基于逆虚拟激励法的掘进机截割机构的载荷识别

发布时间：2018-04-30 18:33

本文选题：掘进机截割机构 + 逆虚拟激励法　；参考：《中北大学》2017年硕士论文

【摘要】：掘进机在高速重载条件下工作,其良好的性能是煤矿安全生产的保障,而其安全性与截割机构的运行状态有很大关系。通过对掘进机截割机构的载荷分析研究,及时精确识别其承受动态载荷,能够为故障预示和演化、寿命预测等提供技术支持。然而截割机构的载荷通常又无法直接测试得到,因此就有必要通过间接手段求解其载荷,载荷识别技术是一种非常有效的方法。本文首先对载荷识别技术进行了研究,载荷识别技术包括频域法、时域法和人工智能方法,由于掘进机截割机构受到的是随机载荷,随机载荷识别一般采用频域法,而其中的逆虚拟激励法计算简单,精度也较高,因此本文通过逆虚拟激励法对掘进机截割机构载荷进行识别。逆虚拟激励法需要得知系统的频响函数逆矩阵,鉴于频响函数求逆存在病态问题,本文引入改进Tikhonov正则化法来改善识别效果,用GCV准则来求解正则参数,并用该方法做了悬臂梁的MATLAB仿真,得出其可以将载荷识别误差降低2.3dB,在改善精度方面较之前方法有较理想的效果。本文通过实验验证了该方法的可行性。由于矿井中的响应测试环境比较复杂,很难顺利和高效地进行测试振动信号的工作,因而本文搭建即地面假岩壁的载荷试验平台,在井上进行了切割假岩壁的试验。要测得掘进机截割机构上的振动响应,必先须对传感器进行优化布置。本文根据模态置信准则建立了传感器的适应度函数,利用粒子群优化算法求解其适应度值作为评价传感器组合优劣的依据,得出了传感器的优化布置方式,进而测得了截割机构的振动响应。此外还需要获得掘进机截割机构的频响函数,本文通过ANSYS中进行瞬态分析求得频响函数。在掘进机截割机构的有限元模型上根据测点布置原则选择激励点施加载荷,根据瞬态动力学分析获得了截割机构上响应测点的位移信号。在MATLAB中写入载荷,和测点位移,分别对它们作傅里叶变换,然后据此求出系统的频响函数。获得其频响函数后,由改进正则化法求解其广义逆矩阵。然后采用逆虚拟激励法,通过实测振动信号建立响应功率谱矩阵,构造虚拟响应向量,求解各虚拟激励向量,合成载荷功率谱矩阵。通过与由实际测得的应变求得截割机构的载荷功率谱对比,得出该方法可以在非固有频率频段内和固有频率频段内分别将载荷识别误差最大降低2.23dB和1.44dB。因此该方法可以提升载荷识别的精度,为更深入地研究掘进机的力学特性及对其进行优化设计提供了依据。
[Abstract]:The good performance of roadheader working under the condition of high speed and heavy load is the guarantee of coal mine safety production, and its safety is closely related to the running state of cutting mechanism. By analyzing and studying the load of the cutting mechanism of the roadheader, the dynamic load can be recognized accurately and timely, which can provide technical support for fault prediction and evolution, life prediction and so on. However, the load of cutting mechanism can not be measured directly, so it is necessary to solve the load by indirect means. Load identification is a very effective method. In this paper, the technology of load identification is studied firstly. The technology of load identification includes frequency domain method, time domain method and artificial intelligence method. Because the cutting mechanism of roadheader is subjected to random load, the random load identification generally adopts frequency domain method. The inverse virtual excitation method is simple in calculation and high in accuracy, so this paper uses the inverse virtual excitation method to identify the load of the cutting mechanism of the tunneling machine. The inverse virtual excitation method needs to know the inverse matrix of the frequency response function of the system. In view of the ill-posed problem of finding the inverse of the frequency response function, the improved Tikhonov regularization method is introduced to improve the recognition effect and the GCV criterion is used to solve the regular parameters. The method is used to simulate the cantilever beam by MATLAB, and it is concluded that the load identification error can be reduced by 2.3 dB, and the accuracy of the method is better than that of the previous method. The feasibility of this method is verified by experiments in this paper. Because the response testing environment in mine is complex, it is difficult to test vibration signals smoothly and efficiently. Therefore, the loading test platform of surface pseudo rock wall is set up in this paper, and the test of cutting false rock wall is carried out in the well. In order to measure the vibration response of the cutting mechanism of the roadheader, the sensor must be arranged optimally. In this paper, the fitness function of the sensor is established according to the modal confidence criterion. The particle swarm optimization algorithm is used to solve its fitness value as the basis for evaluating the combination of sensors, and the optimal arrangement of the sensor is obtained. Then the vibration response of the cutting mechanism is measured. In addition, the frequency response function of the cutting mechanism of roadheader is also needed. The frequency response function is obtained by transient analysis in ANSYS. In the finite element model of the cutting mechanism of the roadheader, the excitation point is selected to apply load according to the principle of measuring point arrangement, and the displacement signal of the response point on the cutting mechanism is obtained according to the transient dynamic analysis. The load and displacement of measuring points are written in MATLAB. The Fourier transform is performed on them, and then the frequency response function of the system is obtained. After the frequency response function is obtained, the generalized inverse matrix is solved by the improved regularization method. Then the response power spectrum matrix is established by using the inverse virtual excitation method and the response power spectrum matrix is established by the measured vibration signal. The virtual response vector is constructed to solve each virtual excitation vector and the load power spectrum matrix is synthesized. Compared with the actual measured strain, the load power spectrum of the cutting mechanism is obtained. It is concluded that this method can reduce the maximum load identification error in the frequency band of the non-natural frequency and the frequency band of the natural frequency, respectively, and reduce the 2.23dB and 1.44 dB respectively. Therefore, this method can improve the accuracy of load identification, and provide a basis for further research on mechanical characteristics and optimization design of roadheader.
【学位授予单位】：中北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TD421.5

【参考文献】