Supply chain resiliency using spatio-temporal feedback

Patent metadata

Field	Value
Publication number	US-12572888-B2
Application number	US-202217578351-A
Country	US
Kind code	B2
Filing date	Jan 18, 2022
Priority date	Jan 18, 2022
Publication date	Mar 10, 2026
Grant date	Mar 10, 2026

Abstract

Spatio-temporal climate forecasts are analyzed and one or more resiliency policies for a supply chain are dynamically generated. The resiliency policy is embedded in a resiliency reasoning graph and a temporal feedback loop is performed based on user feedback regarding the generated resiliency policy and user interaction with the resiliency reasoning graph. One or more machine learning models are updated based on the user feedback and a joint optimization of the machine learning models is re-solved based on the user feedback. The resiliency policy is updated based on the updated machine learning models based on the user feedback and an operation of a supply chain is adjusted based on the updated resiliency policy.

First claim

What is claimed is:

1 . A method comprising: dynamically generating one or more resiliency policies for a supply chain; embedding the resiliency policy in a resiliency reasoning graph; performing a temporal feedback loop based on user feedback regarding the generated resiliency policy and user interaction with the resiliency reasoning graph; updating one or more machine learning models based on the user feedback; re-solving a joint optimization of the one or more machine learning models based on the user feedback; updating the one or more resiliency policies policy based on the updated machine learning models based on the user feedback; and adjusting an operation of a supply chain based on the updated resiliency policy, wherein the adjusting of the operation of the supply chain further comprises physically changing an amount of additional inventory stored in a warehouse.

2 . The method of claim 1 , wherein at least one of the machine learning models is a climate model.

3 . The method of claim 1 , further comprising: building the machine learning models across the supply chain; building counterfactual models across the supply chain based on the machine learning models; performing a multi-objective optimization based on user constraints and the resiliency reasoning graph, wherein the generation of the resiliency policy is based on results of the multi-objective optimization; updating the resiliency reasoning graph, analyzing a first set of factors and updating resiliency reasoning graph creation constraints in response to a determination that the resiliency policy was helpful; and analyzing a second set of factors, generating parameters, and analyzing a root-cause in response to a determination that the resiliency policy was unhelpful.

4 . The method of claim 1 , further comprising: dynamically optimizing one or more inventory management policies by analyzing climatic conditions and disruptive events using climate-aware demand forecasts, climate-aware lead-time predictions, and climate-aware labor shortage predictions; optimizing the one or more resiliency policies that are generated by a resilient inventory planning policy generator and a climate-aware resilient pareto policy generator; and recommending one or more resiliency policies using a multi-objective optimization technique.

5 . The method of claim 1 , further comprising: identifying one or more clusters of nodes in the supply chain based on characteristics of the corresponding node and a position in the supply chain; identifying one or more important nodes in the supply chain for participating in the joint optimization; and enabling communication between node-clusters.

6 . The method of claim 1 , further comprising: identifying a supply chain network and determining node dependencies and possible sequences; starting an estimation local pareto from a most downstream node in the supply chain; identifying, by the corresponding node, an operating region and sharing the identified operating region with an upstream node in the supply chain; identifying one or more next dependent nodes in the supply chain; and repeating the starting and the identifying the next dependent nodes operations until pareto is computed for all nodes of the supply chain.

7 . The method of claim 1 , further comprising: identifying a supply chain network and determining node dependencies and possible sequences; performing sequential optimization to obtain an initial local pareto across all nodes in the supply chain; computing a single performance metric and costs representing performance and cost of all nodes in the supply chain; and generating a global pareto that balances the performance metric and costs in a space of resiliency parameters across the supply chain.

8 . A computer program product, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform operations comprising: dynamically generating one or more resiliency policies for a supply chain; embedding the resiliency policy in a resiliency reasoning graph; performing a temporal feedback loop based on user feedback regarding the generated resiliency policy and user interaction with the resiliency reasoning graph; updating one or more machine learning models based on the user feedback; re-solving a joint optimization of the one or more machine learning models based on the user feedback; updating the one or more resiliency policies policy based on the updated machine learning models based on the user feedback; and adjusting an operation of a supply chain based on the updated resiliency policy, wherein the adjusting of the operation of the supply chain further comprises physically changing an amount of additional inventory stored in a warehouse.

9 . The computer program product of claim 8 , the operations further comprising: building the machine learning models across the supply chain; building counterfactual models across the supply chain based on the machine learning models; performing a multi-objective optimization based on user constraints and the resiliency reasoning graph, wherein the generation of the resiliency policy is based on results of the multi-objective optimization; updating the resiliency reasoning graph, analyzing a first set of factors and updating resiliency reasoning graph creation constraints in response to a determination that the resiliency policy was helpful; and analyzing a second set of factors, generating parameters, and analyzing a root-cause in response to a determination that the resiliency policy was unhelpful.

10 . The computer program product of claim 8 , the operations further comprising: dynamically optimizing one or more inventory management policies by analyzing climatic conditions and disruptive events using climate-aware demand forecasts, climate-aware lead-time predictions, and climate-aware labor shortage predictions; optimizing the one or more resiliency policies that are generated by a resilient inventory planning policy generator and a climate-aware resilient pareto policy generator; and recommending one or more resiliency policies using a multi-objective optimization technique.

11 . The computer program product of claim 8 , the operations further comprising: identifying one or more clusters of nodes in the supply chain based on characteristics of the corresponding node and a position in the supply chain; identifying one or more important nodes in the supply chain for participating in the joint optimization; and enabling communication between node-clusters.

12 . The computer program product of claim 8 , the operations further comprising: identifying a supply chain network and determining node dependencies and possible sequences; starting an estimation local pareto from a most downstream node in the supply chain; identifying, by the corresponding node, an operating region and sharing the identified operating region with an upstream node in the supply chain; identifying one or more next dependent nodes in the supply chain; and repeating the starting and the identifying the next dependent nodes operations until pareto is computed for all nodes of the supply chain.

13 . The computer program product of claim 8 , the operations further comprising: identifying a supply chain network and determining node dependencies and possible sequences; performing sequential optimization to obtain an initial local pareto across all nodes in the supply chain; computing a single performance metric and costs representing performance and cost of all nodes in the supply chain; and generating a global pareto that balances

Assignees

• IBM

Inventors

• Kulkarni Kedar
• MARVANIYA Smitkumar Narotambhai
• Godbole Shantanu R
• Weldemariam Komminist

Classifications

• G06N20/00

  Machine learning · CPC title

• G06Q10/06315

  Needs-based resource requirements planning or analysis · CPC title

• G06Q10/087 Primary

  Inventory or stock management, e.g. order filling, procurement or balancing against orders · CPC title