Scholarlink

<p>Wheat (<i>Triticum aestivum</i> L.) is highly resource-intensive cereal crop of North-Western India with a significant energy footprints. Therefore, efficient energy budgeting and modeling for enhancing resource-use efficiency, while confirming long-term productivity and sustainability is essential. The present study was therefore, conducted to quantify energy inputs/outputs in 94 wheat production systems and develop predictive models using Cobb–Douglas Production Functions (CDPFs) and artificial intelligence techniques viz. artificial neural networks (ANNs) and adaptive neuro-fuzzy inference systems (ANFISs) on data gathered during 2021–22. These findings showed that total input energy of ~ 25.1 GJ ha<sup>−1</sup> yielded ~ 175 GJ ha<sup>−1</sup> of net energy gain. CDPF model exhibited a low adjusted R<sup>2</sup><sub><i>Adj</i></sub> (0.105), indicating diminishing yield response with increasing energy input. Conversely, ANN architecture (11–15-1–1; 11-input layers, two-hidden layers having 15-and 1-neurons, respectively and 1-output layer) and ANFIS (three-stage, 8-sub-networks) models achieved highest modeling accuracy (R<sup>2</sup> = 0.988** and 0.999** (<i>p</i> &lt; 0.01), respectively). The 1:1 correspondence for three-stage ANFIS model (8-sub-networks) revealed robust performance of simulation vs. modeling function. We therefore, propose ANFIS models with multi-layered structures developed by employing fuzzy-rules for precise prediction of wheat yields. These findings underscore the potential of modeling for energy-efficient and yield-optimized wheat production systems, emphasizing the need for policy interventions to promote efficient energy management by rationalizing fertilizer use, energy-efficient farm mechanization and irrigation to enhance ecosystems' sustainability. The superior performance of ANN and ANFIS prototypes highlights their potential as robust decision-support tools for real-time resource optimization and yield forecasting.</p> Graphical abstract <p></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

Energy based wheat yield prediction using robust neural networks and neuro-fuzzy inference systems across North-Western India

  • 期刊论文

  • 发表日期：

    2026年2月19日

查看全文
  • Gagandeep Kaur,
  • Rajni,
  • Jagtar Singh Sivia

摘要

Wheat (Triticum aestivum L.) is highly resource-intensive cereal crop of North-Western India with a significant energy footprints. Therefore, efficient energy budgeting and modeling for enhancing resource-use efficiency, while confirming long-term productivity and sustainability is essential. The present study was therefore, conducted to quantify energy inputs/outputs in 94 wheat production systems and develop predictive models using Cobb–Douglas Production Functions (CDPFs) and artificial intelligence techniques viz. artificial neural networks (ANNs) and adaptive neuro-fuzzy inference systems (ANFISs) on data gathered during 2021–22. These findings showed that total input energy of ~ 25.1 GJ ha−1 yielded ~ 175 GJ ha−1 of net energy gain. CDPF model exhibited a low adjusted R2Adj (0.105), indicating diminishing yield response with increasing energy input. Conversely, ANN architecture (11–15-1–1; 11-input layers, two-hidden layers having 15-and 1-neurons, respectively and 1-output layer) and ANFIS (three-stage, 8-sub-networks) models achieved highest modeling accuracy (R2 = 0.988** and 0.999** (p < 0.01), respectively). The 1:1 correspondence for three-stage ANFIS model (8-sub-networks) revealed robust performance of simulation vs. modeling function. We therefore, propose ANFIS models with multi-layered structures developed by employing fuzzy-rules for precise prediction of wheat yields. These findings underscore the potential of modeling for energy-efficient and yield-optimized wheat production systems, emphasizing the need for policy interventions to promote efficient energy management by rationalizing fertilizer use, energy-efficient farm mechanization and irrigation to enhance ecosystems' sustainability. The superior performance of ANN and ANFIS prototypes highlights their potential as robust decision-support tools for real-time resource optimization and yield forecasting.

Graphical abstract