Procurement, production, inventory, and sales flow processing interact with each other, but traditional cost management focuses more on post-event accounting and struggles to directly support rolling decision-making. Based on two sets of normal operation transaction records from ERPsim, this study constructs a standardized process cost chain within the ERP process boundary. The study uses a rolling-up quantile LightGBM to generate P10, P50, and P90 demand scenarios and compares the high-buffer moving average rule, the service-constrained expected cost strategy, and the risk screening strategy among 847 shared strategies. Two cross-operation online replays show that the quantile LightGBM's P50 prediction WMAPE is 43.85% and 38.74%, respectively, slightly lower than the moving average. The risk screening strategy does not consistently outperform the high-buffer rule calibrated during development: in MAS1→MAS2, it is inferior in cost, immediate fulfillment rate, and actual record box CVaR90; in MAS2→MAS1, its cost is only 0.30% lower, the difference is not significant, and the tail cost is higher. The results indicate that a slight decrease in prediction error does not automatically translate into operational benefits. When using complex controls, enterprises also need to make judgments based on service buffers, procurement lead times, and tail risks .

Luwen Zhang. Research on Cross-Border Tax Compliance Management and Risk Prevention and Control of Enterprises under the Background of Global Tax Transparency. International Journal of Business Management and Economics and Trade (2026), Vol. 7, Issue 2: 81-91. https://doi.org/10.38007/IJBMET.2026.070209.

[1] Smyth C, Dennehy D, Fosso Wamba S, Scott M, Harfouche A. Artificial intelligence and prescriptive analytics for supply chain resilience: A systematic literature review and research agenda[J]. International Journal of Production Research , 2024, 62(23): 8537-8561. DOI: 10.1080/00207543.2024.2341415.

[2] Badakhshan E, Ball P D. Deploying hybrid modeling to support the development of a digital twin for supply chain master planning under disruptions[J]. International Journal of Production Research , 2024, 62(10): 3606-3637. DOI: 10.1080/00207543.2023.2244604.

[3] Guo, X. (2025, April). Research on Financial Trading Algorithms Based on Deep Reinforcement Learning. In 2025 IEEE 3rd International Conference on Control, Electronics and Computer Technology (ICCECT) (pp. 1711-1715). IEEE.

[4] Guo, X. (2025, March). Research on Blockchain-Based Financial AI Algorithm Integration Methods and Systems. In 2025 IEEE International Conference on Electronics, Energy Systems and Power Engineering (EESPE) (pp. 816-821). IEEE.

[5] Yuan, Y. (2026). Research on Memory Management and Dynamic Reasoning Methods for Intelligent Agents Based on Large Language Models.

[6] Zhang, Z. (2026). Research on the Optimization of Payment Risk Dynamic Monitoring Models Driven by High-dimensional Behavioral Features. Advances in Computer and Communication, 7(3).

[7] Xin, H. (2026). Research on Distributed Data Cleaning and Standardized Mapping for Heterogeneous Logistics Data. Advances in Computer and Communication, 7(2).

[8] Ma, X. (2026). Research on Elastic Scaling and Resource Scheduling Strategies for Distributed Systems Facing Traffic Fluctuations.

[9] Wang, N. (2026). Research on Digital Marketing Conversion Improvement Strategies and Predictive Analysis Based on User Behavior Segmentation. Journal of Humanities, Arts and Social Science, 10(7).

[10] Xiaoyu Gu. (2025) Research on Generative AI Psychological Intervention Models Based on Multimodal Emotion Recognition. Advances in Computer and Communication, 6(5), 304-309.

[11] Huijie Pan. Discussion on Low-Latency Computing Strategies in Real-Time Hardware Generation. International Journal of Neural Network (2025), Vol. 4, Issue 1: 57-64.

[12] Zhang, J. (2026). Research on Value Stratification and Precision Marketing of Retail Customers in Banks Based on Clustering Algorithm.

[13] Li, J. (2026). Analysis of Advertising Creativity Generation and User Response Mode Based on AIGC.

[14] Li, J. (2026). Research on the Path and Efficiency of Empowering Accurate Advertising Delivery with Data Infrastructure.

[15] Zhang, Y. (2026). A Framework for LLM-Based Semantic Configuration Understanding and Automated Change Generation in Microservice Orchestration.

[16] Liu, X. (2026). Research on the Application of Artificial Intelligence in Consumer Privacy Data Protection. Procedia Computer Science, 279, 956-965.

[17] Liu, X. (2026). Construction of Consumer Data Privacy Protection System Based On Blockchain Technology. Procedia Computer Science, 282, 2082-2091.

[18] Zhang, Y. (2026). Exploration of Enterprise Big Data Microservice Architecture Based on Domain-Driven Design (DDD). Procedia Computer Science, 282, 1994-2003.

[19] Sun, L. (2026). Resource Scheduling and Elastic Scaling Optimization of Distributed Advertising Systems in Cloud-Native Environments.

[20] Su, J. (2026). Design and Implementation of Android High Reliability Communication Module Based on Hierarchical Architecture.

[21] Chi, C. (2026). Quantitative Evaluation of Tranche-Level Returns and Loss Distributions of Structured Credit Assets under Stress Scenarios.

[22] Wang, Z. (2026). Research on Data Analysis and Quality Prediction of Manufacturing Processes Based on Improved Deep Learning Algorithms. Procedia Computer Science, 281, 1328-1337.

[23] Wang, N. (2026). Research on Integrated Analysis Method of Sales and Operations Data for Management Decision Support.

[24] Han, X. (2026). Research on Manufacturing Deviation Vehicle Performance Transmission Mechanism and Intelligent Compensation Control for Automotive Engineering. Procedia Computer Science, 281, 594-602.

[25] Sun, J. (2026). Automated Feature Engineering and Screening System for Large-Scale Factor Libraries. Procedia Computer Science, 281, 1282-1290.