In case of considering real scale processes, because of many facilities, many numbers of jobs and long target terms, it usually takes long time to make production plan and schedule, and to make matters worse, we are not able to get them with adequate accuracy. There are few papers to solve these problems before, so in this paper, we propose a method to get them faster than ever and with adequate accuracy for these kinds of problems. In this method, we decide time horizon and make plan and schedule for this time horizon. In real processes, there are many complex constraints, so in the above horizon, the problems to be solved are simplified. Once problems are simplified, we solve the simplified problems, and we revise the plans and the schedules as the plans and schedules satisfy original constraints by using the precise simulator imitating the real processes. Then we proceed to next time horizon. We tested this method by computational experiments, and this method is effective for getting plans and schedules with adequate accuracy faster than ever.

In the construction field, the improvement of productivity and the work efficiency are demanded by the introduction of automation and ICT technology for the construction machine. However, it is the fact that the work efficiency and productivity depend on the operator skill of the construction machine in the current construction field. Therefore, the work efficiency will be high in the field with the skilled operator. In this paper, the analysis of feature for the digging operation of an excavator by using a random forest is proposed. A random forest is learned on the basis of skilled work states. The operation difference has been verified by the judgment result of the random forest compared with novice, typical, and professional work states. Moreover, the difference of operations has been considered by state flow models which were made from the judgment result of the random forest.

A new algorithm is proposed to estimate parameters of Single-Input Single-Output system with white noise gain function. The algorithm is an expansion of the previously proposed algorithms proposed by the author with iterative estimation of the variances of noises. This algorithm uses an iterative calculation and it consists of two parts. One part is the estimation of the system parameters by use of the variances of input and output noises. The estimation of the system parameters is one of the least squares identification algorithms using eigenvector. The other part is the estimation of the variances of input and output noises using the white noise gain function of the identified system parameters. Some simulations show the effectiveness of the proposed method.