- 电å舌用于ä¸åŒå“牌è…ä¹³æ ·å“的辨别区分
- å‘布日期ï¼?021-12-29 æµè§ˆæ¬¡æ•°ï¼?217
电åèˆ?/span>用于ä¸åŒå“牌è…ä¹³æ ·å“的辨别区åˆ?/font>
一ã€å®žéªŒç®€è¿?/span>
本实验通过由上海ä¿åœ£å®žä¸šå‘展有é™å…¬å¸æä¾›çš„cTongue电åèˆ?/span>å¯?/span>äº?/span>ç§?/span>è…ä¹³æ ·å“进行气味检测鉴别ã€?/span>
二ã€?/span>Ctongue电åèˆ?/span>介ç»
1.原ç†
电å舌是一ç§ä¸»è¦ç”±äº¤äº’æ•æ„Ÿä¼ 感器阵列ã€ä¿¡å·é‡‡é›†ç”µè·¯ã€åŸºäºŽæ¨¡å¼è¯†åˆ«çš„æ•°æ®å¤„ç†æ–¹æ³•组æˆçš„现代化定性定é‡åˆ†æžæ£€æµ‹ä»ªå™¨ã€‚电åèˆŒåŸºäºŽæƒ°æ€§é‡‘å±žç”µæžæž„æˆç¨³å®šçš„ä¼ æ„Ÿå™¨é˜µåˆ—ï¼Œé€šè¿‡ä¼å®‰ç”µåŒ–å¦è„‰å†²æŠ€æœ¯æ¿€å‘实现原创的组åˆè„‰å†²é©°è±«è°±æ€æƒ³ï¼Œç„¶åŽç»äº¤äº’æ„Ÿåº”è§£æžæŠ€æœ¯æ¥èŽ·å–æµ‹é‡å¯¹è±¡çš„æ•´ä½“ä¿¡æ¯ï¼Œç«‹è¶³äºŽæœ€æ–°çš„电å电路硬件和计算机智能化算法软件,快速ã€å®žæ—¶ã€åœ¨çº¿å®žçŽ°å¯¹äº§å“的整体特å¾è¯„ä»·åŠè‹¥å¹²æˆåˆ†å®šæ€§å®šé‡çš„快速检测与分æžã€?/span>
cTongue电å舌å«ç”?/span>æƒ°æ€§é‡‘å±žä¼ æ„Ÿå™¨ç»„æˆï¼Œæ€§èƒ½ç¨³å®šï¼Œé‡çŽ°æ€§å¥½ï¼Œä½¿ç”¨å¯¿å‘½é•¿ï¼Œæ£€æµ‹ä¿¡æ¯é‡ä¸°å¯Œï¼Œè€Œä¸”清洗简å•,éžå¸¸é€‚åˆäºŽå¤šç±»é£Ÿå“产å“的辨别区分,质é‡è¯„定与真伪辨识ç‰ã€?/span>
2.æ•°æ®åˆ†æžæ–¹æ³•介ç»
2.1 é™ç»´åˆ†æž—有效信æ¯èŽ·å–
é™ç»´åˆ†æžä¸»è¦ä½œç”¨æ˜¯å¯¹æœ‰æ•ˆä¿¡æ¯è¿›è¡Œæ±‡æ€»ï¼ŒæŽ’é™¤æ— ç”¨å†—æ‚ä¿¡æ¯ï¼Œä»Žè€Œå®žçŽ°å¤šç»´åº¦ä¿¡æ¯é«˜æ•ˆç‰¹å¾æå–ï¼?/span>实现低维度数æ®å¯è§†åŒ–ã€?/font>
主è¦ç®—法包å«ï¼?/font>PCAï¼?/font>Principal Component Analysis,主æˆåˆ†åˆ†æžï¼‰ã€?/font>LDAï¼?/font>Linear Discriminant Analysis,线性判别分æžï¼‰ã€LLE (Locally Linear Embedding, 局部线性嵌å…?ã€LE (Laplacian Eigenmapsã€æ‹‰æ™®æ‹‰æ–¯ç‰¹å¾æ˜ å°?/font>)ã€?/font>Isomapï¼?/font>Isometric feature mapping, ç‰è·ç‰¹å¾æ˜ å°„)ã€?/span>T-SNE( t-distributed stochastic neighbor embedding ,tåˆ†å¸ƒéšæœºé‚»åŸŸåµŒå…¥)
2.2分类分林混æ‚ä¿¡æ¯çš„精准有åºåˆ†ç±?/span>
多ç§åˆ†ç±»ç®—法å¯å°†åƒç™¾ç»„原始大数æ®å¿«é€Ÿå‡†ç¡®å»ºç«‹ç±»åˆ«ç‰çº§ã€‚å¯åº”用于:真å‡é‰´åˆ«ã€å“ç§é‰´åˆ«ã€ç”Ÿäº§åœ°æº¯æºã€äº§å“原料鉴别ã€åˆ¶å¤‡å·¥è‰ºé‰´åˆ«ã€å“è´¨ç‰çº§é‰´å®šã€æ„Ÿå®˜è¯„级模拟ã€å·®å¼‚分æžç‰ã€?/font>
主è¦ç®—法包å«ï¼?/font>LDAï¼?/font>Linear Discriminant Analysis,线性判别分æžï¼‰ã€?/font>PLSDAï¼?/font>Partial least squares discrimination analysisï¼Œåæœ€å°äºŒä¹˜æ³•判别)ã€?/font>BPNNï¼?/font>Back Propagation Neural Netï¼?/font>BP神ç»ç½‘络)ã€?/font>SVMï¼?/font>Support Vector Machine,支æŒå‘釿œºï¼‰ã€?/font>KNNï¼?/font>K-NearestNeighbor,最近邻算法ï¼?/font>
2.3å›žå½’åˆ†æž—æ„Ÿå®˜æŒ‡æ ‡çš„å®šé‡é¢„æµ?/span>
多ç§ç®—法选择,æé«˜å®šé‡é¢„测准确度,å¯åº”用äº?/font>对货架期ã€é…æ–¹æµ“åº¦ã€æ„Ÿå®˜æŒ‡æ ‡ç‰çº¿æ€§æŒ‡æ ‡è¿›è¡Œå®šé‡é¢„测ã€?/font>
主è¦ç®—法包å«ï¼?/font>PLSDAï¼?/font>Partial least squares discrimination analysisï¼Œåæœ€å°äºŒä¹˜æ³•判别)ã€?/font>BPNNï¼?/font>Back Propagation Neural Netï¼?/font>BP神ç»ç½‘络)ã€?/font>SVMï¼?/font>Support Vector Machine,支æŒå‘釿œºï¼?/font>
2.4èšç±»åˆ†æž—æœªçŸ¥æ•°æ®æŒ–掘的有效信æ?/span>
èšç±»åˆ†æžæŒ‡å°†ç‰©ç†æˆ–抽象对象的集åˆåˆ†ç»„为由类似的对象组æˆçš„多个类的分æžè¿‡ç¨‹ã€‚它是一ç§é‡è¦çš„人类行为。å¯å¯¹æœªçŸ¥ä¿¡æ¯ä¸å¯»æ‰¾è§„律ã€?/font>应用ï¼?/font>å¼‚å¸¸æ•°æ®æˆ–æ ·å“的排查(排查异味è¯å“)ã€åœ¨å¤§æ‰¹é‡æ•°æ®ä¸æ‰¾å¯»å…±åŒç‚?/font> (对比ä¸åŒäº§åœ°åŽŸæ–™è¯çš„ç›¸ä¼¼æ€§ï¼‰ã€æ•°æ®æŒ–掘未知的信æ¯ï¼ˆç›é€‰è¯å“çš„è´®è—æœŸï¼‰ã€?/span>
主è¦ç®—法包å«ï¼?/font>欧æ°è·ç¦»ï¼?/font>Euclid Distance)ã€é—µå¼è·ç¦»ï¼ˆMinkowski Distance)ã€?/font>马æ°è·ç¦»(Mahalanobis distance)ã€?/font>DBSCANï¼?/font>Density-Based Spatial Clustering of Applications with Noiseã€å¯†åº¦èšç±»ç®—法)ã€?/font>K-Meansï¼?/font>k-means clustering algorithm ã€?/font>kå‡å€¼èšç±»ç®—法)ã€?/font>SOMï¼?/font>Self-organizing feature Mapã€è‡ªç»„ç»‡ç‰¹å¾æ˜ 射网络ï¼?/font>ã€?/font>LVQ (Learning Vector Quantizationã€å¦ä¹ 矢é‡é‡åŒ?/font>)
三ã€å®žéª?/font>
1.æ ·å“ä¿¡æ¯
æ ·å“为ä¸åŒå“牌的è…ä¹³æ ·å“,编å·åˆ†åˆ«ä¸º1ï¼?ï¼?ï¼?ã€?å·ï¼Œæ¯ä¸ªæ ·å“å–相ç‰è´¨é‡ç½®äºŽé¡¶ç©ºè¿›æ ·æ ·å“ç“¶ä¸ï¼Œé™ç½®30分钟,æ¯ä¸ªæ ·å“åš4ä¸ªå¹³è¡Œã€‚å®žéªŒæ ·å“ç¼–å·åŠç¤ºæ„图如下:
æ ·å“ç¼–å·
æ ·å“åç§°
1
王致和大å—è…ä¹?/span>
2
王致和è‡è±†è…
3
王致和白è…ä¹³
4
王致和淡å£é²œé¦™è…ä¹?/span>
5
è€æ‰è‡£çº¢æ²¹è…ä¹?/span>
2.æ ·å“处ç†åŠå®žéªŒå‚æ•?/span>
æ¯ä¸ªæ ·å“ç§°å–相ç‰è´¨é‡ï¼Œç”¨çº¯å‡€æ°´è¿›è¡Œæå–。é‡å–相åŒçš„ä½“ç§¯è¿›è¡Œæ£€æµ‹ã€‚æ ·å“æ£€æµ‹å‰å…ˆè¿›è¡Œä¼ 感器预çƒã€æ¸…æ´—åŠå‚æ•°è®¾ç½®ï¼Œæ ·å“æ£€æµ‹å‚æ•°è§è¡?/font>1ã€?/span>
è¡?/font>1 æ ·å“æ£€æµ‹å‚数表
最大电�/span>
10V
最低电�/span>
-10V
脉冲间隔
100MV
çµæ•åº?/span>
10-4
æ ¹æ®æœ¬æ¬¡å®žéªŒå¯ä»¥å¾—到5个ä¸å?/span>è…ä¹³æ ·å“之间çš?/span>å‘³è§‰å·®å¼‚ã€‚æ ¹æ?/span>PCAæ•°æ®ä¸»æˆåˆ†åˆ†æžå¯ä»¥çœ‹å‡?/span>è…ä¹³5个è…ä¹³æ ·å“å¯ä»¥ç”¨ç”µå舌进行味é“分æžï¼Œå°¤å…¶æ ·å“2与其余è…ä¹³å‘³æ ·å“è·ç¦»è¾ƒè¿œï¼Œå‘³é“差异较å¤?/span>ï¼?/span>说明用ä¿åœ£ç”µå舌是å¯ä»¥å¯¹ä¸åŒè…ä¹³æ ·å“进行区分鉴别çš?/span>。我们通过模型建立和模型预测验è¯?/span>得到基本å¯ä»¥å®Œå…¨åŒºåˆ†æ ·å“的分类模åž?/span>ã€?/span>æ ¹æ®è¯¥åˆ†ç±»æ¨¡åž‹ï¼Œå¯ä»¥å¯¹æœªçŸ?/span>è…ä¹³æ ·å“进行追溯ï¼?/span>或者对ä¸åŒå“牌è…ä¹³æ ·å“进行鉴别ã€?/span>
- 上一篇:谷物å“质分æžç”¨ä»€ä¹ˆä»ªå™¨ï¼Ÿæ·€ç²‰ç³ŠåŒ–度分æžå‘¢ï¼Ÿ
- ä¸‹ä¸€ç¯‡ï¼šæ‰æ¡ƒä»è´¨æž„ç‰¹æ€§çš„ç ”ç©¶
测试方法åº?/b>所在ä½ç½®ï¼šé¦–页 >> 测试方法åº?/a> >> 电å舌用于ä¸åŒå“牌è…ä¹³æ ·å“的辨别区分