智慧水产数据

1、水质关键指标预测预警：基于采集的水塘电导率EC数据，用算法预测下一时间段可能的水塘电导率EC值，如果预测值超标，则系统提前预警，以便业务人员提前进行处理优化，为鱼类提供更适宜的生长环境。
2、水质分析预测数据可用于科研机构，为水产养殖的科学、智能研究提供数据与分析的支持。
3、水质分析预测与优化的经验可帮助其他水产养殖企业，促进水产养殖行业向智能、高效、模式可移植的方向发展。1.数据采集：水塘电导率EC通过相应的传感器监测设备每日多次获得其监测值。
2.数据处理预测：
(1)将1天分为24个时段，每个时段取其前60个时段的“水塘电导率EC”值作为特征值，取其下一个时段的“水塘电导率EC”值作为目标值，用多项式回归及GBDT回归两种算法进行训练及预测，取预测偏差均值更小的作为最后的预测模型，用此模型对当前时段的数据进行预测未来一个时段的水塘电导率EC，得到“水塘电导率EC_下一时段预测”值。以上算法训练每隔一段时间进行再次训练和预测，以获得更新的模型。
(2)如果“水塘电导率EC_下一时段预测”值在0--1800之间则认为无异常，“水塘电导率EC-预测异常”值为“N”，否则“水塘电导率EC-预测异常”为“Y”。

1. Water Quality Key Indicator Prediction and Early Warning: Based on the collected pond electrical conductivity (EC) data, algorithms are utilized to predict the potential pond EC value in the next time period. If the predicted value exceeds the standard, the system will issue an early warning in advance, enabling operational staff to carry out targeted treatment and optimization ahead of schedule, thereby providing a more suitable growth environment for fish.
2. The water quality analysis and prediction data can be provided to scientific research institutions, offering data and analytical support for scientific and intelligent research on aquaculture.
3. The experience derived from water quality analysis, prediction and optimization can assist other aquaculture enterprises, promoting the aquaculture industry to develop towards intelligent, efficient and mode-transplantable directions.
1. Data Collection: The pond electrical conductivity (EC) values are monitored multiple times daily using corresponding sensor monitoring equipment.
2. Data Processing and Prediction:
(1) Divide one day into 24 time intervals. For each time interval, take the EC values of the preceding 60 time intervals as feature values, and take the EC value of the next time interval as the target value. Two algorithms, polynomial regression and Gradient Boosting Decision Tree (GBDT) regression, are employed for training and prediction. The model with the smaller mean prediction deviation is selected as the final prediction model. This model is used to predict the pond EC value for the next time interval based on the data of the current time interval, generating the "Predicted Pond Electrical Conductivity (EC) for the Next Time Interval" value. The aforementioned algorithm training is retrained and executed at regular intervals to obtain updated models.
(2) If the "Predicted Pond Electrical Conductivity (EC) for the Next Time Interval" value falls within the range of 0 to 1800, it is considered non-abnormal, and the "Predicted Abnormal Pond Electrical Conductivity (EC)" value is marked as "N"; otherwise, it is marked as "Y".

搜集汇总

数据集介绍

特点

智慧水产数据集包含702条每日更新的水质监测数据，主要用于水质关键指标的预测预警和科研支持，采用多项式回归及GBDT回归算法进行数据处理和预测。