Observation of the Flow Pattern and Level Control of a Hopper by Using the Image Processing

Recently, the technique of image processing has been used for inspection of industrial products, remote sensing, chracter recognition, and as a visual sensor for robots, etc.. This report is concerned with level control of hopper by using an image processing apparatus. The experimental system consists mainly of the following elements : a) The controlled object is an asymmetrical hopper with a rectangular cross section. b) The detecting element is an ordinary video camera. c) The actuating element is a slide valve which is made in our laboratory. d) The controller is a PC-9801 personal computer. e) SYSTEM9512 is used as the image processing apparatus. f) For the test material, sand is used (mean diameter 355μm). The flow pattern of granules inside the hopper changes at the transitional bed height (H_<tr>). H_<tr> is the bed height at which flow pattern of granules changes from mass flow to core flow or vice versa. The flow pattirn of granuies is uniform in the upper part of H_<tr>. In this sense, it is easy to control the level of the hopper, because the surface of granules in the hopper is kept constant (features of mass flow). We measured H_<tr> by video camera and the data of H_<tr> is treated by the image processing unit (SYSTEM 9512). In this case, it is desired that the level control of granules is performed at the upper part of H_<tr>. The time to transmit the image data to PC-9801,and which is necessary for the image processing, was reduced to maintain good response wave shapes. In order to do so, the area used for data processing of the flow pattern is decreased, and the method of extracting the useful information from the image data is discussed. The following results were obtained through this experiment. 1) By using the level control system with an image processing unit as a detecting element, the control of level becomes easier. 2) There is a problem of relative position between video camera and hopper, so, to get good responses, we treated the problem of relative position by dividing the hopper into three visual areas. 3) It is observed that when inflow rate of granules and width of the hopper are kept constant, H_<tr> is almost constant. This means that the flow pattern is a mass flow state. Thus, H_<tr> becomes the index of control. 4) When inflow rate is constant and width of the hopper becomes wider. H_<tr> becomes greater and the area of uniform flow becomes narrower. Then controlling the level of the hopper is rather difficult in this case. 5) When Q_<in>=Q_<out'>, H_<tr> becomes lower than in the case of Q_<in>=0,and the region of uniform flow becomes wider. Controlling the level is rather easy in this case.