International Journal of Science, Engineering and TechnologyAuthors: Ms. P. K. Bachhav Assistant Professor, Buddhabhushan Birhade, Mihir Binnar, Dinesh Ghare,, Dhananjay Birari
Abstract: Liquidity trouble is not a small-business problem alone. Enterprises of every size and structure run into it when customers pay late, sales arrive in uneven bursts, and operating and debt-service commitments cluster in the same weeks. The accounting and ERP systems most organizations rely on were built to record what has happened; they supply static ledgers and after-the-fact dashboards, and they say little about what could happen next or how a given shock would play out. Recent work shows that machine learning performs well on time-series forecasting, corporate bankruptcy detection, and the control of industrial cyber-physical systems. Yet these strands remain disconnected. Little of the machine learning literature addresses a firm’s own internal finances, since most of it targets high-frequency trading, equity markets, or institutional debt, and digital twins have been built mainly for manufacturing, aerospace, and smart cities rather than corporate finance. FinTwin is our attempt to close that gap: a decision-support system built around a financial digital twin and intended for enterprises across operating scales. Our proposed system pairs data-driven time-series forecasting (ARIMA, XGBoost, and exponential smoothing) with a continuous liquidity risk engine, an interactive Scenario Lab, and a prescriptive Action Simulator whose results feed back into the twin. An Explainable AI (XAI) layer then turns the numerical output into plain-language guidance and automated debtor notifications. To ground the design, we survey 22 leading reference works and examine their conceptual, theoretical, empirical, and architectural contributions.
DOI:
