Evaluation Of College Students' Scientific Research Ability Based On BP Neural Network

In the context of the construction of innovative countries and the reform of national science and technology system,in order to adapt to the national science and technology development trend,it is necessary to train and evalue college students' scientific research ability in colleges and universities.The traditional ability evaluation lacks importance judgment to the evaluation indicator,the importance judgment process is arbitrary and subjective,Based on the above questions,in this paper,BP neural network is used to evaluate the research ability of college students,its purpose is to improve the scientific and accurate of evaluation process and improve the training process of college students.At the same time,it provides a more scientific basis for college to understand students'the scientific research abilityFirstly,based on the process of college students' scientific research ability training and the paper writing process,this paper analyzes the abilities involved in each process and constructs the evaluation indicator system of college students' scientific research ability;Using the combination of formative evaluation and final evaluation,this paper designs the scoring rules of college students' scientific research ability;According to the scoring rules table,this paper designs the network questionnaire and weights the indicators based on the collected questionnaire data and combination weighting method.Then,based on the non-linear characteristics of the evaluation indicator,in order to weaken the randomness and subjectivity of traditional evaluation methods and achieve the scientific and practical evaluation of ability,an evaluation model based on BP neural network is constructed;In order to make the training data comparable,using min-max method to standardize data;According to the selection rules of training samples and test samples,an equal number of samples are selected from the questionnaire data;The number of hidden neurons is determined by trial and error method;The BP neural network based on gradient descent method or quasi-Newton method or LM method is trained by the selected samples,three kinds BP neural network models are validated for this paper's problem by four evaluation indexes such as mean square error,iteration number,generalization ability and prediction accuracy.Experiments show that based on LM algorithm the 8-12-1 single hidden layer BP neural network evaluation model is suitable for this problem.Finally,the prototype system including the demand analysis,database design and the core module is described,BP neural network model management module and ability evaluation module are designed and implemented.The prototype system of college students' scientific research ability evaluation is realized,which provides a practical tool for college students' scientific research management.