Method and apparatus for control and evaluation of pending jobs in a factory

Claim 3

The job scheduling method of claim 1 wherein said step of establishing said Factory Profile curve includes the substep of determining a range enclosing said optimum CI value of said Factory Profile curve and wherein criteria for said releasing step are satisfied if said actual CI curve falls within said range enclosing said optimum CI value.

Claim 4

The job scheduling method of claim 2 wherein said step of establishing said Factory Profile curve of optimum CI value over time includes the substep of adjusting said range about said optimum CI value to provide a wider range for jobs having a higher than normal priority.

Claim 5

The job scheduling method of claim 2 wherein said comparing step includes an error determination based on a minimum of a first sum of squares and a second sum of squares over said processing time for said job being tested for acceptance, said first sum of squares being determined as the difference between the actual CI for said job at each point in said time and the upper bound of said range of optimum CI and said second sum of squares being determined as the difference between the actual CI for said job at each point in said time and the lower bound of said range of optimum CI.

Claim 6

The job scheduling method of claim 2 wherein said new trial start date for acceptance by said comparing step is automatically computed, as by a digital computer, using a BestStartDate algorithm: ______________________________________ Let Ethmax be maximum Threshold Error Let Ethmin (E = Error) be minimum Threshold Error Emin <- Eth Emax <- 0 is the max error without going over the threshold value Sd <- Desired.sub.-- Sd (desired start date), SDc <- Sd, i.e. current Sd is originaliy set to the Desired.sub.-- Sd Nr, N1 no of right shifts and left shifts allowed Nr is a function of order priority, i.e. as order priority increases then Nr becomes smaller; N1 is a management- determined finished good inventory function, stating how early, in the worst case, completion of the order will be acceptable. Note that these two parameters decide on the boundaries of the search space. They can be either predefined or can be decided at the time of order entry. n is shift counter; +ve for right shifts & -ve for left shifts; n <- 0 Wc is the current window.sub.-- no w <- Wc Allocate: Set t <- end.sub.-- of(w) /* t is the max time unit in w, e.g., last day of week w */ Allocate until t OR all steps done otherwise EXIT Evaluate e(t), where t is the end of current window If e(t) > Ethmin THEN BEGIN if e(t) < Emin THEN BestSD <- SDc, Emin <-e(t); /*else the previous SD was better*/ find.sub.-- a.sub.-- better.sub.-- sd; auto allocate i END-begin ELSE /* error term is still smaller than threshold */ BEGIN w <- w+1 Emax <- e(t) GOTO allocate END-begin; If e(t) > Ethmax THEN warning: Promised due date is too good for this priority class; print(BestSD, Emin, Emax); END BestStartDate. PROCEDURE find.sub.-- a.sub.-- better.sub.-- sd: If n=<0 & n>N1 THEN shiftleftORright(-) /* note that N1 is a -ve number */ Else IF n<Nr THEN shiftleftORright(+); ELSE /*UNSUCCESSFUL release */ DO ONE OF THE FOLLOWING Increase job's priority; not recommended if done too often; Decrease job's priority, i.e., increasing Nr the search space this is relaxing constraints; Decrease(make more negative) N1, this is also relaxing WIP constraint; Suspend release until a later time; Other alternatives are changing Eth and CI* etc but these are not recommended since they would have a global effect. END-else; END.sub.-- find.sub.-- a.sub.-- better sd. PROCEDURE shiftleftORright (.sup.˜): /* .sup.˜ is - if shiftleft and + if shiftright */ n <- n.sup.˜ 1 w <- w.sup.˜ 1 Wc <- w SDc <- max(starttime(w), Today) END shiftORright. ______________________________________

Claim 9

The job scheduling system of claim 6 wherein said means for establishing said Factory Profile curve includes a further means for determining a range enclosing said optimum CI value of said Factory Profile curve and wherein said criteria for releasing said one of said jobs into said factory for processing by at least one of said machines on said trial start date are satisfied if said actual CI curve falls within said range enclosing said optimum CI value.

Claim 10

The job scheduling system of claim 7 wherein said means for establishing an optimum CI curve for said shop/includes a further means for adjusting said range about said optimum CI value to provide a wider range for jobs having a higher than normal priority.

Claim 11

The job scheduling system of claim 7 wherein said comparison means further includes an error determination means enabled to compute a minimum of a first sum of squares and a second sum of squares over said processing time for said job being tested for acceptance, said first sum of squares being determined as the difference between the actual CI for said job at each point in said time and the upper bound of said range of optimum CI and said second sum of squares being determined as the difference between the actual CI for said job at each point in said time and the lower bound of said range of optimum CI.

Claim 12

The job scheduling system of claim 7 further including a computational means for computing said start date for acceptance by said comparison means using a BestStartDate algorithm ______________________________________ Let Ethmax be maximum Threshold Error Let Ethmin (E = Error) be minimum Threshold Error Emin <- Eth Emax <- 0 is the max error without going over the threshold value Sd <- Desired.sub.-- Sd (desired start date), SDc <- Sd, i.e. current Sd is originaliy set to the Desired.sub.-- Sd Nr, N1 no of right shifts and left shifts allowed Nr is a function of order priority, i.e. as order priority increases then Nr becomes smaller; N1 is a management- determined finished good inventory function, stating how early, in the worst case, completion of the order will be acceptable. Note that these two parameters decide on the boundaries of the search space. They can be either predefined or can be decided at the time of order entry. n is shift counter; +ve for right shifts & -ve for left shifts; n <- 0 Wc is the current window.sub.-- no w <- Wc Allocate: Set t <- end.sub.-- of(w) /* t is the max time unit in w, e.g., last day of week w */ Allocate until t OR all steps done otherwise EXIT Evaluate e(t), where t is the end of current window If e(t) > Ethmin THEN BEGIN if e(t) < Emin THEN BestSD <- SDc, Emin <-e(t); /*else the previous SD was better*/ find.sub.-- a.sub.-- better.sub.-- sd; auto allocate i END-begin ELSE /* error term is still smaller than threshold */ BEGIN w <- w+1 Emax <- e(t) GOTO allocate END-begin; If e(t) > Ethmax THEN warning: Promised due date is too good for this priority class; print(BestSD, Emin, Emax); END BestStartDate. PROCEDURE find.sub.-- a.sub.-- better.sub.-- sd: If n=<0 & n>N1 THEN shiftleftORright(-) /* note that N1 is a -ve number */ Else IF n<Nr THEN shiftleftORright(+); ELSE /*UNSUCCESSFUL release */ DO ONE OF THE FOLLOWING Increase job's priority; not recommended if done too often; Decrease job's priority, i.e., increasing Nr the search space this is relaxing constraints; Decrease(make more negative) N1, this is also relaxing WIP constraint; Suspend release until a later time; Other alternatives are changing Eth and CI* etc but these are not recommended since they would have a global effect. END-else; END.sub.-- find.sub.-- a.sub.-- better sd. PROCEDURE shiftleftORright (.sup.˜): /* .sup.˜ is - if shiftleft and + if shiftright */ n <- n.sup.˜ 1 w <- w.sup.˜ 1 Wc <- w SDc <- max(starttime(w), Today) END shiftORright. ______________________________________