which must be accurately tested for the specific dataset
AI, artificial intelligence; AR, augmented reality; GPU, graphics processing unit; IR, infrared; IRT, infrared thermography; PLC, programmable logic controller.
An important parameter to check the measurement accuracy of monitoring sensors, is the signal‐to‐noise ratio (SNR) defined as:
(1.2)
where PSignal is the average signal strength, PNoise is the noise strength, ASignal is the root mean square (RMS) of the signal amplitude, and ANoise is the RMS of the signal noise. The SNR parameter can be increased by applying AI as in modern computer numerical control (CNC) machine systems, which are improved by accurate control systems based on ANNs predicting for example surface roughness during tool processing by sound and vibratory signal analyses [50, 51]. The sensor data analyses allow, by means of AI, the prediction of machine failures and malfunctions thus increasing the SNR.
1.2.6 Infrared Thermography in Monitoring Process
The processes controlling the temperature represent many times a fundamental element to guarantee an optimal quality of the worked piece. The temperature check is performed directly on the processed piece or is related to the working environment as for a temperature‐controlled room. Excessive rise in temperature may generate thermal damage in the processed materials. Specifically, monitoring the tool's temperature range is of fundamental importance. A function of IRT is to check during the time the correct processing as for cutting, welding, or surface treatments. The basic physical principles of an IR thermal camera in monitoring processes are indicated in Figure 1.10 (a), indicating all the signals involved for a radiometric measurement which are:
(1.3)
(1.4)
(1.5)
(1.6)
(1.7)
where C + D + E is the total radiated power, σ is the Stefan–Boltzmann constant, εTarget is the emissivity of the object to detect (target), τa is the environment transmittance, TTarget is the temperature of the object, and Ta is the environment temperature.
Figure 1.10 (a) Infrared thermal camera signals. (b) AOV and FOV simplified definitions.
The timing of the measurement acquisition and the movement control of the thermal camera are two relevant factors for object temperature checking. Concerning camera systems, the angle of view (AOV) and the field of view (FOV) in degrees are, respectively, expressed in Figure 1.10b as:
(1.8)
(1.9)
The AOV is a measurement (in degrees) of how much of an object is viewed through the lens, and is measured horizontally, vertically, or diagonally. The FOV is a measurement of object distance and it requires the knowledge of the distance from the optical center of the lens to the object to detect. In a 3D space the FOV is defined horizontally, vertically, and diagonally.
1.2.7 Key Parameters in Supply Chain and AI Improving Manufacturing Processes
AI in manufacturing processes [52, 53] follows the production in different forms, including defect classification, defect prediction, sales prediction, assisted production, RE optimization, automated processing, layout and warehouse optimization, DB security predicting network attacks (cybersecurity risk prediction), logistic optimization, and in general supply chain improvements estimating Key Performance Indicators (KPIs). Important KPIs are:
Count (the amount of product produced in the factory by the last change of machine or the sum of production of the whole shift or week).
Rejection ratio (production of waste affects profitability goals).
Production rate (different speed of machines and processes producing goods).
Production goal (target values for production outputs).
Cycle time (the amount of time for the completion of a task).
Overall equipment effectiveness indicating the efficient utilization of available personnel and machinery.
Idle time (period in which machines are not operating).
Life cycle of a product.
AI can predict the KPI values thus suggesting decision in a BI key and achieving the main goals indicated in Table 1.8.
Table 1.8 Goals achievable in manufacturing industry.
| KPI goal | Description |
|---|---|
| Safety and environment | Number of accidents at work, number of alarms, consumption, waste, recycling of material, etc. |
| Efficiency | Saved materials and resources, saved energy of production lines, improvement of services, maintenance optimization, less production shutdowns, decrease of production time |
| Quality | Percentage of finished product that does not meet the quality criteria, percentage of semi‐finished and raw products that do not meet the quality criteria, size of production losses and waste, internal and external services, etc. |
| Production plan tracking | Production time traceability, logistic layout optimization, improvement of warehouse management, programming efficiently picking processes, etc. |
| Employee's satisfaction | Completed works on time, lost workdays, innovation proposals, product/process innovation proposals, etc. |
In order to improve the KPI, the real issue is to integrate AI in the whole supply chain by following the main scheme of Figure 1.11. The supply chain is characterized by a large number of different variables. A solution to optimize the supply chain is a dynamic monitoring controlling in real time the whole production system. In this way, AI and IoT solutions, by processing digitized data, improve dynamically real‐time production processes: if there is a prediction of sales of a particular product, the production line is addressed by the AI directly for the new strategic product