Skip to main content

On-line Control of Drawing and Blanking Processes and of Quality of the Product by Fusion of Sensors and Artificial Vision Techniques

Final Report Summary - PRO2CONTROL (On-line Control of Drawing and Blanking Processes and of Quality of the Product by Fusion of Sensors and Artificial Vision Techniques)

Forming processes and, in particular, sheet metal blanking and deep drawing are inherently quite unstable manufacturing processes. On one hand, a slight change in the production parameters or in the material quality are usually enough to get the process out of its stability condition and to produce bad parts. On the other hand, when a tooling breakage or excessive wear occurs, the resulting parts are also bad. This situation is especially difficult to detect when producing small size parts, often manufactured in large quantities using high speed and production rate equipment. As a result, production of bad parts goes on until a statistical control is able to detect this and to stop the machine.

The main objective of this project was to develop a complete control system to be fitted on the ensemble press - tooling in order to assure a zero-defect in forming industries consecrated to the manufacturing of small size workparts. The aim was to get a complete diagnosis and control over the quality of the production and of the state of the production goods. As a result, the control system had to be able to integrate the capabilities of state-of-the-art control systems for forming processes and to fulfil the new objective of complete control of the quality of the part. Two complementary methods are considered: use of acoustic emission (AE) and load measurement technologies, which includes a sensor set mounted on the press and the tooling, the electronic needed for the conditioning and a monitoring system installed on a PC, and an artificial vision (AV) system, which is composed by a set of cameras and a flexible image processing system for the detection of defects in real time. The information of both systems is analysed by a software based control system, installed on the same PC, which has to decide the changes in the press state, the tooling condition or the process parameters needed to avoid the occurrence of defects.

The final strategic industrial objectives of the project PRO2CONTROL were to aim towards zero defects at the customer's side and to decrease in at least 25 % the scraps of manufacture, together with improving the worker condition, specially avoiding the manual quality control tasks. Economic objectives are to decrease the manufacturing costs of at least 5 % at the supplier which already use methods of surveillance of the press, and until 20 % at those that use only manual methods by minimising or eliminating costs due to the manual sorting, the costs of premature wear or cracking of tools and the costs of return and customers' scrap.

The project has concluded after 30 months of research activity and the initial objectives have been fulfilled. The first branch of the project was the detection of instabilities in blanking and drawing processes using force and acoustic emission sensors with the purpose of increasing as much as possible the efficiency of the production facilities.

Two monitoring systems have been installed in two different companies, Industrias Alzuaran (in a forming facility consecrated to blanking processes) and Industrias Garita (in a forming facility consecrated to combined drawing and blanking processes). In the forming facility at Industrias Garita consecrated to combined drawing and blanking operations very good results were achieved. The most critical process faults detected at Industrias Garita were tearing of the material, heating and therefore volume increment of the forming punches leading to catastrophic breakages and misalignment of the lateral dies at the tool leading to bad quality parts. At the same time the monitoring system installed in Industrias Alzuaran also offered good results. This monitoring system was prepared to work with three different references and up to nine process faults were detected. The most important process faults detected at the blanking references produced in Industrias Alzuaran were punch breakages, bad evacuated parts from the tool, bad evacuated scraps from the tool and feeding faults.

The second branch of the project was the on line detection of defective parts at the forming facilities using artificial vision techniques with two main purposes: first to avoid the sending of any defective parts to the customers and second to decrease as much as possible the internal production of defective parts. An artificial vision system composed of two intelligent cameras (based on FPGAs) has been developed and integrated into the blanking facility at Industrias Alzuaran. The vision system is able to check the quality of more than 100 parts per minute working at the manufacturing rate and therefore analysing the 100 % of the parts produced.

And third branch of the project was the development of intelligent controllers able to control sheet metal forming processes as human expert operators do but in an automatic way. The advantages of automatic systems over human beings are the achievement of more consistent answers for repetitive tasks, decisions and processes, efficiently and quickly and without any lack of performance because of pressure or tiredness.

So the main objective of the third branch of the project is to develop a control system able to interpret the information coming from both, the monitoring system and the artificial vision system, and make a diagnosis about the current process performance regarding the stability of the process during production and the quality of the parts produced. The system must be able to detect process faults at the forming facility and lack of quality of the parts produced, and to modify the settings of the forming facility in order to avoid such situations.

At the present project the feedback to the forming facility is given via human operators who can check the diagnosis of the expert system in a graphical user interface also developed at the present project. And finally the integration of all the previous developed system was made at Industrias Alzuaran. The final validation was made during the manufacturing of retaining rings in a 125 tons mechanical press using a progressive blanking tool. The production rate is more or less 60 strokes per minute and two parts per stroke are produced. Therefore the production rate is more or less 120 parts per minute. The raw material, CK 67 spring steel, is fed in the progressive tool from a sheet coil using an automatic feeder.

The implementation of an intelligent control system based on artificial intelligent techniques into a blanking facility has been developed. The intelligent control system has been linked to a sensor based monitoring system and to an artificial vision system and emulates the control of the operator in the blanking facility. Compared to human operators, the developed intelligent control system is able to react quicker and with higher precision and reliability.

The expert system has been running in the blanking facility during the last six months connected to the sensor based monitoring system. Up to now, four different references are controlled and nine different process faults have been found and implemented into the expert system. Among the incidences, the most important are: strip feed failures due to bad extracted parts, strip feed failures due to slugs of material inside the tool, misalignment of the strip inside the tool, clogging of the strip to the tool during withdrawal of the ram, detection of broken punches and detection of double parts inside the tool due to bad extracted parts. Besides this, the intelligent controller is able to detect the position within the tool where the incidence has happened. In this way the operator only has to consult in the graphical user interface of the expert system in order to know the incidence, its position in the tool, its cause and the solution.

At the same time during the last two months the expert system has also been connected to the artificial vision system and has been running in the production line. The artificial vision system was able to detect lateral defects like big localised burrs due to small micro cracks in the punches and evaluate the most important dimensions of the produced parts. With this implementation the 100 % of the parts produced at the forming facility are evaluated.

The percentage of defective parts produced at the blanking facility has been slightly decreased from a 0.1 % down to a 0.08 %, which means a 20 %. This decrement in the production of defective parts has been mainly due to the application of the monitoring system although the implementation of the expert system results also in a better resolution of the process faults what leads to a reduction in trials and production of defective parts when solving the process faults. The percentage of faulty parts sent to the customer has also been reduced because nowadays the 100 % of the produced parts are evaluated. Finally, one of the most important achievements of the implementation of the expert system is the reduction in time, and therefore in cost, when solving machine stops. With the introduction of this expert system the operator do not need to look for the fault in the facility but he directly finds in the interface of the expert system a message with the instructions to solve the problem.

Although no measurements have been made due to the complexity of measure this short of variables, it is estimated a reduction of about 50 % in the time that the operator uses to solve the machine stops in the blanking facility after the implementation of the expert system.

Related documents