A method and device for estimating vibration source distance based on scorpion vibration source distance determination mechanism

Abstract

The invention discloses a vibration source distance estimation method and device based on a scorpion vibration source distance determination mechanism, comprising the steps of: using a single three-axis sensor to collect vibration data of a human body walking at different distances, preprocessing the vibration data, and obtaining vibration data corresponding to a single vibration event at different distances; performing Hilbert-Huang transform on the preprocessed vibration data to obtain a marginal spectrum and a Hilbert envelope curve of three-channel vibration data, extracting frequency domain features according to the marginal spectrum, extracting time domain features according to the Hilbert envelope curve, performing feature fusion, and constructing a distance feature database; and using a random forest regression algorithm to calculate a vibration source distance estimation value of a test sample according to the distance feature database. The invention can improve the accuracy of the existing vibration source positioning algorithm for vibration source distance estimation.

Description

Technical Field

[0001] The invention relates to the field of vibration positioning technology, and in particular to a vibration source distance estimation method and device based on a scorpion vibration source distance determination mechanism. Background Art

[0002] For smart home systems, accurate user positioning is the key to ensuring the realization of system functions. From the perspective of data collection, current user positioning methods, such as those based on optical sensors, wearable sensors, radio frequency devices, and acoustic sensors, are limited in use in some scenarios. The vibration sensor-based method is easy to implement, does not require frequent maintenance, has low privacy infringement, does not affect users' daily activities, is not affected by obstacles, and has strong anti-interference capabilities. It can effectively supplement existing information acquisition methods and enhance the scope of application and practicality of smart home systems.

[0003] In terms of positioning indoor users, most of the current vibration positioning solutions are based on the Time Difference of Arrival (TDOA) method. However, there is a dispersion phenomenon in the process of vibration wave propagation, and there is reflection or refraction when the vibration wave encounters the boundary, which causes the collected data to contain many unreliable vibration events, thereby affecting the delay estimation of the signal. At the same time, the unevenness of the propagation medium will also cause the propagation of vibration to be not uniform. Therefore, the positioning accuracy based on the TDOA method is limited in actual scenarios. In view of the limited positioning performance of the current traditional positioning algorithm due to objective factors, with the help of the scorpion's excellent vibration positioning talent, conducting cross-research on bionics and engineering is an effective way to solve the existing dilemma. However, the scorpion-like vibration source positioning algorithm is still not perfect. The existing bionic positioning algorithm mainly constructs a positioning model based on the excitation-inhibition mechanism of the scorpion slit receptor array to estimate the direction of the vibration source, but cannot directly determine the distance information of the vibration source, resulting in limitations in the practical application of the scorpion-like vibration source positioning algorithm.

[0004] The existing scorpion-like vibration source positioning algorithm has limited use. Therefore, it is necessary to construct a vibration source distance estimation model based on the scorpion's biological processing mechanism of vibration information to improve the performance of the existing bionic positioning algorithm. Summary of the invention

[0005] The technical problem to be solved by the present invention is that, in view of the above-mentioned defects of the prior art, a method and device for estimating the source distance based on the scorpion source distance determination mechanism is provided, aiming to solve the problem of limited use of the scorpion-like source positioning algorithm in the prior art.

[0006] The technical solution adopted by the present invention to solve the technical problem is as follows:

[0007] A method for estimating the vibration source distance based on the scorpion vibration source distance determination mechanism, comprising the following steps:

[0008] Vibration data of human walking at different distances are collected by a single three-axis sensor, and the collected vibration data are pre-processed to obtain vibration data corresponding to a single vibration event at different distances;

[0009] The pre-processed vibration data is based on the scorpion's processing mechanism for vibration information. The time domain and frequency domain features are extracted through the Hilbert-Huang transform method, and feature fusion is performed to build a distance feature database.

[0010] A vibration source distance estimation model is constructed based on the distance feature database using the random forest regression algorithm. The vibration source distance is estimated using the constructed vibration source distance estimation model to obtain an estimated value of the vibration source distance and output it.

[0011] The vibration source distance estimation method based on the scorpion vibration source distance determination mechanism, wherein the collected vibration data is preprocessed to obtain the corresponding single vibration event at different distances.

[0012] The corresponding vibration data include:

[0013] The three-channel vibration data collected by the three-axis sensor are subjected to the least square method to eliminate the zero drift, thereby obtaining the first data;

[0014] Use a unified time window to frame the first data so that the start and end times of the three-channel vibration data corresponding to the same vibration event are consistent. Select an appropriate window length to ensure the data integrity of a single vibration event, and obtain three-channel vibration data corresponding to a single vibration event. Obtain vibration data corresponding to each vibration event at different distances, and then construct a second data set.

[0015] The vibration source distance estimation method based on the scorpion vibration source distance determination mechanism, wherein the steps of extracting time domain and frequency domain features from the pre-processed vibration data and performing feature fusion to construct a distance feature database include:

[0016] Drawing on the scorpion's processing mechanism for vibration information, the different vibration source distances are marked by the difference information generated by the longitudinal wave and Rayleigh wave during the propagation process as the distance from the vibration source changes;

[0017] Taking the energy attenuation speed and propagation speed of different vibration waves as clues, the Hilbert-Huang transform method is used to extract the frequency domain features and time domain features of the vibration signal, and feature fusion is performed to construct a distance feature database to reflect the above difference information;

[0018] A Hilbert-Huang transform is performed on each set of three-channel vibration data from the second data set to obtain a marginal spectrum and a Hilbert envelope curve of the three-channel vibration data.

[0019] The vibration source distance estimation method based on the scorpion vibration source distance determination mechanism, wherein the pre-processed vibration data draws on the scorpion's processing mechanism for vibration information, extracts time domain and frequency domain features, and performs feature fusion, and the step of constructing a distance feature database also includes:

[0020] According to the marginal spectrum of the vibration signal, the marginal spectrum is divided into n frequency bands according to a fixed step size, and the energy in each frequency band of the marginal spectrum is calculated. The calculation method is:

[0021]

[0022] Where i represents the sensor x, y, and z axes, i = 1, 2, 3, and f j Indicates the upper frequency limit of the jth frequency band in n frequency bands, j = 1, ..., n, E ij represents the marginal spectrum energy of the jth frequency band of the marginal spectrum of the sensor i-axis data, h i (f) represents the marginal spectrum of the sensor i-axis data;

[0023] Statistical analysis of the energy of each frequency band in the total energy of the sensor's 3-axis vibration signal ij , and use this as the frequency domain feature of the vibration signal:

[0024]

[0025] Among them, EF ij It represents the marginal spectrum energy proportion of the jth frequency band of the marginal spectrum of the sensor i-axis data;

[0026] According to the Hilbert envelope curve of the vibration signal, the value corresponding to the Hilbert envelope curve of the axial data of each sensor at the corresponding moment is counted with a fixed step size each time, a total of m values ​​on each sensor axis are obtained, and normalized, and the normalized time domain feature AF is obtained. ik As time domain features:

[0027]

[0028] Among them, t k A represents the time corresponding to the kth time domain eigenvalue for the Hilbert envelope curve of the sensor i-axis data; i Represents the Hilbert envelope curve of the sensor i-axis data; AF ik represents the kth normalized time domain feature of the sensor i-axis data;

[0029] The feature fusion includes splicing the time domain feature and the frequency domain feature together to obtain a distance feature;

[0030] The constructing of the distance feature database includes acquiring, according to the second data set, distance features of vibration data corresponding to each vibration event at different distances, and thereby obtaining a distance feature database.

[0031] The vibration source distance estimation method based on the scorpion vibration source distance determination mechanism, wherein the steps of constructing a vibration source distance estimation model using a random forest regression algorithm according to a distance feature database, estimating the vibration source distance through the constructed vibration source distance estimation model, obtaining an estimated value of the vibration source distance and outputting it include:

[0032] From the distance feature data set containing c samples, perform c samplings with replacement by a bootstrap sampling method to obtain a sampling set containing c samples;

[0033] According to the T sampling sets containing c training samples, a CART regression tree for distance estimation is trained based on each sampling set;

[0034] When the test set is input, the CART regression trees are combined, and the results of each regression tree are averaged using a simple averaging method to finally obtain an estimated value d of the vibration source distance.

[0035] A vibration source distance estimation device based on the scorpion vibration source distance determination mechanism, comprising: an analog-to-digital conversion module with at least three channels, the front end of the analog-to-digital conversion module is connected to a three-axis vibration sensor, the back end of the analog-to-digital conversion module is connected to a lower computer for data transmission, and the lower computer is connected to an upper computer;

[0036] When the three-axis vibration sensor is in use, the z-axis is perpendicular to the ground, and the x-axis and y-axis are parallel to the ground;

[0037] The upper computer receives the three-channel vibration data from the lower computer, and respectively performs real-time display, data storage and data processing on the three-channel vibration data;

[0038] The vibration source distance estimation device based on the scorpion vibration source distance determination mechanism is used to run the vibration source distance estimation method based on the scorpion vibration source distance determination mechanism.

[0039] A vibration source distance estimation device based on the scorpion vibration source distance determination mechanism, comprising:

[0040] The data acquisition and preprocessing module is used to collect vibration data of human walking at different distances through a single three-axis sensor, preprocess the collected vibration data, and obtain vibration data corresponding to a single vibration event at different distances;

[0041] The distance feature database construction module is used to extract time domain and frequency domain features from the pre-processed vibration data by using the Hilbert-Huang transform method, perform feature fusion, and construct a distance feature database.

[0042] The vibration source distance estimation module is used to construct a vibration source distance estimation model based on the distance feature database using the random forest regression algorithm, estimate the vibration source distance through the constructed vibration source distance estimation model, obtain the estimated value of the vibration source distance and output it.

[0043] An intelligent terminal includes a memory and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by one or more processors, including the method for executing any one of the methods described above.

[0044] A non-transitory computer-readable storage medium, wherein when instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute any one of the methods described above.

[0045] Beneficial effects: The present invention provides a vibration source distance estimation method and device based on the scorpion vibration source distance determination mechanism, comprising the steps of: using a single three-axis sensor to collect vibration data of a human body walking at different distances, preprocessing the vibration data, and obtaining vibration data corresponding to a single vibration event at different distances; performing Hilbert-Huang transform on the preprocessed vibration data to obtain a marginal spectrum and a Hilbert envelope curve of the three-channel vibration data, extracting frequency domain features according to the marginal spectrum, extracting time domain features according to the Hilbert envelope curve, performing feature fusion, and constructing a distance feature database; and using a random forest regression algorithm to calculate a vibration source distance estimation value of a test sample according to the distance feature database. The present invention can improve the accuracy of existing vibration source positioning algorithms for vibration source distance estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is the first flow chart of a vibration source distance estimation method based on the scorpion vibration source distance determination mechanism in the present invention.

[0047] Figure 2 It is a schematic diagram of an electrophysiological experiment of a vibration source distance estimation method based on the scorpion vibration source distance determination mechanism in the present invention.

[0048] Figure 3 The invention discloses an electrophysiological signal of a vibration receptor of a Peter Scorpion according to a vibration source distance estimation method based on a scorpion vibration source distance determination mechanism.

[0049] Figure 4 The invention relates to a bionic indoor vibration source positioning device for a vibration source distance estimation device based on a scorpion vibration source distance determination mechanism.

[0050] Figure 5 It is a single vibration event extraction result of a vibration source distance estimation method based on the scorpion vibration source distance determination mechanism in the present invention.

[0051] Figure 6 It is the marginal spectrum of the sensor's three-axis data when the vibration source distance is 2m and 6m in a vibration source distance estimation method based on the scorpion vibration source distance determination mechanism in the present invention.

[0052] Figure 7 It is a schematic diagram of marginal spectrum feature extraction of a vibration source distance estimation method based on the scorpion vibration source distance determination mechanism in the present invention.

[0053] Figure 8 It is the Hilbert envelope curve of the sensor's three-axis data when the vibration source distance is 2m and 6m in a vibration source distance estimation method based on the scorpion vibration source distance determination mechanism in the present invention.

[0054] Fig. 9 It is a schematic diagram of Hilbert envelope curve feature extraction of a vibration source distance estimation method based on the scorpion vibration source distance determination mechanism in the present invention.

[0055] Fig.10 This is a site diagram of data collection for a vibration source distance estimation experiment of a vibration source distance estimation method based on a scorpion vibration source distance determination mechanism in the present invention.

[0056] Fig.11 It is a functional principle block diagram of a vibration source distance estimation device based on the scorpion vibration source distance determination mechanism in the present invention.

[0057] Fig.12 It is a functional principle block diagram of the intelligent terminal according to an embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0059] A three-axis sensor is used in the embodiment of the present invention. The three-axis acceleration sensor is an electronic device that can measure the acceleration of an object in space. According to the different implementation principles of the sensor, it can be divided into piezoresistive, piezoelectric, capacitive and other types. The essential principle of these sensors is to convert the acceleration of the object into other physical quantities, such as resistance, voltage, capacitance, etc., through sensor elements, and then convert these physical quantities into suitable output signals through conversion and filtering circuits.

[0060] like Figure 1As shown, a method for estimating the vibration source distance based on the scorpion vibration source distance determination mechanism provided by an embodiment of the present invention comprises the following steps:

[0061] Step S100: collecting vibration data of a human body walking at different distances through a single three-axis sensor, preprocessing the collected vibration data, and obtaining vibration data corresponding to a single vibration event at different distances.

[0062] When this step is specifically implemented, it includes: collecting vibration data of a human body walking at different distances through a single three-axis sensor, and then using the least square method to eliminate zero drift of the three-channel vibration data collected by the three-axis sensor to obtain first data;

[0063] Use a unified time window to frame the first data so that the start and end times of the three-channel vibration data corresponding to the same vibration event are consistent. Select an appropriate window length to ensure the data integrity of a single vibration event, and obtain three-channel vibration data corresponding to a single vibration event. Obtain vibration data corresponding to each vibration event at different distances, and then construct a second data set.

[0064] Step S200, for the pre-processed vibration data, the scorpion's processing mechanism for vibration information is used to extract time domain and frequency domain features through the Hilbert-Huang transform method, and feature fusion is performed to construct a distance feature database;

[0065] In this step, specifically, for the pre-processed vibration data, the scorpion's processing mechanism for vibration information is used to mark different vibration source distances through the difference information generated by the longitudinal wave and the Rayleigh wave during the propagation process as the vibration source distance changes;

[0066] Then, taking the energy attenuation speed and propagation speed of different vibration waves as clues, the Hilbert-Huang transform method is used to extract the frequency domain features and time domain features of the vibration signal, and feature fusion is performed to construct a distance feature database to reflect the above difference information;

[0067] Performing Hilbert-Huang transform on each set of three-channel vibration data from the second data set to obtain a marginal spectrum and a Hilbert envelope curve of the three-channel vibration data;

[0068] According to the marginal spectrum of the vibration signal, the marginal spectrum is divided into n frequency bands according to a fixed step size, and the energy in each frequency band of the marginal spectrum is calculated. The calculation method is:

[0069]

[0070] Where i represents the sensor x, y, and z axes, i = 1, 2, 3, and f j Indicates the upper frequency limit of the jth frequency band in n frequency bands, j = 1, ..., n, Eij represents the marginal spectrum energy of the jth frequency band of the marginal spectrum of the sensor i-axis data, h i (f) represents the marginal spectrum of the sensor i-axis data;

[0071] Statistical analysis of the energy of each frequency band in the total energy of the sensor's 3-axis vibration signal ij , and use this as the frequency domain feature of the vibration signal:

[0072]

[0073] Among them, EF ij It represents the marginal spectrum energy proportion of the jth frequency band of the marginal spectrum of the sensor i-axis data;

[0074] According to the Hilbert envelope curve of the vibration signal, the value corresponding to the Hilbert envelope curve of the axial data of each sensor at the corresponding moment is counted with a fixed step size each time, a total of m values ​​on each sensor axis are obtained, and normalized, and the normalized time domain feature AF is obtained. ik As time domain features:

[0075]

[0076] Among them, t k A represents the time corresponding to the kth time domain eigenvalue for the Hilbert envelope curve of the sensor i-axis data; i Represents the Hilbert envelope curve of the sensor i-axis data; AF ik represents the kth normalized time domain feature of the sensor i-axis data;

[0077] The feature fusion includes combining the time domain feature with the frequency domain feature to obtain a distance feature, thereby constructing a distance feature database;

[0078] The distance feature database constructed in the embodiment of the present invention includes obtaining the distance features of the vibration data corresponding to each vibration event at different distances according to the second data set, thereby obtaining the distance feature database.

[0079] Step S300: construct a vibration source distance estimation model using a random forest regression algorithm according to the distance feature database, estimate the vibration source distance using the constructed vibration source distance estimation model, obtain an estimated value of the vibration source distance and output it.

[0080] In this step, c samplings with replacement are performed from the distance feature data set containing c samples by using a bootstrap sampling method to obtain a sampling set containing c samples;

[0081] According to the T sampling sets containing c training samples, a CART regression tree for distance estimation is trained based on each sampling set;

[0082] When the test set is input, the CART regression trees are combined, and the results of each regression tree are averaged using a simple averaging method to finally obtain an estimated value d of the vibration source distance.

[0083] When the present invention is implemented, the biological mechanism used by scorpions to determine the distance of the vibration source is studied through electrophysiological experiments. An electrophysiological experimental platform is built, and the walking feet of Peter's scorpion are tested for vibration source excitation at different distances. It is found that Peter's scorpion can sense the change of the vibration source distance through a single walking foot. On this basis, the tarsal sensory hairs and slit receptors of the scorpion are eliminated respectively, and the electrophysiological responses of different receptors under vibration source stimulation at different distances are tested through the electrophysiological experimental platform. Combined with the characteristics that the slit receptors are sensitive to Rayleigh waves and the tarsal sensory hairs are sensitive to longitudinal waves, it is inferred that the scorpion can sense Rayleigh waves and longitudinal waves through the slit receptors and tarsal sensory hairs. Through the information difference between the longitudinal wave and Rayleigh wave information at different vibration source distances, different vibration source distances are marked, and then the vibration source distance estimation is completed.

[0084] Through Figure 2 The electrophysiological experiment shown in the figure studies the biological mechanism used by scorpions to determine the distance of the vibration source. An electrophysiological experimental platform was built, and the vibration source excitation test of the walking foot of Peter's scorpion was carried out at different distances to obtain the following Figure 3 The electrophysiological signals of the scorpion's walking feet are shown in Figure 1, and it is found that Peter's scorpion can sense the change of the vibration source distance through a single walking foot. On this basis, the tarsal sensory hairs and slit receptors of the scorpion are eliminated respectively, and the electrophysiological responses of different receptors under the stimulation of vibration sources at different distances are tested through the electrophysiological experimental platform. Combined with the characteristics that slit receptors are sensitive to Rayleigh waves and tarsal sensory hairs are sensitive to longitudinal waves, it is inferred that the scorpion can sense Rayleigh waves and longitudinal waves through slit receptors and tarsal sensory hairs. Through the information difference between longitudinal waves and Rayleigh waves at different vibration source distances, different vibration source distances are marked, and then the vibration source distance estimation is completed.

[0085] like Figure 4As shown in the figure, the construction of the bionic indoor vibration source positioning device mainly includes two parts: data acquisition and data processing. The data acquisition is sent to the data acquisition card through a 3-axis accelerometer, and then sent to the host computer for processing. According to research, scorpions perceive the longitudinal waves and Rayleigh waves caused by the movement of prey through the tarsal sensory hairs and slit receptors on their walking feet during the process of positioning and predation. Therefore, in order to better simulate the vibration perception ability of scorpions, the propagation characteristics of longitudinal waves and Rayleigh waves need to be considered when selecting sensors. The motion component of the longitudinal wave is mainly concentrated in the direction parallel to the ground, while the Rayleigh wave can detect the component of the motion of its medium particles in the direction perpendicular or parallel to the ground. When collecting vibration signals, traditional single-axis sensors can generally only detect vibration information perpendicular to the ground. Therefore, in order to obtain more abundant vibration information, the ADXL354 3-axis accelerometer produced by Analog Devices is used. The sensor has two data output modes, analog and digital, and has the characteristics of low noise, low power consumption and low drift. At the same time, due to the use of MEMS technology, the size of the sensor is only 2×2cm. When using a 3-axis sensor, place the sensor's z-axis perpendicular to the ground.

[0086] The sensor backend is connected to the USB-3131 data acquisition card produced by Beijing Smile Technology Co., Ltd. to realize analog-to-digital conversion of vibration signals and data transmission to the PC host computer. The USB-3131 data acquisition card realizes high-bandwidth data transmission based on high-speed USB read-write technology, has 16-bit vertical resolution and 7 ranges with a maximum of ±24V, and is equipped with 24 acquisition channels to meet the needs of multi-channel vibration signal acquisition.

[0087] The data processing part of the bionic indoor vibration source positioning device is completed on the PC side, receiving the vibration signal for real-time display and data storage, and completing the positioning algorithm through MATLAB. During the positioning process, the 3-axis data of a single sensor will be used to estimate the distance to the vibration source.

[0088] During the vibration signal acquisition process, it is easily affected by the noise of the acquisition equipment and the external environment. At the same time, due to the short-term stability of the vibration signal, it is necessary to preprocess the data before using the vibration signal for algorithm implementation. When collecting vibration signals, the sensor will be subjected to vibration shock, causing the collected signal to have zero drift, which in turn leads to a decrease in the signal quality of the effective signal. The least squares method is used to eliminate linear trend terms and possible nonlinear trend terms.

[0089] The original vibration signal generally contains multiple vibration events. In order to extract a single vibration event for analysis, the present invention uses the short-term average / long-term average (STA / LTA) algorithm for vibration detection. The results are as follows: Figure 5 As shown. The STA / LTA algorithm is a method for automatically picking up vibration signals when they arrive. It determines whether a vibration event occurs by the energy change between the effective signal and the background noise in the time domain. LTA represents the signal average value within a long sliding window, which reflects the level of background noise. STA represents the signal average value within a short time window, which is mainly used to obtain information about useful signals. The ratio of STA to LTA can measure the amplitude or energy change of the signal. Since the change of STA is greater than LTA when vibration occurs, the value of STA / LTA will increase. When the value of STA / LTA exceeds a preset threshold, it indicates that a vibration event has occurred. The present invention needs to use a unified time window to process the data from the three axes of the same sensor. Therefore, the STA / LTA values ​​of the three-axis signals are superimposed to detect vibration events. The specific calculation formula of STA / LTA is as follows:

[0090]

[0091]

[0092]

[0093] Among them, STA k (i) represents short-time window average, with a window length of ns; LTA k (i) is the long-term window average, with a window length of nl; k represents the three axes of the sensor; λ is the manually set STA / LTA threshold; CF k (i) is the characteristic function value of the vibration signal at time i, which is used to measure the amplitude or energy change of the signal.

[0094] According to the biological mechanism that scorpions can mark the source distance through the information difference between longitudinal wave and Rayleigh wave information at different source distances, when extracting distance features, it is necessary to analyze the vibration signal according to the propagation characteristics of longitudinal wave and Rayleigh wave, and find a vibration signal feature that can characterize the propagation difference between longitudinal wave and Rayleigh wave. According to the analysis in the previous article, the energy of vibration will decay with the increase of the source distance during the propagation process, and this attenuation is related to the frequency of the vibration itself. The higher the frequency, the faster the attenuation. The frequency of longitudinal wave is higher than that of Rayleigh wave, and the energy decay rate of longitudinal wave is greater than that of Rayleigh wave. The marginal spectrum of the vibration signal is constructed by HHT algorithm, and its frequency domain information is analyzed. The low-frequency part of its marginal spectrum contains more Rayleigh wave information, and the high-frequency part contains more longitudinal wave information. The energy decay rates of high-frequency and low-frequency parts are different, and the energy of high-frequency part decays faster. As the distance from the source increases, the energy proportion in each frequency band will change, the energy proportion of the high-frequency part decreases, and the energy proportion of the low-frequency part increases. On the other hand, according to the propagation characteristics of longitudinal waves and Rayleigh waves, the longitudinal wave and Rayleigh wave information contained in the vibration signals on each axis of the 3-axis sensor is different. Therefore, the energy proportion of the sensor's 3-axis data will change with the distance. Figure 6 The marginal spectra of the sensor's 3-axis data when the vibration source distance is 2m and 6m are shown.

[0095] First, consider the energy attenuation of the vibration signal as a distance feature, and use the HHT algorithm to construct the vibration signal marginal spectrum to complete the energy feature extraction. The specific method is as follows. Figure 7 As shown, according to the marginal spectrum of the vibration signal, the marginal spectrum is divided into n frequency bands according to a fixed step size, and the energy in each frequency band of the marginal spectrum is calculated. The calculation formula is shown in formula 3.18. The proportion of the energy of each frequency band in the total energy of the sensor's 3-axis vibration signal EF is calculated. ij , as shown in the following formula, and this is used as the energy feature of the vibration signal.

[0096]

[0097]

[0098] Where i represents the sensor x, y, and z axes, i = 1, 2, 3; f j Indicates the upper frequency limit of the jth frequency band in n frequency bands, j = 1, ..., n; E ij represents the marginal spectrum energy of the jth frequency band of the marginal spectrum of the sensor i-axis data; h i (f) represents the marginal spectrum of the sensor i-axis data; EF ij It represents the marginal spectrum energy ratio of the jth frequency band of the marginal spectrum of the sensor i-axis data.

[0099] Since the vibration amplitude decays sharply in the form of a negative power function as the distance increases, this may cause the energy features extracted by the marginal spectrum to have a lower degree of distinguishing the source distance when the source distance is large. According to the scorpion source distance determination mechanism, the scorpion marks the source distance by using the difference information between the longitudinal wave and the Rayleigh wave. This difference information can not only include the different energy decay rates, but also consider the characteristics of the longitudinal wave propagating faster and arriving earlier than the Rayleigh wave, and analyze the vibration signal from the time domain perspective. Obtain the Hilbert envelope of the original vibration signal and analyze its time domain information. As the source distance increases, the speed difference between the longitudinal wave and the Rayleigh wave will increase the separation between the two, resulting in obvious changes in the time domain. Figure 8 The Hilbert envelope curve of the sensor's 3-axis data when the vibration source distance is 2m and 6m is shown.

[0100] At this time, using the time domain characteristics of the vibration signal as a distance feature can increase the degree of distinction of distance characteristics when the vibration source distance is large. The specific method is as follows. Fig. 9 As shown in FIG. 1 , according to the Hilbert envelope curve of the vibration signal, the value corresponding to the Hilbert envelope curve of the axial data of each sensor at the corresponding moment is counted with a fixed step size each time, and a total of m values ​​on each sensor axis are obtained, and normalized, and the normalized time domain feature AF is obtained. ik Also as a distance feature.

[0101]

[0102] Among them, t k A represents the time corresponding to the kth time domain eigenvalue for the Hilbert envelope curve of the sensor i-axis data; i Represents the Hilbert envelope curve of the sensor i-axis data; AF ik It represents the kth normalized time domain feature of the sensor i-axis data.

[0103] After feature extraction, the vibration source distance estimation model is trained by the random forest regression algorithm. From the distance feature data set containing c samples, c samples are sampled with replacement by the bootstrap sampling method to obtain a sampling set containing c samples; based on T sampling sets containing c training samples, a CART regression tree for distance estimation is trained based on each sampling set; when the test set is input, the CART regression trees are combined, and the results of each regression tree are averaged using the simple averaging method to finally obtain the estimated value d of the vibration source distance.

[0104] Data collection point along Fig.10The sensor direction is set as shown, starting from a radius of 1.5m and ending at 6m, with a data collection point set every 0.5m. The experimenter stepped on each data collection point in turn, with a stepping frequency of about 90 times / minute, and collected 30s of stepping signals from each data collection point of 20 people. After eliminating the zero drift of the collected vibration signal, the STA / LTA algorithm described above was used to extract vibration events, and 30 samples were obtained for each person at each data collection point, totaling 6000 (30×10×20) samples. 70% of the samples at each data collection point were randomly selected to extract frequency domain features and time domain features, and the distance estimation model was trained using the random forest regression algorithm. 30% of the samples were used as the test set, and the performance of the trained vibration source distance estimation model was tested after data preprocessing and feature extraction. The distance estimation error is shown in Table 1.

[0105] Table 1. Vibration source distance estimation results

[0106]

[0107] Based on the above embodiments, reference Figure 4 As shown, an embodiment of the present invention provides a vibration source distance estimation device based on the scorpion vibration source distance determination mechanism, comprising: an analog-to-digital conversion module with at least three channels, the front end of the analog-to-digital conversion module is connected to a three-axis vibration sensor, the back end of the analog-to-digital conversion module is connected to a lower computer for data transmission, and the lower computer is connected to an upper computer;

[0108] When the three-axis vibration sensor is in use, the z-axis is perpendicular to the ground, and the x-axis and y-axis are parallel to the ground;

[0109] The upper computer receives the three-channel vibration data from the lower computer, and respectively performs real-time display, data storage and data processing on the three-channel vibration data;

[0110] The vibration source distance estimation device based on the scorpion vibration source distance determination mechanism is used to run the vibration source distance estimation method based on the scorpion vibration source distance determination mechanism.

[0111] Exemplary Devices

[0112] like Fig.11 As shown, an embodiment of the present invention further provides a vibration source distance estimation device based on the scorpion vibration source distance determination mechanism, comprising:

[0113] The data collection and preprocessing module 110 is used to collect vibration data of a human body walking at different distances through a single three-axis sensor, preprocess the collected vibration data, and obtain vibration data corresponding to a single vibration event at different distances;

[0114] The distance feature database construction module 120 is used to extract time domain and frequency domain features from the pre-processed vibration data by using the Hilbert-Huang transform method, perform feature fusion, and construct a distance feature database;

[0115] The vibration source distance estimation module 130 is used to construct a vibration source distance estimation model based on the distance feature database using a random forest regression algorithm, estimate the vibration source distance through the constructed vibration source distance estimation model, obtain an estimated value of the vibration source distance and output it, as described above.

[0116] Based on the above embodiments, the present invention further provides an intelligent terminal, whose principle block diagram can be shown as follows: Fig.12 As shown. The intelligent terminal includes a processor, a memory, a network interface, a display screen, and a three-axis vibration sensor connected through a system bus. Among them, the processor of the intelligent terminal is used to provide computing and control capabilities. The memory of the intelligent terminal includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the intelligent terminal is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a vibration source distance estimation method based on the scorpion vibration source distance determination mechanism is implemented. The display screen of the intelligent terminal can be a liquid crystal display screen or an electronic ink display screen.

[0117] Those skilled in the art will understand that Fig.12 The principle block diagram shown in the figure is only a block diagram of a partial structure related to the scheme of the present invention, and does not constitute a limitation on the smart terminal to which the scheme of the present invention is applied. The specific smart terminal may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0118] In one embodiment, a smart terminal is provided, comprising a memory and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by one or more processors, and the one or more programs include instructions for performing the following operations:

[0119] Vibration data of human walking at different distances are collected by a single three-axis sensor, and the collected vibration data are pre-processed to obtain vibration data corresponding to a single vibration event at different distances;

[0120] The pre-processed vibration data is based on the scorpion's processing mechanism for vibration information. The time domain and frequency domain features are extracted through the Hilbert-Huang transform method, and feature fusion is performed to build a distance feature database.

[0121] A vibration source distance estimation model is constructed based on the distance feature database using a random forest regression algorithm. The vibration source distance is estimated using the constructed vibration source distance estimation model to obtain an estimated value of the vibration source distance and output it; the details are as described above.

[0122] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0123] An embodiment of the present invention further provides a non-temporary computer-readable storage medium, wherein when instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute any one of the methods described above.

[0124] In summary, the present invention discloses a vibration source distance estimation method and device based on the scorpion vibration source distance determination mechanism, comprising the steps of: using a single three-axis sensor to collect vibration data of a human body walking at different distances, preprocessing the vibration data, and obtaining vibration data corresponding to a single vibration event at different distances; performing Hilbert-Huang transform on the preprocessed vibration data to obtain a marginal spectrum and a Hilbert envelope curve of three-channel vibration data, extracting frequency domain features according to the marginal spectrum, extracting time domain features according to the Hilbert envelope curve, performing feature fusion, and constructing a distance feature database; and using a random forest regression algorithm to calculate a vibration source distance estimation value of a test sample according to the distance feature database. The present invention can improve the accuracy of existing vibration source positioning algorithms for vibration source distance estimation.

[0125] It should be understood that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for estimating the vibration source distance based on the scorpion vibration source distance determination mechanism, characterized in that: The following steps are involved: Vibration data of human walking at different distances are collected by a single three-axis sensor, and the collected vibration data are preprocessed to obtain vibration data corresponding to a single vibration event at different distances; The pre-processed vibration data is based on the scorpion's processing mechanism for vibration information. The time domain and frequency domain features are extracted through the Hilbert-Huang transform method, and feature fusion is performed to build a distance feature database. A vibration source distance estimation model is constructed using a random forest regression algorithm according to a distance feature database, and the vibration source distance is estimated by using the constructed vibration source distance estimation model to obtain an estimated value of the vibration source distance and output it; The steps of extracting time domain and frequency domain features from the pre-processed vibration data and fusing the features to construct a distance feature database include: Drawing on the scorpion's processing mechanism for vibration information, the different vibration source distances are marked by the difference information generated by the longitudinal wave and Rayleigh wave during the propagation process as the distance from the vibration source changes; Taking the energy attenuation speed and propagation speed of different vibration waves as clues, the Hilbert-Huang transform method is used to extract the frequency domain features and time domain features of the vibration signal, and feature fusion is performed to construct a distance feature database to reflect the above difference information; Performing Hilbert-Huang transform on each set of three-channel vibration data from the second data set to obtain a marginal spectrum and a Hilbert envelope curve of the three-channel vibration data; The step of extracting time domain and frequency domain features from the pre-processed vibration data and performing feature fusion to construct a distance feature database also includes: According to the marginal spectrum of the vibration signal, the marginal spectrum is divided into n frequency bands according to a fixed step size, and the energy in each frequency band of the marginal spectrum is calculated. The calculation method is: in, Represents the sensor x, y, z axis, , Indicates the first The upper frequency limit of the frequency band, , Indicates sensor The marginal spectrum of the axis data is The marginal spectral energy of the frequency band, Indicates sensor The marginal spectrum required by the axis data; Count the proportion of each frequency band energy in the total energy of the sensor's 3-axis vibration signal , and use this as the frequency domain feature of the vibration signal: in, Indicates sensor The marginal spectrum of the axis data is The marginal spectrum energy proportion of each frequency band; According to the Hilbert envelope curve of the vibration signal, the value corresponding to the Hilbert envelope curve of the axial data of each sensor at the corresponding moment is counted at a fixed step size each time, a total of m values ​​on each sensor axis are obtained, and normalized, and the normalized time domain feature is converted into As time domain features: in, Indicates the sensor For the Hilbert envelope curve required by the axis data, the time corresponding to the kth time domain eigenvalue; Indicates sensor The Hilbert envelope curve required by the axis data; Indicates sensor The k-th normalized time domain feature of the axis data; The feature fusion includes splicing the time domain feature and the frequency domain feature together to obtain a distance feature; The constructing of the distance feature database includes acquiring, according to the second data set, distance features of vibration data corresponding to each vibration event at different distances, and thereby obtaining a distance feature database.

2. The vibration source distance estimation method based on the scorpion vibration source distance determination mechanism according to claim 1 is characterized in that: The preprocessing of the collected vibration data to obtain vibration data corresponding to a single vibration event at different distances includes: The three-channel vibration data collected by the three-axis sensor are subjected to the least square method to eliminate the zero drift, thereby obtaining the first data; Use a unified time window to frame the first data so that the start and end times of the three-channel vibration data corresponding to the same vibration event are consistent. Select an appropriate window length to ensure the data integrity of a single vibration event, and obtain three-channel vibration data corresponding to a single vibration event. Obtain the vibration data corresponding to each vibration event at different distances, and then construct the second data set.

3. The vibration source distance estimation method based on the scorpion vibration source distance determination mechanism according to claim 1 is characterized in that: The steps of constructing a vibration source distance estimation model using a random forest regression algorithm according to a distance feature database, estimating the vibration source distance using the constructed vibration source distance estimation model, obtaining an estimated value of the vibration source distance and outputting the estimated value include: From the distance feature data set containing c samples, perform c samplings with replacement by a bootstrap sampling method to obtain a sampling set containing c samples; According to the T sampling sets containing c training samples, a CART regression tree for distance estimation is trained based on each sampling set; When the test set is input, the CART regression trees are combined, and the results of each regression tree are averaged using a simple averaging method to finally obtain an estimated value d of the vibration source distance.

4. A vibration source distance estimation device based on the scorpion vibration source distance determination mechanism, characterized in that: include: An analog-to-digital conversion module with at least three channels, wherein the front end of the analog-to-digital conversion module is connected to a three-axis vibration sensor, the back end of the analog-to-digital conversion module is connected to a lower computer for data transmission, and the lower computer is connected to an upper computer; When the three-axis vibration sensor is in use, the z-axis is perpendicular to the ground, and the x-axis and y-axis are parallel to the ground; The upper computer receives the three-channel vibration data from the lower computer, and respectively performs real-time display, data storage and data processing on the three-channel vibration data; The vibration source distance estimation device based on the scorpion vibration source distance determination mechanism is used to run the method described in any one of claims 1-3.

5. A vibration source distance estimation device based on the scorpion vibration source distance determination mechanism, characterized in that: include: The data acquisition and preprocessing module is used to collect vibration data of human walking at different distances through a single three-axis sensor, preprocess the collected vibration data, and obtain vibration data corresponding to a single vibration event at different distances; The distance feature database construction module is used to extract time domain and frequency domain features from the pre-processed vibration data by using the Hilbert-Huang transform method, perform feature fusion, and construct a distance feature database. A vibration source distance estimation module is used to construct a vibration source distance estimation model based on a distance feature database using a random forest regression algorithm, estimate the vibration source distance using the constructed vibration source distance estimation model, obtain an estimated value of the vibration source distance and output it; The steps of extracting time domain and frequency domain features from the pre-processed vibration data and fusing the features to construct a distance feature database include: Drawing on the scorpion's processing mechanism for vibration information, the different vibration source distances are marked by the difference information generated by the longitudinal wave and Rayleigh wave during the propagation process as the distance from the vibration source changes; Taking the energy attenuation speed and propagation speed of different vibration waves as clues, the Hilbert-Huang transform method is used to extract the frequency domain features and time domain features of the vibration signal, and feature fusion is performed to construct a distance feature database to reflect the above difference information; Performing Hilbert-Huang transform on each set of three-channel vibration data from the second data set to obtain a marginal spectrum and a Hilbert envelope curve of the three-channel vibration data; The step of extracting time domain and frequency domain features from the pre-processed vibration data and performing feature fusion to construct a distance feature database also includes: According to the marginal spectrum of the vibration signal, the marginal spectrum is divided into n frequency bands according to a fixed step size, and the energy in each frequency band of the marginal spectrum is calculated. The calculation method is: in, Represents the sensor x, y, z axis, , Indicates the first The upper frequency limit of the frequency band, , Indicates sensor The marginal spectrum of the axis data is The marginal spectral energy of the frequency band, Indicates sensor The marginal spectrum required by the axis data; Count the proportion of each frequency band energy in the total energy of the sensor's 3-axis vibration signal , and use this as the frequency domain feature of the vibration signal: in, Indicates sensor The marginal spectrum of the axis data is The marginal spectrum energy proportion of each frequency band; According to the Hilbert envelope curve of the vibration signal, the value corresponding to the Hilbert envelope curve of the axial data of each sensor at the corresponding moment is counted at a fixed step size each time, a total of m values ​​on each sensor axis are obtained, and normalized, and the normalized time domain feature is converted into As time domain features: in, Indicates the sensor For the Hilbert envelope curve required by the axis data, the time corresponding to the kth time domain eigenvalue; Indicates sensor The Hilbert envelope curve required by the axis data; Indicates sensor The k-th normalized time domain feature of the axis data; The feature fusion includes splicing the time domain feature and the frequency domain feature together to obtain a distance feature; The constructing of the distance feature database includes acquiring, according to the second data set, distance features of vibration data corresponding to each vibration event at different distances, and thereby obtaining a distance feature database.

6. An intelligent terminal, characterized in that: The device comprises a memory and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by one or more processors, and the one or more programs include being used to execute the method according to any one of claims 1 to 3.

7. A non-transitory computer-readable storage medium, characterized in that: When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method as claimed in any one of claims 1 to 3.

Images