1. Introduction
Talc is a mineral occurring naturally in the form of crystalline hydrated magnesium silicate, with a chemical formula of M
g3SiO
10(OH)
2. Talc has low abrasion, high thermal conductivity and stability, low electrical conductivity, and high oil and grease adsorption [
1]. Due to its unique surface chemistry, lamellar crystal habit, and properties, talc minerals are widely applied commercially and industrial, such as in cosmetics, pharmaceuticals, paints, polymers, and ceramics. Furthermore, the method evaluated the suitable properties of talc, which can contribute to the industry in terms of the efficiency of production planning.
In the talc forming process, many factors, especially moisture, plays an important role. The more accurate the moisture forecasting, the better the quality of the talc pellet. Over the last decade, artificial neural networks (ANNs) have become a popular technique for data prediction due to their high accuracy. Loveday et al. [
2] apply an ANN to decrease the time for palm oil production. The developed model provides a reliable result and high efficiency. Adaptive neuro-fuzzy inference systems (ANFIS) is a kind of ANN usually applied in various fields to study, for example, economic order quality, water level prediction, medicine, and markets [
3]. The ANFIS is an algorithm that combines the advantages of both ANN and fuzzy inference systems (FIS). It has the ability to capture the nonlinear structure of a process and has a rapid learning capacity [
4]. In manufacturing applications, Caydas et al. [
5] developed ANFIS for the prediction of a wire electrical discharge machine. Zhange and Lei [
6] established ANFIS to predict the roughness of laser cutting and improve the quality level of laser cutting. Sen et al. [
7] utilized ANFIS for predicting machining performance parameters of Inconel 690. The result of all research shows that the ANFIS performs with high accuracy with respect to prediction.
In recent years, a hybridization of ANFIS with many optimization algorithms has been introduced to improve the forecasting accuracy of the traditional ANFIS. Abdollahi [
8] introduced a novel hybrid model for forecasting the Australian option price market. In a hybrid process, it consists of an entropy method and ANFIS trained by PSO. Bui et al. [
9] presented three new hybrid ANFIS with cultural, bee, and invasive weed optimization, namely, ANFIS-CA, ANFIS-BA, and ANFIS-IWO for flood susceptibility mapping (FSM). Yaseen et al. [
10] proposed a new hybrid ANFIS with the firefly algorithm for monthly streamflow forecasting. Gocken and Boru [
11] integrate the ANFIS with GA and harmony search (HS) in weather forecasting. However, GA and PSO are commonly combined with ANFIS. Oliverira and Schirru [
12] apply PSO for tuning ANFIS in sensor monitoring compared to ANFIS using one gradient descendent (GD) and GA. It found that the PSO applied in ANFIS gives the best result. Alarifi et al. [
13] combine PSO and GA with ANFIS to improve the prediction performance of the thermophysical properties of Al
2O
3-MWCNT/oil. Based on the result, they found that both of the ANFIS-PSO and ANFIS-GA models are able to predict the thermophysical properties appropriately. Kumar and Hynes [
14] predicted and optimized the surface roughness in thermal drilling by integrating ANFIS and GA. Rezakazemi et al. [
15] employed ANFIS with GA and PSO for the evaluation of H
2-selective mixed matrix membranes (MMMs). The results showed that the ANFIS with PSO is more reliable than the ANFIS with GA and the traditional ANFIS. Sabeti and Deevband [
16] introduced a novel training method of ANFIS by combining PSO and GA to solve the nonlinear dynamical system. The proposed PSOGA method provides the satisfactory results.
On the other hand, the appropriate collection of input data has a fundamental impact on the performance of the ANFIS model. It may lead to a better explanation of the results. There are only a few studies that apply the method in selecting the proper input data to ANFIS. Dariane and Azimi [
17] investigated appropriate input data selection in streamflow using GA and wavelet methods to deal with ANFIS applications. The results show that the performance of the model is improved when GA and wavelets are applied. Jeong et al. [
18] used a wrapper method to select the suitable input variables applied to a neuro-fuzzy model in monthly precipitation forecasting. The uncertainty in forecasting can be reduced effectively. The self-organizing map (SOM), first introduced in 1990 by Kohonen [
19], is an unsupervised learning method in ANN successfully established in data classification, pattern recognition, data compression, and data mining. It can reduce high-dimensional data to low-dimensional data. In manufacturing applications, SOM has been widely used for data clustering [
20,
21]. Some studies apply SOM to cluster the input data for ANFIS. Nourani et al. [
22] identified the input data for ANFIS by SOM and the wavelet transform groundwater level (GWL) and to fill the missing GWL data. The obtained results show that the proposed model can predict reliable accuracy. Amiryousefi et al. [
23] categorized the dataset into two clusters by SOM before feeding into an ANFIS model to predict the mass transfer kinetics in deep-fat frying (DFF). Nasir and Cool [
24] classified the input dataset by the SOM approach and combined it with an ANN or ANFIS. The results show that the proposed model makes a powerful intelligent model for wood machining monitoring.
The moisture in Uttaradit, Thailand is estimated by local traditional drying. It cannot determine the moisture of the talc mineral before the production process. The moisture is measured step-by-step depending on the user experience and it takes a long time to obtain the information. As mentioned previously, the novel techniques based on the combination of ANFIS with GA and PSO, using the SOM-clustering method introduced to improve the forecasting accuracy in the talc pellet forming process. The related theories are briefly described in
Section 2. The schematic and the procedures of the proposed model are indicated in
Section 3. The result description is therefore investigated and discussed in
Section 4. Finally, the overview of this study is concluded in
Section 5.
5. Conclusions
The hybrid model, based on a combination of SOM and ANFIS, is introduced as the proposed model for moisture prediction in the talc forming process in Uttaradit, Thailand. The GA and PSO algorithms are selected as the training algorithms of ANFIS. Five important factors—talc powder, water, temperature, feed speed, and airflow—affecting moisture in the talc pellet forming process were recognized and appropriate data were collected. In order to verify the proposed model, HM-GA, HM-PSO, ANFIS-GA, and ANFIS-PSO are compared. As a result, the HM-PSO model gives a high correlation coefficient for both training and test data with R = 0.9539 and R = 0.9192, respectively. Furthermore, HM-PSO still has a similar RMSE value for training and test data, of about 0.09. For other models, it has a rather large, different RMSE value between training and test data. Therefore, HM-PSO performs more reliably compared to the other algorithms. Since it is a real-world problem occurring in Uttaradit, Thailand, no one applies this method to the talc pellet process. The results, therefore, cannot compare with earlier research. The HM has some limitations according to the optimal parameters: It is only suitable for this study. For other real-world problems, it is necessary to identify the optimal parameters of HM. In this study, it can be said that the idea of raw data management by using SOM to identify a similar group of data is very helpful to obtain the most significantly information to feed into ANFIS. This can reduce the computation time during the training process. The method with clustering can improve the prediction skill compared to the method without clustering, efficiently. For further study, more clustering methods, such as k-means, and more ANFIS training algorithms, such as bee colony or ant colony optimization, should be applied and compared with this study.