Human-computer interaction method for distribution network dispatching based on artificial intelligence speech recognition

Abstract

The present invention relates to a human-computer interaction method for distribution network dispatching based on artificial intelligence speech recognition, comprising the following steps: user-end equipment collects user voice requests, and performs parsing and processing through a parsing and processing module; the parsed voice information is assembled by protocol, and output to an intelligent speech engine through an access layer; the intelligent speech engine normalizes and differentiates the user's request, parses the true intention of the user's request, and transmits key information to a back-end service module that processes the user's intention; the back-end service module processes the user's intention and converts it into a corresponding power grid dispatching operation. This method is conducive to improving the efficiency of distribution network dispatching and the intelligence level of the distribution network dispatching system.

Description

Technical Field

[0001] The present invention belongs to the field of power system distribution automation, and in particular relates to a distribution network dispatching human-computer interaction method based on artificial intelligence speech recognition. Background Art

[0002] At present, the control of distribution network is mainly realized by manual touch interaction. Promoting the application of intelligent voice technology in the daily business of distribution network control and realizing functions such as intelligent voice interaction, voice-to-text conversion, voice semantic analysis, and voice operation control can improve the intelligence level of the intelligent distribution network dispatching system, improve the efficiency of human-computer interaction, and enhance the real-time control capability and daily business handling efficiency of the regional power grid. Summary of the invention

[0003] The purpose of the present invention is to provide a distribution network dispatching human-computer interaction method based on artificial intelligence speech recognition, which is conducive to improving the distribution network dispatching efficiency.

[0004] To achieve the above object, the technical solution adopted by the present invention is: a human-computer interaction method for distribution network dispatching based on artificial intelligence speech recognition, comprising the following steps:

[0005] The user terminal device collects the user's voice request and parses and processes it through the parsing and processing module;

[0006] Perform protocol assembly on the parsed voice information and output it to the intelligent voice engine through the access layer;

[0007] The intelligent voice engine normalizes and differentiates user requests, analyzes the true intent of user requests, and transmits key information to the backend service module that processes user intent;

[0008] The back-end service module processes the user's intentions and converts them into corresponding power grid dispatching operations.

[0009] Furthermore, the analysis and processing link is processed by two modules: a sound signal conversion module and a noise processing module. The sound signal conversion module is used to convert the voice into a format file that can be recognized by a computer, and the noise processing module is used to improve the accuracy of voice collection, thereby extracting the characteristic parameters contained in the user's voice.

[0010] Furthermore, the steps of extracting the user's voice by the sound signal conversion module are as follows:

[0011] First, preprocessing is performed, including sampling and automatic gain control of the input speech signal;

[0012] Secondly, the characteristic parameters of the user's voice are extracted, including the resonance peak of the sound wave and the pitch of the voice;

[0013] Furthermore, algorithm training and pattern library training are used to classify the feature parameters of the extracted speech to complete the semantic output of the speech; through training, knowledge including grid inventory equipment information, grid safety regulations, dispatcher terminology library, and distribution network dispatching and disposal regulations are learned;

[0014] Finally, pattern matching is performed, which mainly involves testing the matching degree between the extracted speech to be tested and the speech in the pattern library, and finally completing the semantic output in the speech.

[0015] Furthermore, microphone array technology is used to acquire the user's voice; the microphone array is a recording system that uses two or more microphones arranged and combined according to certain rules to acquire and process sound in a set space; the microphone array effectively picks up sound through sound source localization, beamforming, noise suppression, anti-reverberation, echo cancellation, and voice enhancement to improve the artificial intelligence voice recognition effect and hear the user's voice clearly.

[0016] Furthermore, the protocol assembly of the parsed voice information is completed through a connection layer application. The connection layer application is mainly responsible for the communication between the device end and the cloud service, and defines the communication protocol between the two. The communication protocol mainly consists of three parts: instructions, events, and end status.

[0017] Furthermore, the intelligent voice engine is used to normalize and differentiate the input of multiple voice sources, accurately analyze the intention of the user's voice, and transmit the key information of the user's voice request to the back-end service module that specifically processes the user's intention.

[0018] Furthermore, the access layer is output to the intelligent voice engine, wherein the access layer is mainly responsible for accessing and forwarding user voice requests, and performing global traffic scheduling and data analysis, and also has security defense functions; when a large number of user voice requests occur, in order to share the pressure of the cloud server, it is responsible for load distribution of user voice requests and sharing the pressure of the server to ensure that the server will not cause user request loss or untimely response under high access conditions.

[0019] Furthermore, the back-end service module is mainly a scheduling module for power grid system functions, which is used to make intelligent decisions and call related distribution network systems to complete response operations based on the semantic requirements in the user's voice, combined with the perception of power grid equipment and status, user habits and interaction context.

[0020] Compared with the prior art, the present invention has the following beneficial effects: Introducing artificial intelligence voice technology into the daily human-computer interaction scenarios of distribution network dispatching is conducive to improving the intelligence level of the intelligent distribution network dispatching system and improving the efficiency of daily business handling. Through machine learning and intelligent interaction, the full-scenario coordination of power grid dispatching is realized, the power grid equipment is simplified, the power grid operation status is learned through the massive collection data of the distribution master station, and the learning and memory of the user's power grid dispatching is added. According to the dispatching user's voice request, the user's potential equipment control needs are predicted, and prompts and suggestions are made to the user in a timely and proactive manner. One or more optimal instructions are selected to respond to better meet the power grid dispatching needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a flow chart of the method implementation of the embodiment of the present invention.

[0022] Figure 2 It is a flow chart for realizing voice opening of a line diagram in an embodiment of the present invention.

[0023] Figure 3 It is a flow chart for implementing the voice-off line diagram in an embodiment of the present invention.

[0024] Figure 4 This is a flowchart of the implementation of voice device positioning in an embodiment of the present invention.

[0025] Figure 5 It is a flow chart for implementing the voice equipment listing in an embodiment of the present invention.

[0026] Figure 6 It is a flow chart for implementing the voice insertion service in an embodiment of the present invention.

[0027] Figure 7 It is a flow chart for realizing voice switch remote control in an embodiment of the present invention.

[0028] Figure 8 It is a flow chart of implementing the voice device call test in an embodiment of the present invention.

[0029] Fig. 9 It is a flow chart for implementing speech statistics query in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0031] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0032] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0033] like Figure 1 As shown, this embodiment provides a distribution network dispatching human-computer interaction method based on artificial intelligence speech recognition, comprising the following steps:

[0034] 1. The user terminal device collects the user's voice request and parses it through the parsing module.

[0035] The analysis and processing link is processed by two modules: a sound signal conversion module and a noise processing module. The sound signal conversion module is used to convert the voice into a format file that can be recognized by a computer, and the noise processing module is used to improve the accuracy of voice collection and then extract the characteristic parameters contained in the user's voice.

[0036] The steps of extracting user voice by the sound signal conversion module are as follows:

[0037] First, preprocessing is performed, including sampling and automatic gain control of the input speech signal;

[0038] Secondly, the characteristic parameters of the user's voice are extracted, including the resonance peak of the sound wave and the pitch of the voice;

[0039] Furthermore, algorithm training and pattern library training are used to classify and process the feature parameters of the extracted speech to complete the semantic output in the speech; according to the feature data in the speech information: object, operation description, different objects are combined with different operation descriptions, and classified processing is performed to achieve different operation purposes, and finally converted into computer language; through training, knowledge including grid inventory equipment information, grid safety regulations, dispatcher terminology library, and distribution network dispatching and disposal regulations are learned;

[0040] Finally, pattern matching is performed, which mainly involves testing the matching degree between the extracted speech to be tested and the speech in the pattern library, and finally completing the semantic output in the speech.

[0041] The present invention adopts microphone array technology to obtain the user's voice; the microphone array is a recording system that uses two or more microphones arranged and combined according to certain rules to obtain and process sound in a set space; the microphone array effectively picks up sound through algorithms such as sound source localization, beamforming, noise suppression, anti-reverberation, echo cancellation, and speech enhancement to improve the artificial intelligence speech recognition effect and hear the user's voice clearly.

[0042] 2. Perform protocol assembly on the parsed voice information and output it to the intelligent voice engine through the access layer.

[0043] The protocol assembly of the parsed voice information is completed through the connection layer application, which is mainly responsible for the communication between the device end and the cloud service, and defines the communication protocol between the two. The communication protocol mainly consists of three parts: instructions, events, and end status.

[0044] 3. The intelligent voice engine normalizes and differentiates the user's requests, analyzes the true intention of the user's request, and transmits key information to the back-end service module that processes the user's intention.

[0045] The intelligent voice engine is used to normalize and differentiate the input of multiple voice sources, accurately analyze the intention of the user's voice, and transmit the key information of the user's voice request to the back-end service module that specifically processes the user's intention.

[0046] The data is output to the intelligent voice engine through the access layer, where the access layer is mainly responsible for accessing and forwarding the user's voice requests, and performing global traffic scheduling and data analysis, while also having security defense functions. When a large number of user voice requests occur, in order to share the pressure on the cloud server, it is responsible for load distribution of the user's voice requests and sharing the pressure on the server to ensure that the server will not cause user request loss, untimely response and other problems under high access conditions.

[0047] 4. The back-end service module processes the user's intentions and converts them into corresponding power grid dispatching operations.

[0048] The back-end service module is mainly a scheduling module for power grid system functions, which is used to make intelligent decisions and call related distribution network systems to complete response operations based on the semantic requirements in the user's voice, combined with the perception of power grid equipment and status, user habits and interaction context.

[0049] The human-computer interaction method for distribution network dispatching based on artificial intelligence speech recognition of the present invention is specifically implemented as follows:

[0050] 1. Artificial intelligence speech recognition and speech synthesis

[0051] It supports voice-to-text conversion, adopts machine learning technology in the field of artificial intelligence, learns the user's spoken pronunciation, grasps the user's pronunciation habits, dialect characteristics, professional vocabulary and other information, and converts the user's speech (audio) into a text.

[0052] (1) Basic voice functions, artificial intelligence voice recognition

[0053] Supports AI speech recognition and speech synthesis of common language and professional terms in cities and prefectures. Supports intelligent noise reduction. Users input through the microphone, filter out noisy noise, and detect the sound collection distance. Supports one-time wake-up and continuous recognition. Supports setting the wake-up word to trigger the start of speech recognition. After wake-up, you can speak multiple times continuously, and the voice will automatically sleep after a period of inactivity.

[0054] (2) Voice Input

[0055] Supports voice instead of keyboard typing, supports instant voice transcription, and supports voice document writing.

[0056] (3) Speech synthesis, text-to-speech

[0057] It supports playing a text in the form of audio according to local language habits and in the way the user is accustomed to.

[0058] 2. Speech model training

[0059] By adopting machine learning technology in the field of artificial intelligence, we master the user's pronunciation habits and dialect characteristics by learning the user's spoken pronunciation, and build a professional model + general voice model recognition library for nine cities to improve the recognition accuracy of commonly used commands and the recognition accuracy of equipment in various cities.

[0060] The speech model training content mainly includes:

[0061] 1) Substation name training

[0062] 2) Route name training

[0063] 3) Equipment name training

[0064] 4) Equipment number name training

[0065] 5) Operation card name training

[0066] 6) Operation intention training, including opening (viewing, browsing, etc.) graphics, closing graphics, equipment positioning, listing, delisting, setting, releasing, remote control, control separation, control combination, switching from operation to cold standby, and other operation ticket instructions, continuing operation, canceling operation, ending operation, calling for test, query, statistics, etc. Operation intention training aims to identify the key fields of each operation intention. If the intention recognition effect is not good during the actual interaction process, the corpus can be manually added to update the model. The richer the corpus, the more the intention can adapt to a more variable language environment, and the higher the recognition accuracy.

[0067] 3. Semantic analysis

[0068] (1) Supports the construction of a voice command structured model based on semantic analysis technology. Through word segmentation, keyword retrieval and other processing, a user's speech can be decomposed into speech that can be understood by the computer.

[0069] (2) Multi-round dialogue: The process of voice interaction involves multiple rounds of interaction with the user, building a conversation scenario, obtaining user input information step by step, and determining the user's operation intention.

[0070] 4. Voice-to-operation function

[0071] (1) Voice modulation

[0072] Supports actively parsing the screen review voice input from the audio device into voice commands containing action commands and operation content, and converting them into screen review commands to achieve voice interaction of screen review services, including normal graphics review and graphics closing.

[0073] The implementation process of voice opening line diagram in this embodiment is as follows Figure 2 shown.

[0074] The implementation process of the voice closing line diagram in this embodiment is as follows Figure 3 shown.

[0075] (2) Voice device positioning

[0076] It supports actively parsing the device positioning commands input by the audio device into voice commands containing action instructions and operation content, and should support converting them into device positioning and graphics opening instructions by matching the device type and device name, to achieve voice interaction of device positioning services. The positioning device types should include station houses, busbars, switches, knife switches, distribution transformers, meter boxes, etc. When there are multiple matching results, user selection should be supported.

[0077] The implementation process of voice device positioning in this embodiment is as follows: Figure 4 shown.

[0078] (3) Voice Sign

[0079] Supports actively parsing the equipment listing voice input by the audio device into voice instructions containing action instructions and operation content, and converting them into listing instructions to realize voice interaction of listing business, including listing and delisting of equipment.

[0080] The implementation process of voice equipment listing in this embodiment is as follows Figure 5 shown.

[0081] (4) Voice positioning

[0082] Supports actively parsing the embedded business voice input from the audio device into voice commands containing action commands and operation content, and converting them into embedded commands to realize voice interaction of embedded business, including remote signal division, remote signal combination, and telemetry setting.

[0083] The implementation process of the voice insertion service in this embodiment is as follows: Figure 6 shown.

[0084] (5) Voice remote control

[0085] It supports actively parsing the switch remote control voice input from the audio device into voice commands containing action commands and operation content, and converting them into switch remote control commands to realize voice interaction of switch remote control services, including remote control verification, remote control execution and other operations of designated switches.

[0086] The implementation process of voice switch remote control in this embodiment is as follows Figure 7 shown.

[0087] (6) Voice low voltage equipment call test

[0088] It supports actively parsing the statistical query voice input from the audio device into voice commands containing action commands and operation content, and converting them into low-voltage equipment test call commands to realize voice interaction of equipment operation data test call services. The test call commands include low-voltage switch status, current voltage of the meter, etc.

[0089] The implementation process of the voice device call test in this embodiment is as follows: Figure 8 shown.

[0090] (7) Voice statistics query

[0091] Supports automatically parsing the statistical query voice input from the audio device into packets

[0092] The voice instructions containing action instructions and operation content are converted into statistical query instructions to realize voice interaction of statistical query business, including opening the statistical query general interface, opening the statistical query formulation interface and other operations.

[0093] The implementation process of voice statistics query in this embodiment is as follows: Fig. 9 shown.

[0094] 5. Voice broadcast reminder

[0095] Supports converting freely organized text content into audio files for playback through the voice platform.

[0096] (1) Smart alarm clock, timed voice reminders, active broadcast of key information, such as reminders for the "six time nodes" for planning and application management, voice broadcast guidance of operation steps, document reading, etc.

[0097] (2) Voice telephone commands and ring closing and opening operation reminders should have knowledge learning functions, build a real-time power grid database, import procedures and typical operation tickets into the knowledge base, and have a reminder function in case of erroneous operation in the instruction ticket.

[0098] (3) Fault abnormal signal reminder function, which provides reminders during fault notification, abnormal judgment, inspection notification, abnormal handling, and so on.

[0099] 6. Voice drafting of operation tickets

[0100] It supports actively parsing the proposed operation ticket commands input by the audio device into operation texts containing operation steps, operation equipment and operation content, and should support sending the operation texts online to the intelligent instruction ticket issuing system.

[0101] Support active recognition of operating equipment based on voice, multiple recognition results of positioning equipment ledger information should support manual selection, and the final text of the operation ticket should support manual editing.

[0102] Without affecting the current business logic of the intelligent ticketing system and the mature networked ordering process, the function of <voice drafting instruction items> is implemented. Its purpose is to reduce the complexity of manually filling in instruction items in the intelligent ticketing system and improve the accuracy of operating equipment and operating steps.

[0103] 7. Voice intelligent shift handover

[0104] It supports recording and storing shift handover recordings through voice, identifying the information of shift handover personnel through fingerprints and recording it in the system, and automatically generating shift handover records.

[0105] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A human-computer interaction method for distribution network dispatching based on artificial intelligence speech recognition, characterized in that: The steps include: The user terminal device collects the user's voice request and converts the information through the analysis and processing link; Perform protocol assembly on the parsed voice information and output it to the intelligent voice engine through the access layer; The intelligent voice engine normalizes and differentiates user requests, analyzes the true intent of user requests, and transmits key information to the backend service module that processes user intent; The backend service module processes the user's intention and converts it into corresponding power grid dispatching operations; The analysis and processing link is processed by two modules: a sound signal conversion module and a noise processing module. The sound signal conversion module is used to convert the voice into a format file that can be recognized by a computer, and the noise processing module is used to improve the accuracy of voice collection, thereby extracting the characteristic parameters contained in the user's voice; the steps of extracting the user's voice by the sound signal conversion module are as follows: First, preprocessing is performed, including sampling and automatic gain control of the input speech signal; Secondly, the characteristic parameters of the user's voice are extracted, including the formant of the sound wave and the pitch of the voice; Furthermore, algorithm training and pattern library training are used to classify the feature parameters of the extracted speech to complete the semantic output of the speech; through training, knowledge including grid inventory equipment information, grid safety regulations, dispatcher terminology library, and distribution network dispatching and disposal regulations are learned; According to the characteristic data in the voice information: objects, operation descriptions, different objects are combined with different operation descriptions, classified and processed to achieve different operation purposes, and finally converted into computer language; The microphone array technology is used to obtain the user's voice; the microphone array is a recording system that uses two or more microphones arranged and combined according to certain rules to obtain and process the sound in a set space; the microphone array effectively picks up the sound through sound source localization, beam forming, noise suppression, anti-reverberation, echo cancellation, and voice enhancement to improve the artificial intelligence voice recognition effect and hear the user's voice clearly; The protocol assembly of the parsed voice information is completed by the connection layer application. The connection layer application is mainly responsible for the communication between the device end and the cloud service, and defines the communication protocol between the two. The communication protocol mainly consists of three parts: instructions, events, and end status; The intelligent voice engine is used to normalize the input of multiple voice sources, output differentiated information, accurately analyze the intention of the user's voice, and transmit the key information of the user's voice request to the back-end service module that specifically processes the user's intention; The access layer is used to output the data to the intelligent voice engine, where the access layer is mainly responsible for accessing and forwarding user voice requests, and performing global traffic scheduling and data analysis, while also having security defense functions. When a large number of user voice requests occur, in order to share the pressure on the cloud server, it is responsible for load distribution of user voice requests and sharing the pressure on the server to ensure that the server does not cause user request loss or untimely response under high access conditions.

2. The human-computer interaction method for distribution network dispatching based on artificial intelligence speech recognition according to claim 1 is characterized in that: The back-end service module is mainly a scheduling module for power grid system functions, which is used to make intelligent decisions and call related distribution network systems to complete response operations based on the semantic requirements in the user's voice, combined with the perception of power grid equipment and status, user habits and interaction context.

Images