Energy technology cost reductions are the result of many innovation trends in the energy system. The energy technology innovation system is increasingly well understood at an aggregate level and using qualitative concepts. However, the quantification of the multiple drivers of energy technology cost reduction trends remains poorly understood. This paper addresses this knowledge gap by presenting a systematic review of current practices. Despite their simplifications, one-factor learning curves (i.e. using a single driver) remain the most popular method for quantitative modelling of energy technology innovation. The role of multiple drivers on cost reductions has been cited in previous studies. This review enriches our understanding of these multiple drivers by examining their impact along different stages of technology development. The review quantifies the variation in these drivers and shows that the development of multi-factor learning curve models and bottom-up cost models are still in their infancy. With a focus on onshore wind and solar PV technologies, the review finds that most of the published multi-factor learning curve analyses are focused on addressing the impact of drivers related to i) manufacturing process improvements (i.e. learning by-doing) and ii) technology feature improvements (i.e. learning by-researching). This means that the other learning drivers such as market dynamics and learning by-interacting across different stakeholders and geographical areas are still poorly quantified, despite their impact on cost reduction being recognised in the innovation literature. There is a danger that misinformed policies are currently being developed in the absence of a good understanding of these multiple drivers.

降低能源技术成本是能源系统中许多创新趋势的结果。能源技术创新体系在总体上和使用定性概念方面越来越为人们所理解。但是，人们对能源技术成本降低趋势的多种驱动因素的量化了解仍然很少。本文通过对当前实践进行系统回顾来解决这一知识鸿沟。尽管简化了，但单因素学习曲线（即使用单个驱动程序）仍然是能源技术创新定量建模的最流行方法。在先前的研究中已经提到了多种驱动因素在降低成本上的作用。通过审查它们在技术开发的不同阶段的影响，这篇综述丰富了我们对这些多种驱动因素的理解。该评论量化了这些驱动因素的变化，并表明多因素学习曲线模型和自下而上的成本模型的开发仍处于起步阶段。该评论着重于陆上风能和太阳能光伏技术，发现大多数已发布的多因素学习曲线分析都集中于解决与驱动器的影响有关的因素，这些驱动因素包括：i）制造工艺的改进（即边做边学）和ii）技术功能的改进（即通过研究学习）。这意味着其他学习驱动因素，例如市场动态和不同利益相关者和地理区域之间的交互学习，尽管创新文献中已认识到它们对降低成本的影响，但其量化仍然很差。