In cognitive radio technology, constant changes in the environment like change in background noise, movements of the users or transmitters, interferences, etc. require the spectrum sensing parameters like detection threshold to be changed. Else, errors in spectrum sensing arises which either causes interference between transmission from primary and secondary users, or the secondary user misses an opportunity of transmission causing under usage of available spectrum bands. Thus an optimal threshold value must be determined for minimum error. Here, initially, a Death Penalty based Differential Evolution (DPDE) algorithm is proposed for constrained optimization and based on the proposed DPDE, a threshold adaptation algorithm is designed for dynamic sensing error minimization in changing environment. The performance of the proposed algorithm is compared with previously proposed gradient descend based threshold adaptation algorithm and was found to be faster and more accurate. © 2016 IEEE.