Prediction of responses to different powdering tests on a galvanneal-coated steel substrate or different tests on other substrates, using computer-based systems and methods

A method of determining a range of values for at least one variable product parameter of a product to be manufactured, said method comprising the steps of:

• providing a computer-readable database which contains product data on several previously manufactured products, as well as test result data for each of said previously manufactured products, said test result data including information on how the previously manufactured products performed when subjected to a first testing procedure;
• analyzing said product data and said test result data using a computer-implemented classification and regression tree analysis;
• generating, based on said computer-implemented classification and regression tree analysis, a binary decision tree having multiple decision result branches, each decision result branch being defined by a result of a binary decision on one of a plurality of product parameters;
• applying at least one desired value of a product parameter to said binary decision tree to eliminate from consideration any decision result branches which are inconsistent with said at least one desired value;
• and
• determining said range based upon which path through the decision result branches which were not eliminated from consideration, yields a best probability that a product with said at least one desired value will respond favorably to said first testing procedure.

A method of performing process control in a manufacturing process, comprising the steps of:

• providing a computer-readable database which contains product data on several previously manufactured products, as well as test result data for each of said previously manufactured products, said test result data including information on how the previously manufactured products performed when subjected to a first testing procedure;
• analyzing said product data and said test result data using a computer-implemented classification and regression tree analysis;
• generating, based on said computer-implemented classification and regression tree analysis, a binary decision tree having multiple decision result branches, each decision result branch being defined by a result of a binary decision on one of a plurality of parameters;
• applying values of product parameters to said binary decision tree to eliminate from consideration any decision result branches which are inconsistent with said values;
• determining at least one probability, based on said values of product parameters, that a product with said values of product parameters will respond favorably to said first testing procedure;
• and
• adjusting at least one step in said manufacturing process if said at least one probability is outside of an acceptable range.

A method of classifying a product, comprising the steps of:

• providing a computer-readable database which contains product data on several previously manufactured products, as well as test result data for each of said previously manufactured products, said test result data including information on how the previously manufactured products performed when subjected to a first testing procedure;
• analyzing said product data and said test result data using a computer-implemented classification and regression tree analysis;
• generating, based on said computer-implemented classification and regression tree analysis, a binary decision tree having multiple decision result branches, each decision result branch being defined by a result of a binary decision on one of a plurality of parameters;
• applying values of product parameters associated with said product to said binary decision tree;
• determining at least one probability, based on said values of product parameters, that said product will respond favorably to said first testing procedure;
• and
• classifying said product as being good, marginal or poor, based on said at least one probability.

A method of classifying a product, comprising the steps of:

• providing a computer-readable database which contains product data on several previously manufactured products, as well as test result data for each previously manufactured product, said test result data including information on how the previously manufactured products performed when subjected to a first testing procedure and information on how at least some of the previously manufactured products performed when subjected to at least a second testing procedure;
• analyzing said product data and said test result data using a computer-implemented classification and regression tree analysis;
• generating, for each of said first and at least a second testing procedures, a binary decision tree based on said computer-implemented classification and regression tree analysis, each binary decision tree having multiple decision result branches, each decision result branch being defined by a result of a binary decision on one of a plurality of parameters;
• applying values of product parameters associated with said product to each binary decision tree;
• determining, based on said values of product parameters as applied to each binary decision tree, probabilities of said product responding favorably when subjected to said first and at least a second powdering tests;
• and
• classifying said product based on which of said first and at least a second testing procedures said product is most likely to satisfy as determined by the probabilities associated with each binary decision tree.

A computer system for use in determining product parameters, said computer system comprising:

• a storage device containing a parameter selection program based on at least one binary decision tree derived from a classification and regression tree analysis of:
• a) product data on previously manufactured products, and
• b) test data on how such previously manufactured products respond to at least one testing procedure;
• an input device for receiving information indicative of at least one desired value of product parameter;
• and
• a processor connected at least indirectly to said storage device and said input device, said processor being adapted to perform said parameter selection program by applying said information to said at least one binary decision tree and generating an output signal indicative of at least one range of at least one other product parameter which, according to said at least one binary decision tree, contains values of said at least one other product parameter which enhance a probability that a product with said at least one desired value of product parameter will respond favorably to said at least one testing procedure.

A computer system for use in performing process control, said computer system comprising:

• a storage device containing a prediction program based on at least one binary decision tree derived from a classification and regression tree analysis of:
• a) product data on previously manufactured products, and
• b) test data on how such previously manufactured products respond to at least one testing procedure;
• an input device for receiving information indicative of at least one product parameter;
• a processor connected at least indirectly to said storage device and said input device, said processor being adapted to perform said prediction program by applying said information to said at least one binary decision tree and generating an output signal indicative of at least one probability that a product with said at least one product parameter will respond favorably to said at least one testing procedure;
• and
• a control mechanism connected at least indirectly to said processor and responsive to said output signal, said control mechanism being adapted to adjust at least one step in a manufacturing process if said at least one probability is outside of an acceptable range.

A computer system for use in classifying products, said computer system comprising:

• a storage device containing a prediction program based on at least one binary decision tree derived from a classification and regression tree analysis of:
• a) a product data on previously manufactured products, and
• b) test data on how such previously manufactured products respond to at least one testing procedure;
• an input device for receiving information indicative of at least one product parameter;
• a processor connected at least indirectly to said storage device and said input device, said processor being adapted to perform said prediction program by applying said information to said at least one binary decision tree and generating an output signal indicative of at least one probability that a product with said at least one product parameter will respond favorably to said at least one testing procedure;
• and
• a classification mechanism responsive to said processor, for identifying said product as good, marginal, or poor based on said at least one probability.