This study proposes a method to improve fractional-order particle swarm optimizer to overcome the shortcomings of traditional swarm algorithms, such as low search accuracy in a high-dimensional space, falling into local minimums, and nonrobust results. In natural phenomena, our controllable fractional-order particle swarm optimizer can explore search spaces in detail to obtain high resolutions. Moreover, the proposed algorithm is memorable, i.e., position updates focus on the particle position of previous and last generations, rendering it conservative when updating the position, and obtained results are robust. For verifying the algorithm’s effectiveness, 11 test functions compare the average value, overall best value, and standard deviation of the controllable fractional-order particle swarm optimizer and controllable particle swarm optimizer; experimental results show that the stability of the former is better than the latter. Furthermore, the solution position found by the controllable fractional-order particle swarm optimizer is more reliable. Therefore, the improved method proposed herein is effective. Moreover, this research describes how a heart disease prediction application uses the optimizer we proposed to optimize XGBoost hyperparameters with custom target values. The final verification of the obtained prediction model is effective and reliable, which shows the controllability of our proposed fractional-order particle swarm optimizer.

Aquaculture has long been a critical economic sector in Taiwan. Since a key factor in aquaculture production efficiency is water quality, an effective means of monitoring the dissolved oxygen content (DOC) of aquaculture water is essential. This study developed an internet of things system for monitoring DOC by collecting essential data related to water quality. Artificial intelligence technology was used to construct a water quality prediction model for use in a complete system for managing water quality. Since aquaculture water quality depends on a continuous interaction among multiple factors, and the current state is correlated with the previous state, a model with time series is required. Therefore, this study used recurrent neural networks (RNNs) with sequential characteristics. Commonly used RNNs such as long short-term memory model and gated recurrent unit (GRU) model have a memory function that appropriately retains previous results for use in processing current results. To construct a suitable RNN model, this study used Taguchi method to optimize hyperparameters (including hidden layer neuron count, iteration count, batch size, learning rate, and dropout ratio). Additionally, optimization performance was also compared between 5-layer and 7-layer network architectures. The experimental results revealed that the 7-layer GRU was more suitable for the application considered in this study. The values obtained in tests of prediction performance were mean absolute percentage error of 3.7134%, root mean square error of 0.0638, and R-value of 0.9984. Therefore, the water quality management system developed in this study can quickly provide practitioners with highly accurate data, which is essential for a timely response to water quality issues. This study was performed in collaboration with the Taiwan Industrial Technology Research Institute and a local fishery company. Practical application of the system by the fishery company confirmed that the monitoring system is effective in improving the survival rate of farmed fish by providing data needed to maintain DOC higher than the standard value.

During the machining process, the computer numerical control machine is susceptible to variations in ambient temperature, cutting heat, and friction within the transmission parts, which generate different heat sources. These heat sources affect the machine structure in different ways, causing deformation of the machine and displacement of the tooltip and workpiece position, ultimately resulting in deviations in machining accuracy. The amount of thermal drift depends on several factors, including the material of the machine components, the cutting conditions, the duration of the machining process, and the environment. This study proposes a hybrid optimization algorithm to optimize the thermal variables of computer numerical control machine tool spindles. The proposed approach combines regression analysis and fuzzy inference to model the thermal behavior of the spindle. Spindle speed and 16 temperature measurement points distributed on the machine are input factors, while the spindle's axial thermal error is considered an output factor. This study develops a regression equation for each speed to account for the different temperature rise slopes and spindle thermal variations at different speeds. The experimental results show that the hybrid thermal displacement compensation framework proposed in this study effectively reduces the thermal displacement error caused by spindle temperature variation. Furthermore, the study finds that the model can be adapted to significant variations in environmental conditions by limiting the machining speed range, which significantly reduces the amount of data needed for model adaptation and shortens the adaptation time of the thermal displacement compensation model. As a result, this framework can indirectly improve product yield. The effects observed in this study are remarkable.

This study proposes the use of an experimental design approach in GSO, and a systematic approach to deal with the hyperparameter settings of GSOs and to provide stable algorithmic performance of GSOs through the experimental design approach. To address these two issues, this study explores the combination of hyperparameters that can improve the performance of GSOs using a uniform design. In addition, the Taguchi method and optimal operations were used to derive an excellent combination of parameters that would provide the best value and robustness of the function to provide a stable performance of GSO. The validity of the performance of the proposed method was tested using ten benchmark functions, including three unimodal, three multimodal, and four restricted multimodal functions. The results were compared with the t-distribution test in addition to the mean and standard deviation to analyze their validity. The results of the t-distribution test showed that the p-values obtained for both UD-GSO and R-GSO were less than 0.05, indicating significant differences compared with GSO for both unimodal and multimodal functions. Two restricted multimodal functions are not significantly different, while the other two are below 0.05, indicating significant differences. This shows that the performance obtained using UD-GSO and R-GSO is more effective than the original GSO. UD-GSO and R-GSO provide better and more robust results than GSO. The main contributions of this paper are as follows: (i) This study proposes a uniform design approach to overcome the difficulties of setting hyperparameters in GSO. (ii) This study proposes a Taguchi method and optimal operation to provide a robust calculation for GSO. (iii) The method applied in this study provides systematic parameter design to solve GSO parameter setting and robust result obtaining.

To build a synchronization error prediction model for the machine tool efficiently, a robust whale optimization algorithm (RWOA) method proposed in this study is applied to the hyperparameter optimization of its model. The proposed RWOA method integrated a non-linear time-invariant inertia weighting (NTIW) method and a Taguchi-based adaptive parameter exploration (ATPE) to improve the performance of WOA and promote robustness. The NTIW method can improve the performance of other algorithms, so this study used the NTIW method in the WOA. In addition, the Taguchi method can get an excellent combination of variables with optimal values and stable performance, making the WOA robust. First, to verify the validity of the proposed RWOA method, 13 benchmark functions were used in this study. The results of the benchmark function tests include the mean, standard deviation, and p-value of the t-distribution test. The results show that 11 of the 13 functions differ significantly. In other non-significant difference functions, the means and standard deviations obtained by the proposed RWOA are considerably better than those obtained by WOA. Since the product cost of machine tools is higher, if a prediction model can be built effectively, it can reduce the cost. Therefore, in this study, the proposed RWOA was used to explore the best hyperparameter combination for the model. From the results, the model’s average MAPE (mean absolute percentage error) was 7.2604% for training data and 9.2603% for the testing data under 30 modeling runs. For the best one in 30 models, the MAPE was 6.8384% for the training data and 6.7372% for the testing data. This model was also introduced into the actual machine in this study, and the experimental results showed the MAPE 6.3447%. The proposed RWOA method effectively explores a suitable synchronization error model for the tool machine.