基于孪生网络的仿真模型智能排序评估方法

doi:10.12305/j.issn.1001-506X.2023.07.16

摘要/Abstract

摘要：

针对动态输出多样本条件下的多备选仿真模型择优问题, 提出了一种基于孪生卷积神经网络(siamese convolutional neural network, SCNN)的仿真模型智能排序评估方法。首先, 将仿真数据和参考数据的一致性度量问题转化为二者特征一致性度量问题。其次, 在分析评估数据特点和对比试验结果的基础上, 确定采用SCNN实现评估数据的特征提取。接下来, 给出基于SCNN的仿真模型排序评估方法, 包括网络结构初步设计、网络参数训练调优和仿真模型排序评估三部分。最后, 通过实例应用, 验证了该方法在评估数据特征提取和仿真模型排序评估方面的有效性。

关键词: 仿真模型排序评估, 孪生卷积神经网络, 多样本数据, 特征提取

Abstract:

Aiming at ranking and selection of simulation models which dynamic outputs and multiple samples, an intelligent ranking evaluation method of simulation models based on siamese convolutional neural network(SCNN) is proposed. Firstly, the consistency measurement problem of simulation data and reference data is transformed into a feature consistency measurement problem. Secondly, based on the analysis of the characteristics of the evaluation data and the comparative test results, it is determined to use SCNN to achieve feature extraction of the evaluation data. Next, a ranking evaluation method for simulation models based on SCNN is presented, which includes three parts: preliminary design of network structure, training and optimization of network parameters, and ranking evaluation of simulation models. Finally, an example application verifies the effectiveness of the proposed method in evaluation data feature extraction and simulation models ranking and selection.

Key words: ranking evaluation of simulation models, siamese convolutional neural network (SCNN), multiple sample data, feature extraction

中图分类号:

TP391.9

杨帆, 马萍, 李伟, 杨明. 基于孪生网络的仿真模型智能排序评估方法[J]. 系统工程与电子技术, 2023, 45(7): 2060-2068.

Fan YANG, Ping MA, Wei LI, Ming YANG. Intelligent ranking evaluation method of simulation models based on siamese network[J]. Systems Engineering and Electronics, 2023, 45(7): 2060-2068.

图/表 15

图1

图2

图3

表1

图4

图5

图6

表2

图7

图8

图9

表3

图10

图11

表4

参考文献 20

1	SARGENT R G . Verification and validation of simulation mo-dels[J]. Journal of Simulation, 2013, 7 (1): 12- 24. doi: 10.1057/jos.2012.20
2	WANG Y N , LI J Q , SUN H B , et al. A survey on VV & A of large-scale simulations[J]. International Journal of Crowd Science, 2019, 3 (1): 63- 86. doi: 10.1108/IJCS-01-2019-0004
3	李伟, 焦松, 陆凌云, 等. 基于特征差异的仿真模型验证及选择方法[J]. 自动化学报, 2014, 40 (10): 2134- 2144.
	LI W , JIAO S , LU L Y , et al. Verification and selection method of simulation model based on feature difference[J]. Journal of Automation, 2014, 40 (10): 2134- 2144.
4	OBERKAMPF W L , DELAND S M , RUTHERFORD B M , et al. Error and uncertainty in modeling and simulation[J]. Reliability Engineering & System Safety, 2002, 75 (3): 333- 357.
5	ARORA H , BANSAL M . Feature extraction through sentiment analysis of tourist sentiments using deep learning techniques like CNN, RNN and LSTM[J]. International Journal of Recent Technology and Engineering, 2020, 9 (1): 2254- 2261.
6	AHMED M Z , MAHESH C . An efficient image based feature extraction and feature selection model for medical data clustering using deep neural networks[J]. Traitement du Signal, 2021, 38 (4): 1141- 1148. doi: 10.18280/ts.380425
7	BENGIO Y , COURVILLE A , VINCENT P , et al. Representation learning: a review and new perspectives[J]. IEEE Trans.on Pattern Analysis and Machine Intelligence, 2013, 35 (8): 1798- 1828. doi: 10.1109/TPAMI.2013.50
8	周玉臣, 方可, 马萍, 等. 基于集成学习的复杂仿真模型验证方法[J]. 系统工程与电子技术, 2018, 40 (9): 2124- 2130.
	ZHOU Y C , FANG K , MA P , et al. Verification method of complex simulation model based on ensemble learning[J]. Systems Engineering and Electronics, 2018, 40 (9): 2124- 2130.
9	BROMLEY J, GUYON I, LECUN Y, et al. Signature verification using a "Siamese" time delay neural network[C]//Proc. of the 6th International Conference on Neural Information Processing Systems, 1994: 737-744.
10	ZHOU Q, XIA H R, YAN H Z, et al. Adaptive multi-feature fusion visual target tracking based on siamese neural network with cross-attention mechanism[C]//Proc. of the 22nd IEEE/ACM International Symposium on Cluster, Cloud and Internet Computing, 2022: 307-316.
11	QI Y, SONG Y Z, ZHANG H, et al. Sketch-based image retrieval via Siamese convolutional neural network[C]//Proc. of the IEEE International Conference on Image Processing, 2016: 4586-4595.
12	徐先峰, 张丽, 郎彬, 等. 引入感知模型的改进孪生卷积神经网络实现人脸识别算法研究[J]. 电子学报, 2020, 48 (4): 643- 647. doi: 10.3969/j.issn.0372-2112.2020.04.004
	XU X F , ZHANG L , LANG B , et al. Research on face recognition algorithm based on improved twin convolution neural network with perceptual model[J]. Acta Electronica Sinica, 2020, 48 (4): 643- 647. doi: 10.3969/j.issn.0372-2112.2020.04.004
13	GU Y F , YANG X B , TIAN L , et al. Structure-aware siamese graph neural networks for encounter-level patient similarity learning[J]. Journal of Biomedical Informatics, 2022, 127, 104027. doi: 10.1016/j.jbi.2022.104027
14	PHREERAPHATTANAKARN T , KIJSIRIKUL B . Text data-augmentation using text similarity with Manhattan siamese long short-term memory for Thai language[J]. Journal of Physics: Conference Series, 2021, 1780 (1): 12- 18.
15	VARIOR R R, SHUAI B, LU J W, et al. A siamese long short-term memory architecture for human re-identification[C]//Proc. of the European Conference on Computer Vision, 2016: 135-153.
16	MESQUITA D B , SANTOS R , MACHARET D G , et al. Fully convolutional siamese autoencoder for change detection in UAV aerial images[J]. IEEE Geoscience and Remote Sensing Letters, 2020, 17 (8): 1455- 1459. doi: 10.1109/LGRS.2019.2945906
17	ZHAO G W , PENG Y X . Semisupervised SAR image change detection based on a siamese variational autoencoder[J]. Information Processing and Management, 2022, 59 (1): 102726. doi: 10.1016/j.ipm.2021.102726
18	LI J G, XU C J, ZHANG T. Similarity measure of time series based on siamese and sequential neural networks[C]//Proc. of the 39th Chinese Control Conference, 2020: 6408-6413.
19	HOU L S, JIN X F, ZHAO Z S. Time series similarity mea-sure via siamese convolutional neural network[C]//Proc. of the 12th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics, 2019.
20	SHEN L L , XIA Y , LI Y P , et al. A multiscale siamese convolutional neural network with cross-channel fusion for motor imagery decoding[J]. Journal of Neuroscience Methods, 2022, 367, 109426. doi: 10.1016/j.jneumeth.2021.109426

名称	尺寸	步进	输出维度
Conv_1	1×202	1	1×3 800×64
Pool_1	1×200	3	1×1 201×64
Conv_2	1×201	1	1×1 001×128
Pool_2	1×201	2	1×401×128
Full	-	-	200

孪生网络	子网络	实验结果	结果分析	是否适用
基本SNN	FC	损失值逐渐趋近于0;数据特征按照模型呈聚集状分布; 训练时间较长	网络结构中参数较多, 导致训练速度较慢, 迭代次数较多	是
SCNN	CNN	损失值逐渐趋近于0;数据特征按照模型呈聚集状分布; 训练时间较短	训练速度相对较快, 调参方便	是
SLSTM	LSTM	损失值下降到一定程度后不再下降; 相同输入条件下数据特征分布较为接近	具有记忆功能, 无法直接处理超长序列, 会导致训练时间过长、梯度消失等问题	否
SAE	AE	损失值几乎不收敛; 相同输入条件下数据特征分布较为接近	对于类内差异性较大的数据, 预训练过程容易导致过拟合	否

样本组别	输入条件	所属模型	标签
1~10	条件1	模型1	1
11~20	条件1	模型2	2
21~30	条件1	模型3	3
31~40	条件1	模型4	4
41~50	条件2	模型1	1
51~60	条件2	模型2	2
61~70	条件2	模型3	3
71~80	条件2	模型4	4
81~90	条件3	模型1	1
91~100	条件3	模型2	2
101~110	条件3	模型3	3
111~120	条件3	模型4	4
121	条件1	参考	-
122	条件2	参考	-
123	条件3	参考	-

[1]	王湖升, 陈伯孝, 叶倾知. 基于箔条干扰实测数据的对抗方法研究[J]. 系统工程与电子技术, 2023, 45(7): 2010-2021.
[2]	马泽煊, 李进, 路艳丽, 陈晨. 融合WaveNet和BiGRU的网络入侵检测方法[J]. 系统工程与电子技术, 2022, 44(8): 2652-2660.
[3]	李郝亮, 陈思伟, 王雪松. 海面角反射器的极化旋转域特性研究[J]. 系统工程与电子技术, 2022, 44(7): 2065-2073.
[4]	朱峰, 蒋倩倩, 林川, 杨啸. 基于支持向量机的典型宽带电磁干扰源识别[J]. 系统工程与电子技术, 2021, 43(9): 2400-2406.
[5]	王力, 刘子奇. WPA-IGA-BP神经网络的模拟电路故障诊断[J]. 系统工程与电子技术, 2021, 43(4): 1133-1143.
[6]	陈浩, 杨俊安, 刘辉. 基于深度残差适配网络的通信辐射源个体识别[J]. 系统工程与电子技术, 2021, 43(3): 603-609.
[7]	张国令, 吴崇明, 李睿, 来杰, 向前. 基于一维堆叠池化融合卷积自编码器的HRRP目标识别方法[J]. 系统工程与电子技术, 2021, 43(12): 3533-3541.
[8]	赵若辰, 王敬东, 林思玉, 顾东泽. 基于卷积神经网络的小型建筑物检测算法[J]. 系统工程与电子技术, 2021, 43(11): 3098-3106.
[9]	张天骐, 范聪聪, 喻盛琪, 赵健根. 基于JADE与特征提取的正交/非正交空时分组码盲识别[J]. 系统工程与电子技术, 2020, 42(4): 933-939.
[10]	弋佳东, 杨洁. 基于IFOA-SA-BP神经网络的雷达信号识别[J]. 系统工程与电子技术, 2020, 42(12): 2735-2741.
[11]	李波, 张琳, 汪文峰, 王冠男. 基于Park-HHT的故障特征提取方法[J]. 系统工程与电子技术, 2020, 42(12): 2944-2952.
[12]	方章闻, 张金艺, 李科, 姜玉稀. 小样本条件下的通信辐射源半监督特征提取[J]. 系统工程与电子技术, 2020, 42(10): 2381-2389.
[13]	刘赢, 田润澜, 王晓峰. 基于深层卷积神经网络和双谱特征的雷达信号识别方法[J]. 系统工程与电子技术, 2019, 41(9): 1998-2005.
[14]	刘立芳, 杨海霞, 齐小刚. 基于线性判别分析的时频域特征提取算法[J]. 系统工程与电子技术, 2019, 41(10): 2184-2190.
[15]	王念滨, 何鸣, 王红滨, 郎泽宇. 基于卷积神经网络的水下目标特征提取方法[J]. 系统工程与电子技术, 2018, 40(6): 1197-1203.