CN-121998801-A - Unmanned education training method and system for traffic industry workers

CN121998801ACN 121998801 ACN121998801 ACN 121998801ACN-121998801-A

Abstract

The invention discloses an unmanned education training method and system for traffic industry workers, and relates to the technical field of education management, wherein the method comprises the steps of acquiring eye image data of workers; the method comprises the steps of carrying out identity verification on a worker according to eye image data, determining a target worker passing the identity verification, obtaining a knowledge point mastering probability vector of the worker through a knowledge tracking model according to learning records of the target worker, screening a course candidate set by taking maximized course knowledge coverage and minimized course difficulty deviation as targets, recommending courses for the target worker through a collaborative filtering recommendation model according to the knowledge point mastering probability vector and the course candidate set, monitoring the learning state of the target worker, and carrying out integral excitation on the target worker according to the learning state. The invention can realize personalized course recommendation according to the teaching of the person, and effectively improve the learning enthusiasm and the training overall effect of the worker.

Inventors

HAO JIATIAN
HE YONGTAO
ZHAI GUANGZHI
TANG XINMIN
WANG YUEFEI
WU ZHONGGUANG
WANG XIANGUANG
WANG WEI
MA WENNING
WANG JI
WEI JIE
MENG DANQI
MA HAIFENG

Assignees

交通运输部科学研究院

Dates

Publication Date: 20260508
Application Date: 20251201

Claims (10)

1. The unmanned education training method for the traffic industry workers is characterized by comprising the following steps of: s1, acquiring eye image data of a worker; S2, carrying out identity verification on the worker according to the eye image data, and determining a target worker passing the identity verification; s3, obtaining a knowledge point mastering probability vector of the target worker through a knowledge tracking model according to the learning record of the target worker, wherein the knowledge tracking model specifically comprises a hypergraph module, a self-attention module, a time dynamic module, a gating fusion module and a prediction module; S4, screening a course candidate set by taking maximization of course knowledge coverage and minimization of course difficulty deviation as targets; S5, grasping probability vectors and the course candidate sets according to the knowledge points, and recommending courses for the target workers through collaborative filtering recommendation models; S6, monitoring the learning state of the target worker; s7, carrying out integral excitation on the target worker according to the learning state.
2. The unmanned educational training method for traffic industry workers according to claim 1, wherein S2 specifically comprises: S201, carrying out identity verification on the worker according to the eye image data; s202, determining the workers passing the identity verification as the target workers, and determining the workers not passing the identity verification as non-target workers.
3. The unmanned educational training method for traffic industry workers according to claim 2, wherein S201 specifically comprises: S2011, according to the eye image data, performing boundary positioning on pupils and irises through circular Hough transformation to obtain pupil boundaries and iris boundaries; S2012, generating an annular ROI area according to the pupil boundary and the iris boundary; S2013, according to the parameters of the pupil boundary and the iris boundary, performing boundary positioning on the eyelid through parabolic Hough transformation to obtain the eyelid boundary; s2014, generating an eyelid mask area according to the eyelid boundary; S2015, combining the annular ROI area and the eyelid mask area to obtain an iris annular area; s2016, mapping the iris annular region from Cartesian coordinates to polar coordinates through a rubber sheet model; S2017, performing feature coding on the iris annular region after mapping to obtain a binary feature template; S2018, acquiring a reference binary feature template of a registered worker in a database; S2019, judging whether the Hamming distance between the binary feature template and the reference binary feature template is smaller than a preset distance, if so, determining that the worker passes the identity verification, and if not, determining that the worker fails the identity verification.
4. The unmanned educational training method for traffic industry workers according to claim 3, wherein S2017 specifically comprises: S2017A, splitting the iris annular region after mapping into a plurality of one-dimensional signals; S2017B, performing complex convolution on each one-dimensional signal through a 1D Log-Gabor filter to obtain a complex response matrix; S2017C, calculating the phase angle of each complex response in the complex response matrix to obtain a phase angle matrix; S2017D, carrying out Gray code quantization on the phase angle matrix to obtain the binary feature template.
5. The unmanned educational training method for traffic industry workers according to claim 1, wherein the learning record is specifically a historical answer sequence of the target worker; The step S3 specifically comprises the following steps: S301, mapping the history answer sequence into a hypergraph in the hypergraph module to obtain a hypergraph incidence matrix; s302, carrying out convolution update on an interaction feature matrix corresponding to the historical answer sequence according to the hypergraph incidence matrix to obtain interaction embedding; s303, in the self-attention module, performing global dependency modeling on the interaction embedding to obtain a time sequence state; s304, in the time dynamic module, carrying out time sequence modeling on the interaction feature matrix to obtain a knowledge state; s305, in the gating fusion module, fusing the time sequence state and the knowledge state to obtain a fused knowledge state; s306, mapping the fused knowledge state into probability in the prediction module to obtain a knowledge point grasping probability vector of the target worker.
6. The unmanned educational training method for traffic industry workers according to claim 1, wherein S4 specifically comprises: S401, setting a first objective function with the aim of maximizing the course knowledge coverage, and setting a second objective function with the aim of minimizing the course difficulty deviation; s402, screening out the course candidate set by adopting a leech optimization algorithm according to the first objective function and the second objective function.
7. The unmanned educational training method for traffic industry workers according to claim 1, wherein the collaborative filtering recommendation model specifically comprises an embedding layer, an embedding propagation layer, a cascading layer, a matching layer and a recommendation layer.
8. The unmanned educational training method for traffic industry workers, according to claim 7, wherein S5 specifically comprises: S501, linearly mapping the knowledge point mastering probability vector and the course candidate set to a dense embedding space in the embedding layer to obtain an initial embedding matrix, wherein the initial embedding matrix comprises a target worker embedding sequence and a course embedding sequence, the target worker embedding sequence comprises a plurality of worker embedding vectors, and the course embedding sequence comprises a plurality of course embedding vectors; S502, constructing a self-connection information graph of the target worker, a self-connection information graph of the course and a transmission information graph between the target worker and the course in the embedding propagation layer according to the initial embedding matrix; S503, carrying out message synthesis on the transmission information graph and the self-connection information graph of the target worker so as to update the embedded sequence of the target worker; S504, synthesizing the message of the transfer information graph and the self-connection information graph of the course to update the course embedding sequence; s505, in the cascade layer, respectively cascading the updated target worker embedding sequence and the course embedding sequence to obtain final worker embedding and final course embedding; s506, in the matching layer, carrying out interactive matching on the final worker embedding and the final course embedding to obtain the matching degree between the target worker and the course; S507, in the recommendation layer, sorting the matching degree in a descending order, and recommending the courses for the target workers according to the sorting order.
9. The unmanned educational training method for traffic industry workers according to claim 1, wherein S7 specifically comprises: S701, calculating a learning integral by adopting a weighting algorithm according to the learning state, wherein the learning state specifically comprises learning duration, concentration degree and assessment score; s702, carrying out point rewards on the target worker according to the learning points.
10. An unmanned educational training system for traffic industry workers, comprising: A processor; A memory having stored thereon computer readable instructions which, when executed by the processor, implement the unmanned educational training method for traffic industry workers of any of claims 1 to 9.

Description

Unmanned education training method and system for traffic industry workers Technical Field The invention relates to the technical field of education management, in particular to an unmanned education training method and system for traffic industry workers. Background The unmanned education training method for the transportation industry workers is a training mode which is developed by means of artificial intelligence, an online learning platform, virtual reality and other technical means for workers engaged in first-line posts such as transportation, road management and vehicle maintenance, and the like, and the method can break through the limitation of traditional training on time, space and teaching materials, so that the workers can flexibly and efficiently acquire professional skills and industry knowledge, and the rapid requirement of digital and intelligent development of the transportation industry is met. Along with the rapid development of the traffic industry to the digital and intelligent directions, the first-line traffic industry workers need to continuously master new skills to adapt to post changes, and the unmanned education and training method can provide individuation and vocational training accessible anytime and anywhere for the workers under the condition that no whole-course on-site instruction of teachers is needed by utilizing technologies such as artificial intelligence, an on-line learning platform, virtual reality and the like. However, the identity verification of the existing unmanned education training method depends on account passwords or face recognition, is easy to be imposter and influenced by factors such as illumination and shielding, has insufficient safety, has a single course recommendation mechanism, fails to consider knowledge coverage and course difficulty matching, is easy to generate recommendation unbalance, and lacks an effective excitation mechanism based on multidimensional behaviors such as learning duration, concentration and achievements, so that the learning power of workers is insufficient, and the overall training effect is poor. Disclosure of Invention The invention provides an unmanned education training method and system for traffic industry workers, which aims to solve the technical problems that the identity verification of the existing unmanned education training method depends on account passwords or face recognition, is easy to be imposter and influenced by factors such as illumination and shielding, has insufficient safety, has a single course recommendation mechanism, cannot consider knowledge coverage and course difficulty matching, is easy to cause recommendation unbalance, and lacks an effective excitation mechanism based on multidimensional behaviors such as learning duration, concentration and achievements, so that the learning power of workers is insufficient. The technical scheme provided by the embodiment of the invention is as follows: First aspect: the embodiment of the invention provides an unmanned education training method for traffic industry workers, which comprises the following steps: s1, acquiring eye image data of a worker; s2, carrying out identity verification on the target worker according to the eye image data, and determining the target worker passing the identity verification; s3, obtaining a knowledge point mastering probability vector of the target worker through a knowledge tracking model according to the learning record of the target worker, wherein the knowledge tracking model specifically comprises a hypergraph module, a self-attention module, a time dynamic module, a gating fusion module and a prediction module; S4, screening a course candidate set by taking maximization of course knowledge coverage and minimization of course difficulty deviation as targets; S5, grasping probability vectors and the course candidate sets according to the knowledge points, and recommending courses for the target workers through collaborative filtering recommendation models; S6, monitoring the learning state of the target worker; s7, carrying out integral excitation on the target worker according to the learning state. Second aspect: The embodiment of the invention provides an unmanned education and training system for traffic industry workers, which comprises the following components: A processor; a memory having stored thereon computer readable instructions which, when executed by the processor, implement the unmanned educational training method for traffic industry workers of the first aspect. Third aspect: An embodiment of the present invention provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the unmanned educational training method for traffic industry workers as described in the first aspect. The technical scheme provided by the embodiment of the invention has the beneficial effects that at least: According to the embodiment of the invention,