论文部分内容阅读
针对风力机振动信号采集过程中易受噪声影响的问题,提出基于过完备原子库的匹配追踪算法对风机振动信号进行处理。该算法能自适应提取和原子相关的信号结构,从而可实现噪声抑制。在匹配追踪算法处理过程中,利用结合梯度信息的改进的粒子群优化算法来寻找最佳原子。仿真结果表明,该算法比标准匹配追踪算法具有更快的运算效率及更高的重构精度。利用该算法对风力发电机齿轮箱振动信号进行去噪处理实验。实验结果表明,去噪后信号信噪比可提高5 d B以上,波形特征更加清晰,并且可以在降噪的同时有效保留故障信息。
Aiming at the problem that the vibration signals of wind turbines are easily influenced by noise during the vibration signal acquisition, a matching pursuit algorithm based on an overcomplete atomic library is proposed to deal with the vibration signals of wind turbines. This algorithm can adaptively extract the signal structure related to the atom, so as to achieve noise suppression. In the process of matching pursuit algorithm, we use the improved particle swarm optimization algorithm combined with gradient information to find the best atom. Simulation results show that this algorithm has faster computing efficiency and higher reconstruction accuracy than the standard matching pursuit algorithm. The algorithm is used to denoise the gearbox vibration signal of wind turbine. The experimental results show that the signal-noise ratio of the denoised signal can be increased by more than 5 d B, the waveform characteristics are clearer, and the fault information can be effectively preserved while reducing noise.