￼52<< INDUSTRY 4.0 There are numerous software on the market (NX-Siemens, Catia, Vericut, ...) that offer the possibility of performing a virtual simula- tion of the manufacturing process. The options of the simulation can change depending on the system but use to include the program, blank, xture, machine tool kinematics... In addition to the reduction of costs in terms of machine set-up, the virtual simulation tools allow a collision-free programming, which is a key factor in the high complexity multi-axis machining operations. Currently there are different ways to optimize the number of valid parts in a process, from predictive methods to direct measurement of manu- factured parts. The rst ones are based on predictive models based on process optimization in terms of parameter variation. These methods use to be based in off-line techniques, where the simulation is used as the start point of the process and helps to nd the best process parameters. However, it does not provide a real-time information about the process or interact with potential unexpected events. Moreover, it is usual that the modelling of manufacturing process requires a nal adjustment of the parameters based on trial and error techniques. A common alternative is to implement inspection operations during the manufacturing processes. The main advantage is the ability of error detection during the process, which avoids unnecessary nishing operations in parts that have been damaged during previous manufac- turing steps. On the other hand, the inspection of the parts during the manufacturing process only provides a binary type data, in which parts can be classi ed as valid or invalid, but there are insuf cient data to explain the reasons for scratches. Process monitoring is probably one of the most characteristic technologies involved in Industry 4.0 However, none of the previous alternatives offers a real-time control over the manufacturing process. Moreover, its application in short series manufacturing use to lead on expensive and complex solutions that rarely are implemented. In this context, the real-time control can be a solution. This type of control is applicable to detect complex phe- nomena such as vibration or thermal damage of the part. Even the different nature of the problems, the main idea is common for almost all the cases. First, it is necessary to know which are the more relevant parameters in order to design a mechanistic model. The model is used to relate the different parameters to the phenomena to be controlled and build a control algorithm. Therefore, the main idea is to modify the process parameters with enough time and accuracy to avoid potential defects. Secondly, it is necessary to install sensors that monitor the process and collect the necessary information to feed the control algo- rithm. Finally, due to the easy access to massive data storage capacity, all the recorded data of the sensors can be saved for future operations. In this way, smart control algorithms are being implemented, which are based on learning from events that are happening and on the basis that these events are likely to be repeated with a given pattern. The sensor to be installed in the machine for monitoring will depend on the type of parameter that needs to be measured. The most common ones are based on thermocouples, accelerometers, gauges... A pos- sibility for machining operation monitoring is the use of the cutting force signal, since they can predict the presence of vibrations, tool wear or excess of machined material among others. For its sensing, piezoelectric sensors are conventionally used because of the reliability Figure 4. Virtual simulation of a combined LMD and machining process [UPV/EHU]. Real-time process monitoring Process monitoring is probably one of the most characteristic techno- logies involved in Industry 4.0. In order to obtain the maximum amount of data from the processes, it is necessary to introduce different moni- toring and sensing systems. There are, in general terms, two different approaches to process monitoring. The rst is based on the massive data analysis. This approach needs to obtain as much data as possible, which will be processed in a next step. The second approach, based on a real-time analysis, is based on a direct analysis and interpretation of the obtained data. This approach requires a fast response system in order to store and compute the data and to give an immediate response to possible events that are detected by the sensors. In order to reduce the number of defective parts and what is known as ‘zero defect production’, Industry 4.0 philosophy tries to introduce both approaches of monitoring. On the one hand, real-time monitoring allows the changing of process conditions depending on the external variables. On the second hand, the massive data storage and off- line analysis can be used for maintenance and long term process optimization. Depending on the speci c manufacturing process, both approaches can be combined for obtaining the best results.