The valve cooling system is a key part to ensure the safe operation of the converter station. Its function is to discharge the heat formed by the components in the valve body, so the accurate evaluation of its cooling capacity is very important. Aiming at the above problems, a bidirectional interval constrained normal cloud model based on ISSO-PP combination weighting is proposed to evaluate the cooling capacity of the converter valve. Firstly, the collected data is preprocessed and the cooling capacity evaluation index system of the valve cooling system is determined. Secondly, AHP and improved CRITIC method are used to calculate the subjective and objective weights of evaluation indicators. And the comprehensive weight is obtained by the improved shark-optimized projection pursuit method （ISSO-PP）. Finally, a forward cloud digital feature calculation method with bidirectional interval constraints is proposed in this paper, based on the data grade evaluation standard as the intermediate type. At the same time, the reverse cloud digital feature calculation method is improved, and the final certainty degree is obtained by combining with the comprehensive weight calculation. The final evaluation result is determined according to the principle of maximum certainty degree.