Abstract: In order to detect the radiation intensity value effectively, an active online monitoring module was designed. Some functions are realized in this module, such as radiation dose detection, radiation dose rate display, overlimit alarm, GPS positioning, 4G data transmission to private cloud. GMcounter tube was used in this module as the detection device for radiation perception. The principles of boosted circuit and pulse counted circuit were also briefly introduced, in order to make GMcounter tube work effectively. As a counting device, the GMcounter tube can’t obtain the dose rate in radiation field directly. Therefore, the mapping relationship must be founded between the counted rate of GMcounter tube and dose rate measured by a radiation dosimeter. Researchers found that the polynomial function can be used as a solution for fitting the two data approximately. In order to solve the coefficients of the polynomial function, some classic algorithms can be used, such as GaussNewton algorithm, LevinbergMarquardt algorithm. However, in some cases, the satisfied simulation results may not be obtained by the classic algorithms. Considered by the good performance of the quantumbehaved particle swarm optimization (QPSO) algorithm for optimization in pathological functions, nondeterministic polynomial (NP) problems, and piecewise nonlinear problems with recursive relations, the QPSO algorithm was proposed to replace the classic algorithms to solve these coefficients. The principle and iterative formula of QPSO algorithm were introduced in this paper, and the control parameters were also analyzed. A linear value method was adopted to change the value of the control parameters in QPSO algorithm with the iteration progresses, and it is proved that the algorithm has relatively balanced between performance and robustness under such parameters. The measured comparison data were obtained by measuring the detectors based on GMcounter tube and the calibrated dosimeter, under different intensity radiation fields with the same conditions. Then the measured data were fitted by QPSO algorithm as a global function with the given function model, and high precision approximation was achieved between the fitted value and the measured dose rate. In order to further optimize the situation where the error is relatively larger between the fitted value and the measured dose rate when the counted rate is relatively smaller, autonomous optimal interval division of the fitting data was achieved based on clustering and data fusion algorithms, and the subjectivity of segmentation selection was avoided. Firstly the Euclidean distance was used to form the preliminary segmentation scheme; then the relatively better segmentation schemes were chosen by optimization with QPSO algorithm; thirdly the correlation distance was used to further optimize segmentation, and the best segmentation lastly was obtained; finally QPSO algorithm was used to calculate the final optimization result. The simulation results show that the fusion algorithm can reduce the error of the fitting function in each data segment effectively, especially for the counting rate is relatively lower, compared with a global fitting function.