Method and apparatus for providing predictions of key performance indicators of a complex manufacturing system

Patent metadata

Field	Value
Publication number	US-11775911-B2
Application number	US-202017609852-A
Country	US
Kind code	B2
Filing date	Mar 27, 2020
Priority date	May 9, 2019
Publication date	Oct 3, 2023
Grant date	Oct 3, 2023

Abstract

A method provides predictions of key performance indicators of a product variant of a product family manufactured by a complex manufacturing system in a manufacturing process. The method provides a manufacturing operation model for each manufacturing operation type used to manufacture a product variant of the product family. Via the complex manufacturing system measured contributions to key performance indicators, process context data and process execution data of manufacturing operations, are provided. The model parameters of the provided manufacturing operation models are learned automatically based on collected process context data, collected process execution data, and measured contributions to key performance indicators, to update the manufacturing operation models. An updated production efficiency model combining updated manufacturing models including the updated manufacturing operation models, to calculate the predictions of the key performance indicators, of the product variant, to be manufactured, depending on a product configuration of the respective product variant, is evaluated.

First claim

The invention claimed is:

1. A computer-implemented method for increasing production efficiency of a complex manufacturing system, the method configured for providing predictions of key performance indicators of a product variant of a product family to be manufactured by machines of the complex manufacturing system in a manufacturing process, which comprises the steps of:
providing a processing unit of the complex manufacturing system with a manufacturing operation model for each manufacturing operation type used to manufacture the product variant of the product family;
providing, via the complex manufacturing system, measured contributions to the key performance indicators, process context data and process execution data of manufacturing operations;
learning automatically, via the processing unit, model parameters of manufacturing operation models based on the process context data collected, the process execution data collected, and the measured contributions to the key performance indicators to update the manufacturing operation models, the learning of the model parameters of the manufacturing operation models includes an iterative adaption of the model parameters which uses a neural network, decision trees or a support vector machine;
evaluating an updated production efficiency model combining updated manufacturing models including updated manufacturing operation models to calculate predictions of the key performance indicators of the product variant to be manufactured by the complex manufacturing system depending on a product configuration of a respective product variant, wherein the key performance indicators predicted for different production variants of a same product variant are evaluated to select automatically by the complex manufacturing system, a most efficient production variant; and
manufacturing by the complex manufacturing system a product variant according to a selected production variant.

2. The computer-implemented method according to claim 1 , which further comprises evaluating the key performance indicators, predicted for the different product variants of a same product family, to select the product variant to be manufactured by the complex manufacturing system.

3. The computer-implemented method according to claim 1 , which further comprises collecting the measured contributions to the key performance indicators, the process context data and the process execution data, during the manufacturing processes having been executed to manufacture the product variants of a same product family.

4. The computer-implemented method according to claim 1 , wherein the key performance indicators depend on efficiency function values measured during manufacturing of the product variant.

5. The computer-implemented method according to claim 3 , wherein the process context data collected during the manufacturing process to manufacture the product variant of the same product family comprise:
a sequence and structure of the manufacturing operations performed during a respective manufacturing process;
a configuration of the product variant manufactured by a sequence of the manufacturing operations; and
time context data indicating a time when the manufacturing process of the product variant of the same product family was executed.

6. The computer-implemented method according to claim 1 , wherein the process execution data of the manufacturing operation model contains continuous and/or discrete variables.

7. The computer-implemented method according to claim 1 , which further comprises during the manufacturing operation, providing the specific contributions to the key performance indicators and the process execution data by a manufacturing execution system of the complex manufacturing system.

8. The computer-implemented method according to claim 7 , which further comprises deriving the measured contributions to the key performance indicators, provided by the manufacturing execution system of the complex manufacturing system, from measurement data provided by sensors of the complex manufacturing system.

9. The computer-implemented method according to claim 5 , which further comprises providing the sequence and structure of the manufacturing operations forming part of the process context data via a product lifecycle management system of the complex manufacturing system.

10. The computer-implemented method according to claim 1 , which further comprises adapting the model parameters iteratively by performing a regression procedure.

11. The computer-implemented method according to claim 1 , which further comprises providing the neural network as a feedforward neural network or a recurrent neural network.

12. The computer-implemented method according to claim 1 , wherein the manufacturing process contains at least one manufacturing process cycle providing data combined to adapt the model parameters of the manufacturing operation model, or used to adapt the model parameters of the manufacturing operation model for each manufacturing process cycle.

Assignees

• Siemens Ag

Inventors

• Baer Felix
• Buerger Niklas
• Klos Hans-Henning
• Paulitsch Christoph
• Schnittger Jens

Classifications

• G06Q10/06393 Primary

  Score-carding, benchmarking or key performance indicator [KPI] analysis · CPC title

• G06Q10/0633

  Workflow analysis · CPC title

• G06Q10/06395 Primary

  Quality analysis or management · CPC title

• G06Q50/04

  Manufacturing · CPC title

• Y02P90/30

  Computing systems specially adapted for manufacturing · CPC title