Teaching or training-oriented virtuality and reality combined teaching method and system
By simulating the aging of physical equipment and the impact of environmental factors in a virtual environment, the virtual control parameters are optimized, solving the problem of idealized response in the virtual environment and improving the authenticity and training effectiveness of virtual-real integrated teaching.
Patent Information
- Application Number
- CN202511285886.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-10
- Publication Date
- 2025-12-12
AI Technical Summary
In existing virtual-real integrated teaching methods, the virtual environment's response to operational control commands ignores the influence of environmental factors, making it difficult for learners to adapt in actual work scenarios and resulting in poor training effectiveness.
By receiving operation control commands from physical model devices, the system obtains real-time environmental parameters of the virtual environment and historical usage information of the physical devices. It then uses an attention mechanism to perform correlation analysis, optimizes virtual control parameters, and simulates the impact of environmental factors and device aging on the virtual model.
It improved the authenticity of teaching and training, enhanced learners' adaptability to the actual operating environment, and improved the effectiveness of practical skills training.
Smart Images
Figure CN121120332A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of virtual-real integration technology, and more specifically, to a virtual-real integrated teaching method and system for teaching or training. Background Technology
[0002] In the current field of teaching and training, the hybrid virtual-real teaching model is gradually gaining attention. It typically includes physical teaching equipment (such as scaled-down architectural equipment) and a virtual learning environment. Learners input operational control commands into the physical teaching equipment, and the virtual learning environment responds to these commands and dynamically updates its animations. However, existing hybrid virtual-real teaching methods still have significant drawbacks, leading to unsatisfactory training outcomes.
[0003] In existing technologies, virtual teaching environments often respond to operational control commands with 100% accuracy. This means that after a command is input, the virtual environment responds exactly as preset, ideally, ignoring the influence of various environmental factors. For example, in construction equipment training, when a learner inputs a lifting command using a scaled-down crane model, the virtual environment only displays an animation of the crane completing the lifting action, without considering the crucial environmental factor of wind. In real-world engineering scenarios, wind strength and direction directly affect the stability of the crane boom and the swing amplitude of the load, requiring operators to adjust their operating strategies in real time based on wind conditions. However, in existing virtual-real hybrid models, the lack of simulation of environmental factors like wind prevents learners from experiencing the complexity and uncertainty of real-world operations, hindering the development of their ability to handle unexpected situations.
[0004] Besides wind, environmental factors such as temperature, humidity, and terrain that may exist in virtual learning environments are often overlooked in current technologies. For example, in training on building water supply and drainage equipment, the different degrees of thermal expansion and contraction of pipes at different temperatures affect the operating status of the equipment. However, current virtual environments do not reflect this change in response to operating commands, making it difficult for learners to understand the potential impact of environmental factors on equipment operation. This idealized response to operating commands often leads to learners struggling to adapt quickly to the various factors in real-world work scenarios after training, resulting in a higher rate of operational errors and significantly diminishing the training effectiveness.
[0005] Therefore, how to incorporate the influence of environmental factors on the response to operational instructions in virtual-real integrated teaching to improve teaching and training effectiveness is a technical problem that urgently needs to be solved. Summary of the Invention
[0006] In response, the present invention provides a virtual-real integrated teaching method, system, electronic device, computer storage medium, and computer program product for teaching or training, to solve at least one of the above-mentioned technical problems.
[0007] A first aspect of the present invention provides a virtual-real hybrid teaching method for instruction or training, comprising the following steps: receiving operation control commands input by a user on a physical model device, parsing and converting the operation control commands into virtual control parameters; acquiring real-time environmental parameters in a virtual environment; acquiring historical usage information of the physical model device, predicting virtual aging parameters based on the historical usage information, and mapping the virtual aging parameters to a virtual model corresponding to the physical model device in the virtual environment; performing correlation analysis on the real-time environmental parameters and the virtual aging parameters of the virtual model using an attention mechanism to determine a set of associated parameter combinations, optimizing and adjusting the virtual control parameters based on the set of associated parameter combinations to obtain target virtual control parameters; and controlling the virtual model in the virtual environment to perform actions according to the target virtual control parameters.
[0008] A second aspect of the present invention provides a virtual-real integrated teaching system for teaching or training, comprising an input unit, an aging analysis unit, a correlation analysis unit, and a control unit; the input unit receives operation control commands input by a user on a physical model device, parses and converts the operation control commands into virtual control parameters, and acquires real-time environmental parameters in the virtual environment; the aging analysis unit acquires historical usage information of the physical model device, predicts virtual aging parameters based on the historical usage information, and maps the virtual aging parameters to the virtual model corresponding to the physical model device in the virtual environment; the correlation analysis unit uses an attention mechanism to perform correlation analysis on the real-time environmental parameters and the virtual aging parameters of the virtual model, determines a set of correlation parameter combinations, optimizes and adjusts the virtual control parameters based on the set of correlation parameter combinations, and obtains target virtual control parameters; the control unit controls the virtual model in the virtual environment to perform actions according to the target virtual control parameters.
[0009] A third aspect of the present invention provides an electronic device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the computer program, when executed by the processor, implements the method as described in any of the preceding claims.
[0010] A fourth aspect of the present invention provides a computer storage medium storing a computer program that can be executed by a processor to implement the method as described in any of the preceding claims.
[0011] A fifth aspect of the present invention provides a computer program product comprising a computer program executable by a processor to implement the method as described in any of the preceding claims.
[0012] This invention utilizes an attention mechanism to perform correlation analysis between real-time environmental parameters and virtual aging parameters, precisely optimizing virtual control parameters. Thus, it retains the realistic tactile feedback of physical equipment operation while incorporating the effects of environmental factors and equipment aging into the virtual model's actions, making virtual actions more closely resemble real-world scenarios. This effectively solves the problem of idealized operational responses and disconnect from reality in existing technologies, significantly improving the authenticity of teaching and training and the effectiveness of practical skills development, helping learners quickly adapt to real-world operating environments. Attached Figure Description
[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0014] Figure 1 This is a flowchart illustrating a virtual-real integrated teaching method for teaching or training disclosed in an embodiment of the present invention.
[0015] Figure 2 This is a schematic diagram of the architecture of a virtual-real integrated teaching or training system disclosed in an embodiment of the present invention.
[0016] Figure 3 This is a schematic diagram of the structure of a virtual-real integrated teaching system for teaching or training disclosed in an embodiment of the present invention. Detailed Implementation
[0017] The following specific embodiments illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0018] Furthermore, the technical features involved in the different embodiments of this application described below can be combined with each other as long as they do not conflict with each other.
[0019] like Figure 1As shown, this embodiment of the invention discloses a virtual-real integrated teaching method for teaching or training, including the following steps: S10, receiving operation control commands input by the user on a physical model device, parsing and converting the operation control commands into virtual control parameters; and obtaining real-time environmental parameters in the virtual environment.
[0020] like Figure 2 As shown, the architecture of the virtual-real integrated teaching or training system of the present invention includes a physical model device, a virtual environment, and a controller. The controller's input is connected to the physical model device, and its output is connected to the virtual environment. The virtual environment can be output to the user through a conventional display screen or VR / AR devices, etc., but the present invention does not specifically limit this.
[0021] The controller receives user input on the physical model equipment and performs parameter conversion. The physical model equipment is usually a scaled-down teaching device (such as a building equipment model). Users input operation control commands (such as start, stop, speed adjustment, steering, etc.) by operating the control components of the physical equipment (such as buttons, joysticks, knobs, etc.). The controller parses these commands and converts them into virtual control parameters (such as action parameters represented in numerical or code form) that can be recognized by the virtual environment.
[0022] At the same time, the controller also acquires real-time environmental parameters in the virtual environment, which is a simulation scene built according to teaching or training needs (such as a virtual scene of a construction site). The real-time environmental parameters include data on various environmental factors in the scene, such as wind speed and direction, temperature, humidity, light intensity, terrain slope, etc., at least one of these parameters, which will have a potential impact on the actions of the virtual model.
[0023] S20, obtain the historical usage information of the physical model device, predict the virtual aging parameters based on the historical usage information, and map the virtual aging parameters to the virtual model corresponding to the physical model device in the virtual environment.
[0024] First, historical usage information of the physical model equipment is acquired. This information includes the cumulative usage time of the equipment, the number and frequency of execution of each operation control command, abnormal load data that occurred during past operations (such as operation records exceeding the rated load), and equipment maintenance records (such as maintenance time, maintenance content, and replacement of parts). Based on this historical data, virtual aging parameters are predicted. Virtual aging parameters are quantitative indicators used to characterize the aging state of the equipment, such as component wear coefficients, performance degradation rates, and response sensitivity reduction coefficients. It is understood that the virtual aging parameters in this invention do not characterize the aging state of the physical model equipment, but rather the aging state of the virtual model of the physical model equipment in a virtual environment; that is, the aging state of the virtual model in the virtual environment is predicted based on the historical usage of the physical model equipment.
[0025] Finally, the calculated virtual aging parameters are mapped onto the virtual model corresponding to the physical device in the virtual environment, so that the virtual model can simulate the aging characteristics of the physical device due to long-term use, making the virtual scene closer to the actual use state of the device.
[0026] S30, an attention mechanism is used to perform correlation analysis on the real-time environmental parameters and the virtual aging parameters of the virtual model to determine a set of associated parameter combinations. Based on the set of associated parameter combinations, the virtual control parameters are optimized and adjusted to obtain the target virtual control parameters.
[0027] The attention mechanism originates from the field of deep learning. Its core idea is to simulate the focusing characteristics of human attention, identifying the factors with a higher degree of influence on the target result among numerous parameters. In this step, the controller performs correlation analysis on real-time environmental parameters and virtual aging parameters based on the attention mechanism. For example, it analyzes whether a certain aging parameter of the equipment (such as wear of transmission components) will have a more significant impact on the equipment's operation under specific wind conditions. By calculating the correlation weights between various parameters, it determines the set of correlation parameter combinations that have the greatest impact on the virtual control parameters (i.e., the key combination of environmental parameters and aging parameters). Based on this set of correlation parameter combinations, the controller optimizes and adjusts the initial virtual control parameters. For example, when the set of correlation parameter combinations includes "strong wind" and "high wear rate," the speed parameter in the virtual control parameters is appropriately reduced to simulate the equipment's behavior under real-world conditions, ultimately yielding target virtual control parameters that better reflect the actual scenario.
[0028] S40, control the virtual model in the virtual environment to act according to the target virtual control parameters.
[0029] After obtaining the target virtual control parameters, the controller drives the virtual model in the virtual environment to perform corresponding actions based on these parameters. For example, the virtual crane model completes a lifting action at an optimized speed and angle, and this lifting action reflects the effects of wind and equipment aging (such as slight swaying of the load due to wind, and slight delay in action response due to aging). Through the dynamic action display of the virtual model, users can intuitively see the virtual effects of operating physical equipment, taking into account environmental factors and equipment aging, thereby improving the realism and effectiveness of teaching or training.
[0030] This invention utilizes an attention mechanism to perform correlation analysis between real-time environmental parameters and virtual aging parameters, precisely optimizing virtual control parameters. Thus, it retains the realistic tactile feedback of physical equipment operation while incorporating the effects of environmental factors and equipment aging into the virtual model's actions, making virtual actions more closely resemble real-world scenarios. This effectively solves the problem of idealized operational responses and disconnect from reality in existing technologies, significantly improving the authenticity of teaching and training and the effectiveness of practical skills development, helping learners quickly adapt to real-world operating environments.
[0031] As an example, the method of predicting virtual aging parameters based on the historical usage information includes: normalizing the historical usage information, which includes cumulative usage time, execution frequency and number of times each operation command is executed, number of times abnormal loads occur and load values, maintenance frequency and maintenance interval duration; inputting the normalized historical usage information into a trained virtual aging prediction model, and outputting virtual aging parameters, which include component wear coefficients and performance degradation coefficients; wherein, the virtual aging prediction model is constructed based on the random forest algorithm.
[0032] The historical usage information of the collected physical model equipment is used to extract information that can describe the usage status of the physical model equipment in multiple dimensions, including: cumulative usage time, which reflects the overall usage intensity of the equipment; the number and frequency of execution of each operation command, which reflects the frequency of use of different functions, and high-frequency operation components are more prone to aging; the number of abnormal loads and the load value, which reflects the additional wear and tear on the equipment, and the greater the load and the more frequent the occurrence, the more serious the aging may be; the number of maintenance and the maintenance interval, which reflects the maintenance status of the equipment, and less maintenance and longer intervals will lead to accelerated aging.
[0033] The historical usage information was normalized to convert historical data of different dimensions and magnitudes into values between 0 and 1, eliminating the impact of differences in data units and ranges, and facilitating the calculation of the subsequent virtual aging prediction model. Specifically, the longer the cumulative usage time, the larger the abnormal load value, and the longer the maintenance interval, the closer the corresponding value is to 1, because these conditions are positively correlated with the degree of equipment aging. This numerical mapping allows for a direct quantification of the impact of each factor on aging, providing standardized feature data for the virtual aging prediction model.
[0034] The virtual aging prediction model is trained using normalized historical usage information as input and the random forest algorithm. The random forest algorithm, by constructing multiple decision trees and combining their results, exhibits strong anti-overfitting ability and prediction accuracy, and can effectively handle multi-dimensional input data. The virtual aging prediction model built based on this algorithm can learn the potential patterns between historical usage information and equipment aging, thereby achieving accurate prediction of virtual aging parameters.
[0035] The normalized historical usage information is input into the trained virtual aging prediction model. The model calculates based on the learned underlying patterns and outputs virtual aging parameters, including component wear coefficients and performance degradation coefficients, with values ranging from 0 to 1. Here, 0 represents no aging, and 1 represents severe aging.
[0036] As an example, the step of using an attention mechanism to perform correlation analysis on the real-time environmental parameters and the virtual aging parameters of the virtual model to determine a set of associated parameter combinations includes: feature encoding the real-time environmental parameters and the virtual aging parameters to obtain parameter vectors; inputting the encoded parameter vectors into a pre-built attention weight calculation module; calculating the correlation weights between each parameter using a multilayer perceptron; filtering out parameter combinations with obvious correlations exceeding a weight threshold to form an initial set of associated parameter combinations; calculating the mutual information values of all parameter combinations not included in the initial set of associated parameter combinations; normalizing the mutual information values; and filtering out parameter combinations with normalized mutual information values higher than a mutual information threshold to form a candidate supplementary set; and using... Hierarchical clustering algorithm clusters the candidate supplementary set, calculates the average association weight and average mutual information value of parameter combinations in each cluster category, and filters out effective cluster categories whose average association weight is higher than the lower limit of association weight and whose average mutual information value is higher than the lower limit of mutual information. The parameter combination corresponding to the cluster center of any effective cluster category is extracted, and each parameter combination in the effective cluster category is sorted from high to low according to association weight. The top M parameter combinations are selected. The parameter combination corresponding to the cluster center is merged with the top M parameter combinations to obtain the subset of indistinct type parameter combinations corresponding to the effective cluster category. All subsets of indistinct type parameter combinations are summarized to form the set of indistinct type parameter combinations, which is then added to the initial set of association parameter combinations to obtain the set of association parameter combinations.
[0037] First, the real-time environmental parameters and virtual aging parameters are feature-encoded, converting non-numerical parameters into computer-recognizable vector forms to ensure a consistent parameter format. Then, the encoded parameter vectors are input into the attention weight calculation module. A multilayer perceptron is used to simulate the nonlinear relationships between parameters, accurately calculating the correlation weights of each parameter combination. Parameter combinations with correlation weights exceeding a weight threshold (e.g., 0.6) are selected as clearly correlated combinations, forming an initial set of correlated parameter combinations. These parameter combinations have a direct and significant impact on the virtual control parameters.
[0038] To avoid overlooking potential influencing factors, for parameter combinations that did not enter the initial set (i.e., whose association weights did not exceed the weight threshold), their mutual information values are further calculated. This value quantifies the dependency between parameters, is not limited by linear relationships, and can capture implicit associations. After normalizing the mutual information values to the range of 0-1, combinations exceeding the mutual information threshold (e.g., 0.3) are selected to form a candidate supplementary set.
[0039] Hierarchical clustering algorithm is used to group parameter combinations with similar features in the candidate supplementary set into one class, reducing redundancy and highlighting potential association patterns. The average association weight (reflecting the overall association strength) and average mutual information value (reflecting the overall dependence) of each cluster are calculated, and dual screening conditions are set (such as average association weight ≥ 0.2 and average mutual information value ≥ 0.25) to retain effective clusters.
[0040] For each valid cluster category, extract the parameter combination corresponding to its cluster center (representing the core association feature of that category); simultaneously, sort the parameter combinations within that valid cluster category according to their association weight, and select the top M parameter combinations (e.g., M=5, where M can also be a set percentage of the total number of parameter combinations in the initial association parameter combination set). Merge the two types of parameter combinations to form a subset of indistinct type parameter combinations corresponding to that valid cluster category. This setup includes both the core features of the category and strongly associated individuals, ensuring the comprehensiveness of indistinct combinations.
[0041] The subsets of indistinct parameter combinations corresponding to each effective cluster category are aggregated to form a complete set of indistinct parameter combinations. This set is then added to the initial set of associated parameter combinations, and after deduplication, the set of associated parameter combinations is obtained. This set contains both obvious and indistinct type combinations, comprehensively covering the complex relationships between environmental and aging parameters, which is beneficial for subsequently determining more accurate target virtual control parameters.
[0042] As an example, the weight threshold is dynamically determined, specifically including: extracting key environmental factors from the real-time environmental parameters, including wind speed, temperature deviation, and humidity saturation rate; using the key environmental factors as input, calculating the initial value of the weight threshold through linear regression; counting the number of parameters in the real-time environmental parameters that exceed the normal working range corresponding to the physical model equipment; adjusting the initial value based on the number of parameters to obtain the final value of the weight threshold.
[0043] First, wind speed, temperature deviation, and humidity saturation rate are extracted as key environmental factors from real-time environmental parameters, as these factors significantly affect equipment operation. Wind speed is quantified from 0 to 12 (e.g., 0 represents no wind, 12 represents a hurricane), directly reflecting its impact on equipment stability. Temperature deviation is the absolute value of the difference between the actual temperature and the equipment's optimal operating temperature (e.g., a deviation of 5 when the optimal temperature is 25℃ and the actual temperature is 30℃), reflecting the impact of temperature deviation on equipment performance. Humidity saturation rate is the ratio of actual humidity to critical humidity (e.g., a saturation rate of 0.75 when the critical humidity is 80% and the actual humidity is 60%), reflecting the potential effect of humidity on equipment aging and operation.
[0044] Next, using key environmental factors as input, the initial value of the weight threshold is calculated through linear regression. The calculation formula is as follows: Initial value of weight threshold = 0.4 + 0.05 × (wind force level / 12) + 0.02 × (temperature deviation value / ΔT) max ) + 0.03 × humidity saturation rate. Where the coefficients (0.05, 0.02, 0.03) reflect the weight of each factor's influence on the threshold (wind has the greatest impact), ΔT max This represents the maximum allowable temperature deviation for the physical model device (e.g., 30℃). The initial value is limited to the range of 0.5-0.7 to ensure that the threshold is neither too high, leading to missed obvious associated combinations, nor too low, resulting in redundant combinations. For example, under conditions of strong winds (level 8), large temperature differences (10℃), and high humidity (saturation 0.9), the initial value can reach 0.67, tending to filter out more obviously associated combinations affected by strong environmental conditions.
[0045] Simultaneously, the system also counts the number of real-time environmental parameters exceeding the normal operating range of the equipment (e.g., wind force > 6, temperature > 35℃ are considered exceeding the range): if the number exceeding the range is ≥ 2 (e.g., strong wind + high temperature simultaneously), it indicates severe environmental interference, and the initial value needs to be lowered by 0.05 (e.g., from 0.65 to 0.6) to include more obviously correlated combinations and adapt to complex environments; if the number exceeding the range is 0 (normal environment), the initial value is raised by 0.05 (e.g., from 0.55 to 0.6) to reduce redundant combinations. The dynamic weight threshold method adopted in this invention allows the weight threshold to better match the actual environmental complexity, ensuring that under different environmental conditions, it can accurately capture combinations that significantly affect virtual control parameters, and can also adapt to changes in environmental complexity in teaching scenarios through quantity adjustment.
[0046] Finally, the correlation weights between the parameters calculated by the multilayer perceptron are compared with the final threshold, and combinations with weights exceeding the threshold are retained. For example, when the final threshold is 0.6, closely related combinations such as "strong wind force + high wear coefficient" and "high temperature deviation + performance degradation coefficient" are included in the initial set of correlation parameter combinations.
[0047] As an example, optimizing and adjusting the virtual control parameters based on the set of associated parameter combinations to obtain target virtual control parameters includes: assigning influence weights to each parameter combination in the set of associated parameter combinations, wherein the influence weights are calculated by weighting the association weights and mutual information values of the parameter combination; constructing a parameter influence coefficient matrix, wherein the elements in the matrix correspond to the mapping relationship between the virtual control parameters and the corresponding parameter combinations, reflecting the degree of influence of the parameter combination on the virtual control parameter; calculating the correction amount corresponding to the virtual control parameter based on the parameter influence coefficients and the influence weights; determining the correction direction based on the physical meaning of the parameter combination; and correcting the virtual control parameter based on the correction direction and the correction amount to obtain the target virtual control parameter; wherein the physical meaning includes reflecting adverse effects and reflecting neutral effects.
[0048] An influence weight is assigned to each combination in the set of associated parameter combinations. This weight is calculated by combining the association weight (reflecting the strength of the direct association between parameters) and the mutual information value (reflecting the potential dependency relationship) of the combination. For example, the formula is: Influence weight = 0.6 × Association weight + 0.4 × Mutual information value.
[0049] A parameter influence coefficient matrix is constructed, where each element corresponds to a mapping relationship between a specific parameter combination and a specific virtual control parameter. The larger the parameter influence coefficient (e.g., in the range of 0-0.2), the more significant the impact of the parameter combination on the corresponding virtual control parameter. For example, the combination of "strong wind force + high wear coefficient" has a higher influence coefficient on the "steering speed" control parameter, but a lower influence coefficient on the "starting torque". By constructing the above matrix, the complex multi-parameter influence relationships are structured, clarifying the differentiated impact of different combinations on each control parameter.
[0050] The correction amount is calculated based on the parameter influence coefficient and influence weight. The formula can be expressed as: Correction amount = Virtual control parameter × ∑(Influence weight × Parameter influence coefficient), which comprehensively reflects the superimposed influence of each parameter combination. Then, the correction direction is determined according to the physical meaning of the parameter combination: if the combination reflects an adverse effect (such as "high temperature + performance degradation"), the control parameter needs to be lowered (target value = original parameter - correction amount) to simulate the real state of limited performance of the physical model equipment; if it is a neutral effect (such as "normal temperature + low wear"), the parameter can be finely adjusted to increase (target value = original parameter × (1 + 0.5 × correction amount)) to reflect environmental adaptability. Finally, through targeted correction, a target virtual control parameter that better fits the actual scenario is obtained.
[0051] like Figure 3 As shown, this embodiment of the invention also provides a virtual-real integrated teaching system for teaching or training, including an input unit 101, an aging analysis unit 102, a correlation analysis unit 103, and a control unit 104.
[0052] The input unit 101 receives operation control commands input by the user on the physical model device, parses and converts the operation control commands into virtual control parameters, and acquires real-time environmental parameters in the virtual environment.
[0053] The aging analysis unit 102 acquires the historical usage information of the physical model device, predicts virtual aging parameters based on the historical usage information, and maps the virtual aging parameters to the virtual model corresponding to the physical model device in the virtual environment.
[0054] The correlation analysis unit 103 uses an attention mechanism to perform correlation analysis on the real-time environmental parameters and the virtual aging parameters of the virtual model, determines a set of correlation parameter combinations, and optimizes and adjusts the virtual control parameters based on the set of correlation parameter combinations to obtain the target virtual control parameters.
[0055] The control unit 104 controls the virtual model in the virtual environment to perform actions according to the target virtual control parameters.
[0056] As an example, the aging analysis unit specifically: normalizes the historical usage information, which includes cumulative usage time, execution frequency and number of times each operation command is executed, number of times abnormal loads occur and load values, maintenance frequency and maintenance interval duration; inputs the normalized historical usage information into a trained virtual aging prediction model, and outputs virtual aging parameters, which include component wear coefficients and performance degradation coefficients; wherein, the virtual aging prediction model is constructed based on the random forest algorithm.
[0057] As an example, the correlation analysis unit 103 specifically: performs feature encoding on the real-time environmental parameters and the virtual aging parameters to obtain parameter vectors; inputs the encoded parameter vectors into a pre-constructed attention weight calculation module; calculates the correlation weights between each parameter using a multilayer perceptron; filters out parameter combinations with obvious correlations exceeding a weight threshold to form an initial set of correlation parameter combinations; calculates the mutual information values of all parameter combinations not included in the initial set of correlation parameter combinations; normalizes the mutual information values; filters out parameter combinations with normalized mutual information values higher than a mutual information threshold to form a candidate supplementary set; and uses a hierarchical clustering algorithm to cluster the candidate supplementary set, calculating the mutual information values of each parameter combination. The average association weight and average mutual information value of parameter combinations in the cluster categories are used to screen out effective cluster categories whose average association weight is higher than the lower limit of association weight and whose average mutual information value is higher than the lower limit of mutual information. The parameter combinations corresponding to the cluster centers of any effective cluster category are extracted, and the parameter combinations in the effective cluster category are sorted from high to low according to their association weight. The top M parameter combinations are selected. The parameter combinations corresponding to the cluster centers are merged with the top M parameter combinations to obtain the subset of indistinct type parameter combinations corresponding to the effective cluster category. The subsets of indistinct type parameter combinations are summarized to form the set of indistinct type parameter combinations. The set of indistinct type parameter combinations is added to the initial set of association parameter combinations to obtain the set of association parameter combinations.
[0058] As an example, the correlation analysis unit 103 specifically: extracts key environmental factors from the real-time environmental parameters, including wind force level, temperature deviation value, and humidity saturation rate; uses the key environmental factors as inputs to calculate the initial value of the weight threshold through linear regression; counts the number of parameters in the real-time environmental parameters that exceed the normal working range corresponding to the physical model equipment; adjusts the initial value based on the number of parameters to obtain the final value of the weight threshold.
[0059] As an example, the correlation analysis unit 103 specifically: assigns an influence weight to each parameter combination in the correlation parameter combination set, the influence weight being calculated by weighting the correlation weight and mutual information value of the parameter combination; constructs a parameter influence coefficient matrix, the elements of which correspond to the mapping relationship between virtual control parameters and corresponding parameter combinations, reflecting the degree of influence of the parameter combination on the virtual control parameter; calculates the correction amount corresponding to the virtual control parameter based on the parameter influence coefficient and the influence weight, determines the correction direction based on the physical meaning of the parameter combination, and corrects the virtual control parameter based on the correction direction and the correction amount to obtain the target virtual control parameter; wherein, the physical meaning includes reflecting adverse effects and reflecting neutral effects.
[0060] This invention also provides an electronic device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the computer program, when executed by the processor, implements the method as described in any of the foregoing embodiments.
[0061] This invention also provides a computer storage medium storing a computer program that can be executed by a processor to implement the methods described in any of the foregoing claims.
[0062] This invention also provides a computer program product comprising a computer program that can be executed by a processor to implement the method as described in any of the foregoing embodiments.
[0063] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0064] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A virtual-real integrated teaching method for instruction or training, characterized in that, The steps include: receiving operation control commands input by the user on the physical model device, parsing and converting the operation control commands into virtual control parameters; obtaining real-time environmental parameters in the virtual environment; obtaining historical usage information of the physical model device, predicting virtual aging parameters based on the historical usage information, and mapping the virtual aging parameters to the virtual model corresponding to the physical model device in the virtual environment. An attention mechanism is used to perform correlation analysis on the real-time environmental parameters and the virtual aging parameters of the virtual model to determine a set of associated parameter combinations. Based on the set of associated parameter combinations, the virtual control parameters are optimized and adjusted to obtain target virtual control parameters. The virtual model in the virtual environment is then controlled to act according to the target virtual control parameters.
2. The virtual-real integrated teaching method for teaching or training according to claim 1, characterized in that: The virtual aging parameters are predicted based on the historical usage information, including: normalizing the historical usage information, which includes cumulative usage time, execution frequency and number of operation commands, number of abnormal load occurrences and load values, maintenance frequency and maintenance interval duration; inputting the normalized historical usage information into a trained virtual aging prediction model, and outputting virtual aging parameters, which include component wear coefficient and performance degradation coefficient; wherein, the virtual aging prediction model is constructed based on the random forest algorithm.
3. The virtual-real integrated teaching method for teaching or training according to claim 2, characterized in that: An attention mechanism is used to perform correlation analysis on the real-time environmental parameters and the virtual aging parameters of the virtual model to determine a set of associated parameter combinations. This includes: encoding the real-time environmental parameters and the virtual aging parameters to obtain parameter vectors; inputting the encoded parameter vectors into a pre-built attention weight calculation module; calculating the correlation weights between parameters using a multilayer perceptron; filtering out parameter combinations with clearly correlated correlations exceeding a weight threshold to form an initial set of associated parameter combinations; calculating the mutual information values of all parameter combinations not included in the initial set; normalizing the mutual information values; filtering out parameter combinations with normalized mutual information values higher than a mutual information threshold to form a candidate supplementary set; and using hierarchical clustering... The method clusters the candidate supplementary set, calculates the average association weight and average mutual information value of parameter combinations in each cluster category, and filters out effective cluster categories whose average association weight is higher than the lower limit of association weight and whose average mutual information value is higher than the lower limit of mutual information. It extracts the parameter combinations corresponding to the cluster centers of any effective cluster category, sorts the parameter combinations in that effective cluster category from high to low according to their association weight, and selects the top M parameter combinations. It merges the parameter combinations corresponding to the cluster centers with the top M parameter combinations to obtain a subset of indistinct type parameter combinations corresponding to that effective cluster category. It summarizes all the subsets of indistinct type parameter combinations to form a set of indistinct type parameter combinations, and supplements the set of indistinct type parameter combinations into the initial set of association parameter combinations to obtain the set of association parameter combinations.
4. A virtual-real integrated teaching method for instruction or training according to claim 3, characterized in that: The weight threshold is dynamically determined, specifically including: extracting key environmental factors from the real-time environmental parameters, including wind force level, temperature deviation value, and humidity saturation rate; using the key environmental factors as input, calculating the initial value of the weight threshold through linear regression; counting the number of parameters in the real-time environmental parameters that exceed the normal working range corresponding to the physical model equipment; adjusting the initial value based on the number of parameters to obtain the final value of the weight threshold.
5. A virtual-real integrated teaching method for instruction or training according to claim 4, characterized in that: The virtual control parameters are optimized and adjusted based on the set of associated parameter combinations to obtain target virtual control parameters. This includes: assigning influence weights to each parameter combination in the set of associated parameter combinations, wherein the influence weights are calculated by weighting the association weights and mutual information values of the parameter combination; constructing a parameter influence coefficient matrix, wherein the elements in the matrix correspond to the mapping relationship between the virtual control parameters and the corresponding parameter combinations, reflecting the degree of influence of the parameter combination on the virtual control parameter; calculating the correction amount corresponding to the virtual control parameter based on the parameter influence coefficients and the influence weights; determining the correction direction based on the physical meaning of the parameter combination; and correcting the virtual control parameter based on the correction direction and the correction amount to obtain the target virtual control parameter. The physical meaning includes reflecting adverse effects and reflecting neutral effects.
6. A virtual-real hybrid teaching system for instruction or training, characterized in that: It includes an input unit, an aging analysis unit, a correlation analysis unit, and a control unit; the input unit receives operation control commands input by the user on the physical model device and parses and converts the operation control commands into virtual control parameters; In addition, the real-time environmental parameters in the virtual environment are obtained; the aging analysis unit obtains the historical usage information of the physical model device, predicts virtual aging parameters based on the historical usage information, and maps the virtual aging parameters to the virtual model corresponding to the physical model device in the virtual environment. The correlation analysis unit uses an attention mechanism to perform correlation analysis on the real-time environmental parameters and the virtual aging parameters of the virtual model, determines a set of correlation parameter combinations, and optimizes and adjusts the virtual control parameters based on the set of correlation parameter combinations to obtain target virtual control parameters; the control unit controls the virtual model in the virtual environment to perform actions according to the target virtual control parameters.
7. A virtual-real integrated teaching system for teaching or training according to claim 6, characterized in that: The aging analysis unit specifically performs the following steps: normalizes the historical usage information, which includes cumulative usage time, execution frequency and number of operation commands, number of abnormal load occurrences and load values, maintenance frequency and maintenance interval duration; inputs the normalized historical usage information into a trained virtual aging prediction model, and outputs virtual aging parameters, which include component wear coefficients and performance degradation coefficients; wherein the virtual aging prediction model is constructed based on the random forest algorithm.
8. An electronic device comprising: At least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, characterized in that: when the computer program is executed by the processor, it implements the method as described in any one of claims 1-5.
9. A computer storage medium, characterized in that: The computer storage medium stores a computer program that can be executed by a processor to implement the method as described in any one of claims 1-5.
10. A computer program product, characterized in that: The computer program product includes a computer program that can be executed by a processor to implement the method as described in any one of claims 1-5.