Contract-based Verification of Digital Twins

Muhammad Naeem, Cristina Seceleanu · 2025

Read Contract-based Verification of Digital Twins by Muhammad Naeem, Cristina Seceleanu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-10993.txt.

Introduction

Muhammad Naeem and Cristina Seceleanu set out a way in 2025 to check that a digital twin behaves as it should. A digital twin is a virtual copy of a real machine or system, often built with a neural network. Checking such a copy is hard, since it can rest on huge data sets and its inner steps are not open to view.

Their method treats the twin as a black box and tests it against written contracts that state what the system must do. A tool runs the twin on chosen inputs, then checks the outputs against a contract model built from timed state machines. It does this on its own, with no need to open the network. The work gives engineers a way to prove a twin meets its stated rules before it guides real equipment.

Catalog record

CREATORMuhammad Naeem, Cristina Seceleanu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE41 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-10993

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Contract-based Verification of Digital Twins Muhammad Naeem, Cristina Seceleanu Abstract Digital twins are becoming powerful tools in industrial applications, offering virtual representations of cyber-physical systems. However, verification of these models remains a significant challenge due to the potentially large datasets used by the digital twin. This paper introduces an innovative methodology for verifying neural network-based digital twin models, in a black-box fashion, by integrating model checking into the process. The latter relies on defining and applying system-level contracts that capture the system’s requirements, to verify the behavior of digital twin models, implemented in Simulink. We develop an automated solution that simulates the digital twin model for certain inputs, and feeds the predicted outputs together with the inputs to the contract model described as a network of timed automata in the Uppaal model checker. The latter verifies whether the predicted outputs fulfill the specified contracts. This approach allows us to identify scenarios where the digital twin’s behavior fails to meet the contracts, without requiring the digital twin’s design technicalities. We apply our method to a boiler system case study for which we identify prediction errors via contract verification. Our work demonstrates the effectiveness of integrating model checking with digital twin models for continuous improvement. 1 Introduction The emergence of Industry 4.0 has propelled Digital Twins (DT) to the forefront of industrial digitalization, offering virtual representations of cyber-physical systems that facilitate precise simulations, analysis, and control [ 19 ] . As industrial systems grow in complexity, ensuring the reliability and correctness of these DT becomes crucial. This work is a part of the Dynamic and Robust Distributed Systems (D-RODS) project aiming to develop an adaptable and dependable framework for the implementation and operation of DT, tackling issues in industrial digitalization including system integration, performance enhancement, and compatibility with legacy systems [ 17 , 8 ] . In the current study, we focus on the Verification and Validation (V&V) of Neural Network-based Digital Twins (NNDT) models using model-checking techniques. NNDT have the potential to predict system behavior, conduct real-time tests, and identify optimization opportunities. However, their effectiveness depends on their ability to represent and accurately respond to real-world conditions [ 7 ] . Recent advancements in model-driven engineering, such as the CoCoSim framework, have demonstrated the value of integrating formal verification techniques with model-based design, especially for multi-periodic discrete systems [ 3 ] . Our research applies a similar concept to the domain of DT, proposing a novel approach that employs model checking to verify monotonicity, functional, and infrastructure contracts specified for black-box NNDT. We propose a methodology for systematically verifying the behavior of DT models implemented in Simulink by defining system contracts, as networks of Uppaal timed automata [ 1 ] . We present an automated solution that connects a simulation environment with model checking, allowing for the continuous validation and improvement of NNDT models. Our approach detects situations where the NNDT generates incorrect outcomes. Due to their dynamic nature and complex interactions with cyber-physical systems, verifying DT models is challenging. Dahmen et al. emphasize the importance of systematic methods to verify and validate DT models [ 5 ] , and propose an approach that breaks down the verification problem into smaller, more manageable components. This research aligns with our methodology of using (sub-)system contracts to verify specific aspects of DT behavior. Given the importance of ensuring the correctness of DT behavior, our research aims to address the following research question: RQ: How can we design an automated solution to model check a black-box NNDT model? To answer it, this paper presents a contract-based approach for verifying black-box NNDT models in Uppaal . The solution simulates the Simulink NNDT model for certain inputs, and feeds the predicted outputs together with the inputs to the contract model described as a network of timed automata in the Uppaal model checker. The latter verifies if the predicted outputs fulfill the specified contracts. In brief, our contributions are as follows: • Model-Checking-based Methodology: We propose a methodology that uses contracts modeled as Uppaal timed automata, and employs model checking to verify the correctness of black-box NNDT. • System Contract Implementation: We define a systematic approach for modeling system contracts that specify expected behaviors, enabling precise verification of model outputs against system requirements. • Practical Validation: Our approach is illustrated through a burner-boiler system case study, showcasing its effectiveness on an industrial application. Our approach facilitates the identification of scenarios where the DT’s behavior fails to meet the contracts, without requiring knowledge of the model’s internal structural and functional details. The remainder of this paper is organized as follows. Section 2 provides the necessary background information. Section 3 details our methodology for verifying NNDT models using contract-based model checking. Section 4 describes our case study of a burner-boiler system, including the DT model implemented in Simulink , as well as the system’s contracts. Section 5 presents the implementation of contracts in Uppaal , for model checking, and Section 6 discusses the verification results. Finally, Section 7 presents and compares to relevant related work, before Section 8 concludes the paper, summarizing our contributions and discussing potential lines of future work. 2 Preliminaries This section presents the background information necessary to understand the concepts and methodologies employed in our study. 2.1 The D-RODS Approach The D-RODS project, proposed a cutting-edge approach for development and operation of DT. This innovative framework aims to address the complexities inherent in industrial digitalization by integrating artificial intelligence (AI) and formal verification techniques. Context: Physical (CP) represents the physical components including machineries and controllers. Context: Learning (CL) layer (Fig. 1 a)) focuses on creating DT models through unsupervised learning. These models are derived from extensive data histories and existing domain expertise. Figure 1 : D-RODS approach details Context: Functionality (CF) and Context: Infrastructure (CI) are similar layers (Fig. 1 b, c), each containing AI and V&V components. These contexts supervise and enhance the development and execution of complex DT, facilitating continuous learning, optimization, and behavior evaluation. At its core, D-RODS seeks to unite DT, AI, and V&V technologies in a novel architectural setup. This integration aims to increase the trustworthiness of AI approaches through formal verification and analysis while optimizing operations, resource utilization, and power consumption. The framework is designed to support high levels of autonomy by ensuring the accuracy and efficiency of employed models through continuous learning and verification. D-RODS is committed to improving system performance forecasting and optimizing resource allocation in the field of AI. The project also aims to integrate formal verification and runtime testing to ensure the accuracy of developed models and continuously monitor operational correctness. Through these advancements, D-RODS aims to advance DT technology, offering a comprehensive framework that addresses the challenges of complex industrial systems while promoting efficiency, reliability, and adaptability. 2.2 Uppaal Timed Automata Model checking is a formal verification technique that systematically and exhaustively verifies whether a system model meets specified requirements. Uppaal [ 1 ] is a state-of-the-art model checker designed for modeling, verification, and simulation of distributed systems. Its modeling framework is based on Timed Automata (TA), which extends traditional finite-state machines by incorporating real-valued clocks. These clocks allow for the precise timing of events, enabling the modeling of systems where timing is critical.

Read the complete text