| Abstrakt: |
An important factor determining the length of industrial trials for ERP systems is the quality of user training for working in the integrated management system, which is integrated at the level of elementary transactions. Obviously, the length of this period and the corresponding losses associated with problems in the operation of the ERP system may be significantly decreased not only by satisfactory design and testing of the system but, in particular, by improving user skills through training. The problem of creating a training program for users of an ERP system at a large metallurgical company is expressed in mathematical terms. In this formulation, the optimization criterion is the total increment in user competence as a result of the program. Solution of this problem is based on network programming, which calls for structurally similar network representation of the target function and constraints. The authors present a general scheme for solution of this problem, as well as an example. In the solution, the individual assessment problems are solved by dichotomous programming. The approximate solutions obtained may be improved by finding the global optimum of the initial problem on the basis of the branch and bound method. The boundaries employed are values of the target function in the approximate solution. In practice, it will be expedient to consider the inverse of the problem considered here: in that case, the criterion is the training cost. In view of the structural similarity of the increment in skills and the training cost, the approach proposed for the direct problem may also be applied to the inverse problem. To generalize the problem considered, user preferences regarding the relative significance of particular programs may be taken into account by appropriate weighting. The basic solution remains unchanged. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink