GKN produces, amongst many other products, Turbine Rear Frames (TRFs). The TRF is a part of a jet engine, and is manufactured by assembling multiple components together in a toroidal structure. Current production of TRFs is highly dependent on manual operations, while the company wants to decrease these manual operations and substitute them with automatic operations. The IFaCOM demonstrator at GKN focuses on demonstrating the automatic assembly and tack welding of components of a TRF section. In the final setup, different subsystems interact together in order to cover all the steps required for obtaining the optimal result at the assembly, from a prediction system using intelligent algorithms for suggesting element configuration, to a cell formed by two robots, welding tools and sensors that demonstrates the physical feasibility of the IFaCOM concept with a TRF section. The demonstrator shows technology for changing a manual assembly process of TRFs into an automated process where the quality requirements are validated and verified. The goal of the automatic assembly process is to achieve a Turbine Rear Frame (TRF) within the drawing tolerances for surface offset in between the different segments, minimizing residual elastic stress in the part and maximizing the plastic strain. The GKN demo’s system analyses and uses data from multiple sensors in upstream operations, during the assembly process and uses the data in the actual assembly tack welding as well as full seam welding. The new approach has a basis in collecting machining data (MAS), geometry data (MES), selecting the best fitting segments (ISP), collecting data from tack welding (ATW) and, if possible, from full seam welding (AP). This opens up for in-depth off line long-term analysis of the assembly process using knowledge from all the different steps of machining of segments (MAS), geometry of segments (MES), differential fit between selected segments (ISP), assembly sensor data (ATW) and if possible data from full seam welding (AP). The IFaCOM Zero Defect approach described above, consisting of three control loops, in the GKN case had the following targets: • For the short-term corrections, an automated system consisting of robots, sensors and a welding source is the central performer. The system assembles and tack welds aerospace Turbine Rear Frames automatically, measuring the process status in real-time and making decisions before applying the necessary corrections by analysing sensor measurements. • An Intelligent Stock Picker (ISP, described in details below) covers the medium-term corrections loop, by calculating the optimal configuration of segments that forms a TRF. Once those geometries are measured, data is analysed and optimal combinations of components for the assembly are suggested. The selected components are picked and placed in optimal positions where they are assembled by the automated assembly system. • The long-term corrections are performed by analysing the data collected by the sensors that are installed in the demonstrator. These data provides information on process trends and point to required improvements of the operation. Diverse software tools have been used in the demonstrator, being mostly developed and programmed during the project with different programming languages (mostly C++ and Python). The different software tools, are able to communicate among them and perform three main tasks in the demonstrator: • Intelligent sorting of components using Genetic Algorithms. The ISP is a genetic algorithm-based software tool that is able to calculate the different configurations of segments that can be assembled with the current stock. Geometrical values of the components in stock at GKN are input to the ISP, and the program calculates all the possible configurations that can be obtained, as well as calculating the manufacturing cost of each configuration and the number of bends. The goal is to minimize the residual stresses accumulated in the final TRF structure. For achieving this goal, the ISP calculated the configurations that minimize the bending of components; and selects components that need to be bent in regions that are easier for a robot to bend. • Real-time control of robotic assembly and tack welding operation. Controlling two robots by analysing the feedback provided by the different sensor systems in real-time is a demanding task that requires solid integration between all the different subsystems. This integration has been achieved by developing specific software that is able to control the different subsystems while orchestrating all the top-level operations needed for the system. The real-time robot control is implemented in the ROS (www.ros.org) framework, using the ROS Control project. ROS has provided a flexible framework for testing different robots and routines. Adjustment and tack welding operations require quasi real-time robot trajectory generations for adjusting the correct offset and welding the correct points. Every segment is different, and every point on the segments is located at an unknown position at the beginning of the operation. The ‘welding robot’, having a laser distance sensor and a welding torch attached to its end-effector, has to orient itself in order to measure the offset between the panels from the correct perspective. The welding torch has to be located strictly perpendicular to the point to be tack welded, as well as placed in the middle of the gap of the two parts that are to be tack welded. These conditions lead to the requirement for the robot orientation to be defined in real-time for every step of the operation. For this, the welding robot uses a laser distance sensor. The robot moves its arm, while the sensor collects profile data of the segments. The set of data is analysed by the PC and creates a virtual approximation of the profile position and orientation. The robot then is reoriented for getting the correct data and welding the points correctly. • Visualization by operator of operation status and progress (Graphical User Interface – GUI). The GUI gets updated in real-time with the information collected by the sensors in the robotic adjustment and tack welding operations. In the GUI is possible to visualize the time elapsed since the beginning of the process, the accumulated force, the total distortion introduced for bending the panels (up left). An overview of the full TRF status is also provided, with colours indicating the progress of the welding process (TRF picture down left, yellow – currently welding, green – done, white – not welded). Also, the operator can check the segment line progress (up right) and in details all the segment joint lines, identified with the weld number (table on the right). Finally, also the actual measurements from the laser sensors and the offset from the nominal profile are displayed in the picture on the right.
Not specified (see website if available)
Associated Key Wiki Items
- Not specified