Cost optimization of blockchain-enabled supply chain system

Abstract

Blockchain Technology-enabled Supply Chain System (BT-enabled SCS) promises to provide trustworthy transactions, better-managed operations, and traceability. BT-enabled SCS is the system using BT to improve the transparency, security, and process integrity of SC. Moreover, a Pharmaceutical Supply Chain (PSC) is a system of drug delivery processes, operations, and organisations. BT-enabled PSC may enable the system to share medical information between systems, track drugs, monitor PSC safely and transparently, reduce delays and human errors, and improve the system's stability, safety, and security. This thesis aims to design a mathematical cost model for BT-enabled SCS, which may assist some companies that evaluate the costs of BT as the main database in their SC system. The second purpose is to minimize the costs of the designed BT-enabled SCS model through Evolutionary Computation (EC) algorithms (CS/ACO/GA) as optimization techniques. We, therefore, identified the cost components of BT-enabled SCS based on the related literature review. The third objective of the thesis is to estimate the costs of the BT-based PSC model, select Evolutionary Supervised Learning algorithms with minimum prediction errors, assign appropriate weights to all cost model components, and determine the cost components of the BT-based PSC model. This study provides a new PSC mathematical cost model, which includes BT, that can improve the safety, performance, and transparency of medical information sharing in a healthcare system. The fourth purpose of this thesis is to determine the most reliable Evolutionary Supervised Learning algorithm(s) with minimum prediction errors, estimate the costs of the BT-based PSC model under uncertain demand, determine the cost components of the multi-function model, and reveal the importance of each cost component. To generate raw data for the BT-enabled SCS model, the authors revised the Operations Research model and Inventory Management model and applied Python software for data generation. Python software also helps us generate raw data for the BT-based PSC cost model and the multi-function BT-enabled PSC cost model under uncertain demand. To reach these goals, we combined four Supervised Learning algorithms (KNN, DT, SVM, and NB) with two EC algorithms for the BT-based PSC cost model and the multi-function BT-enabled PSC cost model under uncertain demand. We applied ACO and FA algorithms for the BT-based PSC cost model and PSO and HS algorithms for the multi-function BT-enabled PSC cost model under uncertain demand. The authors also used the Feature Weighting approach to assign appropriate weights to all cost model components, revealing their importance. Four performance metrics were used to evaluate the cost model, and the Total Ranking Score (TRS) was used to determine the most reliable predictive algorithms. Comparing CS/ACO/GA algorithms, the best solutions for the BT-enabled SCS cost model are CS and ACO with the higher TRS (scored by MSE, RMSE, and ROC), followed by GA standing in the second step. Our findings show that the ACO-NB and FA-NB algorithms perform better than the other six algorithms in estimating the costs of the BT-based PSC cost model with lower errors, whereas ACO-DT and FA-DT show the worst performance. The findings also indicate that the shortage, holding, and expired medication costs more strongly influence the cost model than other cost components. The results also indicate that the HS-NB and PSO-NB algorithms outperform the other six algorithms in estimating the costs of the multi-function BT-enabled PSC model under uncertain demand with lower errors. The findings also illustrate that the Raw Materials cost has a stronger influence on the multi-function model than other components.

Date	14 Feb 2024
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Thien-My Dao (Supervisor) & Tony Wong (Co-supervisor)