An Adaptive Data-Driven Approach to Solve Real-World Vehicle Routing Problems in Logistics

Table 2

Results of testing on instances for 200 and 400 customers.

Instance	Optimal solution (cost) [45]	Proposed algorithm solution (cost)	From the optimal (%)	Number of vehicles optimal [45]	Number of vehicles algorithm	Comments
c1_2_1	2704.57	2704.57	0	20	20	All constraints met
c2_2_1	1931.44	1983.82	0.5238	6	6	All constraints met
r1_2_1	4784.11	4961.81	1.777	20	23	All constraints met, but a greater number of vehicles used.
r2_2_1	4483.16	3827.98	−6.5518	4	8	All constraints met, but a greater number of vehicles used.
rc1_2_1	3602.8	3606.78	0.0398	20	23	All constraints met, but a greater number of vehicles used.
rc2_2_1	3099.53	3169.49	0.6996	6	8	All constraints met, but a greater number of vehicles used.
c1_4_1	7152.02	7152.29	0.0027	40	40	All constraints met
c2_4_1	4116.05	4109.9	−0.0615	12	15	All constraints met, but a greater number of vehicles used.
r1_4_1	10372.31	10400.66	0.2835	40	40	All constraints met
r2_4_1	9210.15	9456.51	2.675	8	9	All constraints met, but a greater number of vehicles used.
rc1_4_1	8573.96	8580.71	0.0675	36	38	All constraints met, but a greater number of vehicles used.
rc2_4_1	6682.37	6679.99	−0.0238	11	12	All constraints met, but a greater number of vehicles used.