AI - Implementation

The following is an example of how a digitisation project can be successfully implemented.

1. Auditing of the production system: In a first step, the respective production components to be monitored (What is to be monitored?) and the expected monitoring results (Which statement of the production monitoring system is expected in which form and which output data formats?
2. In the second step, the Grindaix GmbH analyses the feasibility of the customer's requirements on site at the customer's premises via the logical linking of various possible sensor signals at different points of use (accessibility, insertion, signal interfaces, ...) within the coolant system.
3. specially the signal interfaces have to be chosen very carefully. The Grindaix GmbH carries out a detailed Coolant Audit-E (electronics), in which the interfaces of all notified cooling lubricant system components are analysed for suitability and provides concrete recommendations for conversion (circuit diagram, parts list, hardware, software) exactly where the interfaces for the connection of a cooling lubricant monitoring system are not yet designed according to requirements.
4. Subsequently, the necessary interfaces are produced or used in the customer's production system. The sensors are integrated into the production system, wired and linked to a higher-level control centre (Condition Monitoring Pad).
5. The Condition Monitoring Pad is the control center of the production monitoring system and serves as the central recording of all installed sensor signals. On the software side, upper and lower limit values are defined here as of when a sensor signal value can be identified as a fault. So-called plausibility scenarios are then programmed, which, based on the logical relationship between the most varied production sensor signals, formulate statements that go beyond the individual condition monitoring of a production component (see for example the previously described - differential pressure loss in pipelines). Furthermore, the sampling rates are determined, i.e. how often a sensor signal is to be measured and recorded (saved) in order to enable later Production Data Tracking (GRX-PDT).
6. During commissioning of the system, test runs are performed which do not influence the production operation (status "read"!). Limit values are adjusted, series of measurements are recorded and analyzed for test purposes. The teach-in phase of the production monitoring system begins. In a subsequent interaction phase, the control mechanisms can be activated successively, depending on the customer's requirements, to enable adaptive control of the production components. Once this process is completed, the production monitoring system is accepted. The employees are trained in the handling of the monitoring system and instructed in the maintenance possibilities.

It must be avoided at all times that the sensor integration described here makes the production system too susceptible to possible measurement errors. For this reason, the monitoring system can initially operate in "read-only" mode and transmit error messages to a higher-level control system without intervening in the production process. Further monitoring systems specifically control a wide range of variables within the production unit in order to maintain the coolant supply required by the customer at all times.

Interfaces to machines as signal-setpoint transmitters (e.g. pressure requirement cycles on frequency-controlled pumps) must almost always be created. In most cases these are not available and the system supplier does not allow any intervention in his system electronics without giving a warranty. Therefore, such interfaces must be provided additionally either via Profinet/bus interfaces (bus systems) or by means of isolation amplifiers to ensure galvanic isolation of the electrical signals, either by suitable service providers or by the system supplier itself.

Galvanic isolation or decoupling means avoiding the electrical line between two circuits between which power or signals are to be exchanged. This prevents any influence on the current circuit of the signal transmitter (machine tool, filtration system, pump, ...). This additional effort leads to a higher investment volume and initially represents an obstacle to the implementation of digitalized system technology. However, if the advantages can be calculated, a concrete payback period can be determined without the use of artificial intelligence.