光伏电站发电量预估数据

1、准确预测一个地区分布式光伏场站的整体输出功率，可以提高电网的稳定性，增加电网消纳光电能量的能力，在降低能源消耗成本的同时促进低碳能源发展，实现动态供需状态预测的方法，为绿色电力源网荷储的应用落地提供支持。
2、准确预估光伏电站发电量，可以自动发现一些有故障的设备或者低效电站，提升发电效能。1、逆变器及电站数据采集,将逆变器中计算累计发电量数据,告警数据同步到Maxcompute大数据平台
2、天气数据采集, 通过API获取ERA5气象数据包括光照辐射、云量、温度、湿度等
3、数据特征构建, 在大数据处理平台进行数据预处理，用累计发电量矫正小时平均发电功率，剔除异常数据、归一化。告警次数等指标计算
4、异常数据处理, 天气、设备数据根据经纬度信息进行融合, 并对融合后的数据进行二次预处理操作，剔除辐照度和发电异常的一些数据
5、算法模型训练,基于XGBoost算法模型对历史数据进行训练, 生成训练集并保存至OSS
6、算法模型预测,基于XGBoost算法模型接入OSS训练集对增量数据进行预测, 并评估预测准确率等效果数据，其中误差率=(发电量-预估发电量)/发电量，当误差率低于一定阈值时，该数据预测为准确。预测准确率=预测准确数量/预测数据总量。

Accurately predicting the overall output power of distributed photovoltaic (PV) power stations in a region can enhance power grid stability, improve the grid’s capacity to absorb and accommodate photovoltaic energy, reduce energy consumption costs, promote the development of low-carbon energy, and support the practical deployment and implementation of green power source-grid-load-storage systems through dynamic supply and demand state prediction.

Accurately forecasting the power output of PV power stations can automatically identify faulty equipment or underperforming stations, thereby enhancing overall power generation efficiency.

1. Inverter and power station data collection: Synchronize the cumulative power generation data and alarm data calculated by inverters to the MaxCompute big data platform.
2. Weather data collection: Obtain ERA5 meteorological data including solar irradiance, cloud cover, temperature, humidity and other relevant parameters via API.
3. Data feature construction: Conduct data preprocessing on the big data processing platform, correct the hourly average power generation using cumulative power generation data, remove abnormal data, perform normalization, and calculate indicators such as alarm frequency.
4. Abnormal data processing: Fuse weather and equipment data based on latitude and longitude information, conduct secondary preprocessing on the fused data, and eliminate data with abnormal solar irradiance and power generation.
5. Algorithm model training: Train on historical data using the XGBoost algorithm model, generate a training dataset and save it to OSS.
6. Algorithm model prediction: Use the XGBoost model connected to the OSS-stored training dataset to predict incremental data, and evaluate performance metrics such as prediction accuracy. The error rate is calculated as (actual power generation - predicted power generation)/actual power generation. A prediction is deemed accurate when the error rate is lower than a specified threshold. Prediction accuracy is equal to the number of accurately predicted samples divided by the total number of predicted data samples.

搜集汇总

数据集介绍

特点

光伏电站发电量预估数据集由锦浪科技股份有限公司提供，包含501条记录，每日更新，用于预测光伏电站发电量和提升电网稳定性。数据集包含电站ID、容量、告警数、湿度、云量、温度、地表太阳辐射、地表热辐射、顶部太阳辐射、发电量、等效满发小时数和预估值等字段，应用XGBoost算法模型进行预测。