Global Journal of Computer Science and Technology, D: Neural & Artificial Intelligence, Volume 22 Issue 1

© 2022 Global Journals Vehicle Routing Problem with Time Window Constrain using KMeans Clustering to Obtain the Closest Customer Table 3: Comparison between algorithms [30] ACOMO PS_KPSO PID NUM_CUST C (CNY) CE CS NV C (CNY) CE CS NV C101_25 25 3481.31 19.73 98.4 5 3138.070823 19.50420315 100 3 C101_50 50 9583.7 53.75 99.2 11 5942.717648 33.43172956 100 5 C101_75 75 20195.27 94.87 99.47 17 10639.70989 54.95890959 100 8 C101_100 100 31202.27 129.87 99.6 23 13561.40714 69.03832969 100 10 Results with other test cases In the Solomon-100 dataset, there are three formats of destination grouping. One is a cluster format C), one is a random format (R) and one is a random- clustered format (RC). These three formats have been used for 25, 50, 75 and 100 customers. So other than C101, there are C201, R211, R201 and RC201. The comparison between the proposed algorithm (ACO+K- Means algorithm) and modified Ant Colony algorithm [30] have been given in Table 4. Table 4: Remaining test set results comparison PS_KACO ACOMO PID NUM_CUST TOT_VEH NV C CE CS NV C CE CS C_201_25 25 25 2 215.543 14.864 100 3 613.81 22.28 100 50 25 2 444.961 19.7345 100 6 1232.8 42.46 100 C_201_75 75 25 3 511.09 26.2824 100 9 2177.34 58.81 100 C_201_100 100 25 3 591.557 27.9907 100 13 2221.71 99.08 100 r_201_25 25 25 2 543.693 21.8306 100 5 946.39 45.17 100 r_201_50 50 25 2 1039.39 32.3543 100 6 1404.82 69.75 100 r_201_75 75 25 3 1368.58 44.4869 100 7 2482.75 98.4 100 r_201_100 100 25 4 1995.19 62.9339 100 13 2931.63 111.02 100 r_211_25 25 25 1 375.432 13.1144 100 2 400 24.11 100 r_211_50 50 25 2 1391.42 39.8279 100 7 600 44.87 100 r_211_75 75 25 2 1199.99 35.7638 100 7 873.41 72.94 100 r_211_100 100 25 3 1867.28 55.0745 100 9 1080.64 84.49 100 r_c201_25 25 25 2 454.046 19.9274 100 4 847.18 31.43 100 r_c201_50 50 25 3 974.703 36.1249 100 8 1554.47 80 100 r_c201_75 75 25 4 1623.5 55.0429 100 10 2186.5 121.39 100 r_c201_100 100 25 4 1927.47 61.4963 100 11 2959.41 139.2 100 The data from Table 4 helps in evaluating the effectiveness of the proposed algorithm. Even with increase in the number of customers, be it clustered, random or both, there is barely any increase in the number of vehicles employed. With an average of 2.625 vehicles per case, this greatly affects the total travel, storage, damage and fuel costs while reducing the carbon footprint by a great extent, ultimately helping not only the economy of the organisation but also trying to improve the environmental condition of the Earth. It can be assumed from the results data that there is a high probability of increase in number of customers. As the number of vehicles employed is less, there is scope of increasing customer reach and maybe there is a chance of increasing the speed of delivery. With the new electronic vehicle usage, there will be even more cuts in the carbon footprint value and better customer coverage. Terms The below terms explain the pareto optimal solution which are explained from figure 3 to figure 24. It explains the co-efficient used in capacity of the vehicles and number of customers. C101(25)- Cost efficient Pareto optimal solution set and its multidimension interpolation for 25 customers C101(50)- Cost efficient Pareto optimal solution set and its multidimension interpolation for 50 customers) Cost efficient Pareto optimal solution set and its multidimension interpolation for 75 customers R101(25)- Fuel Co efficient solution set and its multidimension interpolation for 25 customers R101(50)- Fuel Co efficient solution set and its multidimension interpolation for 55 customers R101(75)- Fuel Co efficient solution set and its multidimension interpolation for 75 customers C_201_50 Global Journal of Computer Science and Technology Volume XXII Issue I Version I 60 ( )D Year 2022 e) f)