Scholarlink

<p>Throughout a period exceeding 45&#xa0;years and over 95 orbital launches, the Indian Space Research Organisation (ISRO) has developed an impressive record of cost-effective and sound engineering. However, the telemetry and financial logs gathered throughout this period are seldom studied as a cohesive set. This current study extracts relevant insights from this particular case. The unifying of conventional aerospace variables and machine learning techniques like XGBoost and ensemble learning models introduces a new paradigm in order to assess the effectiveness of launch vehicles. The data set utilized includes a range of missions to the far reaches of space, the Earth, and navigation satellites. This model achieved a classification accuracy of 92.3 percent on historical launch outcomes. Budgetary predictions also demonstrated similar reliability, wherein financial models maintained a root mean square error (RMSE) of $1.18 million. This low variance has direct implications for planning real-world mission budgets. A temporal study of the data set also reveals the maturation of the ISRO over time. Success rates of early developmental missions hovered around 55 percent reliability. Current missions have surpassed 82 percent reliability. This is a direct result of higher manufacturing standards and operation parameters. Feature importance plots reveal payload mass margins and health anomalies to be the most significant indicators of critical failure. This mapping of historical failure modes provides engineers with a statistical baseline to weigh against costs in future orbital missions (T. Chen and C. Guestrin, "XGBoost: A scalable tree boosting system," in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016. [Online]. Available: <a href="https://dl.acm.org/doi/10.1145/2939672.2939785">https://dl.acm.org/doi/10.1145/2939672.2939785</a>; Indian Space Research Organisation (ISRO), "ISRO Missions &amp; Launch Vehicles," 2025. [Online]. Available: <a href="https://www.isro.gov.in">https://www.isro.gov.in</a>; S. Aggarwal and P. Rao, "Machine learning models in spacecraft mission planning and forecasting," INAE Letters, vol. 7, no. 3, pp. 178–193, 2022. [Online]. Available: <a href="https://www.sciencedirect.com/science/article/pii/S127096381930141X">https://www.sciencedirect.com/science/article/pii/S127096381930141X</a>).</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Data-driven launch analytics for ISRO: predictive modeling of mission success and cost optimization through machine learning

  • 期刊论文

  • 发表日期：

    2026年4月1日

查看全文
  • Akshat Pal,
  • Nipun Varshneya,
  • Aastha Singh,
  • O. V. Gnana Swathika

摘要

Throughout a period exceeding 45 years and over 95 orbital launches, the Indian Space Research Organisation (ISRO) has developed an impressive record of cost-effective and sound engineering. However, the telemetry and financial logs gathered throughout this period are seldom studied as a cohesive set. This current study extracts relevant insights from this particular case. The unifying of conventional aerospace variables and machine learning techniques like XGBoost and ensemble learning models introduces a new paradigm in order to assess the effectiveness of launch vehicles. The data set utilized includes a range of missions to the far reaches of space, the Earth, and navigation satellites. This model achieved a classification accuracy of 92.3 percent on historical launch outcomes. Budgetary predictions also demonstrated similar reliability, wherein financial models maintained a root mean square error (RMSE) of $1.18 million. This low variance has direct implications for planning real-world mission budgets. A temporal study of the data set also reveals the maturation of the ISRO over time. Success rates of early developmental missions hovered around 55 percent reliability. Current missions have surpassed 82 percent reliability. This is a direct result of higher manufacturing standards and operation parameters. Feature importance plots reveal payload mass margins and health anomalies to be the most significant indicators of critical failure. This mapping of historical failure modes provides engineers with a statistical baseline to weigh against costs in future orbital missions (T. Chen and C. Guestrin, "XGBoost: A scalable tree boosting system," in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016. [Online]. Available: https://dl.acm.org/doi/10.1145/2939672.2939785; Indian Space Research Organisation (ISRO), "ISRO Missions & Launch Vehicles," 2025. [Online]. Available: https://www.isro.gov.in; S. Aggarwal and P. Rao, "Machine learning models in spacecraft mission planning and forecasting," INAE Letters, vol. 7, no. 3, pp. 178–193, 2022. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S127096381930141X).