Presently, many different types of machines and equipment are in operation at factories throughout the world. These include machines that process or assemble parts used in products, machines used for coating products, and countless others. Many of these machines are automated. However, the inspection process that checks the quality of final products still relies heavily on the faculties of human experts as well as on perceptions based on visual and auditory sensations. OMRON is now studying the possibilities of automating the inspection process using its Knowledge Information Control Technology.

Automotive engines are usually checked by inspectors who can identify a defect by distinguishing an abnormal sound or vibration. The ability to perceive highly subtle anomalies from the normal sound of engine operation and identify a cause of these anomalies is a task that can only be handled by seasoned inspectors with highly tuned intuitive facilities. This means that such experts must always be available, which means high fixed costs. Also, the inspection process and results can vary considerably if the staff changes. To counteract these problems, OMRON has employed its proprietary analysis technology to develop an abnormal sound detection system. The system can detect unusual sounds or vibrations by digitizing the waveforms of these sounds/vibrations and forming a knowledge base. Formerly, such a sound detection system required system engineers to manually analyze a sound that was previously identified by an expert as indicating a defective product. Only after that, the data could be input into a knowledge database for inspection.

OMRON's Knowledge Information Control Technology allows machines themselves to determine which characteristic elements should be extracted from which sounds to identify an abnormality, and which combination of sounds would make it easier for them to discover the causes of an abnormality. By processing an immense amount of data, a process very difficult for humans to handle, and learning from it, it will become possible to create a much more precise knowledge base than has ever existed before.

Besides automobiles, there are many other products and components that are inspected for abnormalities through the detection of sound. For example, PC hard disk checks and leakage checks for water and sewage systems have conventionally depended on the hearing capacity of experts. OMRON's Knowledge Information Control Technology enables this ability of experts to identify problems to be incorporated into machines.

As an example, OMRON's Signarc(*1) abnormal sound detection system makes full use of this technology.

The inspection of printed circuit boards for PCs and mobile phones has also relied heavily on the discerning vision of seasoned experts. A printed circuit board contains many components mounted on it through solder alloy joints. Performing detailed checks to assure correct connection of all components with no irregularities can be a very time-consuming task. Again, experienced operators with their accumulated expertise have traditionally been necessary for the determination of problematic processes and their accompanying solutions.

Now, this inspection work can also be semi-automated with OMRON's Knowledge Information Control Technology. Such a system uses a camera to record each process carried out until a printed circuit board is completed, then checks the recorded image and stores the data obtained. Should a problem be detected, the system will automatically locate the process that created the problem. If done manually, this process of checking and taking countermeasures has to be repeated until all processes run correctly, which means that it can be hours or days before the manufacturing line can be put back into normal operation. OMRON's Knowledge Information Control Technology makes it possible to solve the problem in a few hours or less. By drawing on the speed and other benefits of this technology, OMRON's "Q-up Navi"(*2) solutions are making great strides in quality improvement.

If human sense perceptions or expertise can be put into a database as knowledge, thereby enabling machines to make automatic judgments, it will be possible to maintain a consistently high inspection accuracy regardless of which inspector is used or in which country the inspection takes place. However, because production sites consistently encounter new situations that have never been experienced before, it is impossible for fully automatic machines to properly handle every situation that arises. This is why it is important that human operators and machines can interact more cooperatively. If a machine can warn an operator of problems beyond its experience, the operator in turn can teach the system to interpret the situation, enabling even smarter machines to be created.

Exploiting the Potential of Automated Machines

Return to Knowledge Information Control Technology Top