Applications of FIS, ANN and ANFIS to industrial control

การประยุกต์ใช้ FIS, ANN และ ANFIS กับการควบคุมในอุตสาหกรรม

Name: Prasert Aengchuan

keyword: Fuzzy logic

; Artificial neural network (ANN)

; Adaptive neuro-fuzzy inference system (ANFIS)

; Process control

; Inventory control

Abstract: This dissertation presented the comparison of fuzzy logic (FIS), artificial neural networks (ANN) and adaptive neuro-fuzzy inference system by applying with the problems of process control and inventory control. Conventional industrial control such as process control and inventory control mostly concern to known or deterministic input parameters which is not practical and not realistic for many industries. Existing approaches cannot entirely deal with uncertain input parameters. Recently, the effective models to deal with such kind of problems which use in numerous applications under uncertainty are fuzzy logic approach, artificial neural networks (ANN) and adaptive neuro-fuzzy inference system (ANFIS) approach. So, these approaches were proposed in this dissertation for both process control and inventory control with a case study factory of each application. However, the most suitable model for each problem has not been comparatively studied. The process control application has determined with a case study of gypsum process to compare all three models. For FIS model, inputs data from process were converted to be fuzzy variables. The generated fuzzy rules were utilized to obtain the fuzzy process control parameters. Then these parameters were converted to the process control parameters for using in process control. For ANN model, data from process control were utilized to train and test with the model until getting the lowest error. Then these control parameters were applied to control in the process. Similarly to ANFIS model, data from process control were utilized to train and test with the model until getting the lowest error. Then these control parameters were applied to control in the process. ANFIS model was studied with 3 membership functions ; trapezoidal and triangular (Trap), Gaussian and bell shape. All approaches were compared with the existing process parameters. The results indicated that the proposed ANFIS_Bell model obtained with the best performance. Moreover, the proposed ANFIS_Bell model can reduce production defects approximately 5.2% when implemented with a case study factory. For inventory control application, the fuzzy input parameters ; demand and supply which are uncertain were applied for the inventory system. For FIS model, the developed fuzzy rules were utilized to find out the fuzzy order quantity continuously. The order quantity was adjusted according to the FIS model with the order quantity evaluation algorithm for the inventory model. The output of FIS model was also applied as data for FIS+ANN and FIS+ANFIS models. The FIS+ANFIS model was studied with 3 membership functions ; trapezoidal and triangular (Trap), Gaussian and bell shape. Inventory costs of the proposed models were compared with the stochastic economic order quantity (EOQ). The results represented that the FIS+ANFIS_Gauss model gave the best performance of total inventory cost saving by more than 75% compared to stochastic EOQ model. From the comparison in this study found that for the continuous data such as process control application, ANFIS model was appropriated to implement because of achieving the best performance of prediction accuracy and this model can reduce production defects. For discontinuous data such as inventory control application, FIS+ANFIS model was suitable to implement because of achieving the best performance of prediction accuracy and this model can reduce the inventory costs

Thammasat University. Thammasat University Library

Address: BANGKOK

Email: preserv@tu.ac.th

Name: Busaba Phruksaphanrat

Role: advisor

Created: 2015

Modified: 2021-10-27

Issued: 2021-10-27

วิทยานิพนธ์/Thesis

application/pdf

DOI: 10.14457/TU.the.2015.1549

eng

DegreeName: Doctor of Philosophy

Level: Doctoral Degree

Descipline: Engineering

Grantor: Thammasat University

©copyrights Thammasat University

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	4776prasert.pdf	9.73 MB	3	2023-02-24 13:33:07

ใช้เวลา

0.029629 วินาที