A comprehensive survey on optimizing deep learning models by metaheuristics

Deep neural networks (DNNs), which are extensions of artificial neural networks, can learn higher levels of feature hierarchy established by lower level features by transforming the raw feature space to another complex feature space. Although deep networks are successful in a wide range of problems in different fields, there are some issues affecting their overall performance such as selecting appropriate values for model parameters, deciding the optimal architecture and feature representation and determining optimal weight and bias values. Recently, metaheuristic algorithms have been proposed to automate these tasks. This survey gives brief information about common basic DNN architectures including convolutional neural networks, unsupervised pre-trained models, recurrent neural networks and recursive neural networks. We formulate the optimization problems in DNN design such as architecture optimization, hyper-parameter optimization, training and feature representation level optimization. The encoding schemes used in metaheuristics to represent the network architectures are categorized. The evolutionary and selection operators, and also speed-up methods are summarized, and the main approaches to validate the results of networks designed by metaheuristics are provided. Moreover, we group the studies on the metaheuristics for deep neural networks based on the problem type considered and present the datasets mostly used in the studies for the readers. We discuss about the pros and cons of utilizing metaheuristics in deep learning field and give some future directions for connecting the metaheuristics and deep learning. To the best of our knowledge, this is the most comprehensive survey about metaheuristics used in deep learning field.

     

Design and implementation of a low-cost multi-channel temperature measurement system for photovoltaic modules

An efficient and low-cost temperature logging system with a 16-channel input was developed for measurements of photovoltaic module temperature. This paper reports the principle of operation, design aspects, as well as the experimentation and performance of the simultaneous temperature measurement of 16 solar cells/modules. The system consists of a 16 channel multiplexer, a 12 bit A/D, a differential amplifier and NTC temperature sensors. The temperature range of the sensor is from −20 °C to 120 °C. The simplistic design requires no large internal memory to store data but incorporates a high degree of sensitivity and dynamic range (according to climate condition), thus the cost of the design remains low and makes it suitable for field applications. The system was successfully tested for the operating temperature of a 40-cell mono crystalline Si photovoltaic module under realistic outdoor conditions during a summer and a winter day. The temperature Instrumentation developed for avoidance of special interface card use enabled the successful collection of data from long distances with negligible level of noise.     

Performance comparison of a double-axis sun tracking versus fixed PV system

In the present study, performance results of two double axis sun tracking photovoltaic (PV) systems are analyzed after one year of operation. Two identical 7.9 kWp PV systems with the same modules and inverters were installed at Mugla University campus in October 2009. Measured data of the PV systems are compared with the simulated data. The performance measurements of the PV systems were carried out first when the PV systems were in a fixed position and then the PV systems were controlled while tracking the sun in two axis (on azimuth and solar altitude angles) and the necessary measurements were performed. Annual PV electricity yield is calculated as 11.53 MW h with 1459 kW h/kWp energy rating for 28 fixed tilt angle for each system. It is calculated that 30.79% more PV electricity is obtained in the double axis sun-tracking system when compared to the latitude tilt fixed system. The annual PV electricity fed to grid is 15.07 MW h with 1908 kW h/kWp for the double axis sun-tracking PV system between April-2010 and March-2011. The difference between the simulated and measured energy values are less than 5%. The results also allow the comparison of different solutions and the calculation of the electricity output.