Process control and optimization technique using immunological concepts

A method of optimizing the operation of a process, comprising, during on-line operation of the process:

• determining a multiplicity of process control states during ongoing the on-line operation of the process by determining values for each of a set of process inputs and a set of process outputs that result from the set of process inputs being applied to the process to thereby define a particular process control state, the set of process inputs including a set of manipulated variables (MVs) and a set of disturbance variables (DVs), and the set of process outputs including a set of controlled variables (CVs);
• storing the multiplicity of process control states in a memory;
• detecting a change of at least one of the set of disturbance variables that causes the process to exit from operating at a process steady state condition, the process steady state condition corresponding to a DV steady state of the at least one of the set of disturbance variables and the change causing the at least one of the set of DVs to exit from the DV steady state;
• based on the detected change to the at least one of the set of disturbance variables, operating the process in a regulation operation mode of the on-line operation of the process, including:
• selecting, from the memory, a subset of one or more of the stored multiplicity of process control states, the selected subset of the one or more stored multiplicity of process control states having one or more values that are closer to corresponding one or more values of a process control state corresponding to the detected change than are one or more values of another subset of the one or more of the stored multiplicity of process control states, and
• using the selected subset of one or more of the stored multiplicity of process control states to develop a set of process control inputs to cause the process to operate at a new process steady state condition corresponding to a new DV steady state of the at least one of the set of disturbance variables;
• upon the process operating at the new process steady state condition, operating the process in a stochastic operation mode of the on-line operation of the process, including periodically determining a new set of process control inputs to be used to control the process by predicting the response of the process to a proposed set of process control inputs based on at least one of the selected process control states and analyzing the optimality of the proposed set of process control inputs using an optimality criteria that uses both the proposed set of process control inputs and a set of predicted process control outputs that result from the proposed set of process control inputs;
• and
• removing one of the multiplicity of stored process control states from the memory based on an amount of usage of the one of the stored multiplicity of stored process control states while maintaining other stored process control states in the memory.