Green product certification is an important link between sustainable production and consumption, which are widely used by governments to promote sustainable development. However, the risks in the green product certification are not sufficiently addressed. The purpose of this study is to identify and evaluate the risk of green product certification, for a risk index system and the intelligent risk assessment model were constructed, which could provide support for risk management of green product certification. Green furniture products were taken as an example, as they are closely related to daily life and health. Based on the implementation specification of green product certification on furniture in China, six basic links and 19 certification risk indicators were proposed. Then an intelligent risk assessment model was constructed using the interpretation structure model, analytical network process and probabilistic neural networks methods. Based on the history data of green furniture product certification in China, the results show that on-site inspection stage has the highest risk weight of 0.3457 among the six links. The comprehensiveness of the information input has the highest risk weight of 0.1290 among the 19 indicators. The intelligent risk assessment model could accurately evaluate the risk, the results showed that there is always a certain risk in the green product certification. Therefore, some suggestions were recommended, certification traceability system, professional staff training and intelligent information filling would be helpful to risk management of green product certification. The results of this study could be extended to other fields of green products and be used for other countries to improve the risk management of green product certification.

绿色产品认证是连接可持续生产和消费的重要环节，被各国政府广泛用于促进可持续发展。然而，绿色产品认证中的风险并未得到充分解决。本研究旨在识别和评估绿色产品认证的风险，构建风险指标体系和智能风险评估模型，为绿色产品认证的风险管理提供支持。以绿色家具产品为例，因为它们与日常生活和健康息息相关。根据我国家具绿色产品认证实施规范，提出了6个基本环节和19个认证风险指标。然后利用解释结构模型、解析网络过程和概率神经网络方法构建了智能风险评估模型。根据我国绿色家具产品认证的历史数据，结果表明，现场检查阶段在六个环节中风险权重最高，为0.3457。在19项指标中，信息输入的全面性风险权重最高，为0.1290。智能风险评估模型可以准确评估风险，结果表明绿色产品认证始终存在一定的风险。因此，提出了一些建议，认证追溯体系、专业的人员培训和智能信息填写将有助于绿色产品认证的风险管理。