nuttx/libs/libdsp/lib_observer.c

348 lines
10 KiB
C
Raw Normal View History

/****************************************************************************
* control/lib_observer.c
*
* Copyright (C) 2018 Gregory Nutt. All rights reserved.
* Author: Mateusz Szafoni <raiden00@railab.me>
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
* 3. Neither the name NuttX nor the names of its contributors may be
* used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
* FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
* COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
* OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
* AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
* ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*
****************************************************************************/
/****************************************************************************
* Included Files
****************************************************************************/
#include <stddef.h>
#include <assert.h>
#include <dsp.h>
/****************************************************************************
* Pre-processor Definitions
****************************************************************************/
/****************************************************************************
* Private Functions
****************************************************************************/
/****************************************************************************
* Public Functions
****************************************************************************/
/****************************************************************************
* Name: motor_observer_init
*
* Description:
* Initialize motor observer
*
* Input Parameters:
* observer - pointer to the common observer data
* ao - pointer to the angle specific observer data
* so - pointer to the speed specific observer data
* per - observer execution period
*
* Returned Value:
* None
*
****************************************************************************/
void motor_observer_init(FAR struct motor_observer_s *observer,
FAR void *ao, FAR void *so, float per)
{
DEBUGASSERT(observer != NULL);
DEBUGASSERT(ao != NULL);
DEBUGASSERT(so != NULL);
DEBUGASSERT(per > 0.0);
/* Set observer period */
observer->per = per;
/* Connect angle estimation observer data */
observer->ao = ao;
/* Connect speed estimation observer data */
observer->so = so;
}
/****************************************************************************
* Name: motor_observer_smo_init
*
* Description:
* Initialize motor sliding mode observer.
*
* Input Parameters:
* smo - pointer to the sliding mode observer private data
* kslide - SMO gain
* err_max - linear region upper limit
*
* Returned Value:
* None
*
****************************************************************************/
void motor_observer_smo_init(FAR struct motor_observer_smo_s *smo,
float kslide,
float err_max)
{
DEBUGASSERT(smo != NULL);
DEBUGASSERT(kslide > 0.0);
DEBUGASSERT(err_max > 0.0);
/* Initialize structure */
smo->k_slide = kslide;
smo->err_max = err_max;
}
/****************************************************************************
* Name: motor_observer_smo
*
* Description:
* One step of the SMO observer.
* REFERENCE: http://ww1.microchip.com/downloads/en/AppNotes/01078B.pdf
*
* Below some theoretical backgrounds about SMO.
*
* The digitalized motor model can be represent as:
*
* d(i_s.)/dt = (-R/L)*i_s. + (1/L)*(v_s - e_s. - z)
*
* We compare estimated current (i_s.) with measured current (i_s):
*
* err = i_s. - i_s
*
* and get correction factor (z):
*
* sign = sing(err)
* z = sign*K_SLIDE
*
* Once the digitalized model is compensated, we estimate BEMF (e_s.) by
* filtering z:
*
* e_s. = low_pass(z)
*
* The estimated BEMF is filtered once again and used to approximate the
* motor angle:
*
* e_filtered_s. = low_pass(e_s.)
* theta = arctan(-e_alpha/e_beta)
*
* The estimated theta is phase-shifted due to low pass filtration, so we
* need some phase compensation. More details below.
*
* where:
* v_s - phase input voltage vector
* i_s. - estimated phase current vector
* i_s - phase current vector
* e_s. - estimated phase BEMF vector
* R - motor winding resistance
* L - motor winding inductance
* z - output correction factor voltage
*
* Input Parameters:
* observer - (in/out) pointer to the common observer data
* i_ab - (in) inverter alpha-beta current
* v_ab - (in) inverter alpha-beta voltage
* phy - (in) pointer to the motor physical parameters
*
* Returned Value:
* None
*
****************************************************************************/
void motor_observer_smo(FAR struct motor_observer_s *observer, FAR ab_frame_t *i_ab,
FAR ab_frame_t *v_ab, FAR struct motor_phy_params_s *phy)
{
FAR struct motor_observer_smo_s *smo =
(FAR struct motor_observer_smo_s *)observer->ao;
FAR ab_frame_t *emf = &smo->emf;
FAR ab_frame_t *z = &smo->z;
FAR ab_frame_t *i_est = &smo->i_est;
FAR ab_frame_t *v_err = &smo->v_err;
FAR ab_frame_t *i_err = &smo->i_err;
FAR ab_frame_t *sign = &smo->sign;
float i_err_a_abs = 0.0;
float i_err_b_abs = 0.0;
float angle = 0.0;
/* REVISIT: observer works only when IQ current is high enough */
/* Calculate observer gains */
smo->F_gain = (1 - observer->per*phy->res/phy->ind);
smo->G_gain = observer->per/phy->ind;
/* Saturate F gain */
if (smo->F_gain < 0)
{
smo->F_gain = 0.0;
}
/* Saturate G gain */
if (smo->G_gain > 0.999)
{
smo->G_gain = 0.999;
}
/* Configure low pass filters
*
* We tune low-pass filters to achieve cutoff frequency equal to
* input singal frequency. This gives us constant phase shift between
* input and outpu signals equals to:
*
* phi = -arctan(f_in/f_c) = -arctan(1) = -PI/2
*
* Input signal frequency is equal to the frequency of the motor currents,
* which give us:
*
* f_c = omega_e/(2*PI)
* omega_m = omega_e/poles
* f_c = omega_m*poles/(2*PI)
*
* filter = T * (2*PI) * f_c
* filter = T * omega_m * poles
*
* T - [s] period at which the digital filter is being calculated
* f_in - [Hz] input frequency of the filter
* f_c - [Hz] cutoff frequency of the filter
* omega_m - [rad/s] mechanical angular velocity
* omega_e - [rad/s] electrical angular velocity
*
*/
smo->emf_lp_filter1 = observer->per * observer->speed * phy->poles;
smo->emf_lp_filter2 = smo->emf_lp_filter1;
/* Get voltage error: v_err = v_ab - emf */
v_err->a = v_ab->a - emf->a;
v_err->b = v_ab->b - emf->b;
/* Estimate stator current */
i_est->a = smo->F_gain * i_est->a + smo->G_gain * (v_err->a - z->a);
i_est->b = smo->F_gain * i_est->b + smo->G_gain * (v_err->b - z->b);
/* Get motor current errror */
i_err->a = i_ab->a - i_est->a;
i_err->b = i_ab->b - i_est->b;
/* Slide-mode controller */
sign->a = (i_err->a > 0 ? 1.0 : -1.0);
sign->b = (i_err->b > 0 ? 1.0 : -1.0);
/* Get current error absolute value - just multiply value with its sign */
i_err_a_abs = i_err->a * sign->a;
i_err_b_abs = i_err->b * sign->b;
/* Calculate new output correction factor voltage */
if (i_err_a_abs < smo->err_max)
{
/* Enter linear region if error is small enough */
z->a = i_err->a * smo->k_slide / smo->err_max;
}
else
{
/* Non-linear region */
z->a = sign->a * smo->k_slide;
}
if (i_err_b_abs < smo->err_max)
{
/* Enter linear region if error is small enough */
z->b = i_err->b * smo->k_slide / smo->err_max;
}
else
{
/* Non-linear region */
z->b = sign->b * smo->k_slide;
}
/* Filter z to obtain estimated emf */
LP_FILTER(emf->a, z->a, smo->emf_lp_filter1);
LP_FILTER(emf->b, z->b, smo->emf_lp_filter1);
/* Filter emf one more time before angle stimation */
LP_FILTER(emf->a, emf->a, smo->emf_lp_filter2);
LP_FILTER(emf->b, emf->b, smo->emf_lp_filter2);
/* Estimate phase angle according to:
* emf_a = -|emf| * sin(th)
* emf_b = |emf| * cos(th)
* th = atan2(-emf_a, emf->b)
*/
angle = fast_atan2(-emf->a, emf->b);
#if 1
/* Some assertions
* TODO: simplify
*/
if (angle != angle) angle = 0.0;
if (emf->a != emf->a) emf->a = 0.0;
if (emf->b != emf->b) emf->b = 0.0;
if (z->a != z->a) z->a = 0.0;
if (z->b != z->b) z->b = 0.0;
if (i_est->a != i_est->a) i_est->a = 0.0;
if (i_est->b != i_est->b) i_est->b = 0.0;
#endif
/* Angle compensation.
* Due to low pass filtering we have some delay in estimated phase angle.
*
* Adaptive filters introduced above cause -PI/2 phase shift for each filter.
* We use 2 times filtering which give us constant -PI phase shift.
*/
angle = angle -M_PI_F;
/* Normalize angle to range <0, 2PI> */
angle_norm_2pi(&angle, 0.0, 2*M_PI_F);
/* Store estimated angle in observer data*/
observer->angle = angle;
}