Research On Energy Load Prediction Of Controlled Environment Agriculture Industrial Zone Based On LSTM Algorithm

In the current development of protected horticulture industry,high energy consumption cost has become the main obstacle restricting the high-quality development of protected horticulture.To reduce energy consumption costs and improve production efficiency,it is an effective solution to build a machine learning model-based smart energy Internet for protected horticulture industrial parks.However,the application of traditional energy consumption prediction model technology in protected horticulture buildings lacks verification.Therefore,it is necessary to explore the energy consumption features based on protected horticulture production scenarios.In addition,the energy consumption behavior of protected horticulture buildings has obvious trends and periodicity,which is related to the circadian rhythm of plants and the periodicity of production activities.Therefore,this paper will explore the effect of different time series processing on the improvement of the model,and verify the effect of building energy consumption features on the accuracy of the model based on the production scenarios of protected horticulture.This paper will elaborate on the following aspects:(1)Depict the energy consumption features of the thermal load and electrical load of the protected buildings horticulture building throughout the year;(2)Propose the building energy consumption features based on the protected horticulture production scenarios,and explore the relationship between energy consumption and features(3)Explore the impact of machine learning and deep learning algorithms on time series processing on models,and propose how to develop high-precision models based on time series processingThe study found that the data-driven methods are feasible in protected horticulture energy consumption prediction,with better models such as LSTM and RF models having a decision coefficient of over 0.9.Multiple features constructed based on protected horticulture production scenarios have a good improvement effect,and the coefficient of determination of the model is improved by 2% to 9% for better features.By testing the time series,it is found that the LSTM algorithm has better results in the time interval between 24 and 48 hours.However,in the energy consumption model of protected horticultural buildings,compared with some non-sequential algorithms under time series processing,the LSTM algorithm has a slightly worse performance,which is 3.55% and 2.17% lower than the coefficient of determination of the random forest and gradient boosting tree models,respectively.At the same time,according to the advantages and disadvantages of different models in terms of accuracy,development difficulty,and computing power requirements,a method of how to construct an energy consumption prediction model for facility gardening buildings is proposed.On the basis of the models,a plan for energy transformation based on the energy consumption model of the protected horticulture park is proposed,which is expected to save energy costs of 12.04 million yuan throughout the entire lifecycle of the plan.