Research On Tuning Algorithm Of PID Parameters Based On Fuzzy Inference

As a classical control technique, PID control still plays an essential role in theindustrial control. The parameters of the controller entirely determine the controlperformance. An ideal set of parameters suitable for a controlled object can satisfyhigher requirements, so PID parameters tuning is the most significant step in theprocess of PID controller designing. Industrial control objects which tend to bestrongly non-linear can not be easily obtained the parameter model of the industrialcontrol objects which tend to be strongly non-linear can not be easily obtained, sothe traditional PID tuning algorithms are not applicable in this situation.Of the above problems, a PID tuning algorithm based on fuzzy inference isproposed. The main process of the system talked about is: after setting the targettemperature, PID control is put in use of controlling the system. When thetemperature is steady, the overshoot and the setting time in this control process arestored in the EEPROM of single-chip. At next starting of the system, the data storedin the EEPROM are first read out, then， the data read out are put into the fuzzycontroller as the input. After fuzzy inference process, we can get the correction ofthe PID parameters and the historical factor. The new PID parameters are composedof historical part and future part. The historical part is multiplied by the historicalfactor and historical parameters. The future part is multiplied by the complementaryof the historical factor and new parameters. The process is continued several timesuntil the control performances satisfy the target. Manual is not needed in the entireprocess, so the algorithm is suitable for the complicated control.The control core of the hardware is the STC12C5628AD single-chip. Thetemperature sampling part is the temperature sensor of DS18B20.The man-machineinterface is consist of5keys and LCD5110.The executive part is consist of solidstate relay and heating pot. The software can be divided into task manager part, PID calculating part, fuzzy inference part, control performance measure part andEEPROM data processing part. The core of the algorithms studied in this paper is todetermine the fuzzy inference rules of PID parameter tuning.The fuzzy inferencerules established based on the effect of overshoot and setting time caused by PIDparameters. The determination process of fuzzy inference rules: firstly, using thesoftware PID simulation study can determine the three parameters effect on the twocontrol indicators theoretically. Secondly, actual hardware is used in determining thereal effect of the PID parameters under practical conditions.Experimental results can be seen in the target temperature of50℃and60℃:overshoot and setting time of the fuzzy tuning PID control is significantly better thanthe traditional PID control. When the object controlled changes, the fuzzy tuningPID control can still maintain high quality control. The quality of conventional PIDcontrol declines. With the stable operation, the system can satisfy the system designgoals basically.