Analysis of effective variables on daily electrical load curves of Iran power network

Artificial intelligence implementation framework development for building energy saving

“…The following three problems are proposed as the unmet demands of AI studies: Lack of a systematic AI implementation methodology for effective energy saving at three building levels.As illustrated in Section 1, none successfully integrated AI at the equipment, facility, and whole building levels. A systematic methodology is expected to solve the first unmet demand and implements AI at three levels for effective building energy saving. A method is needed to demonstrate the energy‐saving effects of AI to building owners.According to References [2–114], 31.9% of studies specifically presented data for energy savings or energy cost savings as percentages. Among these studies, 30.6% used simulations to illustrate the energy consumption differences before and after implementing AI, whereas 69.4% reported results based on experimental data.…”

“…The first category is computational intelligence at equipment level. Main control tools are different NNs 2‐29,110‐112 and Fuzzy control 30‐41,113 . Unlike conventional digital or computer control systems that can only perform one‐to‐one control output, Fuzzy tools can incorporate multiple variables into a single feedback sensor, thereby implementing a many‐to‐one control strategy.…”

“…According to References [2–114], 31.9% of studies specifically presented data for energy savings or energy cost savings as percentages. Among these studies, 30.6% used simulations to illustrate the energy consumption differences before and after implementing AI, whereas 69.4% reported results based on experimental data.…”

“…After neural network (NN) or called artificial NN (ANN)-based training, these tools enable the input of multiple values with a single answer output, which is advanced to existing control systems. Represented research studies include ANN-based management of renewable energy sources 2 and optimal control of a heating system in a building 3 ; an ANN tool to predict daylight illuminance in office buildings 4 ; an ANN model development to predict hourly heating energy consumption of a model house 5 ; using ANN to predict performance of wet cooling tower 6 ; a bilevel timetable optimization by ANN 7 ; the comparison between model simulation and ANN for forecasting building energy consumption 8 ; an adaptive NN for chiller data analysis 9 ; NN tools for short-term energy-consumption prediction and power supply feedback control [10][11][12][13][14] ; a spiking NN (SNN) to construct intelligent control, hierarchical architecture, and a multiple agent system (MAS) 15 ; an SNN tool for energy-consumption analysis 16 ; an SNN for daily heating load prediction 17 ; a convolutional NN for short-term load forecasting 18 ; a recurrent NN for load prediction 19,20 ; an RNN for control optimization 21 ; a ward-type NN for air temperature prediction 22,23 ; and that for cooling tower approach temperature prediction 6 ; advanced NN with a modified learning algorithm 24,25 ; a wavelet-based NN for optimization of scheduling control 23,24 ; a supporting vector machine and regression (SVM&R) combined with NN for predicting heating and cooling loads 26,27 ; a simulatedbased NN for forecasting electrical energy consumption 28 ; NNs for demand side management 25 ; and feedforward NN-based indoor-climate control framework development for thermal comfort and energy saving in buildings. 29 Variou...…”

“…AI developments have provided a series of solutions for the whole building. However, a literature survey of 113 studies 2‐114 revealed that energy conservation control for the equipment, facility, and entire site/building accounted for 23.9%, 42.5%, and 33.6%, respectively. None of these studies have successfully integrated AI for saving three‐level building energy conservation, and this may be due to the lack of implementation frameworks.…”