Qass responds to the challenges of digitisation by developing new technologies and applications. With us, the user can shape digitisation and not just experience it. One particular example is the topic of sensor and information fusion. Here, an undefined number of sensors and other data streams are linked together. This is done with consideration of a comprehensible visualization of data and feature extraction for Smart Data. The aim is to improve the significance of the respective process behaviour by combining signals that cannot be clearly assigned and referenced to each other to produce a reliable statement. Now QASS plays out its strengths in data analysis, building high-performance measurement hardware and software programming.
Qass is a manufacturer of piezoelectric and magneto-inductive sensors, which are actively connected to the main unit via a preamplifier measuring chain. Up to four measuring channels (multiplexing) can actively record sensor data simultaneously and subject them to a spectral analysis with 40 µs per Fourier step. The strengths of the system also lie in feature extraction by mathematical calculation and the use of adaptive electrical filters to ensure reliability even in harsh industrial processes. External sensors can also be connected to the interface of the preamplifier and thus benefit from the strengths of the instrument. However, a completely new approach is to use the built-in computer interfaces of the measuring device, because data can now also be read in via USB, network or serial connections, as well as Profi-Net. It is irrelevant from which source the data originates, whether, for example, a bypass of an existing measuring device, a data stream for accompanying logging of the component or a sensor. In general, from this point on, the Qass Analyzer program looks at the data alone and brings it to an adaptive evaluation.
Big Data Analysis
Qass has created two levels for this adaptive evaluation, which turn the measuring device into a universal tool for data analysis and sensor fusion.
The first level is the Qass operator network. An operator is a small program that operates a specific functional task within the measuring instrument. There are three classes of operators. Class 1 accesses the basic functions of the measuring instrument, which mainly concern the storage and provision of data. Class 2 already serves for data analysis. The range of functions of the provided operators is already very diverse and basically includes pattern recognition based on the spectral data. This involves searching for taught-in referenceable objects of the spectral landscape in new data streams. One of the main advantages is the reliable elimination of interference. Class 3 is mostly used for visualization and adaptation of the data, as well as for machine communication. The operator elements can be combined to a data flow model and interactively adapted to the respective measurement task. The class 1 operators obtain their data from the described interfaces. For each sensor or data stream, a separate data analysis is created, which in turn can be linked in cascade to a common evaluation.
If you are still not satisfied with the existing operators, you can activate them yourself at the keyboard. With the topic of scripting, Qass has created the possibility of accessing the functions of the encoder with its own self-created program code.
Special options for adaptation
There are Java script and Python script operators, each of which allows the strengths of the respective programming language. While Java script is object-oriented and can be programmed realistically, Python is more aspect-oriented and functional and allows the integration of libraries such as TensorFLow. This adds the capability of machine learning based on real sensor data to the measurement device. The practical advantages are obvious. In the past, Qass needed months to program a new application for a specific application, but now it can be done within a few days. Thanks to the fast and clean adaptation to the sensor and data content of the production process, more and more applications for the universal measuring device Optimizer4D can be created in such a short time.
The graphic elements of the Cooperate Design can be used, which represents the second level of adaptive evaluation. Under the name PENGui, graphic elements can be created that allow easier access to the first evaluation level, which is sometimes complex. It is up to the user to decide which elements to create. Qass has attached great importance to simplified access to complex measurement technology so that the data can be understood and not interpreted first.
Successful integrations are already the use of current clamps in welding processes and their representation in the spectral analysis with direct statement about the quality of the MIG/MAG welding process under combined use of structure-borne sound or the product QASS QUBE. This is a measuring device for the detection of oil-dependent risk factors in large, self-sufficient drive trains via infrared spectrometry, as well as accompanying interference electricity and detection of early damage in bearings due to structure-borne noise.
QASS continues to research
Our next vision is the establishment of predictive algorithms to improve the reliability of machines. Many machines have “predictive signals” that allow risk assessment and allow the addition of new signals or special behaviour to increase their operational readiness. The data is analyzed centrally on servers and countermeasures are recommended to the user.