[en] Recently we modified the Thomson scattering diagnostic (TS) installed on LHD so that DC levels (V_DC) of all avalanche photodiodes (APD) used for detecting scattered light can be registered every 1 ms, which enabling us to make validity check on TS data taken under very intense plasma radiation. In the line of this task, we first examined how the pulse-performance of an APD degrades as the intensity of continuous light (J_DC) incident to the APD increases. We found two effects are involved in deteriorating the pulse-performance of the APD: (1) the responsivity of the APD to a pulsed light drops as J_DC increases, causing a systematic errors on the deduced electron temperature (T_e) and density (n_e); (2) the frequency response of the APD and the following circuit drops as J_DC increases, which deforms the pulse shape. The bias voltage applied to the APD (V_b) has large influence on these behaviors, showing the best overall performance for a high J_DC around V_b - 0.5 V_r, where V_r is the recommended voltage giving responsivity of 675 kV/W at 1060 nm. Considering these effects together, we set a conservative validity criterion for the pulse APD performance in term of the V_DC: V_DC < 0.5 V. The V_b=0.5 V_r setup gives much reliable T_e-profiles without a collapse in T_e-profile for a much wider range of plasma radiation intensity. With this criterion, we check the validity of T_e-and n_e-profiles of two example data. (author)