Method of analyzing the performance of gas turbine engines

A method of identifying likely contributors to an error caused in measured data from a system which has a plurality of associated system variables, the method including the steps of:

• a) optimizing an error function which relates the measured data to expected data and to a predetermined number of the system variables by adjusting the system variables to obtain an optimized form of the error function;
• b) selecting a subset of the adjusted system variables which contains the adjusted system variables which contribute most to the optimized form of the error function;
• and
• c) analysing all possible combinations of a predetermined number of the variables of the subset, to determine the combination which makes the greatest contribution to the optimized error function, wherein at least one of the above steps is performed by a processor.

A method of identifying likely contributors to an error caused in measured data from a system which has a plurality of associated system variables, the method including the steps of:

• a) optimizing an error function which relates the measured data to expected data and to a predetermined number of the system variables by adjusting the system variables to obtain an optimized form of the error function;
• b) selecting a first subset of the adjusted system variables which contains a plurality of the adjusted system variables which contribute most to the optimized form of the error function;
• c) optimizing a further error function, which does not contain system variables outside the subset, by adjusting the system variables in the first subset;
• d) selecting a second subset of the adjusted system variables which contains the system variables which contribute most to the optimized form of the further error function;
• and
• e) analysing all possible combinations of a predetermined number of the system variables of the second subset, to determine the combination which makes the greatest contribution to the optimized form of the further error function, wherein at least one of the above steps is performed by a processor.