Abstract

With a service life of about 50 years, building structures are extremely durable capital goods. In concrete structures in particular, the quality features such as the loadbearing capacity or the durability, are determined by processes such as casting or concrete compaction that take place in an extremely short time period of a few hours. Therefore, the precise control of both the concrete quality - and here especially of the fresh concrete properties - as well as of the involved processes is of particular importance. The quality characteristics of fresh concrete are currently still determined empirically on the construction site. As a standard method, the flow table test is used to assess the consistency. In addition, the spread out fresh concrete is assessed visually. The use of digital quality control standards is currently failing due to the lack of appropriate sensor and measurement processes for recording and digitally evaluating the fresh concrete properties as well as corresponding control loops. In order to introduce such digital processes into concrete technology, suitable sensors and methods that enable an automatic and digital derivation of the fresh concrete properties during the quality control have to be developed. In order to overcome the limitations of the subjective visual inspection, an image-based method has been developed and validated in this thesis. This method allows a digital recording and evaluation of concrete properties as part of the quality control (flow table test). By combining digital image analysis methods (computer vision) and the use of deep neural networks, concrete properties can be predicted in real time from image data. In this way, the quality control of the fresh concrete is expanded to include a large number of additional parameters. Based on this evaluation, the data can be integrated into a digital control loop. The results of experimental investigations clearly show that the image-based method can be used to record and evaluate concrete properties as a part of the quality control (flow table test). As information which can be extracted from the image data and which can subsequently be used to derive the fresh concrete properties from, both, 3D surface parameters and geometric 2D properties as well as the reflection are suitable. Based on these parameters, the digital control loop offers the possibility to link the concrete production directly with the quality characteristics of the fresh concrete. Thus, Industry 4.0-standards can be established in concrete construction industry.