Research On The Calculation Model And Prediction Optimization Of Cooling Load In Office Buildings Based On π-theorem

The current scale of the construction industry output value has reached record highs,which brings huge building energy consumption that comes with it is the current scale of the construction industry output value has reached record highs,and the huge building energy consumption that comes with it is currently a key concern in the field of energy conservation and emission reduction in China.Nowadays,we advocate carbon peaking and carbon neutrality,we should study how to reduce building loads and energy consumption to achieve the goal of carbon peak and carbon neutrality.For nearly half a century,experts and scholars at home and that the research part has been done abroad on the calculation of building envelope.The establishment and research progress of construction load measurement model is slow.Using the very classical load calculation theory and load measurement algorithm,the construction load problem can be quickly and correctly transmitted to future generations,but more problems need to be solved,and the weight gain.It should be possible to calculate how the building is covered by multiple factors,the correctness of single factor and multi factor calculation,the load degree of equipment under comparative work and the load of equipment under the action of all factors.According to the load measures,there is more research on the load measures of power system,but there is little research on the load of air conditioning system on site,and there is more research on the design and type selection of air conditioning equipment.When installing the load model,the influence of meteorological parameters on the building load is very considerable.There is no influence of internal interference of the building,and there is a certain error.In view of the current research status in the calculation and prediction of cooling load of enclosure structure,propose a cooling load calculation model based on πtheorem,and using neural networks for predictive optimization study of cooling load.The main things are great:（1）This paper confirms the current situation of energy design and forecasting,as well as the needs of household construction and utilization.This paper systematically teaches the methods and theories of actual load calculation and measurement at home and abroad,guides the construction method of cooling load calculation model and measurement model.（2）Based on the dynamic heat transfer process of cooling load,the factors affecting the cooling load of building envelope are analyzed according to the partial correlation coefficient.Eight influencing factors and high connectivity are selected as the basis of model construction.These are the indicators of building level and geometric parameters of building area,such as heat transfer coefficient of external wall,external window coefficient and solar waste heat coefficient,window wall and building level of building.On the π-theory of section analysis method,dimensionless treatment is accepted,and the dimensional characteristics of building environment and thermal insulation performance index are considered.The dynamic simulation of the maximum cooling load and the inversion of the normalized heat transfer coefficient in the warm spring and cold winter section of the building,and the calculation of the cooling load index model in the warm spring and cold winter section of the building is established.Taking the office building as an example,the accuracy of the model is confirmed and analyzed.The calculation model provides a rapid and accurate energy-saving performance for the planning phase of building design,as well as a prediction of energy efficiency retrofits.（3）Based on BP and GA-BP algorithms,the cooling load prediction of office buildings is studied.Firstly,the Spearman change correlation coefficient of input change is analyzed,the input vector is subtracted from the output and normalized.After the initial operation,taking the cooling load of the building as the distribution vector,the simulation calculation is carried out on the pytoch platform to reduce the learning performance and effect of the two algorithms.The results show that after optimizing the BP neural network by using genetic algorithm to simulate the process of natural selection and evolution of organisms,the converter speed of GA-BP measurement model is increased by 2.41 times,the average absolute error is Mae,the average square error is 26.44% and 39.29% respectively,the MSE is reduced,and the Pearson correlation coefficient r is increased by 20.08%.GA-BP algorithm can effectively solve the problem that the training process of the neural network does not jump out of the local optimal solution,proving the computational accuracy of GA-BP and its stability.Figure [37] Table [16] Reference [80]...