Device for detecting normal gaits and tumble behaviors of old people through ground vibration

Abstract

The invention discloses a device for detecting normal gaits and tumble behaviors of old people through ground vibration. The device comprises a vibration sensor module used for collecting ground vibration data; the data processing unit is used for processing and analyzing the vibration data in real time; the power management system is used for providing power supply required by the device; the communication module is used for transmitting the data to external equipment or a system; the vibration sensor module is connected with the data processing unit through a signal line, and the data processing unit receives vibration data from the vibration sensor module; the data processing unit is connected with the power management system to ensure that the data processing unit obtains stable power supply; the beneficial effects of the invention are that the piezoelectric sensor can accurately detect ground vibration, ensure sensitive response to tiny vibration, and accurately capture gaits and tumble events; the filtering algorithm effectively removes noise, improves the stability and reliability of vibration data, and enables the signal processing to be more accurate.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart home and health monitoring, and in particular relates to a device for detecting the normal gait and falling behavior of an elderly person through ground vibration. Background Art

[0002] As the global population aging problem becomes increasingly serious, the home safety of the elderly has received widespread attention; falls are a common accident among the elderly, which may lead to serious injuries or health problems; in order to improve the home safety of the elderly, the technology community has conducted a lot of research and development in the fields of smart homes and health monitoring.

[0003] Smart home technology has developed rapidly in recent years; smart sensors, Internet of Things technology and data analysis methods are constantly improving, providing a variety of solutions for home-based elderly care; smart home systems include smart door locks, environmental monitoring equipment, smart lights, etc. These devices are interconnected through the network to achieve remote control and status monitoring, especially in the home environment of the elderly, smart sensors can monitor health data, environmental status and activities to provide more convenient home management; in the field of health monitoring, technological advances have also significantly improved the ability to monitor the health status of the elderly; for example, smart bracelets, smart scales, smart mattresses and other devices can record the user's physiological data in real time, such as heart rate, body temperature and weight. Through data analysis, these devices can provide health advice or detect abnormalities, thereby helping users to intervene and manage their health early.

[0004] Ground vibration sensing technology uses vibration sensors to capture vibration signals from the ground; accelerometers can detect changes in ground acceleration, and piezoelectric sensors convert vibrations into electrical signals through the piezoelectric effect; the application of these sensors makes it possible to detect events such as gait and falls, and can reflect the user's activity status in real time.

[0005] Data analysis technology involves processing the data collected by sensors to extract relevant features; the analysis of vibration data can include vibration amplitude, frequency, and duration, etc.; these features help to distinguish between normal gait and fall events; for example, through real-time monitoring of vibration amplitude, drastic vibration changes can be identified, thereby inferring possible falls.

[0006] The real-time monitoring system is responsible for processing sensor data instantly to provide current health status updates; the system includes three parts: data acquisition, data processing and result display; the key to real-time monitoring is to process data quickly and accurately, and display the results in real time to help users or caregivers understand the current situation.

[0007] Problems and defects in the prior art:

[0008] In actual applications, current vibration sensors still face challenges in data accuracy and stability. Vibration data may be affected by environmental noise, ground materials and other factors, affecting the accuracy of detection results. Existing equipment may perform inconsistently under different ground materials and environmental conditions, thus affecting its overall performance.

[0009] Data processing capabilities are crucial for real-time monitoring systems. Existing systems may experience processing delays or data loss when faced with large amounts of data. Real-time data processing requires efficient algorithms and computing power to ensure that data can be accurately processed and fed back in a short period of time.

[0010] In the prior art, vibration data is often affected by external interference, such as uneven ground, external vibration sources, etc.; this may cause the system to be unable to accurately distinguish between gait and fall in some cases; in addition, the volatility of vibration data may make it difficult to distinguish between normal gait and fall, especially when the vibration amplitude is close;

[0011] Although many existing health monitoring devices can record and display vibration data in real time, they usually lack further data analysis capabilities; for example, they are unable to provide detailed gait analysis, or are unable to conduct in-depth evaluation and recording of abnormal events; existing technologies focus more on data collection and real-time feedback, and often ignore in-depth mining and analysis of data. Summary of the invention

[0012] The purpose of the present invention is to provide a device for detecting the normal gait and falling behavior of the elderly through ground vibration, thereby overcoming the data accuracy problems and processing capacity limitations in the prior art; through high-precision vibration sensors and efficient data processing technology, it can accurately monitor and analyze ground vibrations to help determine normal gait and falling status.

[0013] To achieve the above object, the present invention provides the following technical solution: a device for detecting the normal gait and falling behavior of the elderly through ground vibration, comprising:

[0014] Vibration sensor module: used to collect ground vibration data;

[0015] Data processing unit: used to process and analyze vibration data in real time;

[0016] Power management system: provides the power supply required by the device;

[0017] Communication module: used to transmit data to external devices or systems;

[0018] The vibration sensor module is connected to the data processing unit via a signal line, and the data processing unit receives vibration data from the vibration sensor module;

[0019] The data processing unit is connected to a power management system to ensure that the data processing unit obtains a stable power supply;

[0020] The data processing unit is connected to the communication module and is used to transmit the processed data to an external system.

[0021] As a preferred technical solution of the present invention, the main features of the shock sensor module include:

[0022] Sensor type: accelerometer, piezoelectric sensor;

[0023] Working range: 0-255 vibration amplitude value;

[0024] Sensitivity: Adjustable to suit different floor materials and usage environments.

[0025] As a preferred technical solution of the present invention, the data processing unit includes

[0026] Data acquisition module: converts the original signal collected by the vibration sensor into digital data;

[0027] Data processor: including microcontroller or digital signal processor, used for filtering, feature extraction and pattern recognition of digitized data;

[0028] Memory: used to store real-time data and historical data for subsequent analysis.

[0029] As a preferred technical solution of the present invention, the key technical features of the data processing unit include:

[0030] Filtering algorithm: Use low-pass filter or median filter to remove noise;

[0031] Generate vibration data: Use the generate_vibration_data() function to generate simulated vibration data;

[0032] Design a bandpass filter: Use the design_bandpass_filter() function to design a bandpass filter; the firwin() function from the scipy.signal module is used to design FIR filters;

[0033] Apply filter: Use apply_filter() function to apply the designed filter to the vibration data; lfilter() function is used to filter the data;

[0034] Data Visualization: Use the matplotlib library to plot the raw and filtered vibration data to help visualize the filtering effect.

[0035] As a preferred technical solution of the present invention, it also includes classification of vibration data, and the specific steps are as follows:

[0036] Data collection: Get data through ground vibration sensors;

[0037] Feature extraction: extract features from vibration data;

[0038] Data labeling: Label known vibration data into different categories for model training;

[0039] Model training:

[0040] k-NN: Classification is performed based on the distance metric of the training samples, and the algorithm makes predictions based on the nearest neighbor classification of new samples;

[0041] SVM: Separate data points of different categories by constructing a hyperplane and find the optimal classification boundary;

[0042] Model prediction: Use the trained model to classify the newly collected vibration data to determine whether the current vibration pattern is normal walking or falling.

[0043] As a preferred technical solution of the present invention, the data acquired by the ground vibration sensor includes the amplitude, frequency and duration of the vibration.

[0044] As a preferred technical solution of the present invention, extracting features from vibration data includes:

[0045] Statistical characteristics: mean, standard deviation, maximum value, minimum value;

[0046] Frequency domain features: frequency components are extracted through Fourier transform;

[0047] Time domain characteristics: peak value and peak interval of vibration signal.

[0048] As a preferred technical solution of the present invention, the main features of the power management system include:

[0049] Power supply type: rechargeable battery or DC power supply can be selected;

[0050] Battery Management: Includes battery status monitoring and charge management functions to ensure battery health and continued device operation.

[0051] As a preferred technical solution of the present invention, the main features of the communication module include:

[0052] Communication protocol: support standard communication protocols TCP / IP, zigbee, BLE;

[0053] Transmission rate: Different transmission rates can be selected according to needs to ensure the real-time and reliability of data.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] Piezoelectric sensors can accurately detect ground vibrations, ensuring sensitive responses to tiny vibrations and accurately capturing gait and fall events;

[0056] The filtering algorithm effectively removes noise, improves the stability and reliability of vibration data, and makes signal processing more accurate;

[0057] The high-speed data processing capability of the MCU enables real-time monitoring, instant feedback on vibration status, and timely response to possible fall events;

[0058] The use of wireless transmission modules improves the flexibility of the equipment, enables data to be transmitted remotely, and facilitates real-time monitoring and management by users. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a structural schematic diagram of the present invention;

[0060] Figure 2 It is the overall principle diagram of the present invention;

[0061] Figure 3 It is a home deployment diagram of the equipment of the present invention;

[0062] In the picture: 1. Two No. 5 batteries; 2. Wireless transmission module; 3. Microcontroller; 4. Pressure sensor. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] Example 1

[0065] See also Figure 1-Figure 3 , which is the first embodiment of the present invention, provides a device for detecting the normal gait and falling behavior of the elderly through ground vibration, comprising:

[0066] Vibration sensor module: used to collect ground vibration data;

[0067] Data processing unit: used to process and analyze vibration data in real time;

[0068] Power management system: provides the power supply required by the device;

[0069] Communication module: used to transmit data to external devices or systems.

[0070] In this embodiment, preferably, the vibration sensor module includes a high-precision accelerometer and / or a piezoelectric sensor; the accelerometer is used to detect the acceleration change of the ground, and the piezoelectric sensor converts the vibration into an electrical signal through the piezoelectric effect; the main features of the sensor module include:

[0071] Sensor type: accelerometer, piezoelectric sensor;

[0072] Working range: 0-255 vibration amplitude value;

[0073] Sensitivity: Adjustable to suit different floor materials and usage environments.

[0074] In this embodiment, preferably, the data processing unit is responsible for real-time collection, processing and analysis of vibration data; the unit includes the following components:

[0075] Data acquisition module: converts the original signal collected by the vibration sensor into digital data;

[0076] Data processor: including microcontroller or digital signal processor (DSP), used for filtering, feature extraction and pattern recognition of digitized data;

[0077] Memory: used to store real-time data and historical data for subsequent analysis;

[0078] The key technical features of the data processing unit include:

[0079] Filtering algorithm: Use low-pass filter or median filter to remove noise and improve data stability;

[0080]

[0081]

[0082]

[0083] Generate vibration data: Use the generate_vibration_data() function to generate simulated vibration data; normal gait signals contain low-frequency vibrations, fall events contain high-frequency vibrations, and random noise is added to simulate the actual situation;

[0084] Design a bandpass filter: Use the design_bandpass_filter() function to design a bandpass filter; the firwin() function from the scipy.signal module is used to design FIR filters;

[0085] Apply filter: Use apply_filter() function to apply the designed filter to the vibration data; lfilter() function is used to filter the data;

[0086] Data Visualization: Use the matplotlib library to plot the raw and filtered vibration data to help visualize the filtering effect.

[0087] The formulas involved in the code are explained as follows:

[0088] Bandpass Filter Design Formulas

[0089] FIR filter coefficient calculation (firwin function)

[0090] formula:

[0091]

[0092] explain:

[0093] h[n]: filter coefficients (impulse response);

[0094] fhigh: high cutoff frequency (normalized to the ratio of sampling frequency);

[0095] flow: low cutoff frequency (normalized to the ratio of the sampling frequency);

[0096] N: filter order;

[0097] π(Pi) is usually taken as 3.141592654 in calculations;

[0098] The firwin function in the formula calculates the coefficients of the bandpass filter based on the input low cutoff frequency and high cutoff frequency; firwin uses the window function method to design the FIR filter;

[0099] Normalized cutoff frequency

[0100] formula:

[0101]

[0102] explain:

[0103] cutoff_frequency: cutoff frequency (low cutoff frequency or high cutoff frequency);

[0104] nyquist_frequency: Nyquist frequency, equal to half of the sampling frequency;

[0105] This formula normalizes the cutoff frequency to the Nyquist frequency for filter design; the filter application formula

[0106] FIR filter application (lfilter function)

[0107] formula:

[0108]

[0109] explain:

[0110] y[n]: output signal after filtering;

[0111] h[k]: filter coefficient (obtained from design);

[0112] x[n]: input signal (vibration data);

[0113] M: the number of filter coefficients (filter order);

[0114] This formula represents the application of an FIR filter to the input signal, where the lfilter function uses this formula to perform a convolution operation;

[0115] Data generation formula

[0116] Sine wave generation

[0117] Formula: data=A·sin(2πf·t)

[0118] explain:

[0119] A: Amplitude;

[0120] f: frequency;

[0121] t: time variable (generated by sampling frequency and data points);

[0122] The formula is used to generate simulated vibration signals, where the normal gait signal is a low-frequency sine wave and the fall signal is a high-frequency sine wave;

[0123] Noise Addition

[0124] Formula: data+=noise

[0125] explain:

[0126] noise: random noise;

[0127] Simulate disturbances and noise in real environments by adding randomly generated noise to the vibration data;

[0128] Summarize

[0129] Design a bandpass filter:

[0130] Use the firwin function to calculate the filter coefficients based on the normalized cutoff frequency;

[0131] Apply filter:

[0132] Use the lfilter function to apply the designed filter coefficients to the vibration data;

[0133] Data Generation:

[0134] Use the sine wave formula to generate vibration data and add noise to simulate the actual situation;

[0135] Feature extraction: Extracting features such as amplitude, frequency, and duration of vibration data to distinguish between normal gait and falls;

[0136] Pattern recognition algorithms: Use support vector machine (SVM), k-nearest neighbor (k-NN) and other algorithms for pattern recognition;

[0137] When using algorithms such as support vector machines (SVM) and k-nearest neighbor (k-NN) for pattern recognition, we can apply these algorithms to extract features from ground vibration data to identify different behavior patterns; specifically, these algorithms can be used to detect and classify ground vibration-based behaviors, such as normal walking and falling events;

[0138] The goal of pattern recognition

[0139] In ground vibration detection applications, the goal of pattern recognition is to classify and identify different behavior patterns based on the data obtained from the ground vibration sensor; for example:

[0140] Normal walking: Ground vibration signals have certain periodicity and characteristics, such as relatively regular changes in vibration frequency and amplitude;

[0141] Fall events: Falls usually produce large vibration peaks or irregular vibration patterns that are different from normal walking;

[0142] Relationship with ground vibration

[0143] Ground vibration data can be used to extract features that reflect different behavior patterns; SVM and k-NN algorithms can be used to classify these features to identify whether it is normal walking or falling; the specific steps are as follows:

[0144] Data collection: Data is obtained through ground vibration sensors; the data usually includes information such as the amplitude, frequency, and duration of the vibration;

[0145] Feature extraction: Extract features from vibration data; for example:

[0146] Statistical characteristics: mean, standard deviation, maximum value, minimum value;

[0147] Frequency domain features: frequency components are extracted through Fourier transform;

[0148] Time domain characteristics: peak value, peak interval, etc. of vibration signal;

[0149] Data labeling: Label known vibration data into different categories (such as normal walking or falling) for model training;

[0150] Model training:

[0151] k-NN: Classification is performed based on the distance metric of the training samples, and the algorithm makes predictions based on the nearest neighbor classification of new samples;

[0152] SVM: Separate data points of different categories by constructing a hyperplane and find the optimal classification boundary;

[0153] Model prediction: Use the trained model to classify the newly collected vibration data to determine whether the current vibration pattern is normal walking or falling.

[0154] Code Sample

[0155] Assuming we already have a dataset containing ground shaking features, here is a simplified example showing how to use k-NN and SVM for classification:

[0156] import numpy as np

[0157] from sklearn.neighbors import KNeighborsClassifier

[0158] from sklearn.svm import SVC

[0159] from sklearn.model_selection import train_test_split

[0160] from sklearn.metrics import accuracy_score

[0161] #Example vibration data (data after feature extraction)

[0162] X = np.array([

[0163] [0.5,0.1,1.2],#Features: mean, standard deviation, peak

[0164] [0.6,0.2,1.3],

[0165] [1.5,0.4,2.0],#Fall event

[0166] [1.6,0.5,2.1],

[0167] #More samples... ])

[0169] y = np.array([0,0,1,1])#Label: 0 = normal walking, 1 = falling

[0170] # Divide into training set and test set

[0171] X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.3,random_state=42)

[0172] #k-NN classification

[0173] knn=KNeighborsClassifier(n_neighbors=3)

[0174] knn.fit(X_train,y_train)

[0175] knn_predictions=knn.predict(X_test)

[0176] knn_accuracy=accuracy_score(y_test,knn_predictions)

[0177] #SVM classification

[0178] svm = SVC (kernel = 'linear')

[0179] svm.fit(X_train,y_train)

[0180] svm_predictions=svm.predict(X_test)

[0181] svm_accuracy=accuracy_score(y_test,svm_predictions)

[0182] print(f"k-NN Accuracy:{knn_accuracy}")

[0183] print(f"SVM Accuracy:{svm_accuracy}")

[0184] Summarize

[0185] Objective: To classify normal walking and falling events by analyzing ground vibration data;

[0186] Data processing: including data collection, feature extraction, data labeling, model training and prediction;

[0187] Model: Use k-NN and SVM algorithms to train and classify data;

[0188] In this way, ground vibration data can be converted into features for pattern recognition and these features can be classified using modern machine learning algorithms, thereby achieving automatic recognition of different behavior patterns.

[0189] The formulas involved in the code are explained as follows:

[0190] k-Nearest Neighbor (k-NN) Classification

[0191] k-NN classification algorithm

[0192] formula:

[0193]

[0194] explain:

[0195] x: the sample to be classified;

[0196] C: the set of all categories;

[0197] xi: samples of the nearest k neighbors;

[0198] d(x,xi): the distance between samples x and xi (usually Euclidean distance is used);

[0199] k: the number of nearest neighbors;

[0200] This formula calculates the distance from sample x to all training samples, selects the nearest k samples, and predicts the category of x based on the categories of these samples;

[0201] Euclidean distance (used to calculate the distance between samples)

[0202] formula:

[0203]

[0204] explain:

[0205] x and xi: feature vectors of two samples;

[0206] xj: the value of sample x on the jth feature;

[0207] xi,j: the value of sample xi on the jth feature;

[0208] This formula calculates the Euclidean distance between samples and is used to determine the nearest neighbor samples;

[0209] Support Vector Machine (SVM) Classification

[0210] Linear Support Vector Machine Decision Function

[0211] formula:

[0212] f(x)=w T x+b

[0213] explain:

[0214] x: eigenvector;

[0215] w: weight vector;

[0216] b: bias term;

[0217] f(x): output value of the decision function;

[0218] If f(x) is greater than zero, the predicted sample belongs to the positive class; if it is less than zero, the predicted sample belongs to the negative class;

[0219] Decision Boundary

[0220] formula:

[0221] w T x+b=0

[0222] explain:

[0223] The decision boundary is the hyperplane that SVM uses to separate different classes;

[0224] This formula represents the decision boundary in the SVM model, that is, the sample prediction category on this boundary is zero;

[0225] Optimization objective of support vector

[0226] formula:

[0227]

[0228] explain:

[0229] The optimization goal is to minimize the bi-norm (sum of squares) of the weight vector w to maximize the classification interval;

[0230] The constraint is to ensure that all samples are correctly classified, that is

[0231] y i (w T xi +b)≥1

[0232] holds for all i.

[0233] Model Evaluation

[0234] Accuracy

[0235] formula:

[0236]

[0237] explain:

[0238] Number of Correct Predictions: The number of samples that the model predicts correctly in the test set;

[0239] Total Number of Predictions: The total number of samples in the test set;

[0240] Summarize

[0241] k-NN algorithm:

[0242] Use Euclidean distance to calculate the distance between samples;

[0243] Predict the category of the sample based on the categories of the nearest k neighbors;

[0244] SVM Classifier:

[0245] Classification is performed through a linear decision function, the goal is to minimize the bi-norm of the weight vector to maximize the classification interval;

[0246] Decision boundaries are used to separate samples of different categories;

[0247] Model Evaluation:

[0248] Use accuracy to evaluate the predictive performance of the classifier;

[0249] In this embodiment, preferably, the main features of the power management system include

[0250] Power supply type: rechargeable battery or DC power supply can be selected;

[0251] Battery Management: Includes battery status monitoring and charge management functions to ensure battery health and continued device operation.

[0252] In this embodiment, preferably, the communication module is used to transmit real-time data to an external device or system, and supports wireless (such as Wi-Fi, Zigbee, Bluetooth) communication; the main features include:

[0253] Communication protocol: support standard communication protocols, such as TCP / IP, zigbee, BLE, etc.

[0254] Transmission rate: Different transmission rates can be selected according to needs to ensure the real-time and reliability of data.

[0255] The operation method of the device comprises the following steps:

[0256] Vibration data collection

[0257] Step A: After the device is started, the vibration sensor module starts to collect ground vibration data;

[0258] Time: The time interval for data collection can be set from 1 second to 10 seconds;

[0259] Temperature: The sensor operating environment temperature should be between -10℃ and 50℃ to ensure normal function;

[0260] Data processing

[0261] Step B: The data processing unit receives the digital signal of the vibration sensor module and performs filtering to remove noise;

[0262] Processing time: The filtering processing time should be controlled within 10 milliseconds;

[0263] Feature extraction: extracting features such as amplitude, frequency and duration of vibration data;

[0264] Pattern recognition: Use pattern recognition algorithms to analyze data to determine whether it is a fall event;

[0265] Step C: The processed data is displayed in real-time (if any) through the display and operation interface.

[0266] Data transmission: Data is transmitted to external systems or devices (if any) through the communication module.

[0267] Connection relationship and control relationship:

[0268] The vibration sensor module is connected to the data processing unit via a signal line, and the data processing unit receives vibration data from the vibration sensor module;

[0269] The data processing unit is connected to the power management system to ensure that the data processing unit obtains a stable power supply;

[0270] The data processing unit is connected to the communication module (optional) to transmit the processed data to an external system;

[0271] The data processing unit is connected to the display and operation interface (optional) to display the device status and data in real time.

[0272] The present invention realizes efficient collection and analysis of ground vibration data through the above technical scheme, improves the accuracy and reliability of the device, and provides a practical new method for gait and fall detection in a home-based elderly care environment.

[0273] Example 2

[0274] See also Figure 2 , which is a second embodiment of the present invention, provides a device for detecting the normal gait and falling behavior of the elderly through ground vibration, comprising:

[0275] Two AA batteries 1

[0276] Position and structure: The upper part of the device is equipped with a battery compartment that holds two AA batteries. The battery compartment is compactly designed to ensure that the batteries can be securely fixed and effectively connected to other components of the device;

[0277] Function: Each No. 5 battery provides a voltage of 1.5V, and the total voltage of two batteries in series is 3V; the battery provides a stable power supply for the device to ensure the normal operation of each electronic component; the battery selection takes into account the power consumption requirements of the device and the ease of use;

[0278] Wireless transmission module 2

[0279] Position and structure: The wireless transmission module 2 is installed near the battery compartment and is compactly designed to fit the space of the device; the wireless module includes an antenna part to enhance the signal transmission capability;

[0280] Function: Responsible for wirelessly sending the vibration data processed by the device to the remote monitoring system or user equipment; according to application requirements, the wireless transmission module 2 can select Bluetooth module, Wi-Fi module or other wireless communication technology, and its main function is to realize remote transmission and real-time monitoring of data;

[0281] Microcontroller 3 (MCU)

[0282] Position and structure: The MCU is located between the battery compartment and the wireless transmission module 2, with appropriate space to ensure good heat dissipation and signal connection; the MCU interface design includes multiple input and output pins for easy connection with other components;

[0283] Function: responsible for data acquisition, processing and control functions; MCU receives vibration signals from pressure sensor 4 and performs real-time data processing, including filtering, feature extraction and pattern recognition; the processed data is sent to external devices through a wireless transmission module; MCU selection takes into account computing power, power consumption and interface flexibility, and commonly used models include ARM Cortex-M series, AVR series, etc.;

[0284] Piezoelectric sensor 4

[0285] Position and structure: The piezoelectric sensor 4 is installed at the bottom of the device and is in direct contact with the ground. The design of the sensor takes into account the high sensitivity and frequency response capability of the vibration signal;

[0286] Function: Used to detect ground vibrations and convert them into electrical signals; the piezoelectric sensor 4 can sense tiny vibration changes, thereby accurately identifying normal gait and fall events; the sensor output signal is transmitted to the MCU for further processing.

[0287] Implementation steps

[0288] Device assembly

[0289] Install the piezoelectric sensor 4: Fix the piezoelectric sensor 4 to the bottom of the device, and ensure good contact between the sensor and the ground to ensure accurate transmission of the vibration signal;

[0290] Connect the battery: Place two AA batteries 1 into the battery compartment. The battery compartment is designed for easy replacement of batteries. The batteries are connected to the MCU via a connecting wire to ensure a stable power supply.

[0291] Install MCU and wireless transmission module 2: Install MCU on the upper part of the device and connect it to the battery and wireless transmission module 2. MCU is responsible for control and data processing, while wireless transmission module 2 is responsible for data transmission.

[0292] Connect the sensor to the MCU: Connect the output signal line of the piezoelectric sensor 4 to the analog input port of the MCU to ensure a stable signal connection and reduce signal interference;

[0293] Test device: Perform functional tests on the device to ensure that the collection, processing and transmission of vibration data are normal; during the test, the device's response ability under different vibration conditions should be checked;

[0294] Data processing and transmission

[0295] Data acquisition: MCU obtains the original vibration signal from the piezoelectric sensor 4. The vibration signal usually contains noise and interference and needs to be pre-processed;

[0296] Signal processing: The MCU uses filtering algorithms (such as bandpass filters) to process the signal, remove high-frequency noise and low-frequency interference, and extract useful vibration characteristics; the filter design takes into account the frequency range and noise characteristics of the signal;

[0297] Feature extraction: The MCU extracts features from the filtered vibration signal, such as the frequency and amplitude of the vibration, to distinguish between normal gait and fall status;

[0298] Pattern recognition: Based on the extracted features, the MCU runs a pattern recognition algorithm (such as k-NN, SVM, etc.) to determine whether the vibration signal belongs to a fall event or a normal gait; the pattern recognition algorithm classifies the training data to achieve accurate recognition of different vibration patterns;

[0299] Data transmission: The processed data and recognition results are sent to the remote monitoring system or user equipment through the wireless transmission module 2 to achieve real-time monitoring and feedback.

[0300] Although the embodiments of the present invention have been shown and described, as detailed above, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A device for detecting the normal gait and falling behavior of the elderly through ground vibration, characterized in that: include Vibration sensor module: used to collect ground vibration data; Data processing unit: used to process and analyze vibration data in real time; Power management system: provides the power supply required by the device; Communication module: used to transmit data to external devices or systems; The vibration sensor module is connected to the data processing unit via a signal line, and the data processing unit receives vibration data from the vibration sensor module; The data processing unit is connected to a power management system to ensure that the data processing unit obtains a stable power supply; The data processing unit is connected to the communication module and is used to transmit the processed data to an external system.

2. The device for detecting the normal gait and falling behavior of the elderly by ground vibration according to claim 1, characterized in that: The main features of the shock sensor module include Sensor type: accelerometer, piezoelectric sensor; Working range: 0-255 vibration amplitude value; Sensitivity: Adjustable to suit different floor materials and usage environments.

3. The device for detecting the normal gait and falling behavior of the elderly by ground vibration according to claim 1, characterized in that: The data processing unit comprises Data acquisition module: converts the original signal collected by the vibration sensor into digital data; Data processor: including microcontroller or digital signal processor, used for filtering, feature extraction and pattern recognition of digitized data; Memory: used to store real-time data and historical data for subsequent analysis.

4. The device for detecting the normal gait and falling behavior of the elderly by ground vibration according to claim 3, characterized in that: The key technical features of the data processing unit include: Filtering algorithm: Use low-pass filter or median filter to remove noise; Generate vibration data: Use the generate_vibration_data() function to generate simulated vibration data; Design a bandpass filter: Use the design_bandpass_filter() function to design a bandpass filter; the firwin() function from the scipy.signal module is used to design FIR filters; Apply filter: Use apply_filter() function to apply the designed filter to the vibration data; lfilter() function is used to filter the data; Data Visualization: Use the matplotlib library to plot the raw and filtered vibration data to help visualize the filtering effect.

5. The device for detecting the normal gait and falling behavior of the elderly by ground vibration according to claim 1, characterized in that: It also includes the classification of vibration data. The specific steps are as follows: Data collection: Get data through ground vibration sensors; Feature extraction: extract features from vibration data; Data labeling: Label known vibration data into different categories for model training; Model training: k-NN: Classification is performed based on the distance metric of the training samples, and the algorithm makes predictions based on the nearest neighbor classification of new samples; SVM: Separate data points of different categories by constructing a hyperplane and find the optimal classification boundary; Model prediction: Use the trained model to classify the newly collected vibration data to determine whether the current vibration pattern is normal walking or falling.

6. The device for detecting the normal gait and falling behavior of the elderly by ground vibration according to claim 5, characterized in that: The data acquired by the ground vibration sensor include the amplitude, frequency and duration of the vibration.

7. The device for detecting the normal gait and falling behavior of the elderly by ground vibration according to claim 5, characterized in that: Extracting features from vibration data includes Statistical characteristics: mean, standard deviation, maximum value, minimum value; Frequency domain features: frequency components are extracted through Fourier transform; Time domain characteristics: peak value and peak interval of vibration signal.

8. The device for detecting the normal gait and falling behavior of the elderly by ground vibration according to claim 1, characterized in that: The main features of the power management system include the type of power supply: rechargeable battery or DC power supply can be selected; Battery Management: Includes battery status monitoring and charge management functions to ensure battery health and continued device operation.

9. The device for detecting the normal gait and falling behavior of the elderly by ground vibration according to claim 1, characterized in that: The main features of the communication module include: Communication protocol: support standard communication protocols TCP / IP, zigbee, BLE; Transmission rate: Different transmission rates can be selected according to needs to ensure the real-time and reliability of data.

Images