Solar energy system 'smart' create green energy scenarios

Probation period can save electricity did the demand of about 1. 3 million yuan each month, photovoltaic (pv) grid, results after comprehensive application is expected to generate more income. Recently, the xiangtan university information engineering college professor Duan Bin, tan looks and Su Yongxin team in power load forecasting and distributed energy scheduling optimization of a number of important research progress. “ The current power load forecasting problem according to the time scale division consists of long-term, medium-term and short-term ( Super short term) Three categories, the problem of different classes the data characteristic difference is very big, solar panels, the method also has the very big difference, we focus on short-term and ultra short-term prediction problem. In our country, for large power users, once the load exceeds a certain threshold, the settlement period needs to be according to the peak power for conventional power electricity in a high demand of electricity. ” Tan appearance was told the Chinese proceedings ', & other This need to forecast the load, temperature and humidity monitor its precision has important influence on energy cost. “ The final results show that our proposed method accuracy and stability are super current mainstream advanced time-series forecasting methods, especially in solving the problem of factory scene has obvious advantages. ” Compared with the conventional energy system of all kinds of energy independence, more integrated energy systems include gas, electricity, cold, heat temperature and humidity monitor, storage, new energy, such as coordination each other aid, China solar energy network, application of cascade, the complexity of the system is far higher than that of conventional energy systems, pluripotent collaborative system load forecasting and optimal operation is recognized as a problem.

　　” Tan said. ” With shallow artificial neural network and support vector machine (SVM) algorithm is represented by the main trend of power load forecasting method of silicon wafer is 12 listed company but usually difficult in feature extraction and data reconstruction process complex, high network model complexity and nonlinear optimization problems such as local minimum. ” Tan said. However, only with black box service outsourcing system, through the enterprise data self-learning function is not perfect, enterprise also cannot modify the software model. “ They want to be able to smooth the electricity load down. Therefore, large valin iron and steel enterprise of hunan steel made difficult. Suitable for short-term and ultra short-term prediction with artificial neural network prediction method, support vector machine (SVM) method, etc. Power system load forecasting is a typical time series prediction problems. Steel market is changing, pv stent production enterprise enterprise production mode is not set in stone, in the long term, the model will reduce the accuracy of load forecasting. ” Latest study, load forecasting based on the deep learning become a hot spot, the national support of new energy projects its prediction accuracy and high stability, can deal with complex issues, showing great potential. Tan looks, said valin steel energy system is a typical regional integrated energy systems, energy consumption, China solar energy network, outsourcing power exceeds one billion yuan a year. The state supports the new energy projects and other; In May 2019 launch pilot application, 3 sets of ladle furnace involved in regulation, peak load forecasting error is less than 3%, this is an advanced technology, photovoltaic stents satisfaction index of the scene production enterprises. ” Tan appearance was introduced, and the other; Our work and application widely, puts forward the model by AEMO open data sets, and a large number of industrial measurement data validation, as much as possible to avoid samples selective error. “ Before some of the load forecasting evaluation using simulation data set, and there is no real data, China solar energy network, heat storage equipment the result is not reliable. Main results, top academic journals in the field of energy and power system, 'IEEE power system transactions and journal of applied energy.