Comparative Study on Estimation Methods of Aircraft IDG Reliability Parameters
-
摘要: 准确估计飞机整体驱动发电机(Integrated drive generator, IDG)的可靠性分布参数, 对掌握该部件的故障变化规律和制定维修策略起到关键性作用。针对飞机IDG故障数据为小样本的特点, 以威布尔分布为例, 采用最小二乘支持向量回归机(LSSVR)、支持向量回归机(SVR)和最小二乘法(LSR)对飞机IDG进行可靠性参数估计。结合实例与蒙特卡罗仿真, 对比分析3种参数估计方法的精度、运行时间以及样本量变化时的稳定性。结果表明, 在小样本情况下, LSSVR的参数估计精度最高, LSR的运行时间最短; 随着样本量的减小, 3种参数估计方法的精度均有所减小, 但LSSVR的稳定性最好。Abstract: Accurately estimating the reliability parameters of the aircraft integral drive generator (IDG) plays a key role in mastering the failure variation law of the component and formulating maintenance strategies. Aiming at the characteristics of aircraft IDG failure data as small samples, taking Weibull distribution as an example, least squares support vector regression (LSSVR), support vector regression (SVR) and least square regression (LSR) are respectively used to estimate the reliability parameters of aircraft IDG. The accuracy, running time and stability of the three parameter estimation methods are compared and analyzed. The results show that in the case of small samples, LSSVR has the highest parameter estimation accuracy and LSR has the shortest running time; as the sample size decreases, the accuracy of the three parameter estimation methods decreases, but LSSVR has the best stability.
-
Key words:
- aircraft IDG /
- parameter estimation /
- small sample /
- LSSVM /
- SVM /
- LSR /
- weibull distribution
-
表 1 飞机IDG的部分故障数据记录
序号 ATA 机号 故障日期1 故障日期2 故障描述 维修措施 飞行时间/FH 1 2400 B5**1 201 5.08.18 2017.10.05 左发供电故障, 测试有代码IDG FAULT 航后更换左发IDG 6 836.7 2 2411 B5**7 2015.08.02 2016.02.10 航后检查左发IDG指示销跳出 航后更换左发IDG油滤 1 008.13 3 241121 B5**6 2015.12.26 2017.09.06 右发IDG的DRIVE灯不亮, 航后测试GCU无代码 航后更换右发IDG 5 103.44 4 2400 B5**5 2015.02.03 2017.09.26 航后检查左发IDG滑油指示销跳出 航后更换左发IDG回油滤和加油滤 1 742.72 5 241111 B5**1 2017.07.24 2018.06.27 航后驾驶舱右发DRIVE灯不工作, 灯光测试正常, 对调GCU故障依旧, 量线发现IDG本体低压电门故障 航后更换右发IDG 3 001.74 6 241111 B5**3 2015.07.28 2017.10.25 过站检查发现右发IDG空气滑油冷却器漏滑油 更换右发IDG空气滑油冷却器 7 870.67 表 2 故障数据、训练样本及参数估计结果
故障间隔时间/FH 训练样本 参数估计结果 xi yi LSR SVM LSSVM 1008.3000 6.9159 -3.0679 -3.0679 -2.9499 -3.0296 1742.7200 7.4632 -2.1458 -2.1458 -2.2458 -2.2939 3001.7400 8.0069 -1.6463 -1.6463 -1.5453 -1.5430 4801.8500 8.4768 -1.0103 -0.9706 -0.9430 -0.9014 5103.4400 8.5377 -0.7717 -0.8910 -0.8652 -0.8200 5658.3900 8.6409 -0.5603 -0.7561 -0.7338 -0.6832 6634.2900 8.8000 -0.3665 -0.5481 -0.5318 -0.4755 7870.6700 8.9709 -0.1836 -0.3246 -0.3159 -0.2571 9372.8800 9.1456 -0.0061 -0.0963 -0.0965 -0.0386 12187.9300 9.4082 0.1713 0.1351 0.2306 0.1784 14724.7800 9.5973 0.3549 0.4942 0.4638 0.5091 16229.9600 9.6946 0.5545 0.5545 0.5830 0.6255 17720.5300 9.7825 0.7902 0.7902 0.6902 0.7307 23561.0000 10.0673 1.1285 1.1285 1.0339 1.0766 表 3 飞机IDG的可靠性参数估计结果
参数估计方法 形状参数β 尺寸参数η LSR 1.295 9 9 914.730 3 SVR 1.174 2 10 029.451 6 LSSVR 1.240 9 9 823.669 4 表 4 飞机IDG(实际数据)的可靠性参数估计结果评价
参数估计方法 误差分析指标/NRMSE 运行时间/s LSR 0.054 2 0.098 0 SVR 0.053 4 6.910 0 LSSVR 0.051 2 3.654 0 表 5 飞机IDG(仿真数据)的可靠性参数估计结果及其评价
参数估计方法 形状参数β 尺寸参数η 误差分析指标NRMSE 运行时间/s LSR 10 329.45 1.418 2 0.060 6 0.077 0 SVR 10 456.718 8 1.448 4 0.060 2 6.101 0 LSSVR 9 923.987 6 1.661 4 0.051 4 5.901 0 -
[1] 刘晓庆, 秦曦, 钱勇, 等. 新冠肺炎疫情对全球航空业影响深远[N]. 中国航空报, 2020-04-24(007)LIU X Q, QIN X, QIAN Y, et al. The COVID-19 has a profound impact on the global airline industry[N]. Journal of Chinese Aeronautics, 2020-04-24(007) (in Chinese) [2] 白龙, 孙楚, 周元钧. 航空机电作动器的混合整流全状态反馈控制[J]. 航空学报, 2016, 37(6): 1940-1952 https://www.cnki.com.cn/Article/CJFDTOTAL-HKXB201606024.htmBAI L, SUN C, ZHOU Y J. Full-state feedback control of a novel hybrid rectifier applied to aircraft electric actuator load[J]. Acta Aeronautica et Astronautica Sinica, 2016, 37(6): 1940-1952 (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-HKXB201606024.htm [3] 孔祥芬, 张利寒, 蔡峻青. EM算法下飞机IDG删失数据的可靠性分析[J]. 机械科学与技术, 2019, 38(7): 1138-1142 https://www.cnki.com.cn/Article/CJFDTOTAL-JXKX201907020.htmKONG X F, ZHANG L H, CAI J Q. Reliability analysis of aircraft IDG failure data censored randomly by EM algorithm[J]. Mechanical Science and Technology for Aerospace Engineering, 2019, 38(7): 1138-1142 (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JXKX201907020.htm [4] 沈安慰, 郭基联, 王卓健. 航空装备现场数据可靠性评估方法有效性分析[J]. 航空学报, 2014, 35(5): 1311-1318 https://www.cnki.com.cn/Article/CJFDTOTAL-HKXB201405016.htmSHEN A W, GUO J L, WANG Z J. Validity analysis of reliability evaluation method in aviation equipment field data[J]. Acta Aeronautica et Astronautica Sinica, 2014, 35(5): 1311-1318 (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-HKXB201405016.htm [5] 王新刚, 张鑫垚, 杨禄杰, 等. 基于磨损退化数据竞争失效的刀具可靠性分析[J/OL]. 中国机械工程, 2020: 1-8[2020-07-02]. http://kns.cnki.net/kcms/detail/42.1294.TH.20191220.0928.002.htmlWANG X G, ZHANG X Y, YANG L J, et al. Competitive failure reliability analysis based on cutting tool wear degradation data[J/OL]. China Mechanical Engineering, 2020: 1-8[2020-07-02]. http://kns.cnki.net/kcms/detail/42.1294.TH.20191220.0928.002.html (in Chinese) [6] 辛龙, 周越文, 翟颖烨, 等. 基于Weibull分布的航空装备部件寿命预测研究[J]. 电光与控制, 2014, 21(12): 102-105 https://www.cnki.com.cn/Article/CJFDTOTAL-DGKQ201412023.htmXIN L, ZHOU Y W, ZHAI Y Y, et al. Weibull distribution based lifetime prediction for components of aerial equipments[J]. Electronics Optics & Control, 2014, 21(12): 102-105 (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-DGKQ201412023.htm [7] FAN Y, WANG S K, ZHOU F, et al. Parameter estimation for small sample censored data based on SVM[J]. Advanced Materials Research, 2010, 145: 31-36 doi: 10.4028/www.scientific.net/AMR.145.31 [8] PAN H S, SHENG A W, WANG Z J, et al. Analysis of MTBF evaluation methods for small sample sizes[C]//2016 11th International Conference on Reliability, Maintainability and Safety (ICRMS). Hangzhou, China: IEEE, 2016: 1-7 [9] 尹浩霖, 王达梦, 马志勇, 亦等. 支持向量回归参数估计在风电机组故障模式分析中的应用[J]. 机械科学与技术, 2018, 37(11): 1755-1761YIN H L, WANG D M, MA Z Y, et al. Application of support vector regression parameter estimation to fault mode analysis in wind turbines[J]. Mechanical Science and Technology for Aerospace Engineering, 2018, 37(11): 1755-1761 (in Chinese) [10] 张新锋, 赵彦. 发动机系统可靠性最小二乘支持向量机分析[J]. 机械设计与制造, 2012(9): 219-221 doi: 10.3969/j.issn.1001-3997.2012.09.078ZHANG X F, ZHAO Y. Reliability analysis of engine systems with least square support vector machine[J]. Machinery Design & Manufacture, 2012(9): 219-221 (in Chinese) doi: 10.3969/j.issn.1001-3997.2012.09.078 [11] PENG Y F, CHENG J S, LIU Y F, et al. An adaptive data-driven method for accurate prediction of remaining useful life of rolling bearings[J]. Frontiers of Mechanical Engineering, 2018, 13(2): 301-310 doi: 10.1007/s11465-017-0449-7 [12] 李书明, 董成利, 黄燕晓. 基于威布尔的发动机涡轮叶片寿命可靠性评估[J]. 中国民航大学学报, 2008, 26(4): 14-17 https://www.cnki.com.cn/Article/CJFDTOTAL-ZGMH200804004.htmLI S M, DONG C L, HUANG Y X. Evaluate reliableness of aero-engine turbine blades based on weibull-distributing[J]. Journal of Civil Aviation University of China, 2008, 26(4): 14-17 (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZGMH200804004.htm [13] 郭媛媛, 孙有朝, 李龙彪. 基于蒙特卡罗方法的民用飞机故障风险评估方法[J]. 航空学报, 2017, 38(10): 221126SUN Y Y, SUN Y C, LI L B. Failure risk assessment method of civil aircraft based on monte carlo method[J]. Acta Aeronautica et Astronautica Sinica, 2017, 38(10): 221126 (in Chinese) [14] TOMEVENYA K M, LIU S J. Probabilistic fatigue-creep life reliability assessment of aircraft turbine disk[J]. Journal of Mechanical Science and Technology, 2018, 32(11): 5127-5132 doi: 10.1007/s12206-018-1010-2 [15] 胡德, 郭刚正. 最小二乘法、矩法和最大似然法的应用比较[J]. 统计与决策, 2015(9): 20-24 https://www.cnki.com.cn/Article/CJFDTOTAL-TJJC201509006.htmHU D, GUO G Z. Modeling and analysis about reliability of rheostatic brake system in multiple failure modes[J]. Statistics & Decision, 2015(9): 20-24 (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-TJJC201509006.htm [16] ZHAO W, TAO T, ZIO E. System reliability prediction by support vector regression with analytic selection and genetic algorithm parameters selection[J]. Applied Soft Computing, 2015, 30: 792-802 doi: 10.1016/j.asoc.2015.02.026 [17] SUYKENS J A K, VANDEWALLE J. Least squares support vector machine classifiers[J]. Neural Processing Letters, 1999, 9(3): 293-300 [18] 何旭, 姜宪国, 张沛超, 等. 基于SVM的小样本条件下继电保护可靠性参数估计[J]. 电网技术, 2015, 39(5): 1432-1437 https://www.cnki.com.cn/Article/CJFDTOTAL-DWJS201505040.htmHE X, JIANG X G, ZHANG P C, et al. SVM based parameter estimation of relay protection reliability with small samples[J]. Power System Technology, 2015, 39(5): 1432-1437 (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-DWJS201505040.htm [19] ALMETWALLY E M, ALMONGY H M, RASTOGI M K, et al. Maximum product spacing estimation of weibull distribution under adaptive type-Ⅱ progressive censoring schemes[J]. Annals of Data Science, 2020, 7(2): 257-279