【作者】 杜金榜；

【导师】 王跃科；

【作者基本信息】 国防科学技术大学 ， 仪器科学与技术， 2011， 博士

【摘要】 旋转机械是坦克等武器装备系统的关键部件之一,其振动状态监测对于提高武器装备的可靠性与安全性、促进视情维修技术的发展具有十分重要的意义。大型旋转机械振动监测向综合、高速、连续和网络化趋势发展。海量监测数据的实时传输与同步存储已成为亟待解决的成本与技术瓶颈问题,解决这些难题的一个有效途径是进行旋转机械振动数据压缩。本文以某型装甲车辆载荷谱测试和发动机全面鉴定试验为背景,立足于测试数据压缩这一仪器科学与技术的学科前沿和新的生长点,从标准化和多学科深层次交叉融合的角度出发,旨在将业已成熟的标准语音编码技术应用于旋转机械振动数据压缩领域。围绕基于语音编码的旋转机械振动数据压缩关键技术与系统实现展开研究。主要内容包括以下几个部分:1.根据旋转机械状态监测与振动数据压缩试验的需求,基于振动信号本质特征,建立了涵盖基频振动、谐波振动、拟周期振动、冲击振动、调制型振动乃至混沌振动等典型状态的广义归一化旋转机械振动信号模型。2.在继承和发展语音、图像及其它领域数据压缩评价方法的基础上,建立了一套较为完备的综合衡量旋转机械振动数据压缩与重构性能的评价方法。主要包括时频域主观视觉验证、压缩效率、算法复杂度、时延特性和时频域重构信号质量等主客观数字评价指标,以及基于主客观数字评价指标线性加权最优的多目标优化综合评价原则。3.针对语音编码与旋转机械振动信号处理的关键技术,从信源产生机理、经典数学模型、基本处理单元、信息冗余消除以及工程应用等不同层面进行广泛的类比分析,定性讨论了语音编码技术在旋转机械振动数据压缩中的适用性。基于广义归一化振动信号模型和综合评价方法,应用ADPCM波形编码、G.729A参数编码、MP3感知编码和EZW变换编码四种代表性语音编码算法进行振动数据压缩与重构仿真试验。从定性和定量的角度,仿真验证了经典语音编码算法在旋转机械振动数据压缩中应用的可行性和有效性。4.在试验对比分析的基础上,从压缩效率、重构信号质量、算法复杂度及适用性等方面,讨论了四种经典语音编码算法在旋转机械振动数据压缩中的综合性能及其优化思路。为进一步提高压缩效率,提出了基于数据稀化和语音编码的旋转机械振动数据联合压缩算法。5.从服务于旋转机械综合、高速、连续和网络化振动测试的角度,借鉴GPIB、VXI、PXI和LXI系统规范,基于多DSP并行处理、嵌入式PC104和MC3一体化等数字化测试技术,自主研制和设计了高性价比的Ⅰ型类GPIB、Ⅱ型和Ⅲ型类VXI/PXI以及基于LXI的IV型旋转机械综合振动测试与实时压缩处理系统,为基于语音编码的旋转机械振动数据压缩技术研究提供了实验平台。6.以多功能轴承-转子实验台为对象构建旋转机械振动测试与数据压缩实验系统。对论文提出的基于语音编码的旋转机械振动数据压缩算法进行了实验研究。进一步验证了基于语音编码的振动数据压缩方法的可行性和有效性。更多还原

【Abstract】 Rotary machines are key parts of weapon equipment and system such as tanks, and the vibration monitoring is much significant to improve the reliability and safety of weapon equipment and to advance the development of condition-based maintenance technology. The development trend of vibration monitoring for large-scale rotary machines is integrated, high-speed, continuous, networking and etc. The real-time transmission and synchronizing storage of jillion monitoring data have been the bottle-neck of cost and technology to be settled urgently. Vibration data compression is an efficient approach. On the material application background of certain-type armor-vehicle’s load-spectrum testing and all-around experimentation of engine, be established in test data compression of the subject foreland and growing point of instrument science and technology, in view of the standardization and deep amalgamation of multi-subject, the dissertation is focused on the key technologies and system implement, in order to apply the standard and ripe speech coding technology to rotary machines’vibration data compression. The contents mainly include:1. The generalized and normalized model of rotary machines’vibration signal has been established according to the requirement of status monitoring and vibration data compression experimentation, which contains various typical vibration status, such as basic frequency, harmonic, pseudo-periodic, under impact, modulated, chaos and so on.2. A relative mature comprehensive evaluation method of rotary machines’vibration data compression and reconstruction has been set up for the sake of weighing the performance, succeed and developed to evaluating methods of speech, image and other data compression. It includes subjective and objective digital value targets, such as vision verify mean opinion score and reconstructed quality in time-domain and frequency-domain, compression efficiency, complexity, time-delay performance and so on. Furthermore, the value principle of multi-target optimization is discussed based on the minimum energy of linear weighed sum of the digital value targets.3. Aimming at the key technologies of speech coding and rotary machines’vibration signal processing, the far-ranging analogy compare and analysis has been accomplished based on various lay of mechanism of signal source happening, typical math model, basic processed unit, principle of information redundancy eliminating, application of engineering practices and etc. The applicability of speech coding technology to rotary machines’vibration data compression has been qualitative analyzed. According to the generalized and normalized model and comprehensive evaluation method, the emulation experimentations of vibration data compression and reconstruction has been done with the four-type representative speech coding arithmetic including ADPCM of wave-shape coding, G.729A of parameter coding, MP3 of apperceive coding and EZW of transform coding. The applicability and validity of speech coding arithmetic is validated by the qualitative and quantitative analysis.4. The integrated performance and optimized consideration of the four-type classical speech coding arithmetic are discussed based on the contrast- experimentation from the aspects of compression efficiency, reconstructed quality, complexity and applicability, etc. In order to improve the compression efficiency, the combining vibration data compression arithmetic has been brought forward based on data dilution and speech coding.5. The integrate vibration testing and real-time compressing systems with high performance-to-price ratio have been independent developed or designed based on the digital testing technologies of multi-DSP parallel procession, embedded PC104 module, MC3 and etc., using for reference of the technology criterion of GPIB, VXI, PXI and LXI system. Which have type-I analogously GPIB, type-II and type-III analogously VXI or PXI, and type-VI of LXI, serving for rotary machines’vibration monitoring of comprehensive, high-speed, continuous and networking trend. The systems have provided experimentation platforms for rotary machines’s integrate vibration testing and data compression based on speech coding.6. The experiment system of rotary machines vibration testing and data compression has been designed based on the multi-function platform of axletree-to-rotor. The arithmetics of rotary machines’vibration data put forward in dissertation based on speech coding are studied by experimentation. And the applicability and validity of speech coding arithmetic to rotary machines’vibration data compression are verified much more.更多还原