Trans RINA, Vol 157, Part A3, Intl J Maritime Eng, Jul-Sep 2015
Figure 4 shows the vessel anomaly behavior in the speedy scenario. As shown in the figure, the yellow circle denotes the normal behavior, while the red square indicates the anomaly behavior of the vessel. In this scenario, the vessel is moving with the speed of more than the maximum speed of the vessel.
The anomaly behaviour can include many cases; e.g. vessels with random movement in the middle of water, vessels with unexpected stops, vessels with a close track in the middle of water, vessels with very short tracks, vessel tracks with many interactions, vessel tracks with many loops, travel over land, deviations from standard routes, speeding, traffic direction violation, etc. [26, 33].
The result is presented in Table 2. As illustrated, the best accuracy result appears from dividing the raw data into 80% for training and 20% for testing. In Figure 5, we can view the line chart of the experimental results. The blue line with the diamond marker shows the memory test result, whereas the red line with the square marker presents the blind test result.
Table 2. The Experimental Result from four Types of Partition Data
50-50
Memory Test
Blind Test
94.78% 94%
Type of Experiments 60-40 95%
93.46%
70-30 80-20 94.05% 95.19%
94.58% 94.72%
Figure 4. Line chart from experimental result
Figure 3. Scenario of vessel anomaly behavior (speedy route)
A model to describe normality is constructed using a training data set. The testing data set will be compared to the training data set to classify them into two categories: normal or anomaly. In the experiments, we use the holdout method. The holdout method partitions data into two subsets called as the training set and testing set [23, 36]. It will give significantly different results depending on how the training and testing data are distributed.
Here, we perform two types of testing, the memory test and blind test. Memory test is the prediction accuracy on training data set. However, blind test is prediction accuracy on testing data. For the classification process, we select four types of experiments:
(i) 50% of the data is used for training, and 50% of data for testing (50-50).
(ii) 60% of the data is used for training, and 40% of data for testing (60-40).
(iii) 70% of the data is used for training, and 30% of data for testing (70-30).
(iv) 80% of the data is used for training, and 20% of data for testing (80-20).
From the figure we can see that that the lowest percentage appears in the (70-30) point which is 94.05%. The memory test result obtains the best accuracy in the (60-40) point which is 95%. From the blind test result, we can see that that the lowest percentage appears in the (60-40) point which is 93.47%. The blind test result gets the best accuracy in the (70-30) point which is 95.19%.
By performing experiment on the AIS data, we are able to identify the anomaly such as vessels with speed deviating from the normal behavior. Not only speed, here we combined the speed and spatial data to define the vessel anomaly behavior. However, the relation between objects (e.g. distance to closest vessel) has not been considered.
7. CONCLUSION In this
paper, we have explored vessel anomaly
behaviour with visual analysis and vessel tracking data using the BNs approach in the speed scenario.
From our experiment, we found that the BNs method can be used for vessel anomaly detection. As for the holdout method, it is shown that the partitions 60% for training and 40% for testing from the data yield the best result for memory test (95%). For the blind test accuracy, the best
A-150
©2015: The Royal Institution of Naval Architects
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76