The unit of luminous intensity candela is one of the seven basic units of the International System of Units (SI). The accuracy of the luminous intensity value is directly related to other parameters. As a new type of cold light source, LED is replacing incandescent lamps and fluorescent lamps as the main lighting source due to its advantages of long life, high luminous efficiency and no pollution. Therefore, accurately measuring the average luminous intensity of LED is one of the important links to effectively evaluate the quality of LED lamps. At present, the international organization for metrology and standardization is exploring the use of LED lamps as a standard for the transmission of photometric values, and establishing a photometric value dissemination system based on LED standard lamps. It has become one of the research hotspots at home and abroad to clarify the variation law of the average luminous intensity of LED lamps with the ambient temperature, to determine the influence of the ambient temperature and its variation, and to ensure the effective traceability of the average luminous intensity of LED lamps. The current related research mostly starts from the perspective of experimental measurement to obtain the average luminous intensity of LED lamps at different ambient temperatures. However, the difference of ambient temperatures in different laboratories directly leads to the deviation of the average luminous intensity of LED lamps, which makes it difficult to obtain standard value of the average luminous intensity of LED lamps. It is even more difficult to establish a photometric value dissemination system based on LED standard lamps.For the problem that LED luminous intensity is extremely sensitive to ambient temperature, a correlation evaluation model between ambient temperature and average luminous intensity of LED based on the principle of heat conduction is established in this paper. However, changes of ambient temperature can cause changes in parameters such as the LED lamps own temperature, band contraction coefficient, temperature coefficient, thermal conductivity, radial size, air heat capacity and other parameters. The luminous intensity of LED lamps is closely related to these parameters. Therefore, it is difficult to quantitatively analyze the variation of LED luminous intensity with ambient temperature using a theoretical models. To this end, a set of average luminous intensity measuring device for LED single tube has been successfully developed. The expanded uncertainty of the average luminous intensity measured by the device is U = 2.0% (k = 2) under near-field and far-field conditions. It is verified that the measurement reliability of the device meets the national calibration specification JJF 1501-2015. The variation law of average luminous intensity for red (R120905, GaAsP, 631 nm), green (R120905, GaAsP, 631 nm) and blue (B120905, GaN, 465 nm) LED standard tubes with ambient temperature is obtained by using the measuring device, respectively. It is known that the length, width and height of the LED chip materials are 3.50 mm, 2.50 mm and 1.00 mm, respectively.The ambient temperature range is 19.0 ℃ ~ 30.0 ℃ and the temperature change interval is 0.5 ℃ in the experiment. The power supply current of LED standard tube is 20.000 mA. According to the national calibration specification JJF 1501-2015, the extended uncertainty U_k of the average luminous intensity measurement results of red, green and blue LED standard tubes with unit temperature change can be obtained at the ambient temperature of 23 ^oC. The results show that the average luminous intensity of the three color LED standard tubes shows a linear attenuation trend with the increase of ambient temperature. 23 ℃ is used as the measurement base point. When the ambient temperature change is controlled at ± 3.5 ℃, the measurement uncertainty of the average luminous intensity for the three LED standard tubes is limited to 2.0% ~ 4.6%, which all meet the U = 1.5% ~ 5.0% (k = 2) required by the national calibration specification JJF 1501-2015. Using this result, the evaluation model is modified. This enables the theoretical calculation of the average luminous intensity and emission wavelength of LED lamps disturbed by ambient temperature to be solved. Therefore, it is suggested that the ambient temperature should be controlled at 23 ℃ ± 3.5 ℃ when measuring the average luminous intensity of LED, which helps to reduce the measurement deviation of the average luminous intensity of LED under different laboratory ambient temperatures. The results provide a useful reference for establishing a photometric value dissemination system based on LED and improving the effectiveness and reliability of the traceability of photometric parameters such as LED luminous flux, illumination and luminance.