CN108875710A - Elevator door speed of service estimation method based on energy threshold algorithm - Google Patents

Abstract

The invention discloses a method for estimating the real-time running speed of an elevator door based on an energy threshold algorithm. The present invention comprises the following steps: step 1, obtaining the acceleration signal of the original elevator door operation through an acceleration sensor; step 2, performing detrending item processing on the obtained acceleration signal; step 3, converting it into frequency domain by fast Fourier transform, and filtering Denoising; step 4, perform Fourier inverse transform on the filtered frequency domain signal, and calculate the energy of the acceleration signal; step 5, set the threshold value for the running phase of the elevator, and remove the jitter noise signal; step 6, the final acceleration signal q(t) is integrally processed to obtain the corresponding real-time speed signal. The invention removes the jitter interference signal existing at the moment of opening and closing of the elevator door and in the gap between the opening and closing of the door, and at the same time ensures that the loss of useful signals is reduced, and the speed signal is obtained by integrating.

Description

Estimation Method of Elevator Door Running Speed Based on Energy Threshold Algorithm

technical field

The invention belongs to the field of electrical control and digital signal processing, and relates to a method for estimating the real-time running speed of an elevator door based on an energy threshold algorithm.

Background technique

As a means of public transportation, the most basic and essential function of an elevator is to ensure that people or goods are delivered to the destination floor smoothly and safely. Ensuring the safe operation of elevators and minimizing losses has become a top priority for government regulators, elevator manufacturers, elevator suppliers and other departments to urgently solve.

In the safe operation of the elevator, the door switch and running speed are one of the important indicators to measure whether the elevator is running normally. The acceleration signal of the elevator door can be used to calculate and estimate its real-time running speed very well, but the acceleration sensor is accompanied by various noises and interferences during the on-site acquisition process, resulting in the integrated speed signal in the process of acceleration signal processing. The error is large, and the result after the second integration is seriously distorted. At the same time, after frequency domain filtering is adopted, there are small-amplitude jitter signals in the static state of the elevator door in the opening and closing gap and at the moment of opening and closing the door. Therefore, the accurate processing algorithm of the acceleration signal of the elevator door is the core part of estimating its real-time speed value and running state.

This patent proposes a method for estimating the real-time running speed of the elevator door based on the combination of the energy threshold algorithm and the time-domain and frequency-domain filtering algorithm. The algorithm can more accurately estimate the acceleration value, speed value and running state of the real-time running elevator door. These parameters will be used as important indicators in the normal operation, safety monitoring and maintenance of the elevator.

Contents of the invention

The present invention processes the collected elevator door acceleration signals through a series of signal processing algorithms, and intends to obtain relatively accurate elevator door running acceleration and speed values, so as to know whether the elevator is running safely, provide accurate judgment of elevator operation, and reduce accidents. The probability of occurrence, improving the efficiency of monitoring and operation and maintenance, and reducing labor costs provide guarantees.

The main implementation process of the patented algorithm is as follows: firstly, the detrending item processing is performed on the measured acceleration signal data of the elevator door switch; Liye inverse transformation, using the energy threshold algorithm to process the transformed time domain signal to filter out the jitter interference signal at the moment of elevator opening and closing and the gap between opening and closing doors; finally, integrate the processed acceleration signal to obtain the real-time speed signal.

The present invention mainly comprises the steps:

Step 1, obtain the acceleration signal of the original elevator door operation through the acceleration sensor;

1-1. Place the acceleration sensor horizontally on one side of the elevator door;

1-2. Manually control the switch of the elevator door, and record the acceleration signal data when the elevator door is opened and closed every time through the acceleration sensor.

Step 2, performing detrending item processing on the obtained acceleration signal;

2-1. For the acquired acceleration signal data, calculate its trend item;

The sampling data of the elevator door acceleration signal obtained by actual measurement is {x _k } (k=1,2,3,...,n), n is the length of the sampling data, for the convenience of calculation, the length of the sampling data is extended to N, And let N=2 ^L , L is the smallest integer that makes N≥n, let the sampling time interval Δt=1, set a polynomial function:

determine the polynomial function The undetermined coefficients a _j (j=0,1,...,m) of each such that the polynomial function The sum of squared errors E with the sampled data x _k is the smallest, that is

The condition that satisfies the extremum value of E is:

Take E in turn to find the partial derivative of a _i , and generate an m+1 element linear equation system:

Solve the equation system to obtain m+1 undetermined coefficients a _j (j=0,1,...,m). In the above formulas, j is the set polynomial order, and its value range is 0≤j≤m.

2-2. Obtain the acceleration signal after the linear trend item is eliminated.

The calculation formula for eliminating the linear trend item is:

When m≥2, it is the trend item of the curve. In actual elevator door acceleration signal processing, m=1, 2, 3 are usually taken. After processing the sampled data with polynomial trend item elimination, y _k is obtained.

Step 3, convert to frequency domain by fast Fourier transform, filter and denoise;

3-1. Discrete Fourier transform (DFT);

Since the actual sampled signal is discrete and the sample length N of the sampled signal in time T is limited, the {y _k } (k=1,2,3,...,n) processed by the detrending term is taken as N points Sequence y(r)(r=0,1,2,...N-1), needs to adopt the discrete algorithm of Fourier transform when carrying out Fourier transform to elevator door acceleration signal, discrete Fourier transform ( DFT) is expressed as:

In the formula: Y(k1) is equivalent to Y(k1·Δf), sampling frequency

3-2. Fast Fourier Transform (FFT);

3-2-1. Due to the limitation of the length of the sampling signal and the calculation cost, the signal y(r) of the detrending item is processed by fast Fourier transform:

The N-point discrete Fourier transform of the N-point sequence y(r) can be expressed as:

Among them, W＝e ^-j2π/N

Using the periodicity of the Fourier variation coefficient W ^(k1)r , that is

W ^(k1)r = W ^k1(r+N) = W ^(k1+N)r

Taking advantage of its symmetry, that is

W ^(k1)r+N/2 = -W ^(k1)r

3-2-2. According to its periodicity, the discrete Fourier transform of the long sequence can be decomposed into the discrete Fourier transform of the short sequence.

The sequence y(r) (r=0,1,2,...,N-1) processed by the detrending item with a sample length of N=2 ^L is first divided into two groups according to the parity of r:

ask for it separately The discrete Fourier transform of the point, the first half is obtained as:

The second half is:

3-2-3. Repeat step 3-2-2 to obtain the FFT transformation result d(r) of y(r) (r=0,1,2,...N-1).

3-3. Perform frequency domain filtering on the amplitude-frequency signal.

Using a finite impulse response FIR digital filter, the difference equation of the FIR filter can be expressed as:

In the formula: d(n1) and p(n1) are the input time-domain signal sequence after fast Fourier transform and the output frequency-domain signal sequence after frequency-domain filtering; b _k3 is the filter coefficient, n1≥0, k3= 0,1,2,...N-1.

The z transformation of the impulse response function h(n) of the FIR filter is the system transfer function, which can be expressed as:

Then its impulse response function is:

Step 4, performing an inverse Fourier transform on the filtered frequency domain signal to calculate the energy of the acceleration signal;

4-1. Inverse Fourier transform of frequency domain signals;

In the formula: f(r) is equivalent to f(rΔt), sampling time interval Δt=1, r=0,1,2...N-1.

4-2. Normalize the acceleration signal;

First, normalize the acceleration signal whose amplitude is less than the set threshold A, and then filter out the jitter interference noise after normalization processing; generally, the amplitude of the jitter interference noise is smaller than the normalization coefficient C, while other The useful signal is greater than C. Therefore, after the acceleration signal is divided by the preset normalization coefficient, further processing can be effectively performed to distinguish the interference signal from the useful signal.

Among them, f(t)=f(r·Δt).

The normalized acceleration signal g(t) is squared to obtain its energy waveform, which makes the difference between the jitter signal and the useful signal larger, and it is more conducive to setting the threshold B to remove most of the jitter signals.

4-3. Energy calculation of acceleration signal.

The energy E1 of the acceleration signal f(t) is defined as:

Square the magnitude of the signal after the inverse Fourier transform.

Step 5, set the threshold for the running stage of the elevator, and remove the jitter noise signal;

5-1. Distinguish low-amplitude jitter and high-amplitude useful signal operating segments, determine the signal operating segment, and minimize the loss of useful signals. Set an accurate threshold B for the squared waveform and make a judgment. The waveform greater than the threshold is retained, and the waveform smaller than the threshold is reset to zero:

And named the judged waveform as energy signal amplitude u(t);

5-2. Multiply the amplitude of the energy signal by the normalization coefficient to obtain the final acceleration signal q(t) as follows:

q(t)=u(t)·C

Step 6. Integrate the final acceleration signal q(t) to obtain the corresponding real-time speed signal r(k5):

The sampling data of the final acceleration signal q(t) is {q _k5 }(k5=1,2,3,...,n), the sampling time step Δt is taken as the integration step in the numerical integration, and the trapezoidal numerical quadrature formula for:

The beneficial effects of the present invention are as follows:

The present invention processes the collected elevator door acceleration signals through a series of signal processing algorithms, and intends to obtain relatively accurate elevator door running acceleration and speed values, so as to know whether the elevator is running safely, provide accurate judgment of elevator operation, and reduce accidents. The probability of occurrence, improving the efficiency of monitoring and operation and maintenance, and reducing labor costs provide guarantees.

The present invention preprocesses the obtained complex original signal by detrending items, fast Fourier transforms it into the frequency domain and filters it, then inverse Fourier transforms it into the time domain, determines the running section according to the energy of the signal, sets the threshold, and removes the difficulty Problem - The jitter interference signal at the moment of opening and closing of the elevator door and the gap between the opening and closing of the door can also ensure that the loss of useful signals is reduced, and the speed signal can be obtained by integrating.

The present invention is for processing and judging the field measured signals, and has strong anti-interference ability in complex situations, so it will have better adaptability and accuracy for the acceleration signal processing under the ideal state, and the accurate speed signal will have a greater impact on the elevator door. Normal operation, maintenance, safety monitoring provide great help.

Description of drawings

Fig. 1 is the original signal diagram of elevator door acceleration;

Fig. 2 is the figure after the original signal of the elevator door acceleration is removed from the trend item;

Fig. 3 is the frequency domain diagram after the elevator door acceleration signal is detrended;

Fig. 4 is the figure after frequency domain filtering after the elevator door acceleration signal goes trend item;

Fig. 5 is converted into a time-domain diagram after frequency-domain filtering of the elevator door acceleration signal;

Fig. 6 is the figure after the energy threshold algorithm is passed through the time domain of the elevator door acceleration signal;

Fig. 7 is a speed diagram obtained by time-domain integration of the elevator door acceleration signal.

Detailed ways

The present invention will be further described below in conjunction with accompanying drawing and example.

As shown in Figures 1-7, the present invention uses a time-domain detrending term, fast Fourier transform to frequency-domain filtering, and inverse Fourier transform to time-domain and takes energy by processing the measured elevator door acceleration signal. Threshold algorithm denoises, integrates to obtain accurate speed signal algorithm, and proposes a real-time elevator door speed estimation method based on energy threshold algorithm.

The present invention firstly obtains the original signal of the elevator door acceleration in disorder and disorder in Fig. 1 through on-site measurement. Figure 2 is the time-domain diagram of the original acceleration signal after the detrending term processing. The image after the detrending term processing is shifted to the horizontal line as a whole, which means that a large amount of DC component noise is eliminated; Lie transform into frequency domain, analyze its spectral characteristics, and set filter parameters; Figure 4 is the figure after setting the passband 8HZ and stopband 35HZ frequency domain filtering for Figure 3 by using FIR and other ripple filtering methods; Figure 5 is the figure after the frequency domain filtering of Figure 4 The time-domain diagram of the inverse Fourier transform; Figure 6 uses the energy threshold algorithm for Figure 5, and performs zero-return processing on the static state of the elevator door in the opening and closing of the door and the small-amplitude jitter signal that exists when the door is closed. Accurate acceleration signal diagram; Figure 7 is the speed signal diagram obtained after time domain integration of the acceleration signal diagram.

The present invention processes the collected elevator door acceleration signals through a series of signal processing algorithms, and intends to obtain relatively accurate elevator door running acceleration and speed values, which are used as important indicators in elevator normal operation, safety monitoring and maintenance.

The specific implementation method is as follows:

Step 1. Obtain the original elevator door running acceleration signal through the acceleration sensor

1-1. Place the acceleration sensor horizontally on one side of the elevator door;

1-2. Manually control the switch of the elevator door, and record the data of the acceleration signal when the elevator door is opened and closed each time through the acceleration sensor, as shown in Figure 1.

Step 2. Perform detrending item processing on the obtained acceleration signal

2-1. Obtain the original acceleration signal data and calculate the trend item;

The sampling data of the actual measured elevator door acceleration signal is {x _k } (k=1,2,3,...,n), n is the length of the sampling data, for the convenience of calculation, the length of the sampling data is extended to N (make N =2 ^L , L is an integer and N≥n), let the sampling time interval Δt=1, set a polynomial function:

determine function The undetermined coefficients a _j (j=0,1,...,m) of each, so that the function The sum of squared errors E with the discrete data x _k is the smallest, that is

The condition that satisfies the extremum value of E is:

Taking E in turn to calculate the partial derivative of a _i can generate a system of linear equations with m+1 elements:

Solve the equation system to obtain m+1 undetermined coefficients a _j (j=0,1,...,m). In the above formulas, j is the set polynomial order, and its value range is 0≤j≤m.

2-2. Obtain the acceleration signal that eliminates the linear trend item.

The calculation formula for eliminating the linear trend item is:

When m≥2, it is the trend item of the curve, usually m=1, 2, 3. After processing the sampled data with polynomial trend item elimination, y _k is obtained.

Step 3. Convert to frequency domain by fast Fourier transform, filter and denoise

3-1. Discrete Fourier transform (DFT);

Treat {y _k }(k=1,2,3,...,n) after detrending item processing as N-point sequence y(r)(r=0,1,2,...N-1) , when performing Fourier transform on the elevator door acceleration signal, the discrete algorithm of Fourier transform needs to be adopted, and the expression of discrete Fourier transform (DFT) is:

In the formula: Y(k1) is equivalent to Y(k1·Δf), sampling frequency

3-2. Fast Fourier Transform (FFT);

3-2-1. This patent uses fast Fourier transform to process the detrending signal:

The N-point discrete Fourier transform of the N-point sequence y(r) can be expressed as:

Among them, W＝e ^-j2π/N

Using the periodicity of the Fourier variation coefficient W ^(k1)r , that is

W ^(k1)r = W ^k1(r+N) = W ^(k1+N)r

Taking advantage of its symmetry, that is

W ^(k1)r+N/2 = -W ^(k1)r

3-2-2. According to its periodicity, the discrete Fourier transform of the long sequence can be decomposed into the discrete Fourier transform of the short sequence.

The sequence y(r) (r=0,1,2,...,N-1) processed by the detrending item with a sample length of N=2 ^L is first divided into two groups according to the parity of r:

ask for it separately The discrete Fourier transform of the point, the first half is obtained as:

The second half is:

3-2-3. Repeat step 3-2-2 to obtain the FFT transformation result d(r) of y(r) (r=0,1,2,...N-1).

3-3. Perform frequency domain filtering on the amplitude-frequency signal.

Using a finite impulse response FIR digital filter, the difference equation of the FIR filter is:

In the formula: d(n1) and p(n1) are the input time-domain signal sequence after fast Fourier transform and the output frequency-domain signal sequence after frequency-domain filtering; b _k3 is the filter coefficient, n1≥0, k3= 0,1,2,...N-1.

The z transformation of the impulse response function h(n) of the FIR filter is the system transfer function:

Then its impulse response function is:

Step 4. Perform inverse Fourier transform on the filtered frequency domain signal to calculate the energy of the acceleration signal

4-1. Inverse Fourier transform of frequency domain signals;

In the formula: f(r) is equivalent to f(rΔt), sampling time interval Δt=1, r=0,1,2...N-1.

4-2. Normalize the acceleration signal;

First, normalize the acceleration signal whose amplitude is less than the set threshold A, and then filter out the jitter interference noise after normalization processing; in general, the amplitude of the jitter interference noise is smaller than the normalization coefficient C, while other The useful signal is greater than C. Therefore, after the acceleration signal is divided by the preset normalization coefficient, further processing can be effectively performed to distinguish the interference signal from the useful signal.

Among them, f(t)=f(r·Δt).

The normalized acceleration signal g(t) is squared to obtain its energy waveform, which makes the difference between the jitter signal and the useful signal larger, and it is more conducive to setting the threshold B to remove most of the jitter signals.

4-3. Energy calculation of acceleration signal.

The energy E1 of the acceleration signal f(t) is defined as:

Square the magnitude of the signal after the inverse Fourier transform.

Step 5. Determine the operating stage of the elevator and set the threshold to remove the jitter noise signal

5-1. Distinguish low-amplitude jitter and high-amplitude useful signal operating segments, determine the signal operating segment, and minimize the loss of useful signals. Set an accurate threshold B according to the waveform, keep the waveforms greater than the threshold, and reset the waveforms smaller than the threshold to zero:

And named the judged waveform as energy signal amplitude u(t);

5-2. Multiply the amplitude of the energy signal by the normalization coefficient to obtain the final acceleration signal q(t) as follows:

q(t)=u(t)·C

Step 6. Integrate the final acceleration signal q(t) to obtain a corresponding real-time speed signal r(k5).

6-1. The sampling data of the final acceleration signal q(t) is {q _k5 }(k5=1,2,3,...,n), and the sampling time step Δt is taken as the integration step in the numerical integration, trapezoidal The numerical quadrature formula is:

Note the following in steps 4 and 5:

The setting of the normalization coefficient C and the threshold B needs to be judged according to the amplitude of the jitter waveform, selected according to the actual situation, and adjusted with the feedback of the filtering effect.

Claims (5)

1. based on the elevator door real-time running speed estimation method of energy threshold algorithm, it is characterized in that comprising the steps: Step 1, obtain the acceleration signal of the original elevator door operation through the acceleration sensor; 1-1. Place the acceleration sensor horizontally on one side of the elevator door; 1-2. Manually control the switch of the elevator door, and record the acceleration signal data when the elevator door is opened and closed every time through the acceleration sensor; Step 2, performing detrending item processing on the obtained acceleration signal; Step 3, convert to frequency domain by fast Fourier transform, filter and denoise; Step 4, performing an inverse Fourier transform on the filtered frequency domain signal to calculate the energy of the acceleration signal; Step 5, set the threshold for the running stage of the elevator, and remove the jitter noise signal; Step 6. Integrate the final acceleration signal q(t) to obtain a corresponding real-time velocity signal. 2. the elevator door real-time running speed estimation method based on energy threshold algorithm according to claim 1, is characterized in that step 2 is specifically realized as follows: 2-1. For the acquired acceleration signal data, calculate its trend item; The sampling data of the elevator door acceleration signal obtained by actual measurement is {x _k } (k=1,2,3,...,n), n is the length of the sampling data, for the convenience of calculation, the length of the sampling data is extended to N, And let N=2 ^L , L is the smallest integer that makes N≥n, let the sampling time interval Δt=1, set a polynomial function: determine the polynomial function The undetermined coefficients a _j (j=0,1,...,m) of each such that the polynomial function The sum of squared errors E with the sampled data x _k is the smallest, that is The condition that satisfies the extremum value of E is: Take E in turn to find the partial derivative of a _i , and generate an m+1 element linear equation system: Solve the equation system to find m+1 undetermined coefficients a _j (j=0,1,...,m); in the above formulas, j is the set polynomial order, and its value range is 0≤j≤ m; 2-2. Obtain the acceleration signal after eliminating the linear trend item; The calculation formula for eliminating the linear trend item is: When m≥2, it is the trend item of the curve; in the actual elevator door acceleration signal processing, m=1, 2, 3 is taken, and y _k is obtained after the polynomial trend item elimination process is performed on the sampled data. 3. the elevator door real-time running speed estimation method based on energy threshold algorithm according to claim 2, is characterized in that step 3 is specifically realized as follows: 3-1. Discrete Fourier transform; Since the actual sampled signal is discrete and the sample length N of the sampled signal in time T is limited, the {y _k } (k=1,2,3,...,n) processed by the detrending term is taken as N points Sequence y(r)(r=0,1,2,...N-1), needs to adopt the discrete algorithm of Fourier transform when carrying out Fourier transform to elevator door acceleration signal, discrete Fourier transform ( DFT) is expressed as: In the formula: Y(k1) is equivalent to Y(k1·Δf), sampling frequency (k1,r=0,1,2,...N-1); 3-2. Fast Fourier Transform (FFT); 3-2-1. Due to the limitation of the length of the sampling signal and the calculation cost, the signal y(r) of the detrending item is processed by fast Fourier transform: The N-point discrete Fourier transform of the N-point sequence y(r) can be expressed as: Among them, W＝e ^-j2π/N Using the periodicity of the Fourier variation coefficient W ^(k1)r , that is W ^(k1)r = W ^k1(r+N) = W ^(k1+N)r Taking advantage of its symmetry, that is W ^(k1)r+N/2 = -W ^(k1)r 3-2-2. According to its periodicity, the discrete Fourier transform of the long sequence can be decomposed into the discrete Fourier transform of the short sequence; The sequence y(r) (r=0,1,2,...,N-1) processed by the detrending item with a sample length of N=2 ^L is first divided into two groups according to the parity of r: ask for it separately The discrete Fourier transform of the point, the first half is obtained as: The second half is: 3-2-3. Repeat step 3-2-2 to obtain the FFT transformation result d(r) of y(r) (r=0,1,2,...N-1); 3-3. Perform frequency domain filtering on the amplitude-frequency signal; Using a finite impulse response FIR digital filter, the difference equation of the FIR filter can be expressed as: In the formula: d(n1) and p(n1) are the input time-domain signal sequence after fast Fourier transform and the output frequency-domain signal sequence after frequency-domain filtering; b _k3 is the filter coefficient, n1≥0, k3= 0,1,2,...N-1; The z transformation of the impulse response function h(n) of the FIR filter is the system transfer function, which can be expressed as: Then its impulse response function is: 4. the elevator door real-time running speed estimation method based on energy threshold algorithm according to claim 3, is characterized in that step 4 is specifically realized as follows: Step 4, performing an inverse Fourier transform on the filtered frequency domain signal to calculate the energy of the acceleration signal; 4-1. Inverse Fourier transform of frequency domain signal; In the formula: f(r) is equivalent to f(rΔt), sampling time interval Δt=1, r=0,1,2...N-1; 4-2. Normalize the acceleration signal; First, normalize the acceleration signal whose amplitude is less than the set threshold A, and then filter out the jitter interference noise after normalization processing; in general, the amplitude of the jitter interference noise is smaller than the normalization coefficient C, while other The useful signal is greater than C; therefore, after dividing the acceleration signal by the preset normalization coefficient, further processing is effective to distinguish the interference signal from the useful signal; Among them, f(t)=f(r·Δt); The normalized acceleration signal g(t) is squared to obtain its energy waveform, which is conducive to setting the threshold B to remove most of the jitter signals; 4-3. Energy calculation of acceleration signal; The energy E1 of the acceleration signal f(t) is defined as: Square the magnitude of the signal after the inverse Fourier transform. 5. the elevator door real-time running speed estimation method based on energy threshold algorithm according to claim 4, is characterized in that step 5 is specifically realized as follows: Step 5, set the threshold for the running stage of the elevator, and remove the jitter noise signal; 5-1. Distinguish low-amplitude jitter and high-amplitude useful signal operating segments, determine the signal operating segment, and minimize the loss of useful signals; set an accurate threshold B for the squared waveform , and make a judgment, keep the waveforms greater than the threshold, and reset the waveforms smaller than the threshold to zero: And named the judged waveform as energy signal amplitude u(t); 5-2. Multiply the amplitude of the energy signal by the normalization coefficient to obtain the final acceleration signal q(t) as follows: q(t)=u(t)·C Step 6. Integrate the final acceleration signal q(t) to obtain the corresponding real-time speed signal r(k5): The sampling data of the final acceleration signal q(t) is {q _k5 }(k5=1,2,3,...,n), the sampling time step Δt is taken as the integration step in the numerical integration, and the trapezoidal numerical quadrature formula for: