1 Introduction
Natural light is an important source of energy in green buildings. The effective use of natural light can improve the visual comfort of indoors and reduce the energy consumption of buildings. As a common shading device, venetian blinds have been widely used in indoor lighting control. However, the angle control of the traditional venetian blinds is mostly manual or electric, and it is necessary to rely on human judgment to change the angle of the venetian blind. The disadvantages are obvious. Therefore, the appearance of automatic venetian blinds overcomes the deficiencies of manual venetian blinds and electric venetian blinds, improving the indoor lighting performance.
In recent years, researchers at home and abroad have proposed a series of methods to intelligently adjust the angle of venetian blinds to obtain a constant desired illuminance, such as genetic algorithm [1], fuzzy control [2], neural network [3], adaptive fuzzy Control [4] and so on. However, in practical applications, if there is a lot of need to maintain a constant illumination, it is necessary to place a plurality of illumination sensors, not only the wiring is complicated, but also the system control is not convenient. In addition, with the use of luminaires, light output attenuation, dust deposition, and wall reflections all make it difficult to provide accurate control values. In order to effectively predict indoor natural illuminance, researchers have also proposed some prediction algorithms in recent years, such as genetic algorithm [5], artificial neural network [6 ~ 7], adaptive neuro-fuzzy inference system [8]. However, most of these algorithms are suitable for the indoor natural illuminance prediction of a single blind blind (Bindings), and the prediction of the indoor natural illuminance of the split blinds is worth exploring.
The segmented venetian blind system has better visual and thermal performance than the traditional venetian blind system. At some point, the venetian blinds are different in height depending on the height of the blades on the window, and the adjusted blade angles are different: The venetian blinds on the upper part of the window are inclined downwards in the indoor direction to reflect the natural light to the ceiling and deep inside the room to obtain a higher Illumination; The venetian blinds in the middle of the window are placed in a horizontal position to directly view the outdoor view; the venetian blinds in the lower part of the window are tilted downwards in the outdoor direction to prevent overheating in the room. This paper simulates the office space model through EnergyPlusTM software, and uses the adaptive neural fuzzy inference systems (ANFIS) to establish the Daylight illuminance prediction based on spiltblinds (DIPSB) model. Simulation analysis was performed in MATLAB.
2 Principles and methods
In this paper, there are three stages in total: First, the office space model is established in EnergyPlusTM software to obtain the illuminance data needed to establish the DIPSB model. Secondly, the AIPSIS model is used to establish the DIPSB model. Finally, the accuracy of the DIPSB model is verified by the illuminance percentage error. .
2. 1 Office space model establishment
EnergyPlusTM is a new building full energy analysis software developed by Lawrence Berkeley National Laboratory, University of Illinois, US Army Construction Engineering Laboratory, Oklahoma State University and other organizations with the support of the US Department of Energy [ 9, 10]. EnergyPlusTM can be used to simulate the building's year-round load and energy consumption. It uses an anisotropic sky model to more accurately simulate sky-scattering intensity on sloping surfaces. The office space model established in EnergyPlusTM software is shown in Figure 1.