System and method for scenario-driven optimization of sourcing cost

Patent metadata

Field	Value
Publication number	US-12079752-B2
Application number	US-202117524220-A
Country	US
Kind code	B2
Filing date	Nov 11, 2021
Priority date	Nov 11, 2021
Publication date	Sep 3, 2024
Grant date	Sep 3, 2024

Abstract

Sourcing is a practice of locating and selecting entities based on set criteria. Determining most appropriate supplier entities at the lowest cost can develop a competitive advantage. However, existing sourcing techniques lack effective ways to target best suppliers for right part bundling and to efficiently optimize sourcing cost for sustained savings. Present disclosure leverages machine learning and optimization methods with technology data to provide a computationally efficient solution. Relationships between parts and corresponding attributes are obtained for parts bundling and best supplier locations are mapped for selected part bundles. A sourcing cost minimizer is used in conjunction with an iterative process to grow or limit supplier capacity by constraining award amount, bundle mix, or number of supplier entities based on multiple alternative scenarios being defined to systemically minimize cost and manage risks driven by changes in customer demand or manufacturing location.

First claim

What is claimed is:

1. A processor implemented method, comprising:
receiving, via one or more hardware processors, an input request from one or more entities;
querying, via the one or more hardware processors, a parts repository based on the input request to identify a plurality of parts;
applying, via the one or more hardware processors, a Jaccard similarity for each pair of parts from the plurality of parts by calculating a similarity score between each set of parts from the plurality of parts to obtain one or more combinations of a set of similar parts;
identifying, via the one or more hardware processors, based on the similarity score, at least one corresponding supplier from a suppliers repository and tagging the at least one corresponding supplier to each of the one or more combinations of the set of similar parts;
performing, via the one or ore hardware processors, a comparison between the similarity score and a threshold;
identifying, via the one or more hardware processors, based on the comparison, one or more target suppliers from the one or more combinations of the set of similar parts and the at least one supplier;
determining, via the one or more hardware processors, a sourcing cost for each set of similar parts corresponding to the one or more identified target suppliers based on one or more pre-defined scenarios;
optimizing, via the one or more hardware processors, a total sourcing cost derived from the sourcing cost for each set of similar parts based on one or more constraints;
predicting, via the one or more hardware processors, a target cost for subsequent time instances using a Neural Network AutoRegressive with eXogenous input (NNARX) model, wherein the NNARX model comprises of three layers, wherein a first input layer comprises neurons representing historical values of the cost, current and historical values for external driver variables, a final output layer comprises a neuron representing the target cost, and a middle layer comprises neurons configured to compute nodal weights of the NNARX model with a rectified linear activation function, wherein each node of the middle layer is connected to one or more nodes of the input layer and to a node of the output layer, wherein the input request is divided into a training set and a testing set, wherein the NNARX model is used on the testing set to compute predicted cost and a forecast error for the testing data and wherein the NNARX model is used with the forecast error to predict the target cost for the subsequent time instances;
performing, via the one or more hardware processors, a comparison of (i) the target cost predicted for the subsequent time instances and (ii) the optimized sourcing cost for each of the one or more pre-defined scenarios; and
identifying, via the one or more hardware processors, a focal scenario from the one or more pre-defined scenarios based on the comparison, wherein the focal scenario serves as an output; and
iteratively performing, via the one or more hardware processors, until the optimized total sourcing cost of the focal scenario under the target cost with a tolerance value by:
identifying one or more objective improvements;
modifying at least one of (i) cost associated with each part, and (ii) the one or more constraints based on the identified objective improvements; and
obtaining a new optimized cost for the focal scenario based on the modified cost and the one or more constraints.

2. The processor implemented method of claim 1 , wherein the one or more pre-defined scenarios comprise at least one of (i) an overall cost reduction for a specific combinatorial set of parts and one or more corresponding target suppliers, (ii) at least one potential supplier identified for fulfilling parts to a specific entity under pre-defined quantity bounds, (iii) at least one potential supplier identified for fulfilling parts to a specific entity at a given spend opportunity, (iv) a maximum number of potential suppliers identified for delivery across entities, or (v) combinations thereof.

3. The processor implemented method of claim 1 , wherein the one or more constraints comprise at least one of demand, a maximum quantity of parts, a minimum quantity of parts, a maximum spend, and a number of suppliers.

4. A system, comprising:
a memory storing instructions;
one or more communication interfaces; and
one or more hardware processors coupled to the memory via the one or more communication interfaces, wherein the one or more hardware processors are configured by the instructions to:
receive an input request from one or more entities;
query a parts repository based on the input request to identify a plurality of parts;
apply a Jaccard similarity for each pair of parts from the plurality of parts by calculating a similarity score between each set of parts from the plurality of parts to obtain one or more combinations of a set of similar parts;
identify, based on the similarity score, at least one corresponding supplier from a suppliers repository and tag the at least one corresponding supplier to each of the one or more combinations of the set of similar parts;
perform a comparison between the similarity score and a threshold; identify based on the comparison, one or more target suppliers from the one or more combinations of the set of similar parts and the at least one supplier;
determine a sourcing cost for each set of similar parts corresponding to the one or more identified target suppliers based on one or more pre-defined scenarios;
optimize a total sourcing cost derived from the sourcing cost for each set of similar parts based on one or more constraints;
predict a target cost for subsequent time instances using a Neural Network AutoRegressive with eXogenous input (NNARX) model, wherein the NNARX model comprises of three layers, wherein a first input layer comprises neurons representing historical values of the cost, current and historical values for external driver variables, a final output layer comprises a neuron representing the target cost, and a middle layer comprises neurons configured to compute nodal weights of the NNARX model with a rectified linear activation function, wherein each node of the middle layer is connected to one or more nodes of the input layer and to a node of the output layer, wherein the input request is divided into a training set and a testing set, wherein the NNARX model is used on the testing set to compute predicted cost and a forecast error for the testing data and wherein the NNARX model is used with the forecast error to predict the target cost for the subsequent time instances;
perform a comparison of (i) the target cost predicted for the subsequent time instances and (ii) the optimized sourcing cost for each of the one or more pre-defined scenarios; and
identify a focal scenario from the one or more pre-defined scenarios based on the comparison, wherein the focal scenario serves as an output; and
iteratively perform until the optimized total sourcing cost of the focal scenario under the target cost with a tolerance value by:
identifying one or more objective improvements;
modifying at least one of (i) cost associated with each part, and (ii) the one or more constraints based on the identified objective improvements; and
obtaining a new optimized cost for the focal scenario based on the modified cost and the one or more constraints.

5. The system of claim 4 , wherein the one or more pre-defined scenarios comprise at least one of (i) an overall cost reduction for a specific combinatorial set of parts and one or more corresponding target suppliers, (ii) at least one potential supplier identified for fulfilling parts to a specific entity under pre-defined quantity bounds, (iii) at least one potential supplier identified for fulfilling parts to a specific entity at a

Assignees

• Tata Consultancy Services Ltd
• Tata Consultancy Ltd Services

Inventors

• Kulkarni Devadatta Madhukar
• Srinivasan Ramakrishnan Sundaram
• Tew Jeffrey David

Classifications

• G06F16/24578

  using ranking · CPC title

• G06N3/04

  Architecture, e.g. interconnection topology · CPC title

• G06N20/00

  Machine learning · CPC title

• G06N5/02

  Knowledge representation; Symbolic representation · CPC title

• G06N3/09

  Supervised learning · CPC title