KR102573461B1 - Smart artificial intelligence multi-fungtion emergency bell system and its control method - Google Patents

Abstract

The present invention relates to a smart artificial intelligence multifunctional emergency bell system and a control method thereof, and is provided with a transmission/reception module for transmitting and receiving the collected voice from a voice collection module that collects voice through a microphone for receiving a user's voice to an external server emergency bell device; And a voice processing module that converts the voice collected from the emergency bell device into text, an emotion analysis module that analyzes the user's emotion from the extracted keywords and text, and the emotion data detected by the emotion analysis module and a pre-stored service. A control center including a service agency linkage module that compares agency information data and provides the emotion data to matched service agencies to request a service; including, a control center that manages a server; It can be built for the purpose of protecting life and property from accidents, crime prevention, and crime prevention. It can be operated even in a narrow place, so it can request help in an emergency situation and spread the emergency situation. It can be used in poor indoor and outdoor environments. there is.

Description

Smart artificial intelligence multi-function emergency bell system and its control method {SMART ARTIFICIAL INTELLIGENCE MULTI-FUNGTION EMERGENCY BELL SYSTEM AND ITS CONTROL METHOD}

The present invention relates to a smart artificial intelligence multifunctional emergency bell system and a control method thereof, and more particularly, to report an emergency situation in response to an emergency situation and necessary services according to a user's voice information, or to provide various public information and maintain facilities, information It relates to a smart artificial intelligence multi-function emergency bell system that can perform inquiries, commercial advertisements, etc., and a control method thereof.

Recently, emergency bell systems are being installed and operated in local governments, public and private sectors nationwide as an active method to respond to violent crimes targeting unspecified persons in blind spots of surveillance mainly in public places, and the scope of application is gradually expanding.

Until now, CCTV surveillance system has been used as the most common crime prevention method for crime prevention and crime prevention. Have.

In particular, in the case of the CCTV monitoring system, since it is impossible to install it in places where it is difficult to collect images, such as accommodations or bathrooms where women live, there is a problem in that it cannot monitor the space and respond quickly in a criminal situation.

In order to overcome these problems, emergency bell systems are being installed and operated in local governments, public and private sectors across the country as an active method to respond to violent crimes targeting unspecified persons in blind spots among public places, and the scope of application is gradually expanding. there is.

Here, the emergency bell system is a system built for the purpose of protecting human life and property from disasters, disasters, incidents, accidents, crime prevention, and crime prevention. am.

As for the emergency bell system, an IoT emergency bell system introducing IoT technology has been released on the market, and in this case, it is used in connection with a broadband network (communication network) such as 3G and LTE.

On the other hand, unmanned information desks (kiosks) for providing public services in public places are used as automated guidance systems that provide public facility information using artificial intelligence and IoT technology. Examples include city halls in major cities, It is also operated for the purpose of providing information, consulting, reservation payment, and management at stations and large hospitals.

Most of these unmanned information desks provide a visual interface (UI) environment using a kiosk, and as large LCD panels are used to maintain visibility, there is a problem in building and using the system in a narrow place. However, since it is exposed for a long time in a high-temperature and high-humidity environment, it has a problem of frequent failure.

In addition, it is inconvenient for the elderly or foreign tourists who are not accustomed to kiosks and Internet searches, and since the unmanned information desks take pictures of users with cameras such as CCTVs, there is a problem in protecting personal information due to invasion of privacy.

1. Korean Patent Registration No. 10-1434502 (2014.08.20.) 2. Korea Patent Registration No. 10-1832446 (2018.02.20.) 3. Korean Patent Publication No. 10-2022-0064702 (2022.05.19.)

The present invention relates to a smart artificial intelligence multifunctional emergency bell system and a control method thereof, and is intended to provide a multifunctional emergency bell system capable of providing various unmanned public services. In particular, it can be installed in narrow places such as indoor and outdoor toilets, It is intended to provide a smart artificial intelligence emergency bell that can be operated even in harsh environments with high temperature and humidity.

According to the present invention, a smart artificial intelligence multifunctional emergency bell system, an emergency bell device having a transmission/reception module for transmitting and receiving the collected voice from a voice collection module that collects voice through a microphone for receiving a user's voice to an external server; And a voice processing module that converts the voice collected from the emergency bell device into text, an emotion analysis module that analyzes the user's emotion from the extracted keywords and text, and the emotion data detected by the emotion analysis module and a pre-stored service. A control center that manages servers including a control unit including a service agency linking module requesting a service by comparing agency information data and providing the emotion data to a matched service agency.

In addition, the control unit may include a keyword extraction module for extracting keywords from the text converted by the voice processing module and providing the extracted keywords to the emotion analysis module.

In addition, the control unit includes a location information collection module that collects information about the location of the voice information collected by the voice collection module, and includes the keyword extracted from the keyword extraction module and the location collected from the location information collection module. By comparing information values, matched matching data may be provided to the service agency linking module.

In addition, the control unit may include an artificial intelligence dialog module outputting pre-stored additional information to the user as a voice when the keyword value extracted from the keyword extraction module is not a keyword corresponding to the location information value collected from the location information collection module; can include

In addition, the artificial intelligence dialogue module generates dialogue list data by storing the text converted in the voice processing module in chronological order, and converts information on the dialogue list data to the keyword extraction module, the emotion analysis module, and the location information. The keyword result data is generated by receiving the result corresponding to the conversation flow by providing each to the collection module, and the keyword result data is received from the service institution linking module and is matched with the service institution information data stored in advance. Service can be requested by providing result data.

delete

delete

delete

delete

delete

As for the smart artificial intelligence multifunctional emergency bell system and its control method of the present invention, it can exhibit the following effects.

First, according to the smart artificial intelligence multifunctional emergency bell system and its control method of the present invention, it is built for the purpose of protecting life and property from disasters, calamities, incidents, accidents, crime prevention, crime prevention, etc. It can also be operated, so it can request help in an emergency situation and spread the emergency situation, and it has the effect of being used in poor indoor and outdoor environments.

Second, according to the smart artificial intelligence multifunctional emergency bell system and its control method of the present invention, the smart artificial intelligence emergency bell device transmits the user's request to the control center in real time to respond to the user's request. there is

Thirdly, according to the smart artificial intelligence multifunctional emergency bell system and its control method of the present invention, it is possible to maximize efficiency while minimizing on-site manpower for maintenance, and to continuously maintain and manage emergency bell devices as well as malfunction and damage. It can show the effect of minimizing the operational blank time by enabling prompt action on the etc.

Fourth, according to the smart artificial intelligence multi-function emergency bell system and its control method of the present invention, it can be used only by voice without a small LCD panel or LCD panel, enabling miniaturization, securing visibility and having little effect even in harsh environments. There is an effect that can be used without any restrictions.

Fifth, according to the smart artificial intelligence multifunctional emergency bell system and its control method of the present invention, it is possible to solve the privacy invasion problem while providing convenience to users (foreigners, etc.) who are not familiar with digital use.

The effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description of the claims.

1 is a schematic diagram showing a smart artificial intelligence multi-function emergency bell system and a control method thereof according to the present invention.
Figure 2 is a diagram showing the configuration of Figure 1;
Figure 3 is a block diagram showing an embodiment of the smart artificial intelligence multi-function emergency bell system of FIG.
4 is a flowchart illustrating a control method of a smart artificial intelligence multi-function emergency bell system according to an embodiment of FIG. 1 .
5 is a flowchart illustrating a control method according to another embodiment of FIG. 4 .

In describing the present invention, the terms or words used in this specification and claims are based on the principle that the inventor can appropriately define the concept of the term in order to best describe his or her invention. should be interpreted as a meaning and concept that corresponds to the technical idea of

In addition, throughout the specification, when a part is said to be 'connected' to another part, this is not only the case where it is 'directly connected', but also the case where it is 'indirectly connected' with another element in between. include

In addition, 'including' a certain component means that other components may be further included, rather than excluding other components unless otherwise stated.

1 is a schematic diagram showing a smart artificial intelligence multi-function emergency bell system and a control method thereof according to the present invention, FIG. 2 is a diagram showing the configuration of FIG. 1, and FIG. 3 is one of the smart artificial intelligence multi-function emergency bell system of FIG. Figure 4 is a block diagram showing an embodiment, Figure 4 is a flow chart showing a control method of a smart artificial intelligence multi-function emergency bell system according to one embodiment of Figure 1, Figure 5 is a flow chart showing a control method according to another embodiment of Figure 4 am.

Hereinafter, an embodiment of a smart artificial intelligence multifunctional emergency bell system and a control method thereof of the present invention will be described with reference to the accompanying drawings.

Referring to FIGS. 1 to 3, the smart artificial intelligence multifunctional emergency bell system according to the present invention is described. The smart artificial intelligence multifunctional emergency bell system includes a speaker outputting an alarm signal as sound and a microphone for receiving a user's voice. An emergency bell device 10 having a transmission/reception module (B5) for transmitting and receiving the voice collected by the voice collection module (B1) for collecting voice through an external server; And a voice processing module (M1) for receiving the voice collected from the emergency bell device (10) and converting it into text, an emotion analysis module (M3) for analyzing the user's emotion from the extracted keywords and text, and an emotion analysis module (M3) A server including a control unit including a service agency linkage module (M4) that compares the emotion data detected in ) with the service agency information data stored in advance and provides the emotion data to the matched service agency 30 to request a service. It includes a control center 20 that manages.

The emergency bell device 10 is installed in an area easily exposed to crime (hereinafter referred to as a 'crime prevention area'), including public toilets, residences, elevators, etc., to request rescue or alarm in an emergency situation led by the user. While notifying, the service institution 30 is provided through the control center 20, and an emergency call can be made with the control center 20 or the service institution 30.

The emergency bell device 10 includes a voice collection module (B1) that collects user's voice information, a rescue request button exposed to the outside and composed of a capacitive touch sensor, and a rescue request signal generated when the rescue request button is operated. and interlocks with the rescue request module (B2) that transmits it to the selected control center (20), the speaker module (B3) that outputs an alarm by the rescue request signal from the rescue request module (B2), and the rescue request module (B2) and a warning light module (B4) that operates by

In addition, the emergency bell device 10 includes a transmission and reception module (B5) that provides voice transmission and reception with the control center 20, and attempts to connect the call with the person in charge of the abnormal voice determination module (B6) and the control center (20) A call connection module (B7) may be included.

The voice collection module B1 may recognize a user's voice to collect and store voice information. The voice collection module B1 may provide the collected voice information to the control center 20 . The voice collection module (B1) is implemented as an interactive human interface that wirelessly interworks with a specific app of the emergency bell device 10 and can interactively perform voice information of a remote manager.

The voice collection module B1 transmits the voice of a certain section among the user's stored voice information to the control center 20 and partially or entirely deletes the collected voice information, which can be applied by the designer's choice.

The rescue request module (B2) generates a rescue request signal when a rescue request button is composed of a capacitive touch sensor exposed to the outside of the emergency bell device 10 and a rescue request button is operated, and one control center 20 selects the rescue request signal. provided by

The rescue request module (B2) has a rescue request button exposed to the outside and outputs an alarm signal in response to an input signal according to the operation of the rescue request button. The rescue request button may be composed of a physical button or, preferably, a capacitive touch sensor as described above in consideration of minimization of failure and convenience of operation.

The rescue request module B2 may output an emergency bell by user manipulation, generate a rescue request signal, and transmit the signal to the control center 20 . The rescue request module (B2) may include a speaker module (B3) that outputs the alarm signal output by the operation of the rescue request button as a sound, that is, an emergency bell.

The speaker module (B3) includes a speaker that outputs an alarm signal to the emergency bell device 10 as sound, and an alarm signal or a manager's voice of the control center 20 described later can be output through this speaker.

The warning light module (B4) operates together with an emergency bell in conjunction with the alarm signal output from the rescue request module (B2), so that it can be displayed so as to visually inform the surroundings of an emergency situation.

Accordingly, the rescue request module (B2) outputs an alarm signal and a rescue request signal together when the rescue request button is operated. The speaker module (B3) generates an alarm signal and the warning light module (B4) informs an emergency activate the bell The rescue request module B2 transmits the output rescue request signal to the control center 20 through the transmission/reception module B5.

The transmission/reception module (B5) wirelessly connects the emergency bell device 10 and the control center 20, can provide information of the emergency bell device 10 to the control center 20, and also transmits and receives information to the control center 20. It is natural that the transmission/reception module corresponding to the module B5 is provided, and since the same actions and effects are exhibited, a detailed description of the transmission/reception module provided in the control center 20 will be omitted.

For example, the transmission/reception module (B5) transmits information of the voice collection module (B1), the rescue request module (B2), the abnormal voice determination module (B6), and the call connection module (B7) to one selected control center (20), or A control signal may be received from the control center 20 .

The transmission/reception module B5 may be connected through a selected wireless communication network, preferably 3G, 4G, 5G, 3rd Generation Partnership Project (3GPP), Long Term Evolution (LTE), World Interoperability for Microwave Access (WIMAX), Wi-Fi, Bluetooth communication, infrared communication, ultrasonic communication, visible light communication (VLC: Visible Light Communication), LiFi, and the like are included, but are not limited thereto.

The abnormal voice determination module (B6) stores the reference voice frequency upon rescue request learned through machine learning, and when an operating signal is input by user voice recognition, the user's voice frequency used for voice recognition and the pre-stored reference voice The frequency is compared to analyze the similarity between the mutual sound frequencies, and the analysis result may be provided to the control center 20 .

Such an emergency bell device 10 can confirm an emergency situation including a crime by enabling a person in charge of the control center 20 to perform a call. In case of an emergency, the user can be instructed to respond to the first action by nurturing the person in charge.

The call connection module (B7) allows the user to attempt a call with the person in charge of the control center (20). The call connection module B7 may include a call connection button of a physical button.

When the call connection module (B7) presses the call connection button, it is received through the microphone and the user's voice is transmitted to the person in charge of the control center (20) by the transmission and reception module (B5), and the voice of the person in charge of the control center (20) is also transmitted. Conversely, it can be output through the speaker module (B3).

This emergency bell device 10 can be requested (requested) through a user's emergency request or a call with a person in charge of the control center (20).

When the user's voice is recognized and keyword derivation and emotion analysis are completed, the control center 20 may provide the service by identifying contextual awareness and service based on this. The control center 20 continuously accumulates and learns voice information about the collected voice information using artificial intelligence based on deep learning techniques, and compares and judges based on the learned content to determine whether a situation has occurred according to the extracted keyword. can be quickly judged.

For example, the control center 20 stores voice information in a database, and deep learning-based artificial intelligence can improve the ability to extract keywords from voice information or classify keywords.

In the control center 20, the more deep learning-based artificial intelligence learns, the higher the accuracy and reliability of recognition of the fire situation of the building, so in the end, it is not managed by a separate manager and unmanned according to emergency situations and voice information. It may be possible to respond to whether or not a situation occurs.

The control center 20 may monitor requests transmitted from the emergency bell device 10 . The control center 20 can promptly respond to a crime by dispatching management personnel to a crime prevention target area according to an emergency situation caused by a crime.

The control center 20 receives and stores the analysis result of the similarity between the notification signal transmitted from the emergency bell device 10 and both voice frequencies, and selectively receives between the two voice frequencies at the request of the control center 20. The analysis result for the degree of similarity may be provided to the service agency 30 as provision information.

More specifically, the control center 20 includes a control unit receiving information from the emergency bell device 10, and the control unit includes a voice processing module (M1), a keyword extraction module (M2), an emotion analysis module (M3), It may include a service agency linkage module (M4), a location information collection module (M5), and an artificial intelligence dialogue module (M6).

The voice processing module M1 may receive the voice collected from the emergency bell device 10 and convert it into text. The voice processing module (M1) receives the voice input and collected information from the voice collection module (B1), analyzes the voice information (component), and converts the voice into a sentence (text).

The voice processing module (M1) is connected by communication with an external server to improve the voice recognition rate in converting voice into text, and updates or receives information that can increase the voice recognition rate from the external server to improve accuracy in converting voice into text. can

In one embodiment, the voice processing module M1 divides the voice signal into small units because it is difficult to analyze the voice signal since the voice signal is represented as a continuous waveform along the time axis and extracts features from each unit to obtain frequency and energy. , voice gender, age, speaker characteristics, etc. can be extracted.

Accordingly, the control unit learns the extracted characteristics as a machine model, that is, learns the relationship between a voice signal and the text for the voice signal by learning through deep learning, and uses a Recurrent Neural Network (RNN), synthesis Deep learning models such as convolutional neural networks (CNNs) and transformers may be used, and since these learning models are already known technologies, a detailed description thereof will be omitted.

The voice processing module M1 may be implemented with a Hidden Markov Model, a Viterbi algorithm, a Weighted Finite State Transducer (WFST), and the like. In the voice processing module M1, a user's voice may be collected or recorded in real time. The voice processing module (M1) stores the field voice collected by the voice collection module (B1), and stores only the voice for a predetermined specific section.

On the other hand, the voice processing module (M1) may be provided in the emergency bell device 10, and at this time, the voice signal stored in consideration of the storage capacity may be transmitted to the control center 20 and then the corresponding voice signal may be deleted from the storage space. there is. Preferably, as in one embodiment of the present invention, the control center 20 is provided with a voice processing module (M1).

In one embodiment, the voice processing module (M1) may store the recorded voice from the time point (t1) when the operation signal is input to the emergency bell device (10) to the set previous time (t2). Preferably, t2 may be a time before 30 seconds from t1.

This voice processing module (M1) is implemented as an interactive human interface that wirelessly links the emergency bell device 10 and the control center 20 to be implemented so that the remote manager's voice command can be interactively performed. can

The voice processing module (M1) is an example of a configuration separated into a voice engine and a conversation engine, but is not limited thereto, and artificial intelligence (AI) that allows interactive execution commands to be executed in a conversational human interface manner. can be understood as an engine.

In addition, the dialogue engine of the voice processing module M1 may be implemented to have a function of converting a voice command into a processing command within the operating system and a function of determining whether a processing command is available and whether the processing command is authorized.

Processing command conversion may be determined by comparison and analysis between a preset processing command and a voice command, and processing command availability and authority may be determined by comparison and analysis of commands that can be processed for each authority of a preset system operator.

The keyword extraction module M2 may extract keywords from the text converted in the voice processing module M1.

For example, the keyword extraction module M2 can extract important contents by automatically summarizing long voice data input by the user. The keyword extraction module M2 recognizes the voice signal and converts it into text, and then processes the text data to determine the sentence structure and keyword context, and can accurately extract the user's requirements.

When the user's voice information is converted into text by voice recognition, the keyword extraction module (M2) can extract key keywords for analyzing user requirements through an automatic summary function. The keyword extraction module M2 can extract keywords only with words in the case of short sentences, but it is difficult to extract the main keywords from the sentences in the case of long sentences.

Therefore, since there is a possibility of errors when main keywords are extracted with one algorithm, accuracy can be improved by extracting main keywords through several algorithms.

For example, the TextRank algorithm used in the keyword extraction module (M2), LexRank, LSA (Latent Semantic Analysis), RAKE (Rapid Automatic Keyword Extraction), N-gram, Latent Dirichlet Allocation (LDA), TF-IDF (Term Frequency-Inverse Document Frequency) technique and Topic Modeling technique can be used.

Here, the TextRank algorithm determines the importance by analyzing the connectivity between words, calculates the ranking by representing each word as a node and the relationship between words as an edge, and increases the connectivity of important words with other words. Can be used with other algorithms.

In addition, the RAKE algorithm extracts key words and phrases, removes stopwords and punctuations from the sentence, calculates the frequency of words and phrases, calculates the scores of different words and phrases, and then the words with high scores. You can extract keywords by extracting phrases and phrases.

In addition, N-gram can extract important words due to words with high frequency by viewing consecutive N words as one unit and calculating the frequency of the corresponding words. In addition, TF-IDF calculates the weight of the word by multiplying the frequency (TF) in which the word appears in the document and the reciprocal of the frequency (IDF) in which the word appears in other documents, and extracts a word with a high weight based on this. can

In addition, topic modeling is a statistical modeling technique such as LDA as a method of extracting main topics or subjects from documents, and it is possible to extract main topics by identifying how topics are distributed in documents.

It is possible to improve accuracy in extracting keywords for user voice information through various algorithms as described above, and the algorithm for extracting keywords will not be limited thereto.

The keyword extraction module M2 may extract keywords from texts provided by the voice processing module M1. The keyword extraction module M2 may extract and classify keywords based on data previously stored in the control unit.

The keyword extraction module (M2) receives information on the location from the location information collection module (M5) and can further accurately define the meaning of the core keyword. Therefore, service matching accuracy can be further improved by adding the keyword extraction module M2. The keyword extraction module M2 may provide the extracted keywords to the emotion analysis module M3.

The emotion analysis module M3 may generate emotion data by receiving voice information and extracted keywords from the voice processing module M1 and the keyword extraction module M2 and analyzing the user's emotion.

The emotion analysis module M3 compares voice information obtained by extracting feature vectors for voice speed, volume pitch, and tone from the collected voice information with information stored through deep learning to generate user emotion data.

The emotion analysis module (M3) creates a model that learns and judges emotions using a machine learning algorithm, and can quickly determine and respond to user emergency situations through emotion analysis, and deep learning-based learning is further added to detect emotions. can be analyzed.

The emotion analysis module M3 performs emotion analysis with simultaneous input voices, and compares with pre-stored data to determine whether the user's state is in an emergency state. The emotion analysis module M3 can improve accuracy by analyzing emotions based on user voice information through machine learning-based or deep learning-based learning.

Support vector machine (SVM), decision tree, random forest, Gaussian kernel-based NB (Naive Bayes), KNN, Naive Bayes, etc. can be used for machine learning-based algorithms. , RNN, CNN, LSTM, GRU, Transformer, etc. can be used for deep learning-based learning.

Such a deep learning model is by way of example, and is not limited thereto, and machine learning capable of converting voice information (signal) into text is also possible by other methods.

The emotion analysis module (M3) can be used by selectively applying it by the user in the above-mentioned machine learning-based or deep learning-based learning, and any one or more of these can be applied to further improve accuracy.

According to an embodiment, when the keyword derivation and sentiment analysis for the user's requirements are completed, context awareness and service may be identified based on this.

emotion information key keyword tranquility surprised ... fear help Service A1 Service A2 ... Service An need Service B1 Service B2 ... Service Bn lack of service C1 service C2 ... Service Cn : : : : : confirmation request Service X1 Service X1 ... service Xn

As shown in Table 1, services can be searched based on a two-dimensional matrix. In Table 1, the X-axis is emotion information, and the Y-axis is the main keyword, and the service can be provided by matching with the situation corresponding to the X-axis and Y-axis.

For example, when a user shows emotions such as fear or surprise by searching for a service agency 30 based on a two-dimensional matrix and analyzing voices collected in a situational awareness process and at the same time identifying keywords that require help, an emergency Depending on the situation, the rescue request module (B2), the warning light module (B4), the abnormal voice determination module (B6), and the call connection module (B7) may be activated to perform the above-mentioned procedure.

Meanwhile, when the controller determines that the keyword extracted from the keyword extraction module M2 is not an emergency situation, the controller receives location information from the location information collection module M5 and drives natural language analysis logic.

The natural language analysis logic matches the extracted keywords so that the user's requirements can be accurately classified because the voice information or the location information of the emergency bell device 10 in the location information collection module (M5) is information about the place where the user requested the service. As a result, service matching accuracy can be improved.

At this time, the service execution result is generated in natural language, and when the meaning is unclear to the user through the speaker in the multi-function emergency bell, it can respond with preset information.

First, in the process of recognizing the situation, when the user shows emotions such as fear or surprise, and at the same time extracts a keyword that requires help, it is determined as an emergency situation, the emergency bell function is activated, and the control center 20 service matches with the police The corresponding keyword (information) is provided to the police station.

It is possible to provide information to the service agency linking module M4 so that the user's requirements are accurately determined and the service agency is matched.

The service agency linking module (M4) compares the data received from the voice processing module (M1), the keyword extraction module (M2), and the emotion analysis module (M3) with pre-stored service agency information data, and the service agency (30) can be matched.

For example, the service agency linking module M4 compares the detected emotion data with pre-stored service agency information data to match the service agency 30, and provides the matched service agency 30 with the emotion data to provide a corresponding service. can request

However, words frequently used by users vary even in one context. That is, many users express various syllables, sentences, words, etc. in the same situation. As such, there is a problem in that many users cannot predict the words they like to use. In addition, the range of word selection is wide, and there are problems such as homonyms, synonyms, synonyms, polysynonyms, dialects, pronunciation, and stress of words.

In this way, when providing public services, the user's service requirements may vary depending on the location (location) of the emergency bell device 10.

Therefore, the control center 20 includes a location information collection module (M5) that collects location information of the emergency bell device 10, so that accuracy can be improved by being predictable in interpreting the user's words.

The location information collection module M5 may collect location information according to the transmission/reception location of the emergency bell device 10 or may collect location information from a device that collects location information such as GPS in the emergency bell device 10.

When the control unit compares the extracted keyword or emotion data with the voice collection location collected by the location information collection module M5 and provides the information to the service agency linkage module M4, the service agency 30 can be linked as described above.

Since the control unit derives the literal meaning requested by the user and matches the service agency 30, there is an effect of improving accuracy.

In addition, the control unit compares the above-described data with the location information provided from the location information collection module M5 before providing the above data to the service institution 30, and provides the information so as to solve the problem corresponding to the user's request. and can improve accuracy.

According to one embodiment, when the control unit receives the voice information "The toilet paper has run out", after performing the above process, the location information (location information of the emergency bell device) provided with the voice information from the location information collection module (M5) ), the emergency bell device 10 installed in the toilet and the park is judged differently according to the corresponding sentence by interpreting the predicate, and learning can be performed accordingly.

That is, the control unit connects the corresponding service to “fall” in the case of a toilet to “no use” according to the location information provided from the location information collection module M5 or the location information where the emergency bell device 10 is installed. can make it

In addition, in the case of a park, the control unit may interpret “~ is on the floor” and link the cleaning service.

Therefore, the control unit can interpret homonyms according to the place, so that the service matching rate can be increased.

In addition, according to another embodiment, when the voice of “Find me Hong Gil-dong” input from the library is recognized, it is recognized that the author Gil-dong Hong or among the book titles Gil-dong Hong is included by the subject interpretation to guide the book or book information of Gil-dong Hong. service can be called.

In addition, the park may recognize Hong Gil-dong as a citizen and provide a service for finding Hong Gil-dong in the park.

The location information collection module M5 may collect location information according to the location information of the place where the voice is collected or the location information where the emergency bell device 10 is installed. The location information collection module (M5) is connected using GPS technology, Wi-Fi AP, Bluetooth, NFC, Bluetooth Low Energy (BLE), etc. between emergency bell devices 10 and collects voice information from the control center 20 can be specified.

Accordingly, the control unit processes the location information provided from the location information collection module (M5) in natural language, compares words that can be interpreted in two or more meanings, such as homonyms and synonyms generated from keywords extracted from the user's voice information, and information can be made more precise.

The derived linguistic meanings support services specialized for the installation location (place), and speech recognition suitable for each service can be provided through machine learning.

In addition, if the keyword value extracted from the keyword extraction module (M2) is not a keyword corresponding to the location information value collected from the location information collection module (M5), the controller outputs pre-stored additional information as a voice to the user. It can be driven by module M6.

The artificial intelligence dialogue module (M6) may attempt to communicate with the emergency bell device (10) in one selected control center (20). The artificial intelligence dialogue module (M6) is a program that answers user's questions through Natural Language Processing (NLP), and can perform dialogue management and natural language generation functions.

The artificial intelligence dialogue module M6 may be an artificial intelligence chatbot capable of analyzing sentences, grasping intentions, generating responses, and outputting answers. The artificial intelligence dialog module (M6) is a matching algorithm that calls a service (function) corresponding to the user's request identified as a result of voice recognition.

The conversation history database of the artificial intelligence dialogue module (M6) is a database that stores conversation contents between the user and the system. The user's dissatisfaction may be identified by analyzing the conversation flow in the conversation history database in the keyword extraction module (M2) and the emotion analysis module (M3).

The artificial intelligence dialogue module (M6) may attempt to communicate with the user. The artificial intelligence dialogue module (M6) stores data on the user's Q&A, builds a natural language processing AI algorithm, a chatbot for dialogue management and natural language generation, a service matching algorithm for responding to user requirements, and a database of conversation history ( learning) can.

Accordingly, the artificial intelligence dialog module (M6) can reflect the user's request by continuously identifying the function desired by the user or the part requiring improvement through Q&A to the user.

The artificial intelligence dialogue module (M6) can attempt a conversation through the microphone and speaker of the emergency bell device 10, collects voice through the microphone, and outputs voice (frequency) through the speaker to attempt a conversation with the user. can do.

On the other hand, in the case of the artificial intelligence dialogue module (M6), it is composed of a button type like the rescue request button, and a call connection can be attempted only when the user directly presses or holds the button.

In addition, the artificial intelligence dialogue module (M6) can communicate with the person in charge of the control center (20) through the modules (B1 to B7) provided in the emergency bell device (10).

At this time, in consideration of the situation in which the user cannot press the button due to the criminal's physical action, a call connection may be automatically attempted when an input signal is output according to the operation of the rescue request button.

For example, when a user makes a rescue request through the rescue request module B2, but the external environment prevents the user from making a rescue request any longer, that is, in a falsely safe situation, the service cannot be provided.

Therefore, the artificial intelligence dialogue module (M6) analyzes the user's voice in real time through the voice processing module (M1) and the emotion analysis module (M3) during the conversation with the user, and if it corresponds to the emergency situation data stored in advance, the service agency linkage module (M4 ) can transmit a rescue request signal.

Alternatively, if the user's voice is analyzed in real time during a conversation between the person in charge of the control center 20 and the user, and it corresponds to emergency data stored in advance, a notification may be provided to the person in charge of the control center 20 so that the person in charge of the control center 20 can recognize an emergency situation. .

On the other hand, if the service type is not certain, an additional query (conversation) may be conducted to the user from the emergency bell device 10. Accordingly, it is possible to improve accuracy by clarifying the service requested by the user.

Here, voice recognition can extract necessary information by pre-processing a voice signal input by a user, analyzing frequency components, and grasping voice characteristics. From the information extracted in this way, words spoken by the user are determined and converted into text.

When an emergency situation occurs when a user is located in a building, if an alarm signal is indiscriminately output, the perpetrator may be stimulated and put into a dangerous situation. The control unit determines whether or not to move indoors by using whether a GPS signal is received, and controls the speaker not to output an alarm sound when a user's scream is recognized after moving indoors.

In addition, according to another embodiment of the present invention, the alarm sound is not output immediately after scream recognition, but when the control center 20 or the guardian hears situation information or sound before and after scream recognition and determines that it is an emergency situation, When the center 20 or the guardian's terminal transmits an alarm output control signal to a portable scream recognizer, it can be output.

After scream recognition, the controller controls the speaker to output an alarm sound when the GPS signal is not received and then received again. The controller may control the speaker to output an alarm sound when moving from inside the building to the outside.

According to the present embodiment, in order to prevent indiscriminate transmission of situation information, the control unit controls the situation when a scream is recognized a plurality of times within a preset time instead of when a scream is recognized once in the voice collection module (B1). Information can be controlled to be transmitted to the control center (20).

However, since only one scream can occur in an emergency situation, if it is recognized only once, it is possible to track the noise (voice) generated thereafter for a predetermined time and deliver the emergency situation to the control center 20. Of course.

The person in charge of the service institution 30 can listen to the recorded voice before the operation of the emergency bell device 10 to understand the situation from the time of occurrence to the previous situation.

By collecting this information, it is possible to help determine whether the emergency bell device 10 is malfunctioning or falsely reported, as well as determining whether management personnel are dispatched to the site.

The service institution 30 may call when a request from an external institution is required among user requests.

According to an embodiment, when the user provides voice information that there is no toilet paper in the bathroom, the control center 20 outputs the toilet paper and toilet paper from the voice processing module M1, the keyword extraction module M2, and the emotion analysis module M3. derives insufficient information and transmits the user's request to the service institution 30 to the manager of the corresponding toilet. After that, the service institution 30 may solve the problem in response to solve the corresponding problem.

The service agency 30 may provide various services according to user settings. For example, the service agency 30 may be associated with a manager monitoring the toilet. At this time, the service organization 30 provides the information received from the control center 20 to the manager as a result value. Thereafter, the administrator may solve the problem requested by the user by providing a service corresponding to the provided result value.

The deep learning distributed processing device continuously learns the characteristics of the object to be monitored that appear in the environmental characteristics (environmental conditions) of the target area to be monitored through deep learning, thereby increasing the accuracy of identification of the object to be monitored.

Including driving control of local video devices, video monitoring monitoring control, and video control system control management through interactive operation method through voice commands instead of video control system operation and execution commands by manual mouse or keyboard operation Everything you do can be controlled remotely.

In particular, the artificial intelligence dialogue module M6 can remotely receive a manager's voice command through a speaker, microphone, or the like, or provide pre-stored question data to provide Q&A with the user. Accordingly, since accurate requests can be resolved through the voice collection module (B1), voice processing module (M1), keyword extraction module (M2), emotion analysis module (M3), etc., accuracy can be improved. It may be such a interactive human interface method.

The voice processing module (M1) and the artificial intelligence dialog module (M6) instruct and process remote commands as if they are talking to each other without separate work, further improving the convenience and efficiency of management. As a countermeasure to this, information is provided to the service institution 30, and thus, it is possible to remotely operate the system quickly, conveniently, and accurately without going through a complicated process.

In addition, in general, when several people talk in a space within a building or children talk loudly, it may be misrecognized as a scream.

Considering this situation, situation information may be transmitted to the control center 20 only when a scream is recognized a plurality of times within a preset time. Preferably, situation information may be transmitted to the control center 20 when it is recognized a plurality of times within one or more preset times stored in the trigger list rather than the volume of the user's voice.

According to one embodiment of the present invention, the transmission/reception module (B5) of the emergency bell device 10 may transmit situation information to the control center 20, and may broadcast an emergency situation message to other peripheral devices.

Transmission of context information and broadcasting of emergency messages may be implemented as separate chips. Broadcasting of the emergency message is transmitted to other emergency bell devices 10 located in a short distance to output an alarm sound to inform the surroundings of an emergency situation, so that other users can receive help.

Emergency bell device 10 may be used through a short-range network communication such as Bluetooth through the transmission and reception module (B5). The emergency bell device 10 may be paired with a mobile communication terminal possessed by a user through a local area network. The emergency bell device 10 periodically transmits situation information according to time to the control center 20 .

The emergency bell device 10 and the control center 20 may include pairing information, and the control center 20 may determine whether or not it is an emergency situation through whether the above-described pairing state is released after voice recognition, and emergency It can be determined that the bell device 10 has a problem.

The control center 20 receives the situation information and determines whether the user is in an emergency situation. For example, after an emergency occurs, if the user's location over time remains unchanged for a long time or moves at high speed, it can be determined that the perpetrator is confined or moved to another place through the perpetrator's vehicle. .

In consideration of this point, the control center 20 determines that an emergency situation occurs when the location information is not updated for more than a preset first threshold time from the scream recognition time or the location information is changed at a preset speed or more.

In addition, the control center 20 may determine whether or not an emergency situation exists by using whether or not a scream is additionally recognized in the ambient sound after the scream recognition.

According to an embodiment of the present invention, ambient sound information is transmitted to an AI server built with a separate artificial neural network model, rather than the control center 20, and the AI server analyzes the ambient sound information to determine whether it is an emergency situation. can

For example, when a sound of a predetermined size or more is recognized and the emergency bell device 10 transmits situation information, and one or more triggers are included in the ambient sound, the control center 20 may determine that an emergency situation has occurred. there is.

The service agency 30 can be connected to various services according to the operator's choice. For example, the toilet management room, the local government police station, the local fire station, the civil complaint situation room of the local government, the police unit in charge of the crime prevention area, autonomous crime prevention unit, management office, etc. , It is possible to provide a service to the user by matching any one or more than one of them.

Referring to FIGS. 2 and 4, an embodiment of a control method of a smart artificial intelligence multifunctional emergency bell system according to the present invention is described. When a user provides voice information, The voice information is collected in the voice collection module (B1) of the emergency bell device (10) (S10).

The collected voice information is transmitted to the voice processing module M1 of the control center 20, and the voice processing module M1 may convert the voice provided from the voice collection module B1 into text (S20). Keywords can be extracted from the textualized sentence in the keyword extraction module (M2) (S30).

At this time, the controller may analyze the user's emotion based on the collected voice information and generate emotion data (S15).

The control unit classifies the user's requested keyword by comparing the textualized keyword and the analyzed emotion data with previously stored emotion data (S40).

The extracted keyword is compared with previously stored service agency information data, and the extracted keyword is provided to the service agency 30 corresponding to the detected keyword to request a service (S56).

Accordingly, the user can receive a service for problem solving from the service organization 30 .

On the other hand, in step S15, the emotion data is to analyze the user's emotion based on the data obtained by comparing the collected voice information with information stored through deep learning and extracting feature vectors for voice speed, volume pitch, and tone. can

In addition, after step S40, the classified keywords are compared with keywords corresponding to pre-stored emergency situations (S50).

Here, the control unit provides an emergency service corresponding to the emergency keyword when the classified keyword is a keyword corresponding to an emergency situation (S60), and the rescue request module (B2) of the emergency bell device (10), the speaker module (B3), The warning light module (B4) can be operated.

At this time, in step S50, the control unit collects (S55) information about the location where voice information is collected if it does not correspond to an emergency situation. Based on the collected location information and the extracted keyword, the service may be requested by providing the extracted keyword to the service agency corresponding to the voice information (S56).

Thereafter, the result of requesting the service may be output to the user as a voice to notify the result (S70). Whether or not to output the alarm signal may be differently controlled according to the current user's location even if it is determined to be an emergency situation.

Referring to FIGS. 2 and 5, another embodiment of the control method of the smart artificial intelligence multifunctional emergency bell system according to the present invention is described. When the user provides voice information, The voice information is collected in the voice collection module (B1) of the emergency bell device (10) (S10).

The collected voice information is transmitted to the voice processing module M1 of the control center 20, and the voice processing module M1 may convert the voice provided from the voice collection module B1 into text (S20). Keywords can be extracted from the textualized sentence in the keyword extraction module (M2) (S30).

At this time, the controller may analyze the user's emotion based on the collected voice information and generate emotion data (S15).

The control unit extracts and classifies the user's requested keyword by comparing the textualized keyword and the analyzed emotion data with previously stored emotion data (S40).

The extracted keyword is compared with previously stored service agency information data, and the extracted keyword is provided to the service agency 30 corresponding to the detected keyword to request a service (S56).

Accordingly, the user can receive a service for problem solving from the service organization 30 .

On the other hand, in step S15, the emotion data is to analyze the user's emotion based on the data obtained by comparing the collected voice information with information stored through deep learning and extracting feature vectors for voice speed, volume pitch, and tone. can

In addition, after step S40, the classified keywords are compared with keywords corresponding to pre-stored emergency situations (S50).

Here, it may be determined whether the classified keyword is an emergency situation (S50). If the classified keyword corresponds to an emergency situation, an emergency service corresponding to the emergency keyword may be provided (S60).

At this time, in step S50, the control unit collects (S55) information about the location where voice information is collected if it does not correspond to an emergency situation. Based on the collected location information and the extracted keyword, the service may be requested by providing the extracted keyword to the service agency corresponding to the voice information (S56).

Meanwhile, in the above step S30, the user may try to have a conversation by outputting the additional information in which the extracted keyword value is pre-stored as a voice to the user (S31). That is, if it fails to extract the main keyword, it may try to have a conversation with the user.

In addition, in the step S31, conversation list data is generated by storing the text converted from the collected voice information in chronological order, keywords are extracted from the information on the conversation list data, and emotions are analyzed and located in the collected voice information. Each information is received, keyword result data for a result corresponding to a conversation flow is generated, and the keyword result data can be provided by calling a service institution matching pre-stored service institution information data according to the keyword result data. .

Here, the conversation attempt may ask the user various questions, and these questions may be useful questions for extracting keywords. In addition, examples of questions may be set by the setter in the control unit in advance. Various questions can be asked by combining questions in response to answers, and stored data learned through machine learning can be loaded and stored to ask questions to the user.

According to the smart artificial intelligence multifunctional emergency bell system and control method of the present invention, it can be built for the purpose of protecting human life and property from disasters, disasters, incidents, accidents, crime prevention, and crime prevention, and can be used in narrow places It can also be operated, so it can request help in an emergency situation and spread the emergency situation, and it has the effect of being used in poor indoor and outdoor environments.

In addition, the smart artificial intelligence emergency bell device according to the present invention can respond to the user's request by transmitting the user's request to the control center in real time, and can maximize efficiency while minimizing the field manpower for maintenance. In addition, continuous maintenance of the emergency bell device, as well as rapid action for malfunction and damage, etc. can be performed, thereby minimizing operational blank time.

In addition, according to the present invention, since it can be used only by voice without a small LCD panel or LCD panel, it can be miniaturized, and it can be used without restrictions in the installation place because it secures visibility and has little effect even in harsh environments, and it can be used without restrictions on digital use. It has the effect of resolving the privacy invasion problem while providing convenience to users (foreigners, etc.).

Of course, the present invention is not limited to the above-described embodiments, and may include a combination of two or more embodiments or a combination of at least one embodiment and known technology as a new embodiment.

Although the preferred embodiments of the present invention have been described in detail above, the present invention is not limited thereto, and those skilled in the art using the basic concept (technical idea) of the present invention defined in the following claims Various modifications and improvements are possible and will be said to belong to the scope of the present invention, and the specific protection scope of the present invention will be clarified by the appended claims.

10: emergency bell device
20: control center
30: service center

Claims (11)

An emergency bell device having a transmission/reception module for transmitting and receiving the voice collected by the voice collection module for collecting voice through a microphone for receiving the user's voice to an external server; and
A voice processing module that receives and converts voices collected from the emergency bell device into text, an emotion analysis module that analyzes the user's emotions from the extracted keywords and text, and the emotion data detected by the emotion analysis module and pre-stored service institutions A control center including a control center for managing a server; including a control unit including a service agency linking module requesting a service by comparing information data and providing the emotion data to a matched service agency;
The control unit,
a keyword extraction module extracting keywords from the text converted by the voice processing module and providing the extracted keywords to the emotion analysis module; And a location information collection module for collecting information on the location of the voice information collected by the voice collection module;
Comparing the keyword extracted from the keyword extraction module with the location information value collected from the location information collection module, and providing matching data to the service institution linking module;
If the keyword value extracted by the keyword extraction module is not a keyword corresponding to the location information value collected by the location information collection module, an artificial intelligence dialog module outputting pre-stored additional information as a voice to the user; smart artificial intelligence further comprising: Intelligent multifunctional emergency bell system.
delete delete delete According to claim 1,
The artificial intelligence dialogue module,
The text converted by the voice processing module is stored in chronological order to generate conversation list data, and information on the conversation list data is provided to the keyword extraction module, the emotion analysis module, and the location information collection module, respectively, so that the conversation flows. Receiving the corresponding result, generating keyword result data, receiving the keyword result data from the service institution linking module and providing the keyword result data to a service institution that matches the previously stored service institution information data to request a service Smart artificial intelligence multifunctional emergency bell system.
According to claim 1,
The artificial intelligence dialogue module,
Smart artificial intelligence multifunctional emergency bell system that analyzes the user's voice in real time through the voice processing module and emotion analysis module during conversation with the user and transmits a rescue request signal to the service agency linkage module if it corresponds to the pre-stored emergency data.
delete delete delete delete delete

Images