Device and method for monitoring a system

A method of employing a device to monitor a system, the device having a processor apparatus having a processor and a storage, an input apparatus having a number of input ports that receive a number of sets of signals that are representative of a number of operational parameters of the system and provide them as input signals to the processor apparatus, an output apparatus that is structured to receive output signals from the processor apparatus, the storage having stored therein a number of routines which, when executed on the processor, cause the device to perform a number of operations comprising:

• performing a learning procedure comprising:
• receiving on a subset of the number of input ports one or more series of first learning signals that are representative of one or more operational parameters of the number of operational parameters;
• receiving on another input port different from the subset a series of second learning signals that are representative of another operational parameter of the number of operational parameters;
• subjecting at least a representation of at least a subset of the one or more series of first learning signals and the series of second learning signals to at least a first machine learning algorithm to obtain at least a first model of the system;
• at each time step in a series of time steps during operation of the system, performing an operational procedure comprising:
• receiving on each of the subset of the number of input ports a first operational signal that is representative of one of the one or more operational parameters;
• receiving on the another input port a second operational signal that is representative of the another operational parameter, wherein the another operational parameter is an operational parameter of primary interest, and wherein the another operational parameter is indicative of a stability of the system;
• inputting at least a representation of each first operational signal into the at least first model to obtain from the at least first model a predicted operational signal that is a prediction of the another operational parameter;
• obtaining an error signal based at least in part upon a difference between the predicted operational signal and the second operational signal;
• subjecting the error signal to a statistical process control algorithm, wherein the first statistical process control algorithm determines whether the predicted operational signal is sufficiently different than the second operational signal;
• subjecting the first operational signals to a second statistical process control algorithm to detect any of the first operational signals outside a range of normal behavior;
• outputting an alarm based on the statistical process control algorithm determining that the error signal exceeds a predetermined threshold;
• and
• outputting another alarm based on the second statistical process control algorithm determining that any of the one or more operational parameters are outside the range of normal behavior.

A device structured to monitor a system, the device comprising:

• a processor apparatus comprising a processor and a storage;
• an input apparatus comprising a plurality of input ports that are structured to receive a plurality of signals indicative of a plurality of operational parameters of the system and to provide the plurality of signals as input signals to the processor apparatus;
• an output apparatus that is structured to receive output signals from the processor apparatus;
• the storage having stored therein a number of routines which, when executed on the processor, cause the device to:
• perform a learning procedure comprising:
• receiving, on secondary input ports of the plurality of input ports, a plurality of secondary learning signals that are indicative of a plurality of secondary operational parameters of the plurality of operational parameters;
• receiving, on a primary input port of the plurality of input ports, primary learning signals indicative of a primary operational parameter of the plurality of operational parameters, wherein the primary operational parameter is indicative of a stability of the system, wherein the primary operational parameter is different than the plurality of secondary operational parameters, and wherein the input port is different than the secondary input ports;
• inputting the primary learning signals and the plurality of secondary learning signals into a machine learning algorithm to obtain a model of the system;
• at each time step in a series of time steps during operation of the system, performing an operational procedure comprising:
• receiving on the primary input port a primary operational signal indicative of the primary operational parameter;
• receiving on each secondary input port a secondary operational signal indicative of a secondary operational parameter of the plurality of secondary operational parameters;
• inputting each secondary operational signal into the model to obtain from the model a predicted primary operational signal that is a prediction of the primary operational parameter;
• obtaining an error signal based on a difference between the predicted primary operational signal and the primary operational signal;
• inputting the error signal into a first statistical process control algorithm, wherein the first statistical process control algorithm is to determine whether the predicted primary operational signal is sufficiently different than the primary operational signal;
• and
• outputting an alarm based on the first statistical process control algorithm determining that the error signal exceeds a predetermined threshold.