Development of an algorithm in CONDITION MONITORING
As already mentioned in the introductory article "Cost reduction and efficiency increase with Condition Monitoring in industry", in the CMS development phase the specific case of each plant will be evaluated, beacuse each system is based on different cause and effect processes.However, there are macro levels to refer to that can be extended to most manufacturing companies. Different algorithms are suggested in the scientific literature as each component has specific failure modes, due for example to consumption, its geometry or dynamic conditions. As an example we can mention the ball bearings, one of the most common components in the mechanical field, whose failure modes can be traced back to the different parts that compose them. For example, due to the different times of consumption and periodicity of impacts, we can trace the fault back to a specific component rather than another.The choice to base the condition-monitoring on an algorithm means that it’s possible to refer to a series of systematic answers and therefore a response mode is associated with each error. This process allows monitoring the machinery conditions and reduce false alarms.The data flow starts from the acquisition of raw information, generally of limited size, often binary, through sensors and subsequently sent an output. This flow can be divided into three main classes:Data cleaning: preparatory procedures activated to remove inconsistent data, empty, damaged measurement files, disconnected sensors and malfunctioning cables etc.Fault Detection: includes all the appropriate procedures to recognize a failure in the system. In this phase output is the reporting of the presence of a fault, but not of the type. In some industrial fields this is enough if the component is to be replaced in its entirety.Fault Diagnostics: includes all procedures that characterize the error of a specific component and the level of component damage. Fault diagnosis techniques are also useful for redesigning a component: detailed fault knowledge may suggest a better design to reduce loads in working conditions, extending the expected life of the component.Implementing an algorithm is the first step to ensure that the whole condition monitoring system is functional to identifying failures and sending a response.
As already mentioned in the introductory article "Cost reduction and efficiency increase with Condition Monitoring in industry", in the CMS development phase the specific case of each plant will be evaluated, beacuse each system is based on different cause and effect processes.
However, there are macro levels to refer to that can be extended to most manufacturing companies. Different algorithms are suggested in the scientific literature as each component has specific failure modes, due for example to consumption, its geometry or dynamic conditions. As an example we can mention the ball bearings, one of the most common components in the mechanical field, whose failure modes can be traced back to the different parts that compose them. For example, due to the different times of consumption and periodicity of impacts, we can trace the fault back to a specific component rather than another.
The choice to base the condition-monitoring on an algorithm means that it’s possible to refer to a series of systematic answers and therefore a response mode is associated with each error. This process allows monitoring the machinery conditions and reduce false alarms.
The data flow starts from the acquisition of raw information, generally of limited size, often binary, through sensors and subsequently sent an output. This flow can be divided into three main classes:
- Data cleaning: preparatory procedures activated to remove inconsistent data, empty, damaged measurement files, disconnected sensors and malfunctioning cables etc.
- Fault Detection: includes all the appropriate procedures to recognize a failure in the system. In this phase output is the reporting of the presence of a fault, but not of the type. In some industrial fields this is enough if the component is to be replaced in its entirety.
- Fault Diagnostics: includes all procedures that characterize the error of a specific component and the level of component damage. Fault diagnosis techniques are also useful for redesigning a component: detailed fault knowledge may suggest a better design to reduce loads in working conditions, extending the expected life of the component.
Implementing an algorithm is the first step to ensure that the whole condition monitoring system is functional to identifying failures and sending a response.