Because of the complexity and diversity of human behavior, social systems are unpredictable and difficult to quantify. With the development of computer technology, scholars have begun to construct artificial systems to simulate real social systems in order to explore the trajectory of social phenomena. In this paper, we combined complex system modeling theory with a computational experimental method to construct an adaptive, multi-agent-based system. For this purpose, we constructed heterogeneous firm and consumer agents. Each agent has its own properties, behavioral rules, and interaction rules, which can be adjusted according to its experiences. Our model is based on an abstraction of the real world. We first applied the model to simulate consumers’ product selection process and firms’ product innovation decision- making process. Then, we analyzed the internal mechanisms affecting consumers’ green demand and firms’ environmental behavior. The experimental results revealed that consumer preferences for products with high environmental performance encourage firms to pursue environmental innovation. When green demand among consumers is sufficiently high, firms can obtain high economic profits when they engage in environmental innovation, whereas when green demand among consumers is low, the government should intervene with mechanisms such as subsidies or incentives to encourage firms to engage in environmental innovation. However, as a rapid increase in green demand would quickly lead to a market monopoly, firms should respond to changes in the market in a timely manner to avoid being eliminated from the market.