改进分数阶Tikhonov正则化的截割煤岩载荷识别方法
Identification method of cutting coal and rock load based on improved fractional Tikhonov regularization
-
摘要: 为探究截齿截割煤岩载荷的有效识别方法，实现截割载荷特征的提取与辨识，根据分数阶微积分理论，将经典的整数阶Tikhonov正则化推广到分数阶模式，构造改进分数阶滤波因子，提出了一种改进分数阶Tikhonov正则化方法和算法。根据时域方法理论建立截割煤岩载荷的识别模型，通过核函数方法将载荷表示为一系列核函数的叠加，测量载荷表示为输入载荷和核函数响应之间的卷积分形式，采用离散化将卷积方程转化为线性方程组，利用改进分数阶正则化方法将反求过程转化为一类无约束优化问题，并采用新超记忆梯度法求解目标函数。研究表明:随着分数阶次的增大，均方根误差(RMSE)及迭代次数值呈先减小后增大的趋势，存在着最小RMSE和最少迭代次数值，可以判断存在最优的分数阶次，即α=05，此时载荷识别效果相对理想。与整数阶和分数阶Tikhonov正则化方法相比较，改进的算法不仅能够保留较小奇异值对应的分量，且也能抑制较大奇异值对应的分量，从而能够有效克服载荷识别的病态性，且被识别载荷与试验载荷的均方根误差(RMSE)分别为0418 2，0388 4，0366 5，及迭代次数分别为19，14，11，具有较高的精度，能够克服其解的光滑性，且载荷细节特征能够较好被识别。据此，改进分数阶正则化方法在截割煤岩载荷识别方面具有更强的抗噪性和鲁棒性，为解决截割煤岩载荷及矿山机械工程中的载荷识别问题提供了一种有效研究方法。
-
关键词:
- 截割煤岩 /
- 载荷识别 /
- 改进分数阶Tikhonov正则化 /
- 无约束优化 /
- 新超记忆梯度法
Abstract: In order to explore the effective identification method of cutting coal and rock load and realize the extraction and identification of cutting load characteristics,according to fractional calculus theory,the classical integral Tikhonov regularization is extended to fractional mode,and an improved fractional filter factor is constructed,an improved frac- tional Tikhonov regularization method and algorithm are proposed. Based on the theory of time domain method,the i- dentification model of the coal and rock load is established. The load is expressed as a series of kernel functions by the kernel function method. The measured load is expressed as the volume integral form between the input load and the response of the kernel function. The convolution equation is converted into a linear system of equations by discretization, andThe improved fractional order regularization method is used to transform the reverse process into a class of uncon- strained optimization problems,and the novel super memory gradient method is used to solve the objective function. The results show that with the increase of fractional order,root mean square error ( RMSE) and iteration number de- crease first and then increase. There are minimum RMSE and minimum iteration number. It can be judged that there is an optimal fractional order,that is α = 0. 5. At this time,the load identification effect is relatively ideal. Compared with the integer order and fractional order Tikhonov regularization method,the improved algorithm can not only retain the components corresponding to smaller singular values,but also suppress the components corresponding to larger singular values,thus which can effectively overcome the ill conditioned of load identification,and the Root mean square error (RMSE) of the identified load and the test load (RE) is 0. 418 2,0. 388 4,0. 366 5,and the number of iterations is 19,14,11,respectively,it has high accuracy,can overcome the smoothness of its solution,and the load details can be well identified. Therefore,the improved fractional regularization method has stronger anti noise and robustness in iden- tifying the load of cutting coal and rock,and provides an effective method for solving the problem of load identification of coal and rock and the problem of load identification in mine mechanical engineering. -
-
期刊类型引用(8)
1. 李明强，王林军，罗城升，陈保家. 基于改进分数阶Tikhonov正则化的桥梁移动载荷识别. 电子测量技术. 2023(15): 59-66 . 
百度学术
2. 张丹，王栋辉，徐鹏，张标，胡金平. 滚筒随机载荷下采煤机行走机构齿销啮合力研究. 煤矿机械. 2022(09): 31-34 . 百度学术
3. 任春平，赵中旭，马化凯. Canny算法与正则化的煤岩破碎图像边缘重构方法. 黑龙江科技大学学报. 2021(01): 43-47 . 百度学术
4. 王鑫，徐兰欣，陈洪月，白杨溪，陈洪岩. 基于弹-塑-断裂理论的镐型截齿截割机理研究与实验验证. 振动与冲击. 2021(03): 165-171+194 . 百度学术
5. 陈少军. 基于微积分和凸集投影技术的图像超分辨率重建技术研究. 西安文理学院学报(自然科学版). 2021(02): 38-42+47 . 百度学术
6. 于宁，孙业新，陈洪月. 基于多源数据融合的采煤机截割载荷预测方法. 中国机械工程. 2021(10): 1247-1253+1259 . 百度学术
7. 刘春生，李德根，任春平. 基于熵权的正则化神经网络煤岩截割载荷谱预测模型. 煤炭学报. 2020(01): 474-483 . 本站查看
8. 任春平，刘若涵，徐鹏. 镐型截齿截割煤岩载荷的区间识别模型与解算方法. 黑龙江科技大学学报. 2020(06): 649-653 . 百度学术
其他类型引用(4)
下载:
图(7)
计量
- 文章访问数: 1108
- HTML全文浏览量: 3
- PDF下载量: 308
- 被引次数: 12
