基于熵权TOPSIS-PSO-ELM的制造企业绿色创新能力评价模型及实证研究

为客观和准确地评价制造企业绿色创新能力,本文构建了制造企业绿色创新能力评价指标体系,提出了基于熵权TOPSIS的粒子群(PSO)优化极限学习机(ELM)集成学习算法的制造企业绿色创新能力评价模型.首先运用熵权法客观确定指标权重,结合TOPSIS测度并综合评价制造企业绿色创新能力,然后将评价值作为先验样本进行极限学习机的训练与测试,训练过程中利用PSO优化极限学习机的网络结构与连接权值,从而对绿色创新能力进行全面的分析和评价.最后以60家制造企业为例进行实证分析,并将熵权TOPSIS-PSO-ELM算法与极限学习机回归拟合算法对比,结果表明:基于熵权TOPSIS-PSO-ELM模型所得评价结果较已有方法更为准确可靠.此外,为进一步提高我国制造企业绿色创新发展能力提出了理论建议.

Study of the Environmental Innovation Capability Evaluation Model of Manufacturing Enterprises Based on Entropy Weighted TOPSIS-PSO-ELM and Empirical Research

In order to objectively and accurately evaluate the environmental innovation capability of manufacturing enterprises, this study establishes anevaluation indicator system of environmental innovation ability of manufacturing enterprises. We propose the evaluation model of environmental innovation capability of manufacturing enterprises based on the integrated learning algorithm which is Entropy weighted TOPSIS and Extreme Learning Machine(ELM)with Particle Swarm Optimization (PSO). First, the Entropy weight method is employed to calculate the weighted index and comprehensive evaluation of environmental innovation capability of manufacturing enterprises by TOPSIS method. Then, the evaluation value is used as a priori sample for the training and testing of Extreme Learning Machine. This model produces a better network architecture and initial connection weights,and trains the traditional backward propagation again by PSO. The environmental innovation ability of the manufacturing enterprises is analyzed and evaluated in a more comprehensive way. Furthermore, an empirical evaluation of the sixty enterprises are taken as the example to illustrate the feasibility of this method, and a comparative analysis of the environmental innovative capability of the enterprises was also carried on. The validity of the prediction model is verified by comparing the proposed the Entropy weighted TOPSIS-PSO-ELM algorithm with traditional ELM regression fitting algorithms. The resultsshows that the evaluation results based on the Entropy weighted TOPSIS-PSO-ELM model is more accurate and reliable than the existing methods. In addition, it provides theoretical suggestions for further improving the environmental innovation capability of manufacturing enterprises in China.