CN113934971A - Long-distance sensing method for underground rescue robot based on LoRa - Google Patents
Long-distance sensing method for underground rescue robot based on LoRa Download PDFInfo
- Publication number
- CN113934971A CN113934971A CN202111204000.XA CN202111204000A CN113934971A CN 113934971 A CN113934971 A CN 113934971A CN 202111204000 A CN202111204000 A CN 202111204000A CN 113934971 A CN113934971 A CN 113934971A
- Authority
- CN
- China
- Prior art keywords
- signal
- robot
- lora
- signals
- human body
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
Abstract
A long-distance sensing method of an underground rescue robot based on LoRa is characterized in that the robot carrying a receiver moves along an underground tunnel, a transmitter is fixed in the underground tunnel, two antennas arranged on the receiver respectively receive signals, and the influence of phase deviation is eliminated by utilizing signal quotient operation; in order to identify the human body activity signal under the interference of the robot moving signal, the signal is subjected to standardized processing by adopting empirical mode decomposition, and the signal received by a receiver is processed while the human body induction signal change mode is kept; in order to better distinguish the robot moving signal and the human body activity signal, a periodic characteristic is defined to quantize the periodic level of the signal, and meanwhile, a fast Fourier transform ratio criterion is adopted as a frequency characteristic to quantize the frequency level of the signal; and finally, the human body activity signals are judged by using a support vector machine classification method, so that the method can accurately sense survivors in the well after the disaster, expand the sensing range and improve the rescue rate.
Description
Technical Field
The invention relates to an underground rescue method, in particular to an underground rescue robot long-distance sensing method based on LoRa, and belongs to the technical field of underground intelligent sensing.
Background
Coal mine safety is an important guarantee for the development of the coal industry, however, frequent disaster accidents not only cause great loss to lives and properties of people, but also bring adverse effects to social stability. Thousands of miners are trapped in a mine every year in the world, and the mine rescue robot can effectively reduce the risk of rescuers as intelligent search and rescue equipment capable of acquiring and transmitting disaster environment parameters and trapped miner information, and becomes a research hotspot in recent years. However, the underground coal mine roadway is long and narrow, and the underground environment after the disaster is more complex and the space is limited, which puts higher requirements on the life perception research of the mine rescue robot. Various downhole communication technologies have been proposed, such as WIFI, bluetooth, and radio frequency signals, but such schemes have limited perceptual performance. On one hand, the severe underground environment may cause severe attenuation of wireless signals, limiting the sensing range; on the other hand, such methods need to analyze the reflected signal from the target to analyze the activity information, however, as the propagation path becomes longer, the signal is more attenuated when being reflected to the receiving end than when being directly propagated to the receiving end, resulting in poor perception function.
In recent years, a technology based on remote radio (loRa) sensing appears in the field of wireless sensing, the sensing range can be further widened, and the loRa has the characteristics of low power consumption, long communication distance, strong anti-interference capability, strong penetration capability and the like, and has natural advantages in long-distance through-wall wireless sensing. Personnel state perception is realized through utilizing the parameter variation of loRa communication signal to realize loRa's communication function and perception function organic combination. Therefore, the research on the long-distance underground sensing by utilizing the LoRa has a good application prospect, the development of the mine rescue robot sensing technology in China can be powerfully promoted, however, most of the sensing technologies based on the LoRa still need to analyze reflected signals, and the instability of sensing is increased.
Disclosure of Invention
The invention aims to provide a long-distance sensing method of an underground rescue robot based on LoRa, which can accurately sense survivors in the underground after a disaster, enlarge the sensing range and improve the rescue rate.
In order to achieve the purpose, the invention provides a long-distance sensing method of an underground rescue robot based on LoRa, which comprises the following steps:
step 1: the robot carrying the receiver moves along the underground tunnel and is used for collecting an LoRa signal transmitted by a transmitter, wherein the LoRa signal is mixed with a robot moving signal and a human body activity signal, and the transmitter is fixedly arranged in the underground tunnel;
step 2: because signals of a pair of receivers and a pair of transmitters are asynchronous, two antennas are arranged on the receivers and respectively receive LoRa signals, the influence of phase offset on the LoRa signals is eliminated by utilizing signal quotient operation, and abnormal values in the LoRa signals are removed by adopting a filter to realize signal smoothing;
and step 3: in order to identify the human body activity signal under the interference of the robot movement signal, the LoRa signal is subjected to standardization processing by adopting empirical mode decomposition, and the LoRa signal received by the receiver is processed while the change mode of the human body activity signal is kept;
and 4, step 4: in order to better distinguish robot moving signals and human body activity signals, two types of feature extraction algorithms are provided, the periodic level of the signals is quantized by defining periodic features, and meanwhile, the frequency level of the signals is quantized by taking a fast Fourier transform ratio criterion as a frequency feature;
and 5: and finally, judging the human activity signals by using a support vector machine classification method, and giving a detection alarm to realize the perception of survivors.
The transmitter of the invention is provided with an LoRa node and a 9dbi directional antenna; the robot carries all mobile sensing devices, a portable power supply provides power to work, and the model of an antenna is USRPX 310; the carrier frequency of signals transmitted by the LoRa node is 915MHz, the linear frequency modulation bandwidth BW is 125KHz, the spreading factor SF is 12, the coding rate CR is 4/8, and the sampling rate is 500 Hz.
The filter model is Savitzky-Golay.
The signal quotient operation in the step 2 specifically comprises:
Rratio＝R1/R2
in the formula: r1Is the signal received by the antenna 1;
R2is the signal received by the antenna 2;
Rratiothe quotient of the signals is represented and is the quotient operation result of the antenna 1 and the antenna 2;
in step 3, in an ideal situation, the robot movement generates a sine-wave-like signal and fluctuates irregularly, which may be due to the influence of receiver inclination, robot motion instability and multipath effect, and the signal induced by the robot movement is less obvious than the sharp and significant signal induced by human body movement. In addition, EMD also avoids the problem of the need to pre-set basis functions for wavelet transforms. In summary, the signal mobility reduction mainly includes two steps of Empirical Mode Decomposition (EMD) processing and Intrinsic Mode Function (IMF) selection;
empirical Mode Decomposition (EMD) processing: empirical Mode Decomposition (EMD) decomposes a signal into a series of components called Intrinsic Mode Functions (IMF) based on local time scale characteristics of data, each Intrinsic Mode Function (IMF) must satisfy two constraints, first, the number of extreme points and the number of zero crossings are extremely the same for the whole data segment, or the difference does not exceed 1; secondly, at any moment, the mean values of an upper envelope formed by a local maximum value and a lower envelope formed by a local minimum value are zero, namely the upper envelope and the lower envelope are locally symmetrical relative to a time axis, Empirical Mode Decomposition (EMD) is an iteration process, after N rounds of Intrinsic Mode Functions (IMF) are extracted and iterated, when the residual error of a component is a monotonic function, the residual error at the moment is stored for the selection of the next Intrinsic Mode Function (IMF), and the iteration is finished;
intrinsic Mode Function (IMF) selection: selecting a median Intrinsic Mode Function (IMF) component from all Intrinsic Mode Function (IMF) components, wherein the component can distinguish signals generated by robot movement and signals generated by human body movement, the Intrinsic Mode Function (IMF) component can embody the physical significance of the signals, among a plurality of Intrinsic Mode Function (IMF) components, the first Intrinsic Mode Function (IMF) component generally carries high-frequency oscillation information, high-frequency components are removed by turns, and finally, a plurality of Intrinsic Mode Function (IMF) components are blurred in a frequency domain and obviously unsuitable for judging whether a signal section belongs to a human body movement signal, therefore, selecting the median IMF component for subsequent processing, wherein the component can not only standardize irregular data, but also keep better frequency characteristics for two types of signals, particularly, residual error is removed, when the number N of the Intrinsic Mode Function (IMF) is an odd number, the selected median IMF is (N-1)/2 IMF, and when the number N of the Intrinsic Mode Functions (IMF) is even, the selected median Intrinsic Mode Function (IMF) is the N/2 Intrinsic Mode Function (IMF).
The feature extraction algorithm in step 4 is as follows: after extracting proper Intrinsic Mode Functions (IMF), defining periodic characteristics and frequency characteristics as characteristics to distinguish signal change modes generated by two types of activities,
periodic characteristics: through experimental tests, the signal generated by the movement of the robot is found to be a sine wave, the amplitude of the signal induced by the human body activity is in a state of fluctuating up and down, because the sudden movement of the human limbs generates an accidental reflection path, according to the difference of the signal generation principle, two characteristics of an upper envelope variation mode and sine wave similarity are introduced to quantize the periodic characteristics of two groups of signals, firstly, an upper envelope variation mode is introduced, the waveform is symmetrical under the constraint of an Intrinsic Mode Function (IMF), therefore, the average envelope curves of the signal generated by the movement of the robot and the signal generated by the human body activity are almost the same, which is probably an x axis, however, only the upper envelope or the lower envelope can reflect the different variation modes between the two signals, the upper envelope surface of the movement signal of the robot is flatter, and the upper envelope surface of the signal generated by the human body activity is more curved, therefore, the period level is expressed by the variance of the upper envelope, the smaller the variance, the higher the period level, and another characteristic characterizing the period level is the similarity between the test signal and the sine wave, the sine wave signal d (t) is modeled, and the similarity calculation is performed according to equation (1),
d(t)＝Asin(2πft+φ) (1)
in the formula: a is the signal amplitude;
f is the frequency;
phi is an initial phase;
t is a time variable;
according to the selected median eigenmode function cn(t) fitting the parameters A, f and φ, the median eigenmode function cnAll peaks of (t) are denoted as p (x)i)i∈[1，L]Wherein x isiSample value corresponding to ith peak, L is peak number, intrinsic mode function component is symmetrical about x axis, average value of all peak values is fitting value of A, given sampling rate fsAnd f is defined as:
to simplify the fitting process of phi, the value of phi is directly determined as pi/2, and then the Intrinsic Mode Function (IMF) component from the first peak to the last peak is intercepted for similarity calculation, and is marked as
The similarity SL can be expressed as the ratio of a to the root mean square error:
in the formula: m is the length of the intercepted signal;
t' is a certain time;
for a robot moving signal, an originally selected Intrinsic Mode Function (IMF) waveform and a waveform of a modeled sine wave are highly overlapped, for a human body moving signal, the two waveforms are crossed, the similarity SL is used as a characteristic for distinguishing the two types of signals, the larger the similarity SL is, the better the periodicity is, and then, PL is defined as the difference of the similarity SL minus an upper envelope, the PL can represent a uniform characteristic of the periodic characteristic, and the larger the PL is, the better the periodicity is;
frequency characteristics: defining a standard FL to represent the definition of frequency distribution, testing FFT results of Intrinsic Mode Function (IMF) components of a robot moving signal and a human body activity signal, wherein the results show that the peak value of the robot moving signal is more clear, and the corresponding frequency is dominant in a frequency domain, while the human body activity signal has a plurality of spread spectrum parts, which are mainly caused by random motion of a plurality of parts of a body, and providing a new measurement method to quantify the definition of the frequency, as shown in a formula (4):
in the formula: e1Represents the 1 st energy peak;
E2represents the 2 nd energy peak;
E3represents the 3 rd energy peak;
the larger the FL, the "sharper" the frequency distribution, and the more likely the signal is to belong to a robot movement signal.
And 5, judging the human body activity signals by a support vector machine classification method, specifically, taking the periodic characteristics and the frequency characteristics as the input of the support vector machine, dividing two types of sample points on two sides of the hyperplane by searching a maximum interval hyperplane, and maximizing the distance from the sample point closest to the hyperplane on the two sides to the hyperplane.
Compared with the prior art, the robot is provided with the receiver, the receiver moves along an underground roadway along with the movement of the robot and is used for collecting the LoRa signal which is transmitted by the transmitter and contains environment and human body activity information, and the transmitter is fixedly arranged in the underground roadway; because the receiving and transmitting pairs are asynchronous, two antennas are arranged on the receiver and respectively receive signals, the influence of phase offset is eliminated by utilizing signal quotient operation, and then an abnormal value is removed by adopting a filter to realize signal smoothing; in order to identify human activity signals under the interference of robot moving signals, EMD is adopted to carry out standardized processing on the signals, and the signals received by a receiver are processed while a human body induction signal change mode is kept; in order to better distinguish signals generated by robot movement and signals generated by human body activity, two types of feature extraction algorithms are provided, the periodic level of the signals is quantized by defining periodic features, and meanwhile, the frequency level of the signals is quantized by taking a fast Fourier transform ratio criterion as a frequency feature; and finally, judging the human activity signals by using a SVM classification method, and giving a detection alarm to realize the perception of survivors. By introducing empirical mode decomposition and periodic frequency characteristics, the invention solves the unique challenge caused by robot movement, namely the interference of the robot can confuse human action signals.
Drawings
FIG. 1 is a diagram of a motion-aware scenario of the present invention;
FIG. 2 is a schematic diagram of the present invention;
FIG. 3 is a diagram of the classification results of the present invention.
In the figure: 1. robot, 2, underworkings, 3, transmitter.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
As shown in fig. 1 and 2, a long-distance sensing method for an underground rescue robot based on LoRa includes the following steps:
step 1: the robot 1 carrying the receiver moves along the underground tunnel 2 and is used for collecting an LoRa signal transmitted by a transmitter 3, wherein the LoRa signal is mixed with the robot moving signal and a human body movement signal, the transmitter 3 is fixedly arranged in the underground tunnel 2, and the transmitter can be arranged at any position which does not influence the movement of the tunnel, such as the top wall above the tunnel in a hanging and fixing way;
step 2: because signals of a pair of receivers and a pair of transmitters are asynchronous, two antennas are arranged on the receivers and respectively receive LoRa signals, the influence of phase offset on the LoRa signals is eliminated by utilizing signal quotient operation, and abnormal values in the LoRa signals are removed by adopting a filter to realize signal smoothing;
and step 3: in order to identify the human body activity signal under the interference of the robot 1 moving signal, the LoRa signal is subjected to standardization processing by adopting empirical mode decomposition, and the LoRa signal received by the receiver is processed while the human body activity signal change mode is kept;
and 4, step 4: in order to better distinguish robot moving signals and human body activity signals, two types of feature extraction algorithms are provided, the periodic level of the signals is quantized by defining periodic features, and meanwhile, the frequency level of the signals is quantized by taking a fast Fourier transform ratio criterion as a frequency feature;
and 5: and finally, judging the human activity signals by using a support vector machine classification method, and giving a detection alarm to realize the perception of survivors.
The emitter 3 is provided with an LoRa node and a 9dbi directional antenna; the robot 1 carries all mobile sensing equipment, a portable power supply provides power to work, and the model of an antenna is USRPX 310; the carrier frequency of signals transmitted by the LoRa node is 915MHz, the linear frequency modulation bandwidth BW is 125KHz, the spreading factor SF is 12, the coding rate CR is 4/8, and the sampling rate is 500 Hz.
The filter model is Savitzky-Golay.
The signal quotient operation in the step 2 specifically comprises:
Rratio＝R1/R2
in the formula: r1Received for antenna 1A signal;
R2is the signal received by the antenna 2;
Rratiothe quotient of the signals is represented and is the quotient operation result of the antenna 1 and the antenna 2;
in step 3, in an ideal situation, the robot movement generates sine-wave-like signals and fluctuates irregularly, which may be due to the influence of receiver inclination, robot motion instability and multipath effect, and compared with sharp and significant signals induced by human body movement, the signals induced by the robot movement are less significant. Furthermore, Empirical Mode Decomposition (EMD) also avoids the problem that the wavelet transform needs to preset basis functions. In conclusion, the signal mobility reduction mainly comprises two steps of Empirical Mode Decomposition (EMD) processing and Intrinsic Mode Function (IMF) selection;
EMD treatment: based on the local time scale characteristics of data, an Empirical Mode Decomposition (EMD) decomposes a signal into a series of components called Intrinsic Mode Functions (IMFs), each IMF must satisfy two constraint conditions, firstly, for the whole data segment, the number of extreme points and the number of zero-crossing points are extremely the same, or the difference value is not more than 1; secondly, at any moment, the mean values of an upper envelope formed by a local maximum value and a lower envelope formed by a local minimum value are zero, namely the upper envelope and the lower envelope are locally symmetrical relative to a time axis, Empirical Mode Decomposition (EMD) is an iteration process, after N rounds of Intrinsic Mode Function (IMF) extraction and iteration, when the residual error of a component is a monotonic function, the residual error at the moment is stored for selection of the next Intrinsic Mode Function (IMF), and the iteration is finished;
selecting an intrinsic mode function IMF: selecting a median intrinsic mode function IMF component from all intrinsic mode function IMF components, wherein the component can distinguish signals generated by robot movement and signals generated by human body movement, the intrinsic mode function IMF component can embody the physical significance of the signals, among a plurality of intrinsic mode function IMF components, the first intrinsic mode function IMF component usually carries high-frequency oscillation information, high-frequency components are removed by turns, and finally, a plurality of intrinsic mode function IMF components are blurred in a frequency domain and obviously not suitable for judging whether a signal section belongs to a human body movement signal or not, so that the median IMF component is selected for subsequent processing, the component can normalize irregular data, better frequency characteristics are kept for two types of signals, particularly, residual errors are removed, when the number N of the intrinsic mode function IMF is an odd number, the selected median IMF is (N-1)/2 IMF, and when the number N of the intrinsic mode functions IMF is an even number, the selected median intrinsic mode function IMF is the N/2 th intrinsic mode function IMF.
The feature extraction algorithm in step 4 is as follows: after extracting proper intrinsic mode function IMF, defining periodic characteristic and frequency characteristic as characteristics to distinguish signal change modes generated by two types of activities,
periodic characteristics: through experimental tests, the signal generated by the movement of the robot is found to be a sine wave, the amplitude of the signal induced by the human body activity is in a state of fluctuating up and down, because the sudden movement of the human limbs generates an accidental reflection path, according to the difference of the signal generation principle, two characteristics of an upper envelope variation mode and sine wave similarity are introduced to quantify the periodic characteristics of two groups of signals, an upper envelope variation mode is firstly introduced, the waveform is symmetrical under the constraint of an intrinsic mode function IMF, therefore, the average envelope curves of the signal generated by the movement of the robot and the signal generated by the human body activity are almost the same, namely, the average envelope curve possibly is an x axis, however, only the upper envelope or the lower envelope curve can reflect the different variation modes between the two signals, the upper envelope surface of the movement signal of the robot is flatter, the upper envelope surface of the movement signal of the human body activity signal is more curved, therefore, the period level is expressed by the variance of the upper envelope, the smaller the variance, the higher the period level, another characteristic characterizing the period level is the similarity between the test signal and the sine wave, the sine wave signal d (t) is modeled, and the similarity calculation is performed according to equation (1),
d(t)＝Asin(2πft+φ) (1)
according to the selected median eigenmode function IMF cn(t) fitting the parameters A, f, and φ, the median eigenmode function IMF cnAll peaks of (t) are denoted as p (x)i)i∈[1，L]Wherein x isiSample value corresponding to ith peak, L is peak number, intrinsic mode function IMF component is symmetrical about x axis, average value of all peak values is fitting value of A, given sampling rate fsAnd f is defined as:
to simplify the fitting process of phi, the value of phi is directly determined as pi/2, and then the intrinsic mode function IMF component from the first peak to the last peak is intercepted for similarity calculation, and is marked as
The similarity SL can be expressed as the ratio of a to the root mean square error:
in the formula: m is the length of the intercepted signal;
t' is a certain time;
for a robot moving signal, the originally selected IMF waveform and the waveform of a modeled sine wave are highly overlapped, for a human body moving signal, the two waveforms are crossed, the similarity SL is used as a characteristic for distinguishing the two types of signals, the larger the similarity SL is, the better the periodicity is, and then, PL is defined as the difference of the similarity SL minus the upper envelope, the PL can represent the uniform characteristic of the periodicity level, and the larger the PL is, the better the periodicity is;
frequency characteristics: defining a standard FL to represent the definition of frequency distribution, testing FFT results of intrinsic mode function IMF components of a robot moving signal and a human body activity signal, wherein the results show that the peak value of the robot moving signal is more clear, and the corresponding frequency is dominant in a frequency domain, while the human body activity signal has a plurality of spread spectrum parts, which are mainly caused by random motion of a plurality of parts of a body, and providing a new measurement method to quantify the definition of the frequency, as shown in formula (4):
in the formula: e1Represents the 1 st energy peak;
E2represents the 2 nd energy peak;
E3represents the 3 rd energy peak;
the larger the FL, the "sharper" the frequency distribution, and the more likely the signal is to belong to a robot movement signal.
And 5, judging the human body activity signals by a support vector machine classification method, specifically, taking the periodic characteristics and the frequency characteristics as the input of the support vector machine, dividing two types of sample points on two sides of the hyperplane by searching a maximum interval hyperplane, and maximizing the distance from the sample point closest to the hyperplane on the two sides to the hyperplane. As shown in fig. 3, the hyperplane divides the sample points of the feature input into two categories, which are the signal generated by the robot movement and the signal generated by the human body activity.
Claims (6)
1. A long-distance sensing method for an underground rescue robot based on LoRa is characterized by comprising the following steps:
step 1: the robot (1) carrying the receiver moves along the underground roadway (2) and is used for collecting LoRa signals transmitted by the transmitter (3), and the transmitter (3) is fixedly installed inside the underground roadway (2);
step 2: installing two antennas on a receiver, respectively receiving the LoRa signals, eliminating the influence of phase offset on the LoRa signals by utilizing signal quotient operation, and then removing abnormal values in the LoRa signals by adopting a filter;
and step 3: identifying a human body activity signal under the interference of a robot (1) moving signal, carrying out standardization processing on an LoRa signal by adopting empirical mode decomposition, and processing the LoRa signal received by a receiver while keeping a human body activity signal change mode;
and 4, step 4: two types of feature extraction algorithms are provided to distinguish robot moving signals and human body activity signals, the periodic level of the signals is quantified by defining periodic features, and meanwhile, the frequency level of the signals is quantified by taking a fast Fourier transform ratio criterion as a frequency feature;
and 5: the method comprises the steps of judging human body activity signals through a support vector machine classification method, taking periodic characteristics and frequency characteristics as input of the support vector machine, dividing two types of sample points on two sides of a hyperplane by searching for a hyperplane with a maximum interval, maximizing the distance from the sample point on the two sides, which is closest to the hyperplane, and giving a detection alarm.
2. A long-distance sensing method for an underground rescue robot based on LoRa (Long distance Rake distance) according to claim 1, characterized in that the emitter (3) is provided with a LoRa node and a 9dbi directional antenna; the robot (1) carries all mobile sensing equipment, a portable power supply provides electric power to work, and the model of an antenna is USRPX 310; the carrier frequency of the signals transmitted by the LoRa node is 915MHz, the linear frequency modulation bandwidth is 125KHz, the spreading factor is 12, the coding rate is 4/8, and the sampling rate is 500 Hz.
3. The LoRa-based underground rescue robot long-distance sensing method is characterized in that the model of the filter is Savitzky-Golay.
4. The long-distance sensing method for the LoRa-based underground rescue robot is characterized in that the signal quotient operation in the step 2 is specifically as follows:
Rratio＝R1/R2
in the formula: r1Is the signal received by the antenna 1;
R2is the signal received by the antenna 2;
Rratiothe quotient is the quotient of antenna 1 and antenna 2.
5. The long-distance sensing method for the LoRa-based underground rescue robot as claimed in claim 3, wherein in the step 3, the empirical mode decomposition processing specifically comprises: based on the local time scale characteristics of data, empirical mode decomposition decomposes a signal into a series of components called intrinsic mode functions, each intrinsic mode function must satisfy two constraint conditions, and firstly, for the whole data segment, the number of extreme points is the same as that of zero-crossing points, or the difference is less than 1; secondly, at any moment, the mean values of an upper envelope formed by a local maximum value and a lower envelope formed by a local minimum value are zero, empirical mode decomposition is an iterative process, after N rounds of intrinsic mode function extraction and iteration, when the residual error of a component is a monotonic function, the residual error at the moment is stored for the selection of the next intrinsic mode function, and the iteration is finished;
selecting an intrinsic mode function: and selecting a median intrinsic mode function component from all the intrinsic mode function components, wherein the component distinguishes robot movement signals and human body movement signals, when the number N of the intrinsic mode functions is an odd number, the selected median IMF is (N-1)/2 IMF, and when the number N of the intrinsic mode functions is an even number, the selected median intrinsic mode function is the N/2 intrinsic mode function.
6. The long-distance sensing method for the LoRa-based underground rescue robot is characterized in that the feature extraction algorithm in the step 4 is as follows: after extracting proper eigenmode functions, defining periodic characteristics and frequency characteristics as characteristics to distinguish the robot and signal change modes generated by human body activities,
periodic characteristics: introducing an upper envelope variation mode and sine wave similarity to quantify the periodic characteristics of two groups of signals, firstly introducing the upper envelope variation mode, wherein the waveform is symmetrical under the constraint of an intrinsic mode function, so that the average envelope curve of a signal generated by the movement of a robot and a signal generated by the activity of a human body is the same and is an x axis, and the other characteristic for representing the periodic characteristics is the similarity between a test signal and a sine wave, modeling a sine wave signal d (t), and calculating the similarity according to the formula (1),
d(t)＝Asin(2πft+φ) (1)
in the formula: a is the signal amplitude;
f is the frequency;
phi is an initial phase;
t is a time variable;
according to the selected median eigenmode function cn(t) fitting the parameters A, f and φ, the median eigenmode function cnAll peaks of (t) are denoted as p (x)i)i∈[1,L]Wherein x isiSample value corresponding to ith peak, L is peak number, intrinsic mode function component is symmetrical about x axis, average value of all peak values is fitting value of A, given sampling rate fsAnd f is defined as:
the value of phi is directly determined as pi/2, and the intrinsic mode function component from the first peak to the last peak is intercepted for similarity calculation and is marked as
The similarity SL is expressed as the ratio of a to the root mean square error:
in the formula: m is the length of the intercepted signal;
t' is a certain time;
for a robot moving signal, the originally selected intrinsic mode function waveform and the waveform of the modeled sine wave are highly overlapped, and for a human body moving signal, the two waveforms are crossed, and the similarity SL is used as a characteristic for distinguishing the robot moving signal and the human body moving signal;
frequency characteristics: FL is "definition" of frequency distribution, and by testing FFT results of eigenmode function components of the robot movement signal and the human activity signal, it indicates that the peak value of the robot movement signal is more "definition", and its corresponding frequency is dominant in the frequency domain, while the human activity signal has a plurality of spread spectrum portions, and the quantization of the frequency distribution "definition" FL is as shown in equation (4):
in the formula: e1Represents the 1 st energy peak;
E2represents the 2 nd energy peak;
E3representing the 3 rd energy peak.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111204000.XA CN113934971B (en) | 2021-10-15 | 2021-10-15 | Long-distance sensing method for underground rescue robot based on LoRa |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111204000.XA CN113934971B (en) | 2021-10-15 | 2021-10-15 | Long-distance sensing method for underground rescue robot based on LoRa |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113934971A true CN113934971A (en) | 2022-01-14 |
| CN113934971B CN113934971B (en) | 2022-05-24 |
Family
ID=79279540
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111204000.XA Active CN113934971B (en) | 2021-10-15 | 2021-10-15 | Long-distance sensing method for underground rescue robot based on LoRa |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113934971B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090116595A1 (en) * | 2007-05-21 | 2009-05-07 | Florida State University | System and methods for determining masking signals for applying empirical mode decomposition (emd) and for demodulating intrinsic mode functions obtained from application of emd |
| WO2012055148A1 (en) * | 2010-10-27 | 2012-05-03 | 中国人民解放军第四军医大学 | Multichannel ultra wide band based (uwb-based) radar life detector and positioning method thereof |
| CN109283930A (en) * | 2018-10-16 | 2019-01-29 | 华南理工大学 | A kind of spherical intelligent inspection robot |
| CN110045348A (en) * | 2019-05-05 | 2019-07-23 | 应急管理部上海消防研究所 | A kind of human motion state classification method based on improvement convolutional neural networks |
| CN111600617A (en) * | 2020-05-12 | 2020-08-28 | 中国科学院软件研究所 | Non-contact sensing method based on LoRa signal of Internet of things |
| CN112132719A (en) * | 2020-08-11 | 2020-12-25 | 湖南大学 | Personnel search and rescue method, device and system for emergency rescue and storage medium |
| CN113163333A (en) * | 2021-05-27 | 2021-07-23 | 上海上晟电子科技有限公司 | LoRa technology-based field life-saving rescue backpack system and method |
-
2021
- 2021-10-15 CN CN202111204000.XA patent/CN113934971B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090116595A1 (en) * | 2007-05-21 | 2009-05-07 | Florida State University | System and methods for determining masking signals for applying empirical mode decomposition (emd) and for demodulating intrinsic mode functions obtained from application of emd |
| WO2012055148A1 (en) * | 2010-10-27 | 2012-05-03 | 中国人民解放军第四军医大学 | Multichannel ultra wide band based (uwb-based) radar life detector and positioning method thereof |
| CN109283930A (en) * | 2018-10-16 | 2019-01-29 | 华南理工大学 | A kind of spherical intelligent inspection robot |
| CN110045348A (en) * | 2019-05-05 | 2019-07-23 | 应急管理部上海消防研究所 | A kind of human motion state classification method based on improvement convolutional neural networks |
| CN111600617A (en) * | 2020-05-12 | 2020-08-28 | 中国科学院软件研究所 | Non-contact sensing method based on LoRa signal of Internet of things |
| CN112132719A (en) * | 2020-08-11 | 2020-12-25 | 湖南大学 | Personnel search and rescue method, device and system for emergency rescue and storage medium |
| CN113163333A (en) * | 2021-05-27 | 2021-07-23 | 上海上晟电子科技有限公司 | LoRa technology-based field life-saving rescue backpack system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113934971B (en) | 2022-05-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kim et al. | Cyclostationary approaches to signal detection and classification in cognitive radio | |
| CN102367743B (en) | Mining wireless frequency-mixing multidata monitoring system | |
| CN109164489A (en) | A kind of earthquake fluid prediction technique based on VMD Yu TK energy operator | |
| US20150180689A1 (en) | Cognitive radio spectrum sensing with improved edge detection of frequency bands | |
| CN102324959B (en) | Frequency spectrum sensing method based on multi-aerial system covariance matrix | |
| CN103532645B (en) | The compression frequency spectrum sensing method that a kind of observing matrix is optimized | |
| CN103297160A (en) | Spectrum sensing method and spectrum sensing device for goodness-of-fit test based on normalized eigenvalues | |
| CN103391143A (en) | Multi-antenna blind spectrum sensing method and system based on eigenvalue | |
| CN106170139B (en) | A kind of frequency spectrum detecting method and system | |
| CN110737023A (en) | mining micro-seismic monitoring signal processing method | |
| Divakaran et al. | Wavelet based spectrum sensing techniques for cognitive radio-a survey | |
| CN113934971B (en) | Long-distance sensing method for underground rescue robot based on LoRa | |
| CN103491551A (en) | Feature vector based weighting-coordinated frequency spectrum sensing method | |
| CN103334781A (en) | Monitoring system for potential safety hazard in methane gas exhausting and mining | |
| CN103169449B (en) | Method and device for detecting respiration signals | |
| CN105763273A (en) | Cognitive radio spectrum sensing method | |
| CN108683985A (en) | A kind of WIFI location fingerprints point prescreening method and storage medium | |
| KR20120086023A (en) | Cognitive radio communication system engine unit using chaotic property | |
| CN111641934B (en) | Method for estimating electromagnetic signal situation distribution based on vehicle-mounted equipment monitoring data | |
| Abbasi et al. | Novel CNN and hybrid CNN-LSTM algorithms for UWB SNR estimation | |
| CN105262506A (en) | Time-frequency overlapped Gaussian amplitude modulation communication signal separation method | |
| Pradhan et al. | S-transformation based integrated approach for spectrum estimation, storage, and sensing in cognitive radio | |
| Lee et al. | Device Free Localization Using Prior Correction-Based Logistic Regression | |
| CN202300532U (en) | Mining wireless frequency mixing multi-data monitoring device | |
| CN101841928A (en) | Electrocardiosignal transmission method in environment of multihop wireless body area network based on adaptive codebook |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |