S curve ramp function

Question

David Cole 2025-5-17

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2177282-s-curve-ramp-function

回答： Sam Chak 2025-5-28

I need help creating a function for an s curve ramp. The code below will create the inserted picture below. However, this code only works for a constant ramp block (connected to x). This code from the piecewise function on the Matlab smf() help file page.

I need two additional inputs (acceleration rate and jerk rate). The output should ramp the input velocity (x) signal based on the acceleration rate (a), and jerk rate (j). I need the block or code to be dynamic. I do not want to put a time value inside of my code. I will be exporting the code to structure text and input the code into a Siemens PLC.

I don't want to include a fixed dt in my code because I will be running the code in a PLC. Maybe it would be better to design this function in simulink and set a fixed time solver and use integrator blocks?

x = 0:0.01:10; % domain

a = 1; % unsaturated interval's left endpoint

b = 8; % unsaturated interval's right endpoint

y = smf(x, a, b);

plot(x, y), grid on

xline(a, '--')

xline(b, '--')

title("S-curve piecewise function")

xlabel("x")

ylabel("Amplitude")

ylim([-0.05 1.05])

%% S-curve piecewise function

function y = smf(x, a, b)

m = (a + b)/2;

y = (x > a & x <= m) .* ( 2*((x-a)/(b-a)).^2) + ...

(x > m & x <= b) .* (1-2*((x-b)/(b-a)).^2) + ...

(x > b);

end

4 个评论
显示 2更早的评论隐藏 2更早的评论

David Cole 2025-5-17

I am working on a PLC and VFD project. My PLC does not have a native s curve block that will take in a continous signal and output a continous signal. For example, the PLC might receive a speed (velocity) signal of 10 m/s from an initial speed of 0 m/s. I need the PLC to generate a signal that goes from 0 to 10 and follows an s curve. The equipment will have an acceleration limit (m/s^2) and and jerk limit (m/s^3). I need the s curve to take both of these limit into consideration a produce a s curve that will ramp the speed from 0 to 10 m/s (as an example). The goal is to produce the code in Simulink to observe the correct behavior and then export to code so it can be used in the PLC.

John D'Errico 2025-5-17

编辑：John D'Errico 2025-5-17

Ah. You have a limit on acceleration AND jerk. Note that the set of curves that satisfy such a constraint will require nonlinear bounds on the curve. It should not be too difficult though.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Sam Chak 2025-5-18

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2177282-s-curve-ramp-function#answer_1565433

在 MATLAB Online 中打开

This might not be what PLC users like to do.

%% 99% Math-free Jerk-based Velocity

% parameter

c = 1e-3; % control parameter for accuracy purposes

% Velocity and Jerk settings

Vref = 10; % desired velocity (specified by PLC user)

Jlimit = 40/49; % Jerk limit (specified by PLC user)

Jon = 1.0; % Jerk on time (specified by PLC user)

Joff = 2/sqrt(Jlimit/Vref) + Jon; % Jerk off time <-- this is math!

Jmid = (Jon + Joff)/2; % Jerk half-time <-- this is math!

% simulation time

tStart = 0; % start running but not producing velocity until Jerk on time

h = 0.1; % sampling time

tEnd = round(1.2*Joff); % stop running

t = (tStart:h:tEnd);

% Jerk specifications on the PLC

tdata = [tStart Jon-c Jon+c Jmid-c Jmid+c Joff-c Joff+c tEnd];

Jdata = [0.0 0.0 Jlimit Jlimit -Jlimit -Jlimit 0.0 0.0];

% maybe call ordinary Jerk function?

d2y = @(t) interp1(tdata, Jdata, t); % OpenPLC can perform interpolation

Jerkode = @(t, y) [ y(2); % dy = y(2) acceleration

d2y(t)]; % d2y = dy(2)/dt jerk

% initial condition

y0 = [0; 0]; % zero velocity, zero acceleration

% set solver options

options = odeset('RelTol', 1e-6, 'AbsTol', 1e-8);

% call ode45 Solver (OpenPLC can solve simple ODEs)

[t, y] = ode45(@(t, y) Jerkode(t, y), t, y0, options);

% True Velocity Profile

tt = linspace(tStart, tEnd, 51);

TrueVel = Vref*smf(tt, Jon, Joff);

% Plot the solution

figure

hold on

plot(t, y(:,1))

plot(tt, TrueVel, 'o')

hold off

grid on

xlabel('Time')

ylabel('Amplitude')

title('Jerk-based S-curve Velocity')

legend('Numerical solution', 'True Velocity Profile', 'location', 'best')

●

%% S-curve membership function (borrowed from Fuzzy Logic Toolbox)

function y = smf(x, a, b)

m = (a + b)/2;

y = (x > a & x <= m) .* ( 2*((x-a)/(b-a)).^2) + ...

(x > m & x <= b) .* (1-2*((x-b)/(b-a)).^2) + ...

(x > b);

end

2 个评论
显示无隐藏无

David Cole 2025-5-18

Thank you for the code. I will start reviewing it immediately. My goal is to test the code and then export the code to structured text, which can be imported into a Siemens PLC. Within the PLC, the structured text can be made into a block (like a Simulink subsystem) that takes in inputs and produces an output. Once the constants are set they should never change. However, the input speed will change continuously throught the day based on the desired speed; for example, the speed might change from 0 to 10 m/s and 10 m/s to 3 m/s. The s curve block should produce an s curve output based on the user defined constants. The block will probably be placed in a 20 milli-second cyclic routine (like using a fixed time solver), but that might change. A 10 milli-second cyclic routine might be better.

I need to be careful about placing a time inside of the routine. I though about using integrators inside on simulink so I could avoid derivatives and odes.

Utilimately, here is what I want:

David Cole 2025-5-18

scurve.slx

Here is my simulink file. The bottom part works but the top part does not ramp down in a curve.

请先登录，再进行评论。

Answer 2

Sam Chak 2025-5-28

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2177282-s-curve-ramp-function#answer_1565533

在 MATLAB Online 中打开

Hi @David Cole

In addition to your user inputs (Vin, accel, and jerk), you also need to provide the "Jerk On" time. You can use my S-curvy function to generate the desired S-shaped velocity profile that satisfies the acceleration and jerk constraints.

If you find my solution helpful, please remember to vote and click "Accept".

%% user inputs

Vref = 10; % Target Velocity

Alimit = 0.5; % hardware Acceleration limit

Jlimit = 0.1; % hardware Jerk limit

Jon = 5.0; % Preferred Jerk-Trigger 'on' time

%% call my S-curvy fun

[t, V] = scurve(Vref, Alimit, Jlimit, Jon);

acc = gradient(V)./gradient(t);

jerk = gradient(acc)./gradient(t);

%% plot results

figure

plot(t, V), grid on

xlabel('Time')

ylabel('Amplitude')

title('I''m S-curve Velocity')

figure

subplot(211)

mv = min(acc);

Mv = max(acc);

yL = 0.2*(Mv - mv);

plot(t, acc), grid on

ylim([mv-yL, Mv+yL])

title('I''m acceleration')

subplot(212)

mv = min(jerk);

Mv = max(jerk);

yL = 0.2*(Mv - mv);

plot(t, jerk), grid on

ylim([mv-yL, Mv+yL])

title('I''m jerk')

%% S-curvy function

function [t, out] = scurve(Vref, Alimit, Jlimit, Jon)

% parameter

c = 1e-3; % control parameter for accuracy purposes

% Jerk settings

Alim = sqrt(Jlimit*Vref);

if Alim <= Alimit

Jlim = Jlimit;

else

Jlim = Alimit^2/Vref;

end

Joff = 2/sqrt(Jlim/Vref) + Jon; % Jerk off time

Jmid = (Jon + Joff)/2; % Jerk half-time

% simulation time

tStart = 0; % start running but not producing velocity until Jerk on time

h = 0.01; % sampling time

tEnd = round(Joff) + Jon; % stop running

t = (tStart:h:tEnd);

% Jerk specifications on the PLC

tdata = [tStart Jon-c Jon+c Jmid-c Jmid+c Joff-c Joff+c tEnd];

Jdata = [0.0 0.0 Jlim Jlim -Jlim -Jlim 0.0 0.0];

% maybe call ordinary Jerk function?

d2y = @(t) interp1(tdata, Jdata, t); % OpenPLC can perform interpolation

Jerkode = @(t, y) [ y(2); % dy = y(2) acceleration

d2y(t)]; % d2y = dy(2)/dt jerk

% initial condition

y0 = [0; 0]; % zero velocity, zero acceleration

% set solver options

options = odeset('RelTol', 1e-6, 'AbsTol', 1e-8);

% call ode45 Solver (OpenPLC can solve simple ODEs)

[t, y] = ode45(@(t, y) Jerkode(t, y), t, y0, options);

out = y(:,1);

end

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。