Grey evaluation and prediction model of safety risk in airworthiness directive cases
-
摘要:
基于1996～2019年的适航指令（CAD）数据，以关联分析得出的典型故障构建评价指标。用风险系数确定白化权函数的边界条件，根据偏差最大化原则确定各指标的权重，提出了基于灰色白化权聚类的安全风险评估模型。利用等时间间隔的各指标数据构成灰色预测模型，并通过优化权重值将预测精度提高了2.88%。预测的故障数量与实际发生相符，证明了预测模型的准确性，可依此有针对地提出改进和预防措施，从而减少故障的发生。
Abstract:Based on the data of China Airworthiness Directive (CAD) in 1996−2019, the evaluation indexes were constructed with the typical failures obtained by correlation analysis. The boundary conditions of whitening weight function were determined by risk coefficient, and the weight of each index was determined according to the principle of maximization of deviation. A safety risk assessment model based on grey whitening weight clustering was proposed. The grey prediction model was constructed by using the index data of equal time interval, and the prediction accuracy was improved by 2.88% by optimizing the weight value. The predicted number of failures was consistent with the actual occurrence, which proved the accuracy of the prediction model. Accordingly, improvement and prevention measures can be put forward to reduce the occurrence of faults.
-
图 1 基于灰色聚类的CAD安全风险评估流程
Figure 1. CAD security risk assessment process based on grey clustering
表 1 A320机型的CAD关联规则挖掘结果
Table 1. CAD association rule mining results of A320 models
故障编号 规则 频率 x1 隔框，裂纹，机身故障→严重影响的 0.11 x2 翼肋，机翼故障→严重影响的 0.22 x3 磨损，油箱、线束故障→严重影响的 0.16 x4 舱门作动筒，液压油渗漏，主起落架故障→危险的 0.12 x5 传感器失效，信号控制组件异常，收放系统故障→危险的 0.08 x6 磨损，发动机短舱吊架故障→危险的 0.07 x7 发动机失速，气路故障→灾难的 0.11 x8 高压引气活门故障、漏气，发动机故障→
灾难的0.06 x9 方向舵，飞行控制系统故障→灾难的 0.07 y1 设计、制造、人为因素分析→设计缺陷 0.44 y2 设计、制造、人为因素分析→制造缺陷 0.29 y3 设计、制造、人为因素分析→人为缺陷 0.27 
下载: 导出CSV
表 2 1996～2019年A320机型CAD关联规则数量统计
Table 2. Statistics of CAD association rules of A320 models in 1996—2019
年份 故障编号 x1 x2 $ \cdots $ x8 x9 y1 y2 y3 1996 5 3 $ \cdots $ 0 1 8 5 4 1997 3 7 $ \cdots $ 0 2 10 6 2 $ \vdots$ $ \vdots$ $ \vdots$ $\vdots$ $ \vdots$ $ \vdots$ $ \vdots$ $ \vdots$ $ \vdots$ 2018 3 4 $ \cdots $ 1 3 10 7 8 2019 3 2 $ \cdots $ 1 4 9 5 4
下载: 导出CSV
表 3 白化权函数
Table 3. White the weight function
参数 k=1 k=2 k=3 j=1 [ , ,0.47,2.18] [1.14,2.85, ,4.56] [3.51,5.23, , ] j=2 [ , ,0.44,1.83] [0.98,2.28, ,3.78] [2.93,4.32, , ] $ \vdots$ $ \vdots$ $\vdots $ $\vdots $ j=9 [ , ,0.44,1.24] [0.75,1.56, ,2.36] [1.87,2.68, , ] g=1 [ , ,1.18,2.44] [2.10,3.37, ,4.64] [4.3,5.57, , ] g=2 [ , ,2.19,3.71] [3.30,4.82, ,6.33] [5.93,7.44, , ] g=3 [ , ,2.63,4.46] [3.97,5.79, ,7.62] [7.13,8.95, , ]
下载: 导出CSV
表 4 1996～2019年CAD安全风险评估结果
Table 4. CAD safety risk assessment results in 1996—2019
年份 Amax,1 Amax,2 灰类（风险等级） 1996 $ {\sigma }_{1}^{1} $ $ {\sigma }_{2}^{3} $ 高 1997 $ {\sigma }_{1}^{1} $ $ {\sigma }_{2}^{3} $ 高 $\vdots $ $\vdots $ $\vdots $ $\vdots $ 2018 $ {\sigma }_{1}^{3} $ $ {\sigma }_{2}^{3} $ 低 2019 $ {\sigma }_{1}^{2} $ $ {\sigma }_{2}^{3} $ 中
下载: 导出CSV
表 5 x7叠加值预测值及误差比较
Table 5. x7 superposition prediction value and error comparison
年份 叠加值 LS-SVM
模型GM（1,1）
模型改进GM（1,1）
模型预测值 离差
平方预测值 离差
平方预测值 离差
平方1996 0 0 0 0 0 0 0 1997 1 0.51 0.24 0.63 0.14 0.69 0.10 $\vdots $ $\vdots $ $\vdots $ $\vdots $ $\vdots $ $\vdots $ $\vdots $ $\vdots $ 2016 25 23.83 1.37 26.05 1.10 25.55 0.30 2017 29 27.68 1.74 28.52 0.23 30.10 1.21 2018 33 31.07 3.72 31.16 3.39 32.83 0.03 2019 34 32.84 1.35 33.98 0.00 34.73 0.53
下载: 导出CSV
表 6 2020年CAD安全风险评估结果
Table 6. CAD safety risk assessment results 2020
参数 实际值 预测值 $ {\sigma }_{1}^{1} $ 0.1217 0.1302 $ {\sigma }_{1}^{2} $ 0.1911 0.1692 $ {\sigma }_{1}^{3} $ 0.5367 0.5373 灰类 低 低
下载: 导出CSV
表 7 2021～2025年CAD安全风险评估结果
Table 7. CAD safety risk assessment results 2021—2025
年份 $ {\sigma }_{1}^{1} $ $ {\sigma }_{1}^{2} $ $ {\sigma }_{1}^{3} $ 灰类 2021 0.1256 0.2819 0.4540 低 2022 0.1417 0.3447 0.4213 低 2023 0.2169 0.2417 0.4246 低 2024 0.2717 0.1453 0.4392 低 2025 0.3537 0.0921 0.4319 低
下载: 导出CSV
-
[1] ZHANG X G,MAHADEVAN S. Bayesian network modeling of accident investigation reports for aviation safety assessment[J]. Reliability Engineering and System Safety,2020,30(20): 860-901. [2] SEUNGWON N,JOHN S. Barrier analysis of an aviation safety assessment model[J]. Incose International Symposium,2018,28(1): 616-627. doi: 10.1002/j.2334-5837.2018.00504.x [3] 张泽京,张曙光,柳旭,等. 无人机系统安全目标水平预估方法[J]. 航空动力学报,2018,33(4): 1017-1024.ZHANG Zejing,ZHANG Guangshu,LIU Xu,et al. Estimated method of target level of safety for unmanned aircraft system[J]. Journal of Aerospace Power,2018,33(4): 1017-1024. (in Chinese) [4] 鲍梦瑶,李果,丁水汀. 基于模型的航空发动机系统安全性研究[J]. 航空动力学报,2016,31(8): 2029-2039.BAO Mengyao,LI Guo,DING Shuiting. Research on model-based safety analysis for aero-engine[J]. Journal of Aerospace Power,2016,31(8): 2029-2039. (in Chinese) [5] 楚娜娜,张曙光,高艳蕾,等. 基于Simscape模型的航空发动机系统安全性分析方法[J]. 航空动力学报,2021,36(4): 885-896.CHU Nana,ZHANG Shuguang,GAO Yanlei,et al. Safety analysis method of aero-engine systems based on Simscape model[J]. Journal of Aerospace Power,2021,36(4): 885-896. (in Chinese) [6] 文军. 基于灰色多层次的航空公司飞行安全评价研究[J]. 中国安全科学学报,2010,20(2): 29-34,178. doi: 10.3969/j.issn.1003-3033.2010.02.005WEN Jun. Airline flight safety risk assessment based on grey hierarchy method[J]. China Safety Science Journal,2010,20(2): 29-34,178. (in Chinese) doi: 10.3969/j.issn.1003-3033.2010.02.005 [7] 葛志浩,徐浩军,刘琳,等. 飞行事故概率模型与风险评估方法[J]. 中国安全科学学报,2008,18(2): 162-165,180. doi: 10.3969/j.issn.1003-3033.2008.02.028GE Zhihao,XU Haojun,LIU Lin,et al. Probability model for flight accidents and risk assessment method[J]. China Safety Science Journal,2008,18(2): 162-165,180. (in Chinese) doi: 10.3969/j.issn.1003-3033.2008.02.028 [8] ZHANG X G,MAHADEVAN S. Ensemble machine learning models for aviation incident risk prediction[J]. Decision Support Systems,2018,116(18): 48-63. [9] DOTHANG T,WOOJIN C. Using machine learning algorithms to predict the risk of small unmanned aircraft system violations in the national airspace system[J]. Journal of Air Transport Management,2020,86(20): 161-171. [10] 王岩韬,赵嶷飞. 基于多算法协作的航班运行风险辨识研究[J]. 中国安全科学学报,2018,28(6): 166-172.WANG Yantao,ZHAO Yifei. Research on flight operations risk identification based on multi-algorithm collaboration[J]. China Safety Science Journal,2018,28(6): 166-172. (in Chinese) [11] 崔杰,党耀国,刘思峰. 基于灰色关联度求解指标权重的改进方法[J]. 中国管理科学,2008,26(5): 141-145. doi: 10.16381/j.cnki.issn1003-207x.2008.05.026CUI Jie,DANG Yaoguo,LIU Sifeng. An improved approach for determining weights of attributes in decision making based on grey incidence[J]. Chinese Journal of Management Science,2008,26(5): 141-145. (in Chinese) doi: 10.16381/j.cnki.issn1003-207x.2008.05.026 [12] 党耀国,耿率帅,尚中举. 基于指标信息量的灰色可能度聚类模型研究[J]. 数学的实践与认识,2020,50(14): 6-11.DANG Yaoguo,GENG Shuaishuai,SHANG Zhongju. Research on the grey clustering model based on information of data index[J]. Mathematics in Practice and Theory,2020,50(14): 6-11. (in Chinese) [13] 段辉明,吴雨,龙杰. 数据信息对灰色GM(1, 1)模型的影响[J]. 统计与决策,2021,37(5): 54-59.DUAN Huiming,WU Yu,LONG Jie. Influence of data information on grey GM(1, 1) model[J]. Statistics and Decision,2021,37(5): 54-59. (in Chinese) [14] ELVIS T,ASUMING F E,DANIEL K,et al. Improvement of gray system model using particle swarm optimization[J]. Journal of Electrical Systems and Information Technology,2021,8(1): 12-27. doi: 10.1186/s43067-021-00036-9 [15] GURCAN C,NEGASH B,NATHAN H. Improved grey system models for predicting traffic parameters[J]. Expert Systems with Applications,2021,177(21): 72-83. [16] 王衍洋,曹义华. 航空运行风险的灰色神经网络模型[J]. 航空动力学报,2010,25(5): 1036-1042.WANG Yanyang,CAO Yihua. Gray neural network model of aviation safety risk[J]. Journal of Aerospace Power,2010,25(5): 1036-1042. (in Chinese) [17] WAN J,XIA Z H,ZHU X P. Operational safety risk assessment of civil aviation based on grey clustering[J]. Journal of Physics,2019,168(3): 109-117. [18] 曹惠玲,胡彦杰. 适航指令设计制造缺陷案例关联分析方法研究[J]. 中国安全科学学报,2021,31(11): 54-63.CAO Huiling,HU Yanjie. Association analysis method of CAD design and manufacturing defect cases[J]. China Safety Science Journal,2021,31(11): 54-63. (in Chinese) -
点击查看大图
计量
- 文章访问数: 163
- HTML浏览量: 117
- PDF量: 46
- 被引次数: 0