PUTM_EV_Data_Acquisition_Card_2022

About the project

It's main task is to receive signals from sensors mounted on the car and analyze them and send to CAN-bus.

Hardware design

The hardware part of the project is the PCB board which is connected to the sensors and CAN-bus. The microcontroller which is used is STM32L4P5RET6. The PCB is supply by 5VDC.

Sensors

It receive signals from :

~~inductive sensors - there are used to measure wheels rotation speed~~
brake pressure sensors - which measure the pressure in the brake system
shutdown circuit sense:

BSPD
Overtravel switch
Driver's kill switch
Left and right kill switches

acceleration - from MPU on board

Program flow diagram

)

DMA and ADC timings consideration

The program is required to collect and relay data @ 100 ms. Two adc peripherals are used to collect data from 3 analog sensors each.

ADC clock frequency: $32 MHz$

Sampling time: $92.5$ cycles - reason: High impedance inputs are used, subject to change

Oversampling: by 256

Sample conversion time: $T_{sample} + 12.5 = 105 cycles$

With oversampling: $T_{total} = 105 cycles * 256 samples = 26880 cycles$

$$ t = \frac{26800 * 3 channels}{32 * 10^6 Hz} = 2.52 * 10^{-3} s = 2.52 ms $$

1D Kalman filter design for the IMU data

$$ \vec{x}(t + 1) = F(t + 1; t) * \vec{x}(t) + G * \vec{u} + w(t) $$ $$ \vec{z} = H * \vec{x}(t) $$

$$ \vec{x} = \begin{bmatrix} a_x \ a_y \ a_z \end{bmatrix} $$

$$ F = \begin{bmatrix} 1 \ 1 \ 1 \end{bmatrix} \space\space\space H = \begin{bmatrix} 1 \ 1 \ 1 \end{bmatrix} \space\space\space G = [1] \space u = [0] $$

$$ P(t + 1) = F(t)P(t) F(t)^T + Q(t) $$

Since there is no way to calculate the change in acceleration, w(t) is estimated to gaussian N(0, $\sigma$) as process noise.

$$ Q(t) = \begin{bmatrix} \sigma_x \space 0 \space 0 \ 0 \space \sigma_y \space 0 \ 0 \space 0 \space \sigma_z\end{bmatrix} $$

$$ \sigma_{process} = \sigma_x = \sigma_y = \sigma_z $$

$$ R = \begin{bmatrix} ? ? ? \ ? ? ? \ ? ? ? \end{bmatrix} $$

R to be selected.