Combination of Bayesian Network for SITRM Approach and Its Application in Industrial Critical Technology Innovation Model Selection

Scenario-based Technology Roadmapping (STRM) can solve the issue that single linear prediction technology roadmapping is difficult to be implemented in a dynamic and unstable environment due to its lack of robustness. Currently, STRM is still under theoretical research, and it has the following limitations: (1) STRM targets at enterprise-level technology roadmapping. Few analysis and researches are given to Scenario-based Industrial Technology Roadmapping (SITRM), (2) there is no analysis about industrial technology innovation model,(3) the effects and update of STRM are rarely discussed. In order to find a new method for planning industrial technology innovation, this paper presents Bayesian Network (BN) quantitative analysis in SITRM on the basis of incorporating the idea of STRM into industrial technology roadmapping. According to the research of SITRM, BN analysis has the following advantages: (1) BN can learn layers or the interrelation in SITRM. In addition, BN can express and conduct analysis of these levels and interrelations by using the Directed Acyclic Graph (DAG), (2) the Conditional Probability Table (CPT) can demonstrate the system’s multiple status and the uncertainty of the logical relation between the multiple status, (3) BN can also display the relation between industrial scenarios and industrial critical technologies even when the data is incomplete. This paper investigates the following aspects: (1) this paper constructs a topology model of SITRM and displays the direct or indirect relation between random variables in ITRM (industrial environments, industrial targets and industrial critical technologies) through the DAG. It solves the issue that relation between cross-level random variables cannot be displayed in ITRM design process. It also uses the CPT to demonstrate multiple status and the uncertainty of the logical relation between multiple statuses in SITRM in order to enhance the robustness of SITRM design, (2) based on probability deduction methods of BN, this paper performs a quantitative analysis of the planning of industrial technology innovation model and innovation path. The paper also provides specific analysis process and calculation model, (3) this paper presents the probability algorithm of the occurrence of relevant nodes based on posterior probability. It also discusses how to reconstruct and update SITRM based on this algorithm, (4) this paper verifies its methods based on the case study of the Light Emitting Diode (LED) industry in Guangdong. However, this paper also leaves some issues to be further discussed. For example, the time dimension of ITRM, the timeliness of industrial critical technologies and how to provide quantitative information for ITRM update.