基于二阶线性系统与非线性系统的微弱信号检测研究

发布时间：2019-03-27 16:10

【摘要】：在现代化生产中，重大关键设备的实时监控和故障诊断具有重要意义。然而，在很多设备的早期故障中，故障信号往往被噪声所淹没而难以被检测出来，所以微弱信号的检测显得尤为重要，受到越来越多的重视。首先课题以二阶线性系统为信号检测处理模型，提出基于二阶线性动力系统参数调节共振的微弱信号处理方法。其基本原理是：把包含有特征频率成分和噪声干扰的待检测信号当作系统的激励，而把线性动力系统看作信号处理模型。通过人为调节系统的固有频率和阻尼比等参数，，使固有频率与待检测信号中的特征频率相等，于是系统响应达到共振，进而使特征信号进一步突出而克服噪声，达到检测特征信号成分的目的。该方法的具体实施是，在得到系统响应最大值随固有频率变化的特性曲线后，根据曲线中的极大值即可识别噪声中的特征信号。这种基于线性系统模型的信号检测，可通过选取合适的系统阻尼比参数来优化检测结果。其次课题以典型的Duffing系统为对象，研究了基于二阶非线性系统随机共振的信息检测方法。该方法通过非线性双稳系统、待测微弱信号和信号中叠加的噪声三者达到最佳匹配，使噪声能量向信号转移，达到微弱信号能量的加强，从而实现微弱故障信号的提取。对于满足绝热近似理论的小参数信号来说，可以直接应用双稳系统随机共振进行信号检测。而对于大参数信号而言，应用变尺度随机共振方法可通过对大频率信号的频率进行线性压缩，使信号转化为满足绝热近似理论的小参数信号，从而实现大参数信号的随机共振检测。Duffing系统本身参数的调节可使系统和信号能够与大噪声达到最佳匹配，提高随机共振检测效果。最后课题将基于二阶线性系统与非线性Duffing系统的信号检测方法分别应用于转子系统故障诊断中，对两种方法进行了比较分析，认为对于采样点数少、噪声较低，且存在大幅值频率干扰的故障信号，基于线性系统的检测方法要优于非线性Duffing系统的随机共振检测。
[Abstract]:In modern production, real-time monitoring and fault diagnosis of important key equipment is of great significance. However, in the early fault of many equipment, the fault signal is often submerged by noise and difficult to detect, so the detection of weak signal is especially important, and more attention has been paid to it. Based on the second-order linear system as the signal detection processing model, a weak signal processing method based on the parameter-adjusted resonance of the second-order linear dynamic system is proposed. The basic principle is that the signal to be detected including the characteristic frequency component and noise interference is regarded as the excitation of the system, while the linear dynamic system is regarded as the signal processing model. By artificially adjusting the natural frequency and damping ratio of the system, the natural frequency is equal to the characteristic frequency of the signal to be detected, so that the response of the system reaches resonance, and then the characteristic signal is further prominent and the noise is overcome. To achieve the purpose of detecting the components of the characteristic signal. The concrete implementation of this method is that the characteristic signal in the noise can be identified according to the maximum value in the curve after the characteristic curve of the maximum value of the system response varying with the natural frequency is obtained. This kind of signal detection based on linear system model can optimize the detection result by selecting the appropriate damping ratio parameters of the system. Secondly, taking the typical Duffing system as the object, the information detection method based on stochastic resonance of second-order nonlinear system is studied. By means of the nonlinear bistable system, the weak signal to be measured and the noise superimposed in the signal are matched optimally, so that the noise energy is transferred to the signal and the weak signal energy is enhanced, thus the weak fault signal is extracted. For the small parameter signal which satisfies the adiabatic approximation theory, the bistable system stochastic resonance can be directly used to detect the signal. For large-parameter signals, the variable-scale stochastic resonance method can be used to transform the large-frequency signals into small-parameter signals which satisfy the adiabatic approximation theory by linearly compressing the frequency of large-frequency signals. So the random resonance detection of large parameter signal can be realized. The adjustment of duffing system parameters can make the system and signal best match with large noise, and improve the effect of random resonance detection. Finally, the signal detection method based on the second-order linear system and the nonlinear Duffing system is applied to the fault diagnosis of the rotor system respectively. The two methods are compared and analyzed, and the results show that the sampling points are less and the noise is lower for the two methods. The detection method based on linear system is better than the stochastic resonance detection of nonlinear Duffing system for the fault signal with large amplitude and frequency interference.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH165.3

【参考文献】