Hi, how can i split Figures over two pages

[juseung Lee]

Hi, iam beginner of Latex 

i have some problem for placing some figures in my code

so could you help me or advice for me in order to solve this problem

i tried to split figures over two pages but, some figures on second page was not placed properly. on first page, it is okay but, it makes space in the upper on second page

i would like to eliminate those space on second page

Thank you

my code is as below

\documentclass{article}

\usepackage{epsfig,amssymb,amsthm,amsmath,multicol,psfrag,cite,subfigure,tabularx,graphicx,captcont}

\begin{document}

\section{Introduction}

Proportional-integral (PI) controllers with anti-windup are widely used in various industry applications. In such systems, it has recently been shown that tracking can be lost due to zero mean measurement noise. Specifically, non-zero tracking error in the form of bias occurs at system output due to the interaction between measurement noise and anti-windup. The phenomenon is named as Noise Induced Tracking Error (NITE). Occurrence of NITE has indeed been confirmed in xerographic printing systems and thermal control systems in real time computing systems .

References provide analysis of NITE for only a particular type of anti-windup but conjectured that it would occur with PI controlled systems with any type of anti-windup. The conjecture is based on the fact that the role of all anti-windups is to prevent control signals from staying into saturation, and it is this very behavior that acts as the main cause of tracking loss when anti-windup is persistently triggered by measurement noise.

The main goal of this note is to provide analyses of NITE with general types of anti-windup. This is achieved by analyzing all possible anti-windup structures presented in recent work of. Specifically,  we have shown that, for static external, static global, dynamic external, dynamic global structures of anti-windups, NITE occurs with measurement noise. Also, for each anti-windup structure, we provide NITE quantification with respect to system parameters. Similarly to, the analyses are carried out using a stochastic averaging theory  to approximate a nonlinear system driven by random process by a deterministic one.

\begin{figure}[!h]

\centering

%-----------------SUBFIGURE -1-------------%

\begin{subfigure}[foo1]{

\includegraphics[width=.8\linewidth] {back_calculation.eps}

\label{PI_AW:subfig1} }

\end{subfigure}

\end{figure}

\begin{figure}[!h]

\centering

% -------------- SUBFIGURE -2 -------------- %

\begin{subfigure}[foo2]{

\includegraphics[width=.8\linewidth] {back_calculation.eps}

\label{PI_AW:subfig2} }%

\end{subfigure}

\end{figure}

\begin{figure}[!h]

\centering

% -------------- SUBFIGURE -3 -------------- %

\begin{subfigure}[foo3]{

\includegraphics[width=.8\linewidth] {back_calculation.eps}

\label{PI_AW:subfig3} }%

\end{subfigure}

\end{figure}

\begin{figure}[!h]

\centering

% -------------- SUBFIGURE -4 -------------- %

\begin{subfigure}[foo4]{

\hspace{1.5cm}

\includegraphics[width=.8\linewidth] {back_calculation.eps}

\label{PI_AW:subfig4} }%

\end{subfigure}

\label{myfigure}

\caption{PI controlled systems with anti-windups}

\end{figure}

\end{document}