The problems of early detection of the machines’ forthcoming malfunction (a part of predictive maintenance) are usually solved by planned inspections or production shutdown. Sometimes the quality control of e.g. air-conditioning, gearboxes etc. also depends on the ability of well-trained and costly professionals.
The Czech SME has developed an algorithm leveraging latest technologies in machine learning and Artificial Intelligence (AI). The company focuses on the early detection and prediction of mechanical malfunction on any machinery. The system monitors different components of machines in order to recognize unpredictable anomalies or known issues such as overheating, damaged bearings, knocking rollers, damage in exhaust pipes, loosening of the engine bolts, etc. The unique pre-processing of the input data allows the neural network to learn and identify the important features quickly with high confidence. The maintenance team is warned by text message or e-mail notification typically 2-7 days in advance.
The target industries are manufacturing, energy, oil and gas, automotive and others.
By monitoring acoustics and vibration of machines via AI algorithms, this service allows for continuous monitoring of critical assets and provides insights for preventive maintenance. Full end-to-end system consisting of sensors, Internet-of-Things (IoT) devices and cloud-based back-end platform for quick deployment.
Some steps in manufacturing and assembly are done by workers or robots that are unable to make quality checks while machines are running. Yet, the quality of these steps or events can be heard. The solution offered can evaluate whether these steps have been completed successfully or not. End-of-Line (EoL) testing is usually done by human operators or a combination of sound recording systems and human supervision. Human means error and unpredictable cost. Using AI algorithms, the solution offered improves acoustic EoL testing by fully emulating human intuition about product sounds and acoustics, transferring essential operator experience and quality system data into the algorithms. The AI-driven method can enhance the accuracy of these checks. This allows for standardized, auditable results, free from possible human errors.
In order to allow high-resolution audio recordings anytime and anywhere the SME has also developed its proprietary mobile recording gadget. This device is a shielded sound card that connects to a mobile phone and allows to record and mark suspicious audio files, upload them to the Czech SME cloud for analysis and check results in real-time.
Another option how to record the sound is a standalone recording and computing IoT device that can be installed to enable audio monitoring of any asset. The abundant edge-computing power enables possibility to run complex machine learning models without a need to transfer data to cloud.
The SME is looking for cooperation based on commercial agreement with technical assistance. The subject of the agreement will be delivering the offered solution (including proprietary mobile recording gadget), consulting how to install the solution, advising clients how to properly operate it and provide services (analysis of results).