Research On Excavator Vibration And Sound Signal Processing Using Time-Frequency Analysis Method And Implementation

Excavator is the typical construction machinery. During the operation, the generated noise will cause environmental noise pollution. And the impact of high noise level can cause the physical and psychological harm to operator. Vibration of the engine is the the main reason of the noise. By collecting and analyzing the noise and vibration signals, it can provide an effective guidance to reduce the vibration and noise. However, the number of excitation sources can cause excessive noise and vibration is large, and the propagation mechanism is complicated with changeable path. Only having an in-depth understanding of the main excitation sources and their transfer rules, it can benefit into achieving the targeted structural improvements.Generally, engineering signals having non-stationary and nonlinear characteristics, time-frequency analysis technique is an effective tool for analyzing such signals. However, the traditional time-frequency analysis methods often suffer the poor time-frequency resolution, unexpected cross-terms etc., which make them difficult to achieve precise analysis of such signals. To solve these problems, in this paper, the research is made in progress in the following areas:1. We propose a novel method which can achieve an ideal time-frequency resolution, called as parametric synchroextracting transform. Firstly, based on the short-time Fourier transform, we propose a synchroextracting operator, which can extract the time-frequency coefficient in the instantaneous frequency region to generate a new time-frequency representation. Then, in order to solve the energy blur problem, we prpose a novel demodulated approach. By constructing the demodulated operator being consistent with the analyzed signal, it enhances the energy concentration greatly. Finally, combining the above two methods, it can achieve an ideal time-frequency representation for non-stationary and nonlinear signals.2. A-weighted sound pressure level noise spectrum is a commonly used sound research method. However in the analysis of time-varying signals, it cannot provide effective dynamic characteristic information, which is not suitble for analyzing the noise feature in non-stationary working conditions. Therefore, based on time-varying analysis and the traditional A-weighted sound pressure level, we propose a novel method, which can be utilized as the analysis tool of non-stationary sound characteristic and distribution.3. The sparse component analysis (SCA) method is applied for modal analysis. And the improvement based on original SCA is developed on matrix estimation and source separation. This study finds that the multi-channel vibration modal signals in time-frequency domain can produce linear shape clustering feature, and each clustering line is corresponding to the linear direction of each matrix column vector. According to the above features, by performing single-mode extraction and single-source detection, we can estimate the column vector respectively, which can achieve a much preciser estimation than conventional methods. Finally, the mono-mode signal can be separated out and the corresponding modal parameter can be identified by single-mode identification method.4. The proposed time-frequency methods are utilized for analyzing the vibration and sound of an excavator. According to the signal analysis of two common conditions, we can have the following conclusion:(1) the main vibration excitation sources are due to the low-level excitation of engine and the first-order pulse excitation of hydraulic, with obvious FM, AM and other non-stationary characteristics; (2) the main noise excitation sources are due to the second order inertia force excitation of engine, cooling fan motivation, incentives and muffler exhaust pump incentive, with the time-varying characteristics and the complated distribution positions around the excavator. The above analysis conclusions can provide the targeted guidance for reducing vibration and noise of the excavator.5. The proposed sparse component analysis method is applied to the modal analysis of excavator frame structure. By processing the 8-channel test signals, a total of 20 order single-mode signals are separated out. According to the identified modal parameters, they may provide more specific reference to the development and improvement of exvator structure.