Thesis for the degree of Doctor of Engineering Sciences in specialty 05.13.06 – information technology. – The Institute of Mathematical Machines and Systems Problems of the National Academy of Sciences of Ukraine, Kyiv, 2021.
The main result of the dissertation research is the solution of the scientific applied problem of increasing the efficiency of food enterprise management processes by developing information technology to support decision making on the basis of modified heuristic and evolutionary methods and algorithms.
The thesis develops a mathematical model of the order execution planning process, which, in contrast to the traditional models, provides a breakdown by management levels and departments, as well as an assessment of the plan effectiveness, taking into account external and internal influences based on the proposed partial criteria and constraints. The mathematical model takes into account all the conditions of the food company and the order execution planning process. These conditions include: the shelf life of raw materials and finished products; the possibility of occurrence and the need to process substandard products; features of execution of each order; features of using technological equipment, etc.
The paper proposes modifications of an «ant colony» and a «wolf pack» algorithms, the application of which allows reconfiguration of the plans while reducing the time necessary to find the optimum. Modification of the «ant colony» algorithm reduces time by using local optimization approaches, modification of the «wolf pack» algorithm – by selecting templates from the best alternative solutions to the problem.
A combined decision-making method based on a «fish shoal» and a modified «wolf pack» algorithms has been developed. This method works stably for small and medium-sized problems and reduces the search time, provided that up to 5-6 partial criteria are taken into account.
A combined method based on «ant colony» and «wolf pack» algorithms, as well as on a «genetic» algorithm has been developed. The method stably works for multi-criteria problems of different dimensions. It is advisable to use it for companies that manufacture a wide range of multicomponent products and require complex technological operations, separated by certain areas of production.
An information technology for order execution planning, which provides the formation and operational reconfiguration of plans in a short period of time through a combination of data mining for pre-classification, as well as the use of modified algorithms and methods based on a combination of algorithms, is developed in the thesis. The proposed technology makes it possible to select a set of algorithms and methods adequate for a particular subject area. The use of data mining provides an increase in the speed of finding solutions through data analysis for previous periods which allows to identify the best sequences of the use of specific technological equipment for the manufacture of products according to current orders. The application of this technology allows to find additional alternative optimal plans for the execution of orders, and the presentation of the results of their comparison in the form of a visual petal diagram helps the responsible person to make an effective decision quickly.
An architecture of a hybrid web-based decision support system has been developed. It implements the proposed information technology and is based on flexible integration in the data lake from statistical and expert sources. The proposed architecture makes it possible to combine the following information technologies: an order planning technology which provides the creation and adjustment of operational and calendar plans through the use of modified and combined algorithms; approaches of expert-modeling system of manufacturing recipes that gives the chance to pick up in production conditions replacement of components in a compounding; a technology to improve the process of analysis and planning of orders at food companies which, in contrast to the known ones, is based on determining the range and adjusting recipes to reduce costs and fast execution of orders.
The process of creating a decision support system using a spiral model of software development has been formalized. In addition to the traditional one, the place of refactoring in such a system is clearly defined. It allows to reduce development time and increase development efficiency in general.
The use of a web-based decision support system that implements the proposed information technology will improve the technical and economic performance of the enterprise.
Keywords: food company management, decision making, decision support systems, order execution planning, combined algorithms, combined decision-making methods, information technology.
