【正文】 and the temperature control error is less than 2. Temperature is the main control parameter in a thermal analyzer. In the process of detecting activation energy of chemical substances, because temperature and reaction rate is exponential relationship, we can obtain reliable, accurate test results only by achieving a precise control of temperature. However, the furnace, which used as the control object of a thermal analysis instrument, is characteristic of nonlinear, large time delayed, timevaried, strong coupled and so on, so the result using conventional PID control method is not ideal. In order to achieve a wide range of highprecision temperature control requirements, we design and implement a temperature control system of microfluidized dynamic thermal analyzer by bining fuzzy control and PID control methods. The main idea is to adopt fuzzy reasoning control methods according to different error E and errorchange Ec to get selftuning PID parameters based on the conventional PID regulator [3, 6], which greatly improved the stability and control accuracy of the temperature control system. The rest of the paper is organized as follows. In Section 2, we provide the temperature control system structure of the thermal analyzer. In Section 3, we describe the design of fuzzy controller. In Section 4, we give the experimental results. Finally, in Section 5, we give the concluding remarks. II. TEMPERATURE CONTROL SYSTEM STRUCTURE OF THE THERMAL ANALYZER Temperature control system of the thermal analyzer is posed of microcontroller ADuC841 as the control core, furnace as the controlled object and PC as the monitor machine. The overall system structure is shown in Figure 1The system uses platinum and rhodium 10type thermocouple to measure the internal temperature of the furnace. The amplified signal input to the microcontroller ADuC841 to get A/D conversion, digital filtering and scaling transformation, etc. By detecting the temperature of the cold junction of thermocouple with the sensor AD590, the cold junction pensation of thermocouple is pleted. After paring the measured temperature with the set temperature value, the controller gets the temperature deviation, pletes the fuzzy PID control algorithm, and then outputs the PWM control signal to drive triac by zerocrossing trigger circuit chip MOC3041 to control its turnon time, to change the heating power and thus to control temperature of the electric furnace. As upper monitoring puter, it is used to configure the temperature control parameters of the electric furnace, to access the detected data, to display its curve and to plete its analysis work. III. .THE FUZZY CONTROLLER of the input and output variables and the membership function The fuzzy controller takes two input and three output structure in the system. The temperature input error is E, error change is Ec. The output of fuzzy controller is Kp, KiandKd. The domain of E,Ec and Kp is {3, 2, 1, 0, 1, 2, 3}. The domain of Ki and Kd is {, , , 0, , , }. All of their fuzzy subsets are {NB, NM, NS, ZO, PS, PM, PB} [2], which represent {Negative large, negative medium, negative small, zero, positive small, positive medium, positive large} correspondingly. When the error is large, it is controlled according to its own characteristics of PID control, the fuzzy controller is not needed at this time, and the output value automatically closes to the given value. When the error bees smaller to a certain extent, the fuzzy control takes effect. The input error, error change and output membrship functions use triangular functions, which are shown in Figure 2 and 3.B. The selftuning rules of PID parameters Kp, Ki and KdWhile |E| is large: In order to speed up the system response speed, larger Kp is chosen. At the same time, in order to avoid the probable differential supersaturation because of e instantaneously being large that make the control effect beyond permission, a smaller Kd is chosen, and in order to avoid larger overshoot, thus the integral is limited by setting Ki = 0. While |E|and |Ec| are middle: In order to make the overshoot of the system response relatively small, Kpis set smaller, and in order to ensure the system response speed, Ki and Kd is chosen proper values. While |E|is small: In order to make the system have better steady state characteristic, Kp and Ki are set larger, and in order to avoid oscillation near the setpoint , Kd is set properly. When |Ec| is small, Kd is set middle, and when |Ec| is large, Kd is set small [4, 5]. According to the above control rules, the control rule table of PID parameters can be obtained. It is shown in Table 1 C. Defuzzification In this paper, we use the weighted average method to the fuzzy evaluation to get the precise control value. The formula of weighted average algorithm is shown in equation。參考文獻1. [D]..2. 張建波,[J]..3. 劉新明，唐逸輝，熱電偶的測溫原理概述，4. 陶磊,薛濤,[J].山西冶金..5. 李洪波,高思田,、納米尺度測量環(huán)境的溫度測量和控制技術[J]..6. [D]..7. 宋延壽,商樹桓,[J]..8. 黃燕巖LabVIEW平臺下熱電偶自動檢定系統(tǒng)的研究[D].. 5657.9. 黃澤銑,:10. [J]..11. 焦麗娜,[J].,.12. [J].,04,.53.13. Liu Linear Thermocouple Temperature Meter Based on Inverse14. Reference Computation Technology and Automation15. (ICICTA),2010 International Conference May 1. 138143.16. 張雄飛,段軍政,[J].,.17. 孫育才,王榮興,[M].北京:清華大學出版社,2005.18. [D].. 2426.19. （電磁兼容）設計與測試案例分析[M].北京:電子工業(yè)出版社,. 125127.20. [M].北京:國防工業(yè)出版社，2005.21. [M].北京:中國電力出版社,2008. 130133.22. 段悅,[J].微型機應用.23. 張震,王劍,[J].,3,27(3)..8084.24. Genix M.，Vairac P.，Cretin temperature surface measurement with intrinsic thermocouple[J].International Journal of Thermal 48(9). 16791682.25. 張玉泉，王元委，[J]..2325.26. Gao Feiyan,Tang Yaogeng,Luo Correction for Thermocouple Vacuum Sensor Based on Multiple Support Vector Technology and Mechatronics Automation(ICMTMA),2010 International Conference March 2010 .2. 81784.27. 徐國文,[J].2010. 12.28. Agee John T.，Masupe Shedden,Setlhaolo neuralnetwork c