華梵大學

This study develops a behaviorally informed evaluation model for Internet of Things (IoT) development platforms by integrating human-centered decision perspectives with a hybrid multi-criteria decision-making (MCDM) approach. Through an extensive literature review and expert consultation, this study identifies and structures the critical factors influencing users' (customer satisfaction) and developers' (strategy selection) perceptions of IoT platform performance. Three stages of purposive surveys were conducted to gather the judgments of experienced IoT specialists. The fuzzy Delphi method (FDM) synthesizes expert judgments to reduce 30 sub-criteria to 16 key factors through fuzzy-based consensus. The analytic hierarchy process (AHP) then converts subjective assessments into consistent and quantifiable weights of the sub-criteria to ensure hierarchical rigor. Finally, VIKOR identifies a compromise solution that balances conflicting criteria and yields robust platform rankings of four representative IoT platforms. This study finds that customer satisfaction outweighs strategic considerations in IoT platform evaluation, with Security and Privacy ranked as the top two critical criteria, followed by Technology, Implementation, Convenience, and Market. Operations optimization, Trust and verification, and Customization emerge as the most influential sub-criteria. Benchmarking four IoT platforms identifies performance gaps and provides a self-assessment tool to optimize resources, align technical performance with user perceptions, and support evidence-based policy.

This paper investigates a new design method for PDtype iterative learning control (ILC). Specific design and adjustment approaches are proposed separately for the P-type control learning gain and the D-type control learning gain in order to accelerate the convergence rate of the learning error. In particular, the adjustment of the D-type control learning gain adopts an intelligent gain adjustment rule based on a fuzzy system. In addition to theoretical proof and analysis, a numerical example is conducted and verified through simulations using Matlab software. Furthermore, the control algorithm is implemented as a digital circuit using the Verilog hardware description language. After its correctness is verified on the Quartus II platform, the design is realized on an FPGA chip to demonstrate its feasibility.

For a robotic manipulator executing a repetitive carriage task, the primary target is to pick up an object from one position and place it at the intended position. In this circumstance, the manipulator is required to only pass certain key positions. This forms a multi-point-to-point trajectory tracking problem for which the controller only stimulates the manipulator at the predetermined instants. This paper considers an iterative learning control scheme for a nonlinear system with unknown parameters under the premise of minimizing the performance index, which is composed of the output tracking error and the rated input increment. For the optimized control scheme, the unknown system parameters are iteratively updated to obey two criteria: to minimize the output estimation error and the rated parameter estimation increment and to minimize only the output estimation error. A theoretical proof shows that the output tracking error is convergent along the iteration axis. Two numerical simulations are provided to demonstrate the effectiveness of the proposed learning scheme.

An inductively coupled plasma chemical vapor deposition (ICPCVD) technology was used to deposit a silicon germanium oxide (SiGeO) film on fused silica (FS), and thermal poling was applied to it. It was found that at a higher poling temperature, this sample could produce a stronger second harmonic generation (SHG) signal than FS under the same poling conditions. The development of this nonlinear optical material will help develop high-performance, low-cost silicon photonic devices.