Simulated Research Dataset for Artificial Intelligence-Based Industrial Accident Risk Prediction in Chemical Process Engineering: A Methodological Resource for Statistical Modeling, Process Monitoring, and Safety Analytics

Here is the publication description: Simulated Research Dataset for Artificial Intelligence-Based Industrial Accident Risk Prediction in Chemical Process Engineering: A Methodological Resource for Statistical Modeling, Process Monitoring, and Safety Analytics Dr. Edwin Gerardo Acuña Acuña, PhD, MBA — Universidad Fidélitas, Costa Rica | ORCID: 0000-0001-7897-4137 This dataset introduces a quantum-enhanced simulated research resource comprising 1,500 independent observations and 48 variables, designed to support the development, validation, and benchmarking of Artificial Intelligence models for industrial accident risk prediction in chemical process engineering environments. Generated through a Quantum-Classical Hybrid Simulator (QC-HS) integrating IBM Qiskit-based quantum circuits with classical statistical distributions, the dataset covers eight scientific domains: process operating conditions, quantum-enhanced sensor readings (QSR), equipment maintenance parameters, human factors, environmental conditions, Statistical Process Control (SPC) metrics, quantum computing simulation metrics, and AI/ML model performance indicators. The dataset provides two primary target variables — a continuous Accident Risk Score (0–100) and a binary Incident Occurred indicator — supporting both regression and classification modeling paradigms. Key statistical characteristics include a mean Accident Risk Score of 29.06 (SD = 6.61), an incident rate of 15.1%, and process capability indices averaging Cp = 1.317 and Cpk = 1.240, reflecting near-threshold industrial process control conditions. The quantum simulation component encompasses variational quantum eigenvalue (VQE) computations across qubit configurations ranging from 5 to 127 qubits, with a mean quantum advantage factor of 2.38 over classical counterparts and quantum error rates consistent with current-generation NISQ devices. Eight chemical process unit types (CSTR Reactors, Distillation Columns, Heat Exchangers, PFR Reactors, Compressor Units, Storage Tanks, Vapor-Liquid Separators, and Absorber Columns) are represented across six NFPA 704 chemical hazard classes. The dataset includes 209 SPC out-of-control events (13.9%) and a fully documented codebook with variable definitions, measurement scales, distributional parameters, and domain classifications. All observations were generated with a fixed random seed (42) ensuring full reproducibility. This resource is designed to serve researchers in quantum machine learning, process safety management, statistical modeling, and cyber-physical systems, providing a structured benchmark for Partial Least Squares Structural Equation Modeling (PLS-SEM), Random Forest and Gradient Boosting classifiers, Quantum Neural Networks (QNN), Variational Quantum Classifiers (VQC), real-time anomaly detection, and Six Sigma process capability analysis. The dataset addresses a documented gap in the literature: no existing public dataset integrates quantum computing simulation metrics with classical chemical process safety analytics in a unified methodological framework. Dataset specifications: N = 1,500 | Variables = 48 | Domains = 8 | Format: XLSX + CSV | Version: 1.0 (2026) | Simulation engine: QC-HS / IBM Qiskit | License: Research use Keywords: quantum machine learning, industrial accident prediction, chemical process safety, statistical process control, simulated dataset, AI safety analytics, VQE, quantum-classical hybrid systems, PLS-SEM, NISQ, process capability, anomaly detection