The three operations, measurement, decision and action, are always present in every type of control. Measured are ambient conditions, in our case the internal illumination, global and reflected solar radiation and the current position of the roller blind. On the basis of the measurements, the controller decides what to do to follow the desired (set point) inside illumination. As a result of the controller’s decision, the system must take an action. This is in our case accomplished with the suitable alternation of the roller blind position. Variable
solar radiation is called a disturbance in the process, because it causes the deviation of the controlled inside illumination from the set point value. Because of the external disturbances, such as changeable solar radiation, the automatic process control for the window geometry alternation is justified. The illumination algorithm was designed and developed progressively during the research procedure. The design of the light control loop was based on experimentation. The first set of measurements was used to examine the relation between the outside solar radiation and the inside daylight illumination with different surfaces of the window and in changeable weather conditions. After getting the experience in this sphere the control algorithm was developed and improved progressively with the aid of experiments. The control algorithm is created in the IdR BLOCK environment, where placing and interconnecting various blocks define the control scheme. To each block a subprogram, which performs the necessary operation of input data, is assigned. Blocks are grouped into block groups, called loops, and they are framed into control algorithm. The proper functioning of the control algorithm is of essential importance for the adequate adjustment and velocity of alternations of the roller blind with regard to external disturbances, i. e. solar radiation changes. The assessment of the control algorithm is subjective. We observed, how the alternation of the window geometry — roller blind positioning — influences the internal lighting and luminous efficacy with respect to the given solar radiation.

For adaptable window geometry fuzzy logic [7,8,9] is a system advantage in controlling, because of the fuzzy controllers’ ability of non­linear mapping between the ambient conditions and the corresponding roller blind position. In fuzzy systems the advantage of reduced complexity is taken into account and it is achieved with subjective estimation of the used information. The principle, which derives directly from human reasoning, is based on a linguistic model, which is expressed with a set of conditional rules, IF-THEN statements. We want the movable shading device to be alternating, as if it was adapted manually to the internal demands and external conditions. We have to find a linguistic model for the optical process as a basis for the fuzzy controller, expressed with a set of fuzzy rules. With the determination of IF-THEN statements and fuzzy sets the input and output relations for each fuzzy controller are determined. The inputs are in our case on-line measured external and internal conditions and the output is an appropriate roller blind position. The designer must select appropriate fuzzy sets with membership functions for all input and output variables (Fig 2).

The input and output numerical values must belong to appropriate membership functions, which are correlated and combined with the aid of logical fuzzy in IF-THEN statements. With the aid of the linguistic terms the measured values are described as sets of ordered pairs: numerical values and degrees of memberships to fuzzy subsets. The correlation, non-linear mapping between two inputs and one output, is graphically represented with a specific 3D shape as shown in Fig. 3. The actual numerical output value, the current position of the roller blind, is located on the 3D shape. The output values depend on the inputs, measured condition values and on the designed fuzzy model.

The control algorithm is defined as cascade control (Fig 4). Fuzzy controller is used as the main controller while PID is the auxiliary one.