Scholarlink

<p>Water treatment and desalination plants are essential for meeting the global demand for safe and clean water. Distillate rate, feed heater temperature, and condenser temperature are critical parameters that impact the performance of the desalination system. A high desalination rate leads to high energy consumption. In contrast, a high feed heater temperature may cause degradation or system failure if it exceeds the thermal limits of the system material. At the same time, the condenser temperature plays a crucial role in the system's frequent maintenance and reliability. Therefore, identifying inefficiencies in the desalination process and optimizing it can reduce energy consumption and maximize freshwater output. The present study proposes an experimental approach to collecting response data in real-time and its optimization using Artificial Intelligence (AI). Initially, the experiments were designed using Design Expert software with a central composite design. A tabletop experimental setup was manufactured and tested for response data collection (distillate rate, feed heater temperature, and condenser temperature) with varied vacuuming degrees (20 to 60 kPa), salinity levels (20,000 to 40,000 ppm), feed temperature (25 to 45 °C) and volume flow (0.3 to 0.5 L/min). Subsequently, a random search-optimized Extreme Gradient Boosting (XGBoost) algorithm is proposed to improve the performance of the proposed desalination device. Statistical analysis is conducted to develop a robust AI-based regression model for predicting the performance of the proposed device. It is observed that the model succeeded in predicting distillate parameters with 1.42% Mean Squared Error (MSE), 2.91% Root Mean Squared Error (RMSE). The observed optimum values of distillate, feed heating temperature, and condensing unit temperature were 7,580 ml/h, 53.8°C, and 49.9°C, respectively. The optimized XGBoost prediction can facilitate immediate adjustments to processes for optimal performance.</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

An experimental evaluation, performance analysis, and improvement of water desalination system using optimized machine learning

  • 期刊论文

  • 发表日期：

    2025年10月17日

查看全文
  • Faizan Ahmed,
  • Nayeemuddin Mohammed,
  • Hiren Mewada,
  • Mohd Sharizal Abdul Aziz,
  • C. Y. Khor

摘要

Water treatment and desalination plants are essential for meeting the global demand for safe and clean water. Distillate rate, feed heater temperature, and condenser temperature are critical parameters that impact the performance of the desalination system. A high desalination rate leads to high energy consumption. In contrast, a high feed heater temperature may cause degradation or system failure if it exceeds the thermal limits of the system material. At the same time, the condenser temperature plays a crucial role in the system's frequent maintenance and reliability. Therefore, identifying inefficiencies in the desalination process and optimizing it can reduce energy consumption and maximize freshwater output. The present study proposes an experimental approach to collecting response data in real-time and its optimization using Artificial Intelligence (AI). Initially, the experiments were designed using Design Expert software with a central composite design. A tabletop experimental setup was manufactured and tested for response data collection (distillate rate, feed heater temperature, and condenser temperature) with varied vacuuming degrees (20 to 60 kPa), salinity levels (20,000 to 40,000 ppm), feed temperature (25 to 45 °C) and volume flow (0.3 to 0.5 L/min). Subsequently, a random search-optimized Extreme Gradient Boosting (XGBoost) algorithm is proposed to improve the performance of the proposed desalination device. Statistical analysis is conducted to develop a robust AI-based regression model for predicting the performance of the proposed device. It is observed that the model succeeded in predicting distillate parameters with 1.42% Mean Squared Error (MSE), 2.91% Root Mean Squared Error (RMSE). The observed optimum values of distillate, feed heating temperature, and condensing unit temperature were 7,580 ml/h, 53.8°C, and 49.9°C, respectively. The optimized XGBoost prediction can facilitate immediate adjustments to processes for optimal performance.