Shulou(Shulou.com)06/01 Report--

This article will explain in detail how to use the neural network time series tool ntstool. The content of the article is of high quality, so the editor will share it for you as a reference. I hope you will have a certain understanding of the relevant knowledge after reading this article.

Data in nature tend to change over time. A time series is a series of statistical data arranged in the order in which they occur. The value of the data in the time series depends on the change of time, and the numerical distribution of the adjacent time has a certain regularity, which shows a certain trend or periodic change law as a whole, so the unknown data can be predicted from the known data. However, the value of each data point is accompanied by randomness, which can not be completely deduced from historical data.

Time series analysis can rely on many mathematical tools. Such as the moving level model, the quadratic moving average model and so on. In the field of artificial intelligence, a variety of intelligent algorithms can also be applied to time series analysis. Prediction can be regarded as a dynamic filtering problem. In neural networks, dynamic divine path networks with tapped delay lines can be used to deal with nonlinear filtering and prediction problems.

MATLAB neural network toolbox provides users with a time series tool ntstooL, which can solve three kinds of time series problems: nonlinear autoregression with external input, nonlinear autoregression without external input, and time delay.

Type ntstool on the command line to open the neural network time series tool

Select NonLinear Autoregressive with External (Exogenous) Input and click the Next button to proceed to the select Data step

Single force Load Example Data set button, eject

Time SeriesDataSet Chooser dialog box, select the last item Fluid Flow in Pipe in the list on the left, single

Click the Impod button to import

Click the Next button to proceed to the validation and Test Data step. Similar to the neural network fitting tool, it is necessary to divide the data set into training samples, verification samples and test samples. The default settings can be used here.

(4) Click the Next button to enter the Network Architecture step. What you need to specify in this step is the number and delay of neurons in the hidden layer, with default values of 10 and 2, respectively. Delay indicates how much data the current output is related to.

Click the Next button to proceed to the Train Network step

Click the Train button and the system starts training. The default number of iterations is 1000.

The dialog box displays the mean square error and correlation R of training samples, verification samples, and test samples. The correlation is between 0: 1, which refers to the degree of coincidence between the target output and the actual output. A value of 1 indicates a complete match, while a value of 0 indicates a mismatch.

After the training is completed, the four buttons on the right side of the dialog box become active. Plot Error Histogram button is used to display error histogram

The yellow vertical bar represents zero error, and it can be seen from the figure that the error is concentrated near zero, and the error is large.

The Plot Response button shows the trend of training train, verification data and test data.

The Plot Error Autocorrelation button is used to display error autocorrelation

In the error autocorrelation diagram, two horizontal red dotted lines represent the confidence interval, and if the error is distributed in the interval, it is acceptable. Several error lines exceed this interval, indicating that the training result is not ideal.

Click the Next button to proceed to the Evaluate Network step. Because you are using the data that comes with MATLAB, there is no appropriate test data. This step is skipped.

Click the Next button to proceed to the save Results step. Similar to the fitting tool, you can save the network and variables in this step, or export the network as a script file or simuLink model

Click the Finish button to complete the prediction process of the time series

On the neural network time series tool ntstool how to share here, I hope the above content can be of some help to you, can learn more knowledge. If you think the article is good, you can share it for more people to see.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.