Method of Transfer Learning for a Specific Production Process of an Industrial Plant

Abstract

A method of transfer learning for a specific production process of an industrial plant includes providing data templates defining expected data for a production process, and providing plant data, wherein the data templates define groupings for the expected data according to their relation in the industrial plant; determining a process instance and defining a mapping with the plant data; determining historic process data; determining training data using the determined process instance and the determined historic process data, wherein the training data comprises a structured data matrix, wherein columns of the data matrix represent the sensor data that are grouped in accordance with the data template and wherein rows of the data matrix represent timestamps of obtaining the sensor data; providing a pre-trained machine learning model using the determined process instance; and training a new machine learning model using the provided pre-trained model and the determined training data.

What is claimed is: 
     
         1 . A method of transfer learning for a specific production process of an industrial plant, comprising:
 providing a plurality of data templates defining expected data for a production process;   providing plant data of the industrial plant, comprising data points of the specific production process, wherein the data points comprise information about input and output of the specific production process;   wherein the data template defines a grouping for the expected data according to their relation in the industrial plant;   determining a process instance of the specific production process, defining a mapping between the plant data (P) to the expected data of the specific production process;   determining historic process data, being historic sensor data relating to the specific production process using the determined process instance;   determining training data using the determined process instance and the determined historic process data, wherein the training data comprises a structured data matrix, wherein columns of the data matrix represent the sensor data that are grouped in accordance with the data template, and wherein rows of the data matrix represent timestamps of obtaining the sensor data;   providing a pre-trained machine learning model using the determined process instance; and   training a new machine learning model using the provided pre-trained model and the determined training data.   
     
     
         2 . The method of  claim 1 , wherein determining the training data comprises preprocessing the historic process data, thereby standardizing a format of the training data. 
     
     
         3 . The method of  claim 2 , wherein preprocessing the historic process data comprises adapting a sampling frequency to a standardized data matrix format. 
     
     
         4 . The method of  claim 2 , wherein preprocessing the historic process data comprises scaling the historic process data to a 0-1 domain. 
     
     
         5 . The method of  claim 2 , wherein preprocessing the historic process data comprises fusing missing data points of the historic process data from available data points of the historic process data. 
     
     
         6 . The method of  claim 2 , wherein preprocessing the historic process data comprises removing outliers from the historic process data. 
     
     
         7 . The method of  claim 1 , wherein the pre-trained model comprises trained weights, and wherein training the new machine learning model comprises adjusting the trained weights. 
     
     
         8 . The method of  claim 1 , wherein the pre-trained machine learning model comprises at least one layer, and wherein training the new machine learning model comprises:
 categorizing each layer using the determined process instance in one of the categories frozen or non-frozen; and   reusing the frozen layers of the pre-trained machine learning model and retraining the non-frozen layers of the pre-trained machine learning model.   
     
     
         9 . The method of  claim 1 , wherein the pre-trained machine learning model comprises at least one layer, and wherein training the new machine learning model comprises:
 categorizing each layer using the determined process instance in one of the categories frozen or non-frozen; and   applying different learning rates on the at least one layer depending on the determination if the layer is a frozen layer or a non-frozen layer.   
     
     
         10 . The method of  claim 1 , wherein the data points comprise input/output names of the specific production process, and wherein the historic process data is determined using the input/output names. 
     
     
         11 . The method of  claim 1 , wherein training the new machine learning model comprises using the data matrix as input for the new machine learning model to obtain a prediction as output from the new machine learning model.