客服信息
我们支持 澳洲论文代写 Assignment代写、加拿大论文代写 Assignment代写、新西兰论文代写 Assignment代写、美国论文代写 Assignment代写、英国论文代写 Assignment代写、及其他国家的英语文书润色修改代写方案.论文写作指导服务
唯一联系方式Q微:7878393
热点文章
最新论文更多>>
Result and Discussion on Thermal Imagery
CHAPTER 5 RESULT AND DISCUSSION |
5.1 Introduction
This chapter contains the results of various machine conditions for regional temperature analysis and for extracting ROI from a thermogram as discussed in chapter 3. The image enhancement implemented in MATLAB via textual programming and have been tested for successful results under different machine conditions. The obtained results are at constant room temperature. The faults of rotary machine have been concluded after scatter plot of regional temperature data, temperature variations along scan line of extracting ROI. The accuracy of the proposed method is checked by comparing the obtained results with original image features to enhance image features. In the present work faults at different machine conditions were considered:
- Single point defect on an outer race
- Normal condition
- Machine unbalance by a circular disc having two holes near outer periphery at 180°
5.2.1 Result and Discussion
a) Normal condition
Thermal images are acquired by thermal imager. These images show the temperature distribution of the whole setup. Each pixel of the image shows temperature of that point. Then images processed by GuideIR analyzer to extract temperature values in an excel sheet. After that excel sheet is imported into Matlab and a single row of pixels selected to draw graph pixels vs temperature. In the ROI of the image there are two bearings which show more temperature than other parts because those bearings are in motion. Temperature rise in both bearings is approximately same.
(a) |
(b) |
(c) |
(d) |
Fig. 5.1 Thermal images of various machine conditions :(a) Thermal image at Normal condition,(b) at Unbalance condition, (c) at Defective bearing condition, (d) Region of interest extracted from original image
b) Unbalance Condition
The procedure is same for acquiring image &its processing to draw graphs. In the unbalance condition, there is more horizontal and vertical forces acting on the bearings. These forces produce vibrations in machines. Due to these vibrations the bearing temperature rises more than normal condition as shown in fig. This additional rise in temperature gives information about some abnormality in the system.
Fig. 5.2 Heat signature of bearings at different machine conditions
c) Bearing defect
An outer race bearing defect seeded by EDM machine to analyze the bearing condition. One side bearing replaced with bearing contains a point defect in outer race. When the machine in running condition balls of bearing strikes to point defect at every rotation. Due to impact of balls on outer race the temperature of bearing rises more than defect free bearing as shown in fig.5.2
5.2.2 Classification of various machine conditions:
To monitor the rotary machine conditions, thermal imager is placed in front of the machine in such a way so that it can acquire thermal images of whole machine. A thermal imager mobIR M8 are used to acquire the images. A software guideIR analyzer
is used to connect the IR camera to laptop to record/store the image on the hard disk of
the computer for further processing and analysis. The ROI has taken out with the help of Matlab programming. The ROI image further used to extract histogram features (Skewness, standard deviation, kurtosis, mean, root mean square). The image acquire by ROI is enhanced by 2D DWT module of mat lab by sym4 wavelet at level 2. Fig.1 shows the original thermal image of machine conditions. For the purpose of rotating machine fault diagnosis and reduction of image processing, computational, ROI is chosen from the original image as a rectangle, size125x25 pixels. This size is likewise applied to other images. Then, histogram feature extraction is carried out to describe the total (30x6) 180 feature values extracted from these ROIs. These features are known as original image features.
Table-5.1 Total no of features extracted |
||||||||||
Machine condition |
No of data files |
Dimension of image data |
Size of ROI |
Total data file |
Using features |
Total feature dimensions (in each level) |
||||
Normal |
10 |
|||||||||
Unbalance |
10 |
160x120 |
125x25 |
30 |
6 |
30x6 |
||||
Bearing defect |
10 |
|||||||||
Table-5.2 Histogram features of Infrared image |
|||||||
Kurtosis |
RMS |
Skewness |
STD |
Mean |
Entropy |
||
816 |
1.7782 |
115.1889 |
-0.0641 |
70.5114 |
91.0863 |
0.6613 |
|
817 |
1.7738 |
115.5996 |
-0.0823 |
70.5888 |
91.5455 |
0.6698 |
|
818 |
1.7666 |
114.9731 |
-0.0619 |
70.5128 |
90.8122 |
0.6715 |
|
Normal |
819 |
1.7742 |
114.7758 |
-0.0486 |
70.4895 |
90.5804 |
0.6672 |
820 |
1.7729 |
115.7663 |
-0.0813 |
70.6754 |
91.6892 |
0.6656 |
|
821 |
1.7713 |
115.7967 |
-0.0824 |
70.6979 |
91.7102 |
0.6728 |
|
822 |
1.7689 |
115.9155 |
-0.0817 |
70.7787 |
91.7979 |
0.6726 |
|
823 |
1.769 |
116.019 |
-0.0877 |
70.7761 |
91.9306 |
0.6795 |
|
824 |
1.7675 |
115.2545 |
-0.0737 |
70.5147 |
91.1667 |
0.6699 |
|
825 |
1.7683 |
116.7201 |
-0.1103 |
70.9378 |
92.6903 |
0.6793 |
|
Table-5.3 |
|||||||
Kurtosis |
RMS |
Skewness |
STD |
Mean |
Entropy |
||
933 |
1.7782 |
115.1889 |
-0.0641 |
70.5114 |
91.0863 |
0.6613 |
|
934 |
1.7738 |
115.5996 |
-0.0823 |
70.5888 |
91.5455 |
0.6698 |
|
935 |
1.7666 |
114.9731 |
-0.0619 |
70.5128 |
90.8122 |
0.6715 |
|
Bearing |
936 |
1.7742 |
114.7758 |
-0.0486 |
70.4895 |
90.5804 |
0.6672 |
defect |
937 |
1.7729 |
115.7663 |
-0.0813 |
70.6754 |
91.6892 |
0.6656 |
938 |
1.7713 |
115.7967 |
-0.0824 |
70.6979 |
91.7102 |
0.6728 |
|
939 |
1.7689 |
115.9155 |
-0.0817 |
70.7787 |
91.7979 |
0.6726 |
|
940 |
1.769 |
116.019 |
-0.0877 |
70.7761 |
91.9306 |
0.6795 |
|
941 |
1.7675 |
115.2545 |
-0.0737 |
70.5147 |
91.1667 |
0.6699 |
|
942 |
1.7683 |
116.7201 |
-0.1103 |
70.9378 |
92.6903 |
0.6793 |
|
Table-5.4 |
|||||||
Kurtosis |
RMS |
Skewness |
STD |
Mean |
Entropy |
||
837 |
1.7845 |
115.4502 |
-0.0699 |
70.6607 |
91.3011 |
0.6775 |
|
838 |
1.7726 |
115.6732 |
-0.0803 |
70.7435 |
91.519 |
0.6719 |
|
839 |
1.7814 |
116.4828 |
-0.1008 |
70.8823 |
92.4339 |
0.6738 |
|
unbalance |
840 |
1.7814 |
114.3545 |
-0.0369 |
70.4288 |
90.0934 |
0.6602 |
841 |
1.783 |
114.691 |
-0.0441 |
70.5222 |
90.4474 |
0.6668 |
|
842 |
1.7775 |
115.665 |
-0.0706 |
70.7808 |
91.4799 |
0.6697 |
|
843 |
1.7934 |
114.6844 |
-0.0375 |
70.5195 |
90.4412 |
0.6678 |
|
844 |
1.7877 |
115.7136 |
-0.0674 |
70.7685 |
91.5508 |
0.6765 |
|
845 |
1.7759 |
115.5185 |
-0.0592 |
70.8346 |
91.2528 |
0.6715 |
|
846 |
1.7749 |
115.3965 |
-0.0613 |
70.7544 |
91.1605 |
0.6713 |
|
In order to observe the original image feature distribution, three features can be arbitrarily selected from the feature set.
Table-5.5 Original image features |
||||
Image no. |
Kurtosis |
Root mean square |
Standard deviation |
|
816 |
1.7782 |
115.1889 |
70.5114 |
|
817 |
1.7738 |
115.5996 |
70.5888 |
|
818 |
1.7666 |
114.9731 |
70.5128 |
|
819 |
1.7742 |
114.7758 |
70.4895 |
|
Normal |
820 |
1.7729 |
115.7663 |
70.6754 |
821 |
1.7713 |
115.7967 |
70.6979 |
|
822 |
1.7689 |
115.9155 |
70.7787 |
|
823 |
1.769 |
116.019 |
70.7761 |
|
824 |
1.7675 |
115.2545 |
70.5147 |
|
825 |
1.7683 |
116.7201 |
70.9378 |
|
Table-5.6 |
||||
Image no. |
Kurtosis |
Root mean square |
Standard deviation |
|
837 |
1.7845 |
115.4502 |
70.6607 |
|
838 |
1.7726 |
115.6732 |
70.7435 |
|
839 |
1.7814 |
116.4828 |
70.8823 |
|
840 |
1.7814 |
114.3545 |
70.4288 |
|
Unbalance |
841 |
1.783 |
114.691 |
70.5222 |
842 |
1.7775 |
115.665 |
70.7808 |
|
843 |
1.7934 |
114.6844 |
70.5195 |
|
844 |
1.7877 |
115.7136 |
70.7685 |
|
845 |
1.7759 |
115.5185 |
70.8346 |
|
846 |
1.7749 |
115.3965 |
70.7544 |
|
Table-5.7 |
||||
Image no. |
Kurtosis |
Root mean square |
Standard deviation |
|
933 |
1.7996 |
113.485 |
69.7871 |
|
934 |
1.7979 |
116.1042 |
70.4447 |
|
935 |
1.7983 |
116.195 |
70.4748 |
|
936 |
1.7945 |
116.4671 |
70.5499 |
|
Bearing defect |
937 |
1.799 |
117.3067 |
70.6926 |
938 |
1.7938 |
116.8189 |
70.6715 |
|
939 |
1.7875 |
115.6822 |
70.4866 |
|
940 |
1.7918 |
116.33 |
70.5077 |
|
941 |
1.7924 |
116.4579 |
70.5408 |
|
942 |
1.7883 |
117.8719 |
70.9786 |
|
Fig.-5.3 Graph of original image features
The fig. 5.3 only gives the information to understand how the features distribute in same machine condition and how the clusters of the features separate in the different conditions. Fig.5.3 shows the distribution of the three- first features involving kurtosis, standard deviation (STD), and the root mean square (RMS) of the original images. It can be seen that the features of machine conditions are not well clustered and overlapped with each other. To increase the separation among the feature clusters, it is necessary to apply an enhancement method to ameliorate the image quality.
The enhancement method based on 2D wavelet symlet 4 wavelet is applied to ameliorate the image quality. The main advantage of this method is that no artificial information is introduced into the enhanced image. In the decomposition of thermal image data from different machine conditions, we apply symlet wavelets of degree 4 and the decomposition level of 2. The wavelet transform has been used as a good image
reparability, and compaction and sparsely features in addition to statistical properties. Having performed decomposition, four kinds of wavelet coefficients approximation coefficients (A), Horizontal details (HD), vertical details (VD), diagonal details (DD), have found from each class of machine conditions data. Coefficients passed through the low pass filter is considered for feature extraction because the low frequency signals contain most important details of the original image. Then histogram features were extracted from images enhanced by sym4 wavelet.
Table-5.8 Enhanced image features |
||||
Image no. |
Kurtosis |
Root mean square |
Standard deviation |
|
816 |
86.4521 |
5.9581 |
5.9419 |
|
817 |
84.5492 |
5.9618 |
5.9386 |
|
818 |
86.4491 |
5.948 |
5.9188 |
|
819 |
86.4491 |
5.948 |
5.9188 |
|
Normal |
820 |
88.018 |
5.9535 |
5.9264 |
821 |
85.2876 |
5.9173 |
5.8861 |
|
822 |
87.8076 |
5.9777 |
5.9479 |
|
823 |
84.8836 |
5.9598 |
5.9233 |
|
824 |
87.8627 |
5.8783 |
5.8421 |
|
825 |
86.0576 |
5.9261 |
5.8903 |
|
Table-5.9 |
||||
Image no. |
Kurtosis |
Root mean square |
Standard deviation |
|
837 |
88.1372 |
5.1349 |
5.1048 |
|
838 |
83.6757 |
5.1133 |
5.0845 |
|
839 |
92.029 |
5.0921 |
5.073 |
|
840 |
90.6656 |
5.2643 |
5.2334 |
|
Unbalance |
841 |
91.3633 |
5.262 |
5.2266 |
842 |
96.429 |
5.1555 |
5.1232 |
|
843 |
87.8128 |
5.2826 |
5.254 |
|
844 |
92.8662 |
5.1222 |
5.0897 |
|
845 |
91.9623 |
5.2015 |
5.1607 |
|
846 |
87.6445 |
5.1467 |
5.107 |
|
Table-5.10 |
||||
Image no. |
Kurtosis |
Root mean square |
Standard deviation |
|
933 |
1.7996 |
113.485 |
69.7871 |
|
934 |
1.7979 |
116.1042 |
70.4447 |
|
935 |
1.7983 |
116.195 |
70.4748 |
|
936 |
1.7945 |
116.4671 |
70.5499 |
|
Bearing defect |
937 |
1.799 |
117.3067 |
70.6926 |
938 |
1.7938 |
116.8189 |
70.6715 |
|
939 |
1.7875 |
115.6822 |
70.4866 |
|
940 |
1.7918 |
116.33 |
70.5077 |
|
941 |
1.7924 |
116.4579 |
70.5408 |
|
942 |
1.7883 |
117.8719 |
70.9786 |
|
(a)
(b)
Fig.-5.4 Images feature distribution graph: (a) Original image features, (b) Enhanced image features
Evidently, after enhancing, the features are well separated into groups which are similar in characteristics and there is no overlap between the machine conditions. This shows that the proposed enhancement method has assisted in improving the image quality.
1
上一篇:The Negative Impacts Of Information Technology On Society 下一篇:Sealed Bid Auction via Active Network