Research On Valuation Of New Energy Enterprise Based On A BP Neural Network Model

In recent years,environmental issues have become a topic of common concern throughout the world due to the mounting problems caused by environment pollution and limited natural resources.In order to strengthen the sustainable development of the ecological environment,relevant policies have been issued at home and abroad to increase investment in environmental protection,which has boosted the rapid development of the new energy industry.With the support of the policy,the new energy industry has ushered in a new round of reform and reconstruction,resulting in a huge influx of capital.The development and application of new energy is an important trend of energy development in the future.Therefore,the valuation of new energy enterprises has received widespread attention.New energy enterprises are also characterized by large initial investment,long payback period and high potential value of enterprises.This thesis makes use of a set of selected features of new energy enterprises and leverages a BP neural network model to evaluate their values.This thesis focuses on the valuation of new energy enterprises based on BP neural network model,and carries out research from the following aspects.Firstly,the author sorts out the existing studies on enterprise valuation methods and the dominating factors that affect enterprise values.Secondly,the thesis gives a brief overview of the traditional valuation methods and their limitations,that motivated us to propose the idea of using BP neural network model to evaluate new energy enterprises.Thirdly,the author compiled a set of public new energy enterprises to carry out the research.This thesis utilized eight categories of indicators as inputs of neural networks,including financial indicators and non-financial ones,that have significant impacts on the value of new energy enterprises.After sample refinement,the author scaled the refined data-set with standardization.A five-layer neural network model,including the input layer,the three hidden layers and the output layer,has been constructed.The hyper-parameters of the neural model are also configured.The author used 90%randomly selected samples to train the neural network model,and regularization methods including dropout and L2 are employed that effectively alleviate the overfitting problem.The remaining 10%of the samples were used to verify the model effectiveness.To further test the model’s generalization ability,a random cross-validation has been carried out,which shows that the model has excellent generalization ability on the evaluation of new energy enterprises.At last,taking the new energy enterprise A as an example,we demonstrate the specific application of the neural network model.When the author evaluated the value of new energy enterprises,the deep neural network model used in this thesis can well fit the nonlinear relationship.The relative error between the predicted new energy enterprise value and the enterprise market value is within 15%,which also verified the validity and applicability of the neural network model in the value evaluation of new energy enterprises.As one of the data-driven methods,neural network model requires minimal assumptions that suppresses the subjectivity involved in the valuation.In a nutshell,the evaluation method based on BP neural network for new energy enterprises extends the application scenarios of artificial intelligence,and it provides a novel technical approach that harbors new opportunities for more improvements on valuation in the future.