SCN时间序列预测模型详解（Matlab代码实现）

     目录

💥1 概述

📚2 运行结果

🎉3 参考文献

👨‍💻4 Matlab代码

---

💥1 概述

SCN（System Change Number 简称 SCN）是当Oracle数据库更新后，由DBMS自动维护去累积递增的一个数字。在Oracle中，有四种SCN，分别为：系统检查点SCN、数据文件检查点SCN、启动SCN、终止SCN。

📚2 运行结果

 

 

🎉3 参考文献

[1]王鑫,吴际,刘超,杨海燕,杜艳丽,牛文生.基于LSTM循环神经网络的故障时间序列预测[J].北京航空航天大学学报,2018,44(04):772-784.DOI:10.13700/j.bh.1001-5965.2017.0285.

👨‍💻4 Matlab代码

主函数部分代码：

% Demo Data regression of SCN
clear;
clc;
close all;
format long;
tic;

%%  Prepare the training (X, T) and test data (X2, T2) 
% X: each row vector represents one sample input.
% T: each row vector represents one sample target.
% same to the X2 and T2.
% Note: Data preprocessing (normalization) should be done before running the program.
filename = 'newdatajune10-1-2.csv';
Originaldata1 = csvread(filename,0,0,[0,0,4318,3]);
X = Originaldata1(1:2880,:);
X2 = Originaldata1(2880:4319,:);
Originaldata2 = csvread(filename,0,4,[0,4,4318,4]);
T = Originaldata2(1:2880);
T2 = Originaldata2(2880:4319);

%% Parameter Setting
L_max =250;                    % maximum hidden node number
tol = 0.0001;                    % training tolerance
T_max = 100;                    % maximun candidate nodes number
Lambdas = [0.5, 1, 5, 10, ...
    30, 50, 100, 150, 200, 250];% scope sequence
r =  [ 0.9, 0.99, 0.999, ...
    0.9999, 0.99999, 0.999999]; % 1-r contraction sequence
nB = 1;       % batch size

%% Model Initialization
M = SCN(L_max, T_max, tol, Lambdas, r , nB);
disp(M);

%% Model Training
% M is the trained model
% per contains the training error with respect to the increasing L
[M, per] = M.Regression(X, T);
disp(M);