基于BP神经网络的科研绩效综合评价模型--以宁夏师范学院为例

Scientific research performance evaluation model based on BP neural network--Take Ningxia Normal University as an Example

首先引用投入产出理论与德尔菲法构建评价指标体系,归一化处理样本数据,取Sigmoid函数作为神经元激活函数,基于BP神经网络算法建立了评价模型.然后运用Python程序训练评价模型得到相应样本的科研绩效评价值及模型达到一定精度时的各评价指标权重值.最后利用该模型进行样本检验和测试,结果表明该模型能科学、公正地评价高校科研绩效,以服务于宁夏师范学院科研管理工作.

In this paper,the input-output theory and Delphi method are used to construct the evaluation index system.By normalizing the sample data and taking the Sigmaid function as the neuron activation function,an evaluation model is established based on the BP neural network algorithm.Then,the Python program is used to train the evaluation model to obtain the scientific research performance evaluation values of the corresponding samples and the weight values of each evaluation index when the model reaches a certain accuracy.Finally,sample inspection and testing using this model show that the model can scientifically and impartially evaluate the scientific research performance of universities,in order to serve the scientific research management work of Ningxia Normal University.