摘要
介绍一种基于神经网络时间序列预测器的传感器数据验证方法。通过神经网络时间序列预测器获得传感器输出的预测值,并与传感器的实际输出数据进行比较,然后依据一定的判定规则,确定传感器数据的有效性和可靠性。最后,在MATLAB软件平台上,利用某热电厂的温度数据对该算法进行了仿真。仿真结果表明,这种数据验证方法的最大预测误差为5.5352℃,未超过设定的阈值,因此该方法是切实可行的。
A method is introduced in this paper, which is used for sensor verification based on neural network time series predictor. The predictive value of the sensor achieved by neural network time series predictor is compared with the sensor actual data. The validity and reliability are confirmed on the basis of some rules. The algorithm is simulated using the temperature data of some thermoelectric works based on MATLAB. The simulation results show that the most predicable error is 5.5352 and not exceeds the threshold value, so this method is visible.
出处
《仪表技术》
2007年第7期45-47,共3页
Instrumentation Technology
