In the first segment of this series of articles, we mentioned the main interest behind wireless vibration monitoring solutions: to integrate, at a lower cost, more equipment into planned maintenance.
Going wireless multiplies the number of sensors installed, with one obvious constraint: autonomy. It would be unfortunate if a technology that was supposed to save time for maintenance crews required them to spend hours replacing batteries. However, long-term autonomy (a minimum of one year in an industrial environment) implies a low consumer network. Exit WIFI and 4G, enter loT networks (LPWAN or others). These networks have at least one characteristic in common: a much lower data flow than we are accustomed to in our daily lives and in wired real time monitoring solutions.
The constraint of autonomy therefore requires reducing both the amount of data sent from the sensors to the central server as well as the intrinsic weight of the data. As we explain in detail in our white papers,“Predictive maintenance: towards a change of paradigm?”, wireless results in a change of paradigm: you have to make choices.
Wireless forces us to structure an expert approach in data acquisition by looking for the right information in the right place at the right time. This type of approach will help make well-informed decisions. We are thus moving from a “Big Data” logic, which still prevails in the minds of many as the solution to establish predictive models of equipment malfunction, to a “Smart Data” approach.
The good news is that this new strategy for data acquisition is both good for industrial performance and for the planet. It generates gains faster in both OEE and maintenance, based on the experience and algorithms developed by experts of vibration analysis, and it acts as a form of digital ecology by limiting, for example, the size of the servers necessary to store data.
However, what is a “smart” acquisition strategy when talking about wireless vibration monitoring? It is first and foremost a differentiated strategy which focuses on deciding the right measurement interval for each piece of equipment (neither too short nor too long) as well as the right basic indicators (representative of the equipment’s health status and evolution).
It is also an adaptive strategy which is based on embedded intelligence to collect (heavier) diagnostic data only when necessary. In other words, the sensors regularly send the basic indicators on the basis of the defined interval. When the system detects a change in status within these indicators, it automatically chooses to either reduce the measurement interval as to track the equipment more closely, or to initiate a thorough measurement in order to carry out a diagnosis. Be careful: in order for the system to automatically adapt its actions, it is important that the communication within the network can be done in both directions (from the sensor to the gateway and vice versa). This is not always the case.