Charging pile service interaction system based on AI technology

By integrating AI technology on charging piles, real-time monitoring and analysis of user operations, the problem of poor stability management capabilities of traditional charging pile systems is solved, the user's first start-up rate and charging experience are improved, and intelligent interactive services are realized.

CN119058461APending Publication Date: 2024-12-03NINGBO JUNSHENG QUNYING INTELLIGENT TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411179232.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

There are shortcomings in stability management of traditional charging pile service interaction systems, resulting in low first-time charging rate and poor charging experience, affecting the confidence of new energy vehicles and the reputation of the industry.

Method used

Adopting AI-based charging pile service interaction system, including AI charging pile, data acquisition module and interactive service module, through image interactive cameras, voice interactive MICs, AI digital human SPI display screen, interactive monitoring and charging controller, 4G module, The main loop output system and gun line tip operation components monitor user operations in real time and analyze faults through deep learning, providing intelligent interactive services.

Benefits of technology

It improves the user's first charging rate and stability, provides intelligent interactive service capabilities, can promptly correct user error operations and accurately locate the cause of failure, and enhances the charging experience and confidence in new energy vehicles.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119058461A_ABST
    Figure CN119058461A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of charging pile service interaction, and discloses an AI technology-based charging pile service interaction system, which comprises an AI charging pile, a data acquisition module and an interaction service module. The AI charging pile realizes real-time monitoring of operation stages and real demands of a user through the AI charging pile, the data acquisition module classifies the AI charging pile into an image data set, a voice data set and a signal data set, the interaction service module combines an AI charging pile operation instruction with an actual operation image of the user, a recognition model is obtained through training, and a guidance data set is generated through analysis. The method has the advantages that faults are analyzed, fault causes are accurately positioned, manual judgment and intervention are not needed, the intelligent interaction service capability is high, the AI digital human SPI display screen is controlled by the central processing unit MCU to explain solution data in a guidance data set to a user in an AI digital human sound picture form, wrong operation of the user is corrected in time, faults and charging guidance solutions are displayed in time, and the user experience is improved. And the first-time charging rate of the user is high and the stability is high.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of charging pile service interaction technology, and specifically to a charging pile service interaction system based on AI technology. Background Art

[0002] As the supporting infrastructure of electric vehicles, charging piles provide necessary charging services for electric vehicles and are an important support for promoting the development of the new energy vehicle industry. Charging piles can be divided into public charging piles, dedicated charging piles and private charging piles. According to the type of current, charging piles are also divided into AC charging piles and DC charging piles, the former is suitable for slow charging, and the latter is suitable for fast charging. For those who are considering buying or using electric vehicles, it is very important to understand the distribution and service quality of nearby charging infrastructure. Choosing a suitable charging pile should not only consider charging speed, but also factors such as charging cost, convenience and compatibility. With the rapid expansion of the new energy vehicle market, the number and quality of charging piles have become an important indicator to measure the level of new energy transportation development in a country or region.

[0003] At present, when the traditional charging pile service interaction system is applied to externally operated charging piles, there are certain deficiencies and untimely stability management. Faced with potential occasional failures or occasional failures of the new energy vehicle battery management system BMS, it is easy to fail to start charging or charge unstably. In addition, in actual use, the installation environment of the charging pile is complex and changeable. When some car owners use the charging pile, they are prone to misoperation due to unfamiliarity with the equipment. There are hundreds of error conditions for the charging pile equipment. On-site managers and remote customer service cannot fully, timely and accurately analyze the source of car owners' complaints, and the equipment fault handling ability is poor. These problems jointly lead to a low first-time charging rate for car owners and a poor charging experience, which in turn reduces their confidence and satisfaction with new energy vehicles and has a negative impact on the overall reputation of the new energy industry. Summary of the invention

[0004] Technical issues solved

[0005] In view of the shortcomings of the prior art, the present invention provides a charging pile service interaction system based on AI technology, which has the advantages of strong intelligent interactive service capabilities, high user first charging rate and strong stability. It solves the problems of poor stability management of traditional charging pile service interaction systems and low user first charging rate and poor experience.

[0006] Technical Solution

[0007] To achieve the above-mentioned object, the present invention provides the following technical solutions: a charging pile service interaction system based on AI technology, comprising an AI charging pile, a data acquisition module and an interactive service module;

[0008] The AI ​​charging pile includes an image interaction camera, a voice interaction MIC, an AI digital human SPI display screen, an interactive monitoring and charging controller, a 4G module, a main circuit output system, and a gun line and gun head operation component;

[0009] The data acquisition module includes a camera image unit, a sound sensor unit and a vehicle-pile communication unit. The camera image unit is connected to the image interaction camera through a network, and collects image data sets from early to late in chronological order. The image data sets include user operation images at all time points. The sound sensor unit is connected to the voice interaction MIC through a network, and collects voice data sets from early to late in chronological order. The voice data sets include user voice data at all time points.

[0010] The interactive service module includes a deep learning unit, a real-time analysis unit and an AI service unit. The deep learning unit is connected to an image processor GPU and a cloud AI deployment via a network. The cloud AI deployment is connected to a database via a network. The database stores and backs up the AI ​​charging pile operation manual ZY. The cloud AI deployment is provided with a CNN algorithm model, and performs model training in combination with the AI ​​charging pile operation manual ZY and an image data set. The recognition model Shmx is analyzed and generated, and is transmitted to the real-time analysis unit via the network. The real-time analysis unit substitutes the image data set, the voice data set and the signal data set into the recognition model Shmx, analyzes and generates a guidance data group Zdsj, and transmits it to the AI ​​service unit via the network. The AI ​​service unit transmits the guidance data group Zdsj to the central processing unit MCU via the network. The central processing unit MCU controls the AI ​​digital human SPI display screen to explain the solution data in the guidance data group Zdsj to the user in the form of the voice and picture of the AI ​​digital human.

[0011] Preferably, the image interaction camera, voice interaction MIC and AI digital human SPI display screen are all arranged on the outer surface of the AI ​​charging pile, receiving external sound and light information in real time and generating sound and image feedback.

[0012] Preferably, the vehicle-pile communication unit is connected to a 4G module via a network, and collects signal data sets in chronological order from early to late, wherein the signal data sets include communication signals output by the AI ​​charging pile, and the data acquisition module transmits image data sets, voice data sets and signal data sets to the interactive service module via the network; the interactive monitoring and charging controller includes an image processor GPU and a central processing unit MCU, and the image processor GPU and the central processing unit MCU are interconnected via a network, the image processor GPU is connected to an image interactive camera and an AI digital human SPI display screen via a network, and the central processing unit MCU is connected to a voice interactive MIC, a main circuit output system and a gun line and gun head operating component via a network.

[0013] Preferably, the expression of the image data set is to The user operation images corresponding to each time point are shown in sequence, t represents the time point when the image interaction camera captures the user operation, x represents the user operation content captured by the image interaction camera, and 1 to n represent that there are n groups of user operation images.

[0014] Preferably, the expression of the speech data set is to The user voice data corresponding to each time point in turn, s represents the time point when the voice interaction MIC recognizes the user voice, g represents the keyword in the user voice recognized by the voice interaction MIC, and 1 to m represents that there are m groups of user voice data.

[0015] Preferably, the expression of the signal data set is to They correspond to the communication signals output by the AI ​​charging pile at different time points, c represents the time point when the AI ​​charging pile outputs the communication signal, h represents the signal type of the communication signal output by the AI ​​charging pile, and the signal types include vehicle-pile message signals and alarm signals, and 1 to o represents the number of times the AI ​​charging pile outputs the communication signal is o times.

[0016] Preferably, the image processor GPU identifies multiple users in the image data set, classifies and selects users who successfully charge for the first time and users who fail to charge for the first time, and marks the operation images of users who successfully charge for the first time as CG, and marks the operation images of users who fail to charge for the first time as FA.

[0017] Preferably, the cloud-based AI deployment is provided with an AI computing chip, and the cloud-based AI deployment analyzes the user's operation sequence in the operation image CG, the user's operation sequence in the operation image FA, and the guidance sequence in the AI ​​charging pile operation manual ZY through the AI ​​computing chip, and compares the user's operation sequence in the operation image CG with the guidance sequence in the AI ​​charging pile operation manual ZY, and the expression is ZY-CG. The analysis obtains the operation sequence in which the first charging is successfully started even if the operation is not performed in the guidance sequence, and marks it as CGC. Then, the user's operation sequence in the operation image FA is compared with the guidance sequence in the AI ​​charging pile operation manual ZY, and the expression is ZY-FA. The analysis obtains the operation sequence in which the first charging fails even if the operation is not performed in the guidance sequence, and marks it as FAC.

[0018] Preferably, the cloud AI deployment substitutes the operation sequence CGC and the operation sequence FAC into the CNN algorithm model, and trains to obtain the recognition model Shmx, whose calculation formula is as follows:

[0019] Shmx=∑CNN (CGC+FAC+ZY)

[0020] In the formula, Shmx represents the recognition model, ∑ CNN (CGC+FAC+ZY) means that the CNN algorithm model performs deep learning based on the operation sequence CGC when the first charging is successful even if the operation is not performed according to the instruction sequence, the operation sequence FAC when the first charging fails even if the operation is not performed according to the instruction sequence, and the AI ​​charging pile operation manual ZY, and trains a recognition model to determine whether the user operation is correct.

[0021] Preferably, the calculation formula of the guidance data set Zdsj is as follows:

[0022]

[0023] In the formula, Zdsj represents the guidance data set, t=s=c represents the user operation images, user voice data and communication signals at the same time point in the extracted image data set, voice data set and signal data set. It indicates that the recognition model determines whether the user operation is correct and obtains solution data, i.e., the guidance data set, based on the user operation image, user voice keywords and communication signals at the current time point.

[0024] Compared with the prior art, the present invention provides a charging pile service interaction system based on AI technology, which has the following beneficial effects:

[0025] 1. The present invention sets an image interaction camera, a voice interaction MIC, an AI digital human SPI display screen, an interactive monitoring and charging controller, a 4G module, a main circuit output system and a gun line and gun head operation component through an AI charging pile to realize real-time monitoring of the user's operation stage and actual needs. The data acquisition module obtains in real time the sound and image information of human-machine operation encountered during the standby and start-stop charging process of the charging pile, as well as the vehicle-pile message and standby alarm signal output by the AI ​​charging pile, and classifies them into image data sets, voice data sets and signal data sets. The interactive service module combines the AI ​​charging pile operation manual ZY with the user's actual operation image, trains to obtain the recognition model Shmx and analyzes to generate the guidance data set Zdsj, analyzes the fault and accurately locates the cause of the fault without human judgment and intervention, solves the problems of poor stability management capability, low user first start-up rate and poor service experience of the traditional charging pile service interactive system, and has strong intelligent interactive service capability.

[0026] 2. The present invention transmits the guidance data group Zdsj to the central processing unit MCU through the network through the interactive service module. The central processing unit MCU controls the AI ​​digital human SPI display screen to explain the solution data in the guidance data group Zdsj to the user in the form of AI digital human voice and picture, corrects the user's erroneous operation in time, and displays fault and charging guidance solutions. The user's first charging rate is high and the stability is strong. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a schematic diagram of the system flow of the present invention. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] Since the traditional charging pile service interaction system is used in external operation of charging piles, there are certain deficiencies and untimely stability management. In the face of potential occasional failures or occasional failures of the new energy vehicle battery management system BMS, it is easy to fail to start charging or charge unstably. In addition, in actual use, the installation environment of charging piles is complex and changeable. When some car owners use charging piles, they are prone to misoperation due to unfamiliarity with the equipment. There are as many as hundreds of charging pile equipment error conditions. On-site managers and remote customer service cannot fully, timely and accurately analyze the source of car owners' complaints, and the equipment fault handling ability is poor. These problems jointly lead to a low first-time charging rate and poor charging experience for car owners, which in turn reduces their confidence and satisfaction with new energy vehicles and has a negative impact on the overall reputation of the new energy industry. Therefore, a charging pile service interaction system based on AI technology is provided. Please refer to Figure 1 ,The system includes AI charging pile, data acquisition module and interactive service module;

[0030] The AI ​​charging pile includes an image interaction camera, a voice interaction MIC, an AI digital human SPI display screen, an interactive monitoring and charging controller, a 4G module, a main circuit output system and a gun line and gun head operation component to realize real-time monitoring of the user's operation stage and actual needs. The image interaction camera, the voice interaction MIC and the AI ​​digital human SPI display screen are all arranged on the outer surface of the AI ​​charging pile to receive external sound and light information in real time and generate sound and image feedback. The AI ​​charging pile is fixedly installed with an interactive monitoring and charging controller and a 4G module, and the interactive monitoring and charging controller and the 4G module are interconnected through a network. The interactive monitoring and charging controller includes an image processor GPU and a central processing unit MCU, and the image processor GPU and the central processing unit MCU are interconnected through a network. The image processor GPU is connected to the image interaction camera and the AI ​​digital human SPI display screen through the network, and the central processing unit MCU is connected to the voice interaction MIC, the main circuit output system and the gun line and gun head through the network. Operation component, 4G module receives data requests from interactive monitoring and charging controller and processes replies. In actual use, the intelligent Internet of Things-cloud AI control platform system or the local AI control charging system can be formed according to the installation environment of the AI ​​charging pile. The intelligent Internet of Things-cloud AI control platform system or the local AI control charging system are both built based on the charging pile service interaction system of AI technology. The difference between the two lies mainly in the way the 4G module processes and replies to data requests from interactive monitoring and charging controller. As the former, the intelligent Internet of Things-cloud AI control platform system receives interactive information feeds of AI charging piles at different stations and different user operations through cloud AI deployment, and trains the recognition model Shmx. As the latter, the local AI control charging system receives interactive information feeds of AI charging piles at the same station and different user operations through the local AI computing chip, and trains the recognition model Shmx. Among them, the local AI computing chip is fixedly installed inside the AI ​​charging pile;

[0031] The data acquisition module includes a camera unit, a sound sensor unit, and a vehicle-pile communication unit. The camera unit is connected to the image interaction camera through the network and collects image data sets from morning to night in chronological order. The image data set includes user operation images at all time points. The expression of the image data set is to The user operation images corresponding to each time point are image data of fixed duration, which contain user appearance features, operation sequence and operation content information. t represents the time point when the image interaction camera captures the user operation, x represents the user operation content captured by the image interaction camera, and 1 to n represent that there are n groups of user operation images;

[0032] The sound sensor unit is connected to the voice interaction MIC through the network, and collects voice data sets from early to late in chronological order. The voice data set includes user voice data at all time points. The expression of the voice data set is to The user voice data corresponding to each time point in turn, s represents the time point when the voice interaction MIC recognizes the user voice, g represents the keyword in the user voice recognized by the voice interaction MIC, the keyword includes but is not limited to help me charge, turn off the charging pile, how to charge or how to charge, 1 to m represents that there are m groups of user voice data;

[0033] The vehicle-pile communication unit is connected to the 4G module through the network and collects signal data sets from morning to night in chronological order. The signal data set includes the communication signal output by the AI ​​charging pile. The expression of the signal data set is to They correspond to the communication signals output by the AI ​​charging pile at different time points. c represents the time point when the AI ​​charging pile outputs the communication signal. h represents the signal type of the communication signal output by the AI ​​charging pile. The signal types include vehicle-pile message signals and alarm signals. The vehicle-pile message is a tool for information transmission between the charging pile and the electric vehicle. The message is used to confirm each other's identity and status to ensure the smooth charging process. The alarm signal is a fault signal issued by the AI ​​charging pile itself when it detects an abnormal situation. It is used to prompt the management center or the user to take corresponding measures to avoid potential safety risks. 1 to o represents the number of times the AI ​​charging pile outputs the communication signal o times;

[0034] The data acquisition module transmits the image data set, voice data set and signal data set to the interactive service module through the network. The image data set, voice data set and signal data set contain the sound and image information of human-machine operation encountered during the standby and start-stop charging process of the charging pile, as well as the vehicle-pile message and standby alarm signal output by the AI ​​charging pile;

[0035] The interactive service module includes a deep learning unit, a real-time analysis unit and an AI service unit. The deep learning unit connects the image processor GPU and the cloud AI deployment through the network. If the manager chooses to build a local AI control charging system to manage the AI ​​charging pile during actual use, the cloud AI deployment will be replaced by the local AI computing chip fixedly installed inside the AI ​​charging pile.

[0036] The image processor GPU identifies multiple users in the image data set, classifies and selects users who successfully enable charging for the first time and users who fail to enable charging for the first time, and marks the operation image of the user who successfully enables charging for the first time as CG, and marks the operation image of the user who fails to enable charging for the first time as FA. The two types of image data selected by classification are the input items of the subsequent CNN algorithm model;

[0037] The cloud AI deployment is connected to the database through the network. The database stores and backs up the AI ​​charging pile operation manual ZY. The cloud AI deployment is provided with an AI computing chip. The cloud AI deployment analyzes the operation sequence of the user in the operation image CG, the operation sequence of the user in the operation image FA, and the instruction sequence in the AI ​​charging pile operation manual ZY through the AI ​​computing chip. The instruction sequence in the AI ​​charging pile operation manual ZY is the hyperparameter of the CNN algorithm model. The operation sequence of the user in the operation image CG is compared with the instruction sequence in the AI ​​charging pile operation manual ZY, and its expression is ZY-CG. The analysis obtains the operation sequence of the first successful charging without following the instruction sequence, and marks it as CGC. Then, the operation sequence of the user in the operation image FA is compared with the instruction sequence in the AI ​​charging pile operation manual ZY, and its expression is ZY-FA. The analysis obtains the operation sequence of the first failed charging without following the instruction sequence, and marks it as FAC.

[0038] The cloud AI deployment substitutes the operation sequence CGC and the operation sequence FAC into the CNN algorithm model and trains the recognition model Shmx.

[0039] Shmx=∑ CNN (CGC+FAC+ZY)

[0040] In the formula, Shmx represents the recognition model, ∑ CNN (CGC+FAC+ZY) means that the CNN algorithm model performs deep learning based on the operation sequence CGC for the first successful charging without following the instruction sequence, the operation sequence FAC for the first failed charging without following the instruction sequence, and the AI ​​charging pile operation manual ZY, and trains a recognition model to determine whether the user operation is correct. The recognition model can not only recognize the user's operation content and operation sequence, but also analyze the fault and accurately locate the cause of the fault by combining the vehicle-side BMS message and I / O module detection signal obtained by the CAN communication on the original charging pile gun line gun head operation components S+ and S-;

[0041] The specific steps for substituting numerical values ​​into the CNN algorithm model for training are as follows:

[0042] 1. The operation image CG of the user who successfully started charging for the first time and the operation image FA of the user who failed to start charging for the first time are used as model input items. At the same time, the cloud AI deployment uses the AI ​​computing chip to analyze the operation sequence CGC of the user who successfully started charging for the first time without following the instruction sequence, the operation sequence FAC of the user who failed to start charging for the first time without following the instruction sequence, and the AI ​​charging pile operation manual ZY, which are also used as model input items and substituted into the CNN algorithm model;

[0043] 2. The convolution layer enhances the user's operation features in the image and encodes the user's operation features in chronological order. Then, the text data in the AI ​​charging pile operation manual ZY is parsed through natural language processing technology NLP;

[0044] 3. The superposition of multiple convolution and pooling layers arranges the operation feature encoding, the fully connected layer processes the dependency between the predicted operation sequences, and the output layer fits and generates multiple charging sequences according to different charging operation sequences and predicts the rationality score of the operation sequence. Among them, the charging operation sequence output by the CNN algorithm model is completely consistent with the AI ​​charging pile operation manual ZY, and the rationality score is 100. The charging operation sequence output by the CNN algorithm model is completely consistent with the operation sequence CGC of the first successful charging without following the instruction sequence, and the rationality score is 60. The charging operation sequence output by the CNN algorithm model is completely consistent with the operation sequence FAC of the first failed charging without following the instruction sequence, and the rationality score is 0.

[0045] The trained CNN algorithm model is the recognition model Shmx, which can perform rationality scoring based on real-time user operation images, user voice data and communication signals. If the rationality score is lower than 60, the recognition model Shmx determines that there are erroneous operations and equipment failures in the user operation images, user voice keywords and communication signals at the current time point;

[0046] The real-time analysis unit substitutes the image data set, voice data set, and signal data set into the recognition model Shmx, analyzes and generates the guidance data set Zdsj, and transmits it to the AI ​​service unit through the network. The calculation formula is as follows:

[0047]

[0048] In the formula, Zdsj represents the guidance data set, t=s=c represents the user operation images, user voice data and communication signals at the same time point in the extracted image data set, voice data set and signal data set. The recognition model determines whether the user operation is correct and obtains solution data, i.e., guidance data set, based on the user operation image, user voice keywords and communication signals at the current time point. It does not require human judgment and intervention, and solves the problems of poor stability management capability, low user first-time charging rate and poor service experience of the traditional charging pile service interaction system, and has strong intelligent interactive service capabilities.

[0049] The AI ​​service unit transmits the guidance data group Zdsj to the central processing unit MCU through the network. The central processing unit MCU controls the AI ​​digital human SPI display screen to explain the solution data in the guidance data group Zdsj to the user in the form of AI digital human's voice and picture, correct the user's wrong operation in time, and display the fault and charging guidance solutions. The user's first charging rate is high and the stability is strong.

[0050] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A charging pile service interaction system based on AI technology, characterized by: Including AI charging pile, data collection module and interactive service module; The AI ​​charging pile includes an image interaction camera, a voice interaction MIC, an AI digital human SPI display screen, an interactive monitoring and charging controller, a 4G module, a main circuit output system, and a gun line and gun head operation component; The data acquisition module includes a camera image unit, a sound sensor unit and a vehicle-pile communication unit. The camera image unit is connected to the image interaction camera through a network, and collects image data sets from early to late in chronological order, and the image data sets include user operation images at all time points; the sound sensor unit is connected to the voice interaction MIC through a network, and collects voice data sets from early to late in chronological order, and the voice data sets include user voice data at all time points; The interactive service module includes a deep learning unit, a real-time analysis unit and an AI service unit. The deep learning unit is connected to an image processor GPU and a cloud AI deployment via a network. The cloud AI deployment is connected to a database via a network. The database stores and backs up the AI ​​charging pile operation manual ZY. The cloud AI deployment is provided with a CNN algorithm model, and performs model training in combination with the AI ​​charging pile operation manual ZY and an image data set. The recognition model Shmx is analyzed and generated, and is transmitted to the real-time analysis unit via the network. The real-time analysis unit substitutes the image data set, the voice data set and the signal data set into the recognition model Shmx, analyzes and generates a guidance data group Zdsj, and transmits it to the AI ​​service unit via the network. The AI ​​service unit transmits the guidance data group Zdsj to the central processing unit MCU via the network. The central processing unit MCU controls the AI ​​digital human SPI display screen to explain the solution data in the guidance data group Zdsj to the user in the form of the voice and picture of the AI ​​digital human.

2. The charging pile service interaction system based on AI technology according to claim 1 is characterized by: The image interaction camera, voice interaction MIC and AI digital human SPI display screen are all arranged on the outer surface of the AI ​​charging pile, receiving external sound and light information in real time and generating sound and image feedback.

3. The charging pile service interaction system based on AI technology according to claim 1 is characterized by: The vehicle-pile communication unit is connected to the 4G module through the network, and collects signal data sets in chronological order from early to late. The signal data set includes the communication signal output by the AI ​​charging pile. The data acquisition module transmits the image data set, voice data set and signal data set to the interactive service module through the network; the interactive monitoring and charging controller includes an image processor GPU and a central processing unit MCU, and the image processor GPU and the central processing unit MCU are interconnected through the network. The image processor GPU is connected to the image interaction camera and the AI ​​digital human SPI display screen through the network, and the central processing unit MCU is connected to the voice interaction MIC, the main circuit output system and the gun line and gun head operation components through the network.

4. The charging pile service interaction system based on AI technology according to claim 3 is characterized by: The expression of the image dataset is: to The user operation images corresponding to each time point are shown in sequence, t represents the time point when the image interaction camera captures the user operation, x represents the user operation content captured by the image interaction camera, and 1 to n represent that there are n groups of user operation images.

5. The charging pile service interaction system based on AI technology according to claim 4 is characterized by: The expression of the speech dataset is: to The user voice data corresponding to each time point in turn, s represents the time point when the voice interaction MIC recognizes the user voice, g represents the keyword in the user voice recognized by the voice interaction MIC, and 1 to m represents that there are m groups of user voice data.

6. The charging pile service interaction system based on AI technology according to claim 5 is characterized by: The expression of the signal data set is to They correspond to the communication signals output by the AI ​​charging pile at different time points, c represents the time point when the AI ​​charging pile outputs the communication signal, h represents the signal type of the communication signal output by the AI ​​charging pile, and the signal types include vehicle-pile message signals and alarm signals, and 1 to o represents the number of times the AI ​​charging pile outputs the communication signal is o times.

7. The charging pile service interaction system based on AI technology according to claim 6 is characterized by: The image processor GPU identifies multiple users in the image data set, classifies and selects users who successfully enable charging for the first time and users who fail to enable charging for the first time, and marks the operation images of users who successfully enable charging for the first time as CG, and marks the operation images of users who fail to enable charging for the first time as FA.

8. The charging pile service interaction system based on AI technology according to claim 7 is characterized by: The cloud-based AI deployment is provided with an AI computing chip. The cloud-based AI deployment analyzes the user's operation sequence in the operation image CG, the user's operation sequence in the operation image FA, and the guidance sequence in the AI ​​charging pile operation manual ZY through the AI ​​computing chip, and compares the user's operation sequence in the operation image CG with the guidance sequence in the AI ​​charging pile operation manual ZY, and the expression is ZY-CG. The analysis obtains the operation sequence in which the first charging is successfully started even if the operation is not performed in the guidance sequence, and the operation sequence is marked as CGC. Then, the user's operation sequence in the operation image FA is compared with the guidance sequence in the AI ​​charging pile operation manual ZY, and the expression is ZY-FA. The analysis obtains the operation sequence in which the first charging fails even if the operation is not performed in the guidance sequence, and the operation sequence is marked as FAC.

9. The charging pile service interaction system based on AI technology according to claim 8 is characterized by: The cloud AI deployment substitutes the operation sequence CGC and the operation sequence FAC into the CNN algorithm model and trains to obtain the recognition model Shmx, whose calculation formula is as follows: Shmx=Σ CNN (CGC+FAC+ZY) In the formula, Shmx represents the recognition model, ∑ CNN (CGC+FAC+ZY) means that the CNN algorithm model performs deep learning based on the operation sequence CGC when the first charging is successful even if the operation is not performed according to the instruction sequence, the operation sequence FAC when the first charging fails even if the operation is not performed according to the instruction sequence, and the AI ​​charging pile operation manual ZY, and trains a recognition model to determine whether the user operation is correct.

10. The charging pile service interaction system based on AI technology according to claim 9 is characterized by: The calculation formula of the guidance data set Zdsj is as follows: In the formula, Zdsj represents the guidance data set, t=s=c represents the user operation images, user voice data and communication signals at the same time point in the extracted image data set, voice data set and signal data set. It indicates that the recognition model determines whether the user operation is correct and obtains solution data, i.e., the guidance data set, based on the user operation image, user voice keywords and communication signals at the current time point.