Tung-Chien Chiang

A novel control for a nonlinear two-dimensional (2-D) overhead crane is proposed. Instead of the complex design procedures used in classic methods, the proposed scheme combines the principles of neural networks (NNs) and variable structure systems (VSS) to derive control signals needed to drive the cart smoothly, rapidly and with limited payload swing. The merits include the robustness and model-free properties of the sliding mode and neural based controllers, respectively. Simulations performed using a scaled 2-D mathematical model of the crane confirm the effectiveness of the proposed method.

Encoders are generally used to track the motion of industrial mechanisms. However, the information obtained by encoders may have errors due to encoder aging or mechanism-design problem. Therefore, information by visual feedback is a better way to track the movement of industrial mechanisms. However, image information costs lots of computing effort so it is not easy to be used in real-time control applications. This manuscript derives a simple but effective visual feedback method to follow the target and the image information is obtained only by a general handy camcorder. Besides, the proposed method can track multi-locations in a meantime. Fast image pattern recognition and localisation of the colour histogram by using a moving tracking block is applied to increase the calculation speed. Finally, the obtained locations information by the proposed visual feedback method is applied in an industrial crane control system to verify the effectiveness.

This study proposes a simple and effective method, termed Principal Component Analysis (PCA) method, to analyze ECG signals for effectively determining the heartbeat case. This method is easily performed and does not require complex mathematic computations. The average time required for processing a 30-minute long of ECG data is less than 1 minute, and the required maximum memory is only about 10 MB. The ECG records available in the MIT-BIH arrhythmia database are utilized to illustrate the effectiveness of the proposed method. The experiment results show the total classification accuracy was approximately 90.85%.

Heat pumps are typically controlled by a programmable logic controller (PLC). However, PLCs are generally expensive and large. This work uses an MCS-51-based controller to control a heat pump, so that the size of the controller of a heat-pump system can be reduced markedly. Moreover, the proposed MCS-51-based controller can detect the status of each heat pump’s sensor automatically, so that a user can know whether a sensor is normal or abnormal, thereby protecting the heat-pump compressor. Knowing the number of a faulty sensor also helps during maintenance. Experimental results demonstrate that the proposed MCS-51-based controller can replace a PLC-based controller and control a heat pump successfully, achieving an inexpensive and effective solution for controlling heat pumps.