A method and an assembly for controlling an internal combustion engine having multiple burners is provided. Combustion measurement data is collected in a burner-specific manner for each burner and assigned to a burner identification identifying the respective burner. Performance measurement data of the internal combustion engine is also collected and used to determine a performance value. A machine learning model is trained by means of the combustion measurement data, the associated burner identifications and the performance measurement data, to generate burner-specific control data which optimizes the performance value when the burners are actuated in a burner-specific manner using the control data. The control data generated by the trained machine learning model is output for the burner-specific actuation of the burners.

A sensitivity parameter storing portion stores, as known sensitivity parameters owned by a flame sensor, reference received light quantity, reference pulse width, probability of regular discharge, and probabilities of non-regular discharge in advance. The discharge probability is calculated based on the number of drive pulses applied to the flame sensor and the number of discharges determined to have occurred in the flame sensor having received the drive pulses. The calculated discharge probability and the known sensitivity parameters are used to calculate the received light quantity per unit time received by the flame sensor.

A system for generating a base point database for use by a full authority digital engine control (FADEC) to control a gas turbine engine includes a memory configured to store a model of the gas turbine engine. The system also includes a model generation processor coupled to the memory and designed to perform an initial simulation of the gas turbine engine using the model to determine ranges of sensitivity of desired engine parameters throughout an operating envelope, divide the operating envelope into a multiple regions based on the ranges of sensitivity of the desired engine parameters, select multiple combinations of base points within each of the multiple regions, perform an additional simulation of the gas turbine engine to determine an accuracy of interpolation between each of the multiple combinations of base points for each of the multiple regions, and select final base points from the multiple combinations of base points based on the accuracy.

There is provided a method for operating a cement plant capable of simultaneously optimizing both combustion in a calciner and a heat consumption rate. The method for operating a cement plant includes: feeding first fuel to a calciner; feeding second fuel for maintaining the inside at a burning temperature to a cement kiln along with combustion primary air, and introducing air for cooling cement clinker to a cooler; and feeding a part of the air as secondary air to the cement kiln, feeding as tertiary air to the calciner, and discharging the rest of the air from the cooler, wherein relation between a first oxygen concentration at an exhaust gas outlet of the calciner and a heat consumption rate determined by the first fuel and the second fuel, and relation between a second oxygen concentration at an exhaust gas outlet of the preheater and the heat consumption rate are beforehand obtained, and amounts of the secondary air and the tertiary air are adjusted such that both the first oxygen concentration and the second oxygen concentration fall within a range including values of the oxygen concentrations at which the heat consumption rate becomes at its minimum.

A sensitivity parameter storing portion stores, as known sensitivity parameters owned by a flame sensor, reference received light quantity, reference pulse width, probability of regular discharge, and probabilities of non-regular discharge in advance. The discharge probability is calculated based on the number of drive pulses applied to the flame sensor and the number of discharges determined to have occurred in the flame sensor having received the drive pulses. The calculated discharge probability and the known sensitivity parameters are used to calculate the received light quantity per unit time received by the flame sensor.

A method of operating a heat releasing reactor producing product gas. The method includes steps of (a) monitoring a current load of the reactor, (b) finding such a numerical value for a current computational maximum momentary load for which at least one product gas factor computed using currently monitored process data with a numerical model of the reactor fulfills an acceptance condition, and selecting the numerical value as the current computational maximum momentary load, (c) indicating the current computational maximum momentary load to the operator and/or, if the current load is (c1) less than the current computational maximum momentary load, (c1i) indicating the operator that the load may be increased, and/or (c1ii) automatically increasing the load, and/or (c2) greater than the current computational maximum momentary load, (c2i) indicating the operator that the load exceeds the current computational maximum boiler momentary load, and/or (c2ii) automatically reducing the boiler load.

Aspects of the present disclosure relate generally to a system including: a computing device in communication with a combustion system, wherein the computing device is configured to perform actions including: issuing an input to the combustion system; determining whether one of a dynamic output and an emission output corresponding to the input to the combustion system exceeds a first boundary condition; and adjusting the input to the combustion system by one of a first step change and a second step change; wherein the first step change corresponds to the dynamic output and the emission output not exceeding the first boundary condition, and the second step change corresponds to one of the dynamic output and the emission output exceeding the first boundary condition, the second step change being less than the first step change.

A method of determining a local temperature anomaly in a fluidized bed combustion boiler system that includes at least three temperature sensors together defining a measurement grid, each sensor representing a measurement point, includes monitoring current operation data of the boiler, including measured bed temperature and at least primary air flow, fuel moisture, main steam flow, flue gas oxygen, and bed pressure, preparing a numerical model among operation data, such as primary air flow, fuel moisture, main steam flow, flue gas oxygen, and bed pressure. The measured bed temperatures measurement points are prepared and calibrated. Bed temperatures for the measurement points are monitored using the numerical model. This obtains computed bed temperatures under normal operation conditions, and the measured bed temperatures are compared with the computed bed temperatures for at least some of the measurement points. If an anomaly threshold is exceeded, determining that a local temperature anomaly is present.

A method of determining a local temperature anomaly in a fluidized bed combustion boiler system that includes at least three temperature sensors together defining a measurement grid, each sensor representing a measurement point, includes monitoring current operation data of the boiler, including measured bed temperature and at least primary air flow, fuel moisture, main steam flow, flue gas oxygen, and bed pressure, preparing a numerical model among operation data, such as primary air flow, fuel moisture, main steam flow, flue gas oxygen, and bed pressure. The measured bed temperatures measurement points are prepared and calibrated. Bed temperatures for the measurement points are monitored using the numerical model. This obtains computed bed temperatures under normal operation conditions, and the measured bed temperatures are compared with the computed bed temperatures for at least some of the measurement points. If an anomaly threshold is exceeded, determining that a local temperature anomaly is present.

Real-time dynamic optimization of a process model in an online model-based process control computing environment. A mixed integer nonlinear programming (MINLP) solver utilizes grouping of first-principle model units to implement constraints of the underlying process. A group identifier parameter and a group complement parameter enable the grouping behavior through association with the first-principles model units.