Wyniki wyszukiwania - BazTech

1

A problem of predictivie scheduling of jobs in a production system

Wosik I.

Wybrane Problemy Inżynierskie

|

2011

|

nr 2

369-374

EN

In the paper a classical model of failures is considered in that successive failurefree times are supposed to have Weibull distributions and are followed by exponentially distributed times of repairs. It is assumed that parameters of these distributions, in general, change with time. Basing on information about the number of failures, failure-free times, repair times, in a number of periods of the same duration in the past parameters of the model are estimated. Next, predictions of the most important reliability characteristics are found using classical regression technique.

2

Predictive scheduling based on knowledge acquired from frequency of machine work disturbances

Wosik I., Skołud B.

Journal of Machine Engineering

|

2011

|

Vol. 11, No. 1-2

54--67

EN

During execution of a schedule some uncertain events may take place for example: resources may become unavailable, machine can be broken. Uncertainty should be included in the process of job scheduling. In the paper the problem to generate a workable, proactive baseline schedule under production constraints and unexpected event occurrence conditions is considered. The proactive baseline schedule protects against anticipated occurrences that may appear during the schedule execution. The machine breaking time is searched and the information is used to generate a robust schedule. In the paper the method of data acquisition basing on probability theory is proposed. The time of machine breaking is acquired from historical data of frequency of machine failure. A numerical example of building a hypothesis H:{the cumulative distribution function of the failure time is a normal distribution}, verification of the hypothesis, and predictive scheduling is presented. The normal distribution is proposed to describe failure time of machine as it gives consideration to a gradual wear process of the machine. The paper is proposition of improving simulation systems such as the Enterprise Dynamics or Taylor and scheduling systems such as Knowledge based Rescheduling System and Multi Objective Immune Scheduling Algorithm.

3

Tuning of a fuzzy system for controlling searching process in Multi Objective Scheduling Immune Algorithm

Wosik I., Skołud B.

Journal of Machine Engineering

|

2009

|

Vol. 9, No. 1

130--143

EN

In the paper the Multi Objective Immune Algorithm (MOIA) for an open job shop scheduling problem (OJSP) is proposed. The OJSP belongs to most both time consuming and most complicated problems in scope of searching space. In the paper schedules are evaluated by using three criteria: makespan, flowtime and total tardiness. MOIA proposes a schedule, which is best one, selected from a set of achieved solutions. An affinity threshold is a parameter that controls equilibrium between searching space and solutions diversity in MOIA. The affinity threshold is defined by using fuzzy logic system. In the paper fuzzy system is tuned by selecting shape, size of fuzzy sets, and fuzzy decisions of an affinity threshold. If the fuzzy system is used then neither the knowledge about the affinity threshold nor influence over searching processes is not required from a decision-maker. The application of the fuzzy system makes the process of decision-making user friendly. In the paper efficiency of MOIA before and after the fuzzy system tuning is compared and computational results are presented

4

Strategie przeszukiwania zbioru rozwiązań niezdominowanych w sensie Pareto

Wosik I.

Inżynieria Maszyn

|

2009

|

R. 14, z. 3

101-120

PL

W artykule podjęto się problemu optymalizacji wielokryterialnej z uwagi na fakt, że jest on kluczowym tematem w wielu dziedzinach, np.: informatyka, badania operacyjne. Poruszono problematykę szeregowania zadań z zastosowaniem wielokryterialnej oceny harmonogramów. Narzędziem wykorzystywanym do optymalizowania jest autorski algorytm immunologiczny optymalizacji wielokryterialnej MOIA (ang. Multi Objective Immune System). MOIA jest systemem wspomagającym podejmowanie decyzji w zakresie optymalizacji wielokryterialnej, z możliwością wyboru jednego z trzech podejść: "a priori", "a posteriori" oraz "podejścia pośredniego". Skoncentrowano się na podejściu "a posteriori", gdzie MOIA generuje zbiór rozwiązań niezdominowanych w sensie Pareto, natomiast do wyboru rozwiązania najlepszego z centralnej części frontu Pareto stosuje się metodę skalaryzacji wektorowego wskaźnika oceny. Podjęto się też zbadania innych metod oceny znalezionych punktów w przestrzeni wielokryterialnej oraz metod zapobiegających preferowaniu jednego kryterium. Poszukuje się, bowiem takiej metody przeszukiwania wielokryterialnej przestrzeni wektorów zmiennych decyzyjnych, która pozwoli znaleźć rozwiązanie dla sprzecznych kryteriów oceny, nie biorąc pod uwagę preferencji nadanych kryteriom. Weryfikacja metod z zastosowaniem MOIA prowadzona jest dla problemu szeregowania zadań w otwartym gnieździe produkcyjnym oraz uproszczonego problemu typu projekt złożonego z systemów przepływowego i montażowego.

EN

In the paper multi criteria optimisation problem is considered, as it is crucial topic in scientific and practice. The paper deals with scheduling problem with application of multi criteria schedules evaluation. Multi criteria immune algorithm MOIA is applied for multicriteria optimization. MOIA is computer aided decision-making system with three possible approaches: "a priori", "a posteriori" and "a middle approach" for multi criteria optimisation. In the paper the "a posteriori" approach is considered. In the "a posteriori" approach, MOIA reaches Pareto solution set. For selecting one best solution from middle part of Pareto front Weighted Aggregation Method has been used. In the paper, another Pareto solution set searching methods are tested. Multi criteria searching space method finding solution for contradictive objectives without giving preferences (criteria priority) is searched. Application of Pareto solution set searching methods in MOIA is verified. Methods are verified for two problems: open job shop scheduling problem and project scheduling problem consisting of flow shop problem and assembly jobs scheduling problem.

5

Algorytmy immunologiczne w szeregowaniu zadań produkcyjnych

Skołud B., Wosik I.

Zarządzanie Przedsiębiorstwem

|

2008

|

Vol. 11, nr 1

47-56

EN

It is important to shorten time of new production order at-tending, time of production process and to improve productivity and to execute the total production programme. Scheduling problem is one of the major issues on the quick response to continuously changing customers' demands. In the paper, the Multi Objective Immune Algorithm (MOIA) for scheduling problems is proposed. The goal of MOIA is to generate a Pareto curve that enables the decision-maker to make explicit trade-offs between conflicting objective functions. The decision-maker can use two, three or four criteria in optimisation process: makespan minimization; total tardiness minimization; total flow time minimization; total idle time of machines minimization. The decision-maker is released from his obligation to have knowledge about used criteria and their interactions in process of objective functions' weights defining. MOIA reaches one most suitable solution from Pareto-optimal solutions set. In order to find best searching strategy of best schedule various researches were made. MOIA has been used for solving flow shop, job shop and open job shop scheduling problems. The production system can deal with sequence dependent setup times or normal setup times, and production batch flow can be parallel or combination of the series flow and parallel flow. MOIA reaches compromise solutions for contradictive criteria of scheduling problems in short period of time.

6

Zastosowanie iteracyjnych metod przeszukiwania do wyznaczania wielkości partii współbieżnej produkcji wieloasortymentowej

Skołud B, Wosik I.

Inżynieria Maszyn

|

2005

|

R. 10, z. 3

95-107

PL

W systemach współbieżnej produkcji wieloasortymentowej trudność sprawia dobór asortymentu do jednoczesnego wykonania oraz dobór liczności partii produkcyjnych dla poszczególnych zadań. Kryterium, które decyduje o tej wartości jest maksymalne wykorzystanie dysponowanych zdolności produkcyjnych, tj. efektywność wykorzystania maszyn, urządzeń, personelu. W artykule przedstawiono możliwość rozwiązania problemu doboru wielkości partii z zastosowaniem dwóch metod: iteracyjnej metody przeszukiwania lokalnego oraz metody przeszukiwania populacyjnego. W pierwszym przypadku zastosowano algorytm największego wzrostu w połączeniu z mechanizmem tworzenia komórek obronnych układu odpornościowego. W drugim przypadku przedstawiono algorytm selekcji immunologicznej, wykorzystujący mechanizmy selekcji, preselekcji i zatłoczenia. Dla każdego algorytmu przeprowadzono symulację, porównano wyniki i wyciągnięto wnioski.

EN

The complexity of concurrent, multiassortment production systems lies in the difficulty of the choice of elements for simultaneous production and batch sizing the particular task. The criterion of maximum utilization of production resources (machines, devices and staff) decides about the batch size. The possibility of two methods: local and population searching methods for production butch size estimation is depicted in the paper. Hill climbing algorithm connected with mechanism of defense cell creation, witch takes place in mammal immune system is used in the first method. The possibility of immune clonal selection algorithm using cloning, pre-selection and overcrowding mechanisms for production batch size estimation is also shown in the paper. The elaborated methods, algorithms of batch size estimation problem, and results of computer simulations are presented.