• ๋Œ€ํ•œ์ „๊ธฐํ•™ํšŒ
Mobile QR Code QR CODE : The Transactions of the Korean Institute of Electrical Engineers
  • COPE
  • kcse
  • ํ•œ๊ตญ๊ณผํ•™๊ธฐ์ˆ ๋‹จ์ฒด์ด์—ฐํ•ฉํšŒ
  • ํ•œ๊ตญํ•™์ˆ ์ง€์ธ์šฉ์ƒ‰์ธ
  • Scopus
  • crossref
  • orcid

  1. (Dept. of Electrical Engineering, Dong-A University, Republic of Korea.)



Ball and beam system, PID control, Gain-scaling factor, Gain-tuning, Dual control

1. ์„œ ๋ก 

ํ—ฌ๋ฆฌ์ฝฅํ„ฐ(Helicopter)์™€ 2์กฑ ๋ณดํ–‰ ๋กœ๋ด‡(Biped Robot)์˜ ์œ„์น˜ ๋ฐ ์ž์„ธ์ œ์–ด์˜ ๊ธฐ๋ณธ์ด ๋˜๋Š” ๋ณผ-๋น” ์‹œ์Šคํ…œ์€ ์„œ๋ณด๋ชจํ„ฐ์™€ ๋น”์ด ์ง์ ‘ ์—ฐ๊ฒฐ๋˜์–ด ๋ชจํ„ฐ์˜ ์ œ์–ด๋ฅผ ํ†ตํ•ด ์‡ ๊ณต์˜ ์œ„์น˜๋ฅผ ์ œ์–ดํ•˜๋Š” ์‹œ์Šคํ…œ์œผ๋กœ์„œ, ์‹œ์Šคํ…œ์˜ ์•ˆ์ •ํ™” ๋ฐ ์ œ์–ด์™€ ๊ด€๋ จํ•˜์—ฌ ๋‹ค์–‘ํ•œ ์ œ์–ด๊ธฐ๋ฒ•์ด ์‚ฌ์šฉ๋˜์–ด ์™”๋‹ค [1]-[5]. PID ์ œ์–ด๊ธฐ๋Š” ๋น„๋ก€, ์ ๋ถ„, ๋ฏธ๋ถ„ ์„ธ ๊ฐ€์ง€ ์ œ์–ด ์š”์†Œ๋ฅผ ๊ฒฐํ•ฉํ•œ ํ”ผ๋“œ๋ฐฑ ์ œ์–ด๊ธฐ๋กœ, ์ œ์–ดํ•˜๊ณ ์ž ํ•˜๋Š” ๋Œ€์ƒ์˜ ์ถœ๋ ฅ๊ฐ’์„ ๋ชฉํ‘œ๊ฐ’(์„ค์ •๊ฐ’)๊ณผ ๋น„๊ตํ•ด ์˜ค์ฐจ๋ฅผ ๊ณ„์‚ฐํ•˜๊ณ , ์ด ์˜ค์ฐจ๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์ œ์–ด ์‹ ํ˜ธ๋ฅผ ๊ณ„์‚ฐํ•˜๋Š” ์ œ์–ด๊ธฐ๋ฒ•์ด๋‹ค [5]-[9].

๋น„๋ก€ ์ œ์–ด๋Š” ํ˜„์žฌ ์˜ค์ฐจ ํฌ๊ธฐ์— ๋น„๋ก€ํ•œ ์ œ์–ด ์‹ ํ˜ธ๋ฅผ ์ œ๊ณตํ•˜์—ฌ ์‹œ์Šคํ…œ์˜ ๋น ๋ฅธ ์‘๋‹ต์„ ๊ฐ€๋Šฅํ•˜๊ฒŒ ํ•˜์ง€๋งŒ, ๊ณผ๋„ํ•œ ๋น„๋ก€ ์ด๋“์€ ์˜ค๋ฒ„์ŠˆํŒ…๊ณผ ์ง„๋™์„ ์œ ๋ฐœํ•  ์ˆ˜ ์žˆ๋‹ค. ์ ๋ถ„ ์ œ์–ด๋Š” ์˜ค์ฐจ์˜ ๋ˆ„์  ํ˜„ํ™ฉ์„ ๋ฐ˜์˜ํ•ด ์ •์ƒ์ƒํƒœ ์˜ค์ฐจ๋ฅผ ์ œ๊ฑฐํ•˜์—ฌ ๋ชฉํ‘œ์น˜์— ์ •ํ™•ํžˆ ๋„๋‹ฌํ•˜๋„๋ก ๋•๋Š” ๋ฐ˜๋ฉด, ๋ฐ˜์‘ ์†๋„๋ฅผ ๋Šฆ์ถ”๊ณ  ๊ณผ๋„ํ•œ ์ง„๋™์„ ๋ฐœ์ƒ์‹œํ‚ฌ ์œ„ํ—˜์ด ์žˆ๋‹ค. ๋ฏธ๋ถ„ ์ œ์–ด๋Š” ์˜ค์ฐจ์˜ ๋ณ€ํ™” ์†๋„๋ฅผ ์˜ˆ์ธกํ•ด ์ œ์–ด ์‹ ํ˜ธ์— ๋ฐ˜์˜ํ•จ์œผ๋กœ์จ ์ง„๋™๊ณผ ์˜ค๋ฒ„์ŠˆํŒ…์„ ์™„ํ™”ํ•˜์—ฌ ์‹œ์Šคํ…œ์„ ๋” ์•ˆ์ •์ ์œผ๋กœ ๋งŒ๋“ ๋‹ค. ์ด๋Ÿฌํ•œ ์„ธ ๊ฐ€์ง€ ์š”์†Œ์˜ ์กฐํ•ฉ์œผ๋กœ PID ์ œ์–ด๊ธฐ๋Š” ๋‹ค์–‘ํ•œ ์‹œ์Šคํ…œ์—์„œ ์‘๋‹ต ์†๋„, ์•ˆ์ •์„ฑ, ์ •ํ™•์„ฑ์„ ๊ท ํ˜• ์žˆ๊ฒŒ ๊ฐœ์„ ํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ํŠนํžˆ ๋ณผ-๋น” ์‹œ์Šคํ…œ๊ณผ ๊ฐ™์€ ๋น„์„ ํ˜• ๋ฐ ๋ถˆ์•ˆ์ •ํ•œ ๋™์  ์‹œ์Šคํ…œ์—์„œ๋„ ํšจ๊ณผ์ ์ธ ์ œ์–ด ์„ฑ๋Šฅ์„ ๋ฐœํœ˜ํ•œ๋‹ค [1],[5]. ํ•˜์ง€๋งŒ, PID ์ œ์–ด๊ธฐ๋Š” ์„ธ ๊ฐ€์ง€ ์ด๋“ ํŒŒ๋ผ๋ฏธํ„ฐ์˜ ์ ์ ˆํ•œ ํŠœ๋‹์ด ํ•„์š”ํ•˜๋ฉฐ, ์ด๋Š” ์‹œ๊ฐ„์†Œ๋ชจ์ ์ด๊ณ  ๋ณต์žกํ•œ ๊ณผ์ •์ด๋‹ค. ํŠนํžˆ ๋น„์„ ํ˜• ์‹œ์Šคํ…œ์—์„œ๋Š” ์ „ํ†ต์ ์ธ ํŠœ๋‹ ๋ฐฉ๋ฒ•(Ziegler-Nichols ๋“ฑ)์˜ ํ•œ๊ณ„๋กœ ์ธํ•ด ๋ฐ˜๋ณต์ ์ธ ์‹œํ–‰์ฐฉ์˜ค๊ฐ€ ๋ถˆ๊ฐ€ํ”ผํ•˜๋ฉฐ, ๋ฏธ๋ถ„ ์ œ์–ด๊ธฐ์˜ ์žก์Œ ๋ฏผ๊ฐ์„ฑ์€ ์‹ค์ œ ๊ตฌํ˜„์—์„œ ์ถ”๊ฐ€์ ์ธ ๊ณ ๋ ค์‚ฌํ•ญ์„ ์š”๊ตฌํ•œ๋‹ค.

๋ณธ ๋…ผ๋ฌธ์—์„œ๋Š” ๋ณผ-๋น”์‹œ์Šคํ…œ์—์„œ ์„œ๋ณด๋ชจํ„ฐ์˜ ๊ฐ๋„๋ฅผ ๊ฐ€์ƒ์ž…๋ ฅ์œผ๋กœ ๊ฐ€์ •ํ•œ๋‹ค. ์ด ์ž…๋ ฅ์ „์••๊ณผ ๋ชจํ„ฐ๊ฐ๋„ 2๊ฐœ์˜ ์ž…๋ ฅ์„ ํ†ตํ•ด 4์ฐจ ์ƒํƒœ๋ฐฉ์ •์‹์„ 2์ฐจ ์ƒํƒœ๋ฐฉ์ •์‹ 2๊ฐœ๋กœ ๋ถ„๋ฆฌํ•˜์—ฌ ํ‘œํ˜„ํ•œ๋‹ค. ์ด ๋•Œ, ๊ณผ๋„ ์‘๋‹ต ์‹œ $u_{1}$์˜ ๋ณ€ํ™”๊ฐ€ ์ž‘๋‹ค๊ณ  ์ œํ•œํ•œ ์ƒํ™ฉ์„ ๊ฐ€์ •ํ•œ๋‹ค. ์ด๋Ÿฐ ๋ถ„๋ฆฌ๊ตฌ์กฐ๋Š” ๋ณต์žก์„ฑ์ด ์ž‘์•„์ง€๊ธฐ์— ๋™์ ํŠน์„ฑ์„ ๋ช…ํ™•ํ•˜๊ฒŒ ๋ถ„์„ํ•˜๊ณ  ๊ฐœ๋ณ„์ ์œผ๋กœ ์ œ์–ด๊ธฐ๋ฅผ ์„ค๊ณ„ํ•  ์ˆ˜ ์žˆ๊ฒŒ ํ•ด์ฃผ๋ฉฐ [1],[5], ๊ณผ๋„์‘๋‹ต์‹œ๊ฐ„, ์˜ค๋ฒ„์ŠˆํŠธ ๋“ฑ ๋™ํŠน์„ฑ ํ‰๊ฐ€๊ฐ€ ์ง๊ด€์ ์œผ๋กœ ๊ฐ€๋Šฅํ•ด ๋น ๋ฅธ ์„ฑ๋Šฅ ํŠœ๋‹์— ์šฉ์ดํ•˜๊ฒŒ ํ•ด์ค€๋‹ค.

๋˜ํ•œ, ๋…ผ๋ฌธ [9]์—์„œ๋Š” DC๋ชจํ„ฐ์—์„œ ์ด๋Ÿฐ PID์ œ์–ด๊ธฐ์˜ ์ด๋“ ํŠœ๋‹์„ ์šฉ์ดํ•˜๊ฒŒ ํ•˜๊ธฐ ์œ„ํ•ด $\epsilon$์„ ์ถ”๊ฐ€ํ•œ $\epsilon$-PID์ œ์–ด๊ธฐ๋ฅผ ์ œ์•ˆํ•˜์˜€๋‹ค. ์ด ์ ์„ ์ฐธ์กฐํ•˜์—ฌ ๋ณผ-๋น” ์‹œ์Šคํ…œ์—์„œ ์„œ๋ณด๋ชจํ„ฐ ๊ฐ๋„์ œ์–ด๋ฅผ ์œ„ํ•ด ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$์„ ์ถ”๊ฐ€ํ•œ $\epsilon_{1}$-PID์ œ์–ด๊ธฐ๋ฅผ ์„ค๊ณ„ํ•˜๊ณ , ์‡ ๊ณต ์œ„์น˜์ œ์–ด๋ฅผ ์œ„ํ•ด $\epsilon_{2}$-PD์ œ์–ด๊ธฐ๋ฅผ ์„ค๊ณ„ํ•˜์˜€๋‹ค. ์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ํ•œ ์œ„์น˜, ๊ฐ๋„ ์ถ”์ข… ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์„ค๊ณ„ํ•˜๊ณ , ๊ฐ ๋‚ด๋ถ€/์™ธ๋ถ€ ํ๋ฃจํ”„์˜ ์ „๋‹ฌํ•จ์ˆ˜๋ฅผ ํ†ตํ•ด ์˜ค๋ฒ„์ŠˆํŠธ, ์ •์ฐฉ์‹œ๊ฐ„, ์ •์ƒ์ƒํƒœ์˜ค์ฐจ๋ฅผ ์ˆ˜์‹์ ์œผ๋กœ ํ‘œํ˜„ํ•˜์—ฌ $\epsilon$์˜ ์œ ์šฉ์„ฑ์„ MATLAB/Simulink๋ฅผ ํ†ตํ•ด ๋ถ„์„ํ•œ๋‹ค.

๋งˆ์ง€๋ง‰์œผ๋กœ ๋ถ„์„ํ•œ $\epsilon$์˜ ์œ ์šฉ์„ฑ์— ๋”ฐ๋ฅธ ์ด๋“ํŠœ๋‹๋ฐฉ์•ˆ์„ ์ œ์•ˆํ•˜๊ณ , Quanser์™€ ์—ฐ๋™๋œ MATLAB/Simulink์™€ ์‹ค์ œ ๋ณผ-๋น” ์žฅ์น˜๋ฅผ ํ†ตํ•ด ์‹คํ—˜์„ ์ง„ํ–‰ํ•œ๋‹ค. ์ด๋ฅผํ†ตํ•ด ์ œ์•ˆํ•œ ํŠœ๋‹๋ฐฉ์•ˆ์—์„œ $\epsilon$์˜ ํšจ์šฉ์„ฑ์„ ์ž…์ฆํ•œ๋‹ค. ํŠนํžˆ ์„œ๋ณด๋ชจํ„ฐ์˜ ๊ฐ๋„์™€ ์‡ ๊ณต์˜ ์œ„์น˜์˜ ์„ผ์‹ฑ๊ณผ์ •์„ ํ†ตํ•ด ์ถ”์ข…๊ฐ’์„ ๋งŒ๋“ค์–ด ์‡ ๊ณต์ด ์›ํ•˜๋Š” ์œ„์น˜๋กœ ์ด๋™ํ•˜๋„๋ก ์ œ์–ด๊ธฐ๋ฅผ ์„ค๊ณ„ํ•˜๊ณ , ์ด๋ฅผ ํ†ตํ•ด ๋ณผ-๋น” ์‹œ์Šคํ…œ์˜ ํŠœ๋‹๋ฌธ์ œ๋ฅผ ํ•ด๊ฒฐํ•˜๊ณ  ์ถœ๋ ฅ ์„ฑ๋Šฅ์ด ํ–ฅ์ƒ๋˜๋Š” ๊ฒƒ์„ ๋ชฉํ‘œ๋กœ ํ•œ๋‹ค.

2. ๋ณผ-๋น” ์‹œ์Šคํ…œ์˜ ์ƒํƒœ๋ฐฉ์ •์‹ ๋ฐ ์„œ๋ธŒ ์‹œ์Šคํ…œ์œผ๋กœ์˜ ์žฌ๊ตฌ์„ฑ

๋‹ค์Œ์€ ๋ณผ-๋น” ์‹œ์Šคํ…œ์˜ ๊ทผ์‚ฌํ™”๋œ ๋™์—ญํ•™ ๋ฐฉ์ •์‹์ด๋‹ค [4].

(1)
$\ddot{x}= K_{bb}\sin\theta$
(2)
$\ddot{\theta}= -\frac{1}{\tau}\dot{\theta}+\frac{K}{\tau}V_{m}$

์—ฌ๊ธฐ์„œ $x, \theta$๋Š” ๊ฐ๊ฐ ์‡ ๊ณต์˜ ์œ„์น˜$[m]$, ๋ชจํ„ฐ์˜ ๊ฐ๋„$[rad]$๋ฅผ ๋‚˜ํƒ€๋‚ด๋ฉฐ $K_{bb}= 0.419[m/s^{2}\cdot rad]$๋Š” BB01์˜ ๋ชจ๋ธ ์ด๋“, $\tau =0.0248[s]$๋Š” SRV02์˜ ์‹œ์ •์ˆ˜, $K =1.53[rad/V\cdot s]$๋Š” SRV02์˜ ์ •์ƒ์ƒํƒœ์ด๋“, $V_{m}$์€ SRV02์˜ ์ž…๋ ฅ ์ „์••์ด๋‹ค.

์ƒํƒœ๋ณ€์ˆ˜๋ฅผ $[x_{1}, x_{2}, x_{3}, x_{4}]^{T}=[x, \dot{x}, \theta , \dot{\theta}]^{T}$๋กœ ์žก๊ณ , ์„œ๋ณด๋ชจํ„ฐ์˜ ๊ฐ๋„์ธ $\theta$๋Š” ๊ฐ€์ƒ์ž…๋ ฅ $u_{1}$์œผ๋กœ ์ •์˜ํ•˜์—ฌ SRV02์˜ ์ž…๋ ฅ์—ญํ• ์„ ํ•œ๋‹ค. BB01 ์ž…๋ ฅ์ „์••์ธ $u_{2}=V_{m}$๊ณผ ์„œ๋ณด๋ชจํ„ฐ์˜ ๊ฐ๋„์ธ $u_{1}=\theta$์œผ๋กœ 2๊ฐœ์˜ ์ž…๋ ฅ์„ ํ†ตํ•ด 4์ฐจ ์ƒํƒœ๋ฐฉ์ •์‹์„ 2์ฐจ ์ƒํƒœ๋ฐฉ์ •์‹ 2๊ฐœ๋กœ ๋ถ„๋ฆฌํ•˜์—ฌ ์„ธ์šด๋‹ค. ์ด ๋•Œ, ๋ณผ-๋น” ๋™์—ญํ•™์—์„œ ๋ชจํ„ฐ๊ฐ๋„์˜ ๋น„์„ ํ˜• ํ•ญ$(\sin \theta)$์ด ๊ฑฐ์˜ 0์— ๊ทผ์ ‘ํ•˜๋ฏ€๋กœ ๊ณผ๋„์‘๋‹ต ์‹œ $u_{1}$์˜ ๋ณ€ํ™”๊ฐ€ ์ž‘๋‹ค๊ณ  ์ œํ•œํ•œ ์ƒํ™ฉ$(\dot{u}_{1}\approx 0)$์„ ๊ฐ€์ •ํ•œ๋‹ค. ์ด๋Ÿฌํ•œ ์ƒํ™ฉ๋ฅผ ์„ค์ •ํ•จ์œผ๋กœ์จ, ๊ฐ 2์ฐจ ๋ชจ๋ธ์€ ๊ฐœ๋ณ„์ ์œผ๋กœ ์ œ์–ด๊ฐ€ ๊ฐ€๋Šฅํ•˜๋ฉฐ ๊ณผ๋„๊ตฌ๊ฐ„๋™์•ˆ ์ƒํ˜ธ ๊ฐ„์„ญํšจ๊ณผ๊ฐ€ ๋ฌด์‹œ๋  ๋งŒํผ ์ž‘๋‹ค๋Š” ๋…ผ๋ฆฌ์  ๊ธฐ๋ฐ˜์ด ๋งˆ๋ จ๋œ๋‹ค.

(3)
$BB01 \begin{cases} \dot{x}_{1}= x_{2}\\ \dot{x}_{2}=K_{bb}\sin u_{1} \end{cases}$
(4)
$SRV02 \begin{cases} \dot{x}_{3}=x_{4}\\ \dot{x}_{4}=-\frac{1}{\tau}x_{4}+\frac{K}{\tau}u_{2} \end{cases}$

์‹ (3), (4)๋Š” ๋ณผ-๋น” ์‹œ์Šคํ…œ์˜ ์ƒํƒœ๋ฐฉ์ •์‹์ด๋‹ค. ์ด๋Ÿฌํ•œ ๋ถ„๋ฆฌ ๊ตฌ์กฐ๋กœ์˜ ์žฌ๊ตฌ์„ฑ์€ ๋ณต์žก์„ฑ์„ ๊ฐ์†Œ์‹œ์ผœ ๊ฐ ์‡ ๊ณต๊ณผ ๋ชจํ„ฐ์˜ ๋™์ ํŠน์„ฑ์„ ๋ช…ํ™•ํ•˜๊ฒŒ ๋ถ„์„ํ•˜๊ณ  ๊ฐœ๋ณ„์ ์œผ๋กœ ์ œ์–ด๊ธฐ๋ฅผ ์„ค๊ณ„ํ•  ์ˆ˜ ์žˆ๋‹ค. ๋˜ํ•œ, ๊ณผ๋„์‘๋‹ต ์‹œ๊ฐ„, ์˜ค๋ฒ„์ŠˆํŠธ ๋“ฑ ๋™ํŠน์„ฑ ํ‰๊ฐ€๊ฐ€ ๊ธฐ์กด 4์ฐจ ์‹œ์Šคํ…œ๋ณด๋‹ค ์ง๊ด€์ ์œผ๋กœ ๊ฐ€๋Šฅํ•ด ๋น ๋ฅธ ์„ฑ๋ŠฅํŠœ๋‹์„ ์šฉ์ดํ•˜๊ฒŒ ํ•ด์ค€๋‹ค.

์—ฌ๊ธฐ์„œ $x_{1}$์€ ์‡ ๊ณต์˜ ์œ„์น˜, $x_{2}$๋Š” ์‡ ๊ณต์˜ ์†๋„, $x_{3}$์ธ $\theta$๋Š” ์„œ๋ณด๋ชจํ„ฐ์˜ ๊ฐ๋„, $x_{4}$๋Š” ์„œ๋ณด๋ชจํ„ฐ์˜ ๊ฐ์†๋„๋ฅผ ๋œปํ•œ๋‹ค. ์ƒํƒœ๋ฐฉ์ •์‹ (3), (4)์— Jacobian ์„ ํ˜•ํ™”๋ฅผ ์ด์šฉํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™์ด ํ‘œํ˜„ํ•  ์ˆ˜ ์žˆ๋‹ค [10].

(5)
$BB01 \begin{bmatrix}\dot{x}_{1}\\\dot{x}_{2}\end{bmatrix}=\begin{bmatrix}0&1\\0&0\end{bmatrix}\begin{bmatrix}x_{1}\\x_{2}\end{bmatrix}+\begin{bmatrix}0\\K_{bb}\end{bmatrix}u_{1}$
(6)
$SRV02 \begin{bmatrix}\dot{x}_{3}\\\dot{x}_{4}\end{bmatrix}=\begin{bmatrix}0&1\\0&-\frac{1}{\tau}\end{bmatrix}\begin{bmatrix}x_{3}\\x_{4}\end{bmatrix}+\begin{bmatrix}0\\\frac{K}{\tau}\end{bmatrix}u_{2}$

์‹ (5), (6)์€ $x$์™€ $\theta$์— ๋Œ€ํ•œ ์ƒํƒœ๋ณ€์ˆ˜๋กœ ์‡ ๊ณต์œ„์น˜์™€ ๋ชจํ„ฐ๊ฐ๋„์˜ ์ถ”์ข…์— ํ•„์š”ํ•œ ์‹œ๋ฎฌ๋ ˆ์ด์…˜ ๋ฐ์ดํ„ฐ๋ฅผ ์–ป๊ธฐ ์œ„ํ•ด์„  $x$๊ณผ $\theta$์— ๋Œ€ํ•œ ๊ธฐ์ดˆ์ ์ธ ๋ฐฉ์ •์‹์„ ์ •์˜ํ•ด์•ผํ•œ๋‹ค. ์ด๋ฅผ ์œ„ํ•ด ๋ณผ-๋น” ์‹œ์Šคํ…œ์˜ ์ƒํƒœ๋ฐฉ์ •์‹์„ ์ •์˜ํ•˜์˜€๋‹ค.

3. ์ด์ค‘ $\epsilon$-PID/PD ์ œ์–ด๊ธฐ ์„ค๊ณ„ ๋ฐ ์‹œ์Šคํ…œ ์•ˆ์ •์„ฑ ๋ถ„์„

๋ณธ ์ œ์–ด๋ชฉํ‘œ๋Š” ๋ณผ-๋น” ์‹œ์Šคํ…œ์—์„œ ์ถ”์ข…ํ•˜๋Š” ๋ ˆํผ๋Ÿฐ์Šค ๊ฐ’์œผ๋กœ ๋น ๋ฅด๊ณ  ์•ˆ์ •ํ•˜๊ฒŒ ์‡ ๊ณต์œ„์น˜๋ฅผ ์ด๋™์‹œํ‚ค๋Š” ๊ฒƒ์ด๋‹ค. ์‹ (5),(6)์— ๋”ฐ๋ฅธ ์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๊ธฐ ์„ค๊ณ„๋ฅผ ์œ„ํ•œ ๋ธ”๋ก์„ ๋„๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

๊ทธ๋ฆผ 1. ์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๊ธฐ์— ๋”ฐ๋ฅธ ๋ณผ-๋น” ์‹œ์Šคํ…œ ๋ชจ๋ธ

Fig. 1. Ball-Beam system model under dual $\epsilon$-PID/PD controller

../../Resources/kiee/KIEE.2025.74.12.2310/fig1.png

Fig. 1์—์„œ $C_{1}(s)$, $C_{2}(s)$๋Š” ๊ฐ๊ฐ์˜ ์„œ๋ธŒ์‹œ์Šคํ…œ์„ ์ œ์–ดํ•˜๋Š” ์ œ์–ด๊ธฐ์ด๋ฉฐ, $d_{1}$์€ ramp์— ๋Œ€ํ•œ ์™ธ๋ž€์ด๋‹ค. $x_{1d}$๋Š” desired $x_{1}$, ์ฆ‰ $x_{1}$์ด ํŠธ๋ž˜ํ‚น ํ•˜๋ ค๋Š” ๊ฐ’, $x_{3d}$๋Š” $x_{3}$์ด ํŠธ๋ž˜ํ‚น ํ•˜๋ ค๋Š” ๊ฐ’์œผ๋กœ ์ •์˜ํ•œ๋‹ค.

3.1 ๋‚ด๋ถ€ ํ๋ฃจํ”„ ์ œ์–ด๊ธฐ ์„ค๊ณ„ ๋ฐ ์•ˆ์ •์„ฑ ๋ถ„์„

3.1.a $\epsilon_{1}$-PID์ œ์–ด๊ธฐ ์„ค๊ณ„

$C_{1}(s)$๋Š” $x_{3}$๊ฐ€ $x_{3d}=u_{1}$๋กœ ์ถ”์ข…์ด ๋˜๋„๋ก ์„ค๊ณ„๋  ์ œ์–ด๊ธฐ ์ด๋ฉฐ $C_{1}(s)$์˜ ์ถœ๋ ฅ์€ $u_{2}$์ด๋‹ค.

2์žฅ์—์„œ์˜ ์„ค์ •๊ณผ ๊ฐ™์ด ๊ณผ๋„์‘๋‹ต ์‹œ $u_{1}$์˜ ๋ณ€ํ™”๊ฐ€ ์ž‘๋‹ค๊ณ  ๊ฐ€ ์ •ํ•˜๋ฉฐ$(\dot{u}_{1}\approx 0)$, ๋”ฐ๋ผ์„œ ์‹ (5)์™€ ์‹ (6)์„ ๋ถ„๋ฆฌํ•˜์—ฌ ํ‘œํ˜„ ํ•œ๋‹ค. ์ถ”์ข…์˜ค์ฐจ๋Š” $e_{3}=x_{3}-u_{1}, e_{4}=x_{4}$๋กœ ์ •์˜ํ•˜๊ณ , ์‹ (6)์„ ์ •๋ฆฌํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(7)
$\dot{e}_{3}= e_{4}$
(8)
$\dot{e}_{4}= -\frac{1}{\tau}e_{4}+\frac{K}{\tau}u_{2}$

์‹ (6)์„ ํ–‰๋ ฌํ˜•ํƒœ $z=[\dot{e}_{3}, \dot{e}_{4}, e_{3}]^{T}$๋กœ ์ •๋ฆฌํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(9)
$\dot{z}=\begin{bmatrix}\dot{e}_{4}\\-\frac{1}{\tau}\dot{e}_{4}+\frac{K}{\tau}\dot{u}_{2}\\\dot{e}_{3}\end{bmatrix}=\begin{bmatrix}z_{2}\\-\frac{1}{\tau}z_{2}+\frac{K}{\tau}\dot{u}_{2}\\z_{1}\end{bmatrix}$

์ƒํƒœ๋ฐฉ์ •์‹์€ $\dot{z}= Az+B\left[-\frac{1}{\tau}z_{2}+\frac{K}{\tau}\dot{u}_{2}\right]$๋กœ ํ‘œํ˜„ํ•  ์ˆ˜ ์žˆ๋‹ค. ์ด ๋•Œ, $\dot{u}_{2}=v$์ด๊ณ , A์™€ B๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(10)
$\dot{z}=\begin{bmatrix}z_{2}\\-\frac{1}{\tau}z_{2}+\frac{K}{\tau}v\\z_{1}\end{bmatrix}=\begin{bmatrix}0&1&0\\0&0&0\\1&0&0\end{bmatrix}z +\begin{bmatrix}0\\1\\0\end{bmatrix}\left[-\frac{1}{\tau}z_{2}+\frac{K}{\tau}v\right]$

์ด์ œ nominal ์‹œ์Šคํ…œ ํŒŒ๋ผ๋ฏธํ„ฐ๋ฅผ ์ด์šฉํ•˜์—ฌ ํ”ผ๋“œ๋ฐฑ ์„ ํ˜•ํ™” ์ œ์–ด๊ธฐ๋ฅผ ๊ตฌํ•  ์ˆ˜ ์žˆ๋‹ค [10].

(11)
$v =\frac{\tau\omega +z_{2}}{K}$

์ด ๋•Œ ๋‚ด๋ถ€ ์ œ์–ด๊ธฐ $\omega$๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์ด ํ‘œํ˜„ํ•  ์ˆ˜ ์žˆ๋‹ค.

(12)
$\omega =\frac{k_{1}}{\epsilon_{1}^{2}}z_{1}+\frac{k_{2}}{\epsilon_{1}}z_{2}+\frac{k_{3}}{\epsilon_{1}^{3}}z_{3}$

์‹ (12)์„ ์‹ (11)์— ๋Œ€์ž…ํ•˜์—ฌ $v =\dot{u}_{2}$ ์— ๋Œ€ํ•œ ์‹์œผ๋กœ ์ •๋ฆฌํ•˜๋ฉด $\epsilon_{1}$-PID์ œ์–ด๊ธฐ๋ฅผ ์„ค๊ณ„ํ•  ์ˆ˜ ์žˆ๋‹ค.

(13)
$v =\frac{\tau k_{1}}{K\epsilon_{1}^{2}}z_{1}+\left(\frac{\tau k_{2}}{K\epsilon_{1}}+\frac{1}{K}\right)z_{2}+\frac{\tau k_{3}}{K\epsilon_{1}^{3}}z_{3}$
(14)
$u_{2}(t)=c_{1}(t) =K_{P 1}(\epsilon_{1})e_{3}+K_{D 1}(\epsilon_{1})\dot{e}_{3}+K_{I 1}(\epsilon_{1})\int_{0}^{t}e_{3}(s)ds$

์ด ๋•Œ $K_{P}(\epsilon_{1}), K_{D}(\epsilon_{1}), K_{I}(\epsilon_{1})$๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(15)
$K_{P 1}(\epsilon_{1})=\frac{\tau k_{1}}{K\epsilon_{1}^{2}}, K_{D 1}(\epsilon_{1})=\frac{\tau k_{2}}{K\epsilon_{1}}+\frac{1}{K}, K_{I 1}(\epsilon_{1})=\frac{\tau k_{3}}{K\epsilon_{1}^{3}}$

3.1.b ๋‚ด๋ถ€ ํ๋ฃจํ”„์˜ ์•ˆ์ •์„ฑ ํŒ๋ณ„

SRV02์˜ ๋™์—ญํ•™๋ฐฉ์ •์‹๊ณผ $\epsilon_{1}$-PID์ œ์–ด๊ธฐ ์‹์˜ ๋ผํ”Œ๋ผ์Šค๋ณ€ํ™˜์„ ํ†ตํ•ด ๊ตฌํ•œ ๊ฐ๊ฐ์˜ ์ „๋‹ฌํ•จ์ˆ˜ $G_{1}(s)$,$C_{1}(s)$๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(16)
$G_{1}(s)=\frac{K}{s(\tau\cdot s+1)}$
(17)
$C_{1}(s)= K_{P 1}(\epsilon_{1})+K_{D 1}(\epsilon_{1})s+\frac{K_{I 1}(\epsilon_{1})}{s}$

์™ธ๋ž€ $d_{1}$์€ 0์œผ๋กœ ๊ฐ„์ฃผํ•˜๊ณ , ๋‚ด๋ถ€ ํ๋ฃจํ”„ ์ „๋‹ฌํ•จ์ˆ˜ $T_{1}(s)$๋ฅผ ํ†ตํ•ด ํŠน์„ฑ๋ฐฉ์ •์‹ $1+G_{1}(s)C_{1}(s)=0$์„ ๊ตฌํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(18)
$\tau s^{3}+(1+K\cdot K_{D1}(\epsilon_{1}))s^{2}+K\cdot K_{P 1}(\epsilon_{1})s +K\cdot K_{I 1}(\epsilon_{1})=0$

์‹ (18)๋ฅผ ํ†ตํ•ด Routh-HurwitzํŒ๋ณ„๋ฒ•์„ ์ด์šฉํ•œ๋‹ค.

$\begin{array}{c|cc} s^3 & \tau & K \cdot K_{P1}(\epsilon_1) \\ s^2 & 1+K \cdot K_{D1}(\epsilon_1) & K \cdot K_{I1}(\epsilon_1) \\ s^1 & \frac{\begin{vmatrix} \tau & K \cdot K_{P1}(\epsilon_1) \\ 1+K \cdot K_{D1}(\epsilon_1) & K \cdot K_{I1}(\epsilon_1) \end{vmatrix}}{-(1+K \cdot K_{D1}(\epsilon_1))} & 0 \\ s^0 & K \cdot K_{I1}(\epsilon_1) & 0 \end{array}$

์ œ์–ด๊ธฐ๊ฐ€ ์•ˆ์ •ํ•œ $k_{1}$,$k_{2}$,$k_{3}$,$\epsilon_{1}$์˜ ๋ฒ”์œ„๋ฅผ ์•„๋ž˜์™€ ๊ฐ™์ด ๊ตฌํ•  ์ˆ˜ ์žˆ๋‹ค.

(19)
$\frac{\frac{0.0248k_{1}}{\epsilon_{1}^{2}}\left(\frac{0.0248k_{2}}{\epsilon_{1}}+2\right)-\frac{0.0248^{2}k_{3}}{\epsilon_{1}^{3}}}{\frac{0.0248k_{2}}{\epsilon_{1}}+2}> 0$
(20)
$\frac{0.0248k_{3}}{\epsilon_{1}^{3}}> 0$

์ด์— ๋งŒ์กฑํ•˜๋Š” $k_{1}$, $k_{2}$, $k_{3}$์˜ ์กฐํ•ฉ์„ ์‘๋‹ต์†๋„๋‚˜ ์•ˆ์ • ์—ฌ์œ (๋งˆ์ง„)๊ด€์ ์œผ๋กœ ๋ณผ ๋•Œ $s^{1}$ํ•ญ์˜ ๊ฐ’์„ ํฌ๊ฒŒ ํ• ์ˆ˜๋ก ๋” ์„ฑ๋Šฅ์ด ์ข‹๋‹ค. ์ด์œ ๋Š” $s^{1}$ํ•ญ์˜ ๊ฐ’์ด ํฌ๋ฉด ๊ทน์ ์ด ์™ผ์ชฝ์— ๋ฉ€์ฐ์ด ์œ„์น˜ํ•˜๋ฏ€๋กœ ์•ˆ์ • ์—ฌ์œ ๊ฐ€ ์ปค์ง€๋ฉด์„œ ๋” ๋น ๋ฅด๊ณ  ์•ˆ์ •์ ์ด๋‹ค.

3.2 ์™ธ๋ถ€ ํ๋ฃจํ”„ ์ œ์–ด๊ธฐ ์„ค๊ณ„ ๋ฐ ์•ˆ์ •์„ฑ ๋ถ„์„

3.2.a $\epsilon_{2}$-PD์ œ์–ด๊ธฐ ์„ค๊ณ„

$C_{2}(s)$๋Š” $x_{1}$์ด $x_{1d}$๋กœ ์ถ”์ข… ๋˜๋„๋ก ์„ค๊ณ„๋  ์ œ์–ด๊ธฐ์ด๋ฉฐ, $C_{2}(s)$์˜ ์ถœ๋ ฅ์€ $u_{1}$์ด๋‹ค. 2์žฅ์—์„œ ๊ฐ€์ •์„ ํ†ตํ•ด ๋ถ„๋ฆฌ๊ตฌ์กฐ๋ฅผ ์„ค์ •ํ•˜์˜€๊ธฐ์—, ์™ธ๋ถ€ ํ๋ฃจํ”„ ์„ค๊ณ„์— ์žˆ์–ด์„œ ๋‚ด๋ถ€ ํ๋ฃจํ”„๋Š” ์ œ์™ธํ•œ๋‹ค.

์ถ”์ข…์˜ค์ฐจ๋Š” $e_{1}=x_{1}-x_{1d}$, $e_{2}=x_{2}-\dot{x}_{1d}$๋กœ ์ •์˜ํ•˜๋ฉฐ $x_{1d}$๋Š” ๋ ˆํผ๋Ÿฐ์Šค์ด๋ฏ€๋กœ $\dot{x}_{1d}$๋Š” 0์ด๋‹ค. ์‹ (5)๋ฅผ ์ •๋ฆฌํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(21)
$\dot{e_{1}}=e_{2}$
(22)
$\dot{e_{2}}=K_{bb}u_{1}$

์‹ (5)๋ฅผ ํ–‰๋ ฌํ˜•ํƒœ $z=[\dot{e}_{1}, \dot{e}_{2}]^{T}$ํ˜•ํƒœ๋กœ ์ •๋ฆฌํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(23)
$\dot{z}=\begin{bmatrix}\dot{e}_{2}\\K_{bb}\dot{u}_{1}\end{bmatrix}=\begin{bmatrix}z_{2}\\K_{bb}\dot{u}_{1}\end{bmatrix}$

์ƒํƒœ๋ฐฉ์ •์‹์€ $\dot{z}= Az+B[K_{bb}\dot{u}_{1}]$์œผ๋กœ ํ‘œํ˜„ํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ ์ด ๋•Œ $\dot{u}_{1}=v$์ด๊ณ , A์™€ B๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(24)
$\dot{z}=\begin{bmatrix}z_{2}\\K_{bb}v\end{bmatrix}=\begin{bmatrix}0&1\\0&0\end{bmatrix}z +\begin{bmatrix}0\\1\end{bmatrix}[K_{bb}v]$

ํ”ผ๋“œ๋ฐฑ ์„ ํ˜•ํ™” ์ œ์–ด๊ธฐ๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(25)
$v=\frac{\omega}{K_{bb}}$

์ด ๋•Œ ๋‚ด๋ถ€ ์ œ์–ด๊ธฐ $\omega$๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์ด ํ‘œํ˜„ํ•  ์ˆ˜ ์žˆ๋‹ค.

(26)
$\omega =\frac{k_{4}}{\epsilon_{2}^{2}}z_{1}+\frac{k_{5}}{\epsilon_{2}}z_{2}$

์‹ (26)์„ ์‹ (25)์— ๋Œ€์ž…ํ•˜์—ฌ $v =\dot{u}_{1}$์— ๋Œ€ํ•œ ์‹์œผ๋กœ ์ •๋ฆฌํ•˜๋ฉด $\epsilon_{2}$-PD์ œ์–ด๊ธฐ๋ฅผ ์„ค๊ณ„ํ•  ์ˆ˜ ์žˆ๋‹ค.

(27)
$v =\frac{k_{4}}{K_{bb}\epsilon_{2}^{2}}z_{1}+\frac{k_{5}}{K_{bb}\epsilon_{2}}z_{2}$
(28)
$u_{1}(t)=c_{2}(t)= K_{P 2}(\epsilon_{2})e_{1}+K_{D 2}(\epsilon_{2})\dot{e}_{1}$

์ด ๋•Œ $K_{P 2}(\epsilon_{2}), K_{D 2}(\epsilon_{2})$๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(29)
$K_{P 2}(\epsilon_{2})=\frac{k_{4}}{K_{bb}\epsilon_{2}^{2}}, K_{D 2}=\frac{k_{5}}{K_{bb}\epsilon_{2}}$

3.2.b ์™ธ๋ถ€ ํ๋ฃจํ”„์˜ ์•ˆ์ •์„ฑ ํŒ๋ณ„

BB01์˜ ๋™์—ญํ•™๋ฐฉ์ •์‹๊ณผ $\epsilon_{2}$-PD์ œ์–ด๊ธฐ ์‹์˜ ๋ผํ”Œ๋ผ์Šค๋ณ€ํ™˜์„ ํ†ตํ•ด ๊ตฌํ•œ ๊ฐ๊ฐ์˜ ์ „๋‹ฌํ•จ์ˆ˜ $G_{2}(s)$, $C_{2}(s)$๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(30)
$G_{2}(s)=\frac{K_{bb}}{s^{2}}$
(31)
$C_{2}(s)= K_{P 2}(\epsilon_{2})+K_{D 2}(\epsilon_{2})s$

ํŠน์„ฑ๋ฐฉ์ •์‹ $1+ G_{2}(s)C_{2}(s)=0$๊ตฌํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(32)
$\epsilon_{2}^{2}s^{2}+\epsilon_{2}\cdot k_{5}s+k_{4}=0$

์‹ (32)์„ ํ†ตํ•ด Routh-HurwitzํŒ๋ณ„๋ฒ•์„ ์ด์šฉํ•œ๋‹ค.

$\begin{matrix}s^{2}\\s^{1}\\s^{0}\end{matrix}\left |\begin{matrix}\epsilon_{2}^{2}& k_{4}\\\epsilon_{2}\cdot k_{5}& 0\\k_{4}& 0\end{matrix}\right .$

์ œ์–ด๊ธฐ๊ฐ€ ์•ˆ์ •ํ•œ $k_{4}$,$k_{5}$,$\epsilon_{2}$์˜ ๋ฒ”์œ„๋ฅผ ์•„๋ž˜์™€ ๊ฐ™์ด ๊ตฌํ•  ์ˆ˜ ์žˆ๋‹ค.

(33)
$\epsilon_{2}\cdot k_{5}>0$
(34)
$k_{4}>0$

Remark 1. ๋‹ค์Œ๊ณผ ๊ฐ™์ด 4์ฐจ ์‹œ์Šคํ…œ์„ 2์ฐจ ์‹œ์Šคํ…œ์œผ๋กœ ๋ถ„๋ฆฌํ•˜์—ฌ, ๊ฐ 2์ฐจ ๋ชจ๋ธ์—์„œ ๊ฐœ๋ณ„์ ์œผ๋กœ ์ œ์–ด๊ธฐ๋ฅผ ์„ค๊ณ„ํ•˜์˜€๋‹ค. ๊ทธ๋ฆฌ๊ณ  ๊ธฐ์กด PID ์ œ์–ด๊ธฐ์— ๋น„ํ•ด ๊ฐ ์ œ์–ด๊ธฐ์—์„œ ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$, $\epsilon_{2}$๋ฅผ ์ถ”๊ฐ€ํ•˜์—ฌ ๋‹จ์ผ ํŒŒ๋ผ๋ฏธํ„ฐ๋กœ PID์˜ ์ „์ฒด ์ด๋“ ํŠœ๋‹์„ ๋ฏธ์„ธํ•˜๊ฒŒ ์กฐ์ ˆํ•˜๊ณ  ์‰ฝ๊ฒŒ ์„ฑ๋Šฅ์„ ์žฌ์กฐ์ •ํ•  ์ˆ˜ ์žˆ๋„๋ก ํ•œ๋‹ค. ์ด์— ๋Œ€ํ•œ ์ด๋“ ํŠœ๋‹ ๊ณผ์ •์˜ ์ฐจ๋ณ„์ ์—์„œ ์œ ์šฉ์„ฑ์„ ๋ถ„์„ํ•ด๋ณด๋„๋ก ํ•˜์ž.

4. ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$, $\epsilon_{2}$์˜ ์œ ์šฉ์„ฑ ๋ถ„์„ ๋ฐ ์ด๋“ ํŠœ๋‹๋ฐฉ์•ˆ

3์žฅ์—์„œ ์„ค๊ณ„ํ•œ ์ด์ค‘ $\epsilon$-PID/PD ์ œ์–ด๊ธฐ์„ ์‹œ๋ฎฌ๋ ˆ์ด์…˜ํ•˜์—ฌ, ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$, $\epsilon_{2}$์ด ์˜ค๋ฒ„์ŠˆํŠธ, ์ •์ฐฉ์‹œ๊ฐ„๊ณผ ์ •์ƒ์ƒํƒœ์˜ค์ฐจ์— ์–ด๋–ค ์˜ํ–ฅ์„ ๋ฏธ์น˜๋Š”์ง€ ๋ถ„์„ํ•œ๋‹ค.

4.1 ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$์˜ ์œ ์šฉ์„ฑ ๋ถ„์„

4.1.a $\epsilon_{1}$๋ณ€ํ™”์— ๋”ฐ๋ฅธ ๋‚ด๋ถ€ ํ๋ฃจํ”„ ์ถœ๋ ฅ($\theta$)์˜ ์˜ค๋ฒ„์ŠˆํŠธ์™€ ์ •์ฐฉ์‹œ๊ฐ„ ๋ถ„์„

์™ธ๋ž€ $d_{1}$์€ 0์œผ๋กœ ๊ฐ„์ฃผํ•˜๋ฉฐ ์Šคํ…์‘๋‹ต์—์„œ ๋‚˜์˜ค๋Š” ์˜ค๋ฒ„์ŠˆํŠธ์™€ ์ •์ฐฉ์‹œ๊ฐ„์„ ๋ถ„์„ํ•ด๋ณด์ž. 3์žฅ์˜ ๋‚ด๋ถ€ ํ๋ฃจํ”„์˜ ํŠน์„ฑ๋ฐฉ์ •์‹์ธ ์‹ (18)์˜ $s^{3}$์˜ ๊ณ„์ˆ˜๋ฅผ 1๋กœ ์ •๋ฆฌํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(35)
$s^{3}+\left(\frac{k_{2}}{\epsilon_{1}}+\frac{2}{\tau}\right)s^{2}+\frac{k_{1}}{\epsilon_{1}^{2}}s+\frac{k_{3}}{\epsilon_{1}^{3}}=0$

์‹ (33)์„ 2๊ฐœ์˜ ์šฐ์„ธ๊ทน์ ๊ณผ 1๊ฐœ์˜ ์‹ค์ˆ˜๊ทน์ ($\alpha$)์„ ๊ฐ€์ง€๋Š” ์‹์œผ๋กœ ์ •๋ฆฌํ•œ๋‹ค.

(36)
$(s+\alpha)(s^{2}+2\zeta\omega_{n}s+\omega_{n}^{2}) =s^{3}+(2\zeta\omega_{n}+\alpha)s^{2}+(2\zeta\omega_{n}\alpha +\omega_{n}^{2})s+\omega_{n}^{2}\alpha$

์ด์ œ ๊ฐ $s^{2}$,$s^{1}$,$s^{0}$ํ•ญ์˜ ๊ณ„์ˆ˜๋ฅผ ๋น„๊ตํ•œ๋‹ค.

(37)
$\frac{k_{2}}{\epsilon_{1}}+\frac{2}{\tau}=2\zeta\omega_{n}+\alpha$
(38)
$\frac{k_{1}}{\epsilon_{1}^{2}}=2\zeta\omega_{n}\alpha +\omega_{n}^{2}$
(39)
$\frac{k_{3}}{\epsilon_{1}^{3}}=\omega_{n}^{2}\alpha$

์ด ๋•Œ, ์‹ค์ˆ˜ ๊ทน์ ์ด $\alpha\gg \zeta\omega_{n}$๋ฅผ ๋งŒ์กฑํ•œ๋‹ค๊ณ  ๊ฐ€์ •ํ•˜๋ฉด ์‹ (37), (38), (39)๋ฅผ ํ†ตํ•ด $\alpha , \zeta , \omega_{n}$์„ ์ •์˜ํ•  ์ˆ˜ ์žˆ๋‹ค.

(40)
$\alpha =\frac{k_{2}}{\epsilon_{1}}+80.64516129$
(41)
$\omega_{n}=\sqrt{\frac{k_{3}}{\alpha\epsilon_{1}^{3}}}$
(42)
$\zeta =\frac{\frac{k_{1}}{\epsilon_{1}^{2}}-\omega_{n}^{2}}{2\omega_{n}\alpha}$

์‹ (40), (41)๋ฅผ ์‹ (42)์— ๋Œ€์ž…ํ•œ ๋’ค ์˜ค๋ฒ„์ŠˆํŠธ, ์ •์ฐฉ์‹œ๊ฐ„์—์„œ $\epsilon_{1}$์˜ ์˜ํ–ฅ์„ ๋ถ„์„ํ•œ๋‹ค.

(43)
$\zeta =\frac{k_{1}-\frac{k_{3}}{k_{2}+\frac{2}{\tau}\epsilon_{1}}}{2\sqrt{\left(k_{2}+\frac{2}{\tau}\epsilon_{1}\right)k_{3}}}=\frac{\left(k_{2}+\frac{2}{\tau}\epsilon_{1}\right)k_{1}-k_{3}}{2\sqrt{\left(k_{2}+\frac{2}{\tau}\epsilon_{1}\right)^{3}k_{3}}}$
(44)
$Mp=100e^{\frac{-\pi\zeta}{\sqrt{1-\zeta^{2}}}}[\%]$
(45)
$T_{s}(2\%)=\frac{4.6}{\zeta\omega_{n}} [sec]$

์‹ (43), (44)์—์„œ ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$์˜ 0.4๋ถ€ํ„ฐ 1.4๊นŒ์ง€์˜ ๋ณ€ํ™”์— ๋”ฐ๋ผ ๋ชจํ„ฐ ๊ฐ๋„($\theta$)์˜ ์˜ค๋ฒ„์ŠˆํŠธ๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์ด ๋‚˜ํƒ€๋‚œ๋‹ค.

๊ทธ๋ฆผ 2. $\epsilon_{1}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ๋ชจํ„ฐ ๊ฐ๋„($\theta$)์˜ ์˜ค๋ฒ„์ŠˆํŠธ

Fig. 2. Overshoot of $\theta$ vs $\epsilon_{1}$

../../Resources/kiee/KIEE.2025.74.12.2310/fig2.png

Fig. 2๋ฅผ ํ†ตํ•ด ์ด๋“์กฐ์ ˆ์š”์†Œ๊ฐ€ ๊ฐ์†Œํ• ์ˆ˜๋ก ๋ชจํ„ฐ ๊ฐ๋„($\theta$)์˜ ์˜ค๋ฒ„์ŠˆํŠธ๊ฐ€ ๊ฐ์†Œํ•˜๋Š” ๊ฒƒ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค. ๋˜ํ•œ, ์‹ (45)์—์„œ ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$์˜ 0.4๋ถ€ํ„ฐ 1.4๊นŒ์ง€์˜ ๋ณ€ํ™”์— ๋”ฐ๋ผ ๋ชจํ„ฐ ๊ฐ๋„($\theta$)์˜ ์ •์ฐฉ์‹œ๊ฐ„์€ ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

๊ทธ๋ฆผ 3. $\epsilon_{1}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ๋ชจํ„ฐ ๊ฐ๋„($\theta$)์˜ ์ •์ฐฉ์‹œ๊ฐ„

Fig. 3. Settling time of $\theta$ vs $\epsilon_{1}$

../../Resources/kiee/KIEE.2025.74.12.2310/fig3.png

Fig. 3๋ฅผ ํ†ตํ•ด ์ด๋“์กฐ์ ˆ์š”์†Œ๊ฐ€ ๊ฐ์†Œํ• ์ˆ˜๋ก ๋ชจํ„ฐ ๊ฐ๋„($\theta$)์˜ ์ •์ฐฉ์‹œ๊ฐ„์ด ๊ฐ์†Œํ•˜๋Š” ๊ฒƒ์„ ํ™•์ธํ•œ๋‹ค.

4.1.b $\epsilon_{1}$๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์™ธ๋ถ€ ํ๋ฃจํ”„ ์ถœ๋ ฅ$(x)$์˜ ์˜ค๋ฒ„์ŠˆํŠธ์™€ ์ •์ฐฉ์‹œ๊ฐ„ ๋ถ„์„

3์žฅ์—์„œ ์„ค๊ณ„ํ•œ ์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๊ธฐ๋ฅผ MATLAB๋ฅผ ํ†ตํ•ด ๊ตฌํ˜„ํ•˜์—ฌ ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ๋ณผ-๋น” ์‹œ์Šคํ…œ์˜ ์ถœ๋ ฅ์ธ ์‡ ๊ณต์˜ ์œ„์น˜($x$)์˜ ์„ฑ๋Šฅ์— ๋Œ€ํ•ด ๋ถ„์„ํ•œ๋‹ค. ์‹œ๋ฎฌ๋ ˆ์ด์…˜์„ ํ†ตํ•ด $\epsilon_{1}$๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์‡ ๊ณต์œ„์น˜($x$)์˜ ์˜ค๋ฒ„์ŠˆํŠธ๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

๊ทธ๋ฆผ 4. $\epsilon_{1}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์‡ ๊ณต์œ„์น˜($x$)์˜ ์˜ค๋ฒ„์ŠˆํŠธ

Fig. 4. Overshoot of $x$ vs $\epsilon_{1}$

../../Resources/kiee/KIEE.2025.74.12.2310/fig4.png

Fig. 4๋ฅผ ํ†ตํ•ด $\epsilon_{1}$์˜ ๊ฐ์†Œ๋Š” ์‡ ๊ณต์œ„์น˜($x$)์˜ ์˜ค๋ฒ„์ŠˆํŠธ๋ฅผ ๊ฐ์†Œ์‹œ์ผœ์ค€๋‹ค๋Š” ๊ฒƒ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.

๋‹ค์Œ์€ $\epsilon_{1}$๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์‡ ๊ณต์œ„์น˜($x$)์˜ ์ •์ฐฉ์‹œ๊ฐ„์„ ํ™•์ธํ•œ๋‹ค.

๊ทธ๋ฆผ 5. $\epsilon_{1}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์‡ ๊ณต์œ„์น˜($x$)์˜ ์ •์ฐฉ์‹œ๊ฐ„

Fig. 5. Settling time of $x$ vs $\epsilon_{1}$

../../Resources/kiee/KIEE.2025.74.12.2310/fig5.png

์ด๋ฒˆ Fig. 5๋ฅผ ํ†ตํ•ด์„œ๋„ $\epsilon_{1}$์˜ ๊ฐ์†Œ๋กœ ์‡ ๊ณต์œ„์น˜($x$)์˜ ์ •์ฐฉ์‹œ๊ฐ„์„ ๋‹จ์ถ•์‹œํ‚ค๋Š” ์„ฑ๋Šฅ๊ฐœ์„ ์ด ์ผ์–ด๋‚˜๊ณ ์žˆ์Œ์„ ํ™•์ธ ํ•  ์ˆ˜ ์žˆ๋‹ค.

2, 3์žฅ์—์„œ SRV02์™€ BB01์„ ๋ถ„๋ฆฌํ•˜์—ฌ ์ƒํƒœ๋ฐฉ์ •์‹์„ ํ‘œํ˜„ํ–ˆ์ง€๋งŒ, SRV02์˜ ์ถœ๋ ฅ์ธ $\theta$๋Š” BB01์˜ ์ž…๋ ฅ์œผ๋กœ์„œ์˜ ์—ญํ• ์„ ํ•œ๋‹ค. ๊ทธ๋Ÿฌ๋ฏ€๋กœ Fig. 2, Fig. 4์™€ Fig. 3, Fig. 5์—์„œ $\epsilon_{1}$์˜ ๊ฐ์†Œ๋Š” ๋‚ด๋ถ€ ํ๋ฃจํ”„ ์ถœ๋ ฅ(๋ชจํ„ฐ ๊ฐ๋„)๊ณผ ์™ธ๋ถ€ ํ๋ฃจํ”„ ์ถœ๋ ฅ(์‡ ๊ณต ์œ„์น˜)์˜ ์„ฑ๋Šฅ์„ ๋™์‹œ์— ํ–ฅ์ƒ ์‹œ์ผœ์ฃผ๋Š” ๊ฒฝํ–ฅ์„ฑ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.

4.1.c $\epsilon_{1}$๋ณ€ํ™”์— ๋”ฐ๋ฅธ ๋‚ด๋ถ€ ํ๋ฃจํ”„ ์ถœ๋ ฅ($\theta$)์˜ ์ •์ƒ์ƒํƒœ์˜ค์ฐจ ๋ถ„์„

๋‚ด๋ถ€ ํ๋ฃจํ”„์˜ ์ถœ๋ ฅ์ธ ๋ชจํ„ฐ ๊ฐ๋„($\theta$)์˜ ramp์— ๋Œ€ํ•œ ์˜ค์ฐจ์‹ ํ˜ธ $E(s)$๋Š” ์•„๋ž˜์™€ ๊ฐ™์ด ๊ณ„์‚ฐ๋œ๋‹ค.

(46)
$E(s)=R(s)-Y(s)\\ =\left[1-\frac{C_{1}(s)G_{1}(s)}{1+C_{1}(s)G_{1}(s)}\right]R(s)-\frac{d_{1}G_{1}(s)}{s^{2}[1+C_{1}(s)G_{1}(s)]}\\ =\left[\frac{s^{3}+\frac{1}{\tau}s^{2}}{s^{3}+\left(\frac{k_{2}}{\epsilon_{1}}+\frac{2}{\tau}\right)\\s^{2}+\frac{k_{1}}{\epsilon_{1}^{2}}s+\frac{k_{3}}{\epsilon_{1}^{3}}}\right]R(s) \\ -\frac{d_{1}K}{\tau s^{4}+\left(\frac{\tau k_{2}}{\epsilon_{1}}+2\right)s^{3}+\frac{\tau k_{1}}{\epsilon_{1}^{2}}s^{2}+\frac{\tau k_{3}}{\epsilon_{1}^{3}}s}$

์—ฌ๊ธฐ์„œ ์™ธ๋ž€ $d_{1}$์— ๋Œ€ํ•œ ๋ถ„์„์€ ์ œ์–ด๊ธฐ์˜ ๊ฐ•์ธ์„ฑ์„ ๋ถ„์„ํ•˜๊ธฐ ์œ„ํ•จ์ด๋ฉฐ, ์‹ (46)์„ ํ†ตํ•ด ์ •์ƒ์ƒํƒœ์˜ค์ฐจ $e_{ss}$๋ฅผ ์ •๋ฆฌํ•œ๋‹ค.

(47)
$e_{ss}=\lim_{s \to 0}s E(s) =\lim_{s \to 0}\left[\frac{s^{4}+\frac{1}{\tau}s^{3}}{s^{3}+\left(\frac{k_{2}}{\epsilon_{1}}+\frac{2}{\tau}\right)s^{2}+\frac{k_{1}}{\epsilon_{1}^{2}}s+\frac{k_{3}}{\epsilon_{1}^{3}}}\right]R(s) -\frac{d_{1}K}{\tau s^{3}+\left(\frac{\tau k_{2}}{\epsilon_{1}}+2\right)s^{2}+\frac{\tau k_{1}}{\epsilon_{1}^{2}}s+\frac{\tau k_{3}}{\epsilon_{1}^{3}}} =\frac{d_{1}K\epsilon_{1}^{3}}{\tau k_{3}}$

์‹ (47)๋ฅผ ํ†ตํ•ด ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$์˜ 0.4๋ถ€ํ„ฐ 1.4๊นŒ์ง€์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์ •์ƒ์ƒํƒœ์˜ค์ฐจ๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

๊ทธ๋ฆผ 6. $\epsilon_{1}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ๋ชจํ„ฐ ๊ฐ๋„($\theta$)์˜ ์ •์ƒ์ƒํƒœ์˜ค์ฐจ

Fig. 6. Steady-state error of $\theta$ vs $\epsilon_{1}$

../../Resources/kiee/KIEE.2025.74.12.2310/fig6.png

Fig. 6๋ฅผ ํ†ตํ•ด ์ด๋“์กฐ์ ˆ์š”์†Œ๊ฐ€ ๊ฐ์†Œํ• ์ˆ˜๋ก ๋ชจํ„ฐ ๊ฐ๋„($\theta$)์˜ ์ •์ƒ์ƒํƒœ์˜ค์ฐจ๊ฐ€ ๊ฐ์†Œํ•˜๋Š” ๊ฒƒ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.

4.2 ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{2}$์˜ ์œ ์šฉ์„ฑ ๋ถ„์„

4.2.a $\epsilon_{2}$๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์™ธ๋ถ€ ํ๋ฃจํ”„ ์ถœ๋ ฅ$(x)$์˜ ์˜ค๋ฒ„์ŠˆํŠธ์™€ ์ •์ฐฉ์‹œ๊ฐ„ ๋ถ„์„

์‹ค์ œ ์‹œ์Šคํ…œ์—์„œ ๋‚ด๋ถ€ ํ๋ฃจํ”„์˜ ๋ถˆ์™„์ „ํ•œ ๋ถ„๋ฆฌ ๋ฐ ์ƒํ˜ธ์ž‘์šฉ์ด ์™ธ๋ถ€ ํ๋ฃจํ”„ ๋™ํŠน์„ฑ์— ์œ ์˜๋ฏธํ•œ ์˜ํ–ฅ์„ ๋ฏธ์น  ์ˆ˜ ์žˆ๋‹ค. ์ด๋ฅผ ๊ณ ๋ คํ•œ ์ข€ ๋” ์ •ํ™•ํ•œ ์‹œ์Šคํ…œ ๋ถ„์„์„ ์œ„ํ•˜์—ฌ ์•ž์—์„œ ์ œ์‹œํ•œ ๋‹จ์ˆœํ™”๋œ ์‹œ์Šคํ…œ์ด ์•„๋‹Œ ์‹ (3)์—์„œ์˜ ์‹ค์ œ ์‹์€ ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(48)
$\dot{x_{2}}= K_{bb}\sin x_{3}+K_{bb}\sin u_{1}-K_{bb}\sin u_{1} = K_{bb}\sin u_{1}+K_{bb}(\sin x_{3}-\sin u_{1})$

์ด ๋•Œ $(\sin x_{3}-\sin u_{1})$๋Š” ๋ถ„๋ฆฌ๊ตฌ์กฐ์—์„œ ๊ฐ 2์ฐจ ์‹œ์Šคํ…œ์— ๋ฐœ์ƒํ•˜๋Š” crossover term์ด๊ณ  ์ด๋ฅผ $w(t)$๋กœ ์ •์˜ํ•œ๋‹ค. $w(t)$๋Š” ์‹ (3)์—์„œ ํŒŒ์ƒ๋˜๊ธฐ ๋•Œ๋ฌธ์— $\epsilon_{2}$์— ๋Œ€ํ•œ ํ•จ์ˆ˜์ด๋ฉฐ, $x_{3}$๊ฐ€ $u_{1}$์„ ์ถ”์ข…ํ•˜๋Š” ์ƒํ™ฉ์—์„œ๋Š” 0์ด๋˜๋Š” ์„ฑ์งˆ์ด๋‹ค($\triangle w(t)\to 0$). ์ด๋ฅผ ์™ธ๋ถ€ํ๋ฃจํ”„ ์ถœ๋ ฅ์˜ ๋ถ„์„์— ๋ฐ˜์˜ํ•˜๊ธฐ ์œ„ํ•ด ๋‚ด๋ถ€ ํ๋ฃจํ”„์— ์˜ํ•œ crossover term์„ ํŒŒ๋ผ๋ฏธํ„ฐ ํ•จ์ˆ˜์ธ $T_{cross}(\epsilon_{2})$๋กœ ์ •์˜ํ•˜์ž. ์™ธ๋ถ€ ํ๋ฃจํ”„์˜ ํŠน์„ฑ๋ฐฉ์ •์‹์—์„œ ์˜ค๋ฒ„์ŠˆํŠธ์™€ ์ •์ฐฉ์‹œ๊ฐ„์„ ๋ถ„์„ํ•˜๋Š”๋ฐ ๊ฐ€์žฅ ์ง๊ด€์ ์ธ ์˜ํ–ฅ์„ ์ค„ ์ˆ˜ ์žˆ๋Š” ๋Œํ•‘ ๊ณ„์ˆ˜์ธ $s^{1}$์— $T_{cross}(\epsilon_{2})$๋ฅผ ์ถ”๊ฐ€ํ•˜์—ฌ $s^{2}$์˜ ๊ณ„์ˆ˜๋ฅผ 1๋กœ ์ •๋ฆฌํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(49)
$s^{2}+\frac{k_{5}+T_{cross}(\epsilon_{2})}{\epsilon_{2}}s+\frac{k_{4}}{\epsilon_{2}^{2}}=0$

์‹ (49)๋ฅผ 2๊ฐœ์˜ ์šฐ์„ธ๊ทน์ ์„ ๊ฐ€์ง€๋Š” ์‹์œผ๋กœ ์ •๋ฆฌํ•œ๋‹ค.

(50)
$s^{2}+2\zeta\omega_{n}s+\omega_{n}^{2}=0$

์ด์ œ ๊ฐ $s^{1}$,$s^{0}$ํ•ญ์˜ ๊ณ„์ˆ˜๋ฅผ ๋น„๊ตํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(51)
$\frac{k_{5}+T_{cross}(\epsilon_{2})}{\epsilon_{2}}=2\zeta\omega_{n}$
(52)
$\frac{k_{4}}{\epsilon_{2}^{2}}=\omega_{n}^{2}$

์‹ (51), (52)๋ฅผ ํ†ตํ•ด $\zeta$, $\omega_{n}$์„ ์ •์˜ํ•  ์ˆ˜ ์žˆ๋‹ค.

(53)
$\omega_{n}=\frac{\sqrt{k_{4}}}{\epsilon_{2}}$
(54)
$\zeta =\frac{k_{5}+T_{cross}(\epsilon_{2})}{2\sqrt{k_{4}}}$

์‹ (54)์—์„œ ๊ฐ์‡ ๋น„๋Š” $\epsilon_{2}$์™€ ์ง์ ‘ ๋น„๋ก€ํ•˜์ง€ ์•Š์œผ๋ฉฐ, ๊ต์ฐจํ•ญ $T_{cross}(\epsilon_{2})$์˜ ํ•จ์ˆ˜ ๊ตฌ์กฐ์— ๋”ฐ๋ผ $\epsilon_{2}$์— ๋Œ€ํ•œ ๊ฐ„์ ‘์  ์˜์กด์„ฑ์„ ๊ฐ€์ง„๋‹ค. ๋”ฐ๋ผ์„œ $\epsilon_{2}$๊ฐ€ ์ฆ๊ฐ€ํ•˜๋ฉด $T_{cross}(\epsilon_{2})$๊ฐ€ ์–ด๋–ป๊ฒŒ ์ •์˜๋˜๋Š”์ง€์— ๋”ฐ๋ผ ๊ฐ์‡ ๋น„ ์—ญ์‹œ ๊ทธ ํšจ๊ณผ์— ๋งž์ถฐ ๋ณ€ํ™”ํ•œ๋‹ค. ์šฐ์„  ์ง์ ‘์ ์ธ ์˜ํ–ฅ์„ ์ฃผ์ง€ ์•Š์œผ๋ฏ€๋กœ $\epsilon_{2}$์˜ ๋ณ€ํ™”๋Š” ์™ธ๋ถ€ ํ๋ฃจํ”„ ์ถœ๋ ฅ$(x)$์˜ ์˜ค๋ฒ„์ŠˆํŠธ์— ์˜ํ–ฅ์„ ์ฃผ์ง€ ์•Š๋Š”๋‹ค๊ณ  ์ •๋ฆฌํ•˜๊ณ  ์‹ค์ œ ์‹คํ—˜์—์„œ $T_{cross}(\epsilon_{2})$๊ฐ€ ์–ด๋–ค ์˜ํ–ฅ์„ ๋ฏธ์น˜๋Š”์ง€ ์‚ดํŽด๋ณด๋„๋กํ•˜์ž. ๋‹ค์Œ์œผ๋กœ ์‹ (51)์„ ์‹ (45)์— ๋Œ€์ž…ํ•˜๋ฉด ์ •์ฐฉ์‹œ๊ฐ„์€ ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(55)
$T_{s}(2\%)=\frac{9.2\epsilon_{2}}{k_{5}+T_{cross}(\epsilon_{2})}$

์‹ (55)์—์„œ ์ •์ฐฉ์‹œ๊ฐ„์€ ๋ถ„๋ชจ์— ์กด์žฌํ•˜๋Š” ๊ต์ฐจํ•ญ $T_{cross}(\epsilon_{2})$์˜ ํ•จ์ˆ˜ ๊ตฌ์กฐ์— ๋”ฐ๋ผ $\epsilon_{2}$์— ๋Œ€ํ•œ ๋‹จ์ˆœ ๋น„๋ก€๊ด€๊ณ„๊ฐ€ ๋‹ฌ๋ผ์งˆ ์ˆ˜ ์žˆ๋‹ค. ๋”ฐ๋ผ์„œ ์ •์ฐฉ์‹œ๊ฐ„์˜ $\epsilon_{2}$์˜ ์˜์กด์„ฑ์€ $T_{cross}(\epsilon_{2})$์˜ ์„ฑ์งˆ์— ์˜ํ•ด ๊ฒฐ์ •๋˜๋ฉฐ, ์‹ค์ œ ์‹คํ—˜์—์„œ $T_{cross}(\epsilon_{2})$๊ฐ€ ์–ด๋–ค ์˜ํ–ฅ์„ ๋ฏธ์น˜๋Š”์ง€ ์‚ดํŽด๋ณด๋„๋ก ํ•˜์ž.

4.2.b $\epsilon_{2}$๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์™ธ๋ถ€ ํ๋ฃจํ”„ ์ถœ๋ ฅ$(x)$์˜ ์ •์ƒ์ƒํƒœ ์˜ค์ฐจ ๋ถ„์„

์ด๋ฒˆ์—๋Š” ์™ธ๋ถ€ ํ๋ฃจํ”„์˜ ํŠน์„ฑ๋ฐฉ์ •์‹์—์„œ ์ถœ๋ ฅ์˜ ์ •์ƒ์ƒํƒœ์˜ค์ฐจ๋ฅผ ๋ถ„์„ํ•˜๋Š”๋ฐ ์ง๊ด€์ ์ธ ์˜ํ–ฅ์„ ์ค„ ์ˆ˜ ์žˆ๋Š” ์‹œ์Šคํ…œ ํƒ€์ž…๊ณผ ๊ด€๋ จ๋œ ๊ณ„์ˆ˜์ธ $s^{0}$์— $T_{cross}(\epsilon_{2})$๋ฅผ ์ถ”๊ฐ€ํ•œ๋‹ค. ์ถœ๋ ฅ์ธ ์‡ ๊ณต์œ„์น˜($x$)์˜ ์˜ค์ฐจ์‹ ํ˜ธ $E(s)$๋Š” ์•„๋ž˜์™€ ๊ฐ™์ด ๊ณ„์‚ฐ๋œ๋‹ค.

(56)
$E(s)=R(s)-Y(s)=[1-T_{2}(s)]R(s) =\left[\frac{\epsilon_{2}^{2}s^{2}}{\epsilon_{2}^{2}s^{2}+k_{5}\epsilon_{2}s+k_{4}+T_{cross}(\epsilon_{2})}\right]R(s)$

์—ฌ๊ธฐ์„œ $T_{2}$์€ ๋‚ด๋ถ€ ํ๋ฃจํ”„์˜ ์ „๋‹ฌํ•จ์ˆ˜์ด๋ฉฐ, ์‹(56)์„ ํ†ตํ•ด ์ •์ƒ์ƒํƒœ์˜ค์ฐจ $e_{ss}$๋ฅผ ์ •๋ฆฌํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(57)
$e_{ss}=\lim_{s \to 0}s E(s) =\lim_{s \to 0}\left[\frac{\epsilon_{2}^{2}s^{3}}{\epsilon_{2}^{2}s^{2}+k_{5}\epsilon_{2}s+k_{4}+T_{cross}(\epsilon_{2})}\right]R(s) =0$

์‹ (57)์„ ํ†ตํ•ด ์ถœ๋ ฅ($x$)์˜ ์ •์ƒ์ƒํƒœ์˜ค์ฐจ์— $\epsilon_{2}$๊ฐ€ ๊ด€์—ฌํ•˜์ง€ ์•Š๋Š” ๊ฒƒ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ์ •์ƒ์ƒํƒœ์˜ค์ฐจ ๋˜ํ•œ ์„œ๋ธŒ์‹œ์Šคํ…œ์˜ ๋ถ„๋ฆฌ๊ตฌ์กฐ ๊ฐ€์ •์— ์˜ํ•œ ์ •๋ฆฌ์ด๊ธฐ์— ์‹ค์ œ ์‹คํ—˜์—์„œ์˜ ๋ณ€์ˆ˜์ƒํ™ฉ์„ ์‚ดํŽด๋ณด๋„๋กํ•œ๋‹ค.

4.3 ์ด๋“ ์š”์†Œ์˜ ํŠœ๋‹ ๋ฐฉ์•ˆ

์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๊ธฐ์˜ ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$, $\epsilon_{2}$ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ํŠน์„ฑ์„ ์š”์•ฝํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

ํ‘œ 1. $\epsilon_{1}$, $\epsilon_{2}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์‘๋‹ตํŠน์„ฑ

Table 1. Responses characteristics variations vs $\epsilon_{1}$, $\epsilon_{2}$

$\epsilon_{1}$
decrease
$\theta$ $Mp$, $e_{ss}$ and $T_{s}$ decrease
$x$ $Mp$, $e_{ss}$ and $T_{s}$ decrease
$\epsilon_{1}$
increase
$\theta$ $Mp$, $e_{ss}$ and $T_{s}$ increase
$x$ $Mp$, $e_{ss}$ and $T_{s}$ increase
$\epsilon_{2}$
decrease
$x$ $Mp$, $e_{ss}$ consistent
$\epsilon_{2}$
increase
$x$ $Mp$, $e_{ss}$ consistent

Table 1๊ณผ ๊ฐ™์ด $\epsilon_{1}$์„ ๊ฐ์†Œ์‹œํ‚ค๋ฉด ๋‚ด๋ถ€ ํ๋ฃจํ”„ ์ „๋‹ฌํ•จ์ˆ˜์˜ ์ถœ๋ ฅ $\theta$์˜ ์‘๋‹ต ์„ฑ๋Šฅ์ด ๊ฐœ์„ ๋˜์–ด, ์ด๋กœ ์ธํ•ด ์™ธ๋ถ€ ํ๋ฃจํ”„๋ฅผ ํ†ตํ•œ ์ถœ๋ ฅ ๋ณ€์ˆ˜ $x$์˜ ์ œ์–ด ์„ฑ๋Šฅ๋„ ํ–ฅ์ƒ๋œ๋‹ค.

๊ทธ๋Ÿฌ๋‚˜ ์‹ (14), (28)๋ฅผ ํ†ตํ•ด $\epsilon_{1}$, $\epsilon_{2}$์˜ ๊ฐ์†Œ๋Š” $V_{m}$ ๋ฐ $\theta$์„ ์ฆ๊ฐ€์‹œํ‚ค๊ธฐ์—, ์ง€๋‚˜์น˜๊ฒŒ ์ž‘๊ฒŒ ์„ค์ •ํ•˜๋ฉด ์ œ์–ด ์ž…๋ ฅ$V_{m}$ ๋ฐ $\theta$๊ฐ€ ๋ชจ๋‘ ํฌํ™”๋˜์–ด ์ œ์–ด๊ธฐ๊ฐ€ ๋ฐ๋“œ๋ฐด๋“œ ๊ตฌ๊ฐ„์— ๋จธ๋ฌด๋ฅด๊ฒŒ ๋˜๊ณ , ์ด๋กœ ์ธํ•ด ์ œ์–ด ๋ฐ˜์‘์ด ์ •์ฒด๋  ๋ฟ๋งŒ ์•„๋‹ˆ๋ผ ์„ผ์„œ ๋…ธ์ด์ฆˆ๋กœ ์ธํ•œ ์ฑ„ํ„ฐ๋ง์ด ๋ฐœ์ƒํ•œ๋‹ค.

๋”ฐ๋ผ์„œ ๊ณผ๋„ํ•œ ํฌํ™”ํ˜„์ƒ์€ ์˜ค๋ฒ„์ŠˆํŠธ์™€ ์ •์ƒ์ƒํƒœ์˜ค์ฐจ์— ์˜ํ–ฅ์„ ์ฃผ์ง€์•Š๋Š” $\epsilon_{2}$-PD์ œ์–ด๊ธฐ์˜ ์ด๋“ $\epsilon_{2}$๋ฅผ ์ฆ๊ฐ€์‹œ์ผœ ๋ณด์ƒํ•˜๋Š” ๋ฐฉ์‹์œผ๋กœ ํ•ด๊ฒฐํ•˜๊ณ ์žํ•œ๋‹ค.

4์žฅ์—์„œ ๋ถ„์„ํ•œ ์ด๋“์กฐ์ ˆ์š”์†Œ์˜ ํŠน์ง•์„ ๋ฐ”ํƒ•์œผ๋กœ ์ด๋“ํŠœ๋‹ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ๋‹ค์Œ๊ณผ ๊ฐ™์ด ์ œ์•ˆํ•œ๋‹ค.

๊ทธ๋ฆผ 7. ์„ฑ๋Šฅ ๋ฐ ์•ˆ์ •์„ฑ ํ™•๋ณด๋ฅผ ์œ„ํ•œ $\epsilon_{1}$, $\epsilon_{2}$ ํ™œ์šฉ ๊ฒŒ์ธ ํŠœ๋‹ ์•Œ๊ณ ๋ฆฌ์ฆ˜

Fig. 7. Gain tuning algorithm using $\epsilon_{1}$, $\epsilon_{2}$ for performance and safety

../../Resources/kiee/KIEE.2025.74.12.2310/fig7.png

5. ๋ณผ-๋น” ์‹œ์Šคํ…œ ์ œ์–ด ์‹คํ—˜ ๊ฒฐ๊ณผ

์ด๋ฒˆ ์žฅ์—์„œ๋Š” 3์žฅ์—์„œ ๊ตฌํ•œ Routh-Hurwitz ์•ˆ์ •์กฐ๊ฑด ๋ฒ”์œ„์ธ ์‹ (19), (20)๊ณผ ์‹ (33), (34)์— ๋งŒ์กฑํ•˜๋Š” ์ด๋“์กฐ์ ˆ์š”์†Œ ๊ฐ’ $k_{1}=50$, $k_{2}=0.07$, $k_{3}=$ $120$, $k_{4}=$$0.05$, $k_{5}=10$์„ ์ด์šฉํ•˜๊ณ  ์•ˆ์ •์กฐ๊ฑด์— ๋Œ€์ž…ํ•จ์— ๋”ฐ๋ฅธ $\epsilon_{1}$, $\epsilon_{2}$์˜ ๋ฒ”์œ„๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(58)
$\epsilon_{1}>0.028892$
(59)
$\epsilon_{2}>0$

$k_{1}$, $k_{2}$, $k_{3}$, $k_{4}$, $k_{5}$ ๊ฐ’์„ ์„ ์ •ํ•œ ๊ธฐ์ค€์€ ๋ณธ ์‹คํ—˜์˜ ๋ชฉ์ ์ธ ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$, $\epsilon_{2}$์˜ ํšจ์šฉ์„ฑ์„ ์ž…์ฆํ•˜๊ธฐ ์œ„ํ•ด, $\epsilon_{1}$, $\epsilon_{2}$์˜ ๋ณ€ํ™” ๋ฒ”์œ„์—์„œ $V_{m}$ ๋ฐ $\theta$๊ฐ€ ๋ชจ๋‘ ํฌํ™”ํ˜„์ƒ์ด ์ผ์–ด๋‚˜์ง€ ์•Š๊ณ  ์ถœ๋ ฅ ์„ฑ๋Šฅ์˜ ๋ณ€ํ™”๋ฅผ ๋ช…ํ™•ํ•˜๊ฒŒ ๋ณด์—ฌ์ค„ ์ˆ˜ ์žˆ๋Š” ๊ฐ’์œผ๋กœ ์„ ์ •ํ•˜์˜€๋‹ค. ๋ ˆํผ๋Ÿฐ์Šค๋ฅผ ๋น”์˜ ์ค‘์•™์ธ $0[m]$๋กœ ์„ค์ •ํ•œ ๋’ค, ์‹ค์ œ ๋ณผ-๋น” ์‹œ์Šคํ…œ ์‹คํ—˜์„ ์ง„ํ–‰ํ•œ๋‹ค. ์ด ์‹คํ—˜์—์„œ ์„œ๋ณด๋ชจํ„ฐ์˜ ๊ฐ๋„ ์ œ์–ด๊ธฐ($C_{2}(s)$)์—๋Š” $\epsilon_{1}$-PID์ œ์–ด๊ธฐ๊ฐ€ ์‚ฌ์šฉ๋˜์—ˆ๊ณ , ์‡ ๊ณต์˜ ์œ„์น˜ ์ œ์–ด๊ธฐ($C_{1}(s)$)์—๋Š” $\epsilon_{2}$-PD์ œ์–ด๊ธฐ๊ฐ€ ์‚ฌ์šฉ๋˜์—ˆ๋‹ค. ์‚ฌ์šฉ๋œ ์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๊ธฐ์—์„œ ์„ ์ •ํ•œ ์ด๋“์กฐ์ ˆ์š”์†Œ $k_{1}, k_{2}, k_{3}, k_{4}, k_{5}$๊ฐ’์œผ๋กœ ์ œํ•œ๋˜๋Š” ํŠœ๋‹๋ฌธ์ œ๋ฅผ ์•ˆ์ •๋ฒ”์œ„ ์•ˆ์—์„œ $\epsilon_{1}$, $\epsilon_{2}$์˜ ๋ณ€ํ™”๋กœ ์ œ์–ด๊ธฐ ์„ฑ๋Šฅ์„ ํ–ฅ์ƒ์‹œ์ผœ ํ•ด๊ฒฐํ•˜๊ณ , ์ ์ ˆํ•œ ๊ฒŒ์ธ๊ฐ’๋“ค์„ ์ฐพ๋Š” ๊ณผ์ •์„ ์šฉ์ดํ•˜๊ฒŒ ๋งŒ๋“ ๋‹ค.

๊ทธ๋ฆผ 8. ๋ณผ-๋น” ์‹œ์Šคํ…œ์˜ ์‹คํ—˜ ๊ตฌ์„ฑ

Fig. 8. Experimental Setup of the Ball-Beam System

../../Resources/kiee/KIEE.2025.74.12.2310/fig8.png

5.1 $\epsilon_{1}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์‡ ๊ณต์˜ ์œ„์น˜ ๋ถ„์„

Fig. 7์—์„œ ์ œ์•ˆํ•œ ์ด๋“ํŠœ๋‹ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ๋”ฐ๋ผ $\epsilon_{1}$์˜ ๊ฐ์†Œ๊ฐ€ ๋ณผ-๋น” ์‹œ์Šคํ…œ ์ถœ๋ ฅ ์„ฑ๋Šฅ์„ ๊ฐœ์„ ํ•˜๋Š”์ง€ ์‹คํ—˜์„ ํ†ตํ•ด ํ™•์ธํ•œ๋‹ค.

์šฐ์„  $\epsilon_{1}=1, \epsilon_{2}=1$์ธ ์ƒํƒœ๋กœ ์‹คํ—˜ ์ธก์ •์„ ์‹œ์ž‘ํ•˜๋ฉฐ ์‡ ๊ณต์˜ ์œ„์น˜๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์ด ๋‚˜ํƒ€๋‚œ๋‹ค.

๊ทธ๋ฆผ 9. ($\epsilon_{1}=1, \epsilon_{2}=1$)์ผ ๋•Œ ์‡ ๊ณต์˜ ์œ„์น˜๊ถค์ 

Fig. 9. Ball position under when ($\epsilon_{1}=1, \epsilon_{2}=1$)

../../Resources/kiee/KIEE.2025.74.12.2310/fig9.png

Fig. 9์„ ํ†ตํ•ด ๊ธฐ์กด ์ด์ค‘ PID/PD์ œ์–ด๊ธฐ($\epsilon_{1}=1$,$\epsilon_{2}=1$)๋ฅผ ํ†ตํ•œ ๋ณผ-๋น”์ œ์–ด๋Š” ์˜ค๋ฒ„์ŠˆํŠธ์™€ ์ •์ƒ์ƒํƒœ ์˜ค์ฐจ๊ฐ€ ํฐ ๋ฌธ์ œ์ ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.

๊ทธ๋Ÿฌ๋ฏ€๋กœ Fig. 7์˜ ์ฒซ ๋ฒˆ์งธ ์กฐ๊ฑด์— ๋งŒ์กฑํ•˜์ง€ ์•Š๊ธฐ์— $\epsilon_{1}$๊ฐ’์„ 0.2์”ฉ ์ค„์ด๋ฉฐ $\epsilon_{1}=0.8$, $\epsilon_{1}=0.6$, $\epsilon_{1}=$$0.4$์ผ ๋•Œ ์‡ ๊ณต์˜ ์œ„์น˜๋ฅผ ํ™•์ธํ•œ๋‹ค.

๊ทธ๋ฆผ 10. ($\epsilon_{1}=0.8, \epsilon_{2}=1$)์ผ ๋•Œ ์‡ ๊ณต์˜ ์œ„์น˜๊ถค์ 

Fig. 10. Ball position when ($\epsilon_{1}=0.8, \epsilon_{2}=1$)

../../Resources/kiee/KIEE.2025.74.12.2310/fig10.png

๊ทธ๋ฆผ 11. ($\epsilon_{1}=0.6, \epsilon_{2}=1$)์ผ ๋•Œ ์‡ ๊ณต์˜ ์œ„์น˜๊ถค์ 

Fig. 11. Ball position when ($\epsilon_{1}=0.6, \epsilon_{2}=1$)

../../Resources/kiee/KIEE.2025.74.12.2310/fig11.png

๊ทธ๋ฆผ 12. ($\epsilon_{1}=0.4, \epsilon_{2}=1$)์ผ ๋•Œ ์‡ ๊ณต์˜ ์œ„์น˜๊ถค์ 

Fig. 12. Ball position when ($\epsilon_{1}=0.4, \epsilon_{2}=1$)

../../Resources/kiee/KIEE.2025.74.12.2310/fig12.png

์‹คํ—˜๊ฒฐ๊ณผ๋ฅผ ํ†ตํ•ด 4์žฅ์—์„œ ๋ถ„์„ํ•œ $\epsilon_{1}$๊ฐ’์ด ๊ฐ์†Œํ• ์ˆ˜๋ก ์˜ค๋ฒ„์ŠˆํŠธ์™€ ์ •์ƒ์ƒํƒœ์˜ค์ฐจ ์„ฑ๋Šฅ์ด ๊ฐœ์„ ๋˜๋Š” ๊ฒƒ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ ์ˆ˜์น˜๋ฅผ ํ‘œ๋กœ ์ •๋ฆฌํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

ํ‘œ 2. $\epsilon_{1}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์ถœ๋ ฅ ์„ฑ๋Šฅ

Table 2. Output responses vs $\epsilon_{1}$

Settling time(2%)
(sec)
Over shoot
(%)
$\left | e_{ss}\right |$
(m)
$\epsilon_{1}$=1 16.38 56.1 0.0128
$\epsilon_{1}$=0.8 7.75 17.03 0.0025
$\epsilon_{1}$=0.6 4.654 1.21 0.00032
$\epsilon_{1}$=0.4 4.972 0.17 0.00011

Table 2์— ์˜ํ•ด ์šฐ๋ฆฌ๋Š” ๊ฐ€์žฅ ์‡ ๊ณต์˜ ์œ„์น˜ ์„ฑ๋Šฅ์ด ํ–ฅ์ƒ๋œ $\epsilon_{1}=0.4$๋ฅผ ์‚ฌ์šฉํ•˜๋„๋ก ํ•˜์ž.

5.2 $\epsilon_{2}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์ž…๋ ฅ ์ „์•• ๋ถ„์„

๋ณผ-๋น” ์‹œ์Šคํ…œ ์ž…๋ ฅ์ „์••์˜ ํฌํ™”๋„๋Š” ํ•˜๋“œ์›จ์–ด์ ์œผ๋กœ โ€“10V์—์„œ 10V ๊ตฌ๊ฐ„์ด๋‹ค. 4์žฅ์˜ Fig. 7์—์„œ ์ œ์•ˆํ•œ ํŠœ๋‹๋ฐฉ์•ˆ๊ณผ ๊ฐ™์ด $\epsilon_{1}$์˜ ๊ฐ์†Œ์— ๋”ฐ๋ฅธ ์ž…๋ ฅ์ „์••์˜ ํฌํ™”ํ˜„์ƒ์„ $\epsilon_{2}$๋ฅผ ์ฆ๊ฐ€์‹œํ‚ค๋ฏ€๋กœ์จ ๋ณด์ƒํ•ด ์ค„ ์ˆ˜ ์žˆ๋Š”์ง€ ํ™•์ธํ•œ๋‹ค. Fig. 12์˜ ์ƒํ™ฉ์—์„œ $\epsilon_{2}$์˜ ์ฆ๊ฐ€์— ๋”ฐ๋ฅธ ์ž…๋ ฅ์ „์••์˜ ๋ณ€ํ™”๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

๊ทธ๋ฆผ 13. $\epsilon_{2}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์ž…๋ ฅ์ „์••

Fig. 13. $V_{m}$ vs $\epsilon_{2}$

../../Resources/kiee/KIEE.2025.74.12.2310/fig13.png

Fig. 13๋ฅผ ํ†ตํ•ด $\epsilon_{2}=1$์ผ ๋•Œ, 0์ดˆ์—์„œ 5์ดˆ ๊ตฌ๊ฐ„์— ํฌํ™”ํ˜„์ƒ์ด ์ผ์–ด๋‚˜๋Š” ๊ฒƒ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ๋˜ํ•œ ๋…ธ์ด์ฆˆ ํ˜„์ƒ์œผ๋กœ์ธํ•œ ์ž…๋ ฅ์ „์••์˜ ์ŠคํŒŒ์ดํฌ๊ฐ€ ์ผ์–ด๋‚˜ ์ž…๋ ฅ์„ ํํŠธ๋ŸฌํŠธ๋ฆฌ๋Š” ํ˜„์ƒ์ด ์ผ์–ด๋‚œ๋‹ค. $\epsilon_{2}$์˜ ์ฆ๊ฐ€๋Š” ์ž…๋ ฅ์ „์••์„ ๋‚ฎ์ถฐ ํฌํ™”ํ˜„์ƒ์„ ์™„ํ™”ํ•ด์ฃผ๋ฉฐ ์ž…๋ ฅ์ „์••์˜ ์ŠคํŒŒ์ดํฌ ํ˜„์ƒ์„ ํ˜„์ €ํžˆ ์ค„์—ฌ ๋…ธ์ด์ฆˆ ํ˜„์ƒ์œผ๋กœ๋ถ€ํ„ฐ ๋ฒ—์–ด๋‚˜๊ณ  ์žˆ์Œ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค. ๊ฐ๊ฐ ์ž…๋ ฅ์ „์••์—์„œ์˜ ์‡ ๊ณต์˜ ์œ„์น˜ ์ถœ๋ ฅ ์„ฑ๋Šฅ์€ ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

ํ‘œ 3. $\epsilon_{2}$์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ์ถœ๋ ฅ ์„ฑ๋Šฅ

Table 3. Output responses changes based on variations in variable $\epsilon_{2}$

Settling time (2%)
(sec)
Over shoot
(%)
$\left | e_{ss}\right |$
(m)
Variance of $V_{m}$
[$V^{2}$]
$\epsilon_{1}$=0.4
$\epsilon_{2}$=1
4.972 0.17 0.00011 3.1083
$\epsilon_{1}$=0.4
$\epsilon_{2}$=2
4.676 0.23 0.00022 1.7112
$\epsilon_{1}$=0.4
$\epsilon_{2}$=3
4.488 0.52 0.0004 1.2651

Table 3์—์„œ $\epsilon_{2}$์˜ ์ฆ๊ฐ€๋ฅผ ํ†ตํ•ด ์ž…๋ ฅ์ „์••์˜ ๋ถ„์‚ฐ ๊ฐ’์ด ๊ฐ์†Œ๋˜๋Š” ๊ฒƒ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ์ด๋Š” ํ‰๊ท  ์ „์••๊ฐ’์—์„œ ๋ฉ€๋ฆฌ ํผ์ง€๋Š” ์ŠคํŒŒ์ดํฌ ํ˜„์ƒ์ด ์ค„์–ด๋“œ๋Š” ๊ฒƒ์„ ์ˆ˜์น˜์ ์œผ๋กœ ๋ณด์—ฌ์ฃผ๋Š” ๊ฒƒ์ด๋ฉฐ ์ œ์–ด ์ž…๋ ฅ ์„ฑ๋Šฅ์ด ํ–ฅ์ƒ๋˜์—ˆ์Œ์„ ๋ณด์—ฌ์ค€๋‹ค.

๋˜ํ•œ, Fig. 12์˜ ์‡ ๊ณต ์œ„์น˜ ๊ถค์ ์—์„œ ์ŠคํŒŒ์ดํฌ ํ˜„์ƒ์ด ๋‚˜ํƒ€๋‚˜์ง€ ์•Š์•˜์ง€๋งŒ, ์ถ”๊ฐ€ ์‹คํ—˜ ๊ฒฐ๊ณผ $\epsilon_{1}$์„ ๋” ๋‚ฎ๊ฒŒ ์„ค์ •ํ•  ๊ฒฝ์šฐ ์œ„์น˜ ๊ถค์ ์—๋„ ์ŠคํŒŒ์ดํฌ ํ˜„์ƒ์ด ๋ฐœ์ƒํ•จ์„ ํ™•์ธํ•˜์˜€๋‹ค. ๋”ฐ๋ผ์„œ ์ž…๋ ฅ์ „์••์˜ ์ŠคํŒŒ์ดํฌ ์ €๊ฐ์€ ์ถœ๋ ฅ ๊ถค์  ์•ˆ์ •์„ฑ ํ™•๋ณด์—๋„ ๊ธฐ์—ฌํ•  ์ˆ˜ ์žˆ๋Š” ์ค‘์š”ํ•œ ์ œ์–ด ์„ฑ๋Šฅ๊ฐœ์„  ํšจ๊ณผ์ž„์„ ์˜๋ฏธํ•œ๋‹ค.

4์žฅ 4.2.a์™€ 4.2.b ๋ถ„์„์˜ ๊ฒฐ๋ก ์œผ๋กœ ์–ธ๊ธ‰ํ–ˆ๋“ฏ์ด ์‹ค์ œ ์‹คํ—˜์—์„œ $\epsilon_{2}$๋ฅผ ์ฆ๊ฐ€์‹œํ‚ค๋ฉด $T_{cross}(\epsilon_{2})$๊ฐ€ ์ปค์ง€๋Š”๋ฐ, ์ด๋กœ ์ธํ•ด ๋ฏธ์„ธํ•˜์ง€๋งŒ ์˜ค๋ฒ„์ŠˆํŠธ๊ฐ€ ์ฆ๊ฐ€ํ•˜๊ณ , ์ •์ฐฉ์‹œ๊ฐ„์€ ๋‹จ์ถ•๋˜๋Š” ๊ฒฝํ–ฅ์„ ๋ณด์˜€์œผ๋ฉฐ, ์ •์ƒ์ƒํƒœ์˜ค์ฐจ ๋˜ํ•œ ์†Œํญ ์ฆ๊ฐ€ํ•จ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค. ์ด๋Š” ์™ธ๋ถ€ ํ๋ฃจํ”„๋ฅผ ํ•ด์„ํ•  ๋•Œ, ์‹ค์ œ๋กœ๋Š” 4์ฐจ์ธ ๋ณผ-๋น” ์‹œ์Šคํ…œ ๋ชจ๋ธ์„ ๋ถ„๋ฆฌ๊ตฌ์กฐ์—์„œ ๋‚ด๋ถ€ ํ๋ฃจํ”„๋ฅผ ์ œ์™ธํ•˜๊ณ  ํ•ด์„ํ•˜๋ฉฐ ๋ฐœ์ƒํ•œ crossover term์˜ ๊ฐ„์„ญ ํšจ๊ณผ์— ์˜ํ•œ ๋ฏธ์„ธํ•œ ์ฐจ์ด์ด๋‹ค. Table 2์™€ ํ•จ๊ป˜ ์ „์ฒด ๋ณผ-๋น” ์‹œ์Šคํ…œ์˜ ์„ฑ๋Šฅ์œผ๋กœ์จ ๊ณ ๋ คํ–ˆ์„ ๋•Œ, ์ด๋Ÿฌํ•œ ๋ฏธ์„ธํ•œ ์ฐจ์ด์ ์ด ์žˆ์ง€๋งŒ ์‰ฝ๊ฒŒ ๋ณด์ •์ด ๊ฐ€๋Šฅํ•˜๋ฉฐ 4์ฐจ ์‹œ์Šคํ…œ์ด ์•„๋‹Œ 2์ฐจ ๋ชจ๋ธ๋กœ์˜ ๊ฐ„ํŽธํ•œ ํ•ด์„์„ ํ•  ์ˆ˜ ์žˆ๋‹ค๋Š” ์ ๊ณผ ์ „์•• ํฌํ™”๋ฅผ ๋ณด์ƒํ•ด์ค€๋‹ค๋Š” ํฐ ์ด์ ์ด ์žˆ๋‹ค๋Š” ๊ฒƒ์„ ์ „๋‹ฌํ•˜๊ณ ์ž ํ•œ๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ์šฐ๋ฆฌ๋Š” ์˜ค๋ฒ„์ŠˆํŠธ, ์ •์ƒ์ƒํƒœ์˜ค์ฐจ์™€ ๋…ธ์ด์ฆˆ ํ˜„์ƒ ์˜ˆ๋ฐฉ์„ ๊ณ ๋ คํ•ด $\epsilon_{2}=2$๋ฅผ ์‚ฌ์šฉํ•˜๊ธฐ๋กœ ํ•œ๋‹ค.

5.3 ์ด์ค‘ PID/PD์ œ์–ด๊ธฐ์™€ ์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๊ธฐ ๋น„๊ต

๋‹ค์Œ์€ ์ด์ค‘ PID/PD์ œ์–ด๊ธฐ($\epsilon_{1}=1, \epsilon_{2}=1$)์™€ ์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๊ธฐ($\epsilon_{1}=0.4, \epsilon_{2}=2$)์˜ ์„ฑ๋Šฅ๋น„๊ต์ด๋‹ค.

๊ทธ๋ฆผ 14. ์ด์ค‘ PID/PDvs ์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๊ธฐ ์„ฑ๋Šฅ ๋น„๊ต

Fig. 14. Dual PID/PD vs dual $\epsilon$-PID/PD controllers

../../Resources/kiee/KIEE.2025.74.12.2310/fig14.png

๋‹ค์Œ์œผ๋กœ ์‚ฌ๊ฐํŒŒ ์ฐธ์กฐ์‹ ํ˜ธ์—์„œ์˜ ์ œ์–ด๊ธฐ์˜ ์‘๋‹ต ํŠน์„ฑ์„ ๋น„๊ตํ•œ๋‹ค.

๊ทธ๋ฆผ 15. ์‚ฌ๊ฐํŒŒ ์ฐธ์กฐ์‹ ํ˜ธ์—์„œ์˜ ์ด์ค‘ PID/PD์ œ์–ด๊ธฐ vs ์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๊ธฐ ์„ฑ๋Šฅ ๋น„๊ต

Fig. 15. Dual PID/PD controller vs dual $\epsilon$-PID/PD controller for a square-wave reference

../../Resources/kiee/KIEE.2025.74.12.2310/fig15.png

Fig. 14, Fig. 15์™€ Table 2, Table 3๋ฅผ ํ†ตํ•ด ์ถ”๊ฐ€ ์ด๋“์š”์†Œ($\epsilon_{1}=$$0.4$, $\epsilon_{2}=$$2$)๋ฅผ ์‚ฌ์šฉํ•œ ์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๊ธฐ๊ฐ€ ๊ธฐ์กด ์ด์ค‘ PID/PD์ œ์–ด๊ธฐ ๋Œ€๋น„ ์ •์ฐฉ์‹œ๊ฐ„์ด 71.45 [%] ๋‹จ์ถ•๋˜์—ˆ๊ณ , ์˜ค๋ฒ„์ŠˆํŠธ๊ฐ€ 99.59 [%]๊ฐ์†Œ, ์ •์ƒ์ƒํƒœ์˜ค์ฐจ๊ฐ€ 98.28[%] ๊ฐ์†Œํ•˜์˜€์Œ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.

Remark 2. ๋ณธ ๋…ผ๋ฌธ์˜ ๊ธฐ์—ฌ๋„๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(i) ๋” ๋งŽ์€ ๊ณ ์ฐจ ์‹œ์Šคํ…œ์—์„œ๋„ ๋ณต์žกํ•œ ํ•ด์„์˜ ๋‚œ์ด๋„๋ฅผ ์„œ๋ธŒ์‹œ์Šคํ…œ์œผ๋กœ์˜ ๋ถ„๋ฆฌ๊ตฌ์กฐ๋ฅผ ํ†ตํ•ด ๋‚ฎ์ถ”์–ด, ๋™์ ํŠน์„ฑ์„ ๊ตฌ๋ณ„ํ•˜๋ฉฐ ๋ช…ํ™•ํ•˜๊ฒŒ ๋ถ„์„ํ•˜๋Š” ๋ฐฉ๋ฒ•์œผ๋กœ ์ ์šฉ๋  ์ˆ˜ ์žˆ๋‹ค.

(ii) ๋ณผ-๋น” ์‹œ์Šคํ…œ๋ณด๋‹ค๋„ ์ด๋“์กฐ์ ˆ์š”์†Œ๊ฐ€ ๋งŽ์€ ๋ณต์žกํ•œ ์‹œ์Šคํ…œ์—์„œ ์ถ”๊ฐ€ ์ด๋“์กฐ์ ˆ์š”์†Œ๋ฅผ ํ™œ์šฉํ•ด ์ „์ฒด ์ด๋“ ํŠœ๋‹์„ ๋ฏธ์„ธํ•˜๊ฒŒ ์กฐ์ ˆํ•˜๊ณ  ์‰ฝ๊ฒŒ ์„ฑ๋Šฅ์„ ์žฌ์กฐ์ •ํ•  ์ˆ˜ ์žˆ๋Š” ์ƒˆ๋กœ์šด ํŠœ๋‹๋ฐฉ์•ˆ์œผ๋กœ ์ ์šฉ๋  ์ˆ˜ ์žˆ๋‹ค.

6. ๊ฒฐ ๋ก 

๋ณผ-๋น” ์‹œ์Šคํ…œ์—์„œ ์‡ ๊ณต์˜ ์œ„์น˜๋ฅผ ์ œ์–ดํ•˜๊ธฐ ์œ„ํ•ด ๋ชจํ„ฐ ๊ฐ๋„๋ฅผ ๊ฐ€์ƒ์ž…๋ ฅ์œผ๋กœ ๊ฐ€์ •ํ•œ ์„œ๋ธŒ ์‹œ์Šคํ…œ์œผ๋กœ์˜ ์žฌ๊ตฌ์„ฑ๊ณผ ์ด๋“์กฐ์ ˆ์š”์†Œ๊ฐ€ ํฌํ•จ๋œ ์ด์ค‘ $\epsilon$-PID/PD์ œ์–ด๊ธฐ๋ฅผ ์ œ์•ˆํ•˜์˜€๋‹ค. ์ด๋“์กฐ์ ˆ์š”์†Œ $\epsilon_{1}$, $\epsilon_{2}$์˜ ์œ ์šฉ์„ฑ์„ ํŒŒ์•…ํ•˜๊ธฐ ์œ„ํ•ด์„œ ๋‚ด๋ถ€ ํ๋ฃจํ”„์™€ ์™ธ๋ถ€ ํ๋ฃจํ”„ ์‹œ์Šคํ…œ์„ ๋ถ„์„ํ•˜๊ณ , ๊ฐ ์ด๋“์กฐ์ ˆ์š”์†Œ์˜ ๋ณ€ํ™”์— ๋”ฐ๋ฅธ ํŠน์„ฑ์— ๋งž์ถฐ ์ด๋“ํŠœ๋‹๋ฐฉ์•ˆ์„ ์ œ์•ˆํ•˜์˜€๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ๋ณผ-๋น” ์‹œ์Šคํ…œ์˜ ์ •์ฐฉ์‹œ๊ฐ„, ์˜ค๋ฒ„์ŠˆํŠธ ๊ทธ๋ฆฌ๊ณ  ์ •์ƒ์ƒํƒœ์˜ค์ฐจ ์„ฑ๋Šฅ์„ ๊ฐœ์„ ์‹œ์ผœ์ฃผ๋Š” ์ด๋“์กฐ์ ˆ์š”์†Œ๋ฅผ ๋„์ถœํ•  ์ˆ˜ ์žˆ์—ˆ๋‹ค. ์ œ์•ˆ๋œ ์ œ์–ด๊ธฐ์„ค๊ณ„ ๋ฐ ์ด๋“ํŠœ๋‹๋ฐฉ์•ˆ์€ ์ถ”ํ›„ ๋‹ค์–‘ํ•œ ์‹œ์Šคํ…œ์— ํ™•์žฅ๋˜์–ด ์ ์šฉ๋  ๊ฒƒ์„ ๊ธฐ๋Œ€ํ•œ๋‹ค.

Acknowledgements

โ€œThis research was supported by the Regional Innovation System & Education(RISE) program through the Institute for Regional Innovation System & Education in Busan Metropolitan City, funded by the Ministry of Education(MOE) and the Busan Metropolitan City, Republic of Korea.(2025-RISE-02-003-044)โ€

References

1 
M. F. Rahmat, H. Wahid, N. A. Wahab, 2010, Application of intelligent controller in a ball and beam control system, International journal on smart sensing and intelligent systems, Vol. 3, No. 1, pp. 45-60Google Search
2 
J. Hauser, S. Sastry, P. Kokotovic, 1992, Nonlinear control via approximate input-output linearization: The ball and beam example, IEEE Transactions on Automatic Control, Vol. 37, No. 3, pp. 1987-1993Google Search
3 
J.-S. Kim, H.-L. Choi, 2010, Robust control design using the ฮต-sliding surface for ball and beam system, The transactions of the Korean Institute of Electrical Engineers, Vol. 59, No. 8, pp. 1444-1448Google Search
4 
K.-T. Lee, H.-L. Choi, 2017, Switching control of ball and beam system using partial state feedback: Jacobian and two-step linearization methods, The transactions of the Korean Institute of Electrical Engineers, Vol. 66, No. 5, pp. 819-832Google Search
5 
A. Ahmad, N. Syazreen, 2023, Modeling and hybrid pso-woa-based intelligent PID and state-feedback control for ball and beam systems, IEEE Access, Vol. 11, pp. 137866-137880DOI
6 
Y. Joo, 2025, Stability analysis on iterative PID control design algorithm for maximizing proportional gain, The transactions of The Korean Institute of Electrical Engineers, Vol. 74, No. 2, pp. 332-338DOI
7 
I. D. Diaz-Rodriguez, S. Han, S. P. Bhattacharyya, 2019, Analytical Design of PID ControllersGoogle Search
8 
H.-S. Lee, J.-S. Park, H.-L. Choi, 2024, Quadrotor altitude control with experimental data-based PID controller, Journal of IKEEE, Vol. 28, No. 2, pp. 14-22Google Search
9 
H.-L. Choi, Y.-H. Chang, Y. Oh, J.-T. Lim, 2008, On robust position control of DC motors by ฯต-PID controller and its application to humanoid robot arms, Vol. 41, No. 2, pp. 14449-14453DOI
10 
H. K. Khalil, J. W. Grizzle, 2002, Nonlinear SystemsGoogle Search

์ €์ž์†Œ๊ฐœ

๊น€์„ฑํ˜‘(Sung-Hyup Kim)
../../Resources/kiee/KIEE.2025.74.12.2310/au1.png

2025 : BS degree in Electrical Engineering, Dong-A University.

์ตœํ˜ธ๋ฆผ(Ho-Lim Choi)
../../Resources/kiee/KIEE.2025.74.12.2310/au2.png

Reference to Journal of the Institute of Korean Electrical and Electronics Engineers Vol. 22, No. 1.