Intelligent security alarm system and method based on wireless networking

Abstract

The invention discloses an intelligent security alarm system and method based on wireless networking, and relates to the technical field of Internet of Things and intelligent security, and the system comprises a security data collection module, a security feature extraction module, a security anomaly detection module, a security alarm module and an intelligent security management module. The security and protection data acquisition module acquires human body infrared, environmental sound and three-dimensional vibration data, extracts security and protection characteristics, detects movement and sound abnormity, constructs an alarm model, carries out graded alarm, issues information through a wireless technology, and displays regional security and protection event distribution through a thermodynamic diagram. According to the invention, wireless acquisition and intelligent fusion of multi-modal security data are realized, the accuracy and real-time performance of anomaly detection are significantly improved, security event distribution is visually reflected through hierarchical alarm and thermodynamic diagram display, rapid response and decision are facilitated, the false alarm rate is reduced, system deployment and maintenance are simplified, and the system reliability is improved. And the intelligence and practicability of the security and protection system are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical fields of the Internet of Things and intelligent security, and particularly relates to an intelligent security alarm system and method based on wireless networking. Background Technique

[0002] The traditional security alarm system is a security protection network composed of various components such as sensors, function keys, detectors, and actuators. This system is usually regarded as the brain of the security system for homes or specific places. Functionally, the traditional security alarm system is mainly used to achieve security protection goals such as fire prevention, anti-theft, gas leakage alarm, and emergency assistance. It relies on intelligent control network technology and is managed and controlled by a microcomputer. When incidents such as bandit intrusion, theft, fire spread, gas leakage, or emergency assistance occur, it automatically issues an alarm. From the perspective of system composition, the traditional security alarm system includes key components such as front-end detectors, intermediate transmission parts, and alarm hosts. These components work together to detect environmental changes in real time, collect and process environmental data, and once an abnormal situation is detected, an alarm is issued through the alarm device to remind relevant personnel to take countermeasures. In summary, the traditional security alarm system has become an indispensable security guarantee means in all walks of life with its reliable security protection function and flexible application scenarios.

[0003] In order to solve the problem of high false alarm rate caused by insufficient multi-modal perception fusion in the traditional security alarm system, the existing technology is to use multiple sensors to work in parallel and achieve security protection through independent signal detection and analysis. However, this method still has the situation of information isolation between sensors and lack of effective fusion, which leads to frequent false alarms when the system faces complex and changeable security threats, affecting the stability and reliability of the system. At the same time, the problems of complex wired networking deployment and poor scalability in the traditional security alarm system also need to be solved. Wired networking not only requires cumbersome wiring work, but also once the layout is completed, subsequent expansion and adjustment are extremely inconvenient, which limits the flexibility and adaptability of the security system and cannot meet the diverse and dynamic security protection needs of modern places. Therefore, in view of the above problems, an intelligent security alarm system and method based on wireless networking are proposed. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent security alarm system and method based on wireless networking to solve the problems raised in the above background technique.

[0005] To solve the above technical problems, the technical solution adopted by the present invention is: In the first aspect, an intelligent security alarm system based on wireless networking includes a security data acquisition module, a security feature extraction module, a security anomaly detection module, a security alarm module, and an intelligent security management module;

[0006] The security data acquisition module collects and preprocesses the human body movement infrared signal intensity data, environmental sound waveform data, and three-dimensional vibration waveform data;

[0007] The security feature extraction module extracts features from the preprocessed human body movement infrared signal intensity data, environmental sound waveform data, and three-dimensional vibration waveform data, and constructs a security feature sequence list;

[0008] The security anomaly detection module is used to combine the security feature sequence list to perform abnormal movement detection and abnormal sound recognition;

[0009] The security alarm module constructs a security alarm model based on the security feature sequence list and the analysis results of the security anomaly detection module, and outputs an alarm probability;

[0010] The intelligent security management module performs hierarchical alarms based on the analysis results of the security alarm model, uses wireless networking technology to publish security alarm information, and presents the distribution of regional security events in the form of a heat map.

[0011] A further improvement of the technical solution of the present invention is that the security anomaly detection module includes a movement detection unit and a sound recognition unit:

[0012] The movement detection unit constructs an abnormal movement detection model based on the extracted infrared intensity change rate feature and outputs an abnormal movement probability;

[0013] The sound recognition unit constructs an abnormal sound recognition model based on the audio frequency band energy ratio feature and the first three-order Mel cepstral coefficient feature, and outputs an abnormal voiceprint probability.

[0014] A further improvement of the technical solution of the present invention is that in the security data acquisition module, the acquisition and preprocessing process of the human body movement infrared signal intensity data, environmental sound waveform data, and three-dimensional vibration waveform data includes:

[0015] Adopt a self-organizing network protocol to install pyroelectric sensor nodes at building entrances and exits, corridor channels, and key perimeter areas to collect human body movement infrared signal intensity data. Among them, the distance between pyroelectric sensor nodes does not exceed 8 meters, the vertical installation height is 1.2 - 1.5 meters, and the tilt angle is between 15 degrees and 30 degrees. Automatically construct a mesh topology through a dynamic routing algorithm;

[0016] Deploy a four-element microphone array on the diagonal of the building interior space and under the outdoor eaves to collect environmental sound waveform data. Among them, the distance between the four-element microphone arrays is 1.5 meters, the distance from the ground is 2.2 to 2.5 meters, and it is directly connected to the gateway through a star topology. The four-element microphone array nodes use the carrier sense multiple access with collision avoidance mechanism to compete for the channel;

[0017] Deploy triaxial piezoelectric acceleration sensors at the door and window frames of the building and the base of the safe. The axes of the triaxial piezoelectric acceleration sensors are perpendicular to the monitoring plane, fixed with epoxy resin glue, and connected through a hybrid networking architecture to collect three-dimensional vibration waveform data;

[0018] Preprocess the collected human movement infrared signal intensity data, environmental sound waveform data, and three-dimensional vibration waveform data;

[0019] When the change rate of the human movement infrared signal intensity is detected to exceed 0.5 microwatts per square centimeter per second, activate the historical data record, generate a 10-hertz time series, calculate the arithmetic mean of five consecutive sampling points, replace the current point data, and eliminate instantaneous interference pulses with a duration less than 0.5 seconds;

[0020] Calculate the signal energy of the environmental sound waveform data within each 10-millisecond window. When the energy in three consecutive windows exceeds three times the silent baseline, it is determined as the starting point of the valid segment, and when the energy drops below the baseline by more than 50 milliseconds, it is determined as the termination point. Divide the sampling point values within the valid segment by the maximum absolute value of the original data to normalize its maximum absolute value to 1;

[0021] The triaxial piezoelectric acceleration sensor node calibrates the local clock based on the Precision Time Protocol, making the deviation between the three-dimensional vibration waveform data and the infrared signal timestamp less than 5 milliseconds. According to the readings of the built-in temperature sensor, adjust the zero offset of the triaxial acceleration according to the preset temperature drift coefficient.

[0022] A further improvement of the technical solution of the present invention lies in: in the security feature extraction module, the process of constructing the security feature sequence table includes:

[0023] Divide the difference between two adjacent human movement infrared signal intensity sampling points by the sampling interval to obtain the infrared intensity change rate sequence;

[0024] Perform short-time Fourier transform on the normalized valid audio segment, divide each 25 milliseconds into a window, with adjacent windows overlapping by 12.5 milliseconds. Divide the sum of the squared amplitudes of the frequency points in the 500 - 2000 Hz frequency band by the sum of the squared amplitudes of the frequency points in the 0 - 8000 Hz frequency band to obtain the audio frequency band energy ratio within the calculation window;

[0025] Pass the valid audio segment through a triangular filter bank with 40 Mel scale distributions. The output of the triangular filter is the logarithm of the signal energy of the corresponding frequency band. Perform discrete cosine transform on the logarithms of the signal energies of the corresponding frequency bands output by the 40 triangular filters, and take the first three-order Mel cepstral coefficients as the feature components;

[0026] Calculate the total vibration energy integral E of the three-dimensional vibration waveform data within a 1-second window, and perform dynamic baseline normalization on the total energy integral to obtain the normalized vibration energy The calculation process is as follows:

[0027]

[0028] where a x (i), a y (i), and a z (i) respectively represent the accelerations of the i-th sampling point on the x-axis, y-axis, and z-axis, Δt is the sampling interval time, N is the number of sampling points, μ is the average vibration energy in the past 1 hour, and σ is the standard deviation of the vibration energy in the past 1 hour;

[0029] Integrate the extracted infrared intensity change rate sequence, audio frequency band energy ratio, first three-order Mel cepstral coefficients, and normalized vibration energy to form a security feature sequence table.

[0030] A further improvement of the technical solution of the present invention lies in: in the mobile detection unit, the process of constructing an abnormal movement detection model and outputting an abnormal movement probability includes:

[0031] Based on the isolation forest algorithm, use the historical normal infrared intensity change rate sequence data to train the decision tree, preset the number and depth of the decision trees in the edge gateway of the intelligent security alarm system. Each decision tree selects the mean and variance of the normal infrared intensity change rate of the window as the splitting features. The gateway receives the latest window mean and variance, inputs them into the isolation forest model, and constructs an abnormal movement detection model;

[0032] Traverse the decision tree, calculate the baseline path length c(n) and the average path length L of the sample in the decision tree p , and after standardization, it is converted into an abnormal movement probability P1. The calculation process is as follows:

[0033]

[0034] where n is the number of samples in the training data set, γ is the Euler-Mascheroni constant, and P1 ∈ [0, 1].

[0035] A further improvement of the technical solution of the present invention lies in: in the voice recognition unit, the process of constructing an abnormal voice recognition model and outputting an abnormal voiceprint probability includes:

[0036] Perform maximum-minimum normalization on the audio frequency band energy ratio feature in the 500 - 2000 Hz frequency band, and perform standard deviation normalization on the first three-order Mel cepstral coefficients to construct a 4D voiceprint feature vector including the audio frequency band energy ratio, the first Mel coefficient, the second Mel coefficient, and the third Mel coefficient;

[0037] Based on the support vector machine classification model architecture, select the radial basis kernel function, map the 4D voiceprint feature vector to a high-dimensional space, input the historical environmental sound waveform dataset into the support vector machine classification model architecture, solve for the optimal hyperplane, select the sample points closest to the hyperplane as the support vectors β, and preset the trained hyperplane parameters to the edge gateway of the intelligent security alarm system to construct an abnormal sound recognition model;

[0038] Input the 4D voiceprint feature vector into the abnormal sound recognition model, calculate the distance d from it to the hyperplane, and use the Sigmoid function to map this distance to the abnormal voiceprint probability P2. The calculation process is as follows:

[0039]

[0040] Among them, α i is the support vector weight, y i is the support vector label, normal = 1, abnormal = -1, K is the calculation result of the radial basis kernel function, b is the bias term of the classification hyperplane, P2 ∈ [0, 1]. Set the upper and lower limits of the threshold range of the abnormal voiceprint probability, and adopt a hierarchical transmission strategy for data transmission.

[0041] A further improvement of the technical solution of the present invention lies in: in the security alarm module, the process of constructing a security alarm model and outputting an alarm probability includes:

[0042] The security alarm module receives the abnormal movement probability, abnormal voiceprint probability, and normalized vibration energy feature, and integrates them into a joint feature vector Adopt a logistic regression model architecture to construct a security alarm model and calculate the alarm probability P. The calculation process is as follows:

[0043]

[0044] Among them, w1, w2, w3, w4, and w5 are the weight parameters of each input feature, and b is the bias term;

[0045] Based on the historical alarm event data and normal environment data, optimize the weight parameters of each input feature of the logistic regression with the goal of minimizing the cross-entropy loss. The trained weight vector and bias term are solidified into the edge gateway flash memory, and differential parameter update packets are received through a low-power wide-area communication channel every month;

[0046] Input the latest collected abnormal movement probability, abnormal voiceprint probability, and normalized vibration energy feature into the security alarm model to obtain the current alarm probability P.

[0047] A further improvement of the technical solution of the present invention lies in: in the intelligent security management module, the process of performing hierarchical alarm based on the analysis result of the security alarm model includes:

[0048] According to the analysis results of the security alarm model, different alarm levels are divided, namely high-risk alarm level, medium-risk alarm level and low-risk alarm level, and corresponding alarm probability threshold ranges are matched for each alarm level;

[0049] If the alarm probability is within the alarm probability threshold range of the high-risk alarm level, it is determined as an intrusion event, triggering an audible and visual alarm and pushing the real-time position coordinates to the cloud. If the alarm probability is within the alarm probability threshold range of the medium-risk alarm level, the local buzzer is activated to sound intermittently and the gateway LED flashes. If the alarm probability is within the alarm probability threshold range of the low-risk alarm level, the event log is stored. The event log includes the timestamp, sensor location and alarm probability. The environmental baseline is statistically calculated every 24 hours, and the moving average and standard deviation of the alarm probability in the past 24 hours are calculated to adaptively adjust the alarm probability threshold.

[0050] A further improvement of the technical solution of the present invention lies in: in the intelligent security management module, the process of using wireless networking technology to publish security alarm information and presenting the distribution of regional security events in the form of a heat map includes:

[0051] Each sensor node is deployed through a self-organizing network protocol, star topology and hybrid networking architecture, and the alarm event data including position coordinates, timestamp and alarm probability is transmitted to the edge gateway. High-risk alarms are directly transmitted to the cloud through a low-power wide-area communication channel, with a delay not exceeding 200 ms. Medium-risk alarms are transmitted to the gateway for aggregation through the self-organizing network protocol in multiple hops, and forward error correction coding is enabled. Low-risk alarms are batch uploaded to the cloud after being stored locally for 24 hours;

[0052] The intelligent security management module calibrates the timestamp of each sensor node based on the precise time protocol. The position coordinates of the sensor nodes are pre-stored in the cloud database, and the discrete alarm events are mapped to the building floor grid to calculate the event density D(x,y) of each grid point. The calculation process is as follows:

[0053]

[0054] Among them, N is the total number of alarm events in the current period, P i is the alarm probability of the i-th event, K is the kernel function, and h is the bandwidth parameter automatically adjusted according to the sparsity of the event spatial distribution. If the event spacing exceeds 5 meters, then h = 2 meters; if the event spacing is within 2 meters, then h = 1 meter;

[0055] Normalize the event density of each grid point to the interval [0, 1], set the upper and lower limits of the normalized density threshold. If the event density of a grid point exceeds the upper limit of the normalized density threshold, it is determined as high density, and an RGB gradient from dark red to purple is used. If the event density of a grid point is within the range of the upper and lower limits of the normalized density threshold, it is determined as medium density, and an RGB gradient from orange to red is used. If the event density of a grid point is lower than the lower limit of the normalized density threshold, it is determined as low density, and an RGB gradient from light green to yellow is used. Based on WebGL technology, render the grid density value as a heat map layer and overlay it on the building floor plan. The client pulls the latest density matrix from the cloud every 10 seconds;

[0056] The intelligent security management module selects the optimal data transmission path through a dynamic routing algorithm, weighted by signal strength and node power. It combines edge computing nodes to perform local density pre-computation and data compression, and adopts a dual-path concurrent method of low-power wide-area communication and self-organizing network to synchronously implement the energy efficiency management strategy. It dynamically adjusts the sensor sampling rate and node sleep cycle according to the event density, and realizes network self-healing based on the heartbeat detection mechanism.

[0057] In a second aspect, an intelligent security alarm method based on wireless networking is implemented based on an intelligent security alarm system based on wireless networking according to any one of the above claims 1-9, and includes the following steps:

[0058] S1. Collect and preprocess the human body movement infrared signal strength data, environmental sound waveform data, and three-dimensional vibration waveform data;

[0059] S2. Extract features from the preprocessed human body movement infrared signal strength data, environmental sound waveform data, and three-dimensional vibration waveform data, and construct a security feature sequence table;

[0060] S3. Based on the security feature sequence table, construct an abnormal movement detection model and output the abnormal movement probability;

[0061] S4. Based on the security feature sequence table, construct an abnormal sound recognition model and output the abnormal voiceprint probability;

[0062] S5. Based on the security feature sequence table and the analysis results of the security anomaly detection module, construct a security alarm model and output the alarm probability;

[0063] S6. Perform hierarchical alarm based on the analysis results of the security alarm model, use wireless networking technology to publish security alarm information, and present the distribution of regional security events in the form of a heat map.

[0064] Due to the adoption of the above technical solutions, the technical progress achieved by the present invention compared with the prior art is:

[0065] 1. The present invention provides an intelligent security alarm system and method based on wireless networking. Through multi-modal data collection and feature extraction, the accuracy and comprehensiveness of security monitoring are effectively improved, abnormal behaviors and environmental changes can be identified more accurately, and the probabilities of false alarms and missed alarms are reduced.

[0066] 2. The present invention provides an intelligent security alarm system and method based on wireless networking. By adopting wireless networking technology, the complexity of system deployment and maintenance is simplified, the flexibility and scalability of the system are enhanced, and it can adapt to security requirements of different scales and environments.

[0067] 3. The present invention provides an intelligent security alarm system and method based on wireless networking. Through the intelligent security management module, hierarchical alarm and heat map display are realized, providing intuitive and efficient monitoring means for security personnel, facilitating rapid response and handling of security incidents, and improving the overall security efficiency and safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0069] Figure 1 It is a block diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0070] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0071] Embodiment 1, as Figure 1 shown, the present invention provides an intelligent security alarm system based on wireless networking, including a security data collection module, a security feature extraction module, a security anomaly detection module, a security alarm module, and an intelligent security management module;

[0072] Security data acquisition module, which acquires and preprocesses human body movement infrared signal intensity data, environmental sound waveform data, and three-dimensional vibration waveform data. It installs pyroelectric sensor nodes at building entrances, corridors, and key perimeter areas using a self-organizing network protocol to acquire human body movement infrared signal intensity data. Among them, the distance between pyroelectric sensor nodes does not exceed 8 meters, the vertical installation height is 1.2 - 1.5 meters, and the tilt angle is between 15 degrees and 30 degrees. It automatically constructs a mesh topology through a dynamic routing algorithm. It deploys four-element microphone arrays along the diagonal of the building interior space and under the outdoor eaves to acquire environmental sound waveform data. Among them, the distance between four-element microphone arrays is 1.5 meters, and the distance from the ground is 2.2 to 2.5 meters. It is directly connected to the gateway through a star topology. Four-element microphone array nodes use the Carrier Sense Multiple Access with Collision Avoidance (CSMA / CA) mechanism to compete for the channel. It deploys three-axis piezoelectric accelerometers on the door and window frames and the base of the safe. The axes of the three-axis piezoelectric accelerometers are perpendicular to the monitoring plane, fixed with epoxy resin glue, and connected through a hybrid networking architecture to acquire three-dimensional vibration waveform data. It preprocesses the acquired human body movement infrared signal intensity data, environmental sound waveform data, and three-dimensional vibration waveform data. When the change rate of the human body movement infrared signal intensity is detected to exceed 0.5 microwatts per square centimeter per second, it activates the historical data record, generates a 10-hertz time series, calculates the arithmetic mean of five consecutive sampling points, replaces the current point data, eliminates instantaneous interference pulses with a duration less than 0.5 seconds, calculates the signal energy of the environmental sound waveform data within each 10-millisecond window. When the energy in three consecutive windows exceeds three times the silent baseline, it is determined as the starting point of the valid segment, and when the energy drops below the baseline for more than 50 milliseconds, it is determined as the termination point. It divides the sampling point values within the valid segment by the maximum absolute value of the original data to normalize its maximum absolute value to 1. Three-axis piezoelectric accelerometer nodes calibrate the local clock based on the Precision Time Protocol to make the deviation between the three-dimensional vibration waveform data and the infrared signal timestamp less than 5 milliseconds. According to the readings of the built-in temperature sensor, it adjusts the zero offset of the three-axis acceleration according to the preset temperature drift coefficient;

[0073] The security feature extraction module extracts features from the pre - processed human body moving infrared signal intensity data, environmental sound waveform data, and three - dimensional vibration waveform data, constructs a security feature sequence table. Divide the difference between two adjacent sampling points of the human body moving infrared signal intensity by the sampling interval to obtain the infrared intensity change rate sequence. Perform short - time Fourier transform on the normalized effective audio segment, divide a window every 25 milliseconds, with adjacent windows overlapping by 12.5 milliseconds. Divide the sum of the squared amplitudes of the frequency points in the 500 - 2000 Hz frequency band by the sum of the squared amplitudes of the frequency points in the 0 - 8000 Hz frequency band to obtain the audio band energy ratio within the calculation window. Pass the effective audio segment through a set of 40 triangular filters distributed on the Mel scale. The output of the triangular filter is the logarithm of the signal energy in the corresponding frequency band. Perform discrete cosine transform on the logarithms of the signal energies in the corresponding frequency bands output by the 40 triangular filters, and take the first three - order Mel - cepstral coefficients as feature components. Calculate the total vibration energy integral E of the three - dimensional vibration waveform data within a 1 - second window, and perform dynamic baseline normalization on the total energy integral to obtain the normalized vibration energy The calculation process is as follows:

[0074]

[0075] Where a x (i), a y (i) and a z (i) represent the accelerations at the i - th sampling point on the x - axis, y - axis, and z - axis respectively, Δt is the sampling interval time, N is the number of sampling points, μ is the average vibration energy in the past 1 hour, σ is the standard deviation of the vibration energy in the past 1 hour. Integrate the extracted infrared intensity change rate sequence, audio band energy ratio, the first three - order Mel - cepstral coefficients, and the normalized vibration energy to form a security feature sequence table;

[0076] The security anomaly detection module is used to combine the security feature sequence table for abnormal movement detection and abnormal sound recognition, including a movement detection unit and a sound recognition unit. The movement detection unit is based on the isolation forest algorithm, trains decision trees using historical normal infrared intensity change rate sequence data, and pre - sets the number and depth of the decision trees in the edge gateway of the intelligent security alarm system. Each decision tree selects the mean and variance of the normal infrared intensity change rate in the window as the splitting features. The gateway receives the latest window mean and variance, inputs them into the isolation forest model to construct an abnormal movement detection model, traverses the decision trees, calculates the baseline path length c(n) and the average path length L of the sample in the decision tree p , and after standardization, it is converted into an abnormal movement probability P1. The calculation process is as follows:

[0077]

[0078] Among them, n is the number of samples in the training dataset, γ is the Euler-Mascheroni constant, P1 ∈ [0, 1]. The voice recognition unit performs maximum-minimum normalization on the energy ratio feature of the 500-2000 Hz audio frequency band and standard deviation normalization on the first three-order Mel cepstral coefficients, constructs a 4D voiceprint feature vector including the energy ratio of the audio frequency band, the first-order Mel coefficient, the second-order Mel coefficient, and the third-order Mel coefficient. Based on the support vector machine classification model architecture, the radial basis kernel function is selected to map the 4D voiceprint feature vector to a high-dimensional space. The historical environmental sound waveform dataset is input into the support vector machine classification model architecture to solve the optimal hyperplane. The sample points closest to the hyperplane are selected as the support vectors β, and the trained hyperplane parameters are preset to the edge gateway of the intelligent security alarm system to construct an abnormal sound recognition model. The 4D voiceprint feature vector is input into the abnormal sound recognition model, and the distance d from it to the hyperplane is calculated. The Sigmoid function is used to map this distance to the abnormal voiceprint probability P2. The calculation process is as follows:

[0079]

[0080]

[0081] Among them, α i is the support vector weight, y i is the support vector label, normal = 1, abnormal = -1, K is the calculation result of the radial basis kernel function, b is the bias term of the classification hyperplane, P2 ∈ [0, 1]. Set the upper and lower limits of the threshold range of the abnormal voiceprint probability, and adopt a hierarchical transmission strategy for data transmission;

[0082] The security alarm module constructs a security alarm model based on the security feature sequence list and the analysis results of the security anomaly detection module, and outputs the alarm probability. The security alarm module receives the abnormal movement probability, the abnormal voiceprint probability, and the normalized vibration energy feature, and integrates them into a joint feature vector Adopt the logical regression model architecture to construct a security alarm model and calculate the alarm probability P. The calculation process is as follows:

[0083]

[0084] Among them, w1, w2, w3, w4, and w5 are the weight parameters of each input feature, b is the bias term. Based on the historical alarm event data and normal environment data, the weight parameters of each input feature of the logical regression are optimized with the goal of minimizing the cross-entropy loss. The trained weight vector and bias term are solidified to the flash memory of the edge gateway. Receive the differential parameter update packet through the low-power wide-area communication channel every month, input the latest collected abnormal movement probability, abnormal voiceprint probability, and normalized vibration energy feature into the security alarm model, and obtain the current alarm probability P;

[0085] The intelligent security management module conducts hierarchical alarms based on the analysis results of the security alarm model, uses wireless networking technology to publish security alarm information, and presents the distribution of regional security events in the form of a heat map. It divides different alarm levels according to the analysis results of the security alarm model, namely high-risk alarm level, medium-risk alarm level, and low-risk alarm level, and matches corresponding alarm probability threshold ranges for each alarm level. If the alarm probability is within the alarm probability threshold range of the high-risk alarm level, it is determined as an intrusion event, triggering an audible and visual alarm and pushing the real-time position coordinates to the cloud. If the alarm probability is within the alarm probability threshold range of the medium-risk alarm level, it activates the local buzzer to sound intermittently and the gateway LED to flash. If the alarm probability is within the alarm probability threshold range of the low-risk alarm level, it stores the event log, and the event log includes the timestamp, sensor location, and alarm probability. It statistically calculates the environmental baseline every 24 hours, calculates the moving average and standard deviation of the alarm probability in the past 24 hours, and adaptively adjusts the alarm probability threshold. It deploys each sensor node through the ad hoc network protocol, star topology, and hybrid networking architecture, and transmits the alarm event data including the position coordinates, timestamp, and alarm probability to the edge gateway. High-risk alarms are directly transmitted to the cloud through the low-power wide-area communication channel with a latency of no more than 200 ms. Medium-risk alarms are transmitted to the gateway for aggregation through the ad hoc network protocol in multiple hops, and forward error correction coding is enabled. Low-risk alarms are batch uploaded to the cloud after being stored locally for 24 hours. The intelligent security management module calibrates the timestamp of each sensor node based on the precise time protocol. The position coordinates of the sensor nodes are pre-stored in the cloud database, and the discrete alarm events are mapped to the building floor grid to calculate the event density D(x, y) of each grid point. The calculation process is as follows:

[0086]

[0087] Among them, N is the total number of alarm events in the current period, P iLet \(P_i\) be the alarm probability for the \(i\)-th event, \(K\) be the kernel function, and \(h\) be the bandwidth parameter automatically adjusted according to the sparsity of the event space distribution. If the event spacing exceeds 5 meters, then \(h = 2\) meters; if the event spacing is within 2 meters, then \(h = 1\) meter. Normalize the event density of each grid point to the interval \([0, 1]\), set the upper and lower limits of the normalized density threshold. If the event density of a grid point exceeds the upper limit of the normalized density threshold, it is determined as high density, and an RGB gradient from dark red to purple is used. If the event density of a grid point is within the range of the upper and lower limits of the normalized density threshold, it is determined as medium density, and an RGB gradient from orange to red is used. If the event density of a grid point is lower than the lower limit of the normalized density threshold, it is determined as low density, and an RGB gradient from light green to yellow is used. Render the grid density value as a heat map layer based on WebGL technology and overlay it on the building floor plan. The client pulls the latest density matrix from the cloud every 10 seconds. The intelligent security management module selects the optimal data transmission path based on the signal strength and node power through a dynamic routing algorithm, combines with edge computing nodes to perform local density pre-calculation and data compression, and synchronously implements an energy efficiency management strategy in a way of concurrent dual paths of low-power wide-area communication and self-organizing network. Dynamically adjust the sensor sampling rate and node sleep cycle according to the event density, and achieve network self-healing based on the heartbeat detection mechanism.

[0088] Embodiment 2, as Figure 1 shown, based on Embodiment 1, the present invention provides a technical solution: an intelligent security alarm method based on wireless networking, which is implemented based on the above-mentioned intelligent security alarm system based on wireless networking, and includes the following steps:

[0089] S1. Collect and preprocess the human body movement infrared signal intensity data, environmental sound waveform data, and three-dimensional vibration waveform data;

[0090] S2. Extract features from the preprocessed human body movement infrared signal intensity data, environmental sound waveform data, and three-dimensional vibration waveform data, and construct a security feature sequence table;

[0091] S3. Based on the security feature sequence table, construct an abnormal movement detection model and output the abnormal movement probability;

[0092] S4. Based on the security feature sequence table, construct an abnormal sound recognition model and output the abnormal voiceprint probability;

[0093] S5. Based on the security feature sequence table and the analysis results of the security anomaly detection module, construct a security alarm model and output the alarm probability;

[0094] S6. Perform hierarchical alarm based on the analysis results of the security alarm model, use wireless networking technology to publish security alarm information, and present the regional security event distribution in the form of a heat map.

[0095] As described above, it is only the specific implementation manner of this application. However, the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application. Therefore, the protection scope of this application shall be subject to the protection scope of the claims described.

Claims

1. An intelligent security alarm system and method based on wireless networking, characterized in that: Including security data acquisition module, security feature extraction module, security anomaly detection module, security alarm module and intelligent security management module; The security data acquisition module collects and preprocesses human movement infrared signal strength data, environmental sound waveform data and three-dimensional vibration waveform data; The security feature extraction module extracts features from the pre-processed human movement infrared signal strength data, environmental sound waveform data, and three-dimensional vibration waveform data to construct a security feature sequence table; The security anomaly detection module is used to perform abnormal movement detection and abnormal sound recognition in combination with the security feature sequence table; The security alarm module constructs a security alarm model based on the security feature sequence table and the analysis results of the security anomaly detection module, and outputs an alarm probability; The intelligent security management module performs graded alarms based on the security alarm model analysis results, uses wireless networking technology to publish security alarm information, and presents the distribution of regional security events in the form of a heat map.

2. According to claim 1, the intelligent security alarm system based on wireless networking is characterized in that: The security anomaly detection module includes a motion detection unit and a sound recognition unit: The movement detection unit constructs an abnormal movement detection model based on the extracted infrared intensity change rate feature and outputs the abnormal movement probability; The sound recognition unit constructs an abnormal sound recognition model based on the audio frequency band energy proportion characteristics and the first three-order Mel-frequency cepstral coefficient characteristics, and outputs the abnormal voiceprint probability.

3. According to claim 2, the intelligent security alarm system based on wireless networking is characterized in that: In the security data acquisition module, the acquisition and preprocessing process of human movement infrared signal strength data, environmental sound waveform data and three-dimensional vibration waveform data includes: Using the self-organizing network protocol, pyroelectric sensor nodes are installed at the building entrances, corridors and key perimeter areas to collect infrared signal strength data of human movement; Deploy a four-unit microphone array on the diagonal of the building's interior space and under the outdoor eaves to collect ambient sound waveform data; Deploy triaxial piezoelectric accelerometers on the door and window frames and safe bases of the building to collect three-dimensional vibration waveform data; The collected human movement infrared signal intensity data, environmental sound waveform data and three-dimensional vibration waveform data are preprocessed. When the intensity change rate of the human movement infrared signal is detected to exceed 0.5 microwatts per square centimeter per second, the historical data record is activated, a 10 Hz time series is generated, the environmental sound waveform data is normalized, the normalized effective audio segment is obtained, the three-dimensional vibration waveform data is time-calibrated, and the three-axis acceleration zero point offset is adjusted according to the preset temperature drift coefficient.

4. The intelligent security alarm system based on wireless networking according to claim 3 is characterized in that: In the security feature extraction module, the process of constructing the security feature sequence table includes: Based on the pre-processed human body movement infrared signal strength data, the infrared intensity change rate is obtained by dividing the difference between two adjacent human body movement infrared signal strength sampling points by the sampling interval, and the infrared intensity change rate sequence is combined; Perform short-time Fourier transform on the normalized valid audio segment, divide the calculation window, and then obtain the audio frequency band energy ratio within the calculation window, wherein a window is divided every 25 milliseconds, and adjacent windows overlap for 12.5 milliseconds. The sum of the square amplitudes of the frequency points in the 500-2000 Hz frequency band is divided by the sum of the square amplitudes of the frequency points in the 0-8000 Hz frequency band, so as to obtain the audio frequency band energy ratio within the calculation window; The effective audio segment is passed through a triangular filter bank with 40 Mel scale distributions. The triangular filter outputs the logarithm of the signal energy of the corresponding frequency band. The logarithm of the signal energy of the corresponding frequency band output by the 40 triangular filters is subjected to discrete cosine transform, and the first three order Mel cepstral coefficients are taken as feature components. Calculate the total vibration energy integral E of the three-dimensional vibration waveform data within a 1-second window, perform dynamic baseline normalization on the total energy integral, and obtain the normalized vibration energy The extracted infrared intensity change rate sequence, audio frequency band energy proportion, first three-order Mel-frequency cepstral coefficients and normalized vibration energy are integrated to form a security feature sequence table.

5. The intelligent security alarm system based on wireless networking according to claim 4 is characterized in that: In the movement detection unit, the process of constructing an abnormal movement detection model and outputting the abnormal movement probability includes: Based on the isolation forest algorithm, the historical normal infrared intensity change rate sequence data is used to train the decision tree. The number and depth of the decision trees are preset on the edge gateway of the intelligent security alarm system. Each decision tree selects the mean and variance of the window normal infrared intensity change rate as split features. The gateway receives the latest window mean and variance, inputs them into the isolation forest model, and builds an abnormal movement detection model. Traverse the decision tree and calculate the baseline path length c(n) and the average path length L of the sample in the decision tree p , and is converted into the abnormal movement probability P1 after standardization.

6. The intelligent security alarm system based on wireless networking according to claim 5 is characterized in that: In the voice recognition unit, the process of constructing an abnormal voice recognition model and outputting the abnormal voiceprint probability includes: The energy proportion feature of the 500-2000 Hz audio frequency band is normalized to the maximum and minimum, and the standard deviation of the first three order Mel cepstral coefficients is normalized to construct a 4-dimensional voiceprint feature vector including the energy proportion of the audio frequency band, the first order Mel coefficient, the second order Mel coefficient, and the third order Mel coefficient; Based on the support vector machine classification model architecture, the radial basis kernel function is selected to map the 4-dimensional voiceprint feature vector to a high-dimensional space. The historical environmental sound waveform dataset is input into the support vector machine classification model architecture to solve the optimal hyperplane. The sample point closest to the hyperplane is selected as the support vector β. The trained hyperplane parameters are preset to the edge gateway of the intelligent security alarm system to build an abnormal sound recognition model. Input the 4D voiceprint feature vector to the abnormal sound recognition model, calculate its distance d from the hyperplane, use the Sigmoid function to map the distance to the abnormal voiceprint probability P2, set the upper and lower limits of the threshold range of the abnormal voiceprint probability, and use a hierarchical transmission strategy for data transmission.

7. The intelligent security alarm system based on wireless networking according to claim 6 is characterized in that: In the security alarm module, the process of constructing a security alarm model and outputting an alarm probability includes: The security alarm module receives the abnormal movement probability, abnormal voiceprint probability and normalized vibration energy features, integrates them into a joint feature vector, uses the logistic regression model architecture, builds a security alarm model, and calculates the alarm probability P; Based on historical alarm event data and normal environment data, the weight parameters of each input feature of logistic regression are optimized with the goal of minimizing cross entropy loss. The trained weight vector and bias item are solidified in the flash memory of the edge gateway, and the differential parameter update package is received every month through the low-power wide-area communication channel; The latest collected abnormal movement probability, abnormal voiceprint probability and normalized vibration energy characteristics are input into the security alarm model to obtain the current alarm probability P.

8. The intelligent security alarm system based on wireless networking according to claim 7 is characterized in that: In the intelligent security management module, the process of performing graded alarm based on the security alarm model analysis results includes: According to the analysis results of the security alarm model, different alarm levels are divided into high-risk alarm level, medium-risk alarm level and low-risk alarm level, and the corresponding alarm probability threshold range is matched for each alarm level; If the alarm probability is within the alarm probability threshold range of the high-risk alarm level, it is determined to be an intrusion event, the sound and light alarm is triggered, and the real-time location coordinates are pushed to the cloud. If the alarm probability is within the alarm probability threshold range of the medium-risk alarm level, the local buzzer is started to sound intermittently and the gateway LED flashes. If the alarm probability is within the alarm probability threshold range of the low-risk alarm level, the event log is stored. The event log contains the timestamp, sensor location and alarm probability. The environmental baseline is counted every 24 hours, the moving average and standard deviation of the alarm probability in the past 24 hours are calculated, and the alarm probability threshold is adaptively adjusted.

9. The intelligent security alarm system based on wireless networking according to claim 8 is characterized in that: In the intelligent security management module, the process of using wireless networking technology to publish security alarm information and presenting the distribution of regional security events in the form of a heat map includes: Each sensor node is deployed through self-organizing network protocol, star topology and hybrid network architecture, and the alarm event data including location coordinates, timestamp and alarm probability are transmitted to the edge gateway. The high-risk alarm level is directly transmitted to the cloud through the low-power wide-area communication channel with a delay of no more than 200ms. The medium-risk alarm level is transmitted to the gateway aggregation through the self-organizing network protocol in multiple hops, and forward error correction coding is enabled. The low-risk alarm level is uploaded to the cloud in batches after being stored locally for 24 hours. The intelligent security management module calibrates the timestamp of each sensor node based on the precise time protocol. The location coordinates of the sensor nodes are pre-stored in the cloud database, and the discrete alarm events are mapped to the building plane grid to calculate the event density D(x,y) of each grid point. Normalize the event density of each grid point to the interval [0,1], set the upper and lower limits of the normalized density threshold, if the event density of the grid point exceeds the upper limit of the normalized density threshold, it is judged as high density, using a dark red to purple RGB gradient, if the event density of the grid point is within the range of the upper and lower limits of the normalized density threshold, it is judged as medium density, using an orange to red RGB gradient, if the event density of the grid point is lower than the lower limit of the normalized density threshold, it is judged as low density, using a light green to yellow RGB gradient, based on WebGL technology, the grid density value is rendered as a thermal layer, superimposed on the building plan, and the client pulls the latest density matrix from the cloud every 10 seconds; The intelligent security management module uses a dynamic routing algorithm to select the optimal data transmission path based on signal strength and node power weighting, combines edge computing nodes to perform local density pre-calculation and data compression, and adopts a dual-path concurrent approach of low-power wide-area communication and self-organizing network to simultaneously implement energy efficiency management strategies, dynamically adjust the sensor sampling rate and node sleep cycle based on event density, and achieve network self-healing based on the heartbeat detection mechanism.

10. An intelligent security alarm method based on wireless networking, implemented based on an intelligent security alarm system based on wireless networking as claimed in any one of claims 1 to 9, characterized in that: The following steps are involved: S1, collecting and preprocessing human body movement infrared signal strength data, environmental sound waveform data and three-dimensional vibration waveform data; S2, extracting features from the pre-processed human movement infrared signal strength data, environmental sound waveform data, and three-dimensional vibration waveform data, and constructing a security feature sequence table; S3. Based on the security feature sequence table, an abnormal movement detection model is constructed to output the abnormal movement probability; S4. Based on the security feature sequence table, an abnormal sound recognition model is constructed to output the abnormal voiceprint probability; S5. Based on the analysis results of the security feature sequence table and the security anomaly detection module, a security alarm model is constructed and an alarm probability is output; S6. Perform graded alarms based on the analysis results of the security alarm model, use wireless networking technology to publish security alarm information, and present the distribution of regional security events in the form of a heat map.

Images