Multi-user interaction virtual reality teaching system

Abstract

The invention belongs to the field of virtual reality teaching, and particularly relates to a multi-user interaction virtual reality teaching system. Comprising a single-person operation module, an operation summarization module, an interaction analysis module and a teaching feedback module. The single-person operation module is used for acquiring single-person operation data of students in a virtual reality environment, and the single-person operation data comprises student information, single-person connection steps, single-person operation steps, a single-person operation range and equipment parameter changes; the operation summarization module comprises a plurality of standard ranges and a global operation range formed by the plurality of standard ranges; and the single-person operation module is used for receiving all the single-person operation data, fusing the data into global operation data, and performing spatial alignment on the single-person operation data according to the relationship between the global operation range and the single-person operation range. According to the scheme, through space-time sequence alignment, real-time parameter verification, accurate error tracing and dual-channel feedback, the problems of slow teaching progress and low teaching efficiency caused by the limitation of the attention range of students in collaborative teaching are solved.

Description

TECHNICAL FIELD

[0001] The scheme belongs to the field of virtual reality teaching, and particularly relates to a multi-person interactive virtual reality teaching system. BACKGROUND

[0002] In the field of practical teaching such as mechanical assembly and equipment maintenance, multi-person cooperation to complete complex operations is a core and key teaching mode. This mode not only simulates the actual needs of team cooperation in real work scenes, but also cultivates students' cooperation consciousness and comprehensive practical ability, so it occupies an irreplaceable position in related teaching activities. With the development of virtual reality technology, multi-person interactive virtual reality teaching has gradually become an important teaching means in this field. However, the existing multi-person interactive virtual reality teaching system has many drawbacks, which seriously restricts the improvement of teaching efficiency.

[0003] In the existing virtual reality teaching, due to the natural limitation of the students' attention range, the students can only focus on their own operation and the local operation link adjacent to themselves in the cooperation process, and cannot clearly know the influence of their behavior on other collaborators outside the attention range, so it is difficult to accurately understand the role of their operation in the whole task. This situation makes the team cooperation lack overall coherence, and once an error or accident occurs in the operation process, the operation process often needs to be restarted from the beginning, and a large amount of repeated operation will significantly prolong the teaching time and greatly reduce the teaching efficiency.

[0004] More notably, when multiple people cooperate, even if the operation process of a single student is completely correct, the result may not meet the expectation due to problems in the cooperation and coordination of multiple people. At the same time, the operation of a single person is greatly affected by personal subjective factors, and the attention and attention range of a single person in operation are limited in itself, which makes the team need to correct through multiple experimental operations in the cooperation process to gradually find the correct coordination method. This process greatly increases the time cost of teaching and slows down the teaching progress. SUMMARY

[0005] The purpose of the scheme is to provide a multi-person interactive virtual reality teaching system to solve the problem of slow teaching progress when students interactively cooperate in teaching through virtual reality technology.

[0006] In order to achieve the above purpose, the scheme provides a multi-person interactive virtual reality teaching system, comprising: a single-person operation module for acquiring single-person operation data of students in a virtual reality environment, the single-person operation data comprising student information, single-person connection steps, single-person operation steps, single-person operation range, and device parameter changes; The operation summary module includes a plurality of standard ranges and a global operation range composed of the plurality of standard ranges; is configured to receive all single-person operation data and fuse the single-person operation data into global operation data, align the single-person operation data in space according to a relationship between the global operation range and the single-person operation range, align the single-person operation data in time sequence according to a comparison relationship between a single-person connection step and a single-person operation step, and generate an initial operation set; The interaction analysis module includes a global initial parameter and a standard parameter interval of each standard range; is configured to take each device parameter of a current global operation range as an initial parameter, take single-person operation data from the initial operation set in sequence, combine the device parameter of the single-person operation data with the initial parameter, calculate a current device parameter in a standard range after a single-person operation step as a first parameter, and compare the first parameter with a standard parameter interval corresponding to the standard range; then, the interaction analysis module is configured to take the first parameter as the initial parameter, repeatedly take single-person operation data from the initial operation set and combine the single-person operation data with the initial parameter to calculate the first parameter, until the first parameter is not in the standard parameter interval or the single-person operation data in the initial operation set has been all taken out and used to calculate the first parameter; finally, the interaction analysis module is configured to generate operation results according to a comparison result of the first parameter and the standard parameter interval. The teaching feedback module is configured to generate completion information according to operation results or generate interruption information by taking single-person operation data and corresponding initial parameters and first parameters taken out by the interaction analysis module last time.

[0007] The principle and technical effect of the scheme are as follows: first, the single-person operation module acquires single-person data including operation ranges and steps, the operation summary module aligns the single-person operation data in space according to standard ranges and global ranges, aligns the single-person operation data in time sequence according to connection steps and operation steps, and generates an initial operation set. In this way, fragmented actions of students are integrated into a global action chain, the students clearly understand the position and role of their own operation in the global, the problem of limited attention range is solved, and the coordination of multi-person operation is ensured. For example, in virtual engine assembly, the operation summary module generates a set of "cylinder cover positioning → bolt pre-tightening" after alignment, so that student A and student B know the operation correlation. If the prior art uses the scheme to integrate multiple steps into a global action chain, an external coordinate mapping and timestamp unification protocol needs to be additionally introduced, and real-time completion is not possible. Because the prior art lacks a standard range as an extensible intermediate layer, the operation ranges and steps of different teaching scenes are quite different, the alignment algorithm must be repeatedly rewritten for different models and devices, the universality is extremely poor, and the effect of corresponding the operation of students to the global action cannot be achieved.

[0008] Secondly, the interaction analysis module takes the global initial parameter as the starting point, and sequentially takes single-person operation data, calculates the first parameter combined with the device parameter change, and compares it with the standard interval. If it is out of range, it will be terminated and the operation result will be generated. In this way, the student's single-person operation data can be timely warned in the virtual reality environment, and the error can be found in the operation teaching process and the subsequent operation steps are terminated, saving computing power while reducing the time consumption in the error operation steps, thereby further saving the teaching time and improving the teaching efficiency. The prior art needs to be calculated after batch processing, lacks standard parameter interval, cannot timely stop and prompt, and is prone to repeated operation and low efficiency.

[0009] Furthermore, when the interaction analysis module terminates the deduction, the teaching feedback module obtains the last single-person operation data and the corresponding parameter, generates the interruption information, realizes accurate error tracing, quickly locates the problem and parameter change for the teacher and students, such as in the hydraulic debugging, the teaching feedback module sends "student D's 7th step valve pressure is out of range" to the student corresponding to the student information in the single-person operation data according to the operation result. The scheme also reduces the error correction time, and does not need to backtrack the entire operation process, but only needs to correct the error single-person operation data, and other students do not need to re-operate, which greatly improves the teaching efficiency. Even if the prior art records all action logs, it needs to search for key frames in the massive data manually, and cannot automatically bind and output the unique granular data as in the scheme at the moment of out of range, which leads to time-consuming and laborious error correction process and is difficult to quickly solve the problem.

[0010] In addition, the teaching feedback module displays the completion information or interruption information containing specific data according to the operation result. Through this double-channel feedback mechanism, when the students complete the correct and complete cooperation, the students are given positive feedback, which enhances the students' learning confidence and sense of achievement, and at the same time, the teacher knows the completion of the teaching task. At the same time, when the operation is abnormal, the accurate interruption information can provide a clear improvement direction for the students, so that the students know where they have made mistakes and how to correct them, and improve the learning effect.

[0011] In summary, the scheme solves the problems of slow teaching progress and low teaching efficiency caused by the limited attention range of students in cooperative teaching through space-time alignment, real-time parameter verification, accurate error tracing and double-channel feedback.

[0012] Further, the single-person operation module further comprises a single-person prediction model; training data of the single-person prediction model comprises historical single-person operation data of the student, standard operation steps corresponding to the single-person operation range of the student, and standard equipment parameter changes; the single-person prediction model learns the association between historical operation and each standard data through training, and predicts the operation step of the student at the next moment as a predicted step based on the current single-person operation data and the initial parameter, and predicts the equipment parameter change of the student at the next moment as a predicted parameter change, uses the prediction result as an analysis basis to analyze the operation result using the interactive analysis module, combines the operation result, the predicted step, and the predicted parameter change to generate a prediction result prompt, and sends the prediction result prompt to the student terminal corresponding to the student information.

[0013] In the mechanical assembly virtual reality teaching scene, when the student performs part installation operation, the single-person prediction model predicts the installation angle deviation that may occur at the next moment and the corresponding equipment parameter change based on the historical installation data, the standard steps corresponding to the installation range, and the parameters, and the interactive analysis module analyzes that the deviation will cause subsequent students to be unable to smoothly perform part connection, and the teaching feedback module sends a prompt containing the predicted installation angle problem and the influence on the overall assembly to the student. In this way, not only can the student correct the deviation in time before operation and reduce the error operation, but also the overall cooperation progress can be avoided due to the single-person operation problem, the teaching process is smooth, the student can intuitively understand the association between his own operation and the global task, and the importance of cooperation is deepened.

[0014] Further, the single-person prediction model is further used to evaluate the proficiency of the student by counting the number of training times of the same operation range and operation step in the training data of the single-person prediction model, determine the proficiency of the student in the corresponding operation range and operation step in combination with the proportion of the historical single-person operation data meeting the standard parameters and the preset minimum operation times, and store the proficiency in association with the student information; when the interactive analysis module calculates the first parameter, the student information in the single-person operation data participating in the calculation of the first parameter is obtained, and the standard parameter interval is dynamically adjusted according to the proficiency corresponding to the student information.

[0015] By automatically converting the historical practice times and the standard meeting degree of each student into individual proficiency, and then tightening or widening the operation allowed range in real time, the novice receives a prompt when there is a slight deviation in screwing the bolt, so as to quickly practice the skill; the veteran does not have to repeatedly debug the small values due to the widened range, and the whole assembly line is therefore smooth and unobstructed. The novice constantly obtains instant success in strict limits, and the learning confidence is multiplied; the veteran maintains high efficiency in a relaxed limit, and avoids boredom. At the same time, the virtual parts are less worn due to less boundary crossing, the server reset times are greatly reduced, the teacher is also relieved from repeatedly adjusting the interval, the preparation time is reduced, the teaching efficiency of the present scheme is further improved, and at the same time, the personalized interactive teaching method also improves the teaching quality of the present scheme.

[0016] Further, the single-person operation module regulates the collection time of single-person operation data based on the proficiency of the student; when the proficiency of the student is less than a preset minimum proficiency, the single-person operation module acquires a standard operation step corresponding to a current operation range of the student as a central step, and acquires a previous step of the central step as a pre-step; when the single-person operation module does not completely acquire the pre-step, the collection time of the central step is prolonged, and an assistance prompt is generated according to the relationship between the pre-step and the central step, and the assistance prompt is sent to the student information corresponding to the central step.

[0017] The scheme directly converts the proficiency of the student into a rhythm control mode of data collection, and brings three effects of "novice slowdown-prompt synchronization-collaborative sprint". When the proficiency is lower than the minimum threshold, the pre-step is automatically set as a limiting mode, and the central step is not released until the pre-step is completely collected, so as to avoid misoperation of the novice due to skipping the key preparation; meanwhile, the collection is delayed, which is equivalent to slowing down the task rhythm of the novice on the virtual time axis, so that the teacher or the peer has a window for intervention. The relationship between the pre-step and the central step generates an assistance prompt in real time, which is equivalent to pushing what to do next in front of the novice, reducing the groping time and relieving the teacher's inspection pressure. For the proficient student, the scheme does not prolong the collection, and the process maintains a high speed. The overall teaching rhythm is changed from "the slowest person" to "speed adjustment on demand", so that the progress of the class is not dragged down by the individual novice, and the novice can also consolidate each step in a safe rhythm, realizing the win-win of efficiency and quality. In addition, this way also controls the collection by differentiation, which avoids the collaboration confusion caused by the fast operation of the novice, and guarantees the efficient progress of the experienced person, so that the collection of the single-person operation module is more in line with the actual teaching needs, and the collaboration fluency and teaching effectiveness of the multi-person interactive virtual reality teaching system are further improved.

[0018] Further, the interaction analysis module has built-in safety collaboration specifications, the safety collaboration specifications include standard parameter intervals corresponding to each operation type and dangerous intervals, and an operation range affected by the dangerous interval is stored in association with the dangerous interval; when the interaction analysis module calculates the first parameter, if it is detected that the first parameter is in the dangerous interval, all single-person operation steps before the current single-person operation step in the initial operation set are queried as review steps, the student information of the review steps and the latest single-person operation data in the current operation step time sequence are acquired as dangerous data, and if the single-person operation range in the dangerous data has an overlapping area with the dangerous range, the teaching feedback module sends a dangerous prompt according to the student information in the dangerous data.

[0019] Through the accurate identification of the dangerous interval and the associated storage of the dangerous range by the interaction analysis module, in the scene of mechanical gearbox assembly, the pre-associated steps and corresponding students that cause the current dangerous operation can be quickly located, such as, in the process of teaching collaborative operation, the overlap of the dangerous range between student B and student A is found in time, and the safety accidents caused by non-standard operation are avoided; at the same time, the teaching feedback module sends a prompt, so that the relevant students can clearly understand the cause of the danger and the checking direction, which not only reduces the time cost of the teacher's one-by-one checking, but also improves the students' awareness of operation association, and can also ensure the continuity of multi-person collaborative teaching through timely intervention, and balance the operation safety and teaching efficiency.

[0020] Further, the interaction analysis module also has a built-in delay proficiency coupling unit, which is used to collect the estimated value of the current network communication delay in real time, and takes the estimated value and the dangerous interval as inputs; the delay proficiency coupling unit maps the estimated value to the tolerable delay upper bound according to the proficiency level corresponding to the student information, and dynamically extends the data confirmation window of the student corresponding to the student information when the first parameter is not in the dangerous interval and is less than the preset dangerous threshold from the dangerous interval, if the proficiency corresponding to the student information is lower than the preset minimum proficiency and the estimated value exceeds the tolerable delay upper bound, the interaction analysis module immediately adds a virtual compensation step to the initial operation set, the virtual compensation step takes the midpoint of the dangerous interval as the trigger condition, delays the collection time of the current single operation step by Δt, Δt is determined according to the correlation between proficiency and estimated value, and at the same time, the priority of the danger prompt is improved by one level; if the student proficiency is higher than the preset threshold, the single operation module keeps the same collection frequency.

[0021] The delay proficiency coupling unit analyzes the network communication delay estimate value and the dangerous interval, dynamically adjusts the data confirmation window combined with the student proficiency, extends the collection time for low-proficiency students and adds a virtual compensation step when the first parameter is close to the dangerous interval, which not only avoids invalid operation data of novice due to network delay, but also fills the delay gap through virtual steps to prevent the disruption of collaborative rhythm. When low-proficiency students face delay over-limit, the priority of the danger prompt is improved and the collection time is delayed, which not only allows novices to get key warning information first, but also reserves operation correction space for them, reducing dangerous operations caused by panic.

[0022] For high-proficiency students, the original collection rhythm is maintained to avoid reducing operation efficiency due to meaningless delay compensation, realizing differentiated collaborative management of "buffer for novices and rhythm for experienced hands", which ensures operation safety and smoothness without additional system load. The virtual compensation step takes the midpoint of the dangerous interval as the trigger condition, so that the delay processing is accurately matched with the risk level, avoiding rhythm disorder caused by excessive compensation, and further optimizing the stability of multi-person collaborative operation.

[0023] Further, the interaction analysis module further comprises a residual risk score unit; the residual risk score unit is configured to accumulate the residual risk score according to the length of time that the first parameter stays in the dangerous interval each time the first parameter enters the dangerous interval, and when the residual risk score exceeds a preset score threshold, the standard parameter interval corresponding to the current student information is tightened by a preset percentage, and the tightened interval is broadcast to all student terminals.

[0024] The residual risk score unit calculates the residual risk score according to the length of time that the first parameter stays in the dangerous interval, and when the score exceeds the preset threshold, the standard parameter interval of the corresponding student is tightened and broadcast, which allows the present scheme to accurately quantify the risk accumulation degree of the student's operation, avoiding the risk of being ignored by a single short border crossing; for students who frequently approach or enter the dangerous interval, the tightened standard parameter interval is equivalent to increasing the operation precision requirement, which can force them to operate normally and reduce habitual misoperation; at the same time, the tightened interval is broadcast to all student terminals, which allows other students to know the current risk control standard, forming a mutual supervision and mutual improvement teaching atmosphere, which not only enhances the students' risk awareness, but also reduces the pressure of teachers monitoring the operation of each student in real time, improving the overall teaching safety while promoting the overall improvement of teaching quality.

[0025] Further, the safety strategy module obtains the current single operation data of each student collected by the interaction analysis module, combines the standard parameter interval, the dangerous interval and the corresponding dangerous range, sets the dangerous range as a negative reward area through a reinforcement learning algorithm, and generates an initial probability distribution of the first parameter of each student within the standard parameter interval; subsequently, the safety strategy module cooperatively adapts the initial probability distribution to the device parameter changes and predicted steps of other students to obtain a predicted step probability distribution of the first parameter within the standard parameter interval; when it is detected that the first parameter in the predicted step probability distribution of any student falls into the dangerous interval or is less than a preset dangerous interval distance from the dangerous interval, the safety strategy module immediately sends an adjustment instruction containing a preset safety action parameter to the student terminal according to the student information, and synchronizes the instruction to other student terminals.

[0026] The safety policy module is based on real-time operation data of the interactive analysis module, combines standard parameter intervals and dangerous intervals to construct a reinforcement learning model, sets the dangerous range as a negative reward area to generate an initial probability distribution, and then obtains a predicted step probability distribution through collaborative adaptation, so that the probability distribution not only fits the individual operation characteristics, but also adapts to the group collaboration needs, realizing accurate conversion from individual data to group collaborative strategy. Secondly, the trigger condition is set as "the first parameter falls into or approaches the dangerous interval", which forms a logical closed loop with the safety control target. When such risks are detected, adjustment instructions containing preset safety parameters are pushed according to student information, which can not only correct individual dangerous operations, but also through synchronous instructions let other student terminals predict risks in advance, reserve reaction time for group collaborative avoidance, and avoid chain risks caused by information lag.

[0027] At the same time, the preset safety action parameters and the probability distribution model are linked, so that the adjustment instructions have both the scientific nature of algorithm optimization and the pertinence of individual adaptation, which not only solves the problem of lack of operation guidance in single warning in the prior art, but also improves the safety and fluency of multi-person collaboration through group synchronization mechanism, greatly reducing the probability of group operation conflict caused by individual operation errors.

[0028] Further, the interactive analysis module sends a preset calling instruction to the safety policy module immediately after detecting that the residual risk score exceeds the preset score threshold through the residual risk score unit; the safety policy module receives the dangerous interval in the current single-person operation data synchronized by the interactive analysis module after responding to the calling instruction, and maps the dangerous interval as a temporary negative reward area, which is incorporated into the original negative reward system as a temporary increment; when calculating the predicted step probability distribution of the next round of first parameters, the safety policy module preferentially pushes the probability distribution result containing the temporary negative reward area to the student terminal with the highest residual risk score; the temporary negative reward area is automatically removed after the predicted step probability distribution calculation of the next round of first parameters is completed, and does not affect the setting of the negative reward area corresponding to the original dangerous range.

[0029] When the residual risk score exceeds the limit, the interaction analysis module calls the safety policy module in real time, converts the dangerous interval into a temporary negative reward area and integrates it into the original reinforcement learning system, realizes the dynamic binding of risk accumulation and real-time safety policy, and makes the response of the present scheme to high-risk operations more targeted; the setting of the temporary negative reward area can strengthen the avoidance weight of dangerous operations in subsequent parameter calculation, reducing the probability of repeated boundary crossing of high-risk students from the algorithm level; the probability distribution result containing the temporary negative reward area is preferentially pushed to the student with the highest risk score, which can accurately alert the student and promote the student to actively adjust the operation, realizing personalized risk intervention; the temporary negative reward area is automatically removed in the next round of calculation and does not affect the original system, which not only avoids the interference of excessive control on the normal teaching rhythm, but also ensures the flexibility of the negative reward mechanism, improves the operation standardization of high-risk students, and also maintains the stability of the overall safety policy of the system, further reducing the chain risk in group operation. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 The figure is a structural schematic diagram of the function module of the multi-person interactive virtual reality teaching system in the embodiment of the present application. DETAILED DESCRIPTION

[0031] The concept and technical effects of the present application will be described below in conjunction with the embodiments to clearly and completely understand the purpose, features and effects of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all embodiments, and other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative labor are within the scope of the present application: As Figure 1 shown, the multi-person interactive virtual reality teaching system comprises: A single-person operation module for obtaining single-person operation data of students in a virtual reality environment, wherein the single-person operation data comprises student information, single-person connection steps, single-person operation steps, single-person operation range and device parameter changes; An operation summary module comprising a plurality of standard ranges and a global operation range composed of a plurality of standard ranges; for receiving all single-person operation data and fusing it into global operation data, aligning the single-person operation data in space according to the relationship between the global operation range and the single-person operation range, and then aligning the single-person operation data in time sequence according to the comparison relationship between the single-person connection steps and the single-person operation steps to generate an initial operation set; The interaction analysis module includes a global initial parameter and a standard parameter interval of each standard range; the current global operation range is used as the initial parameter, and the single-person operation data is taken out from the initial operation set in sequence, the device parameter change of the single-person operation data is combined with the initial parameter, the current device parameter in the standard range after the single-person operation step is calculated as the first parameter, and the first parameter is compared with the standard parameter interval corresponding to the standard range; the first parameter is taken as the initial parameter, and the single-person operation data is taken out from the initial operation set in sequence and combined with the initial parameter to calculate the first parameter, until the first parameter is not in the standard parameter interval, or the single-person operation data in the initial operation set has been taken out and used to calculate the first parameter; finally, operation results are generated according to the comparison result of the first parameter and the standard parameter interval. The teaching feedback module generates completion information according to the operation result, or generates interruption information according to the single-person operation data, the corresponding initial parameter and the first parameter taken out by the interaction analysis module last time.

[0032] The single-person operation module further includes a single-person prediction model; the training data of the single-person prediction model includes historical single-person operation data of students, standard operation steps corresponding to the single-person operation range of the students and standard device parameter changes; the single-person prediction model learns the association between the historical operation and each standard data through training, predicts the operation step of the students at the next moment as the prediction step based on the current single-person operation data and the initial parameter, predicts the device parameter change of the students at the next moment as the prediction parameter change, uses the prediction result as the analysis basis to analyze the operation results by using the interaction analysis module, combines the operation results, the prediction step and the prediction parameter change to generate a prediction result prompt, and sends the prediction result prompt to the student terminal corresponding to the student.

[0033] In an embodiment of the present scheme, in a mechanical assembly teaching scene, historical single-person operation data (operation steps, bearing installation force change, assembly time, etc. recorded by the single-person operation module) of students performing bearing and shaft sleeve matching operations in the past, and standard operation steps (the "shaft sleeve positioning → bearing trial assembly → uniform force application" steps preset in the bearing assembly standard range in the operation summary module) and standard device parameters (the force application range and coaxiality error threshold included in the bearing matching standard parameter interval in the interaction analysis module) corresponding to the operation are collected. Through machine learning algorithm, the above data is trained, and the model learns the association rules between problems such as excessive force application leading to coaxiality error and step skipping causing assembly jam in historical operation and standard requirements.

[0034] When the student is currently performing the bearing fitting operation, the model captures the current behavior data of the student's hand force application speed, assembly angle, etc. in real time, combines the rules formed by training to predict the operation steps (prediction step) that may occur next time due to sudden increase of force application leading to bearing deflection and the corresponding equipment parameter changes (prediction parameter change) that may cause the coaxiality error to exceed 0.1mm.

[0035] The prediction step and the prediction parameter change are input into the interactive analysis module as analysis basis, the interactive analysis module calls the timing relationship of the initial operation set after the bearing assembly that needs to be installed by other students, analyzes that the deflection will lead to the operation result that the sealing ring cannot be fitted; the teaching feedback module generates the prediction result prompt "the current force application speed is too fast, it is predicted that the bearing will be deflected next time, which will affect the subsequent sealing ring installation", and sends it to the virtual reality terminal of the student (bound with the student information).

[0036] By early warning individual operation risk, reducing single-time mistakes; by predicting global collaboration impact, avoiding multiple operation chain failures; by instant prompting to strengthen standard memory, improving students' standard operation ability, the three effects form a synergistic mechanism, which is significantly better than the existing technology of post-correction mode.

[0037] Specifically, the single-person prediction model is also used to evaluate the student's proficiency by counting the number of training times of the same operation range and operation step in the single-person prediction model training data, and combining the proportion of historical single-person operation data meeting the standard parameters and the preset minimum operation number to determine the student's proficiency in the corresponding operation range and operation step, and store it in association with the student information. When the interactive analysis module calculates the first parameter, the student information in the single-person operation data participating in the calculation of the first parameter is obtained, and the standard parameter interval is dynamically adjusted according to the corresponding proficiency of the student information.

[0038] In an embodiment of the present scheme, the student proficiency evaluation sub-process of the single-person prediction model is as follows: S10: input the following 3 kinds of data into the single-person prediction model: a) historical single-person operation data, including student information, single-person operation range, single-person operation step, and equipment parameter change; b) corresponding standard operation steps and standard equipment parameter changes; c) preset minimum operation number Nmin (default 5, can be modified by teacher background).

[0039] S20: count the following two kinds of data: 1) the number of training times C of the same <single-person operation range, single-person operation step> combination; 2) the proportion R (0-1) of historical equipment parameter changes falling within the corresponding standard parameter interval.

[0040] S30: Calculate proficiency: Proficiency S = min(1, (C / Nmin) R^α), where α ∈ [0.8, 1.2] is defaulted to 1.0 by the system; The result interval of proficiency S is 0-1, which is mapped to four levels of novice, intermediate, proficient, and expert, and stored in the "Student-Proficiency Table" (associated with the primary key of "Student Information").

[0041] The dynamic interval adjustment sub-process of the interaction analysis module is as follows: 1) When the interaction analysis module is ready to deduce a single-person operation data, immediately query the "Student Information" carried by the data, read its proficiency S corresponding to <single-person operation range, single-person operation step>.

[0042] 2) The standard parameter interval [Pmin, Pmax] is real-time scalable according to the following rules: New interval width = (Pmax - Pmin) (1 - β S); Where β ∈ [0.3, 0.5], default 0.4.

[0043] Novice S ≈ 0 → interval tightens 30-50%; veteran S ≈ 1 → interval almost unchanged.

[0044] 3) If the tightened interval leads to out-of-bound, the interaction analysis module immediately triggers "Interrupt Information", and prompts through the teaching feedback module pop-up window "Student ID-Zhang San, 3rd step torque exceeds upper limit 8%, please recalibrate."

[0045] Specifically, the single-person operation module adjusts the collection time of single-person operation data based on student proficiency; when the student proficiency is less than the preset minimum proficiency, the single-person operation module obtains the standard operation step corresponding to the current operation range of the student as the center step, and obtains the previous step of the center step as the pre-step; when the single-person operation module does not completely obtain the pre-step, the collection time of the center step is extended, and the assistance prompt is generated according to the relationship between the pre-step and the center step, and the assistance prompt is sent to the student information corresponding to the center step.

[0046] More specifically, when the proficiency of the student is lower than the preset threshold, the single-person operation module immediately locks the current center step and retrieves the pre-step that the student must complete; if the pre-step has not been completely collected, on the one hand, the data collection time of the center step is automatically delayed to give the student more time to make up, and on the other hand, visual / voice assistance prompts are generated in real time to accurately point to the missing action; only when all the pre-steps are confirmed to be completed, the center step is reopened for collection; for the student who has reached the proficiency standard, the single-person operation module keeps the default rhythm and does not make any delay and prompt, so that the novice can obtain forced practice and immediate guidance on the key actions, avoid step skipping mistakes, and at the same time ensure that the proficient students can advance with the highest efficiency, so as to realize the speed adjustment of the classroom rhythm according to the needs, without being slowed down by individual novices.

[0047] Specifically, the interactive analysis module has built-in safety cooperation specifications, the safety cooperation specifications include standard parameter intervals corresponding to each operation type and dangerous intervals, and the operation range affected by the dangerous interval is stored in association with the dangerous interval as a dangerous range; when the first parameter is calculated, if the first parameter is detected to be in the dangerous interval, all single-person operation steps before the current single-person operation step in the initial operation set are immediately queried as review steps, the student information of the review steps and the latest single-person operation data in the time sequence of the current operation step are obtained as dangerous data, and if the single-person operation range in the dangerous data has an overlapping area with the dangerous range, the teaching feedback module sends a danger prompt according to the student information in the dangerous data.

[0048] In an embodiment of the present scheme, the operation data of the current operation step of student A, "bearing cover fastening", is calculated by the interactive analysis module, and the first parameter shows that the bearing cover bolt torque is in the dangerous interval (for example, exceeds the upper limit of the standard parameter interval). At this time, the interactive analysis module starts the danger investigation, queries all single-person operation steps before "bearing cover fastening" from the initial operation set as review steps, and finds that the latest one in the time sequence is the "bearing seat positioning" step completed by student B. At this time, student B may still stay in the single-person operation range. After obtaining the single-person operation data (dangerous data) of student B, it is detected that the operation range is "left bearing seat mounting area of transmission case shell", and the dangerous range associated and stored with the current dangerous interval is "left bearing seat and adjacent bearing cover mounting area of transmission case shell", and there is an overlapping range. The teaching feedback module immediately sends a danger prompt "the operation range of the bearing seat positioning completed by you overlaps with the dangerous range of the current bearing cover fastening operation. Please leave as soon as possible." to the terminal of student B.

[0049] Further, the interaction analysis module also internally embeds a delay proficiency coupling unit, which is used to collect the estimated value of the current network communication delay in real time, and takes the estimated value and the dangerous interval as inputs together; the delay proficiency coupling unit maps the estimated value to the upper bound of the tolerable delay according to the proficiency level corresponding to the student information, and dynamically extends the data confirmation window of the student corresponding to the student information when the first parameter is not in the dangerous interval and is less than the preset dangerous threshold from the dangerous interval; if the proficiency corresponding to the student information is lower than the preset minimum proficiency and the estimated value exceeds the upper bound of the tolerable delay, the interaction analysis module immediately appends a virtual compensation step to the initial operation set, and the virtual compensation step takes the midpoint of the dangerous interval as the trigger condition to delay the collection time of the current single-person operation step by Δt, which is determined according to the correlation between the proficiency and the estimated value, and at the same time, the priority of the danger prompt is promoted by one level; if the student proficiency is higher than the preset threshold, the single-person operation module keeps the collection frequency unchanged.

[0050] In an embodiment of the present scheme, when the network communication delay fluctuates, the delay proficiency coupling unit first judges the risk distance of the first parameter in combination with the dangerous interval, and regulates and controls the operation close to danger but not exceeding the boundary according to the proficiency level. For low-proficiency students, the virtual compensation step added when the delay exceeds the limit will simultaneously lock the current operation interface, and unlock it after the compensation time ends, avoiding mistakes by beginners in the delay; after the priority of the danger prompt is promoted, it is displayed in the form of a pop-up window on top to ensure that beginners receive it first. For high-proficiency students, the original rhythm is maintained, and the delay is monitored in real time in the background, and only when the delay may trigger a chain risk will it be fine-tuned, without disturbing the operation rhythm.

[0051] For example, in the teaching of electronic circuit welding, when a novice student welds a resistor, the first parameter is close to the dangerous interval, and at this time the network delay suddenly increases and exceeds its upper limit of tolerance. The interaction analysis module immediately appends a virtual compensation step, suspends the operation interface and delays the single-person operation module collection time, and at the same time, the prompt "current network delay is high, please check the welding point temperature" is displayed on top. A skilled student next to him is welding a capacitor, although there is the same delay, but the collection rhythm is maintained, and only the delay data is recorded in the background.

[0052] This design provides double protection for beginners on the edge of delay and risk, and does not disturb the operation of experienced hands, balancing the safety and efficiency of cooperation.

[0053] The interaction analysis module also includes a residual risk integration unit; the residual risk integration unit is used to accumulate the residual risk integration according to the length of stay of the first parameter in the dangerous interval each time the first parameter is detected to enter the dangerous interval, and when the residual risk integration exceeds the preset integration threshold, the standard parameter interval corresponding to the current student information is tightened by a preset percentage as a whole, and the tightened interval is broadcast to all student terminals.

[0054] In one specific embodiment of the present scheme, in the virtual hydraulic device debugging teaching, the student C operates multiple times to make the hydraulic pipeline pressure temporarily enter the dangerous interval, and the residual risk score is continuously accumulated and exceeds the preset threshold. At this time, the residual risk score unit immediately tightens the standard parameter interval of all students as a whole, and pushes a prompt to all student terminals: "The system enters the conservative mode, the current hydraulic pipeline pressure standard interval is tightened, please pay attention to the operation precision." Due to the tightening of the parameter interval, the sensitivity of the operation interface of student C to pressure control is improved, forcing him to adjust the valve more carefully; other students also realize the risk of operation due to the interval adjustment, and standardize the operation, and the teacher can take this opportunity to focus on explaining the key points of pressure control, reducing the subsequent collective operation errors.

[0055] The safety strategy module also includes a safety strategy module that obtains the current single operation data of each student collected by the interaction analysis module, combines the standard parameter interval, the dangerous interval and the corresponding dangerous range, sets the dangerous range as a negative reward area through a reinforcement learning algorithm, and generates an initial probability distribution of the first parameter in the standard parameter interval of each student; subsequently, the safety strategy module cooperatively adapts the initial probability distribution to the equipment parameter changes and predicted steps of other students to obtain a predicted step probability distribution of the first parameter in the standard parameter interval; when it is detected that the first parameter in the predicted step probability distribution of any student falls into the dangerous interval or is less than the preset dangerous interval from the dangerous interval, the safety strategy module immediately sends an adjustment instruction containing a preset safety action parameter to the student terminal according to the student information, and synchronizes the instruction to other student terminals.

[0056] Specifically, the single operation data collected by the interaction analysis module includes operation time, operation force parameter (such as virtual device pressing intensity) and historical operation error rate (such as the deviation percentage of the last 3 operations from the standard parameter), which provides more comprehensive individual operation feature input for reinforcement learning.

[0057] During the cooperative adaptation process, the safety strategy module first calculates the time axis coincidence degree of the initial probability distribution and the predicted steps of other students, and if the coincidence degree exceeds 70%, parameter fine-tuning is started to reduce the action conflict probability; at the same time, the device parameter change rate (such as virtual instrument adjustment speed) is referred to, and higher weight is given to high-speed changing parameters.

[0058] In addition to containing preset safety action parameters, the adjustment instruction also carries a risk level identifier (divided into "warning" and "emergency" levels according to the distance of the first parameter from the dangerous interval), the "warning" level instruction only prompts the adjustment direction, and the "emergency" level instruction directly locks the dangerous operation component and displays the standard operation path.

[0059] After the student terminal executes the adjustment instruction, the operation correction data is returned to the security policy module, the module updates the reward weight of the reinforcement learning model by comparing the probability distribution of the prediction step before and after the correction, and realizes dynamic optimization.

[0060] Specifically, the interaction analysis module immediately sends a preset calling instruction to the security policy module after detecting that the residual risk score exceeds the preset score threshold through the residual risk score unit; the security policy module receives the dangerous interval in the current single-person operation data synchronized by the interaction analysis module in response to the calling instruction, maps the dangerous interval to a temporary negative reward area, and incorporates it as a temporary increment into the original negative reward system; when calculating the probability distribution of the prediction step of the next round of first parameters, the security policy module preferentially pushes the probability distribution result containing the temporary negative reward area to the student terminal with the highest residual risk score; the temporary negative reward area is automatically removed after the calculation of the probability distribution of the prediction step of the next round of first parameters is completed, and does not affect the setting of the negative reward area corresponding to the original dangerous range.

[0061] The above is only an embodiment of the present application, and common knowledge such as specific structures and characteristics in the scheme is not described in detail. It should be noted that for those skilled in the art, without departing from the structure of the present application, a number of modifications and improvements can be made, which should be considered as the protection scope of the present application, and these will not affect the effect and practicality of the present application. The protection scope of the present application should be subject to the content of its claims, and the specific implementation mode and the like recorded in the specification can be used to explain the content of the claims.

Claims

1. A multi-user interactive virtual reality teaching system, characterized in that, The application relates to a teaching system and method for teaching students to operate equipment in a virtual reality environment. The system comprises: a single-person operation module for obtaining single-person operation data of students in a virtual reality environment, wherein the single-person operation data comprises student information, single-person connection steps, single-person operation steps, single-person operation ranges and device parameter changes; an operation summary module comprising a plurality of standard ranges and a global operation range composed of the standard ranges; the single-person operation data is received and fused into global operation data, the single-person operation data is spatially aligned according to the relationship between the global operation range and the single-person operation range, the single-person operation data is time-series aligned according to the comparison relationship between the single-person connection steps and the single-person operation steps, and an initial operation set is generated; an interactive analysis module comprising a global initial parameter and a standard parameter interval of each standard range; the device parameters of the current global operation range are used as the initial parameters, the single-person operation data is sequentially taken out from the initial operation set, the device parameter changes of the single-person operation data are combined with the initial parameters, the current device parameters in the standard range after the single-person operation steps are calculated as first parameters, the first parameters are compared with the standard parameter interval corresponding to the standard range, the first parameters are used as the initial parameters, and the single-person operation data is repeatedly taken out from the initial operation set and combined with the initial parameters to calculate the first parameters until the first parameters are not in the standard parameter interval or the single-person operation data in the initial operation set has been completely taken out and used to calculate the first parameters; finally, operation results are generated according to the comparison results of the first parameters and the standard parameter interval; 2. The multi-user interactive virtual reality teaching system of claim 1, wherein: a teaching feedback module for generating completion information according to the operation results or generating interruption information according to the single-person operation data, the initial parameters and the first parameters taken out by the interactive analysis module last time.

3. The multi-user interactive virtual reality teaching system of claim 2, wherein: The single-person operation module further comprises a single-person prediction model; the training data of the single-person prediction model comprises historical single-person operation data of students, standard operation steps corresponding to the single-person operation range of the students and standard device parameter changes; the single-person prediction model learns the association relationship between the historical operation and the standard data through training, predicts the operation step of the students at the next moment as a prediction step based on the current single-person operation data and the initial parameters, predicts the device parameter changes of the students at the next moment as a prediction parameter change, uses the prediction results as the analysis basis to analyze the operation results by using the interactive analysis module, combines the operation results, the prediction step and the prediction parameter change to generate a prediction result prompt, and sends the prediction result prompt to a student terminal corresponding to the student information. The single-person prediction model is further used to evaluate the proficiency of the students by counting the training times of the same operation range and operation step in the training data of the single-person prediction model, determine the proficiency of the students in the corresponding operation range and operation step by combining the proportion of the historical single-person operation data meeting the standard parameters and the preset minimum operation times, and store the proficiency in association with the student information; When the interactive analysis module calculates the first parameters, the student information in the single-person operation data participating in the calculation of the first parameters is obtained, and the standard parameter interval is dynamically adjusted according to the proficiency corresponding to the student information.

4. The multi-user interactive virtual reality teaching system of claim 3, wherein: The single-person operation module controls the collection time of single-person operation data based on the proficiency of the student; when the proficiency of the student is less than a preset minimum proficiency, the single-person operation module acquires a standard operation step corresponding to a current operation range of the student as a central step, and acquires a previous step of the central step as a previous step; when the single-person operation module does not completely acquire the previous step, the collection time of the central step is extended, and an assistance prompt is generated according to the relationship between the previous step and the central step, and the assistance prompt is sent to the student information corresponding to the central step.

5. The multi-user interactive virtual reality teaching system of claim 4, wherein: The interaction analysis module internally stores a safety cooperation specification, the safety cooperation specification includes a standard parameter interval corresponding to each operation type and a dangerous interval, and an operation range affected by the dangerous interval is stored in association with the dangerous interval as a dangerous range; When the interaction analysis module calculates the first parameter, if it is detected that the first parameter is in the dangerous interval, all single-person operation steps before the current single-person operation step in the initial operation set are immediately queried as review steps, the student information of the review steps and the latest single-person operation data in the current operation step time sequence are acquired as dangerous data, and if the single-person operation range in the dangerous data has an overlapping area with the dangerous range, the teaching feedback module sends a danger prompt according to the student information in the dangerous data.

6. The multi-user interactive virtual reality teaching system of claim 5, wherein: The interaction analysis module also internally stores a delay proficiency coupling unit, which is used to collect an estimated value of the current network communication delay in real time, and the estimated value and the dangerous interval are used as inputs; The delay proficiency coupling unit maps the estimated value to a tolerable delay upper bound according to the proficiency level corresponding to the student information, and dynamically extends the data confirmation window of the student corresponding to the student information when the first parameter is not in the dangerous interval and is less than a preset danger threshold from the dangerous interval, if the proficiency corresponding to the student information is lower than the preset minimum proficiency and the estimated value exceeds the tolerable delay upper bound, the interaction analysis module immediately appends a virtual compensation step to the initial operation set, the virtual compensation step takes the midpoint of the dangerous interval as a trigger condition, delays the collection time of the current single-person operation step by Δt, and Δt is determined according to the association between the proficiency and the estimated value, and the priority of the danger prompt is improved by one level; If the proficiency of the student is higher than the preset threshold, the single-person operation module has a constant collection frequency.

7. The multi-user interactive virtual reality teaching system of claim 6, wherein: The interaction analysis module also includes a residual risk integral unit; the residual risk integral unit is used to accumulate the residual risk integral according to the length of stay of the first parameter in the dangerous interval each time the first parameter is detected to enter the dangerous interval, and when the residual risk integral exceeds a preset integral threshold, the standard parameter interval corresponding to the current student information is tightened by a preset percentage, and the tightened interval is broadcast to all student terminals.

8. The multi-user interactive virtual reality teaching system of claim 7, wherein: The safety policy module obtains the current single-person operation data of each student collected by the interaction analysis module, combines the standard parameter interval, the dangerous interval and the corresponding dangerous range, sets the dangerous range as a negative reward area through a reinforcement learning algorithm, and generates an initial probability distribution of the first parameter of each student within the standard parameter interval; subsequently, the safety policy module cooperatively adapts the initial probability distribution to the equipment parameter changes and the predicted steps of other students, and obtains a predicted step probability distribution of the first parameter within the standard parameter interval; when it is detected that the first parameter in the predicted step probability distribution of any student falls into the dangerous interval or is less than a preset dangerous interval distance from the dangerous interval, the safety policy module immediately sends an adjustment instruction containing a preset safety action parameter to the terminal of the student according to the student information, and synchronizes the instruction to other student terminals.

9. The multi-user interactive virtual reality teaching system of claim 8, wherein: The interaction analysis module immediately sends a preset calling instruction to the safety policy module after detecting that the residual risk score exceeds the preset score threshold through the residual risk score unit; After responding to the calling instruction, the safety policy module receives the dangerous interval in the current single-person operation data synchronized by the interaction analysis module, maps the dangerous interval as a temporary negative reward area, and incorporates the temporary negative reward area as a temporary increment into the original negative reward system; when calculating the predicted step probability distribution of the next round of first parameters, the safety policy module preferentially pushes the probability distribution result containing the temporary negative reward area to the student terminal with the highest residual risk score; The temporary negative reward area is automatically released after the calculation of the predicted step probability distribution of the next round of first parameters is completed, and does not affect the setting of the negative reward area corresponding to the original dangerous range.

Images