Unmanned toll lane intelligent voice response method and device based on large language model, equipment and medium

Abstract

The embodiment of the invention discloses an unmanned toll lane intelligent voice response method and device based on a large language model, equipment and a medium, and the method comprises the steps: firstly, carrying out the collection of the data information of a vehicle through a sensing device based on an unmanned toll lane, and generating input data; afterwards, based on the architecture principle of the model, the big language model is finely adjusted through preset historical work order data and preset rules and regulations, and a charging scene big language model is generated; then, on the basis of a fine-tuning model, semantic understanding is carried out on input data, and a problem complexity value is generated; and meanwhile, performing emotional state judgment on the input data to generate an emergency degree value. And finally, a response scheme is set based on the problem complexity value and the emergency degree value, and unmanned toll lane intelligent voice response is completed. According to the embodiment of the invention, the method achieves the precise understanding of complex semantics, guarantees the voice response intelligence and accuracy in an unattended lane, and improves the passing efficiency and user experience.

Description

Technical Field

[0001] The present disclosure relates to intelligent voice response technology, and in particular to an intelligent voice response method, device, equipment and medium for unmanned toll lanes based on a large language model. Background Art

[0002] With the continuous development of highways and the increase in traffic volume, the traditional manual toll collection model can no longer meet the needs of modern traffic management. Toll station staff not only need to collect vehicle tolls and issue cards, but also need to deal with sudden abnormal situations, which leads to high workload and long duty time, and increased safety risks. In addition, the high labor cost required for manual toll collection puts highway operating units under tremendous pressure.

[0003] Although the current unmanned lane artificial intelligence customer service system has achieved automated responses to a certain extent, it still has some shortcomings. First of all, the existing technical solutions mainly rely on rule engines and preset question and answer libraries, which are difficult to deal with complex customer inquiries. This single response method makes the system insufficiently flexible and adaptable, unable to meet the diverse needs of drivers and passengers, and affects the interactive experience. Secondly, the processing of emotional judgment is also relatively limited. Existing solutions usually require the introduction of additional facial image data to judge the customer's emotions, which not only increases the complexity of technical implementation, but may also cause privacy concerns.

[0004] Therefore, one or more methods are needed to solve the above problems.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0006] The purpose of the present disclosure is to provide an intelligent voice response method, device, equipment and medium for unmanned toll lanes based on a large language model, thereby overcoming one or more problems caused by limitations and defects of related technologies at least to a certain extent.

[0007] According to one aspect of the present disclosure, there is provided an intelligent voice response method for an unmanned toll lane based on a large language model, comprising:

[0008] Based on the sensing equipment of the unmanned toll lane, the input data is generated by collecting the data information of the vehicle;

[0009] Based on the multi-head attention mechanism of the Transformer architecture of the large language model, the large language model is fine-tuned by pre-setting historical work order data and pre-setting rules and regulations to generate a large language model for charging scenarios;

[0010] Based on the charging scenario language model, the real intention of the driver and passenger is generated by semantically understanding the input data, and the complexity of the real intention is judged to generate a problem complexity value;

[0011] Based on the charging scenario large language model, generating an urgency value by judging the emotional state of the input data;

[0012] Based on the problem complexity value and urgency value, the intelligent voice response of the unmanned toll lane is completed by setting the intelligent voice response plan.

[0013] In an exemplary embodiment of the present disclosure, the data information of the vehicle is collected, including:

[0014] A sensor is provided at the front end of the unmanned toll lane. When a vehicle passes the sensor, the vehicle is identified by a visual detection device to generate vehicle identification information.

[0015] A microphone is externally installed in the middle of the unmanned toll lane, and the microphone is used to recognize the customer service request voice and generate an audio request signal;

[0016] Input data is generated by integrating the vehicle identification information and the audio request signal.

[0017] In an exemplary embodiment of the present disclosure, the large language model is fine-tuned by presetting historical work order data and presetting rules and regulations, including:

[0018] Based on the Transformer architecture model, the ability to handle faults in charging scenarios is enhanced by presetting historical work order data, and an enhanced fault handling solution is generated;

[0019] Based on the Transformer architecture model, the ability to respond to regulatory issues is enhanced through preset rules and regulations, and enhanced solutions for responding to regulatory issues are generated;

[0020] Based on the enhanced fault handling solution and the enhanced regulatory issue response solution, a charging scenario large language model is generated by fine-tuning the large language model.

[0021] In an exemplary embodiment of the present disclosure, the real intention of the driver and passenger is generated by semantically understanding the input data, and the complexity of the real intention is judged, including:

[0022] Based on the vehicle identification information, dynamic traffic information is generated by searching the vehicle history information in the system database;

[0023] Based on analog-to-digital conversion technology, the audio request signal is sampled and quantized to generate an original digital signal, and the original digital signal is preprocessed to generate a processed digital signal;

[0024] Based on modern speech recognition technology, the feature vectors of the processed digital signal are extracted and matched to generate acoustic scores, and the phrase sequence is predicted by the acoustic scores to generate recognized text information;

[0025] An integrated input package is generated by integrating the identified text information and dynamic traffic information, and a model input data packet is generated by performing data format processing on the integrated input package.

[0026] In an exemplary embodiment of the present disclosure, the real intention of the driver and passenger is generated by semantically understanding the input data, and the complexity of the real intention is judged, including:

[0027] Based on the large language model of the charging scenario, the real intention of the driver and passengers is generated by performing semantic reasoning on the input data packet of the model;

[0028] Based on the charging scenario large language model, a question complexity value is generated by determining the complexity of the real intention.

[0029] In an exemplary embodiment of the present disclosure, judging the emotional state of the input data includes:

[0030] Based on the toll scene large language model, the emotional state of the driver and passengers is generated by evaluating the model input data packet;

[0031] Based on the emotional state, the urgency of the problem is judged through the charging scenario large language model to generate an urgency value.

[0032] In an exemplary embodiment of the present disclosure, the answering scheme of the intelligent voice is set, including:

[0033] When the complexity value and the urgency value of the question are less than the preset values, a reply text is generated by using the charging scenario large language model based on a preset knowledge base to generate a reply text;

[0034] Based on speech synthesis technology, the reply text is converted into natural language to generate a reply voice;

[0035] A loudspeaker is externally provided in the middle of the unmanned toll lane, and the reply voice is played through the loudspeaker to complete the intelligent voice response of the unmanned toll lane;

[0036] When the complexity value and urgency value of the problem are not less than the preset values, the intelligent voice response of the unmanned toll lane is completed by intervening manual customer service.

[0037] In one aspect of the present disclosure, there is provided an intelligent voice response device for an unmanned toll lane based on a large language model, comprising:

[0038] Input data acquisition module, used to collect vehicle data information;

[0039] Large language model fine-tuning module, used to fine-tune the large language model by pre-setting historical work order data and pre-setting rules and regulations;

[0040] A question complexity value determination module is used to perform semantic understanding on the input data and determine the complexity of the generated real intent;

[0041] An urgency value determination module, used to determine the emotional state of the input data;

[0042] The voice response module is used to set the response plan for intelligent voice.

[0043] In one aspect of the present disclosure, there is provided an electronic device, comprising:

[0044] Processor; and

[0045] A memory having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by the processor, implement the method according to any one of the above items.

[0046] In one aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method according to any one of the above items is implemented.

[0047] Based on the embodiment of the present disclosure, first, based on the perception equipment of the unmanned toll lane, the data information of the vehicle is collected to generate input data. Afterwards, based on the architectural principle of the model, the large language model is fine-tuned by pre-setting historical work order data and pre-setting rules and regulations to generate a large language model for the toll scene. Then, based on the fine-tuned model, the problem complexity value is generated by semantically understanding the input data. At the same time, the urgency value is generated by judging the emotional state of the input data. Finally, based on the problem complexity value and the urgency value, a response plan is set to complete the intelligent voice response of the unmanned toll lane. Therefore, the embodiment of the present disclosure provides a method for responding to the intelligent voice of the unmanned toll lane with a fine-tuned large language model, which achieves accurate understanding of complex semantics, ensures the intelligence and accuracy of voice responses in unmanned lanes, and improves traffic efficiency and user experience.

[0048] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure.

[0049] The technical solution of the present disclosure is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings, which constitute a part of the specification, illustrate embodiments of the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0051] The present disclosure may be more clearly understood from the following detailed description with reference to the accompanying drawings, in which:

[0052] Figure 1 A flowchart of an intelligent voice response method for an unmanned toll lane based on a large language model according to an embodiment of the disclosed method;

[0053] Figure 2 A decision logic flow chart of an intelligent voice response method for an unmanned toll lane based on a large language model according to an embodiment of the disclosed method;

[0054] Figure 3 A structural block diagram of an intelligent voice response device for an unmanned toll lane based on a large language model according to an embodiment of the disclosed method;

[0055] Figure 4 A block diagram of an electronic device according to an embodiment of the method disclosed herein. DETAILED DESCRIPTION

[0056] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be comprehensive and complete and will fully convey the concepts of the example embodiments to those skilled in the art. The same reference numerals in the figures represent the same or similar parts, and thus their repeated description will be omitted.

[0057] In addition, the described features, structures or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced without one or more of the specific details, or other methods, components, materials, devices, steps, etc. may be adopted. In other cases, known structures, methods, devices, implementations, materials or operations are not shown or described in detail to avoid blurring the various aspects of the present disclosure.

[0058] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or these functional entities or parts of functional entities may be implemented in one or more software hardened modules, or these functional entities may be implemented in different networks and / or processor devices and / or microcontroller devices.

[0059] In the embodiment of the present disclosure, firstly, a method for intelligent voice response of unmanned toll lane based on a large language model is provided; Figure 1 As shown in , the intelligent voice response method for unmanned toll lane based on a large language model may include the following steps:

[0060] Step S110, based on the sensing device of the unmanned toll lane, the data information of the vehicle is collected to generate input data;

[0061] Step S120, based on the multi-head attention mechanism of the Transformer architecture of the large language model, the large language model is fine-tuned by pre-setting historical work order data and pre-setting rules and regulations to generate a large language model for charging scenarios;

[0062] Step S130, based on the charging scenario large language model, the real intention of the driver and passenger is generated by semantically understanding the input data, and the complexity of the real intention is judged to generate a problem complexity value;

[0063] Step S140, generating an urgency value by judging the emotional state of the input data based on the charging scenario large language model;

[0064] Step S150, based on the problem complexity value and the urgency value, the intelligent voice response scheme is set to complete the intelligent voice response of the unmanned toll lane.

[0065] Below, as Figure 2 As shown, an intelligent voice response method for an unmanned toll lane based on a large language model in an embodiment of the present disclosure will be further explained.

[0066] In step S110, input data may be generated by collecting data information of the vehicle based on the sensing device of the unmanned toll lane.

[0067] In some optional embodiments of this example, when a vehicle enters an unmanned toll lane, the vehicle is identified by a sensor disposed at the front end of the lane. Afterwards, the vehicle type, license plate number and other vehicle identification information are identified by a visual detection device disposed in the lane, so that the vehicle information can be subsequently obtained in the system knowledge base.

[0068] At the same time, voice prompts will be automatically played in the unmanned toll lane (which will guide drivers and passengers in subsequent payment, consultation and other operations), and the external microphone in the middle of the unmanned toll lane will recognize the voice of customer service requests (such as charging rules, road condition inquiries, reporting equipment failures, etc.) made by drivers and passengers to generate audio request signals.

[0069] Finally, the recognized vehicle identification information and audio request signal are integrated into input data and sent to the large language model for semantic understanding and emotional state determination.

[0070] In step S120, based on the multi-head attention mechanism of the Transformer architecture of the large language model, the large language model can be fine-tuned by pre-setting historical work order data and pre-setting rules and regulations to generate a large language model for charging scenarios.

[0071] Prior to some optional embodiments of this example, the large language model of the Transformer architecture was used as a prototype. According to its multi-head attention mechanism and the actual situation of the toll stations at the entrances and exits of the highway, the large language model was fine-tuned by presetting historical work order data and presetting rules and regulations in the large language model to construct a large language model suitable for the toll scenario of this example.

[0072] The large language model built using the Transformer architecture is an innovative method with the self-attention mechanism as the core, which can significantly improve the processing efficiency and effect of the language model. The calculation formula of the multi-head attention mechanism is as follows:

[0073]

[0074] MultiHead(Q,K,V)=Concat(head1,…,head h )W O

[0075] head i =Attention(QW i Q ,KW i K ,VW i V )

[0076] Where Q (Query) represents the query vector; K (Key) represents the key vector; V (Value) represents the value vector; d k Represents the dimension of the key vector, used to scale the inner product; softmax is used for weight normalization; W i Q , W i K , W iV W represents the linear transformation matrix of the query, key, and value vectors, which is used to generate a specific representation for each attention head; O Represents the linear mapping matrix after multi-head attention; T represents transposition (interchanging the rows and columns of the matrix).

[0077] Using this large language model as a prototype allows the model to focus on the information in the input sequence from multiple different perspectives and generate richer representations.

[0078] Then, the large language model is fine-tuned in combination with the highway toll collection scenario to generate a large language model for the toll collection scenario. The large language model for the toll collection scenario can focus more on the toll collection scenario.

[0079] Specifically, it is possible, but not limited to, to fine-tune the large language model through the pre-set historical highway work order data to generate a fault handling enhancement solution. This solution can enhance the ability of the large language model to handle equipment failures and common problems in specific scenarios. When similar problems occur, corresponding solutions can be quickly given.

[0080] In addition, the large language model is fine-tuned using highway regulations to generate an enhanced solution for responding to regulatory questions. This solution enables the large language model to accurately respond to regulatory questions (such as toll standards, speed limits, etc.) and ensure compliance and accuracy when providing answers.

[0081] In step S130, based on the charging scenario language model, the real intention of the driver and passengers can be generated by semantically understanding the input data, and the complexity of the real intention can be judged to generate a problem complexity value.

[0082] In some optional embodiments of this example, before semantic understanding of the input data is performed through the charging scenario large language model, the input data needs to be preprocessed to facilitate model analysis.

[0083] First, based on the vehicle identification information collected by the above-mentioned visual detection device, a search is performed in the preset system database to obtain the historical information of the current vehicle, such as where the vehicle came from and where it went, so as to further obtain dynamic traffic information such as charging rules, current lane usage, traffic flow, whether there are accidents, etc.

[0084] At the same time, based on the analog-to-digital conversion technology, the collected audio request signal is sampled and quantized to form a processable original digital signal, and then the original digital signal is denoised, normalized, and framed to generate a clearer processed digital signal.

[0085] Then, based on modern speech recognition technology, speech features are extracted from the processed digital signal and converted into feature vectors.

[0086] Then, the feature vector is input into the acoustic model, and the acoustic score of the feature vector is calculated. Then, the relevant speech segments are identified based on the acoustic score, and possible phrase sequences are predicted. The predicted phrase sequence is decoded to generate recognized text information.

[0087] Finally, the recognized text information and dynamic traffic information are integrated into an input package. After data format processing, the generated model input data packet is input into the charging scenario large language model for processing.

[0088] In some optional embodiments of this example, the charging scenario large language model first receives and parses the user's question in the model input data packet, performs natural language processing on it, and understands the syntactic structure and semantic meaning.

[0089] Afterwards, by analyzing keywords and context, and combining historical data and the current scenario, we can gain an in-depth understanding of the user's question (whether they are seeking factual answers, conceptual explanations, or solutions) and determine the true intentions of the current driver and passengers.

[0090] After understanding the real intentions of the driver and passengers, the preset knowledge base is queried (the knowledge base covers highway regulations, toll station operating procedures, common troubleshooting methods and steps, etc.), from which the complexity value of the answer to the current question is obtained, and a question complexity value is generated for subsequent judgment.

[0091] By converting the problem complexity into an estimate comparison, we can more accurately assess the complexity of the problem. For example, some recurring equipment problems will be judged as low value and can be solved automatically by the system without human intervention.

[0092] In step S140, an urgency value may be generated by judging the emotional state of the input data based on the charging scenario large language model.

[0093] In some optional embodiments of this example, simultaneously with the above steps, the historical conversation information in the model input data packet is analyzed through the charging scenario large language model to obtain the current emotional state of the driver and passengers (calm, stable, angry, anxious, etc.).

[0094] Then, based on the emotional state, the urgency of the questions raised by the driver and passengers is judged through the large language model of the charging scenario, and an urgency value is generated for subsequent judgment.

[0095] In step S150, the intelligent voice response of the unmanned toll lane can be completed by setting the intelligent voice response plan based on the problem complexity value and the urgency value.

[0096] In some optional embodiments of this example, first, preset standard values ​​for the complexity of the question and the urgency of the emotion are set in the system according to different charging scenarios. Then, the complexity value and the urgency value of the question are compared with the preset values.

[0097] When the problem complexity value and urgency value are not less than the preset values, it proves that the problem is too complicated or the situation is too urgent. The system determines that the automatic reply cannot handle it and transfers it to manual customer service to ensure that the problem is handled professionally.

[0098] When the complexity and urgency of the question are less than the preset values, the toll scene language model extracts the answer to the driver's true intention from the knowledge base, and combines it with the driver's historical conversation information to generate a compliant and complete reply text. For example, when the driver is calm and stable, the urgency of the question can generally be determined to be low, and a normal automatic reply process can be arranged to avoid manual intervention and improve efficiency.

[0099] Then, by using a speech synthesis algorithm, the reply text generated by the charging scene language model is converted into natural language, thus ensuring the fluency and clarity of the reply voice. Specifically, the reply text generated by the charging scene language model is recorded as R, and the text information is converted into a text embedding vector R emb , and then embed the text into the vector R through the text-to-speech model emb Converted into an audio signal S.

[0100] Finally, the generated reply voice is played to the driver and passengers through the external speaker in the middle of the unmanned toll lane, which indicates the end of the single intelligent voice interaction process.

[0101] This complete process demonstrates the efficient operation of the intelligent voice response method for unmanned toll lanes based on a large language model. This method can not only process conventional voice commands, but also intelligently determine whether human intervention is required through sentiment analysis and problem complexity assessment, thereby ensuring automation efficiency while improving user experience and accuracy in problem solving.

[0102] It should be noted that, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0103] In addition, in this exemplary embodiment, an intelligent voice response device for an unmanned toll lane based on a large language model is also provided. Figure 3 As shown, the intelligent voice response device 300 for an unmanned toll lane based on a large language model may include: an input data acquisition module 310, a large language model adjustment module 320, a problem complexity value determination module 330, an urgency value determination module 340 and a voice response module 350.

[0104] in:

[0105] An input data collection module 310 is used to collect data information of the vehicle;

[0106] A large language model fine-tuning module 320 is used to fine-tune the large language model by presetting historical work order data and presetting rules and regulations;

[0107] A question complexity value determination module 330 is used to perform semantic understanding on the input data and determine the complexity of the generated real intent;

[0108] The urgency value determination module 340 is used to determine the emotional state of the input data;

[0109] The voice response module 350 is used to set the response plan of the intelligent voice.

[0110] The unmanned toll lane intelligent voice response device based on a large language model in the embodiment of the present disclosure corresponds to the embodiment of the unmanned toll lane intelligent voice response method based on a large language model in the present disclosure, and the relevant contents can be referenced to each other, which will not be repeated here. The beneficial technical effects corresponding to the unmanned toll lane intelligent voice response device based on a large language model in the embodiment of the present disclosure can be referred to the corresponding beneficial technical effects of the above-mentioned corresponding exemplary method part, which will not be repeated here.

[0111] It should be noted that although several modules or units of the unmanned toll lane intelligent voice response device 300 based on a large language model are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be concretized in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into multiple modules or units to be concretized.

[0112] Below, reference Figure 4The electronic device according to the embodiment of the present disclosure is described. The electronic device may be any one or both of the first device and the second device, or a stand-alone device independent of them, and the stand-alone device may communicate with the first device and the second device to receive the collected input signals from them.

[0113] Figure 4 A block diagram of an electronic device according to an embodiment of the present disclosure is illustrated.

[0114] like Figure 4 As shown, the electronic device includes one or more processors and memory.

[0115] The processor may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.

[0116] The memory may store one or more computer program products, and the memory may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may include, for example, a read-only memory (ROM), a hard disk, a flash memory, etc. One or more computer program products may be stored on the computer-readable storage medium, and the processor may run the computer program product to implement the various embodiments of the present disclosure described above and / or other desired functions.

[0117] In one example, the electronic device may further include: an input device and an output device, and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0118] In addition, the input device may also include, for example, a keyboard, a mouse, and the like.

[0119] The output device can output various information to the outside, including determined distance information, direction information, etc. The output device can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.

[0120] Of course, to simplify, Figure 4 Only some of the components related to the present disclosure in the electronic device are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device may further include any other appropriate components.

[0121] In addition to the above methods and devices, an embodiment of the present disclosure may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the method according to various embodiments of the present disclosure described in the above part of this specification.

[0122] The computer program product may be written in any combination of one or more programming languages ​​to write program code for performing the operations of the disclosed embodiments, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0123] In addition, an embodiment of the present disclosure may also be a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, enables the processor to execute the steps of the method according to various embodiments of the present disclosure described in the above part of this specification.

[0124] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0125] The basic principles of the present disclosure are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present disclosure are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present disclosure. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, and are not limiting. The above details do not limit the present disclosure to the necessity of adopting the above specific details to be implemented.

[0126] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0127] The block diagrams of the devices, apparatuses, equipment, and systems involved in this disclosure are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including," "comprising," "having," and the like are open words, referring to "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or," and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0128] The method and apparatus of the present disclosure may be implemented in many ways. For example, the method and apparatus of the present disclosure may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present disclosure are not limited to the order specifically described above, unless otherwise specifically stated. In addition, in some embodiments, the present disclosure may also be implemented as a program recorded in a recording medium, which includes machine-readable instructions for implementing the method according to the present disclosure. Therefore, the present disclosure also covers a recording medium storing a program for executing the method according to the present disclosure.

[0129] It should also be noted that in the apparatus, device and method of the present disclosure, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present disclosure.

[0130] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0131] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. An intelligent voice response method for unmanned toll lanes based on a large language model, characterized in that: include: Based on the sensing equipment of the unmanned toll lane, the input data is generated by collecting the data information of the vehicle; Based on the multi-head attention mechanism of the Transformer architecture of the large language model, the large language model is fine-tuned by pre-setting historical work order data and pre-setting rules and regulations to generate a large language model for charging scenarios; Based on the charging scenario language model, the real intention of the driver and passenger is generated by semantically understanding the input data, and the complexity of the real intention is judged to generate a problem complexity value; Based on the charging scenario large language model, generating an urgency value by judging the emotional state of the input data; Based on the problem complexity value and urgency value, the intelligent voice response of the unmanned toll lane is completed by setting the intelligent voice response plan.

2. The method according to claim 1, characterized in that By collecting vehicle data information, including: A sensor is provided at the front end of the unmanned toll lane. When a vehicle passes the sensor, the vehicle is identified by a visual detection device to generate vehicle identification information. A microphone is externally installed in the middle of the unmanned toll lane, and the microphone is used to recognize the customer service request voice and generate an audio request signal; Input data is generated by integrating the vehicle identification information and the audio request signal.

3. The method according to claim 1, characterized in that The large language model is fine-tuned by presetting historical work order data and presetting rules and regulations, including: Based on the Transformer architecture model, the ability to handle faults in charging scenarios is enhanced by presetting historical work order data, and an enhanced fault handling solution is generated; Based on the Transformer architecture model, the ability to respond to regulatory issues is enhanced through preset rules and regulations, and enhanced solutions for responding to regulatory issues are generated; Based on the enhanced fault handling solution and the enhanced regulatory issue response solution, a charging scenario large language model is generated by fine-tuning the large language model.

4. The method according to claim 2, characterized in that: By semantically understanding the input data, the real intention of the driver and passengers is generated, and the complexity of the real intention is judged, including: Based on the vehicle identification information, dynamic traffic information is generated by searching the vehicle history information in the system database; Based on analog-to-digital conversion technology, the audio request signal is sampled and quantized to generate an original digital signal, and the original digital signal is preprocessed to generate a processed digital signal; Based on modern speech recognition technology, the feature vectors of the processed digital signal are extracted and matched to generate acoustic scores, and the phrase sequence is predicted by the acoustic scores to generate recognized text information; An integrated input package is generated by integrating the identified text information and dynamic traffic information, and a model input data packet is generated by performing data format processing on the integrated input package.

5. The method according to claim 4, characterized in that By semantically understanding the input data, the real intention of the driver and passengers is generated, and the complexity of the real intention is judged, including: Based on the large language model of the charging scenario, the real intention of the driver and passengers is generated by performing semantic reasoning on the input data packet of the model; Based on the charging scenario large language model, a question complexity value is generated by determining the complexity of the real intention.

6. The method according to claim 4, characterized in that By judging the emotional state of the input data, including: Based on the toll scene large language model, the emotional state of the driver and passengers is generated by evaluating the model input data packet; Based on the emotional state, the urgency of the problem is judged through the charging scenario large language model to generate an urgency value.

7. The method according to claims 5-6, characterized in that: By setting the intelligent voice response plan, including: When the complexity value and the urgency value of the question are less than the preset values, a reply text is generated by using the charging scenario large language model based on a preset knowledge base to generate a reply text; Based on speech synthesis technology, the reply text is converted into natural language to generate a reply voice; A loudspeaker is externally provided in the middle of the unmanned toll lane, and the reply voice is played through the loudspeaker to complete the intelligent voice response of the unmanned toll lane; When the complexity value and urgency value of the problem are not less than the preset values, the intelligent voice response of the unmanned toll lane is completed by intervening manual customer service.

8. An intelligent voice response device for unmanned toll lanes based on a large language model, characterized in that: include: Input data acquisition module, used to collect vehicle data information; Large language model fine-tuning module, used to fine-tune the large language model by pre-setting historical work order data and pre-setting rules and regulations; A question complexity value determination module is used to perform semantic understanding on the input data and determine the complexity of the generated real intent; An urgency value determination module, used to determine the emotional state of the input data; The voice response module is used to set the response plan for intelligent voice.

9. An electronic device, characterized in that: include: A memory for storing a computer program product; A processor is used to execute the computer program product stored in the memory, and when the computer program product is executed, it implements the method described in any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method described in any one of claims 1 to 7 is implemented.

Images