Campus bullying prevention system based on voice recognition and big data analysis

Abstract

The invention relates to the technical field of information technology and artificial intelligence application, and discloses a campus anti-bullying system based on voice recognition and big data analysis. Comprising a front-end device, a central processing unit, a voice communication real-time talkback module, a prevention reminding module, a disposal reminding module and a user report interface, wherein the front-end device further comprises an AI voice recognition terminal device, a microphone and a sensor module and is used for monitoring environment sound in a monitoring blind area in a campus in real time. By integrating AI voice recognition, a deep learning algorithm and a voiceprint recognition technology, voice features related to campus bullying events can be accurately captured and recognized in a monitoring blind area, and by combining an Ai artificial intelligence large model analysis platform, the system can effectively analyze the grade of the bullying events, intelligently trigger event prevention or event disposal notification pushing, and improve the safety of the bullying events. And the historical event behaviors are subjected to data archiving, so that the accuracy of bullying event identification is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of information technology and artificial intelligence application technology, and specifically to a campus anti-bullying system based on speech recognition and big data analysis. Background Art

[0002] School bullying is a widespread phenomenon around the world, causing serious harm to students' mental and physical health. Although education departments and school administrators in various countries have taken a variety of measures to deal with this problem, such as strengthening communication between teachers and students and establishing psychological counseling mechanisms, in actual implementation, bullying behavior is often difficult to detect and stop in time, especially in blind spots such as toilets, dormitories, locker rooms and remote corridors.

[0003] Deficiencies of existing technologies: Low recognition accuracy: Traditional campus anti-bullying systems mainly rely on technical means such as video surveillance and keyword matching to monitor the campus environment. However, video surveillance cannot be installed in toilets, locker rooms and other areas due to factors such as privacy protection, making these places high-incidence areas for bullying. At the same time, keyword matching technology has significant deficiencies in practical applications. It can only trigger alarms based on preset keywords and cannot distinguish specific contexts and content. This method is prone to false alarms or missed alarms. School administrators are often troubled by a large number of invalid alarms, which increases the management burden and may miss real bullying incidents.

[0004] Difficulty in identifying people involved in the incident: In existing technologies, although bullying incidents can be recorded and confirmed through video surveillance or on-site witnessing, it is still difficult to identify people involved in incidents that occur in blind spots of surveillance. Traditional methods make it difficult to accurately locate the identities of participants, especially when multiple people appear at the same time. It is impossible to effectively determine who the specific participants are and which class they are from, which not only hinders the rapid handling of the incident, but also affects the subsequent investigation and management of the incident.

[0005] To this end, those skilled in the art provide a campus anti-bullying system based on speech recognition and big data analysis to solve the problems raised in the above background technology. Summary of the invention

[0006] In view of the shortcomings of the existing technology, the present invention provides a campus anti-bullying system based on voice recognition and big data analysis, which solves the problems of low recognition accuracy and difficulty in identifying people related to the incident in the existing prevention system.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a campus anti-bullying system based on voice recognition and big data analysis, including a front-end device, a central processing unit, a voice call real-time intercom module, a prevention reminder module, a disposal reminder module, and a user report interface;

[0008] The front-end device also includes an AI voice recognition terminal device, a microphone and a sensor module, which are used to monitor the ambient sound in real time in the monitoring blind spot on campus and upload the data to the central processing unit for processing;

[0009] The central processing unit also includes a natural language processing module, a deep learning algorithm, and a voiceprint recognition module, which are used to receive and analyze the voice data source transmitted by the front-end device;

[0010] The voice call real-time intercom module also includes basic data storage management functions, real-time communication protocols, and data encryption, which are used to handle the connection establishment of the event handling end, thereby improving the efficiency of preventing and handling bullying incidents.

[0011] The prevention and reminder module also includes a real-time monitoring and prevention module and an emergency handling module, which are used to monitor voice data in the campus in real time, a real-time scene sound monitoring and prevention module, a voice behavior abnormality monitoring submodule, and a risk assessment submodule. The information provided by each of the above submodules is used to assess the risk level of each potential event and determine which situations need to be handled. It is used to directly connect with the handling module and cooperate with the central processing unit to identify and prevent bullying incidents.

[0012] The prevention and treatment module also includes a module for receiving prevention reminder instruction data, an instant prevention notification module, and a multi-level authority management module; through instant communication technology, different roles are set to receive prevention information and respond quickly, effectively shortening the time difference from discovery to intervention and improving processing efficiency;

[0013] The user report interface also includes a management platform and a report generation module, which are used to provide a visual operation interface for school administrators to view the status of the monitored area in real time and receive and manage alarm information.

[0014] Preferably, the monitoring blind area includes one or more of access control, playground, passage, corridor, canteen, classroom, supermarket, toilet or remote corridor.

[0015] Preferably, the natural language processing module uses the TF-IDF algorithm for processing, and the calculation formula is: TF-IDF(t,d)=TF(t,d)×IDF(t);

[0016] Among them, TF(t,d) represents the frequency of word t appearing in document d; IDF(t) represents the inverse document frequency of word t.

[0017] Preferably, the deep learning algorithm adopts a bidirectional LSTM algorithm, and its calculation steps are:

[0018] A. Forward LSTM calculation: Forward LSTM from x1 to xT Processing the input sequence sequentially in time steps, it computes the hidden state at each time step t and cell status

[0019] B. Reverse LSTM calculation: Reverse LSTM from x T It processes the input sequence in reverse order of time steps to x1. It computes the hidden state at each time step t and cell status

[0020] Preferably, the voiceprint recognition module uses a composite calculation of a deep neural network and a convolutional neural network, and the steps of using the module are as follows:

[0021] A. Use convolutional neural network to extract time-frequency features from speech signals;

[0022] B. Input the extracted features into a deep neural network for high-dimensional feature learning and classification;

[0023] C. Finally, a vector representing the speaker's characteristics is output for comparison and identification.

[0024] Preferably, the data encryption uses asymmetric TLS / SSL to encrypt the data.

[0025] Preferably, the real-time monitoring and prevention module includes a data transmission unit, which is specifically LoRa, for transmitting information on potential bullying behavior detected during the monitoring process to the security management center or relevant responsible personnel on campus.

[0026] Preferably, the front-end device is connected to the central processing unit via a network, the central processing unit is connected to the database storage management via a network, the central processing unit is electrically connected to the prevention reminder module, and the prevention reminder module is connected to the user report interface via a network.

[0027] The campus anti-bullying method based on speech recognition and big data analysis is used for the campus anti-bullying system based on speech recognition and big data analysis, including the following steps:

[0028] S1. Data collection: The front-end equipment includes AI voice recognition terminal equipment, microphone and sensor modules to continuously monitor the ambient sound in the blind spots of the campus and transmit the captured voice data to the central processing unit in real time;

[0029] S2, data preprocessing: After receiving the voice data, the central processing unit first cleans and formats the data, extracts the voice feature vector, and inputs it into the natural language processing module for further analysis;

[0030] S3. Speech analysis: The natural language processing module uses a preset algorithm to analyze speech data, identify keywords and patterns, and determine whether there is potential dangerous behavior. If potential bullying behavior is identified, the speech data and analysis results will be input into the deep learning algorithm module for more in-depth analysis;

[0031] S4, Pattern Recognition: The deep learning algorithm module is combined with the voiceprint recognition module, and the deep neural network and convolutional neural network are used to further analyze the voice signal to identify specific behavior patterns, event types, and related personnel. The voice call real-time intercom module is enabled at this stage to ensure that when an abnormality is found, a voice call can be made immediately with the event handling team to coordinate and handle it in real time;

[0032] S5. Data transmission: Once suspected bullying behavior is confirmed, the prevention reminder module will send a prevention signal to the campus security management center or relevant responsible persons through the data transmission unit. At the same time, the disposal reminder module will be activated to transmit the prevention information to the designated security management personnel through instant messaging technology, and determine the handling authority of different roles according to the severity of the incident to ensure rapid response and handling;

[0033] S6. Prevention notification: The voice call real-time intercom module continues to support on-site disposal, providing real-time communication, data encryption and recording functions to ensure the safety and effectiveness of the entire intervention process;

[0034] S7. Record archiving: The user report interface provides a visual operation interface through the management platform, allowing school administrators to view the status of the monitoring area in real time, receive and manage alarm information, and the central processing unit stores all relevant voice data, analysis results and prevention information in the database, and generates reports for subsequent action analysis and reference.

[0035] The present invention provides a campus anti-bullying system based on speech recognition and big data analysis.

[0036] Beneficial effects:

[0037] 1. By integrating AI speech recognition, deep learning algorithms and voiceprint recognition technology, the present invention can accurately capture and identify speech features related to campus bullying incidents in monitoring blind spots. Combined with the AI ​​artificial intelligence large model analysis platform, the system can effectively analyze the level of bullying incidents, intelligently trigger event prevention or event handling notification push, and archive historical event behaviors, greatly improving the accuracy of bullying incident identification.

[0038] 2. The present invention provides a powerful management platform that displays the status of the monitored area in real time and provides in-depth security situation analysis for school administrators through automatically generated analysis reports. The platform not only improves the efficiency of campus security management, but also provides strong decision-making support for management to formulate or adjust security strategies through data visualization and report generation functions, thereby effectively improving the security management level of the entire campus. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the system framework of the present invention;

[0040] Figure 2 It is a schematic diagram of the front-end device framework of the present invention;

[0041] Figure 3 It is a schematic diagram of the central processing unit framework of the present invention;

[0042] Figure 4 It is a schematic diagram of the database storage management framework of the present invention;

[0043] Figure 5 This is a schematic diagram of the prevention reminder module framework of the present invention;

[0044] Figure 6 It is a schematic diagram of the user report interface framework of the present invention;

[0045] Figure 7 This is a schematic diagram of the framework of the disposal reminder module of the present invention;

[0046] Figure 8 This is a schematic diagram of the framework of the voice call real-time intercom module of the present invention;

[0047] Fig. 9 It is a schematic diagram of the steps of the present invention. DETAILED DESCRIPTION

[0048] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings 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.

[0049] Embodiment 1:

[0050] Please see attached Figure 1 , an embodiment of the present invention provides a campus anti-bullying system based on voice recognition and big data analysis, including a front-end device, a central processing unit, a voice call real-time intercom module, a prevention reminder module, a disposal reminder module, a user report interface and a database storage management;

[0051] The front-end device is connected to the central processing unit through a network, the central processing unit is connected to the database storage management through a network, the central processing unit is electrically connected to the prevention reminder module, and the prevention reminder module is connected to the prevention disposal module and the user report interface through a network.

[0052] Specifically, the real-time sound collection and preliminary processing functions of the front-end equipment ensure that the system can effectively capture potential bullying behaviors in the monitoring blind spots and provide data support at an early stage. Through efficient voice recognition technology, the system can respond quickly and realize timely monitoring of potential security threats on campus;

[0053] The multi-level analysis capabilities of the central processing unit, combined with voice recognition and big data processing technology, ensure the system's high accuracy and efficiency in processing complex voice data. Through real-time analysis and pattern recognition, the system can quickly and accurately identify potential bullying behaviors, provide strong preventive support, and prevent the incident from further deteriorating;

[0054] Please see attached Figure 8 The voice call real-time intercom module provides reliable data storage and instant messaging processing capabilities, ensuring that all data in the system can be managed securely and effectively. By adopting real-time communication protocols and data encryption technology, this module can not only handle the establishment of connections at the event handling end, but also improve the efficiency of bullying incident prevention and disposal. Combined with the big data analysis platform and the AI ​​artificial intelligence large model analysis platform, the system can effectively analyze the level of bullying incidents, intelligently trigger event prevention or event handling notification push, and archive historical event behaviors, greatly improving the accuracy of bullying incident identification, providing strong data support for school management, and further protecting the privacy of students;

[0055] The prevention reminder module realizes the rapid and extensive transmission of prevention information, effectively covering the entire campus area. Through a variety of signal transmission modes, the system can ensure that prevention information can be promptly conveyed to relevant personnel in any environment, providing strong support for timely intervention and prevention of bullying behavior;

[0056] The user report interface provides an intuitive and easy-to-use operation experience, allowing school administrators to efficiently monitor and manage campus security. Through the visualization of data and the generation of regular reports, management can better grasp the security dynamics on campus, so as to make timely and effective decisions and further improve the overall security level of the campus.

[0057] Please see attached Figure 2The front-end equipment also includes AI voice recognition terminal equipment, microphones and sensor modules, which are used to monitor environmental sounds in real time in the monitoring blind spots on campus and upload data to the central processing unit for processing; the monitoring blind spots include one or more of toilets, dormitories, locker rooms, stairwells or remote corridors.

[0058] Specifically, it is used to monitor the real-time environmental sound in the blind spots of the campus (such as toilets, dormitories, locker rooms, stairwells or remote corridors, etc.), and upload the collected voice data to the central processing unit for further processing. The AI ​​voice recognition terminal device captures subtle environmental sounds through a highly sensitive microphone, and combined with a sensor module, it can perceive and record the sound changes in the surrounding environment in real time. The design of the device takes into account the balance between privacy protection and security monitoring on campus. In areas where cameras cannot be installed, sound monitoring is used to identify and prevent potential bullying behaviors.

[0059] Through the combination of AI voice recognition terminal equipment and microphone and sensor modules, the system can operate effectively in various monitoring blind spots, ensuring that hidden toilets, dormitories, and other unmonitored areas such as locker rooms and remote corridors are fully covered. The front-end equipment can not only identify threatening language, but also capture sound characteristics, identify fighting, calling for help and other behaviors that may be related to bullying. The words involved include but are not limited to "beat you to death", "beat him", "wait for me" and some "swearing" words. The collected data will be uploaded to the central processing unit in real time to provide accurate original data for subsequent data analysis and pattern recognition. Through these precise sound monitoring and transmission functions, the front-end equipment plays a key role in improving the overall recognition accuracy and real-time performance of the system, effectively enhancing the overall prevention ability and response speed of the campus anti-bullying system.

[0060] Please see attached Figure 3 ,The central processing unit also includes a natural language processing module, a deep learning algorithm, and a voiceprint recognition module for receiving and analyzing the voice data transmitted by the front-end device;

[0061] The natural language processing module uses the TF-IDF algorithm for processing, and the calculation formula is: TF-IDF(t,d)=TF(t,d)×IDF(t);

[0062] Among them, TF(t,d) represents the frequency of word t appearing in document d; IDF(t) represents the inverse document frequency of word t.

[0063] The deep learning algorithm uses a bidirectional LSTM algorithm, and its calculation steps are:

[0064] A. Forward LSTM calculation: Forward LSTM from x1 to x TProcessing the input sequence sequentially in time steps, it computes the hidden state at each time step t and cell status

[0065] B. Reverse LSTM calculation: Reverse LSTM from x T It processes the input sequence in reverse order of time steps to x1. It computes the hidden state at each time step t and cell status

[0066] The voiceprint recognition module uses a composite calculation of a deep neural network and a convolutional neural network. The steps for using it are as follows:

[0067] A. Use convolutional neural network to extract time-frequency features from speech signals;

[0068] B. Input the extracted features into a deep neural network for high-dimensional feature learning and classification;

[0069] C. Finally, a vector representing the speaker's characteristics is output for comparison and identification.

[0070] Specifically, the central processing unit is the core computing and analysis center of the system, responsible for receiving and analyzing voice data transmitted by the front-end device. It integrates natural language processing modules, deep learning algorithms, and voiceprint recognition modules. Through the collaborative work of these modules, the system can efficiently and accurately analyze and identify voice data collected in the monitoring blind area, providing strong technical support for real-time monitoring and prevention of campus anti-bullying.

[0071] First, the natural language processing module uses the TF-IDF (term frequency-inverse document frequency) algorithm to process and analyze the text content in the speech data. The TF-IDF algorithm is a statistical method used to evaluate the importance of a word in a document collection. It is calculated as follows: The TF-IDF value consists of two parts, namely term frequency (TF) and inverse document frequency (IDF).

[0072] In the processing and analysis of voice data, the central processing unit further uses deep learning algorithms, especially the bidirectional LSTM (Long Short-Term Memory Network) algorithm, to analyze the time series data in the voice. The bidirectional LSTM can simultaneously consider the contextual information of the input sequence, which is divided into two LSTM processing processes: forward and reverse. Through the forward and reverse processing of the bidirectional LSTM, the system can comprehensively analyze the contextual information in the voice, more accurately identify potential bullying behaviors or emergencies, and enhance the system's semantic understanding ability in complex contexts.

[0073] In addition, the voiceprint recognition module of the central processing unit uses a composite calculation method of deep neural network (DNN) and convolutional neural network (CNN), which is specifically used to identify and confirm the identity of the speaker. Through the collaborative work of the above modules, the central processing unit can conduct in-depth analysis of the voice data from the front-end device, not only accurately identify potential bullying behaviors, but also lock and identify information related to class, number of people and location in the voice content, providing comprehensive technical support for campus anti-bullying. The efficient operation of these modules greatly improves the overall recognition accuracy and prevention response speed of the system.

[0074] Please see attached Figure 4 ,Database storage management also includes a big data analysis platform,data encryption, which is used to store and analyze the collected voice data;,data encryption uses asymmetric TLS / SSL to encrypt data;

[0075] Specifically, the big data analysis platform is one of the core components of this module, which is used to store, process and analyze large amounts of voice data. The platform can process massive amounts of data and use advanced algorithms to mine potential patterns and trends. For example, it can analyze voice data from different time periods and regions to identify areas and times where high-frequency bullying incidents occur, thereby providing data support to school management and helping them optimize safety precautions. In addition, the big data analysis platform can also predict possible bullying incidents in the future by reviewing historical data, helping schools take preventive measures in advance.

[0076] In order to ensure data security during storage and transmission, the data encryption mechanism uses asymmetric TLS / SSL (Transport Layer Security Protocol / Secure Sockets Layer Protocol) technology to encrypt data. Specifically, TLS / SSL encryption technology uses asymmetric encryption algorithms to ensure the confidentiality and integrity of data during transmission. In the initial stage of data transmission, the system uses the public key to encrypt the symmetric key, and the recipient uses the private key to decrypt it, thereby securely establishing an encrypted channel. After that, all data transmission is carried out within this secure channel, and symmetric key encryption is used to encrypt the actual data transmission content, which not only improves transmission efficiency, but also ensures that the data will not be eavesdropped or tampered with during the entire transmission process.

[0077] Please see attached Figure 5 ,The prevention reminder module also includes a real-time monitoring and prevention module, an ,emergency handling module for real-time monitoring of voice data in the campus, a real-time ,scene sound monitoring and prevention module, a voice behavior abnormality monitoring ,submodule, and a risk assessment submodule. The information provided by the ,above submodules is used to assess the risk level of each potential event and ,determine which situations need to be handled and are used to directly connect ,with the handling module to cooperate with the central processing unit to ,identify and prevent bullying incidents;

[0078] Specifically, when the central processing unit detects suspected bullying behavior through natural language processing, deep learning algorithm and voiceprint recognition module, the real-time monitoring and prevention module is immediately activated. After calculation by the real-time scene sound monitoring and prevention module, the voice behavior abnormality monitoring sub-module and the risk assessment sub-module, this key information is quickly transmitted to the security management center or relevant responsible personnel on campus through the built-in data transmission unit.

[0079] It is worth noting that real-time monitoring transmission. LoRa is a long-distance, low-power wireless communication technology that is very suitable for information transmission in a large area such as a campus. Through LoRa technology, information on potential bullying behaviors detected during monitoring can be reliably transmitted to various security nodes on campus, whether it is a security management center that is far away or the relevant personnel directly responsible for security, they can receive alarms in the shortest time.

[0080] The core of this design lies in the long-distance transmission capability and low power consumption of LoRa, which enables prevention information to be accurately and timely transmitted even in areas with poor network coverage or long distances. This ensures the efficient operation of the system in various complex environments. No matter how large the campus is or how complex the building layout is, the system can maintain stable prevention functions.

[0081] Through the efficient operation of the real-time monitoring and prevention module, combined with the long-distance transmission capability of the LoRa data transmitter unit, the system can immediately issue an alarm when bullying behavior begins to emerge, minimizing the impact of the incident.

[0082] Please see attached Figure 7 The prevention and disposal module also includes a module for receiving expected reminder instruction data, an instant prevention notification module, and a multi-level authority management module; through instant communication technology, different roles can be set to quickly respond to the authority of receiving prevention information, which effectively shortens the time difference from discovery to intervention and improves processing efficiency;

[0083] Please see attached Figure 6 ,The user report interface also includes a management platform and a report ,generation module, which is used to provide school administrators with a visual ,operation interface to view the status of the monitoring area in real time, ,and receive and manage alarm information.

[0084] Specifically, through the management platform, managers can view the status of all monitoring areas on campus in real time, including the current voice data monitoring situation, triggered alarm information, and the security situation of each area. The management platform presents this information in a graphical and data-visualized manner, allowing managers to intuitively understand and grasp the security status of the entire campus. At the same time, the platform also allows users to customize settings as needed, such as adjusting monitoring sensitivity, configuring alarm reception methods, etc., to meet security management needs in different scenarios.

[0085] The report generation module is another key function in the user report interface, which is used to automatically generate various analysis reports to provide data support for school management decisions. The report generation module can summarize the monitoring data, alarm records and processing status in the system regularly or on demand to generate detailed security analysis reports. These reports not only include daily security monitoring, but also provide trend analysis and prediction of potential security risks on campus through the support of the big data analysis platform. Through these reports, managers can quickly understand the changes in the security situation on campus, identify high-risk areas or time periods, and formulate or adjust security management strategies accordingly.

[0086] Embodiment 2:

[0087] Please see attached Figure 7 , an embodiment of the present invention provides a campus anti-bullying method based on speech recognition and big data analysis, comprising the following steps:

[0088] S1. Data collection: The front-end equipment includes AI voice recognition terminal equipment, microphone and sensor modules to continuously monitor the ambient sound in the blind spots of the campus and transmit the captured voice data to the central processing unit in real time;

[0089] S2, data preprocessing: After receiving the voice data, the central processing unit first cleans and formats the data, extracts the voice feature vector, and inputs it into the natural language processing module for further analysis;

[0090] S3. Speech analysis: The natural language processing module uses a preset algorithm to analyze speech data, identify keywords and patterns, and determine whether there is potential dangerous behavior. If potential bullying behavior is identified, the speech data and analysis results will be input into the deep learning algorithm module for more in-depth analysis;

[0091] S4, Pattern Recognition: The deep learning algorithm module is combined with the voiceprint recognition module, and the deep neural network and convolutional neural network are used to further analyze the voice signal to identify specific behavior patterns, event types, and related personnel. The voice call real-time intercom module is enabled at this stage to ensure that when an abnormality is found, a voice call can be made immediately with the event handling team to coordinate and handle it in real time;

[0092] S5. Data transmission: Once suspected bullying behavior is confirmed, the prevention reminder module will send a prevention signal to the campus security management center or relevant responsible persons through the data transmission unit. At the same time, the disposal reminder module will be activated to transmit the prevention information to the designated security management personnel through instant messaging technology, and determine the handling authority of different roles according to the severity of the incident to ensure rapid response and handling;

[0093] S6. Prevention notification: The voice call real-time intercom module continues to support on-site disposal, providing real-time communication, data encryption and recording functions to ensure the safety and effectiveness of the entire intervention process;

[0094] S7. Record archiving: The user report interface provides a visual operation interface through the management platform, allowing school administrators to view the status of the monitoring area in real time, receive and manage alarm information, and the central processing unit stores all relevant voice data, analysis results and prevention information in the database, and generates reports for subsequent action analysis and reference.

[0095] Although embodiments of the present invention have been shown and described, it will be appreciated 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 that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The campus anti-bullying system based on speech recognition and big data analysis is characterized by: It includes front-end equipment, central processing unit, voice call real-time intercom module, prevention reminder module, disposal reminder module, and user report interface; The front-end device also includes an AI voice recognition terminal device, a microphone and a sensor module, which are used to monitor the ambient sound in real time in the monitoring blind spot on campus and upload the data to the central processing unit for processing; The central processing unit also includes a natural language processing module, a deep learning algorithm, and a voiceprint recognition module, which are used to receive and analyze the voice data source transmitted by the front-end device; The voice call real-time intercom module also includes basic data storage management functions, real-time communication protocols, and data encryption, which are used to handle the connection establishment of the event handling end, thereby improving the efficiency of bullying incident prevention and handling; The prevention and reminder module also includes a real-time monitoring and prevention module and an emergency handling module, which are used to monitor voice data in the campus in real time, a real-time scene sound monitoring and prevention module, a voice behavior abnormality monitoring submodule, and a risk assessment submodule. The information provided by each of the above submodules is used to assess the risk level of each potential event and determine which situations need to be handled. It is used to directly connect with the handling module and cooperate with the central processing unit to identify and prevent bullying incidents. The prevention and treatment module also includes a module for receiving prevention reminder instruction data, an instant prevention notification module, and a multi-level authority management module; through instant communication technology, different roles are set to receive prevention information and respond quickly, effectively shortening the time difference from discovery to intervention and improving processing efficiency; The user report interface also includes a management platform and a report generation module, which are used to provide a visual operation interface for school administrators to view the status of the monitored area in real time and receive and manage alarm information.

2. The campus anti-bullying system based on speech recognition and big data analysis according to claim 1 is characterized in that: The monitoring blind area includes one or more of access control, playground, passage, corridor, canteen, classroom, supermarket, toilet or remote corridor.

3. The campus anti-bullying system based on speech recognition and big data analysis according to claim 1 is characterized in that: The natural language processing module uses the TF-IDF algorithm for processing, and the calculation formula is: TF-IDF(t,d)=TF(t,d)×IDF(t); Among them, TF(t,d) represents the frequency of word t appearing in document d; IDF(t) represents the inverse document frequency of word t.

4. The campus anti-bullying system based on speech recognition and big data analysis according to claim 1 is characterized in that: The deep learning algorithm adopts a bidirectional LSTM algorithm, and its calculation steps are: A. Forward LSTM calculation: Forward LSTM from x1 to x T Processing the input sequence sequentially in time steps, it computes the hidden state at each time step t and cell status B. Reverse LSTM calculation: Reverse LSTM from x T It processes the input sequence in reverse order of time steps to x1. It computes the hidden state at each time step t and cell status 5. The campus anti-bullying system based on speech recognition and big data analysis according to claim 1 is characterized in that: The voiceprint recognition module uses a composite calculation of a deep neural network and a convolutional neural network, and its use steps are as follows: A. Use convolutional neural network to extract time-frequency features from speech signals; B. Input the extracted features into a deep neural network for high-dimensional feature learning and classification; C. Finally, a vector representing the speaker's characteristics is output for comparison and identification.

6. The campus anti-bullying system based on speech recognition and big data analysis according to claim 1 is characterized in that: The data encryption uses asymmetric TLS / SSL to encrypt data.

7. The campus anti-bullying system based on speech recognition and big data analysis according to claim 1 is characterized in that: The real-time monitoring and prevention module includes a data transmission unit, which is specifically LoRa, for transmitting information on potential bullying behaviors detected during the monitoring process to the security management center or relevant responsible personnel on campus.

8. The campus anti-bullying system based on speech recognition and big data analysis according to claim 1 is characterized in that: The front-end device is connected to the central processing unit via a network, the central processing unit is connected to the database storage management via a network, the central processing unit is electrically connected to the prevention reminder module, and the prevention reminder module is connected to the user report interface via a network.

9. A campus anti-bullying method based on speech recognition and big data analysis, used in a campus anti-bullying system based on speech recognition and big data analysis as claimed in any one of claims 1 to 8, characterized in that: The following steps are involved: S1. Data collection: The front-end equipment includes AI voice recognition terminal equipment, microphone and sensor modules to continuously monitor the ambient sound in the blind spots of the campus and transmit the captured voice data to the central processing unit in real time; S2, data preprocessing: After receiving the voice data, the central processing unit first cleans and formats the data, extracts the voice feature vector, and inputs it into the natural language processing module for further analysis; S3. Speech analysis: The natural language processing module uses a preset algorithm to analyze speech data, identify keywords and patterns, and determine whether there is potential dangerous behavior. If potential bullying behavior is identified, the speech data and analysis results will be input into the deep learning algorithm module for more in-depth analysis; S4, Pattern Recognition: The deep learning algorithm module is combined with the voiceprint recognition module, and the deep neural network and convolutional neural network are used to further analyze the voice signal to identify specific behavior patterns, event types, and related personnel. The voice call real-time intercom module is enabled at this stage to ensure that when an abnormality is found, a voice call can be made immediately with the event handling team to coordinate and handle it in real time; S5. Data transmission: Once suspected bullying behavior is confirmed, the prevention reminder module will send a prevention signal to the campus security management center or relevant responsible persons through the data transmission unit. At the same time, the disposal reminder module will be activated to transmit the prevention information to the designated security management personnel through instant messaging technology, and determine the handling authority of different roles according to the severity of the incident to ensure rapid response and handling; S6. Prevention notification: The voice call real-time intercom module continues to support on-site disposal, providing real-time communication, data encryption and recording functions to ensure the safety and effectiveness of the entire intervention process; S7. Record archiving: The user report interface provides a visual operation interface through the management platform, allowing school administrators to view the status of the monitoring area in real time, receive and manage alarm information, and the central processing unit stores all relevant voice data, analysis results and prevention information in the database, and generates reports for subsequent action analysis and reference.

Images