Application of adaptive strategy for supply chain agent

Abstract

With the tremendous increase in the globalization of trade the corresponding supply chains supporting the manufacture, distribution and supply of goods has become extremely complex. Intelligent agents can help with the problem of effective management of these complex supply chains. In this paper we introduce the design, implementation and testing of an intelligent agent for handling procurement, customer sales, and scheduling of production in a stylized supply chain environment. The supply chain environment used in this paper is modeled after the trading agent competition that is held annually to choose the best agent for managing a supply chain. Our supply chain agent, which we call SCMaster, uses dynamic inventory control and various reinforcement learning techniques like Q-learning, Softmax, ε-greedy, and sliding window protocol to make our agent adapt dynamically to the changing environment created by competing agents. A multi-agent simulation environment is developed in Java to test the efficacy of our agent design. Two competing agents are created modeled after the winners of past trading agent competitions and are tested against our agent in various experimental designs. Results of simulations show that our agent has better performance compared to the other agents.

Appendices

Appendix 1: Meta-learning algorithm

Appendix 2: Procurement RFQ creation algorithm

Appendix 3: Component catalog

Components	Base price	Supplier	Description
100	1000	Pintel	Pintel CPU, 2.0 GHz
101	1500	Pintel	Pintel CPU, 5.0 GHz
110	1000	IMD	IMD CPU, 2.0 GHz
111	1500	IMD	IMD CPU, 5.0 GHz
200	250	Basus, Macrostar	Pintel motherboard
210	250	Basus, Macrostar	IMD motherboard
300	100	MEC, Queenmax	Memory, 1 GB
301	200	MEC, Queenmax	Memory, 2 GB
400	300	Watergate, Mintor	Hard disk, 300 GB
401	400	Watergate, Mintor	Hard disk, 500 GB

Appendix 4: Bill of materials

SKU	Components	Cycles required for assembly
1	100, 200, 300, 400	4
2	100, 200, 300, 401	5
3	100, 200, 301, 400	5
4	100, 200, 301, 401	6
5	101, 200, 300, 400	5
6	101, 200, 300, 401	6
7	101, 200, 301, 400	6
8	101, 200, 301, 401	7
9	110, 210, 300, 400	4
10	110, 210, 300, 401	5
11	110, 210, 301, 400	5
12	110, 210, 301, 401	6
13	111, 210, 300, 400	5
14	111, 210, 300, 401	6
15	111, 210, 301, 400	6
16	111, 210, 301, 401	7

Appendix 5: Parameters for the simulation

Parameter	Simulation setting
Length of simulation	250 days
Agent assembly cell capacity	2000 cycles/day
Nominal capacity of supplier assembly lines	550 components/day
Acceptable purchase ratio for single-source suppliers	0.75
Acceptable purchase ratio for two-source suppliers	0.45
Average number of customer RFQs per product on a day	13
Average number of demand per product on a day	200
Range of lead time (due date) for customer RFQs	3–12 days from the day the RFQ is received
Range of penalties for customer RFQs	10% of the customer reserve price annually
Customer reserve price	75–125% of nominal price of the PC components
Annual storage cost rate	37.5% of nominal price of components
The number of RFQs to a supplier per component	5

Appendix 6: Bidding performance analysis

SKU (product type)	SCMaster		Agent-D		Agent-T
	Avg. order price	Difference with Agent-D	Avg. order price	Difference with Agent-T	Avg. order price	Difference with SCMaster
1	979.56	26.30	953.27	40.17	913.09	− 66.47
2	1020.98	190.98	830.00	− 129.53	959.53	− 61.45
3	1017.12	31.34	985.78	26.37	959.41	− 57.71
4	1053.77	47.99	1005.79	2.83	1002.96	− 50.81
5	1256.23	374.69	881.54	− 292.55	1174.09	− 82.14
6	1292.17	462.17	830.00	− 390.96	1220.96	− 71.20
7	1284.33	91.14	1193.19	− 29.53	1222.72	− 61.61
8	1323.93	86.22	1237.71	− 27.68	1265.40	− 58.53
9	979.82	23.67	956.16	44.81	911.34	− 68.48
10	1020.46	190.46	830.00	− 130.87	960.87	− 59.59
11	1017.59	32.53	985.06	30.12	954.93	− 62.66
12	1042.88	152.38	890.50	− 101.70	992.20	− 50.68
13	1230.90	80.12	1150.78	− 5.79	1156.57	− 74.33
14	1280.64	78.42	1202.22	− 10.88	1213.09	− 67.54
15	1278.90	77.19	1201.70	− 7.70	1209.40	− 69.50
16	1326.45	85.67	1240.78	− 27.44	1268.23	− 58.22
Order success rate (order/offer)	3472/3566 = 0.97		3885/8632 = 0.45		7660/10,920 = 0.70