【作者】 张邦基；

【导师】 于德介；

【作者基本信息】 湖南大学 ， 机械工程， 2010， 博士

【摘要】 日趋严重的能源紧缺与环境污染问题已经向传统内燃机汽车提出了挑战,发展新型节能环保车已是大势所趋。并联式混合动力汽车(Parallel Hybrid Electric Vehicle:PHEV)是实现新能源汽车产业化的重点车型；能量控制策略是影响PHEV动力性能、经济性能和排放性能的重大关键技术,变速箱关键零部件的振动特性和故障诊断是影响PHEV整车性能和舒适性的关键技术,这两方面的研究对发展PHEV具有十分重要的理论意义与应用价值。PHEV能量控制策略研究的主要内容包括：发动机和驱动电机性能优化与控制、电池组荷电状态估计以及各组成单元之间的能量分配等。现行的PHEV发动机性能研究主要以燃油经济性为优化与控制目标,采用的方法是根据发动机的万有特性,利用多项式拟合方法确定发动机的最佳经济运行线或最佳动力运行线,使发动机沿其中一条曲线工作；而对发动机的经济性和排放性两者同时进行优化与控制,由于发动机的排放性能复杂,相关的研究则较少。能量控制策略方面的研究,主要有以车速和转矩等为主要参数的控制策略、以燃油经济性和排放性以及电池组的荷电状态等为控制变量的能量控制策略,基于智能信息处理的能量优化控制策略则研究较少,或局限于理论研究,缺乏实际应用。基于电池组电压、电流和内阻等变量的电池组荷电状态估计,由于PHEV影响因素多、车辆工况复杂等原因,预测方法需要进一步改进且预测精度需要进一步提高。由于PHEV存在动力耦合和能量再生分配,虽然变速箱结构与传统内燃机汽车相同,但其工况更复杂,振动噪声特性差异很大。研究PHEV变速箱及其关键零部件的振动噪声特性,利用振动故障诊断技术分析和诊断变速箱的运行状况,对提高PHEV整车动力性能和改善舒适性具有十分重要的作用。国内外在机械故障诊断方面有大量研究见诸文献,但对PHEV振动特性分析和故障诊断则缺乏深入的研究。基于上述背景,本文以智能信息处理理论和方法为主线,主要完成了两个方面的研究工作：将神经网络、模糊控制应用于PHEV能量总成控制,将小波分析、粗集理论应用于PHEV变速箱关键零部件的振动故障诊断。论文主要创新点如下：1.在长丰猎豹混合动力轻型越野车(型号：CJY6470PHEV)发动机特性实验的基础上,建立了发动机万有特性的神经网络模型,并对神经网络模型的结构和参数进行了优化设计；将神经网络与网格插值方法融合,在发动机万有特性模型中进行数据挖掘,利用多项式和样条函数等数值计算方法获得了CJY6470PHEV发动机的最优运行特性及其与电子节气门之间的数值关系,通过调整节气门开启角改变发动机的功率和转矩,实现了对发动机特性的优化控制。2.基于系统层次的控制器结构设计与分析方法,设计了CJY6470PHEV能量总成控制系统,提出了节气门开启角及行驶挡位的优化控制方法,确定了CJY6470PHEV能量分配的全混合控制策略,建立了各组成单元的数学模型,分析了车辆行驶的功率、扭矩与汽油发动机、交流感应电机驱动系统的功率、扭矩以及镍氢动力电池组荷电状态之间的关系；基于PHEV中发动机驱动系统和电机驱动系统产生和消耗的能量应处于一种动态平衡状态的思想,提出了动力电池组荷电状态估计的能量平衡方法；实现了发动机在最优工况下的运行控制、能量在发动机和电机之间的合理分配以及对电池组荷电状态的控制。3.基于对模糊逻辑的基本理论以及PHEV中人机相互影响的特点和各驱动系统之间的能量分配策略的分析,设计了CJY6470PHEV模糊逻辑能量总成控制系统,建立了能量回馈制动模型和正常行驶模型,确定了T-S(Takagi-Sugeno:T-S)模糊控制中输入量和输出量之间的函数关系；简化了控制系统结构,实现了各组成单元特性的优化与能量的匹配。4.将基于粗集理论的振动故障诊断方法应用于PHEV变速箱滚动轴承的故障诊断,分析了属性约简引起的信息丢失问题和冗余信息的互补性,提出了基于信息互补的粗集诊断新方法,利用多个约简的互补性进行了PHEV变速箱滚动轴承的故障诊断,得到了与工程实际更加吻合的结果。5.分析了小波变换方法进行PHEV变速箱滚动轴承故障诊断时的局限性,也分析了常用特征参数对故障的不敏感性,提出了基于小波变换的能量特征提取和基于粗集理论的诊断规则获取新方法,在不能确定故障特征频率的情况下,该方法具有更高的诊断准确率。更多还原

【Abstract】 The challenge for the traditional gas-engine vehicle has been faced due to the shortage of the energy and the environment pollution. Thus, it is necessary to develop the new energy-saving and environment-friendly vehicles in future, and the Parallel Hybrid Electric Vehicle (PHEV) has become the preferred model to realize the industrialization of the new energy auto. Energy control strategy is the key technique to determine the power performance, economical performance and emission performance. And the vibration and fault diagnosis is the key to the drive comfort. The researches on these two aspects have a very important theoretical meaning and application value to the development of new energy vehicles.The main contents of PHEV energy control strategy studies include:the engine and drive motor performance optimization and control, the battery packs state of charge (SOC) estimation, as well as the energy distribution between the various constituent units. The present studies, aiming to the optimization of the engine performance, are to determine the optimum economical routine or optimum power performance routine using polynomial fit technique according to the universal characteristics of the engine. However, it is noted that only one routine can be determined for the optimization of the engine performance based on this method. Due to the complexity of emission performance, there are quite a few studies on the optimization of both fuel economy and the less emissions of engine. Furthermore, the present studies on PHEV energy control strategy mainly include:control through some parameters such as velocity and torque, control along with fuel economy or emissions, as well as control through some variables such as the battery packs SOC. However, there are quite a few studies on PHEV energy optimization and control based on intelligent information processing. Due to the limitation of theory, it lacks of practical application. The accuracy of estimation of SOC based on battery voltage, current and resistance needs to be further improved because of more factors and more complex working conditions of vehicles.Because of Parallel hybrid electric vehicle (PHEV) taking on dynamic coupling and regenerative braking energy distribution, although the transmission of PHEV is the same as the traditional internal combustion engine vehicles, there are more complex working conditions and more difference in vibration and noise characteristics. It is very important to study the vibration and noise characteristics of PHEV transmission and to monitor the transmission operating using vibration fault diagnosis technology so as to enhance vehicle dynamic performance and improve comfort. There have been a large number of researches on mechanical fault diagnosis at home and abroad, whereas it lacks of in-depth research on PHEV vibration analysis and fault diagnosis.As analysis previously, based on the intelligent information processing theory and methods, there are two aspects completed in this paper:neural network and fuzzy logic are applied to PHEV energy management, wavelet transform and rough set theory are applied to the vibration fault diagnosis of PHEV transmission and its key components. The main research contents and innovations are as follows:1. The researches on the engine numerical modeling method. Based on CJY6470PHEV engine experiments, the engine model of universal characteristics is established using the neural network method. And the structure and parameters of the neural network model are optimal designed. Integrating neural network and grid interpolation, the data mining is developed for the engine model of universal characteristic. And using numerical method such as polynomial and spline functions, the optimum operating characteristics and its numerical relation with the electronic throttle of CJY6470PHEV engine are obtained. It is concluded that the engine characteristics optimal control can be achieved by adjusting the throttle accordingly.2. The researches on the design method and control strategy of the PHEV energy management system. Through the design of CJY6470PHEV energy management system, the optimal relationship between the throttle angle and the vehicle shifts is obtained, and the full hybrid control strategy of the energy distribution of the CJY6470PHEV is determined, also the mathematical model of the various constituent units is established. According to the analysis of the relationship among the vehicle power and torque, the engine and the drive motor’s power and torque, and the battery packs SOC, a new method based on the dynamic energy equilibrium of electric motor drive system and engine drive system is originality proposed to estimate the battery packs SOC. The results show that the HEV energy management system has the capability of making the engine run at the optimal operating conditions, and guarantee the energy reasonable distribution of the electronic motor and engine according to the formulated strategy under the permission of battery SOC.3. Based on the basic theory of fuzzy logic and analysis of the human-machine interaction characteristics of HEV and its energy distribution strategy, the fuzzy logic energy management system of CJY6470PHEV is developed using the T-S (Takagi-Sugeno:T-S) fuzzy control model include the energy regenerative system and the energy management system in the process of normal driving. The T-S fuzzy model adapts the variable of the system state or the function of the input variables as a suffix of the if-then fuzzy rules, which can describe both the fuzzy controller and the dynamic model of controlled object, and to determine a linear function of the input and output, so as to simplify the model of control system. The results show that this system is appropriate to control the energy distribution between the engine and the motor, and enable motor working by a high efficiency in accordance with the manner determined.4. Vibration fault diagnosis method based on rough set theory is applied to PHEV transmission roller bearing. According to the analysis of the information loss caused by attribute reduction and the complementary redundant information, a new diagnosis method of complementary information and rough set theory-based is proposed. In this method, malfunction diagnosis is based on the multiple complementary reductions. So that in spite of the lack of some collecting information, malfunction diagnosis can be completed using other information. Consequently, this method is more suitable for engineering practice.5. Based on the basic theory of wavelet transform, the limitation of using wavelet transform to the roller bearing fault diagnosis is pointed out, namely it needs to calculate fault characteristic frequency accurately. And also the insensitivity of fault diagnosing using the common parameters is analyzed. A new diagnosis method is proposed for the energy feature extraction based on wavelet transform and for the rule acquisition based on set theory. The results showed that the fault diagnosis, based on wavelet transform and rough set theory, can be more appropriate in the engineering practice and have a higher accuracy.更多还原