Research On Working Mode Management Of Data Acquisition System Based On Time Prediction Algorithm

Life and reliability of chip are very important now in equipment.Temperature rise will cause equipment life to reduce greatly,fault cannot be found in time results in lower reliability and common solution is that equipment goes into sleep mode during idle to dissipate as soon as possible,through the fault diagnosis to find fault in time.In order not to affect the equipment use,these functions will run during idle.Some problems will occur when those functions run,rapid fluctuation of the system power when quick start reduce life of power supply module,operation of fault diagnosis during idle may affect the equipment starts,so how to reasonable use of idle time,how to reasonable arrangement of dormancy and fault diagnosis mode during system idle become an important research subject.The research of this paper is to study the mode management of data acquisition system which works repeatedly.The modes studied in this paper include normal working mode,shallow sleep mode,deep sleep mode and self-check mode.Meanwhile,different mode management schemes are designed for different idle time length.A indicator is design to evaluate the effectiveness of mode management in this paper.The indicator is probability of effective management,whch is probability of choosing right mode management plan and carrying out corresponding mode smoothly.The aim of this paper is that system can switch from idle in 1 seconds into normal working state,and ensure probability of effective management greater than 90% at the same time.Mode management in this paper is based on idle time prediction algorithm,appropriate pattern management scheme will be selected according to the predicted idle time.The exponential smoothing algorithm,ARIMA algorithm,the GM(1,1)algorithm and the average of exponential smoothing algorithm and GM(1,1)algorithm are compared in this paper,and exponential smoothing algorithm and the GM(1,1)algorithm are choosed to design Mode management strategy.The weighted average of these two algorithms is pointed out to improve the effect of pattern management through algorithm comparison.designs.The mode management strategy of this paper is designed base on different input quantity of these two algorithms.The mode management strategy is tested in this paper,and the best threshold value and weight combination are found to make probability of effective management reached 90% through the different threshold and weight combination in test,and goal of this paper is achieved.Finally,after analysis,the finally mode management strategy compared with the strategy based on other prediction algorithm,in the priority of quick start,deep sleep time will improve at the same time the burden of power supply module reduce,the times of fault diagnosis wil be more.It shows that goal of this paper is achieved in mode management research.