System and method for performance and health monitoring to optimize operation of a pulverizer mill

A system for performance and health monitoring to optimize operation of a pulverizer mill, the system comprises:

• an input/output interface;
• one or more hardware processors;
• a memory in communication with the one or more hardware processors, wherein the one or more first hardware processors are configured to execute programmed instructions stored in the one or more first memories, to:
• receive a plurality of data as input data from one or more data sources, wherein the plurality of data comprises:
• a plurality of current and historical measurements by a plurality of sensors installed on the pulverizer mill and upstream-downstream equipment, comprising of flow rates, pressures, temperatures, concentrations and control system set points;
• a plurality of current and historical coal properties data comprising of a hardness index, a calorific value and an ultimate and proximate analysis of coal;
• and
• a maintenance and design data comprising of the pulverizer mill and, historical record of maintenance, dimensional and design specific information of the pulverizer mill;
• preprocess the plurality of data and storing the preprocessed plurality of data in a database, wherein the pre-processing further comprises:
• identification and removal of outliers from the plurality of data;
• imputation of missing data from the plurality of data;
• and
• synchronization and integration of a plurality of variables from the one or more data sources using the residence time of a plurality of units of the pulverizer;
• soft sense in real time a plurality of key performance and health parameters of the pulverizer mill and coal quality parameters using the preprocessed plurality of data and a set of soft sensing models,
• wherein the set of soft sensing models comprises a set of physics-based models for real-time determination of grinding effectiveness preceded by data-driven models, wherein the set of physics-based models solve a set of differential and algebraic equations representing transport, drying and grinding of coal within different zones of the pulverizer mill, and wherein the grinding effectiveness is captured by a set of coal grinding matrices representing a fraction of coal moving from a coarser bin to a next finer bin over a length of time, and the coal grinding matrix represents the grinding effectiveness, and the coal grinding matrix is used for predicting mill current pressure and mill differential pressure, wherein the physics-based models are used for the real-time determination of the grinding effectiveness by optimizing the grinding effectiveness value such that the predicted output comprising the coal size distribution matches real measurement comprising coal size distribution at the inlet and outlet of mill in the plant;
• and
• the data-driven models include machine learning regression, Kalman filter, clustering, or classification;
• post establishing the grinding effectiveness of the pulverizer mill, predict by a plurality of condition monitoring models, a plurality of key operating events of the pulverizer mill using the soft sensed plurality of key performance and health parameters, coal quality parameters and the preprocessed plurality of data, wherein the key performance parameters comprises outgoing coal fineness, internal coal accumulation, outgoing coal flow, mill current, mill differential pressure and coal moisture, wherein the coal quality parameters comprises coal calorific value, coal hardness, coal incoming moisture, wherein the adverse conditions include a high operation cost, an operating regime shift, coal property variations, component degradation, safety risks and environmental risks pertaining to an operation of the pulverizers, monitoring operating regime changes and fuel quality changes, wherein the component degradation includes monitoring indicators of roller, table, and gearbox degradation, such as vibrations and current, wherein the plurality of condition monitoring models use the preprocessed data, maintenance information, mill design and mill specific information, predicted performance and coal quality parameters from the set of soft sensing models, wherein the plurality of condition monitoring models observe trends in coal accumulated which is soft sensed by soft sensing models in each zone of mill over a time and raise an alarm if there is a shift in the coal accumulated trends;
• estimate a current health index and a future health index using a plurality of health index models, wherein the health index provides indication of one or more of suboptimal operation, deterioration of components and possible faults and anomalies, despite variations in upstream-downstream conditions, ambient conditions, coal property variations and operating regime changes and wherein the health index is a function of measured parameters, the soft sensed plurality of key performance and health parameters, coal quality parameters and maintenance and design data of the pulverizer mill, wherein the plurality of health index models are pre-generated models;
• receive real-time sensor data from the one or more data sources and mimicking performance of the pulverizer mill in real time;
• compare the current health index and the future health index with a threshold health index, diagnose the current health and future health of the pulverizer mill, wherein the diagnosis comprises of identifying and localization of faults, a root cause analysis of the faults, and estimating a remaining useful life (RUL) of the pulverizer mill and its components;
• provide a set of operating recommendations to the pulverizer mill to mitigate the adverse conditions in the pulverizer mill based on the plurality of key operating events and the diagnosis;
• recommend corrective actions in terms of the plurality of key performance and health parameters of the pulverizer mill and coal quality parameters;
• provide optimal maintenance and scheduling recommendations using maintenance optimizer models, based on the estimated current health index, the future health index and the estimated remaining useful life of the pulverizer mill;
• and
• calibrate existing models to adapt to a current performance and requirements of the pulverizer mill and the plant.

One or more non-transitory machine-readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause:

• receiving a plurality of data as input data from one or more data sources, wherein the plurality of data comprises:
• a plurality of current and historical measurements by a plurality of sensors installed on the pulverizer mill and upstream-downstream equipment, comprising of flow rates, pressures, temperatures, concentrations and control system set points;
• a plurality of current and historical coal properties data comprising of a hardness index, a calorific value and an ultimate and proximate analysis of coal;
• and
• a maintenance and design data comprising of the pulverizer mill and, historical record of maintenance, dimensional and design specific information of the pulverizer mill;
• preprocessing the plurality of data and storing the preprocessed plurality of data in a database, wherein the pre-processing further comprises:
• identification and removal of outliers from the plurality of data;
• imputation of missing data from the plurality of data;
• and
• synchronization and integration of a plurality of variables from the one or more data sources using the residence time of a plurality of units of the pulverizer;
• soft sensing in real time, via the one or more hardware processors, a plurality of key performance and health parameters of the pulverizer mill and coal quality parameters using the preprocessed plurality of data and a set of soft sensing models,
• wherein the set of soft sensing models comprises a set of physics-based models for real-time determination of grinding effectiveness preceded by data-driven models, wherein the set of physics-based models solve a set of differential and algebraic equations representing transport, drying and grinding of coal within different zones of the pulverizer mill, and wherein the grinding effectiveness is captured by a set of coal grinding matrices representing a fraction of coal moving from a coarser bin to a next finer bin over a length of time, and the coal grinding matrix represents the grinding effectiveness, and the coal grinding matrix is used for predicting mill current pressure and mill differential pressure, wherein the physics-based models are used for the real-time determination of the grinding effectiveness by optimizing the grinding effectiveness value such that the predicted output comprising the coal size distribution matches real measurement comprising coal size distribution at the inlet and outlet of mill in the plant;
• and
• the data-driven models include machine learning regression, Kalman filter, clustering, or classification;
• post establishing the grinding effectiveness of the pulverizer mill, predicting, via the one or more hardware processors, by a plurality of condition monitoring models, a plurality of key operating events of the pulverizer mill using the soft sensed plurality of key performance and health parameters, coal quality parameters and the preprocessed plurality of data, wherein the key performance parameters comprises outgoing coal fineness, internal coal accumulation, outgoing coal flow, mill current, mill differential pressure and coal moisture, wherein the coal quality parameters comprises coal calorific value, coal hardness, coal incoming moisture, wherein the plurality of key operating events are associated with adverse conditions in the pulverizer mill, wherein the adverse conditions include one or more of a high operation cost, an operating regime shift, coal property variations, component degradation, safety risks and environmental risks pertaining to an operation of the pulverizer mill, monitoring operating regime changes and fuel quality changes, wherein the component degradation includes monitoring indicators of roller, table, and gearbox degradation, such as vibrations and current, wherein the plurality of condition monitoring models use the preprocessed data, maintenance information, mill design and mill specific information, predicted performance and coal quality parameters from the set of soft sensing models, wherein the plurality of condition monitoring models observe trends in coal accumulated which is soft sensed by soft sensing models in each zone of mill over a time and raise an alarm if there is a shift in the coal accumulated trends;
• estimating, via the one or more hardware processors, a current health index and a future health index using a plurality of health index models, wherein the health index provides indication of one or more of suboptimal operation, deterioration of components and possible faults and anomalies, despite variations in upstream-downstream conditions, ambient conditions, coal property variations and operating regime changes and wherein the health index is a function of measured parameters, the soft sensed plurality of key performance and health parameters, coal quality parameters and maintenance and design data of the pulverizer mill, wherein the plurality of health index models are pre-generated models;
• receiving, via a digital twin, real-time sensor data from the one or more data sources and mimicking performance of the pulverizer mill in real time;
• comparing, via the one or more hardware processors, the current health index and the future health index with a threshold health index, diagnose the current health and future health of the pulverizer mill, wherein the diagnosis comprises of identifying and localization of faults, a root cause analysis of the faults, and estimating a remaining useful life (RUL) of the pulverizer mill and its components;
• providing, via the one or more hardware processors, a set of operating recommendations to the pulverizer mill to mitigate the adverse conditions in the pulverizer mill based on the plurality of key operating events and the diagnosis;
• recommending, via the one or more hardware processors, corrective actions in terms of the plurality of key performance and health parameters of the pulverizer mill and coal quality parameters;
• providing, via the one or more hardware processors, optimal maintenance and scheduling recommendations using maintenance optimizer models, based on the estimated current health index, the future health index and the estimated remaining useful life of the pulverizer mill;
• and
• calibrating, via the one or more hardware processors existing models to adapt to a current performance and requirements of the pulverizer mill and the plant.