Abstract The Precision Fish Farming (PFF) approach was applied to the estimation of fish oxygen consumption of rainbow trout in a raceway farm. A dynamic model, simulating the evolution of Dissolved Oxygen concentration, was identified: the daily oscillation of fish oxygen consumption rate was simulated by means of a sinusoidal function. The model was applied to the data set collected during a four-week field study, which was carried out in July 2019. Water temperature and Dissolved Oxygen concentration were measured with an hourly frequency in farm influent and effluent. Fish biomass was monitored on a daily basis by combining the data provided by a state-of-the art system for non-invasive estimation of fish weight distribution with mortalit counting. The monitoring period was partitioned into two time-windows, as fish was not fed during the first two weeks. These windows were further partitioned into a calibration and validation set. Three model parameters, i.e. the average daily respiration rate, the amplitude of its daily oscillation, and its phase were estimated by fitting the model output to the time series of DO concentration in the effluent. The results of the calibration show that: 1) the daily average oxygen consumption rate is consistent with the literature; 2) the amplitude of the daily oscillation when fish is regularly fed is more than twice that estimated for fasting fish. The results of the validation suggest that the model could be used implement a cost-effective automatic control of oxygen supply, based on the short-term prediction of oxygen demand.

中文翻译：

---

估算水道中虹鳟鱼 (Oncorhynchus mykiss) 的耗氧量：精确养鱼方法

摘要 将精准养鱼（PFF）方法应用于水道养殖场虹鳟鱼耗氧量的估算。确定了模拟溶解氧浓度演变的动态模型：通过正弦函数模拟鱼耗氧率的每日振荡。该模型应用于在 2019 年 7 月进行的为期 4 周的实地研究期间收集的数据集。每小时测量一次农场进水和出水中的水温和溶解氧浓度。通过将最先进的系统提供的数据结合起来，对鱼的重量分布进行非侵入性估计和死亡计数，每天监测鱼的生物量。监测周期分为两个时间窗口，因为前两周没有喂鱼。这些窗口被进一步划分为校准和验证集。三个模型参数，即平均每日呼吸速率、其每日振荡的幅度和它的相位是通过将模型输出拟合到流出物中DO 浓度的时间序列来估计的。标定结果表明：1）日均耗氧率与文献一致；2) 定期喂鱼时的每日振荡幅度是禁食鱼估计值的两倍多。验证结果表明，基于对需氧量的短期预测，该模型可用于实施具有成本效益的供氧自动控制。三个模型参数，即平均每日呼吸速率、其每日振荡的幅度和它的相位是通过将模型输出拟合到流出物中DO 浓度的时间序列来估计的。标定结果表明：1）日均耗氧率与文献一致；2) 定期喂鱼时的每日振荡幅度是禁食鱼估计值的两倍多。验证结果表明，基于对需氧量的短期预测，该模型可用于实施具有成本效益的供氧自动控制。三个模型参数，即平均每日呼吸速率、其每日振荡的幅度和它的相位是通过将模型输出拟合到流出物中DO 浓度的时间序列来估计的。标定结果表明：1）日均耗氧率与文献一致；2) 定期喂鱼时的每日振荡幅度是禁食鱼估计值的两倍多。验证结果表明，该模型可用于实现基于需氧量的短期预测的氧气供应的具有成本效益的自动控制。通过将模型输出拟合到流出物中 DO 浓度的时间序列来估计其相位。标定结果表明：1）日均耗氧率与文献一致；2) 定期喂鱼时的每日振荡幅度是禁食鱼估计值的两倍多。验证结果表明，基于对需氧量的短期预测，该模型可用于实施具有成本效益的供氧自动控制。通过将模型输出拟合到流出物中 DO 浓度的时间序列来估计其相位。标定结果表明：1）日均耗氧率与文献一致；2) 定期喂鱼时的每日振荡幅度是禁食鱼估计值的两倍多。验证结果表明，基于对需氧量的短期预测，该模型可用于实施具有成本效益的供氧自动控制。