Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Destine Mashava has a total of 16 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Digital Twins for Battery-Powered and Electric Rail Vehicles: Architectures, Prognostics, Validation and Research Priorities—A Critical Review

    by Destine Mashava, Eriyeti Murena, Lameck Mugwagwa, Tshifhiwa Nenzhelele, Khumbulani Mpofu

    Battery-powered and electric rail vehicles are an emerging route to decarbonising lines where overhead electrification is uneconomic, and they move a large share of vehicle dependability onto the battery system, its thermal management and power-electronic interfaces. Digital twins (DTs) virtual representations continuously calibrated to a physical asset are widely proposed to manage such systems, yet railway-specific evidence remains thin. This narrative critical review consolidates the foundations needed to design a DT for a battery-powered rail vehicle: energy-supply architectures and railway duty characteristics, lithium-ion ageing and failure mechanisms, battery model families, data infrastructure and edge–cloud partitioning, prognostics and health management, and verification, validation and standards. A four-layer reference architecture that keeps safety-related protection on the vehicle while using the cloud for calibration and fleet analytics is proposed. The review finds that the enabling methods are mature in road-vehicle and laboratory contexts but that railway-specific validation, uncertainty-aware decision support, twin qualification criteria and demonstrated operational benefit are lacking in the literature examined. A prioritised research agenda is presented. The review is narrative rather than systematic; gaps are stated relative to the sources examined.

  • Open Access

    The Daydreaming Factory: A Twin-Calibrated Simulation Framework for Proactive Reliability Exploration of Battery-Powered Rail Vehicles—A Critical Review and Research Agenda

    by Destine Mashava, Eriyeti Murena, Tshifhiwa Nenzhelele, Khumbulani Mpofu, Tawanda Mushiri

    Battery-powered and electric rail vehicles move substantial dependability risk onto battery systems, thermal management, power electronics and software for which no mature fleet failure history exists. Conventional reliability methods depend on failure modes identified beforehand and calibrated by field experience that is not yet available. This critical review examines how twin-calibrated simulation can be used to explore failure possibilities proactively. The “Daydreaming Factory” a conceptual construct proposed here, not an established term is positioned against computational ideas of model-based imagination and is organised through three analytical devices: a taxonomy of virtual environments for reliability engineering, a causal-chain test from environment to reliability outcome, and a five-level evidence-maturity scale. The enabling methods are reviewed: multi-domain co-simulation, scenario generation and rare-event search, fault injection, degradation integration, and verification, validation and uncertainty quantification. A six-component reference framework with an explicit credibility layer is specified. Assessed against the sources examined, the individual methods are mature in adjacent domains, but integrated railway-specific implementations, field validation, qualification pathways and measured reliability benefits are largely absent. A research agenda and candidate performance indicators are proposed. The review is narrative rather than systematic and the framework has not been experimentally validated.

  • Open Access

    Lightweight Pixel-Wise Lesion Segmentation and Severity Estimation of Tobacco Leaf Disease from Close-Up Images: A Group-Aware Benchmark with Robustness Analysis

    by Destine Mashava, Mthulisi Ngwenya, Tawanda Mushiri, Willy Stephen Tounsi Fokui, David Sinkhonde

    Lesion area is the basis of disease severity rating in tobacco, but visual rating is subjective, and most deep-learning studies classify whole images on random splits that can leak near-duplicate photographs. This study evaluates lesion segmentation and severity estimation on 3,485 close-up tobacco leaf images (800 × 800 px) with pixel-level masks, using group-aware training, validation and test partitions (3 independent splits) and a lightweight pixel-wise gradient-boosting classifier operating on colour, texture and image-normalised features. The full model reached a Dice coefficient of 0.837 ± 0.020 (IoU 0.720 ± 0.029), compared with 0.187 for the best Otsu-threshold baseline. Predicted lesion area agreed closely with annotation (Pearson r = 0.981; mean absolute error 0.39 percentage points) and reproduced a four-grade severity scale with 90.0% accuracy (quadratic-weighted κ = 0.913). Without retraining, texture features improved robustness to occlusion and brightness change but caused a collapse under Gaussian noise (Dice 0.143); noise-aware training and a median pre-filter restored Dice to 0.763 and 0.666, respectively. Inference took about 79 ms per image on one CPU core. The data carry no disease-class labels, so no pathogen identification is claimed. The results provide a reproducible CPU-only baseline and a transparent account of failure modes for lesion-level phenotyping of tobacco.

  • Open Access

    TEXT-DERIVED MINING SUPPLY-CHAIN RISK FOR RISK-AWARE DEMAND FORECASTING IN ZIMBABWE’S COAL AND MANUFACTURING VALUE CHAINS: A RESEARCH PROTOCOL

    by Antony G. Foya, Destine Mashava, Takudzwa M. Muhla, Innocent Mapindu

    Industrial value chains built around coal in Zimbabwe, spanning mining, power, metallurgical processing and manufacturing, face interacting disruptions: electricity interruptions, transport constraints, foreign-exchange shortages, regulatory change, climate stress, equipment failure and commodity-price volatility. These disruptions usually appear first as text, in news reports, utility notices, statutory instruments and corporate announcements, and only later in production or sales records. This paper specifies a reproducible research protocol that converts Zimbabwe-focused text into supply-chain risk indices and embeds them in demand and volatility forecasts. Latent Dirichlet Allocation (LDA) identifies persistent structural themes and BERTopic identifies emerging, context-dependent ones. Expert-validated topic probabilities are aggregated into lagged weekly or monthly risk indices that enter the mean and variance equations of ARIMAX-GARCH-X models. The protocol sets out an eight-family risk taxonomy; a multi-case portfolio described by process, product and sector, never by name; information-availability and leakage controls; a nested model ladder; six hypotheses stated in advance with matched statistical tests; human-in-the-loop decision rules; and minimum reporting standards. No empirical forecast improvement is claimed, because the firm-level and text-corpus data have not yet been assembled. The contribution is a testable and auditable design for data-constrained emerging-market settings, intended to support procurement, inventory, production scheduling, logistics, maintenance and industrial-policy planning.

  • Open Access

    Deep Learning for Multi-Component Gearbox Fault Diagnosis on the PHM 2009 Benchmark: A Critical Review and Comparative Evaluation of Time–Frequency Representations and Network Architectures

    by Destine Mashava, Takudzwa M. Muhla, Tafadzwa Wachenuka, David Sinkhonde

    Concurrent faults in gears, bearings and shafts remain among the hardest cases for vibration-based condition monitoring of industrial gearboxes, because their signatures overlap, modulate one another and shift with speed and load. The 2009 PHM Society gearbox has become the standard public benchmark for this multi-component problem, yet published results on it are scattered across input types, network designs and validation schemes, which makes headline accuracies hard to compare. This article presents a critical review of deep-learning diagnosis on this benchmark and the wider gearbox literature, organised around three design choices: signal representation, model family and evaluation protocol. To ground the discussion, we re-analyse a controlled experiment in which three convolutional architectures were trained on 10,240 images per representation (continuous wavelet transform scalograms, short-time Fourier transform spectrograms and wavelet synchrosqueezed images) covering eight spur-gearbox health states. A compact five-stage batch-normalised network reached 99.90%, 99.85% and 99.66% test accuracy with wavelet, Synchrosqueezed and spectrogram inputs, respectively, but the margins among the best models correspond to only a few of the 2,048 test images. On a separate held-out validation subset, accuracy fell to 94.53%, 92.03% and 77.34%, making the representation effect far more visible. Comparison with published capsule-attention, architecture-search and transformer studies shows that reported accuracies on the same benchmark range from below 80% to above 99.5%, mainly because of differences in splitting protocol. We conclude with a reporting checklist and research agenda for leakage-aware, physics-informed and deployable gearbox diagnosis.