New　product　development　process　is　complex,　involving　many　decisions　that　based　on　incomplete　and　unrepresentative　weak　signal　information.　In　this　case,　cognitive　bias　will　affect　decision-making　significantly,　leading　to　development　failure.　Therefore,　we　propose　a　weak　signal　evolution　prediction　method　to　enrich　weak　signal　information　and　exclude　unrepresentative　weak　signal　information.　Specifically,　the　weak　signal　evolution　prediction　method　contains　three　parts.　Firstly,　on　the　basis　of　matter-element　extension　model,　the　system　diagram　approach　is　used　to　construct　evolutionary　three-process　prediction　index　system.　Secondly,　evolutionary　matter　elements,　classical　and　joint　domains,　correlation　functions,　and　weight　coefficients　are　determined.　Finally,　the　discriminant　criterion　is　improved　to　divide　the　evolution　evaluation　stage,　and　then　weak　signal　evolution　prediction　method　is　constructed　on　the　foundation　of　improved　matter-element　extension　model.　We　take　perovskite　photovoltaic　cells　as　an　example　to　verify　the　performance　of　the　weak　signal　evolution　prediction　method　proposed　in　this　paper.　Results　showed　that　the　method　proposed　in　this　paper　has　higher　accuracy　and　applicability,　which　is　more　in　line　with　practical　situations.　The　research　results　further　expand　the　matter　element　extension　model,　providing　a　new　approach　to　reduce　cognitive　bias　in　the　process　of　new　product　development　for　enterprises,　and　offering　new　insights　for　new　product　development.　Meanwhile,　this　study　also　has　shortcomings　such　as　a　limited　research　perspective　and　a　single　case　selection.　IEEE