Method and system for regime-based treatment optimization of industrial assets - Patents.com

Abstract

To provide a method and system for processing optimization based on a period of control of various assets of industrial manufacture and processing plant facilities or installations.SOLUTION: A method and system for processing optimization comprises: identifying a period of operation control of an asset, in which the period of operation control is identified by calculating a period-of-control similarity score of each asset from a period-of-control sequence of consolidated industrial data and comparing it with a given threshold; grouping periods of operation control of industrial assets and the like together by conciliating a period of control of a period-of-control database and the period of operation control; identifying an optimization setting by performing processing optimization on each group of industrial assets in order to maximize an output or minimize processing that takes a loss / cost into account and installation constraints; identifying any fundamental cause of abnormality as detected by analyzing and diagnosing the abnormality with an irreconcilable operation period of control; and assuring the reliability of a constituent element by estimating the remaining years of durability of the constituent element in the asset having the irreconcilable operation period of control, thereby preventing a malfunction.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] (CROSS REFERENCE TO RELATED APPLICATIONS) This application claims priority to Indian Patent Publication No. 202021043590 (Patent Document 1), filed on October 7, 2020.

[0002] The present disclosure herein relates generally to the field of industrial data analytics, and more particularly to the analysis and management of multiple industrial assets. Regime The present invention relates to a method and system for process optimization based on [Background technology]

[0003] Performance indicators such as productivity, product quality, energy consumption, availability, maintenance costs, percentage of emergency work, etc. are used to monitor the performance of processes, equipment, and assets associated with manufacturing and processing industries. Today, industries face the challenges of achieving ambitious manufacturing goals, minimizing industrial energy consumption, achieving emission standards, and customizing industrial products while dealing with wide variations in raw material quality. Industrial assets strive to continuously improve their performance indicators by adjusting a few parameters that are known to affect processes and equipment.

[0004] Many in the industry operation The equipment has numerous dynamic behaviors that change depending on the input materials used in the process, operation strategies, equipment wear and tear, and maintenance activities. Operational Regime Examples of such devices are reactors, gas turbines, combustion engines, etc. To perform process optimization and / or predictive maintenance (fault detection, fault diagnosis, and RUL estimation) for these devices, the device is typically monitored, particularly when monitoring a large number of such devices. Operational RegimeFor example, an original equipment manufacturer (OEM) that supplies gas turbines to CCGT (combined cycle gas turbine) industrial facilities provides remote monitoring services to its end users. During remote monitoring, data such as flow, pressure, temperature, vibration, etc. obtained from different locations are collected. operation The equipment data is sent to the Central Remote Monitoring Center, where the Gas Turbine Specialist Department analyzes the data and Operational Regime The present invention identifies, detects process faults, and suggests appropriate measures to the gas turbine operator to prevent the faults and / or improve the performance of the turbine. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Indian Patent Publication No. 202021043590 Summary of the Invention [Problem to be solved by the invention]

[0006] Traditional techniques such as Mahalanobis distance (MD) and principal component analysis (PCA) have been used for industrial data. Regime However, these techniques are not effective when they do not take into account the temporal relationships that exist in industrial data. [Means for solving the problem]

[0007] The embodiments of the present disclosure provide technical improvements as a solution to one or more of the above-mentioned technical problems identified by the inventors in conventional systems. For example, in one embodiment, Regime Abstract: A system and method for process optimization based on a .

[0008] In one aspect, multiple industrial assets Regime The present invention provides a processor-implemented method for process optimization based on a method for optimizing a process for a plurality of industrial assets, the method including one or more steps of receiving a plurality of data from one or more databases of a plurality of industrial assets at a predetermined frequency, and pre-processing the received plurality of data according to one or more predefined data formats, removing redundant data, unifying sampling frequencies, identifying and removing outliers, imputing missing data, and synchronizing and integrating a plurality of variables obtained from the one or more databases.

[0009] The processor-implemented method includes the steps of calculating a plurality of soft sensor data using the preprocessed data for the plurality of industrial assets; integrating the preprocessed plurality of data and the calculated plurality of soft sensor data to obtain integrated data for each of the plurality of industrial assets; and identifying one or more of the plurality of industrial assets from the integrated data using hierarchical clustering on the segmented data. Operational Regime As used herein, hierarchical clustering includes obtaining one or more segmented data from the integrated data and identifying one or more Regime Corresponding to each of the sequences Regime Calculate the similarity score; Regime A pre-specified threshold for the similarity score and the calculated Regime Compare similarity scores to identify one or more of the industrial assets Operational Regime Further, the identified one or more Operational Regime grouping the plurality of industrial assets into one or more groups based on the plurality of soft sensor data, optimizing a plurality of critical performance parameters of each of the plurality of industrial assets using the plurality of soft sensor data, and finally, recommending one or more process optimization settings for the one or more variables of the plurality of industrial assets.

[0010] Moreover, the processor implementation method Regime At least one past event stored in the database Operational RegimeBased on a comparison with one or more Operational Regime At least one discord from Operational Regime and determining, using a plurality of data-driven anomaly detection models, at least one unidentified Operational Regime and detecting the presence of an anomaly in the plurality of assets using the plurality of data-driven anomaly diagnosis models. Operational Regime Identify at least one cause of the abnormality detected in the non-abnormal discordance. Operational Regime Note that the optimization model is retuned based on adaptive learning on . Finally, based on the identified causes of anomalies, the remaining useful life (RUL) of multiple assets is estimated to advise early warning of failure of one or more assets.

[0011] In another aspect, industrial assets Regime The present invention provides a system for process optimization based on a method for optimizing a processing time based on a plurality of variables obtained from one or more databases of an industrial fabrication plant facility. The system includes an input / output interface configured to receive a plurality of data from one or more databases of an industrial fabrication plant facility at a predetermined frequency, at least one memory storing a plurality of instructions, and one or more hardware processors communicatively coupled to the at least one memory, the one or more hardware processors configured to execute the plurality of instructions stored in the at least one memory. The system is further configured to preprocess the received plurality of data for verifying availability of the received plurality of data, removing redundant data, unifying sampling frequencies, identifying and removing outliers, imputing missing data, and synchronizing and integrating a plurality of variables obtained from the one or more databases. The system further includes: calculating a plurality of soft sensor data using the preprocessed data of the plurality of industrial assets; integrating the preprocessed plurality of data and the calculated plurality of soft sensor data to obtain integrated data for each of the plurality of industrial assets; and using hierarchical clustering on the segmented data to obtain an integrated data for one or more of the plurality of industrial assets from the integrated data. Operational Regime As used herein, hierarchical clustering is configured to identify one or more Regime Obtain a sequence and Regime Corresponding to each of the sequences Regime Calculate the regime similarity score (RSS), Regime A pre-specified threshold for the similarity score and the calculated Regime Compare similarity scores to identify one or more of the industrial assets Operational Regime Further, the identified one or more Operational Regime grouping the plurality of industrial assets into one or more groups based on the plurality of soft sensor data, optimizing a plurality of critical performance parameters of each of the plurality of industrial assets using the plurality of soft sensor data, and finally, recommending one or more process optimization settings for the one or more variables of the plurality of industrial assets.

[0012] Moreover, the system Regime At least one past event stored in the database Operational Regime Based on a comparison with one or more Operational Regime At least one discord from Operational Regime Determine whether at least one unidentified anomaly was detected using multiple data-driven anomaly detection models. Operational Regime The method is configured to detect the presence of anomalies in a plurality of assets using a plurality of data-driven anomaly diagnosis models. Operational Regime Identify at least one cause of the abnormality detected in the non-abnormal discordance. Operational Regime Note that the optimization model is retuned based on adaptive learning regarding . The system is configured to estimate remaining useful lives (RULs) of multiple assets based on the identified causes of anomalies to advise early warning of failure of one or more assets.

[0013] In yet another aspect, a non-transitory computer readable medium, when executed by one or more processors on a system, causes the one or more processors to perform a program for controlling a plurality of industrial assets. RegimeThe method includes one or more steps such as receiving a plurality of data from one or more databases of a plurality of industrial assets at a predetermined frequency and pre-processing the received plurality of data according to one or more predefined data formats, removing redundant data, unifying sampling frequencies, identifying and removing outliers, imputing missing data, and synchronizing and integrating a plurality of variables obtained from the one or more databases.

[0014] The processor-implemented method includes calculating a plurality of soft sensor data using the preprocessed data for the plurality of industrial assets; integrating the preprocessed plurality of data and the calculated plurality of soft sensor data to obtain integrated data for each of the plurality of industrial assets; and identifying one or more of the plurality of industrial assets from the integrated data using hierarchical clustering on the segmented data. Operational Regime As used herein, hierarchical clustering includes obtaining one or more segmented data from the integrated data and identifying one or more Regime Corresponding to each of the sequences Regime Calculate the similarity score; Regime A pre-specified threshold for the similarity score and the calculated Regime Compare similarity scores to identify one or more of the industrial assets Operational Regime Further, the identified one or more Operational Regime grouping the plurality of industrial assets into one or more groups based on the plurality of soft sensor data, optimizing a plurality of critical performance parameters of each of the plurality of industrial assets using the plurality of soft sensor data, and finally, recommending one or more process optimization settings for the one or more variables of the plurality of industrial assets.

[0015] Moreover, the processor implementation method Regime At least one past event stored in the database Operational Regime Based on a comparison with one or more Operational RegimeAt least one discord from Operational Regime and determining, using a plurality of data-driven anomaly detection models, at least one unidentified Operational Regime and detecting the presence of an anomaly in the plurality of assets using the plurality of data-driven anomaly diagnosis models. Operational Regime Identify at least one cause of the abnormality detected in the non-abnormal discordance. Operational Regime Note that the optimization model is retuned based on adaptive learning about . Finally, the remaining useful life (RUL) of multiple assets is estimated based on the identified causes of anomalies to advise early warning of failure of one or more assets.

[0016] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.

[0017] The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and, together with the description, explain the disclosed principles. [Brief description of the drawings]

[0018] [Figure 1] 1 illustrates an exemplary system for regime-based process optimization of multiple industrial assets of an industrial fabrication facility or process, according to an embodiment of the present disclosure. [Diagram 2] FIG. 1 illustrates an architecture diagram of a system for regime-based process optimization of an industrial asset, according to an embodiment of the present disclosure. [Diagram 3] FIG. 1 is a functional block diagram illustrating regime-based process optimization of an industrial asset, according to an embodiment of the present disclosure. [Figure 4] 1 is a workflow illustrating anomaly detection and diagnosis, and asset sub-grouping, according to an embodiment of the present disclosure. [Diagram 5] 1 is a flow diagram illustrating a method for regime-based process optimization of multiple industrial assets of an industrial fabrication plant facility and process according to some embodiments of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] Representative embodiments are described with reference to the accompanying drawings. In the figures, one or more leftmost digits of a reference number identify the figure in which the reference number first appears. Whenever convenient, the same reference numbers are used throughout the figures to refer to the same or similar parts. Examples and features of the disclosed principles are described herein, however, modifications, adaptations, and other implementations are possible without departing from the spirit and scope of the disclosed embodiments. It is intended that the following detailed description be considered as merely exemplary, with a true scope and spirit being indicated by the following claims.

[0020] Embodiments herein relate to the management of various assets in industrial fabrication and processing plant facilities or equipment. Regime The present invention provides a method and system for process optimization based on a time-domain model of various assets using temporal hierarchical clustering. Operational Regime Identify the integrated industrial data Regime For each asset from the sequence Regime Calculate the similarity score (RSS) and compare it with a given threshold, Operational Regime In this specification, the identified Operational Regime of Operational Regime Database Operational Regime In harmony with similar Operational Regime Group together industrial assets that have the same performance and cost. Perform process optimization for each group of industrial assets and identify optimal settings to optimize output or minimize process and equipment constraints that take into account losses / costs. Operational Regime Analyze anomalies in and perform diagnostics to identify the root cause of any anomalies detected. Operational Regime Estimate the remaining useful life of components in an asset having non-abnormal anomalies to ensure the reliability of the components and prevent failures. Regime Adaptive learning for active models (which can be physics-based, data-driven, or hybrid) is continuous or on demand, depending on its configuration.

[0021] Referring now to the drawings, and more particularly to FIGS. 1-5, in which like reference characters indicate corresponding features consistently throughout the figures, preferred embodiments are shown and will be described in connection with the following exemplary systems and / or methods.

[0022] FIG. 1 illustrates an example of an industrial fabrication and processing plant facility or equipment, according to an example embodiment. Regime 1 illustrates a block diagram of a system (100) for process optimization based on. Although the disclosure is described with respect to the system (100) being implemented on a server, it may be understood that the system (100) may comprise one or more computing devices (102), such as a laptop computer, a desktop computer, a notebook, a workstation, a cloud-based computing environment, and the like. It will be understood that the system 100 may be accessed through one or more input / output interfaces 104-1, 104-2, ..., 104-N, collectively referred to as I / O interfaces (104). Examples of I / O interfaces (104) may include, but are not limited to, a user interface, a portable computer, a personal digital assistant, a handheld device, a smart phone, a tablet computer, a workstation, and the like. The I / O interfaces (104) are communicatively coupled to the system (100) through a network (106).

[0023] In some embodiments, the network (106) may be a wireless or wired network, or a combination thereof. In some examples, the network (106) may be implemented as a computer network, i.e., one of different types of networks, such as a virtual private network (VPN), an intranet, a local area network (LAN), a wide area network (WAN), the Internet, and the like. The network 106 may be a dedicated network or a shared network that represents an association of different types of networks that communicate with each other using various protocols, for example, Hypertext Transfer Protocol (HTTP), Transmission Control Protocol / Internet Protocol (TCP / IP), and Wireless Application Protocol (WAP). Furthermore, the network (106) may include various network devices, including routers, bridges, servers, computing devices, and storage devices. The network devices within the network (106) may interact with the system (100) through communication links.

[0024] The system (100) supports a variety of connectivity options, such as Bluetooth, USB, ZigBee, and other mobile services. The network environment allows for the connection of the various components of the system (100) using any communication link, including the Internet, a WAN, a MAN, and the like. In one exemplary embodiment, the system (100) is implemented to operate as a standalone device. In another embodiment, the system (100) may be implemented to operate as a loosely coupled device in a smart computing environment. The components and functionality of the system (100) are described in further detail below.

[0025] Referring to Figure 2, multiple industrial assets Regime An exemplary system for process optimization based on is illustrated. It will be appreciated that industrial assets herein refer to static assets such as one or more industrial units of a processing plant facility, entire processing or fabrication plant facilities with serial and / or parallel processing units, as well as mobile assets or equipment such as aircraft engines, automobile engines, and the like. Industrial plant facilities process inputs in the form of raw materials and generate products and by-products, and possibly solid waste, liquid waste, and gaseous emissions. Industrial plant facilities are typically used to generate inputs in an environment. Operated Typically, environmental conditions such as ambient temperature, pressure, and humidity are determined by the operation Herein, we use time series hierarchical clustering to estimate the Operational Regime Identify in real time. Operational Regime and a prediction model for key performance parameters, operation The process optimization is performed on a group-by-group basis using quality constraints, manipulated variables (MVs), and pre-specified threshold limits for MVs specific to the identified dominance times. The optimization results are advised for multiple assets. However, the identified Operational Regime but, Regime Historical data available in the database Operational Regime and analyzing the received real-time data for anomalies / failures, whereby detecting a fault in at least one of the multiple assets, estimating remaining useful life (RUL) and time to failure (TTF) of the multiple assets based on the identified cause of anomaly, and advising early warning of failure of one or more assets.

[0026] Regime The optimization module based on interacts with one or more industrial assets via a communication layer, Operation DataIt receives real time and non real time data from several industrial process databases like base, laboratory database, maintenance database, environmental database etc. Regime The optimization module based on the above preprocesses the factory facility data continuously or on demand depending on its configuration, uses the preprocessed data and the soft sensor to obtain simulated data, combines the simulated data and the preprocessed data to obtain integrated data, and uses the integrated data to perform prediction, Regime It provides services such as identification, classification, detection, diagnosis and prognosis, process optimization, model performance monitoring, and adaptive learning for active models (which can be physics-based, data-driven, or hybrid). Regime The output of the various services is presented to the user through various interfaces that are part of the optimization module based on the.

[0027] In one embodiment, the system (100) is configured to Regime The system (100) comprises at least one memory (102) with instructions, one or more databases (112), one or more input / output interfaces (104), and one or more hardware processors (108) communicatively coupled to the at least one memory (102) for executing the modules (110) herein. Additionally, the system includes a pre-processing module (114), a computation module (116), a data integration module (118), and a Regime The system includes an identification module (120), a grouping module (122), an anomaly detection and diagnosis (ADD) module (124), an RUL estimation module (126), an adaptive learning module (128), a process optimization module (130), and an advisory module (132).

[0028] Referring to Figure 3, a functional block diagram illustrates one or more modules of the system (100). Herein, the input / output interface (104) is configured to receive real-time and non-real-time data from various databases at a predetermined frequency (e.g., once per second, once per minute, once per hour, etc.), where the frequency is configurable by a user. The real-time data includes: Operation Data and environmental data. Operation Data includes temperature, pressure, flow, and vibration recorded by sensors within the industrial asset and obtained from processes and equipment within the equipment of the industrial asset. Operation Data is obtained from a distributed control system (DCS), an OPC server, etc. Operation Data The environmental data may be stored in a database or historian. Environmental data such as ambient temperature, atmospheric pressure, ambient humidity, and rainfall may also be recorded by sensors and stored in the environmental database. Non-real-time data includes data obtained from laboratories and maintenance activities. Laboratory data consists of properties (e.g., chemical composition, size distribution, concentration, density, viscosity, calorific value, microstructural composition, etc.) related to raw materials, products, by-products, solid and liquid wastes, and emissions tested in laboratories. Laboratory data is typically stored in and retrieved from a laboratory information management system (LIMS), a relational database (RDB), or a SQL database. Information related to process and equipment conditions, factory facility operating conditions, maintenance activities performed on industrial equipment, etc. is stored in and retrieved from a maintenance database.

[0029] In a preferred embodiment, the pre-processing module (114) of the system (100) is configured to perform pre-processing of real-time and non-real-time data received from multiple databases of an industrial fabrication facility. Pre-processing involves removing redundant data, unifying sampling frequencies, filtering data, identifying and removing outliers, imputing missing data, synchronizing data by incorporating appropriate lags, and integrating variables from various data sources.

[0030] In a preferred embodiment, the computation module (116) of the system (100) is configured to compute the plurality of soft sensor data using the pre-processed data of the plurality of industrial assets. In some cases, some applications or environments may have access requirements, extreme operation It should be noted that sometimes it is not possible to directly use physical sensors due to constraints such as conditions, disturbances in the environment, sensor limitations, etc. Furthermore, the information collected from physical sensors may not be sufficient to capture the behavior of the asset. In such cases, soft sensors (physics-based soft sensors or data-driven soft sensors) are used to generate additional information about the factory facility. Examples of soft sensors include the temperature in the ignition zone of a furnace, the concentration of products or by-products inside a reactor, etc. The soft sensors (physics-based and data-driven) are stored in a soft sensor database.

[0031] It will be appreciated that industrial data may be recorded for multiple sensors (typically hundreds of sensors). Some of these sensor variables are known as manipulated variables (MV) and can be adjusted / changed by an operator. Some of these variables are known as disturbance variables (DV) and act externally on the industrial asset and cannot be adjusted / changed. Variables other than manipulated and disturbance variables are known as process variables and cannot be adjusted / changed directly but can be affected by changes in MV and DV.

[0032] In a preferred embodiment, the data integration module (118) of the system (100) is configured to integrate the pre-processed data and the calculated soft sensor data to obtain integrated data for each of the industrial assets. Regime Used to identify the

[0033] In a preferred embodiment, the system (100) Regime The identification module (120) uses hierarchical clustering on the segmented data to identify one or more of the plurality of industrial assets from the integrated data. Operational Regime In this specification, Regime The identification module (120) identifies one or more Regime Get a sequence and use one or more histories Regime Corresponding to each of the sequences Regime Calculate the similarity score (RSS). Regime The identification module (120) is Regime Similarity score threshold and calculated Regime Compare similarity scores to identify one or more of the industrial assets Operational Regime may be identified.

[0034] Referring to FIG. 4, the flow chart is as follows: Regime The functions of the identification module (120) are illustrated in the following description. RegimeThe identification module (120) receives as input integrated data containing variables from each industrial asset. The input time series data is divided into overlapping / non-overlapping time segments containing a fixed number of time instances that can be configured by the user. For each time segment, a distance metric between each pair of variables is calculated. The distance can be Dynamic Time Warping, braycurtis, canberra, chebyshev, cityblock, correlation, cosine, dice, euclidean, hamming, jaccard, jensenshannon, kulsinski, matching, minkowski, rogerstanimoto, russellrao, seuclidean, sokalmichener, sokalsneath, squared euclidean, yule. Additionally, hierarchical clustering is used to group similar variables called clusters. Starting with each variable in a separate cluster, at each step we join the two most similar clusters into a new single cluster, and repeat this until convergence, resulting in a hierarchical clustering tree.

[0035] It will be appreciated that a set of clusters for the variables is obtained by cutting the dendrogram perpendicular to the distance axis at a predefined height. After cutting the dendrogram, one or more clusters of variables are obtained, and herein, a unique number is assigned to each variable within each of the one or more clusters, and a segment is generated for each variable. Regime Obtain a sequence for a few segments. Regime A sample sequence is shown in Table 1.

[0036] [Table 1]

[0037] By segment Regime After obtaining the sequence, Regime Calculate the similarity score (RSS) and view the history Regime Similar in the database Regime Check if the sequence is available. Regime Available in sequences and databases Regime The RSS may be calculated using a weighted distance between sequences. The distance can be Euclidean, cityblock, cosine, euclidean, l1, l2, manhattan, braycurtis, canberra, chebyshev, correlation, dice, hamming, jaccard, kulsinski, 'mahalanobis', minkowski, rogerstanimoto, russellrao, seuclidean, sokalmichener, sokalsneath, square, yule. For example, the weighted Euclidean distance is shown in Equation (1).

[0038]

number

[0039] In the above formula, the segment Regime The sequence obtained Regime Similarity and History Regime Existing in the database Regime If any of the sequences is greater than a predefined threshold, then Regime is an existing Operational Regime The time segment will match one of the existing Operational Regime Corresponding to Regime Assign ID. History Regime In the database Regime The sequence is for the historical data segment. Regime This is obtained by applying discrimination. Regime Similar after applying sequence Regime Group together segments that have the sequence and calculate the percentage of sequences in each group that have a percentage higher than a certain threshold. Regime Sequence Regime Add to database Newly identified Regime Retune the prediction model with respect to . The sequences with a percentage lower than the threshold are sent to the anomaly detection and diagnosis module and, if anomalous, are added to the ADD database.

[0040] In a preferred embodiment, the grouping module (122) of the system (100) comprises one or more Operational Regime Identified with respect to Regime The present invention is configured to group a plurality of industrial assets into one or more groups based on the ID. Regime It will be grouped with other industrial assets that have an ID. Regime One or more for any of the sequences Regime If the RSS of the sequence is less than a predefined threshold, a segment is identified. Operational Regime teeth, Operational Regime Historical data available in the database Operational Regime Note that this does not match any of the above. The data segments are then checked for anomalies / failures.

[0041] In a preferred embodiment, the diagnostic module (124) of the system (100) Operational RegimeThe system is configured to detect anomalies in and performs diagnosis to identify the root cause of the failure and the sensors that contribute to the anomaly. It uses statistical deep learning techniques for anomaly detection and diagnosis. Some multivariate statistical techniques are Principal Component Analysis, One-Class Support Vector Machine, Isolation Forest, Local Outlier Factor, Mahalanobis Distance, Elliptic Envelope, and deep learning techniques are LSTM Autoencoder, Dense Autoencoder, and Convolutional Autoencoder.

[0042] In another embodiment, one or more Operational Regime If the RSS of is less than a predefined threshold and the data segment is not reported as anomalous by the diagnostic module (124), the RSS obtained for this segment is Regime Sequence to a new Regime As Regime Add to database.

[0043] In a preferred embodiment, the RUL estimation module (126) of the system (100) is configured to estimate the health of one or more components of an industrial asset to estimate the remaining useful life (RUL) or time to failure (TTF) due to a detected anomaly. If the RUL of a component is predicted in advance, measures such as inspection, maintenance, or replacement can be taken before the component fails. operation Data-driven techniques involving machine learning and deep learning approaches, physics-based techniques, and combinations thereof are used to build the diagnosis module (124) and the RUL estimation module (126).

[0044] In a preferred embodiment, the adaptive learning module (128) of the system (100) detects discordant and non-anomalous Operational RegimeThe method is configured to recalibrate one or more predictive models according to the identified time series segments. Regime but Regime Discordant and non-anomalous if not available in the database and not identified as anomalous Operational Regime of Regime New to the database Regime Note that the new Regime Using data from Regime Recalibrate the forecasting models related to the new Regime The data obtained from the study is used to calibrate physics-based models.

[0045] In a preferred embodiment, the process optimization module (130) of the system (100) is configured to optimize a number of key performance parameters for each group of a number of industrial assets. operation Identify the decision and use the prescribed process, equipment, and operation of industrial assets within the constraints operation and / or achieve maintenance objectives. operation The goal is to maximize output, quality, efficiency, etc., or to minimize energy consumption, emissions, operation cost, maintenance costs, etc. The forecast model, lower and upper bounds on the manipulated variables, and Regime The process optimization module (130) stores the objectives and constraints functions for each of the prediction models, the manipulated variables, and the identified Regime and selecting suitable manipulated variable bounds and constraints for each of the grouped industrial assets obtained from the optimization database, and performing optimization to identify optimized values ​​for the manipulated variables (MVs) for a given set of objectives and constraints.

[0046] It will be appreciated that data-driven models include models built using statistical, machine learning, or deep learning techniques such as variations of regression (multiple linear regression, stepwise regression, forward regression, backward regression, partial least squares regression, principal component regression, Gaussian process regression, polynomial regression, etc.), decision trees and variations thereof (random forests, bagging, boosting, bootstrapping), support vector regression, k-nearest neighbor regression, spline fitting and variations thereof (e.g., multi adaptive regression splines), artificial neural networks and variations thereof (multilayer perceptrons, recurrent neural networks and variations thereof, e.g., long short term memory networks, as well as traditional neural networks), and time series regression models.

[0047] Additionally, data-driven models also include one-class or multi-class classification, scoring, or diagnostic models based on statistics, machine learning, or deep learning, such as principal component analysis, Mahalanobis distance, isolation forest, random forest classifiers, one-class support vector machines, artificial neural networks and variations thereof, elliptic envelope, and autoencoders (e.g., dense auto-encoders, LSTM autoencoders, CNN autoencoders). Data-driven models can be point models (which do not consider temporal relationships between data instances for prediction) or time series models (which consider temporal relationships between data instances for prediction).

[0048] Moreover, data-driven models also include reduced order models or response surface models of physics-based models. The response variables include key process parameters at the processing plant facility and can be one or more of productivity, yield, cycle time, energy consumption, waste generation, emissions, quality parameters, equipment condition, availability, mean time between failures, number of unplanned shutdowns, operating costs, maintenance costs, or weighted combinations of the above indicative of the plant facility, process, and / or equipment condition. Predictions derived from various models can be used by plant operators or engineers to assess the plant facility's performance and efficiency. operation to make informed decisions, keep a check on possible anomalies, classify the condition / health of plant facilities, identify the root cause of detected anomalies, estimate the remaining useful life of various processes or equipment, and achieve desired levels of critical process parameters. operation This helps to optimize the

[0049] In a preferred embodiment, the advisory module (132) of the system (100) is Regime For each group, one or more optimal settings for one or more manipulated variables identified by the process optimization are recommended.

[0050] In another embodiment, the creation of an adaptive learning knowledge base is illustrated. Regime Identification and grouping of industrial assets Regime A first set of data ( Operation Data The first set of data may be stored in multiple databases for each industrial asset (e.g., laboratory data, environmental data, maintenance data, soft sensing data inferred using physics-driven or data-driven soft sensors, etc.) and may be brought to a common processor via a data communications network.

[0051] Referring to Figure 5, the industrial assets RegimeThe present invention relates to a processor-implemented method for optimizing processing based on a process flow diagram, the process flow diagram being associated with one or more industrial assets in an industrial fabrication or processing plant facility. Regime It will be appreciated that the one or more industrial assets are part of an optimization process module based on the above-mentioned. As used herein, one or more industrial assets refers to static assets, such as one or more industrial devices in a processing plant facility, an entire processing or fabrication plant facility with serial and / or parallel processing devices, as well as moving assets or equipment, such as aircraft engines, automobile engines, and the like.

[0052] Initially, in step (502), a plurality of data are received from one or more databases of a plurality of industrial assets at a predetermined frequency. The plurality of data comprises real-time data and non-real-time data. The one or more databases include: Operation Data Includes a base, laboratory database, maintenance database, and environmental database.

[0053] In a preferred embodiment, the next step (504) is to pre-process the received data to verify availability, remove redundant data, unify sampling frequency, filter data, identify and remove outliers, impute missing data, and synchronize by incorporating appropriate lags and integrating variables from one or more databases. It is noted that the models include one or more data-driven models, one or more physics-based models, and one or more hybrid models.

[0054] In a preferred embodiment, in a next step (506), the pre-processed data of the plurality of industrial assets is used to calculate a plurality of soft-sensor data, where the plurality of soft-sensors includes physics-based soft-sensors and data-driven soft-sensors.

[0055] In a preferred embodiment, the next step (508) is integrating the pre-processed data and the calculated soft sensor data to obtain integrated data for each of the industrial assets.

[0056] In a preferred embodiment, the next step (510) is to identify from the integrated data one or more of the plurality of industrial assets using time series hierarchical clustering on the segmented data. Operational Regime Hierarchical clustering identifies one or more Regime obtaining a sequence; and Regime Corresponding to each of the sequences Regime A step of calculating a similarity score and a predefined Regime Similarity score threshold and calculated Regime Compare similarity scores to identify one or more of the industrial assets Operational Regime and identifying the

[0057] In a preferred embodiment, the next step (512) is to Operational Regime Group multiple industrial assets into one or more groups based on:

[0058] In a preferred embodiment, the next step (514) is: Regime At least one past event stored in the database Operational Regime Based on a comparison with one or more identified Operational Regime At least one discord from Operational Regime Determine.

[0059] In a preferred embodiment, the next step (516) is to use multiple data-driven anomaly detection models to discord Operational Regime Detect the presence of one or more anomalies in and use multiple data-driven anomaly diagnosis models to identify inconsistencies across multiple assets Operational Regime Identifying at least one cause of one or more anomalies detected in.

[0060] In a preferred embodiment, the next step (518) is to identify discordant and non-abnormal Operational Regime Recalibrate one or more predictive models according to the identified time series segments. Regime but Regime Discordant and non-anomalous if not available in the database and not identified as anomalous Operational Regime The new Regime As Regime Note that new Regime Using data from Regime Recalibrate the forecasting models related to the new Regime The data obtained from the study is used to calibrate physics-based models.

[0061] In a preferred embodiment, the next step (520) is to estimate the remaining useful life (RUL) based on the identified causes of anomalies to provide early warning of failure of one or more components in one or more assets.

[0062] Identified for a time series segment Regime but Regime If it is not available in the database and is not identified as anomalous, Regime The new Regime As Regime Note that new Regime and use data from the system's adaptive learning model to develop new Regime Recalibrate the forecasting models related to the new Regime The physics-based model is also retuned using data obtained from the. The retuned predictive model is used in optimizing a plurality of key performance parameters for each group of a plurality of industrial assets.

[0063] In a preferred embodiment, the next step (522) is to determine the predictive model, the objectives and constraints, and each Regime Group-specific instrumental variables and their limits are used to estimate the performance of multiple industrial assets. RegimeSeveral important performance parameters are optimized for the group.

[0064] In a preferred embodiment, the next step (524) is to Regime One or more optimal settings of one or more manipulated variables are recommended for each group.

[0065] Identified for one or more industrial assets Operational Regime but, Regime Historical data available in the database Operational Regime It will be appreciated that if the time series data segments do not match any of the above, the time series data segments are analyzed for anomalies / faults. To perform anomaly detection and diagnosis on the time series data segments, the asset's previous normal Regime Utilizing anomaly detection and diagnosis modules available for the time series segments. When a process or equipment fault is detected, diagnosis is performed to identify the root cause of the fault. Furthermore, Operational Regime but Regime If it is not available in the database and is not identified as an anomaly, Operational Regime The new Regime As Regime Add to database. New Regime Using the integrated data from the adaptive learning module, Regime We also develop predictive models related to new Regime The company uses integrated data from the datacenter to calibrate physics-based models relevant to industrial assets.

[0066] The written description describes the subject matter herein to enable any person skilled in the art to make and use the embodiments. The scope of the subject matter embodiments is defined by the claims, and may include other modifications that occur to those skilled in the art. Such other modifications are intended to fall within the scope of the claims if they have similar elements that do not differ from the claim language, or if they include equivalent elements that have insignificant differences from the claim language.

[0067] Embodiments of the present disclosure are directed to a method for performing model-based process optimization and predictive maintenance (fault detection, fault diagnosis, and RUL estimation) on a multitude of monitored industrial assets, both static and moving. Operational Regime and Regime Accordingly, embodiments herein address the unsolved problem of identifying shifts in one or more industrial assets of an industrial fabrication process. Regime The present invention provides a system and method for optimization based on

[0068] It will be understood that the scope of protection extends to such programs and also to computer readable means having messages therein, such computer readable storage means encompassing program code means for implementing one or more steps of the method when the program runs on a server or mobile device, or any suitable programmable device. The hardware device may be any type of device that can be programmed, including any type of computer, such as a server or personal computer, or any combination thereof. The device may also include means that can be hardware means, such as an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a combination of hardware and software means, such as an ASIC and an FPGA, or at least one microprocessor and at least one memory in which a software module is located. Thus, the means may include both hardware means and software means. The embodiments of the methods described herein may be implemented in the form of hardware and software. The device may also include software means. Alternatively, the embodiments may be implemented on different hardware devices, for example using multiple CPUs.

[0069] The embodiments herein may comprise hardware and software elements. The embodiments implemented in software include, but are not limited to, firmware, resident software, microcode, etc. The functions performed by the various modules described herein may be implemented by other modules or in combination with other modules. As intended herein, a computer usable or computer readable medium may be any device that can contain, store, transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, device, or apparatus.

[0070] The illustrated steps are presented to explain the exemplary embodiments shown, and it should be recognized that continuing technological developments will change the manner in which a particular function is accomplished. These examples are presented herein for illustrative purposes, and not for limitation. Moreover, for convenience of description, boundaries of basic functional components have been arbitrarily defined herein. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately accomplished. Alternatives (including equivalents, extensions, variations, deviations, etc. of the embodiments described herein) will be apparent to one of ordinary skill in the relevant art(s) based on the teachings contained herein. Such alternatives are within the scope and spirit of the disclosed embodiments. Additionally, the terms "comprising," "having," "containing," and "including," as well as other similar forms, are intended to be equivalent in meaning and are intended to be open-ended in that the item or items following any one of these terms are not intended to be an exhaustive enumeration of such item or items, nor are they intended to be limited to only the item or items listed. It must also be noted that as used in this specification and the appended claims, the singular forms "a," "an," and "the" include plural references unless the context clearly dictates otherwise.

[0071] Moreover, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer-readable storage medium refers to any type of physical memory that may store information or data readable by a processor. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing one or more processors to perform steps or stages not inconsistent with the embodiments described herein. It should be understood that the term "computer-readable medium" includes tangible items and excludes carrier waves and transient signals, i.e., non-transitory. Examples include random access memory (RAM), read-only memory (ROM), volatile memory, non-volatile memory, hard drives, CD ROM, DVDs, flash drives, disks, and any other known physical storage medium.

Claims

1. 1. A processor-implemented method for advising one or more optimal settings for operation of a plurality of industrial assets, comprising: receiving a plurality of data for the plurality of industrial assets from one or more data sources via one or more hardware processors; pre-processing the received plurality of data according to one or more predefined data formats via the one or more hardware processors; calculating, via the one or more hardware processors, a plurality of soft-sensor data using the pre-processed data of the plurality of industrial assets, the plurality of soft-sensors including physics-based soft-sensors and data-driven soft-sensors; integrating the pre-processed data and the calculated soft-sensor data via the one or more hardware processors by combining the pre-processed data and the soft-sensor data to obtain integrated data for each of the industrial assets; identifying, via the one or more hardware processors, one or more operation regimes of the plurality of industrial assets from the integrated data using temporal hierarchical clustering on one or more segments of the integrated data, the temporal hierarchical clustering comprising: obtaining, as an input, one or more sequences of the one or more operation regimes from the integrated data comprising one or more variables for each of the plurality of industrial assets, the step comprising: Divide the input time series data into multiple fixed number of overlapping or non-overlapping time segments; generating a dendrogram of said one or more variables by grouping said one or more variables; cutting the tree perpendicular to a distance axis at a predefined height to obtain clusters for the one or more variables; by assigning a unique number to each variable within each of the clusters to obtain a unique number for each of the time segments and the one or more sequences based on the tree; calculating a regime similarity score corresponding to each of the one or more regime sequences using a weighted distance between the obtained one or more regime sequences of the time segment and the regime sequences available in the database to check whether a similar regime sequence is available in the database; and comparing the calculated regime similarity scores to a predefined threshold of the regime similarity scores to identify the one or more operating regimes for the plurality of industrial assets. and grouping, via the one or more hardware processors, the plurality of industrial assets into one or more groups based on the identified one or more operation regimes; optimizing, via the one or more hardware processors, a plurality of key performance parameters of each of the groups of the plurality of industrial assets using a plurality of predictive models specific to the identified operating regimes; recommending, via the one or more hardware processors, the one or more optimal settings of one or more manipulated variables for each of the groups of the plurality of industrial assets based on the optimization of the plurality of critical performance parameters; 23. A processor-implemented method comprising:

2. determining, via the one or more hardware processors, at least one discordant operating regime from the one or more identified operating regimes based on a comparison with at least one past operating regime stored in a regime database by determining that a regime similarity of the one or more sequences to an existing regime sequence is less than the predefined threshold; detecting, via the one or more hardware processors, the presence of one or more anomalies in the discordant operating regime using a plurality of data-driven anomaly detection models; identifying, via the one or more hardware processors, at least one cause of the one or more detected anomalies in the discordant operation regimes of the plurality of industrial assets using a plurality of data-driven anomaly diagnostic models; via the one or more hardware processors, estimating a remaining useful life (RUL) based on the identified causes of the anomaly to provide an early warning of failure of one or more components within one or more assets; The processor implementation method of claim 1 , further comprising:

3. The processor-implemented method of claim 2 , further comprising retuning the plurality of predictive models based on an adaptive learning process regarding non-abnormal and discordant operating regimes.

4. 2. The processor-implemented method of claim 1, wherein the multiple data sources comprise a factory automation system, a data historian, a distributed control system, a laboratory information management system, an environmental database, a maintenance database, and an operations database.

5. The pre-processing step includes: removing noise and outliers from the received data via the one or more hardware processors; achieving a uniform sampling frequency for received time series data via said one or more hardware processors; imputing missing values ​​of the received plurality of data via the one or more hardware processors; synchronizing the complemented data from the multiple data sources via the one or more hardware processors; The method of claim 1 , comprising:

6. The processor-implemented method of claim 2 further comprising adding the determined inharmonious operation regime to the regime database.

7. 1. A system for regime-based process optimization of a plurality of industrial assets, comprising: an input / output interface for receiving a plurality of data for the plurality of industrial assets from one or more data sources; one or more hardware processors; a memory in communication with the one or more hardware processors, the one or more hardware processors executing programmed instructions stored in the memory; pre-processing the received plurality of data in accordance with one or more predefined formats of the data; using the pre-processed data for the plurality of industrial assets to calculate a plurality of soft-sensor data for a plurality of soft-sensors, including physics-based soft-sensors and data-driven soft-sensors; integrating the pre-processed data and the calculated soft-sensor data by combining the pre-processed data and the soft-sensor data to obtain integrated data for each of the industrial assets; identifying one or more operating regimes of the plurality of industrial assets from the integrated data using temporal hierarchical clustering on one or more segments of the integrated data; grouping the plurality of industrial assets into one or more groups based on the identified one or more operating regimes; optimizing a plurality of key performance parameters of each of the groups of the plurality of industrial assets using a plurality of predictive models specific to the identified operating regime; recommending one or more optimal settings for one or more manipulated variables for each of the groups of the plurality of industrial assets; With memory configured as wherein the temporal hierarchical clustering comprises: obtaining, as an input, one or more sequences of the one or more operation regimes from the integrated data comprising one or more variables for each of the plurality of industrial assets, the step comprising: Divide the input time series data into multiple fixed number of overlapping or non-overlapping time segments; generating a dendrogram of said one or more variables by grouping said one or more variables; cutting the tree perpendicular to a distance axis at a predefined height to obtain clusters for the one or more variables; assigning a unique number to each variable within each of the clusters to obtain a unique number for each of the time segments and the one or more regime sequences based on the dendrogram; By doing this, steps, calculating a regime similarity score corresponding to each of the one or more sequences using a weighted distance between the obtained one or more sequences of the time segment and the regime sequences available in the database to check whether similar regime sequences are available in the database; and comparing the calculated regime similarity scores to a predefined threshold of the regime similarity scores to identify the one or more operating regimes for the plurality of industrial assets. A system comprising:

8. The one or more hardware processors execute programmed instructions stored in the memory, determining at least one discordant operating regime from the one or more identified operating regimes based on a comparison with at least one past operating regime stored in a regime database by determining that a regime similarity of the one or more sequences to an existing regime sequence is less than the predefined threshold; Detecting the presence of an anomaly in the at least one unidentified operating regime using a plurality of data-driven anomaly detection models; identifying at least one cause of the detected anomaly in the operation regime of the plurality of industrial assets using a plurality of data-driven diagnostic detection models; estimating a remaining useful life (RUL) based on the identified causes of the anomaly to provide an early warning of failure of one or more components within one or more of the assets. The system according to claim 7 , configured to:

Images