【正文】 to all aspects of performance of the system are decreased. Even the advanced controller will also for this reason caused the reduction of the performance of control system quality. For the controlled object is a firstorder link, according to the timedelay systems in the delay factor and time constant ratio, the size of the can be divided into large timedelay systems (tau/T ) and the general timedelay systems (tau/T ). If the system is a large delay, mon controller to effect on it is very difficult to get a satisfactory result. Sum up delay is a difficult and had to solve the problem, if the delay time parameter of controlled object, the greater the control of the system is more difficult , and the control of timedelay systems how to directly influence the effect of the actual production process, if can effectively control for timedelay systems, most of the problems can be solved, industrial production problems appeared in the delay link. So, can put forward to solve time delay problem is of great significance for industrial production.Discussed in this article is about the time delay problem, the main content of the study is in the control system of the application of Smith forecast device were analyzed, and the improved algorithm is put forward, and the fuzzy PID control algorithm is bined with Smith system applied to the temperature control. Subject es from a thermostat product of the enterprise. The temperature controller is mainly used in packaging machine heating equipment. At present the problem of the thermostat is packing machine for heating start big overshoot, steadystate levels, adjust the time is long, etc. The purpose of this paper is to discuss a suitable algorithm is applied to the temperature controller to solve the above problems. study status of delay phenomenonTime lag phenomenon is widespread, and the temperature control is a typical timedelay problem, for it can be divided into the control of the method based on model and no model. PID control, Smith prediction control, sliding mode variable structure control, robust control is a control method of the model. Smith fuzzy forecast method, adaptive fuzzy method, predictive control, neural network and expert system and so on are no model control method. After years of development, rapid development of intelligent control, control method is no longer oneness, more hot research direction is to classical control method bined with intelligent control method. Smith put forward and applied to forecastIn 1957, o. j. Smith can pensate the timedelay factor of a forecast control algorithm, this method is very effective to solve time delay problem, it is first estimated the controlled system to the outside world under the action of dynamic condition, according to the actual situation through operation can eliminate the characteristic equation of time delay factor, the aim is to make the controller action in advance, to pensate for delay link of influence on system, guarantee the system run time does not have a large impact, the steady state operation astatic, time delay factor is pensated at the same time also will improve the system response speed.Will be abstracted mathematical model of control system, using Smith estimated time delay pensation method in theory can offset time delay, the effects of the control system can be restored to normal without delay factor to affect the normal model. However, Smith estimates of deficiency is also appear in the controlled object model here, if the model of controlled object and forecast of model mismatch, Smith prediction control play a role is very small, resulting in a decline in the performance of the system, serious when even affect the stability of the system. If does not match the parameters of the large difference, Smith forecast not only will not reduce the effects of time delay, estimate the device itself will also be a serious damage factor of the jamming system.In understanding the Smith estimates of Achilles39。 heel, and Bartly Gikes R F T M gain adaptive pensation method. This method with the traditional Smith estimates of the difference is that it increased the three tools: multiplier, division arithmetic unit and leading differential link. To increase operational links to do the model output and the forecast process output division, the resulting value is able to estimate the gain setting. The estimate method has more advantages than the traditional Smith, further improve the stability and rapidity of the Smith forecast device was improved, on the basis of the original added a control link, and make the feedback channel is not a unit of negative feedback, but add a firstorder lowpass filter. This method pared with the ordinary Smith prediction control shows the advantage of better is the steady state characteristics of the system, the matching problem of the model is not very strict. Introduce gain forecast adaptive pure lag pensation controller, using the rate of change of the system output and estimated ratios identification of the output rate of change of controlled object change trend of gain, can real time part of the magnification, forecast the controller to adjust the magnification, equivalent to gain feedback from the adjustment, the real output of the system and forecast the output of the division, to gain online adjust to the change of the tracking system, to ensure that the system requirements. The Smith prediction control, adaptive and fuzzy control, bining the advantages of the proposed fuzzy adaptive Smith prediction control and fully dig Smith prediction control potential. PID controlFor system control problems in the process of industrial production is vital, PID control as important in the classical control theory was widely used control method, and it can be seen that PID control theory is easy to understand, the controller structure is relatively simple, robust and a series of advantages. For using PID controller, it is important pa