For each function, there are types of floating-point numbers and fixed-point numbers. Since the usage methods are the same, here we only take 32 as a floating-point number as an example.

One. BasicMathFunctions

1. Absolute value

pDst[n] = abs(pSrc[n]), 0 <= n < blockSize

Example

float32_t *pSrc;
float32_t *pDst;
uint32_t blocksize;
arm_add_f32(pSrc,pDst,blocksize);

2. Sum

pDst[n] = pSrcA[n] + pSrcB[n],      0 <= n < blockSize.
example

float32_t *pSrcA;
float32_t *pSrcB;
float32_t *pDst;
uint32_t blocksize;
arm_add_f32(pSrcA,pSrcB,pDst,blocksize);

3. Dot multiplication

sum = pSrcA[0]*pSrcB[0] + pSrcA[1]*pSrcB[1] + ... + pSrcA[blockSize-1]*pSrcB[blockSize-1]
example

float32_t *pSrcA;
float32_t *pSrcB;
float32_t *result;
uint32_t blocksize;
arm_dot_prod_f32(pSrcA,pSrcB,blocksize,result);

4. Multiplication

sum = pSrcA[0]*pSrcB[0] + pSrcA[1]*pSrcB[1] + ... + pSrcA[blockSize-1]*pSrcB[blockSize-1]
example

float32_t *pSrcA;
float32_t *pSrcB;
float32_t *pDst;
uint32_t blocksize;
arm_mult_f32(pSrcA,pSrcB,blocksize,pDst);

5. Opposite numbers

pDst[n] = -pSrc[n],      0 <= n < blockSize.   
example

float32_t *pSrc;
float32_t *pDst;
uint32_t blocksize;
arm_mult_f32(pSrcA,pDst,blocksize);

6. Offset

pDst[n] = pSrc[n] + offset,      0 <= n < blockSize.  
example

float32_t *pSrc;
float32_t offset;
float32_t *pDst;
uint32_t blocksize;
arm_offset_f32(pSrc,offset,pDst,blocksize);

7. Subtraction

pDst[n] = pSrcA[n] - pSrcB[n],   0 <= n < blockSize.    

Example

float32_t *pSrcA;
float32_t *pSrcB;
float32_t *pDst;
uint32_t blocksize;
arm_sub_f32(pSrcA,pSrcB,pDst,blocksize);

8. Scale factor

pDst[n] = pSrc[n] * scale,      0 <= n < blockSize.   

Example

float32_t *pSrc;
float32_t scale;
float32_t *pDst;
uint32_t blocksize;
arm_scale_f32(pSrc,scale,pDst,blocksize);

Two. FastMathFunctions

The functions in this block are some math.h functions that are used daily, so I won’t go into details

float32_t    arm_cos_f32(float32_t x) ；

float32_t arm_sin_f32(float32_t x) ；

arm_status arm_sqrt_q31(q31_t in, q31_t * pOut)；

Three.StatisticsMathFunctions

This piece is mainly for some functions that are often used in the process of statistical data analysis and processing.

1. Maximum

Calculate the maximum value in the array and return the maximum value in the array and the position of the maximum value in the array.

float32_t *pSrc;
float32_t pResult;
uint32_t blocksize;
uint32_t pIndex;
arm_max_f32(pSrc,blocksize,&pResult,&pIndex);

2. Minimum

Calculate the minimum value in the array and return the position of the maximum value and the maximum value in the array. Similar to the above usage, no further explanation.

arm_min_f32(pSrc,blocksize,&pResult,&pIndex);

3. Average

Result = (pSrc[0] + pSrc[1] + pSrc[2] + ... + pSrc[blockSize-1]) / blockSize; 

float32_t *pSrc;
float32_t pResult;
uint32_t blocksize;
arm_mean_f32(pSrc,blocksize,&pResult);

4. Power

Result = pSrc[0] * pSrc[0] + pSrc[1] * pSrc[1] + pSrc[2] * pSrc[2] + ... + pSrc[blockSize-1] * pSrc[blockSize-1]; 

float32_t *pSrc;
float32_t pResult;
uint32_t blocksize;
arm_power_f32(pSrc,blocksize,&pResult);

5. Standard deviation

Result = sqrt((sumOfSquares - sum2 / blockSize) / (blockSize - 1)) 
where:
sumOfSquares = pSrc[0] * pSrc[0] + pSrc[1] * pSrc[1] + ... + pSrc[blockSize-1] * pSrc[blockSize-1] 
sum = pSrc[0] + pSrc[1] + pSrc[2] + ... + pSrc[blockSize-1]   

float32_t *pSrc;
float32_t pResult;
uint32_t blocksize;
arm_std_f32(pSrc,blocksize,&pResult);

6. Root Mean Square

Result = sqrt(((pSrc[0] * pSrc[0] + pSrc[1] * pSrc[1] + ... + pSrc[blockSize-1] * pSrc[blockSize-1]) / blockSize));

float32_t *pSrc;
float32_t pResult;
uint32_t blocksize;
arm_rms_f32(pSrc,blocksize,&pResult);

7. Variance

Same as above

arm_var_f32(pSrc,blocksize,&pResult);

Four. SupportFunctions

Support functions mainly include data copy, assignment and type conversion. Type conversion generally does not need to be performed in DSP, so it is omitted here.

1. Data copy

pDst[n] = pSrc[n];      0 <= n < blockSize.

void arm_copy_f32(float32_t * pSrc, float32_t * pDst, uint32_t blockSize) 

2. Data filling

pDst[n] = value;      0 <= n < blockSize.

void arm_fill_f32(float32_t value, float32_t * pDst, uint32_t blockSize) 

Five. ContrillerFunctions

What is mainly introduced here is the function part of PID control.

PID controller is also called proportional integral derivative controller, and it is currently the most widely used controller. There are many methods for tuning PID parameters on the Internet. I think it is more important to first understand the meaning of PID parameters. Here is a link, https://www.zhihu.com/question/23088613/answer/81176620.

y[n] = y[n-1] + A0 * x[n] + A1 * x[n-1] + A2 * x[n-2]
A0 = Kp + Ki + Kd
A1 = (-Kp ) - (2 * Kd )
A2 = Kd
The function is defined as follows

static __INLINE float32_t arm_pid_f32(arm_pid_instance_f32 * S, float32_t in) 

typedef struct
{
float32_t A0; /**< The derived gain, A0 = Kp + Ki + Kd . */
float32_t A1; /**< The derived gain, A1 = -Kp - 2Kd. */
float32_t A2; /**< The derived gain, A2 = Kd . */
float32_t state[3]; /**< The state array of length 3. */
float32_t Kp; /**< The proportional gain. */
float32_t Ki; /**< The integral gain. */
float32_t Kd; /**< The derivative gain. */
} arm_pid_instance_f32;

Initialization function

void arm_pid_init_f32(arm_pid_instance_f32 * S, int32_t resetStateFlag)
This function is to obtain A0, A1, A2 through this function after configuring Kp, Ki, Kd by the user. The second parameter is the initialization flag bit, set 1 to initialize.

Reset function

void arm_pid_reset_f32(arm_pid_instance_f32 * S)