3.5.38
Intuition Badi picture (YE KYUN exist karta hai)Ek spacecraft (ya drone, ya thermostat) ka ek goal hota hai — ek target attitude, altitude, ya speed.
Real system hamesha thoda off rehta hai. "Main jahan rehna chahta hun" aur "main jahan hun" ke beech ka gap error hai.
Ek controller ek aisi machine hai jo error ko corrective action mein badal deta hai. PID sabse simple, sabse zyada use hone wali recipe hai: yeh error ke present (P), uske past (I), aur uske future (D) ko dekhta hai, aur teeno ko ek command mein blend karta hai.
TEEN terms kyun? Kyunki akela ek term hamesha ek predictable tarike se fail hota hai — aur har naya term pichle term ki bimari ka ilaaj hai.
Definition Core quantities
Setpoint r ( t ) Process variable y ( t ) Error e ( t ) = r ( t ) − y ( t ) Control signal u ( t ) function hai u = f ( e ) f
f e ( t )
Uski abhi ki value → abhi mistake kitni badi hai?
Uski accumulated history → kya hum consistently ek direction mein off rahe hain?
Uska rate of change → kya mistake badh rahi hai ya ghatt rahi hai?
Ye exactly value, integral, derivative hain — woh teen cheezein jo calculus aapko kisi function se extract karne deta hai.
Intuition Proportional kyunAgar aap target se door hain, toh zor se push karo; agar paas hain, toh dhire karo. Correction proportional hai error ke — jaise steering harder karte ho jitna zyada aap apni lane se drift kar chuke ho.
u P ( t ) = K p e ( t )
K p
Common mistake Steel-man: "BasK p Kyun sahi lagta hai: bada K p Flaw — steady-state error. Maan lo gravity ya drag continuously aapko oppose kar raha hai (ek constant disturbance d K p e ss = d e ss = K p d = 0. K p shrink kar sakte ho, lekin kabhi kill nahi kar sakte — kyunki zero error zero output dega, aur phir disturbance se koi nahi ladega. Worse, bahut bada K p oscillate / unstable bana deta hai.
Fix: aapko ek aise term ki zaroorat hai jo tab bhi push karta rahe jab instantaneous error tiny ho → integral.
Integral yaad rakhta hai . Ek chhota sa error bhi, agar persist karta rahe, accumulator mein pile up hota rehta hai jab tak output itna bada na ho jaye ki use eliminate kar sake. Yeh woh steady-state offset khatam kar deta hai jo P chod jaata hai.
u I ( t ) = K i ∫ 0 t e ( τ ) d τ
HOW it removes steady-state error (derivation): Steady state mein e = e ss u d t d u I = K i e u I d t d u I = 0 ⇒ K i e ss = 0 ⇒ e ss = 0 jab tak koi bhi error baaqi hai, integrator output change karta rehta hai , aur ekmaatra resting state zero error hai.
Common mistake Steel-man: "Integral sirf help karta hai, toh badaK i Kyun sahi lagta hai: zyada memory = offset zyada jaldi khatam hoga.
Flaw — integral windup & overshoot. Agar actuator saturate ho jaaye (max thrust reach ho) jab error bada ho, toh integral ek bahut bada stored value accumulate karta rehta hai. Jab aap finally target reach karte ho, woh stored value aapko isse bahut aage le jaata hai pehle ki woh unwind ho.
Fix: K i anti-windup use karein (saturate hone par clamping/integrating band karo).
Intuition Derivative kyunDerivative predict karta hai . Agar error tezi se gir raha hai, aap overshoot karne wale ho — toh abhi braking shuru karo. Yeh damping add karta hai, jaise ek shock absorber jo rapid change ka resist karta hai.
u D ( t ) = K d d t d e ( t )
HOW it damps: agar y r e e ˙ < 0 u D < 0
Common mistake Steel-man: "Derivative future dekhta hai, toh bahut zyada add karo."Kyun sahi lagta hai: zyada anticipation = smoother landing.
Flaw — noise amplification. Differentiation high-frequency measurement noise ko multiply karta hai. Ek jittery sensor with small fast wiggles se enormous e ˙ K d u Fix: chhota K d filtered derivative use karo).
Worked example Example 1 — Attitude hold, pure P vs P+IEk reaction wheel satellite ke pitch angle ko control karta hai. Target r = 0 ∘ d = 2 u = d
Pure P , K p = 4 e ss = d / K p = 2/4 = 0. 5 ∘ Ye step kyun? Rest par u = K p e ss d I add karo: ab integrator ramp karta hai jab tak u = 2 e ss = 0 Kyun? Steady state force karta hai u ˙ I = K i e = 0 ⇒ e = 0
Worked example Example 2 — Discrete step (ek control update compute karo)Given K p = 2 , K i = 0.5 , K d = 1 , Δ t = 0.1 e k − 1 = 1.0 e k = 0.8 I p r e v = 0.3 ∑ e j Δ t
Integral: I k = I p r e v + e k Δ t = 0.3 + 0.8 ( 0.1 ) = 0.38 u I = K i I k = 0.5 ( 0.38 ) = 0.19 Kyun: naya error rectangle accumulate karo.
Proportional: u P = K p e k = 2 ( 0.8 ) = 1.6 Kyun: abhi par react karo.
Derivative: u D = K d Δ t e k − e k − 1 = 1 ⋅ 0.1 0.8 − 1.0 = − 2.0 Kyun: error gir raha hai, toh brake karo.
u k = 1.6 + 0.19 − 2.0 = − 0.21 Interpretation: bhale hi error positive hai, strong falling-rate D ko dominate karta hai → hum ease/reverse karte hain overshoot avoid karne ke liye.
Worked example Example 3 — Tuning intuitionSystem oscillate karta hai aur kabhi settle nahi hota: K d K i K p Kyun: har symptom us term se map hota hai jiska woh kaam hai.
Recall Ise ek 12-saal ke bacche ko explain karo (reveal karne ke liye click karo)Socho ek toy boat ko ek pond mein ek dot tak steer karna hai.
P: boat dot se jitni door hai, utna hard tum wheel ghoomte ho.I: agar hawa ise off nudge karti rehti hai aur P ise thoda short chhodta rehta hai, toh tum irritate ho jaate ho aur extra turn add karte ho ki yeh kitni der se galat hai — jab tak woh finally dot par aa jaaye.D: jab boat dot ki taraf zoom kar rahi hoti hai, tum jaldi un-turning shuru karte ho taaki woh isse aage na nikal jaaye.
Present, past, future — teen helpers ek boat steer kar rahe hain.
Mnemonic Teen kaam yaad karo"P-I-D = Push, Integrate the past, Damp the future."
Ya: P = Present, I = Is-history, D = Destiny.
Symptoms: P ush bahut weak → slow; koi I nahi → offset; koi D nahi → overshoot.
Control loop mein error signal kya hota hai? e ( t ) = r ( t ) − y ( t )
Continuous PID control law likhein. u = K p e + K i ∫ 0 t e d τ + K d e ˙
Pure-P controller constant disturbance ke under steady-state error kyun chhodta hai? Equilibrium par u = K p e ss d e ss = d / K p = 0 d
Kaun sa term steady-state error eliminate karta hai, aur kyun? Integral: steady state ko chahiye u ˙ I = K i e = 0 ⇒ e = 0
Kaun sa term damping add karta hai / overshoot reduce karta hai? Derivative — yeh e ˙
Bada K d Yeh high-frequency sensor noise amplify karta hai, control chatter cause karta hai (fix: derivative filter karo).
Integrator windup kya hai aur iska fix kya hai? Actuator saturated rehne par integral accumulate hota rehta hai, bada overshoot cause karta hai; anti-windup clamping se fix karo.
Discrete PID update equation do. u k = K p e k + K i ∑ j e j Δ t + K d Δ t e k − e k − 1
Symptom → term map karo: system kabhi settle nahi hota / oscillate karta hai. K d
Symptom → term map karo: system constant offset ke saath settle hota hai. K i
Feedback control loops — PID controller block ki ek choice hai.Steady-state error and system type — kyun I "type number" badhata hai.Stability & the Routh–Hurwitz criterion — gains stability ko kaise affect karte hain.Laplace transforms & transfer functions — C ( s ) = K p + K i / s + K d s Ziegler–Nichols tuning — systematic gain selection.Reaction wheels & attitude control — GNC actuators jinhein PID commands karta hai.Sensor noise & filtering — derivative ko low-pass filter kyun chahiye.
Integral term Ki integral e
