System and method for just in time characterization of raw materials

Patent metadata

Field	Value
Publication number	US-12373762-B2
Application number	US-202017753109-A
Country	US
Kind code	B2
Filing date	Aug 20, 2020
Priority date	Aug 20, 2019
Publication date	Jul 29, 2025
Grant date	Jul 29, 2025

Abstract

In an industrial plant, various equipment are used to handle processing of raw materials. Considering complexities involved in the processes and the equipment, constant monitoring is required to obtain desired results. The disclosure herein generally relates to industrial process and equipment monitoring, and, more particularly, to data analysis for Just In Time (JIT) characterization of raw materials in any process industry. The system collects real-time plant data among other inputs, and performs characterization of raw materials being used in the plant. The characterization involves categorizing the raw materials into different classes. The class information is further used to predict performance of the industrial plant, and in turn to generate recommendations for optimization of the industrial plant.

First claim

What is claimed is:

1. A processor implemented method for characterization of materials based on plant data, comprising:
receiving the plant data from an industrial plant as input, via one or more hardware processors;
determining, by processing the plant data via the one or more hardware processors, change in one or more raw materials used in the industrial plant, wherein the change in the one or more raw materials is detected at least at a plant level or an equipment level, wherein detecting for a change in an operating regime or an equipment change or degradation and when the operating regime or the equipment is changed then a user is notified and digital twin services are triggered,
wherein when the operating regime or the equipment is not changed, then the one or more hardware processors identifies the one or more raw materials is changed and identifies the equipment where the one or more raw materials is changed,
wherein the change in the operating regime includes a load change or a change of active pulverizers in thermal power plants, change of product grade manufactured in the industrial plant, and wherein determining the change in the one or more raw materials comprises:
pre-processing the plant data by cleaning and merging the plant data and saving the cleaned and merged plant data to a database, wherein the plant data comprises at least one of a) data collected from industrial plant sensors, and b) soft sensors and synthetic data generated through a plurality of computer simulations, wherein the plant data is used to train a plurality of predictive models of the industrial plant and the generated synthetic data is used for class identification and class extraction purposes;
filtering preprocessed plant data; and
determining the change in one or more of the raw materials based on a change in an observed pattern of the filtered plant data for at least one equipment over one or more successive time periods in real-time, wherein the change in the one or more raw materials is determined in comparison with a list of raw materials used, information on initial states of the raw materials, addition or removal of raw materials, and due to change of a raw material from a first form to a second form based on a result of one or more chemical reactions in the industrial plant and wherein the change in the one or more raw materials is determined when difference between the filtered plant data of the successive time periods is beyond a threshold;

determining at least one class that matches each of the one or more raw materials, using at least one material class identification model, via the one or more hardware processors, wherein the determined at least one class is a newly defined class or is from a set of pre-defined classes, wherein the classes of the one or more raw materials are automatically identified based on one or more types of data and wherein the one or more types of data comprises of real-time operating data, past operating data, material characteristics, maintenance data, design data, ambient conditions data and soft sensed data;
predicting material characteristics for each of the one or more raw materials, via the one or more hardware processors, wherein the material characteristics of a raw material comprise one or more directly measurable characteristics and one or more directly non-measurable characteristics, wherein the one or more directly measurable characteristic s comprise chemical composition, physical composition, physical properties, shape or size of the raw material, form or state of the raw material, and inherent chemical properties and wherein the one or more directly non-measurable characteristics comprising kinetic parameters pertaining to material chemical and physical transformations predicted using one or more sensors in the industrial plant;
quantifying the predicted one or more directly non-measurable characteristics for each of the one or more raw materials, via the one or more hardware processors, by reading operating conditions in real-time and predicting the industrial plant performance in real-time, wherein an industrial plant performance is compared against measured industrial plant performance obtained from sensors in real-time and unknown material characteristics are tuned for matching the predicted industrial plant performance with the measured industrial plant performance with desired accuracy, through an internal optimization loop,
wherein the predicted one or more directly non-measurable characteristics once identified are stored in the database, results in learning the material characteristics and adjusting the unknown material characteristics;

creating one or more permutations and combinations of material signature parameters, wherein the material signature parameters are stored in a memory and are divided into one or more operating regimes;
using a machine learning based clustering on the plant data to obtain clustering results for the one or more permutations and combinations of material signature parameters, wherein the machine learning based clustering is performed for all sets of parameters combinations varying hyperparameters and a best set of clustering results are obtained by comparing a separation index for each of the clustering results, the separation index being a number indicating how the plant data has separated into one or more diverse clusters from a machine learning perspective and a ground truth perspective;
identifying the clusters with a best homogeneity score or index within the obtained clustering results, wherein the identified clusters are homogenous clusters both in terms of the plant data points and the material characteristics are isolated as new material classes, and
wherein the steps of creating the one or more permutations and combinations of the material signature parameters, using the machine learning based clustering on the plant data to obtain the clustering results, obtaining the best set of clustering results by comparing the separation index and identifying the clusters with the best homogeneity score or index are repeated for all clusters having the homogeneity score or index below a threshold of homogeneity score or index, until all the clusters are assigned a material class;

selecting at least one of a plurality of predictive models associated with at least one of the predicted material characteristics and the determined at least one class of the one or more raw materials, via the one or more hardware processors,
wherein the database enables recording and re-use of different types of data and information comprising raw material properties and usage, operating data, processed data, simulated data, models, algorithms, optimization and other decisions, expert knowledge, equipment and maintenance records, environmental conditions and plant information, and
wherein the database is configured to collect, store and utilize data from multiple plants at a time;

detecting transition period of the one or more raw materials in real-time, via the one or more hardware processors,
wherein when a material change detection model identifies the raw material transition is not completed, then the material change detection model passes this information to a different set of transition models, and when the material change detection model identifies a completion of the raw material change and stabilization of the plant, then information is passed on to the selected at least one of the plurality of predictive models,
wherein the transition models are a separate set of material identification, plant predictive models built for handling transition periods between two raw materials, being processed;

predicting performance of the industrial plant, using the selected at least one of the plurality of predictive models, via the one or more hardware processors;
generating at least one recommendation or an alert to optimize performanc

Assignees

• Tata Consultancy Services Ltd

Inventors

• Deodhar Anirudh Makarand
• Baikadi Abhishek
• Nistala Sriharsha
• Kumar Rajan
• Gupta Ashit
• Subramanian Sivakumar
• Runkana Venkataramana
• Pandya Rohan

Classifications

• G06N20/00

  Machine learning · CPC title

• G06Q50/04

  Manufacturing · CPC title

• G06Q10/067

  Enterprise or organisation modelling · CPC title

• G06Q10/04

  Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" (market predictions or forecasting for commercial activities G06Q30/0202) · CPC title

• Y02P90/30

  Computing systems specially adapted for manufacturing · CPC title