Analysis and correction of supply chain design through machine learning

Patent metadata

Field	Value
Publication number	US-10832196-B2
Application number	US-201816147311-A
Country	US
Kind code	B2
Filing date	Sep 28, 2018
Priority date	Aug 31, 2018
Publication date	Nov 10, 2020
Grant date	Nov 10, 2020

Abstract

A dynamic supply chain planning system for analysis of historical lead time data that uses machine learning algorithms to forecast future lead times based on historical lead time data, and to divide historical lead time data into clusters based on seasonality and linearity. The machine learning results are further processed to adjust future planned lead times and to identify sources in the supply chain that contribute to large deviations between historical planned lead times and actual lead times.

First claim

The invention claimed is:

1. A dynamic supply chain planning system comprising:
a database comprising historical lead time data related to a supply chain;
a machine learning server in communication with the database, the machine learning server comprising a machine learning service and a data preparation module;
an analytics module in communication with the database; and
a user interface in communication with the analytics module,
wherein:
the machine learning server is in communication with the database;

the data preparation module is in communication with the database to obtain the historical lead time data;
the data preparation module processes the historical lead time data into a processed historical lead time dataset for use by the machine learning service;
the machine learning service comprises at least one of a forecasting module and a clustering module;
the forecasting module is configured to forecast future lead times of the supply chain based on the processed historical lead time data set;
the clustering module is configured to provide a plurality of clusters based on pre-defined features;
the machine learning server uploads the future lead times and the plurality of clusters to the database;
and wherein the analytics module is configured to:
receive from a user via the user interface, tolerance criteria for separation of the historical lead time data into a plurality of tolerance zones based on median absolute differences between actual lead time data and historical planned lead time data; and
separate the historical lead time data into the plurality of tolerance zones;

and wherein in relation to the forecasting module, the analytics module is configured to:
obtain future lead times forecast by the forecasting module from the database; and
replace future planned lead times by the future lead times forecast by the forecasting module in accordance with an action criterion defined by the user for each tolerance zone;

and wherein in relation to the clustering module, the analytics module is configured to:
obtain the plurality of clusters from the database;
separate the plurality of clusters in accordance with the plurality of tolerance zones;
further separate the plurality of clusters according to lead, time identifiers; and
identify within the separated clusters, one or more clusters that have the largest median absolute differences between actual lead times and planned lead times.

2. The system of claim 1 , wherein the forecasting module:
divides the processed historical lead time data into a first portion of data points and a second portion of data points, the first portion consisting of more than 50% of the data points, the second portion consisting of less than 50% of the data points;
trains each of a plurality of machine learning forecasting algorithms on the first portion and subsequently tests each of the plurality of machine learning forecasting algorithms on the second portion;
selects that forecasting machine learning algorithm which has a greatest accuracy in testing the second portion; and
retrains the selected forecasting machine learning algorithm on the processed historical lead time data.

3. The system of claim 2 , wherein the first portion consists of 51%-80% of the data points, and the second portion consists of 20%-49% of the data points, such that the first and second portion total 100% of the data points.

4. The system of claim 3 , wherein the plurality of machine learning algorithms comprise Adaboost Regressor, FB Prophet, linear regression, mean and median.

5. The system of claim 1 wherein the historical lead time data is over a period of two years.

6. The system of claim 1 , wherein the forecasting module provides monthly future lead time data based on a monthly average of daily future lead time data.

7. The system of claim 1 , wherein the clustering module:
separates the processed historical lead time data into groups based on a density of data points and linearity of historical lead times;
performs a cluster analysis based on one or more pre-defined features within each group, with the one or more pre-defined features selected from seasonality and linearity;
prunes each cluster; and
divide each cluster into a plurality of subclusters.

8. The system of claim 1 , wherein the analytics module forms part of a planning system that, prior to an up-to-date change in the historical lead time data:
processes the plurality of tolerance zones;
replaces future planned lead times by the future lead times forecast by the forecasting module in accordance with the action criterion; and
identifies the clusters that have the largest median absolute differences between the actual lead times and the planned lead times.

9. A dynamic supply chain planning system comprising:
a processor operable to execute instructions stored in a memory, the memory including specific instructions for forecasting future lead times of the supply chain based on historical lead time data, and execution of the specific instructions causes the processor to:
receive the historical lead time data from a database;
process the historical lead time data by removing outlier data;
divide the processed historical lead time data into a first portion of data points and a second portion of data points, the first portion consisting of more than 50% of the data points, the second portion consisting of less than 50% of the data points;
train each of a plurality of machine learning forecasting algorithms on the first portion and subsequently test each of the plurality of machine learning forecasting algorithms on the second portion;
select that forecasting machine learning algorithm which has a greatest accuracy in testing the second portion;
retrain the selected forecasting machine learning algorithm on the processed historical lead time data;
use the selected forecasting machine learning algorithm to forecast the future lead times of the supply chain;
upload the future lead times to the database;
receive from a user, tolerance criteria for separation of the historical lead time data into a plurality of tolerance zones based on median absolute differences between actual lead time data and historical planned lead time data;
prior to an up-to-date change in the historical lead time data, separate the historical lead time data into the plurality of tolerance zones; and
prior to any up-to-date change in the historical lead time data, replace future planned lead times by the future lead times in accordance with an action criterion defined by the user for each tolerance zone.

10. The system of claim 9 , wherein:
the first portion consists of 55%-80% of the data points, and the second portion consists of 20%-49% of the data points, such that the first and second portion total 100% of the data points; and
the plurality of machine learning algorithms comprise Adaboost Regressor, FB Prophet, linear regression, mean and median.

11. The system of claim 9 , wherein:
the historical lead time data is over a period of two years; and
the future lead times are based on a monthly average of daily future lead time data.

12. A dynamic supply chain planning system comprising:
a processor operable to execute instructions stored in a memory, the memory including specific instructions for grouping historical lead time data of a supply chain into a plurality of clusters, wherein execution of the specific instructions causes the processor to:
receive the historical lead time data from a database;
process the historical lead time data by removing outlier data;
separate the processed historical lead time data into groups based on a density of da

Assignees

• Kinaxis Inc

Inventors

• Oliveira Almeida Marcio
• Lin Zhen
• Haji Soleimani Behrouz
• Nourashrafeddin Seyednaser
• Bisson-Krol Chantal

Classifications

• G06Q10/06315 Primary

  Needs-based resource requirements planning or analysis · CPC title

• G06N5/022

  Knowledge engineering; Knowledge acquisition · CPC title

• G06F16/906

  Clustering; Classification · CPC title

• G06N5/048

  Fuzzy inferencing · CPC title

• G06F16/285

  Clustering or classification · CPC title