Nonlinear technological processes are characterized by strong dynamic coupling, multivariable interactions, and significant uncertainty caused by time-varying operating conditions and external disturbances. These characteristics substantially limit the performance of conventional control strategies, such as classical PID and fixed rule-based controllers, which often lack sufficient adaptability and robustness in complex industrial environments. Even conventional fuzzy controllers may exhibit degraded performance when operating regimes change rapidly or when multiple competing control actions must be evaluated simultaneously. Consequently, the development of advanced intelligent control approaches for nonlinear and uncertain systems remains a critical challenge in modern automation and control engineering. This paper proposes a quantum-inspired fuzzy inference-based intelligent control framework for nonlinear technological processes. The proposed approach integrates a fuzzy inference system as the core decision-making mechanism with quantum-inspired computational principles, including probabilistic state representation, parallel evaluation of alternative control actions, and adaptive weighting mechanisms. Unlike true quantum computation, the proposed method employs quantum-inspired concepts at the algorithmic level to enhance decision flexibility and robustness while remaining fully compatible with classical real-time control platforms. Fuzzy inference enables the incorporation of expert knowledge and linguistic uncertainty, whereas the quantum-inspired structure allows simultaneous assessment of multiple control scenarios within each control cycle. The controller is implemented in a closed-loop architecture and coupled with a dynamic nonlinear process model. Real-time process measurements are used to adapt inference parameters online, while quantum-inspired weighting reinforces favorable control actions and suppresses suboptimal ones. Simulation results demonstrate faster transient response, improved disturbance rejection, and higher operational efficiency compared to classical PID and conventional fuzzy controllers. The proposed methodology is applicable to a wide range of automation and control systems, including energy conversion and thermal processing units.

Modern automation and control systems applied in energy-conversion units, mechatronic platforms, and industrial processes increasingly operate under conditions characterized by strong nonlinear dynamics, parametric uncertainties, and time-varying external disturbances. These factors significantly degrade the performance of conventional control approaches, including PID and rule-based intelligent controllers, particularly in terms of robustness, transient response, and adaptability. Consequently, the development of advanced control frameworks capable of systematically addressing uncertainty and nonlinear behavior remains a critical challenge in automation and control engineering. This study proposes a quantum-inspired photon–spin control framework for robust automation of nonlinear dynamic systems operating under uncertain conditions. The controlled process is modeled using nonlinear state-space equations representative of automation-oriented dynamic plants. System states are mapped into a photon–spin probabilistic representation, where spin-like state variables and photon-inspired energy encoding enable a superposition-based description of multiple possible system behaviors within a unified control-oriented structure. This representation allows parallel evaluation of alternative control actions under uncertainty. A quantum-inspired inference and decision mechanism based on interference-driven selection logic is employed to identify the most probable optimal control action from the superposed state space. The selected control signal is decoded and applied to the plant through classical actuators, forming a hybrid quantum-inspired/classical feedback control loop that remains fully compatible with standard automation hardware and numerical simulation environments, without requiring physical quantum devices. Simulation studies conducted under significant parametric variations (up to ±15%) and external disturbances demonstrate that the proposed framework achieves approximately 25% reduction in steady-state control error and a 20-30% improvement in transient performance compared to conventional intelligent control strategies. The results confirm the effectiveness and robustness of the proposed approach for next-generation automation and control applications operating under uncertain and dynamically varying conditions.

Drone-based inspection has become an effective tool for improving safety and efficiency in construction applications; however, designing flight paths that balance coverage, inspection resolution, and limited flight time remains challenging. Conventional path planning approaches typically apply uniform flight patterns and fixed image resolutions across entire construction sites, leading to redundant scanning in low-complexity areas and insufficient inspection of critical regions. This paper presents an adaptive drone path planning framework for construction applications that integrates fractal image processing with deep learning-based hazard detection.

The proposed approach first captures a preliminary image of the construction site and applies a fractal quadtree algorithm to partition the site into regions of varying spatial resolution based on visual complexity. These partitions are clustered into multiple altitude levels, enabling resolution-aware path planning in which drones are deployed at different heights to efficiently inspect regions with distinct complexity requirements. High-complexity areas are assigned finer resolutions and lower flight altitudes, while low-complexity areas are inspected at coarser resolutions from higher altitudes.

To enable automated safety inspection, a YOLO-based deep learning model is employed to identify construction hazards from images captured by drone-mounted cameras. The detection model is trained offline using labeled construction site imagery and is capable of recognizing multiple hazard types under varying environmental conditions. Simulation results using real construction site images demonstrate that the proposed method significantly reduces the number of required scan locations compared to traditional random walk and zigzag flight patterns while maintaining sufficient image quality for reliable hazard detection. The proposed framework provides an efficient and scalable solution for adaptive drone-based construction site inspection.

Continuous stirred tank reactors (CSTRs) are challenging to control because of their nonlinear dynamics and the strong interaction between concentration and temperature. These challenges become more pronounced when operating conditions vary and reaction kinetics are uncertain, which often limits the effectiveness of conventional control strategies. In this study, a data-driven predictive control approach is developed for a generic nonlinear CSTR by integrating a Long Short-Term Memory (LSTM) neural network within an MPC framework. A benchmark exothermic CSTR described by coupled mass and energy balance equations with Arrhenius-type kinetics and jacket heat exchange is used as a reference process. Reactor concentration and temperature are selected as the state variables, while the coolant temperature serves as the manipulated input. The first-principles model is employed to generate operational data and to evaluate closed-loop performance. Dynamic simulation data are generated over 300 min with a sampling time of 0.5 min and cover multiple operating regions. Disturbances in feed concentration (±10%) and feed temperature (±5 K), together with 1% measurement noise, are introduced to reflect realistic operating conditions. An LSTM network with two hidden layers of 32 units each is trained to perform multi-step prediction of reactor states. On unseen test data, the model achieves root-mean-square errors of approximately 0.02 kmol/m³ for concentration and 2.0 K for temperature. The trained LSTM is embedded into an MPC scheme with a prediction horizon of 10 steps and explicit input and temperature constraints. Closed-loop simulations indicate that the proposed LSTM-based MPC improves set-point tracking and disturbance rejection compared with conventional PID control and nominal model-based MPC, while achieving reduced overshoot and faster stabilization. The results suggest that data-driven predictive control provides a practical alternative for nonlinear CSTR systems when accurate mechanistic models are difficult to obtain.

The control of biogas production in anaerobic digestion systems is inherently challenging due to pronounced nonlinear dynamics, biological uncertainty, long time delays, and limited availability of reliable online measurements. In response, a range of advanced control strategies—including model-based, data-driven, and artificial intelligence-assisted approaches—have been proposed to enhance methane productivity while maintaining process stability. Nevertheless, a systematic and quantitatively grounded comparison of these strategies from an application-oriented perspective remains limited. This study presents a comparative assessment of novel biogas control strategies using a Multi-Criteria Decision Analysis (MCDA) framework. The evaluated alternatives include fuzzy supervisory control, adaptive neuro-fuzzy inference systems (ANFIS), mechanistic model predictive control (MPC), data-driven MPC employing machine-learning predictors, reinforcement learning-based control, and hybrid architectures that integrate soft sensors with intelligent supervisory layers. Eight evaluation criteria were defined to reflect the requirements of full-scale anaerobic digestion systems, including stability and risk prevention, methane productivity, constraint handling capability, robustness to feedstock variability, sensor practicality, implementability in PLC/SCADA environments, explainability, and lifecycle effort. The MCDA results indicate that hybrid strategies combining soft sensing with supervisory control achieved the highest aggregated performance score (0.82 on a normalized scale), followed by fuzzy (0.76) and ANFIS-based (0.74) supervisory controllers. MPC-based strategies exhibited superior constraint handling performance (criterion scores above 0.85) but were comparatively penalized due to higher modeling and implementation effort. The reported literature suggests that data-driven predictive control can improve methane yield by approximately 5–10%, while intelligent supervisory control supported by soft sensors may reduce acidification risk indicators by 20–30% relative to baseline operation. Reinforcement learning approaches demonstrated high theoretical optimization potential but the lowest industrial readiness. Overall, the proposed MCDA framework highlights hybrid intelligent control architectures as the most balanced solution, offering a practical compromise between performance enhancement, robustness, and deployability in biogas production systems.

The mining sector is undergoing a profound digital transformation driven by the integration of intelligent mining systems that enhance operational efficiency, safety, productivity, and ultimately sustainability. Despite the growing global adoption of smart mining technologies, their uptake within South Africa’s hard rock mining industry remains uneven and underexplored. This study investigates the underlying factors enabling the adoption of intelligent mining systems in South Africa’s hard rock mining industry using exploratory factor analysis. Employing a quantitative research design, data were collected through a structured questionnaire administered to actively practising mining professionals in South Africa. An exploratory factor analysis was employed to uncover the latent structures underlying the observed adoption variables. The findings reveal three distinct, statistically robust enabler clusters that collectively shape intelligent mining adoption. These include continuous awareness and knowledge development, enabling regulatory framework and public acceptance, and governmental incentive and support. The extracted factor structure demonstrates strong internal consistency and explanatory power, providing empirical evidence of the multifaceted nature of intelligent mining adoption in a developing nation context. This study contributes to the smart mining and Industry 4.0 body of knowledge by shifting the focus from barriers to actionable enablers, offering a nuanced understanding of the conditions necessary for successful digital transformation in the hard rock mining space. The results provide valuable insights for mining firms, policymakers, and technology providers seeking to accelerate the deployment of intelligent mining systems in South Africa and comparable global mining jurisdictions

Harsh and extreme marine and environmental conditions have a great impact on offshore energy systems, oil and gas platforms, and wind farms. Extreme waves, corrosion from seawater/microorganisms, and the remote nature of these facilities demand continuous monitoring to prevent costly total shutdowns or accidents caused by corrosion and wave-induced vibrations. Condition monitoring of offshore facilities involves using multi-source data fusion from sensors to enable real-time data analytics, coupled with AI models to obtain facility insights. Digital twin provides integrated real-time analysis, alarms, and an AI-based approach that represents a paradigm shift in condition monitoring for offshore facilities by offering a virtual replica that can simulate real-time degradation.

We reviewed traditional methods and compared them to recent advances that fused physics-based models and AI into a digital twin. We review technologies such as blockchain, CNN-based algorithms, physics-augmented AI, physics-informed neural networks (PINNs), graph neural networks (GNNs), federated learning, and their suitability in creating offshore facilities digital twins. We investigated digital twin architectures, implementation and integration strategies, challenges, and provided a forward-looking roadmap.

Our review reveals that traditional approaches often struggle with multi-fault complexity and data corruption from harsh environments, leading to higher false positives/negatives in fault detection and limited predictive accuracy. In contrast, digital twins integrating physics-informed models achieve superior predictive performance, due to their robust handling of noisy/multi-source data. The digital twins approach reported lower root-mean-square errors and better generalisation in corrosion-fatigue and structural fatigue scenarios, leading to improved forecasting accuracy for unplanned downtime, maintenance costs, and the useful lives of facilities and equipment. Our review also highlighted a research-to-practice gap in the scalability of the proposed solution, data privacy, and cross-operator data-sharing capabilities.

Predictive maintenance enabled by condition monitoring and fault diagnosis (CMFD) has emerged as a critical operational capability for improving equipment reliability and production continuity. However, empirical evidence based on real operational data linking CMFD to performance outcomes remains limited within the operations and supply chain management literature. This study examines how condition monitoring intensity and fault diagnosis effectiveness influence maintenance efficiency, unplanned downtime, and operational performance using secondary industrial datasets from manufacturing environments. Drawing on maintenance strategy and operations performance theory, a causal framework is developed connecting CMFD capabilities to productivity, cost efficiency, and service reliability through maintenance effectiveness. The analysis employs panel regression and mediation techniques to evaluate performance changes associated with predictive maintenance interventions across multiple equipment units and time periods. The findings demonstrate that enhanced monitoring frequency and improved fault detection accuracy significantly reduce unplanned downtime and maintenance costs while increasing throughput and delivery reliability. Maintenance effectiveness is shown to mediate the relationship between CMFD capabilities and operational performance outcomes. This research contributes to operations and supply chain management literature by empirically establishing predictive maintenance as a strategic operational capability rather than solely a technical tool. Managerially, the results provide evidence-based justification for investments in CMFD technologies as drivers of operational efficiency, resilience, and sustainable performance.

The condition monitoring and early fault diagnosis of motor drive systems play a critical role in ensuring the reliability and availability of industrial robots operating in smart manufacturing environments. Motor drives in robotic applications are exposed to nonlinear dynamics, variable mechanical loads, and progressive degradation, which significantly limit the effectiveness of conventional model-based fault detection and diagnosis (FDD) methods due to parameter uncertainty and unmodeled dynamics. This paper proposes a data-driven fault detection framework based on a hybrid Convolutional Neural Network–Recurrent Neural Network (CNN–RNN) observer for the continuous condition monitoring of motor drive control systems in industrial robots. The CNN component enables the automatic extraction of fault-sensitive features from multichannel sensor signals, while the RNN component captures temporal dependencies associated with fault evolution and degradation processes. The observer-based structure allows residual-like information to be implicitly learned from operational data without requiring an explicit analytical model of the system. The esults demonstrate that the proposed CNN–RNN observer achieves a fault detection accuracy of 98.4%, outperforming conventional observer-based diagnostic approaches by 12.5%. Moreover, incipient faults are reliably detected up to 350 operating hours prior to critical failure while maintaining a false-positive rate below 1.2%. These results confirm the effectiveness of deep learning-based observers as a practical solution for the condition monitoring and predictive maintenance of motor drive systems in industrial robotic applications.

The grinding process is a common choice for finishing of mechanical parts in industrial practice where both surface quality and integrity are required to be maintained at sufficiently high levels. As it is not always possible to obtain all the necessary information for process monitoring through experimental measurements, it is often necessary to develop numerical models, which can be validated based on experimental data and then used to predict various outcomes of the grinding process such as the temperature or the stress field in the workpiece. Nevertheless, when specific responses are required to be predicted in real time, numerical models cannot be directly used due to their computational cost and thus, machine learning methods can be employed as an alternative choice. In order to determine a method which can achieve both the required level of accuracy and reduced computational cost, two different models, namely NARX (nonlinear autoregressive exogenous model) and LSTM (long-short term memory), are compared for a case of peripheral grinding of steel components under different process conditions. Both machine learning models are trained based on data from a validated numerical model, and their accuracy regarding the prediction of temperature field in every case is evaluated through various criteria.