This article explores an enhanced method for solving high-frequency forward problem Partial Differential Equations (PDEs) using Physics-Informed Neural Networks (PINNs). While PINNs have demonstrated success across multiple domains, they often face challenges in accurately capturing high-frequency components due to the inherent "spectral bias" of neural networks, where the network tends to focus on low-frequency features. To overcome these limitations, we introduce a novel approach that utilizes spline interpolation for signal augmentation and frequency decomposition, followed by transfer learning to progressively incorporate higher-frequency components into the learning process. The proposed method ensures smoother loss landscapes, making it easier for the network to learn high-frequency components. The proposed method is evaluated using a well-known forced mass-spring system, demonstrating the enhanced ability to capture high-frequency dynamics compared to conventional PINN methods.

Formal verification of PLC programs requires transforming control logic into a mathematical model. Since PLC languages lack a formal semantics, multiple transformation methods exist, based on various code interpretations. This work introduces a novel approach leveraging Single Static Assignments (SSA) to ensure a faithful PLC code transformation while enhancing model checking scalability. By systematically tracking variable assignments, the method accurately captures execution dependencies and preserves control logic. Experimental results demonstrate promising improvements in verification efficiency.

PLC programs play a major role in the operation of automated systems, controlling and monitoring various industrial processes. Formal verification methods, notably model checking, have been extensively studied to ensure the proper functioning of PLC programs. To implement this verification method, various formal modeling approaches have been proposed. However, most of them are often complex and require extensive expertise for implementation. An alternative is proposed, aimed at simplifying the formalization process.

Despite advancements in their performance, cyber-physical systems remain susceptible to cyberattacks. This paper focuses on the turnout system in railway transportation infrastructure. For such systems, the data monitored is geographically dispersed, which makes it vulnerable: attackers may either conceal failures or trigger unnecessary maintenance decisions. To address this issue, a threat assessment paradigm is proposed in order to provide maintenance operators with a threat indicator based on field data. The method is developed and validated over a real railway turnout data set.

Industrial Control Systems (ICS) are present in many fields, sometimes critical infrastructures, and perform complex functions to carry out their missions. However, since the beginning of the century, they have been the target of increasingly specific attacks whose consequences can be catastrophic. Thus, new security approaches at the border between IT security and safety have emerged to detect malicious activities in ICS. S.A.F.E. (Security Approach based on Filter Execution) is one of them as it enables the order analysis issued by the Programmable Logic Controller (PLC) based on behavioral models. Indicators are used to trigger automation-oriented detection mechanisms. One of these indicators, the distance, measures the minimal number of orders from the PLC to apply from the current state before reaching a prohibited state. This computation is based on the Dijkstra algorithm which is time consuming and complex. This paper presents a comparative study of different algorithms for optimizing distance computation in a cybersecurity context.

Since the beginning of the century, Industrial Control Systems (ICS) are sujbected to attacks targetting their actuator controls. Although active research has been done, most of the approaches fail to secure them from strikers with high automation knowledge due to their environment and requirements. This paper provides a ICS comprehensive view, outlines the existing works and presents the authors' cybersecurity vision as well as the approach developed by G-SCOP Center of Research. From its deficiencies, further investigation is pointed out by giving the orientation for designing behavioral model based Intrustion Detection System (IDS) based on the equipment degradation and addressing one of the limitation of the current approach on a single action chain.

This paper proposes a method for real time diagnosis against product quality drifts in an automated manufacturing system. We use Logical Diagnosis model to reduce the search space of suspected equipment in the production flow, which is then formulated as a Bayesian network to compute risk priority for each equipment, using joint and conditional probabilities. The objective is to quickly and accurately localize the possible fault origins and support effective decisions on corrective maintenance. The key advantages offered by this method are (i) reduced unscheduled equipment breakdowns, and (ii) increased and stable production capacities, required for success in highly competitive and automated manufacturing systems. Moreover, this is a generic method and can be deployed on fully or semi automated manufacturing systems.

This paper proposes an approach to accurately localize the origin of product quality drifts, in a flexible manufacturing system (FMS). The logical diagnosis model is used to reduce the search space of suspected equipment in the production flow; however, it does not help in accurately localizing the faulty equipment. In the proposed approach, we model this reduced search space as a Bayesian network that uses historical data to compute conditional probabilities for each suspected equipment. This approach helps in making accurate decisions on localizing the cause for product quality drifts as either one of the equipment in production flow or product itself.

This paper proposes an estimation method for the confidence level of feedback information (CLFI), namely the confidence level of reported information in computer integrated manufacturing (CIM) architecture for logic diagnosis. We studied the factors affecting CLFI, such as the measurement system reliability, production context, position of sensors in the acquisition chains, type of products, reference metrology, preventive maintenance and corrective maintenance based on historical data and feedback information generated by production equipments. We introduced the new ‘CLFI’ concept based on the Dynamic Bayesian Network(DBN) approach, Naïve Bayes model and Tree Augmented Naïve Bayes model. Our contribution includes an online confidence computation module for production equipments data and an algorithm to compute CLFI.