Wear State Prediction of Crusher Liner Based on LMS of Improved Dustpan Tongue Function and BAS-LSSVM
-
摘要: 针对破碎机衬板磨损难以预测的问题,提出一种改进箕舌线函数的LMS和BAS-LSSVM的衬板磨损状态预测方法。首先,在最小均方误差算法(LMS)的基础上,引入改进的箕舌线函数,提出改进箕舌线函数的LMS算法,将其用于衬板超声回波信号的声时(TOF)的计算中;其次,通过TOF计算出衬板厚度,并根据衬板磨损前后的厚度变化得出磨损量;最后利用天牛须算法(BAS)优化最小二乘支持向量机(LSSVM)的惩罚因子
$\gamma $ 和其核函数中的标准化参数$\sigma $ ,将磨损量作为预测模型的输入,衬板磨损阶段作为输出,建立BAS-LSSVM衬板磨损预测模型。结果表明,该方法对动锥衬板和定锥衬板的识别准确率分别达到了94.44%和95.56%,能够有效预测出衬板的磨损状态。Abstract: In order to solve the problem that the wear of crusher liner is difficult to predict, a new method for predicting the wear state of crusher liner based on LMS and BAS-LSSVM with improved half-pan tongue function is proposed. Firstly, based on the least mean square error (LMS) algorithm, an improved dustpan tongue function LMS was introduced to calculate the acoustic time (TOF) of the ultrasonic echo signal of the lining board. Secondly, the thickness of liner was calculated by TOF, and the wear amount was obtained according to the thickness change of liner before and after wear. Finally, a BAS-LSSVM liner wear prediction model was established by optimizing the penalty factor of least squares support vector machine (LSSVM) and the standardized parameters in its kernel function by using the Beetle Antennae Search algorithm (BAS). The wear amount was taken as the input of the prediction model, and the lining wear stage was taken as the output. The experimental results show that the identification accuracy of moving cone liner and fixed cone liner can reach 94.44% and 95.56% respectively, which can effectively predict the wear state of the crusher liner.-
Key words:
- wear state prediction /
- LMS /
- TOF calculation /
- LSSVM
-
表 1 人工测量衬板各测点厚度
衬板 测点编号 厚度/mm 定锥衬板 A1 45.48 A2 46.07 A3 48.86 A4 48.91 动锥衬板 B1 11.09 B2 11.89 B3 13.74 B4 14.89 表 2 4种方法计算衬板各测点测厚的相对误差
测点 平均相对误差/% 互相关法 传统LMS
算法基于箕舌线的
LMS算法本文提出
方法A1 0.336 0.307 0.261 0.222 A2 0.268 0.256 0.246 0.242 A3 0.446 0.406 0.365 0.316 A4 0.376 0.357 0.346 0.338 B1 0.302 0.298 0.295 0.291 B2 0.296 0.287 0.278 0.272 B3 0.526 0.505 0.489 0.467 B4 0.365 0.358 0.343 0.324 表 3 衬板各测点磨损量及相对误差
测点 人工检测
磨损量/mm超声检测
磨损量/mm相对误差/
%A1 0.93 0.88 5.38 A2 1.14 1.11 2.63 A3 1.32 1.40 5.71 A4 2.07 1.93 6.28 B1 0.86 0.81 5.81 B2 0.97 1.02 4.90 B3 1.13 1.15 2.61 B4 1.55 1.51 2.58 表 4 3种优化算法的初始参数设置
算法名称 参数名称 对应值 GA 最大迭代次数 200 种群数量 20 交叉概率 0.9 变异概率 0.1 WOA 鲸鱼数量 20 最大迭代次数 200 BAS 步长因子c1、c2 3、10 最大迭代次数 200 表 5 3种测试函数运行10次的平均最优解
测试函数 Sphere Griewank Rosenbrock GA 1.0426 0.2758 0.2244 WOA 0.6193 0.0623 0.0446 BAS 0.1258 0.0014 0.0715 表 6 3种预测模型预测准确率
% 算法 衬板 磨合磨损阶段 稳定磨损阶段 破坏磨损阶段 平均预测准确率 GA-LSSVM 定锥 90 80.0 86.67 85.56 动锥 90 86.67 86.67 87.78 WOA-LSSVM 定锥 86.67 90 90 88.89 动锥 90 86.67 93.33 90 BAS-LSSVM 定锥 93.33 93.33 96.67 94.44 动锥 93.33 96.67 96.67 95.56 -
[1] 王汝杰, 王威, 宋仁伯, 等. 热处理工艺参数对超高强度衬板用合金钢组织及冲击磨料磨损性能的影响[J]. 材料导报, 2016, 30(S2): 610-615.WANG R J, WANG W, SONG R B, et al. Effect of heat treatment parameters on microstructure and impact-abrasive wear resistance of ultra-high strength alloyed steel for liner[J]. Materials Review, 2016, 30(S2): 610-615. (in Chinese) [2] 姚一民. 球磨机磨球运动学分析及对衬板影响的研究[D]. 武汉: 武汉理工大学, 2010YAO Y M. The study of motion analysis of milling ball and to the liner influence on the ball mill[D]. Wuhan: Wuhan University of Technology, 2010. (in Chinese) [3] LINDQVIST M, VERTSSON C M. Development of wear model for cone crushers[J]. Wear, 2006, 261(3-4): 435-442. doi: 10.1016/j.wear.2005.12.010 [4] HOSEINI M R, WANG X D, ZUO M J. Estimating ultrasonic time of flight using envelope and quasi maximum likelihood method for damage detection and assessment[J]. Measurement, 2012, 45(8): 2072-2080. doi: 10.1016/j.measurement.2012.05.008 [5] 陈建, 孙晓颖, 林琳, 等. 基于单周期互相关滤波的超声波TOF检测方法[J]. 仪器仪表学报, 2014, 35(3): 664-669. doi: 10.19650/j.cnki.cjsi.2014.03.025CHEN J, SUN X Y, LIN L, et al. Accurate ultrasonic TOF measurement method based on monocycle cross-correlation filtering[J]. Chinese Journal of Scientific Instrument, 2014, 35(3): 664-669. (in Chinese) doi: 10.19650/j.cnki.cjsi.2014.03.025 [6] 李鹏, 蔡玉雷, 陈金立. 采用椭圆算法和Canny算子的超声TOF估计方法[J]. 仪器仪表学报, 2014, 35(10): 2385-2391. doi: 10.19650/j.cnki.cjsi.2014.10.031LI P, CAI Y L, CHEN J L. Ultrasonic TOF estimation method using ellipse algorithm and Canny operator edge detection[J]. Chinese Journal of Scientific Instrument, 2014, 35(10): 2385-2391. (in Chinese) doi: 10.19650/j.cnki.cjsi.2014.10.031 [7] 者娜, 杨剑锋, 刘文彬, 等. KPCA和改进SVM在滚动轴承剩余寿命预测中的应用研究[J]. 机械设计与制造, 2019(11): 1-4. doi: 10.3969/j.issn.1001-3997.2019.11.001ZHE N, YANG J F, LIU W B, et al. Research on application of KPCA and improved SVM in residual life prediction of rolling bearings[J]. Machinery Design & Manufacture, 2019(11): 1-4. (in Chinese) doi: 10.3969/j.issn.1001-3997.2019.11.001 [8] 崔建国, 滑娇娇, 董世良, 等. 基于SK和LSSVM的飞机空气制冷机寿命趋势分析[J]. 振动. 测试与诊断, 2016, 36(3): 471-477.CUI J G, HUA J J, DONG S L, et al. Life trend analysis of aircraft air refrigerator based on SK and LSSVM[J]. Journal of Vibration, Measurement & Diagnosis, 2016, 36(3): 471-477. (in Chinese) [9] 罗小燕, 陈慧明, 卢小江, 等. 基于网格搜索与交叉验证的SVM磨机负荷预测[J]. 中国测试, 2017, 43(1): 132-135. doi: 10.11857/j.issn.1674-5124.2017.01.027LUO X Y, CHEN H M, LU X J, et al. Forecast of SVM mill load based on grid search and cross validation[J]. China Measurement & Test, 2017, 43(1): 132-135. (in Chinese) doi: 10.11857/j.issn.1674-5124.2017.01.027 [10] 马凯, 王平波, 武彩. 一种基于正态分布曲线的变步长LMS算法[J]. 计算机仿真, 2019, 36(9): 295-299. doi: 10.3969/j.issn.1006-9348.2019.09.059MA K, WANG P B, WU C. A variable-step-size LMS algorithm based on normal distribution curve[J]. Computer Simulation, 2019, 36(9): 295-299. (in Chinese) doi: 10.3969/j.issn.1006-9348.2019.09.059 [11] 火元莲, 龙小强, 连培君, 等. 一种类箕舌线函数的变步长归一化自适应滤波算法[J]. 电子与信息学报, 2021, 43(2): 335-340. doi: 10.11999/JEIT191003HUO Y L, LONG X Q, LIAN P J, et al. A kind of versoria function normalized adaptive filtering algorithm[J]. Journal of Electronics & Information Technology, 2021, 43(2): 335-340. (in Chinese) doi: 10.11999/JEIT191003 [12] 宋志雄, 喻翌, 赵海全. 基于箕舌线函数的变步长自适应谐波电流检测算法[J]. 电力系统自动化, 2013, 37(22): 54-59. doi: 10.7500/AEPS201209109SONG Z X, YU Y, ZHAO H Q. Variable-step adaptive harmonic current detection algorithm based on versiera function[J]. Automation of Electric Power Systems, 2013, 37(22): 54-59. (in Chinese) doi: 10.7500/AEPS201209109 [13] 黄沁元, 谢罗峰, 殷国富, 等. 基于变分模态分解和天牛须搜索的磁瓦内部缺陷声振检测[J]. 振动与冲击, 2020, 39(17): 124-133. doi: 10.13465/j.cnki.jvs.2020.17.017HUANG Q Y, XIE L F, YIN G F, et al. Acoustic-vibration detection for internal defects of magnetic tile based on VMD and BAS[J]. Journal of Vibration and Shock, 2020, 39(17): 124-133. (in Chinese) doi: 10.13465/j.cnki.jvs.2020.17.017 [14] 闫重熙, 陈皓. 基于改进天牛须搜索算法优化LSSVM短期电力负荷预测方法研究[J]. 电测与仪表, 2020, 57(6): 6-11. doi: 10.19753/j.issn1001-1390.2020.06.002YAN Z X, CHEN H. Research of LSSVM short-term load forecasting method based on the improved beetle antennae search algorithm[J]. Electrical Measurement and Instrumentation, 2020, 57(6): 6-11. (in Chinese) doi: 10.19753/j.issn1001-1390.2020.06.002 [15] 舒星, 刘永刚, 申江卫, 等. 基于改进最小二乘支持向量机与Box-Cox变换的锂离子电池容量预测[J]. 机械工程学报, 2021, 57(14): 118-128. doi: 10.3901/JME.2021.14.118SHU X, LIU Y G, SHEN J W, et al. Capacity prediction for lithium-ion batteries based on improved least squares support vector machine and box-cox transformation[J]. Journal of Mechanical Engineering, 2021, 57(14): 118-128. (in Chinese) doi: 10.3901/JME.2021.14.118