VII Международная студенческая научная конференция Студенческий научный форум - 2015

The purpose of the article is to increase the efficiency of logistic management of the transport system "point of departure - point of destination" ("PDep - PDes"). The purpose achievement is possible using the methods of the theory of fuzzy sets and neural networks, allowing to consider various ambiguities and indeterminacy in the processes of material flow’s management in real-time.

In recent years much attention is paid to a problem of planning, organization and management of work of the logistic delivery systems of goods in the conditions of risk and uncertainty of environment. However, existing methods of control consider influence of uncertain factors only when studying separate links and elements of channels of products distribution. Besides, probabilistic-statistical model are used mainly at disclosure of influence of uncertainty in problems of logistic management [2].

Effective delivery of goods in conditions of the dynamics and environmental indeterminacy associates with a high level of logistic management. This raises the problem of creation of the new organization of cargo delivery process that ensures the operation’s stability of separate links in the logistics delivery chain depending on the requirements to the process of delivery [1].

The efficiency of the logistics cargo management is achieved through coordinated interaction of separate parts and elements of the logistics chain of cargo delivery in the system “place of departure - place of destination" by rapid response to the incoming current information about changing the conditions at all stages of the transport process.

It’s necessary to process a large flow of complex information about the current state of the transport system’s objects, managing cargo in real time. So the use of artificial neural networks is proposed [4].

The process of cargo delivery in the system "PDep - PDes" is presented by the following stages: shipment at the shipper enterprise; the transportation of goods to the destination with selecting the type of transport and route; unloading at the destination point.

Each stage is characterized by a set of parameters that reflect the level of transport service, which at the design stage are fuzzy linguistic variables, so it is reasonable to use fuzzy sets [3].

It’s required to choose the most efficient route of cargo delivery with the required level of the transport service adequacy. For solving the problem the method based on fuzzy neural network ANFIS in MATLAB has been proposed. The General sequence of the process of developing hybrid network models can be represented by the following steps:

1. Preparing the file with the training data.

2. You should open the ANFIS editor (comand “anfisedit”) and upload a file with the training data.

3. After preparation and uploading the training data you can generate the structure of the FIS system (Sugeno type).

4. After generation the hybrid network, you can visualize its structure.

5. Before the hybrid network training, you should set the training parameters (the method of hybrid network training, the level of training error, the number of cycles).

6. Further configuration options of built and trained hybrid network can be done using standard graphical tools of the package Fuzzy Logic Toolbox.

7. The next step is testing of fuzzy system and the output of results in the visualization.

8. Now we should verify the adequacy of the constructed fuzzy model of hybrid network. To solve this task, you should use the function “evalfis”. The results of the forecast for the specific input is obtained by typing the following code in the command area:

x = [1 70 7 3 70 3000] % «The input parameters»;

fis = readfis('set1.fis') % «Upload file»;

y = evalfis(x, fis) % «Conclusion of the forecast».

The obtained value of the output of the linguistic variable “adequacy of transport service” for membership function y=0,7 is the result of solving the problem of fuzzy inference for the proposed linguistic values of input variables at a certain stage of rational delivery route selection. It means the less time of cargo delivery to the destination.

The obtained results give grounds to speak about the possibility of practical application of the neuro-fuzzy network to improve the efficiency of logistic management of the cargo delivery transport system.

Further researches consist in developing fuzzy situation networks for operational decision-making in the management of all participants interaction at the delivery system in real time.

Conclusions:

1. The proposed approach of coordinated interaction of the transport system elements and parts " PDep - PDes " allows to reduce the time of cargo delivery.

2. On the basis of this experiment we can make the conclusion that the use of ANFIS for solving the problem of coordination of logistics chains interaction is very promising. The disadvantage is that the quality of the results depends on the quality of experimental data or training data. Therefore, the selection of training data is an important process in using ANFIS.

List of references:

1. Gubenko V.K. Efficiency of the logistics system of cargo delivery/ V.K. Gubenko, Y.A. Nefedov, A.A. Lamsin // Visnyk Priazovskogo sovereign technical University. - 2007. - № 17, p. 23.

2. Nefedov Y.A. Logistic management of the transport system maintenance "metallurgical enterprise - port" in real time: dis. on competition of a scientific degree of cand. the technology. sciences: 05.22.01 / Nefedova Yana Igorevna. X., 2010, p. 167.

3. Leonenko A.V. Fuzzy modeling with MATLAB and fuzzyTECH / A.V. Leonenko - SPb.: BHV-Petersburg, 2003, p. 45.

4. Stovba S.D. Designing fuzzy systems by means of MATLAB / S.D. Stovba – M.: 2007, p.89.

5