Abstract: Once major equipment are broken down suddenly, if it cannot be responded in time, the production is suspended, and life cost, environment cost and economy cost are happened. Time-tight, urgent and high complexity are the characteristics of emergency repairs. For such practical problems of emergency repair, based on the above characteristics, aiming at minimizing the direct and indirect losses of emergency repairs, the serial and parallel relationships of logic sequence of critical tasks and non-critical tasks are considered. A forward optimization model for “task-personnel” matching is constructed. However, even if the optimal value of the model is still higher than the benchmark cost value, in order to solve this kind of problem, in view of the adjustable space for completion time, the inverse optimal value model is constructed. It is driven by the benchmark cost value. It is followed the idea of reverse thinking, which is the causes are found out from the results, and the task time is reverse-derived. In view of the inverse optimal value model has the characteristics of 0-1 mixed integer, nonlinear, bi-level and NP-hard, the hybrid “genetic-integer linear planning” algorithm is designed. The parallelization of genetic algorithm with the global search capability of integer linear planning are combined. Relevant data are obtained based on the actual investigation and interview of the relevant experts of the enterprise. The numerical analysis are showed that: the forward optimization method cannot get the benchmark value. Based on inverse optimal value method, both the intelligent algorithm and the exact algorithm can get the benchmark value with different “task-personnel” matching scheme and completion time. It shows that for the cost control, both the combination of personnel assignment scheme and completion time standard should be considered, and only one of them cannot achieve the expected. The inverse competition value method can ensure the realization of the target cost and reduce the cost of the total repair time. The “task-personnel-time” assignment themes are given by the algorithm. Intelligent algorithm solution speed has significant advantages, which is more suitable for such time-critical, urgent task and complex problems as emergency repair. The method proposed in this study can be applied to many fields such as project schedule control, standard working time design of human resource management, resource allocation decision-making from the perspective of target management, etc. It can give inspiration to the problem for inverse inferring decision parameters according to the expected results of decision-making.

Key words: inverse optimal value model; 0-1 mixed integer nonlinear bi-level programming; mixed genetic-integer linear programming algorithm; “task-person” matching; time optimization; emergency repair project