CN-121985072-A - Intelligent management and control method and system for tunnel telephone broadcasting system

CN121985072ACN 121985072 ACN121985072 ACN 121985072ACN-121985072-A

Abstract

The invention relates to the technical field of tunnel information, in particular to an intelligent management and control method and system of a tunnel telephone broadcasting system; the on-site video and audio data are collected, the event label and the urgency are identified by using the event classification model, the comprehensive priority is calculated by combining the section dynamic weight, the multipath emergency calls are intelligently ordered, the priority processing of high-risk events is ensured, and omission is avoided. And automatically uploading video and audio data to the cloud, and enabling a manager to remotely monitor the site environmental sound of any telephone terminal, so as to discover abnormality (such as quarry and abnormal equipment sound) in advance, thereby improving the early warning capability. And carrying out voice recognition on the connected call, extracting event entity combinations, automatically recommending the best matched broadcast template based on the broadcast template prediction model, reducing the complicated process of switching operation among a plurality of systems by a manager, and obviously improving the emergency response efficiency. The intelligent level and the disposal speed of the tunnel emergency management are improved by realizing the cooperation of multiple systems such as emergency telephone, broadcasting, monitoring and the like.

Inventors

LIAO ZHUANGZHI
LI QIANG
DONG YUXI
Tong Xingbin

Assignees

四川九通智路科技有限公司
四川乐西高速公路有限责任公司
四川高路信息科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260407

Claims (10)

1. An intelligent control method for a tunnel telephone broadcasting system is characterized by comprising the following steps: When emergency calls from a plurality of tunnel extensions are received at the same time, acquiring video and audio data collected by monitoring equipment in a group where the tunnel extensions are located, wherein the collected video and audio data are added into an uploading queue and uploaded to a cloud end for backup; Identifying the video and audio data based on a pre-constructed event classification model to obtain an event tag of the position of the tunnel extension and the emergency degree of the event tag, wherein the event tag comprises fire disaster, traffic accident, personnel help, equipment fault and no fault; weighting the emergency degree of the event label of each tunnel extension based on the section weight of the section where the tunnel extension is positioned to obtain the comprehensive priority of a plurality of emergency calls; Sequencing a plurality of emergency calls based on the comprehensive priority, and carrying out voice recognition and event entity extraction on the emergency calls in the process of being connected to obtain an event entity combination of the current emergency call; And predicting the probability model based on the pre-constructed broadcast templates and the probability of the event entity combination for predicting a plurality of broadcast templates required by the manager, and sending one or more broadcast modules with highest probability to the manager.
2. The intelligent management and control method of a tunnel telephone broadcasting system according to claim 1, wherein the network structure of the event classification model comprises: Video coding branch for extracting space-time characteristics of video data based on pre-trained 3DCNN module to obtain video characteristics ; An audio coding branch for extracting a spectrogram of audio data based on a pre-trained VGGish module to obtain audio features ; A multi-modal fusion layer for characterizing the video And the audio features Fusion is carried out to obtain the joint characteristics And based on the full connection layer, the joint characteristics Dimension reduction is carried out to obtain dimension reduction characteristics ; A multi-task output head for based on the dimension-reducing feature Predicted event tags and based on the dimension-reduction features And outputting the degree of urgency.
3. The intelligent management and control method of a tunnel telephone broadcasting system according to claim 2, wherein the process of constructing the event classification model comprises: s1, acquiring an audio and video data sample and a tag of the audio and video data, wherein the tag comprises a labeling event tag and a labeling emergency degree; s2, splitting the audio and video data samples into audio data samples and video data samples, and respectively inputting the audio data samples and the video data samples into the audio coding branch and the video coding branch to obtain a prediction event label and a prediction urgency outputted by the multi-task output head; S3, calculating the classification loss of the predicted event label and the labeling event label based on a pre-constructed classification loss function, and calculating the regression loss of the predicted emergency degree and the labeling emergency degree based on a pre-constructed regression loss function; s4, adjusting internal parameters of a network structure of the event classification model based on the classification loss and the regression loss back propagation; s5, repeating the steps S2-S4 until training is completed, and obtaining an event classification model.
4. The intelligent control method for a tunnel telephone broadcasting system according to claim 1, wherein the determining method for the section weights of the sections where the plurality of tunnel extensions are located comprises: Acquiring static parameters and dynamic parameters of a plurality of sections, wherein the static parameters comprise gradient, curve radius and historical accident rate, the dynamic parameters comprise average vehicle speed, vehicle flow, time occupancy, CO concentration, smoke concentration and rainfall intensity, and the dynamic parameters are statistical average values in target duration before a current time point; normalizing the static parameters and the dynamic parameters to obtain normalized static parameters and normalized dynamic parameters, wherein the values of the normalized static parameters and the normalized dynamic parameters are positively correlated with risks; Weighting the plurality of normalized static parameters based on the pre-constructed static parameter weights to obtain static weights, and weighting the plurality of normalized dynamic parameters based on the pre-constructed dynamic parameter weights to obtain dynamic weights; fusing the static weight and the dynamic weight to obtain section weights of sections where the tunnel extensions are located 。
5. The intelligent control method of a tunnel telephone broadcasting system according to claim 4, wherein the static parameter weight is determined based on expert evaluation, and the dynamic parameter weight determining method comprises: Calculate the first The section is at Specific gravity of dynamic parameters Wherein the specific gravity The mathematical expression of (2) is: In the formula, Represent the first The dynamic parameters of the normalization are used to normalize the dynamic parameters, Representing the number of segments; Based on the specific gravity Calculate the first Information entropy of normalized dynamic parameters And calculate the information entropy Calculate the first Coefficient of difference of normalized dynamic parameters Wherein the first Information entropy of normalized dynamic parameters And coefficient of difference The mathematical expressions of (a) are respectively: coefficient of difference for multiple dynamic parameters Carrying out normalization processing to obtain dynamic parameter weights of a plurality of dynamic parameters 。
6. The intelligent management and control method of a tunnel telephone broadcasting system according to claim 1, wherein the voice recognition and event entity extraction are performed on the emergency call in progress to obtain the event entity combination of the current emergency call, comprising: when an emergency call is connected, collecting real-time voice data, and identifying the real-time voice data to obtain an identification text; word segmentation processing is carried out on the identification text to obtain a plurality of words; converting the words into word vectors, and calculating cosine similarity between the word vectors and any event entity in a pre-constructed event entity library; When cosine similarity between the word vector and any event entity is greater than or equal to a preset similarity threshold, the word vector is used as the event entity, and event entity combinations are constructed based on all event entities extracted in the dialogue process, wherein each event entity combination comprises two event entities.
7. The intelligent management and control method for a tunnel telephone broadcasting system according to claim 6, wherein the method for constructing the broadcasting template prediction probability model comprises the following steps: Obtaining a plurality of historical emergency call event records, wherein the historical emergency call event records include event entity combination samples extracted from voice samples A broadcast template finally selected by a manager; counting each event entity combination sample from a plurality of historical emergency call event records And takes M event entity combination samples with highest frequency as key event entity combination And combining the other event entity combination samples as a low-frequency entity combination ; Constructing key event entity combinations Conditional probability of (2) Low frequency physical combination Conditional probability of (2) Wherein the conditional probability And the conditional probability The mathematical expression of (2) is: In the formula, Indicating when event entities are combined into Time broadcast template The probability of the selection is determined by the selection, A broadcast template is represented and is displayed, Representing key event entity combinations The total number of times of occurrence, In order to smooth the coefficient of the coefficient, A set of templates is represented and, Indicating the number of broadcast templates to be used, Indicating when event entities are combined into Time broadcast template The number of times selected; Indicating when event entities are combined into Time broadcast template The probability of the selection is determined by the selection, Representing low frequency event entity combinations The total number of times of occurrence, Indicating when event entities are combined into Time broadcast template The number of times selected; and constructing a broadcast template prediction probability model based on the conditional probabilities of the various event entity combinations.
8. The intelligent management and control method of a tunnel telephone broadcasting system according to claim 7, wherein predicting probability models based on pre-constructed broadcasting templates and the probability of the event entity combination predicting a plurality of broadcasting templates required by a manager, comprises: Substituting the event entity combinations into a pre-constructed broadcast template prediction probability model to obtain conditional probabilities of the event entity combinations; the probability of each broadcast module is extracted from the conditional probabilities of the combination of the event entities, and the probabilities of each broadcast module are summed to obtain the total probability of the probabilities of each broadcast module.
9. The intelligent management and control method of a tunnel telephone broadcasting system according to claim 1, further comprising: And acquiring broadcasting states of a plurality of sections in the tunnel, and feeding back the broadcasting states and the real-time playing content to the manager, wherein the broadcasting states comprise directional broadcasting, FM broadcasting or non-broadcasting.
10. An intelligent management and control system for a tunnel telephone broadcasting system, comprising: The system comprises an acquisition module, a cloud backup module and a cloud backup module, wherein the acquisition module is used for acquiring video and audio data acquired by monitoring equipment in a group where the tunnel extension is located when emergency calls from a plurality of tunnel extensions are received at the same time, wherein the acquired video and audio data are added into an uploading queue and uploaded to the cloud backup; The identification module is used for identifying the video and audio data based on a pre-constructed event classification model to obtain an event tag of the position of the tunnel extension and the emergency degree of the event tag, wherein the event tag comprises fire, traffic accidents, personnel help, equipment faults and no faults; The priority calculating module is used for weighting the emergency degree of the event label of each tunnel extension based on the section weight of the section where the plurality of tunnel extensions are positioned to obtain the comprehensive priority of a plurality of emergency calls; The entity extraction module is used for sequencing a plurality of emergency calls based on the comprehensive priority, and carrying out voice recognition and event entity extraction on the connected emergency calls to obtain an event entity combination of the current emergency call; and the broadcast mapping module is used for predicting the probability model based on the pre-constructed broadcast templates and the probability of the event entity combination for predicting the various broadcast templates required by the manager, and sending one or more broadcast modules with highest probability to the manager.

Description

Intelligent management and control method and system for tunnel telephone broadcasting system Technical Field The invention relates to the technical field of tunnel information, in particular to an intelligent management and control method and system of a tunnel telephone broadcasting system. Background Telephone broadcasting systems in tunnels, also known as tunnel emergency broadcasting systems or tunnel emergency broadcasting systems, are important facilities for tunnel authorities to issue emergency speech information to vehicles and personnel in tunnels. The system consists of a core control system, a transmission network, hardware equipment (a transmitter, a receiver, a loudspeaker, a microphone and the like) and a software platform. The system is mainly used for rapidly starting when traffic accidents, fires and other emergency events occur in the tunnel, issuing escape instructions to vehicles and personnel in the tunnel, guiding safe evacuation and the like. At present, most of telephone systems, broadcasting systems, monitoring systems, sensing systems and the like in tunnels are emergency telephone systems and broadcasting systems which operate independently, and a unified management and control platform is lacking. Resulting in a loss of critical functionality of the telephone broadcast system, for example: (1) When a plurality of alarms are simultaneously input, the alarms cannot be intelligently ordered according to the emergency degree or the position, operators can only answer the alarms one by one, and high-priority events are easy to miss. (2) The site environment sound of any telephone terminal in the tunnel cannot be monitored remotely, and abnormality (such as quarry and abnormal equipment sound) cannot be perceived in advance. (3) The operator needs to select broadcast content, broadcast position and the like according to the telephone answering result, and the operator needs to switch different systems to operate, which is complex. Disclosure of Invention Therefore, the present invention is directed to an intelligent control method and system for a tunnel telephone broadcasting system, so as to solve the above technical problems. In order to achieve the above purpose, the present invention adopts the following technical scheme: The invention relates to an intelligent management and control method of a tunnel telephone broadcasting system, which comprises the following steps: When emergency calls from a plurality of tunnel extensions are received at the same time, acquiring video and audio data collected by monitoring equipment in a group where the tunnel extensions are located, wherein the collected video and audio data are added into an uploading queue and uploaded to a cloud end for backup; Identifying the video and audio data based on a pre-constructed event classification model to obtain an event tag of the position of the tunnel extension and the emergency degree of the event tag, wherein the event tag comprises fire disaster, traffic accident, personnel help, equipment fault and no fault; weighting the emergency degree of the event label of each tunnel extension based on the section weight of the section where the tunnel extension is positioned to obtain the comprehensive priority of a plurality of emergency calls; Sequencing a plurality of emergency calls based on the comprehensive priority, and carrying out voice recognition and event entity extraction on the emergency calls in the process of being connected to obtain an event entity combination of the current emergency call; And predicting the probability model based on the pre-constructed broadcast templates and the probability of the event entity combination for predicting a plurality of broadcast templates required by the manager, and sending one or more broadcast modules with highest probability to the manager. In one embodiment of the present application, the network structure of the event classification model includes: Video coding branch for extracting space-time characteristics of video data based on pre-trained 3DCNN module to obtain video characteristics ; An audio coding branch for extracting a spectrogram of audio data based on a pre-trained VGGish module to obtain audio features; A multi-modal fusion layer for characterizing the videoAnd the audio featuresFusion is carried out to obtain the joint characteristicsAnd based on the full connection layer, the joint characteristicsDimension reduction is carried out to obtain dimension reduction characteristics; A multi-task output head for based on the dimension-reducing featurePredicted event tags and based on the dimension-reduction featuresAnd outputting the degree of urgency. In an embodiment of the present application, the process for constructing the event classification model includes: s1, acquiring an audio and video data sample and a tag of the audio and video data, wherein the tag comprises a labeling event tag and a labeling emergency degree; s2, splitting t