An Investigation Of Customer Demand Forecast System In The Field Of Industry Gas

Most of the customers in the industrial gas industry are large-scale manufacturing enterprises,such as steel,chemical and other enterprises,of which the production line cannot be easily interrupted,which requires the gas supply cannot be interrupted as well.At the same time,most customers will delegate gas companies to directly manage their gas supply,which requires a lot of resources to monitor and manage the gas supply of these customers.In the current process,the gas companies rely on planners to predict customer demand,and make distribution plans,the efficiency is very low.How to improve the efficiency has become an urgent problem to be solved.The purpose of this thesis is to find the most suitable forecast model for the industrial gas industry,developed a customer demand forecast and order automatic generation system to improve the gas supply monitoring,management of the automation degree and reduce labor costs.First,we carry out the research and optimization of the forecast model.We compare multiple linear regression,general neural network,Gaussian process regression and support vector machine in predicting continuous data,and proposes a hybrid forecasting model.In the prediction of intermittent data,a deep neural network is adopted.Through a large number of experiments,the prediction accuracy of each model is fully evaluated,and compared with a world-class commercial forecast software Demantra.Through the experiment,we can see that the hybrid model proposed by the author is about 25% higher than the commercial prediction software Demantra in the continuous data prediction.Then,we analyze the current processes of the gas company,find the lowest efficiency and obvious defect in the whole process,and design the corresponding solutions and systems.The system is used to replace some of the manual operations in the original process.In this way,we can improve the efficiency and make up the defect in the process.Finally,the optimized forecasting model is applied in the development of the system.By comparing the actual distribution plan and the forecast distribution plan,we find that the system can improve the efficiency of the whole process by eliminating about 30% of the labor in the process of customer forecasting and distribution planning.Thus,the original purpose of this thesis gets meet.