Electronics in Advanced Research Industries. Alessandro Massaro
integration plays an important role in the upgrading of production processes. Technological approaches able to transform the production into an auto‐adaptive system are horizontal, vertical, and end to end integration [62]. Figure 1.15 defines all the steps involved in the horizontal and vertical integration in Industry 5.0. Specifically, horizontal integration concerns the solution in adaptive robotics, the AI and AR implementations, and the hardware and software tools used for the production upgrade. Besides, vertical integration is mainly oriented on the supply chain including raw material management, production management, manufacturing process applications, assembly processes, logistics and commercial strategies.
Figure 1.15 Scheme of horizontal, vertical and end to end integration in Industry 5.0.
Among the tools most used by companies to improve production processes are computer‐aided manufacturing (CAM) and CAD. The use of these tools makes it possible to significantly speed up the development cycle thus, accelerating the response capacity and product adaptation to preferences and needs of customers. Furthermore, CNC machines can be managed by an AI engine by optimizing the production processing. 3D printing and additive manufacturing [63], complete the full scenario of SM facilities in modern fabrication processes where machine scheduling and process simulation are optimized by AI [64]. Furthermore, ANNs are able to predict the breakage of mechanical components and their degradation, permitting formulation of innovative processes [65]. Concerning CNC processing, the CAD tools are directly interconnected with the machines, and, by means of a feedback system, automatically adapt the working machine parameters, by controlling working tolerances (see the basic feedback system of Figure 1.16). The feedback system is improved by the AI engine able to control production processes, and change production processing according to the optimized CAD design.
Figure 1.16 CAD and CNC interconnected by a feedback system.
Different data mining tools such as Rapid Miner, Weka, R Tool, Konstanz Information Miner (KNIME), and Orange Canvas [66] implement AI algorithms by means of programmable objects using GUIs. The GUIs represent an easy way to integrate the AI tools into the information system. The objects, named blocks or widgets, are programmed to extract data from the DB, and to provide outputs useful for the control and the actuation. Moreover, the GUIs are suitable for parameter setting: each object is a node [67] representing a specific function of data pre‐processing, data processing and the output stages interconnected to the PLC. Each node is interconnected to other ones by structuring a data workflow representing the AI algorithm to run. The auto‐adaptive solutions are executed by automatically updating the dataset to process, and by planning the automated execution of the AI workflow by cron. The interventions useful to upgrade the transformation of the production optimizing automation are:
Adding new sensors monitoring critical production point to check.
Changing some of the old machines of the old production line with new ones, addressing the production for an intelligent actuation.
Adding new robotic elements able to develop new functions such as the in line automatic rejection of defects.
Using smart local PCs able to process locally production data.
Setting thresholds alerting conditions enabling alarms.
Definition of priority electrical loads important to ensure continuous production.
Interconnection and synchronization of PLC systems of the same production line.
Upgrading the technology adding possible P&P enabling Simple Object Access Protocol (SOAP) services.
Upgrading the ERP integrating AI and advanced BI algorithms.
Implementation of a big data system suitable for data fusion and data collection.
Introduction of new robotic and mechatronic systems based on a new concept of control.
Upgrading the CAD design tools evaluating dynamically the working tolerances.
Advanced RE and rapid prototyping.
Possibility to launch production process simulations by introducing supervised workflow representing a specific production.
Adopting AI algorithms integrated in the BPM tool as process mining.
Introduction of AR technology supporting technology transfer and assembly processes.
Development of the vertical, horizontal and end to end integrations by means of the new technologies.
Improvement of the product quality.
Upgrading the IT infrastructure.
Implementation of auto‐adaptive algorithms supporting production.
Possibility to switch dynamically the production adapting the marketing in the new market where supply and demand are constantly changing.
Using new technologies in order to decrease costs.
Improvement of data security.
Using CAM tools upgrading RE approaches.
1.5 Basic Concepts of Artificial Intelligence
AI in industrial applications is a discipline to aid data analysis, capable of self‐learning from the data, to predict and classify data, and to formulate KPIs. The prediction is referred to a machine failure, or in more extensive cases to the evolution of production malfunctions. AI contributes to main areas in sustainable manufacturing such as:
supply chain management
predictive maintenance
quality control
energy consumption optimization
Of industrial interest is the self‐classification of images classifying and predicting production defects. AI algorithms, as for deep learning (DL), can simultaneously process multiple variables with different “calculation weight.” The AI models are specific for the application and therefore are not generic; the training dataset to be analyzed must be chosen accurately. A particular AI algorithm is the ANN. A neural network is composed of:
Input layer (level designed to receive information from outside in order to learn to recognize and process the same information received).
Hidden layer (layer connecting the input level with the output level, and helping the neural network to learn the complex relationships analyzed by the data; often there is more than one hidden level as for DL neural networks).
Output layer (final layer showing the results of what the algorithm has processed).
Each connection between neurons is associated with a weight determining the importance of the input value (input variable xi). The initial weights are set randomly. Each neuron, on the other hand, is characterized by an activation of a mathematical function. Figure 1.17a shows a basic neural network and Figure