Furnace controller and method of operating a furnace

A method for controlling operation of a furnace to melt a material that comprises metal, the method comprising:

• storing data parameters related to the charge material, furnace condition and operation data for the furnace for a plurality of different cycles of operation of the furnace in a non-transitory computer readable medium of a computer device;
• creating or determining one or more cycle performance parameters based on the stored data parameters;
• creating a historian data store based on the stored data parameters and the cycle performance parameters, the historian data store being created such that outlier data for each of the one or more cycle performance parameters is not included in the historian data store, the outlier data for each of the one or more cycle performance parameters being data for that cycle performance parameter that is outside of a pre-selected variance range for an average value for the cycle performance parameter;
• after removal of the outlier data to create the historian data store, determining x-variables from the historian data store for one or more cycles of operation of the furnace and feeding the x-variables into a regression model to determine a relationship between at least one of the x-variables with at least one y-variable to define at least one reference cycle, each y-variable corresponding to a respective one of the cycle performance parameters;
• receiving real-time data from sensors of the furnace;
• comparing the real-time data from the sensors of the furnace to the at least one reference cycle to determine whether an adjustment to one or more furnace operational parameters is needed;
• upon determining that a difference from the at least one reference cycle exists that meets or exceeds a significance threshold based on the comparing of the real-time data from the sensors of the furnace to the at least one reference cycle, adjusting operation of the furnace so that operation of the furnace is adjusted to converge toward a pre-selected furnace performance defined by the at least one reference cycle and
• determining the relationship between x-variables and y-variables via the regression model to identify one or more significant x-variables, insignificant x-variables being x-variables determined to have an effect that is below a pre-selected significance threshold and each of the one or more significant x-variables having an effect that is above the pre-selected significance threshold;
• utilizing the real-time data from the sensors of the furnace for the identified one or more significant x-variables to calculate at least one desired cycle performance parameter value for use in improving furnace operation for a cycle of furnace operation.