环境噪声问题也日益突出,给人们的生产和生活带来了日趋严重的影响。该问题已逐步成为人们广泛关注的一个热点问题。本文以大连市主城区环境区域噪声问题为工程背景,研究基于计算智能、系统理论的综合评价方法和声振检测与故障诊断技术及其在城市环境噪声相关问题中的应用。城市环境噪声问题内涵丰富,涉及面广,本文重点在噪声测量布点优化、噪声质量综合评价及交通噪声控制等方面开展研究工作。该课题在理论上属机械、环境科学、声学、计算机科学、系统和控制工程等交叉学科前沿课题的基础理论和应用基础研究;在实践上,具有相当的工程应用前景,可推广应用于水、大气污染等其它环境领域问题的监测和评价,及各种车辆变速箱和机械传动系统的出厂噪声质量检测。研究的计算智能类算法更可推广于求解多种复杂工程系统的优化问题。因而,无论从理论上还是实践上,该问题都是一个亟待较好解决的重要问题。 城市环境区域噪声的测量布点优化问题在理论上可归结为一个离散型0-1规划问题,属组合优化范畴,是NPC问题,难以很好地求解;其噪声质量评价则属于系统工程的综合评价问题,具有多指标、多层次、多关联等特点,只依靠评审专家对其进行人为的定性评价已难以满足对于复杂工程系统评价的科学性和客观性要求;而汽车变速箱出厂噪声质量的检测和故障诊断属于城市交通噪声控制的一个重要问题,为避免车间背景环境噪声影响,目前常规检测变速箱噪声超标与否的方法是采用声级计在消声室中测得声压级的方式,费时费力,且无法在线测量。由此可见,解决这些问题存在较大的难度和复杂性。结合国情,本文针对上述问题主要进行了以下的研究工作。 (1)给出课题背景、研究意义,综述了城市环境区域噪声领域相关问题的解决方法及其国内外的代表性工作,并在此基础上归纳总结,进而明确了本文的相关研究思想及求解策略。 (2)提出了并行混合粒子群免疫算法(PHPSO-IA)。该算法以粗粒度并行遗传算法(PGA)为基础,针对其早熟和收敛速度慢等缺陷,给出一系列相应的改进措施,主要有:a.依据改进的自适应交叉、变异算子对子群体分类,实行多种群的并行演化;b.根据不同类型子群体的特点,引入不同模式的基于粒子群优化的个体PSO更新算子,其主要目的是充分利用粒子群算法收敛快的特点;c.加入免疫的思想可起到双重效果,一方面基于提出的可调等比排序的免疫选择算子能有效防止早熟,另一方面模拟免疫记忆和接种过程的子群体信息交换策略可加快收敛速度;d.对于优秀个体的混沌序列搜索能显著改善混合算法的局部搜索性能。PHPSO-IA有连续型和0-1离散型两种版本,可同时用于解决上述两类问题,是一种通用的混合计算智能类算法。通过若干典型函数优化和背包问题算例验证了该算法的可行性和有效性。鉴于传统的城市噪声等网格测更多还原

【Abstract】 With the acceleration of worldwide urbanization course, the rapid expansion of city scales and the continuous development of city construction, problems of urban environmental noise are becoming more and more serious. They have brought increasingly severe influence on people’s working and living. And these problems have gradually become hot spots that arouse extensive attention. Taking urban environmental noise in Dalian as engineering background, this dissertation studies computational intelligence, comprehensive evaluation and technique of acoustic vibration measurement & default diagnosis as well as their application to the relevant problems of urban environmental noise. The problems of urban environmental noise involve abundant contents and relate to many factors. This dissertation puts study emphases on optimization of the noise measurement points, comprehensive evaluation of noise quality and traffic noise control. Theoretically, this subject concerns to mechanical engineering, environmental science, acoustics, computer science, system & control engineering. And it can be ascribed as the frontier fundamental and applied research of cross-discipline subject. In practice, this subject has extensive engineering application prospects, such as monitoring and evaluation of water and air pollution in the field of environmental protection as well as the delivery test of noise quality for various vehicle gearboxes and mechanical drive systems. Meanwhile, the proposed computational intelligence algorithm can also be applied to many optimization problems of complex engineering systems. Altogether, this subject is of great importance and urgency in terms of theory and practice. But it is still far from being solved satisfactorily.The optimization problem of measurement points of urban environmental noise can be boiled down to a discrete 0-1 programming problem. Since it belongs to NPC problems, it’s hard to solve this problem very well; Urban noise quality evaluation belongs to comprehensive evaluation problems in system engineering, which possesses the characteristic of multiple indices, multi-layers, strong correlation, etc. Only by qualitative evaluation of relevant experts, it is rather difficult to meet the requirements of scientificity and objectivity for complex engineering systems; Delivery test of noise quality and relevant fault diagnosis for automobile gearboxes is one of the important aspects of urban traffic noise control. To avoid the influence of workshop background noise, the common method of testing whether the noise generated by gearboxes exceeds the delivery test standard or not is to measure the noise sound pressure level by sound level meter in anechoic chambers. But this method is rather time-consuming and laborious as well as it’s difficult to realize the measurement directly on production line. As above stated, it’s clear that there exist considerable difficulty and complexity to solvethese problems. According to the state of our nation, this dissertation carries out the following research works for them.1. Research engineering background and significance are explained. Moreover, the corresponding solutions and typical research works at home and abroad for relevant problems in the field of urban environmental noise are surveyed. On the basis of them, the research methods and solution strategies of this dissertation are made clear further.2. A parallel hybrid PSO immune algorithm (PHPSO-IA) is proposed based on parallel genetic algorithm (PGA) in this dissertation. To overcome the two main defects, i.e. premature convergence and slow convergence rate, of the traditional parallel genetic algorithm, PHPSO-IA makes a series of improvements on PGA as follows, (a) Subpopulations are classified as several types according to modified crossover and mutation operators and all types of subpopulations evolve parallelly.(b) In accordance with the characteristics of different types of subpopulations, different types of individual PSO update operators are introduced. And it aims at making full use of the fast convergence property of particle swarm optimization.(c) Introducing immunity theory into PHPSO-IA has double functions. One is that proposed immune selector based on adjustable geometric-progression rank-based selection can prevent the algorithm from premature effectively. The other is that convergence rate can be accelerated by individual migration strategy between subpopulations based on imitation of immune memory mechanism and inoculation procedure, (d) The chaos sequence search for superior individuals can improve local searching performance of the hybrid algorithm considerably. PHPSO-IA is a universal hybrid computational intelligence algorithm. It contains continuous and 0-1 discrete versions in this dissertation and can be adopted to solve these two kinds of problems. Examples of typical function optimization and knapsack problems show that PHPSO-IA is feasible and effective. In view of lack of consideration for non-uniform distribution of real acoustic field when locating measurement points by traditional noise measuring method with equal grids, the idea of locating points by the method with unequal grids are given based on the strategy of element partition in finite element analysis (FEA). Moreover, on the basis of it, PHPSO-IA is adopted to solve the optimization problems of measurement points in the field of urban environmental noise. The results are satisfactory. It can greatly reduce the consumption of manpower and material resources during the course of measuring noise in future and open a gate for realizing automatic monitoring and evaluation of urban environmental noise.3. Multilevel modified set pair analysis method based on C/S & B/S mixed mode is presented. And it is applied to comprehensive evaluation for urban environmental noise quality. First of all, to overcome the defect that it’s not convenient to decide the concrete pollution class when traditional set pair analysis(SPA) is applied to the field of environmental quality evaluation, the method of modified set pair analysis (MSPA) as well as its evaluation criteria based on integrated connection degree are proposed. Taking the evaluation of urban environmental noise quality in Dalian as example, it verifies that the multilevel MSPA is reasonable, effective and practical by comparing it with multilevel fuzzy comprehensive evaluation, whose weight vectors are both given by AHP. Evaluation results by above two methods are consistent and coincide with objective reality. Moreover, multilevel MSPA based on C/S & B/S mixed mode is presented further. It possesses the characteristics of flexibility, convenience and maintainability. As well as, it can realize the resource and information sharing for relevant personnel at different geographic positions. The above network-based multilevel MSPA is beneficial to making an evaluation more objective and accurate as well as raising relevant working efficiency. To contribute to actual engineering applications, making the relevant algorithms and methods proposed in this dissertation as well as SoundPLAN and Maplnfo integrated, the overall design planning and modular framework structure of the evaluation software system for urban environmental noise are presented. And the corresponding software system is programmed. All of this work benefits promoting research progress of relevant theory and methods in the field of comprehensive evaluation of urban environmental noise.4. Traffic noise is the foremost source of urban environmental noise. And the vehicle noise is the key factor of it. From the point of view of controlling the noise quality of automobile gearboxes before leaving factories, a new on-line noise vibration measurement method and the fault diagnosis method for automobile gearboxes exceeding noise standard are proposed. The former can realize the delivery test of gearboxes’ noise on production line. The derived relevant formula and the concrete vibration measurement procedure are given, which can effectively prevent from being affected by the background noise and obtain the sound pressure level of gearboxes’ noise on production line by testing the vibration acceleration signals of boxes. The latter is a fault diagnosis method based on local wave theory and octave analysis. The local wave method can decompose complex signals into finite local wave components and make them separate in time-frequency coordinate system. It also can avoid the mutual influence of sidebands of different frequencies. The octave analysis can further calculate the corresponding frequency-band sound pressure levels of components and find out the frequency within a frequency band that radiates maximum noise. Thus, the fault source can be final decided. The abovementioned two methods are combined into a complete flowchart. It makes the delivery test of gearbox noise more convenient and effective. This work is beneficial to updating gearbox delivery quality and improving the performance of whole vehicle. Therefore, it contributes tocontrolling traffic noise and even total urban environmental noise effectively.This subject is a collaborative research project of Institute of Vibration Engineering of Dalian University of Technology and Dalian Environment Monitoring Center. And it is financially supported by the National Natural Science Foundation of China (No. 50475155). Altogether, it not only benefits the development of science and technology in related fields theoretically, but also has great economic returns and social benefits. And it can be spread through and applied to various relevant problems stated-above.更多还原