An interactive command recommendation method, device and medium for CAD software

Abstract

This application discloses an interactive command recommendation method, device, and medium for CAD software, belonging to the field of computer-aided design software technology, and solves the problem of low accuracy in existing operation command recommendation results. The method includes: based on the user's design state and selected elements, fusing the probability values ​​of the next-stage operation commands corresponding to a preset general probability transition matrix and a preset personal probability transition matrix, respectively, to obtain the final probability value of the predicted next-stage operation commands; constructing a recommended command list suitable for the current user based on the final probability value; and dynamically updating the preset personal probability transition matrix based on the probability value of the operation commands actually selected by the current user and the historical selection probability values ​​corresponding to the selected operation commands, so as to predict the next recommended command list based on the updated preset personal probability transition matrix, thereby achieving dynamic optimization of interactive command recommendation.

Description

Technical Field

[0001] This application relates to the field of computer-aided design software technology, and in particular to an interactive command recommendation method, device and medium for CAD software. Background Technology

[0002] Computer-aided design (CAD) software, as a core tool in modern engineering design, has been widely used in many fields such as machinery, architecture, and electronics. In recent years, with the maturity of cloud computing technology, cloud-based CAD has emerged. Cloud-based CAD handles complex model calculations, data storage, and rendering tasks on the server side, while the user client is mainly responsible for displaying the interactive interface and responding to user input.

[0003] However, as cloud-based CAD software becomes increasingly powerful, its command sets have also become larger and more complex. Faced with numerous menus, toolbars, and function panels, users often need to spend a lot of time navigating the interface and searching for specific commands, which not only disrupts the continuous design flow but also significantly reduces design efficiency.

[0004] To improve user experience, some existing CAD systems have attempted to introduce command recommendation mechanisms, aiming to predict user intent and proactively provide the next command that might be used. However, existing recommendation mechanisms are limited by the isolation of single-machine data, and their recommendation logic is often based on preset static rules or extremely limited local historical operation data. Secondly, even some CAD systems with cloud infrastructure typically only build a single, general recommendation model, using this model to provide similar recommendation lists for users with different experience levels, professional fields, and operating habits. This results in a serious disconnect between the recommendations of existing mechanisms and the actual needs of users, leading to low recommendation accuracy. Summary of the Invention

[0005] This application provides an interactive command recommendation method, device, and medium for CAD software to solve the following technical problem: the recommendation results of existing recommendation mechanisms are seriously out of touch with the actual needs of users, and the recommendation accuracy is low.

[0006] The embodiments of this application adopt the following technical solutions:

[0007] This application provides an interactive command recommendation method for CAD software. It includes: acquiring the user's design state and selected elements in a cloud-based CAD design environment; wherein the design state includes at least one of sketch mode and non-sketch mode; based on the design state and selected elements, fusing the probability values ​​of the next-stage operation commands corresponding to a preset general probability transition matrix and a preset personal probability transition matrix used to predict the next-stage operation commands to obtain the final probability values ​​of the predicted next-stage operation commands; constructing a recommended command list suitable for the current user based on the final probability values ​​of the predicted next-stage operation commands; and dynamically updating the preset personal probability transition matrix based on the probability values ​​of the operation commands actually selected by the current user and the historical selection probability values ​​corresponding to the selected operation commands, so as to predict the next recommended command list based on the updated preset personal probability transition matrix, thereby achieving dynamic optimization of interactive command recommendation.

[0008] In one implementation of this application, before fusing the probability values ​​of the next-stage operation commands corresponding to the preset general probability transition matrix and the preset individual probability transition matrix used to predict the next-stage operation command, the method further includes: obtaining the historical operation command sequence of the group users on the cloud CAD platform; wherein, the historical operation command sequence includes at least one of the following: timestamp, design state, selected element, and the final executed command; based on the historical operation command sequence of the group users, determining the general next-stage operation commands corresponding to different combinations of design states and selected elements; based on the frequency of the general next-stage operation commands corresponding to different combinations, determining the general probability values ​​of the next-stage operation commands corresponding to different combinations, and constructing a preset general probability transition matrix based on the combination information and the general probability values ​​of the next-stage operation commands corresponding to the combination information; wherein, the elements in the preset general probability transition matrix represent the probability that the group users execute different operation commands given a design state and selected element.

[0009] In one implementation of this application, after constructing a preset general probability transition matrix based on the general probability value of the next-stage operation command corresponding to the combined information, the method further includes: comparing the general probability value with a preset probability threshold; deleting general probability values ​​that are less than the preset probability threshold; wherein, the preset probability threshold is used to avoid a general probability value of 0, so as to ensure that all commands have a basic probability of being recommended.

[0010] In one implementation of this application, before fusing the probability values ​​of the next-stage operation commands corresponding to the preset general probability transition matrix and the preset personal probability transition matrix used to predict the next-stage operation command, the method further includes: obtaining the historical operation command sequence of the current user on the cloud CAD platform; determining the individual next-stage operation commands corresponding to different design states and selected elements based on the historical operation command sequence of the current user; determining the individual probability values ​​of the individual next-stage operation commands corresponding to different combinations based on the frequency of the individual next-stage operation commands corresponding to different combinations, so as to construct a preset personal probability transition matrix based on the individual probability values; wherein, the elements in the preset personal probability transition matrix represent the probability that the current individual user executes different operation commands given a design state and selected elements.

[0011] In one implementation of this application, based on the design state and selected element, the probability values ​​of the next-stage operation command corresponding to the preset general probability transition matrix and the preset individual probability transition matrix used to predict the next-stage operation command are fused to obtain the final probability value of the predicted next-stage operation command. Specifically, this includes: determining weight coefficients for balancing group experience and individual habits based on the historical operation command data corresponding to the current user and the historical operation command data corresponding to the group users; obtaining the first probability value of the next-stage operation command corresponding to the current design state and selected element in the preset general probability transition matrix; wherein the current design state and selected element correspond to one or more next-stage operation commands; obtaining the second probability value of the next-stage operation command corresponding to the current design state and selected element in the preset individual probability transition matrix; and weightedly fusing the first probability value and the second probability value corresponding to the same next-stage operation command based on the weight coefficients to obtain the final probability value of the next-stage prediction.

[0012] In one implementation of this application, a recommended command list suitable for the current user is constructed based on the predicted final probability values ​​of different operation commands in the next stage. Specifically, this includes: sorting the predicted different operation commands in the next stage according to their respective final probability values ​​to construct an initial recommended command list; constructing a parametric association graph based on the current user's current operation data and historical operation data; wherein the parametric association graph includes at least one of the following: parent-child relationship, dimension-driven relationship, and geometric constraint relationship between different operation commands; comparing the parametric association graph with a preset CAD domain knowledge rule base, and predicting design conflicts that may occur after executing the recommended commands based on the comparison results; retrieving resolved cases from the historical case library whose similarity to the current design conflict scenario is greater than a preset similarity threshold, and determining the operation commands used in the resolved cases as compensatory operation commands; binding the compensatory operation commands with the recommended commands, and optimizing the initial recommended command list based on the bound recommended commands to obtain the final recommended command list.

[0013] In one implementation of this application, after constructing a recommended command list suitable for the current user, the method further includes: comparing the operation command sequence generated by the current user for the current task with a preset parameterized template; wherein the preset parameterized template consists of historical operation command sequences that meet high-frequency conditions, and also includes parameter value rules corresponding to the historical operation command sequences; when the comparison result meets the triggering conditions of the preset parameterized template, a reference command sequence is determined from the preset parameterized template; using the current design state and the real-time geometric attributes corresponding to the selected element as input, parameter values ​​are dynamically filled for each command in the reference command sequence to generate a command sequence instance adapted to the current task; the command sequence instance is simulated and executed, and when the simulation execution result meets the preset execution conditions, the current user is provided with a one-click option to execute the instantiated command sequence; when the user triggers the option to execute the instantiated command sequence, the instantiated sequence is executed in sequence, and a parameter determination interface is provided to the user before each operation command is executed.

[0014] In one implementation of this application, a preset personal probability transition matrix is ​​dynamically updated based on the probability value of the operation command actually selected by the current user and the historical selection probability value corresponding to the selected operation command. Specifically, this includes: selecting a preset number of historical operation commands based on the chronological order of the historical operation commands executed by the current user; determining the recent operation probability value for executing different next-stage operation commands under a given design state and selected element based on the preset number of historical operation commands; weighting and fusing the recent operation probability value with the corresponding probability value in the preset personal probability transition matrix based on a preset attenuation factor to determine the updated probability value; when there is no corresponding historical operation command for a given design state and selected element, using the calculated recent operation frequency as the updated probability value; and replacing the data in the preset personal probability transition matrix based on the updated probability value to dynamically update the preset personal probability transition matrix.

[0015] This application provides an interactive command recommendation device for CAD software, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: obtain the design state and selected elements corresponding to the user in a cloud-based CAD design environment; wherein the design state includes at least one of sketch mode and non-sketching mode; based on the design state and selected elements, fuse the probability values ​​of the next-stage operation commands corresponding to a preset general probability transition matrix and a preset personal probability transition matrix used to predict the next-stage operation commands, respectively, to obtain the final probability value of the predicted next-stage operation commands; construct a recommended command list suitable for the current user based on the predicted final probability values ​​of different next-stage operation commands; dynamically update the preset personal probability transition matrix based on the probability value of the operation commands actually selected by the current user and the historical selection probability values ​​corresponding to the selected operation commands, so as to predict the next recommended command list based on the updated preset personal probability transition matrix, thereby achieving dynamic optimization of interactive command recommendation.

[0016] This application provides a non-volatile computer storage medium storing computer-executable instructions. These instructions are configured to: acquire the user's design state and selected elements in a cloud-based CAD design environment; wherein the design state includes at least one of sketch mode and non-sketch mode; based on the design state and selected elements, fuse the probability values ​​of the next-stage operation commands corresponding to a preset general probability transition matrix and a preset personal probability transition matrix used to predict the next-stage operation commands to obtain the final probability value of the predicted next-stage operation commands; construct a recommended command list suitable for the current user based on the predicted final probability values ​​of different next-stage operation commands; and dynamically update the preset personal probability transition matrix based on the probability value of the operation commands actually selected by the current user and the historical selection probability values ​​corresponding to the selected operation commands, so as to predict the next recommended command list based on the updated preset personal probability transition matrix, thereby achieving dynamic optimization of interactive command recommendations.

[0017] The above-mentioned technical solutions adopted in this application embodiment can achieve the following beneficial effects: Firstly, by obtaining the design state and specific selected elements, this application embodiment determines the user's current design intent, establishing an accurate context for operation command recommendations. Secondly, this application embodiment innovatively integrates a general probability transition matrix reflecting group design patterns with a personal probability transition matrix reflecting individual operating habits, ensuring the rationality of recommendations for new users while achieving personalized adaptation for experienced users, solving the problem of traditional recommendation mechanisms being disconnected from actual user needs. Furthermore, the recommendation command list constructed based on probability prediction in this application embodiment can effectively filter irrelevant commands, prioritizing the commands most likely to be needed by the user, significantly reducing menu search time and improving efficiency. Finally, by continuously updating the personal probability transition matrix through real-time learning of actual user operations, the recommendation model is continuously optimized as user experience grows, improving the accuracy and adaptability of recommendations. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0019] Figure 1 A schematic diagram of a command prediction system provided in an embodiment of this application;

[0020] Figure 2 A flowchart illustrating an interactive command recommendation method for CAD software, provided as an embodiment of this application;

[0021] Figure 3 This is a schematic diagram of an interactive command recommendation device for CAD software provided in an embodiment of this application.

[0022] Figure label:

[0023] 200: Device for recommending interactive commands for CAD software; 201: Processor; 202: Memory. Detailed Implementation

[0024] This application provides an interactive command recommendation method, device, and medium for CAD software.

[0025] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0026] Figure 1 A schematic diagram of a command prediction system provided in an embodiment of this application is shown below. Figure 1 As shown, the command prediction system includes a browser client, a server scheduling client, and a command prediction server. The browser client's data collection module records user actions and current states (such as edit mode, selected objects, etc.), and forwards the data along with the user ID to the command prediction server via the server scheduling client. The prediction server retrieves the corresponding user information based on the user ID, and fuses the probability values ​​of the next-stage operation commands corresponding to the general probability transition matrix and the personal probability transition matrix based on the user's actions, generating a personalized recommended command list, which is then returned to the front end for display via the scheduling client. When a user accepts a recommended command, this selection action is transmitted back via the scheduling client and stored in the database for real-time updates of personal behavior data, driving dynamic optimization of the personal matrix.

[0027] Figure 2 This document provides a flowchart of an interactive command recommendation method for CAD software, as illustrated in an embodiment of this application. Figure 2 As shown, the recommended method for interactive commands in CAD software includes the following steps:

[0028] Step 101: Obtain the user's design status and selected elements in the cloud-based CAD design environment.

[0029] In one implementation of this application, the browser-side data collection module collects the user's design state and selected elements. The design state can be divided into sketch state and non-sketching state, and the selected elements include information such as sketches, features, points, lines, and surfaces. When the user selects an operable element, the command prediction function is triggered, sending the state and selected element to the server scheduler, which then forwards them to the command prediction server to predict the next stage of operation commands for the current user.

[0030] Step 102: Based on the design state and selected elements, the probability values ​​of the next stage operation commands corresponding to the preset general probability transition matrix and the preset personal probability transition matrix used to predict the next stage operation command are fused to obtain the final probability value of the predicted next stage operation command.

[0031] In one implementation of this application, a sequence of historical operation commands from a group of users is obtained on a cloud-based CAD platform. This sequence includes at least one of the following: timestamp, design state, selected element, and the final executed command. Based on this sequence, different combinations of design states and selected elements are identified, along with their corresponding general next-stage operation commands. Based on the frequency of these general next-stage operation commands for each combination, a general probability value for each next-stage operation command is determined. A pre-defined general probability transition matrix is ​​then constructed based on the combination information and the corresponding general probability values ​​of the next-stage operation commands. Each element in the pre-defined general probability transition matrix represents the probability that the group of users will execute different operation commands given a specific design state and selected element.

[0032] Specifically, the system collects historical operation command sequences from the log database of the cloud-based CAD platform. Each operation record includes at least a timestamp, design status, selected element, and the final executed command. Statistical analysis is performed on the operation records to construct a mapping relationship between conditions and events. Specifically, each operation record is scanned and treated as a sample of an event occurring under specific conditions. The condition is the combination of design status and selected element, and the event is the final executed command. A frequency statistics table is constructed by counting the number of times each different command is executed after each unique combination of conditions across all records. For example, the system might count the number of times a user subsequently executes the Insert Hole Wizard and the number of times a chamfer is inserted under the condition of selecting a cylindrical surface in part mode.

[0033] After obtaining the frequency statistics table, the frequencies are converted into a probability distribution through probability calculation. For each combination of conditions, a general probability value for executing the command is calculated. The calculation formula is the number of times the command appears under this condition divided by the total number of times all commands appear under this condition. Finally, all probability values ​​are organized into a preset general probability transition matrix. This preset general probability transition matrix is ​​only updated after a preset time has elapsed or after receiving an update command from the user.

[0034] Specifically, the general probability transition matrix in this application embodiment is constructed based on the historical operation command sequence data of all users on the platform, describing the probability distribution of a group of users executing the next operation in a specific state, denoted as... .in, Indicates the user's current design state. This represents the element selected by the user. This represents the modeling command to be predicted. That is, the group of users in the state When selecting an element Execute command The probability of.

[0035] In one implementation of this application, after constructing the preset general probability transition matrix, the method further includes: comparing the general probability values ​​with a preset probability threshold, and deleting general probability values ​​that are less than the preset probability threshold. The preset probability threshold is used to avoid a general probability value of 0, ensuring that all commands have a base probability of being recommended.

[0036] Specifically, this application employs a probability smoothing mechanism. To prevent commands from being excluded from recommendation due to their probabilities being zero in the general probability transition matrix, a general probability is set for any command. The minimum value is ( (>0 and is a very small value), ensuring that all commands have a base probability of being recommended.

[0037] In one implementation of this application, the historical operation command sequence of the current user is obtained from a cloud-based CAD platform. Based on the historical operation command sequence corresponding to the current user, the individual user's next-stage operation command is determined for different combinations of design states and selected elements. Based on the frequency of the individual user's next-stage operation command corresponding to different combinations, the individual user's probability value for the individual user's next-stage operation command corresponding to different combinations is determined, and a preset individual probability transition matrix is ​​constructed based on the individual user probability value. The elements in the preset individual probability transition matrix represent the probability that the current individual user will execute different operation commands given a design state and selected elements.

[0038] Specifically, a preset personal probability transition matrix is ​​constructed based on a construction method similar to that of a preset general probability transition matrix. The personal probability transition matrix in this embodiment is calculated and generated based on a single user's recent operation records, reflecting the specific user's operating habits, and is denoted as... .in, Represents the current user. i.e., user In state Select element Execute operation command at time The probability of.

[0039] Step 103: Based on the predicted final probability values ​​of different operation commands for the next stage, construct a list of recommended commands suitable for the current user.

[0040] In one implementation of this application, weighting coefficients are determined based on the historical operation command data corresponding to the current user and the historical operation command data corresponding to a group of users to balance group experience and individual habits. A first probability value for the next-stage operation command corresponding to the current design state and the selected element is obtained from a preset general probability transition matrix; wherein the current design state and the selected element correspond to one or more next-stage operation commands. A second probability value for the next-stage operation command corresponding to the current design state and the selected element is obtained from a preset individual probability transition matrix. Based on the weighting coefficients, the first probability value and the second probability value corresponding to the same next-stage operation command are weighted and fused to obtain the final probability value for the next-stage prediction.

[0041] Specifically, this embodiment dynamically calculates weight coefficients based on the current user's proficiency to achieve an adaptive balance. The calculation process can be based on the user's historical operation data and group user data. Specifically, firstly, a user proficiency index is calculated, which is the ratio of the number of user operations to the average number of operations per user on the platform. Simultaneously, an operation stability index is calculated, evaluated by calculating the entropy of the user's recent operation sequence; the lower the entropy value, the more concentrated the command usage and the higher the stability. Finally, the weight coefficient is obtained by weighted fusion of user proficiency and operation stability. When a user is more proficient and has more stable operating habits, the weight value is larger, and the system relies more on their personal habits; conversely, for new users or users with unstable operations, the weight value is smaller, and the system relies more on general experience.

[0042] Furthermore, when predictions are needed for the current design state and selected element, two query operations are performed in parallel. First, a search is performed in the preset general probability transition matrix using the current design state and selected element as the joint key to obtain the first probability value corresponding to all possible next-stage operation commands. This value represents the behavioral tendency of the group of users under the same conditions. Simultaneously, the same query is performed in the preset individual probability transition matrix to obtain the second probability value corresponding to the current user.

[0043] Furthermore, after obtaining the first and second probability values ​​for the same command, they are weighted and fused using calculated dynamic weight coefficients to obtain the final probability value. After completing the weighted fusion calculation for all candidate commands, a list containing all commands and their final probability values ​​is obtained. Subsequently, the commands are sorted from highest to lowest according to their final probability values. After sorting, the system can apply subsequent filtering and optimization logic to ultimately generate a highly personalized list of recommended commands arranged in descending order of recommendation probability, which is then presented to the user.

[0044] Specifically, in this embodiment, the probability values ​​of the next-stage operation commands corresponding to the general probability transition matrix and the individual probability transition matrix are fused to obtain the final probability value of the next-stage prediction. The calculation formula is as follows:

[0045] ;

[0046] In the formula, Weighting coefficients (0≤ ≤1), used to balance the influence of general group operating experience and individual user operating habits; After integration, the user's status Select element Execute command The final probability.

[0047] In one implementation of this application, the predicted operation commands for the next stage are sorted based on their final probability values ​​to construct an initial recommended command list. A parametric association graph is constructed based on the current user's current operation data and historical operation data; the parametric association graph includes at least one of the following: parent-child relationship, dimension-driven relationship, and geometric constraint relationship between different operation commands. The parametric association graph is compared with a preset CAD domain knowledge rule base, and based on the comparison results, design conflicts that may occur after executing the recommended commands are predicted. In the historical case library, resolved cases with a similarity greater than a preset similarity threshold to the current design conflict scenario are retrieved, and the operation commands used in the resolved cases are identified as compensatory operation commands. The compensatory operation commands are bound to the recommended commands, and the initial recommended command list is optimized based on the bound recommended commands to obtain the final recommended command list.

[0048] Specifically, based on the final probability values ​​predicted in the next stage, all candidate operation commands are sorted from highest to lowest probability, and the top N commands are selected to form an initial recommended command list. A parametric relationship graph is dynamically constructed based on the CAD model currently being edited by the user. This graph uses the model's feature tree as its core data source and is automatically generated by parsing the dependencies between features. Nodes in the graph represent features such as Extrude 1, Chamfer 2, and Grass. Figure 3 Directed edges represent the relationships between features. Specifically, the following relationships are extracted:

[0049] Parent-child relationship: For example, stretch 1 feature depends on grass Figure 3 , then grass Figure 3 It is the parent feature of stretch 1, and there is a parent-child relationship edge between the two.

[0050] Dimension-driven relationship: For example, if the positioning dimension of hole 1 is driven by datum surface 1, then there is a dimension-driven edge between datum surface 1 and hole 1.

[0051] Geometric constraint relationships: For example, an instance of feature array 1 depends entirely on the geometry of the feature source feature, and there are geometric constraint edges between them.

[0052] Furthermore, the constructed parametric association diagram is compared with a predefined CAD domain knowledge rule base. This rule base includes different parametric association diagrams corresponding to multiple operation commands, as well as design conflicts corresponding to different parametric association diagrams. Each command in the initial recommendation list is then associated with the current parametric association. Figure 1 The data is then fed into the rules engine for matching. Once a rule is matched, it predicts that executing the recommended command may trigger a corresponding design conflict.

[0053] Furthermore, when a potential conflict is predicted for a recommended command, a historical case database stored in the cloud is retrieved. This database records numerous operation sequences that have successfully resolved similar design conflicts. During the retrieval, the current conflict scenario is converted into a feature vector, and its similarity is calculated with the conflict scenario vectors recorded in the case database. All resolved cases with a similarity greater than a preset threshold are selected, and the effective pre-operations taken by the user before executing the risky command in these cases are analyzed. These verified effective pre-operations are identified as compensatory operation commands in the current scenario. The retrieved compensatory commands are bound to the original risky command to form an operation strategy package. Subsequently, the initial recommendation list is optimized: the newly generated operation strategy package is inserted as an advanced option at the beginning of the recommendation list, with a visual label added for explanation. The final generated list of recommended commands considers both model integrity and operational safety.

[0054] In one implementation of this application, the sequence of operation commands generated by the current user for the current task is compared with a preset parameterized template. The preset parameterized template consists of historical operation command sequences that meet high-frequency conditions, and also includes parameter value rules corresponding to the historical operation command sequences. When the comparison result meets the triggering conditions of the preset parameterized template, a reference command sequence is determined from the preset parameterized template. Using the current design state and the real-time geometric attributes corresponding to the selected element as input, parameter values ​​are dynamically filled into each command in the reference command sequence to generate a command sequence instance adapted to the current task. The command sequence instance is simulated for execution. If the simulation execution result meets the preset execution conditions, the current user is provided with a one-click option to execute the instantiated command sequence. When the user triggers the option to execute the instantiated command sequence, the instantiated sequence is executed sequentially, and a parameter determination interface is provided to the user before each operation command is executed.

[0055] Specifically, the cloud server in this embodiment stores multiple parameterized templates. These templates are high-frequency operation sequences with clear engineering semantics, mined by analyzing historical data from different users. Each template not only contains an ordered list of commands but also defines the parameter value rules for each command. For example, for the drilling command, the rule might be defined as: Hole diameter D = diameter of the currently selected cylindrical surface * coefficient K1; Hole depth H = thickness of the currently selected plate + offset. When a user performs an operation, the user's current task context, including design state, type of selected element, and geometric attributes, is compared in real time with the trigger conditions of all templates. For example, the trigger conditions could be set as: selected element type is planar, current mode is non-sketch mode, and pre-selected hole features exist on the planar surface. If a match is successful, the trigger conditions of the template are deemed met.

[0056] After matching a suitable parametric template, its contained command sequence is identified as the reference command sequence. Then, the dynamic instantiation engine is activated. This engine takes the current design state and the real-time geometric properties of the selected element as input, and calculates specific parameter values ​​for each command in the sequence according to the predefined parameter rules in the template. For example, if the user selects a hole with a diameter of 5mm, the diameter and depth of the countersunk hole are automatically calculated according to the rules, thus generating a specific command sequence instance that perfectly adapts to the current model size [drilling (D=5mm), chamfering (D=9mm, H=3mm)]. This process transforms the template into a personalized, immediately executable operation plan.

[0057] Furthermore, before recommending the instance to the user, it is simulated. Each command in the sequence is run through the simulator, and its geometric feasibility is checked. Preset execution conditions may include: no geometric errors are generated during command execution, and the generated features meet basic engineering rationality. If the simulation execution successfully passes all checks, the instance is deemed feasible. At this point, a non-modal prompt box will pop up in the user interface, clearly displaying the brief steps and key parameters of the sequence, and providing a one-click execution button. When the user clicks one-click execution, the system begins to execute the instantiation sequence in sequence. Before executing each command, a parameter confirmation interface will be displayed. This interface displays all parameters automatically calculated by the system according to the template rules, and allows the user to make final confirmations or fine-tuning. After the user confirms, the command is executed, and then the system automatically proceeds to the confirmation interface for the next command.

[0058] Step 104: Based on the probability value of the operation command actually selected by the current user and the historical selection probability value corresponding to the selected operation command, dynamically update the preset personal probability transition matrix, and predict the next recommended command list based on the updated preset personal probability transition matrix to achieve dynamic optimization of interactive command recommendation.

[0059] In one implementation of this application, based on the chronological order of historical operation commands executed by the current user, a preset number of historical operation commands are selected. Based on these preset number of historical operation commands, the recent operation probability value for executing different next-stage operation commands under a given design state and selected element is determined. Based on a preset attenuation factor, the recent operation probability value is weighted and fused with the corresponding probability value in a preset personal probability transition matrix to determine the updated probability value. When no corresponding historical operation command exists for a given design state and selected element, the calculated recent operation frequency is used as the updated probability value. The data in the preset personal probability transition matrix is ​​replaced based on the updated probability value to dynamically update the preset personal probability transition matrix.

[0060] Specifically, in order to capture users' recent operating habits in real time, the personal probability transition matrix needs to be dynamically updated. The embodiment of this application uses the exponential moving average algorithm to achieve this.

[0061] First, operation logs and frequency calculation. Specifically, the system continuously records the user's recent... This operation behavior, based on the record, calculates the recent operation frequency of executing different next-stage operation commands in a specific scenario. ,in This represents the user's current state. To select an element, To execute the command, Indicates the state Select element Execute command The frequency.

[0062] Secondly, the matrix update formula is used. Specifically, the update formula for the individual probability transition matrix is ​​as follows:

[0063] ;

[0064] In the formula, The individual probabilities before the update. For the updated individual probabilities, The attenuation factor (0 < <1), the larger the value, the more the update result depends on historical probability data, and the smaller the value, the more it focuses on recent operation frequency.

[0065] Specifically, when a user uses the service for the first time or when the individual probability transition matrix is ​​empty, the calculated frequencies are directly applied. As the initial individual probability transition matrix.

[0066] In this embodiment, to eliminate useless or unusable commands in the current scenario, a pre-filter is applied to the recommendation results, retaining only those relevant to the current design state. Select element Matching available commands ensure that the recommendation list pushed to the user has practical operational significance.

[0067] Figure 3 This is a schematic diagram of an interactive command recommendation device for CAD software, provided as an embodiment of this application. Figure 3As shown, an interactive command recommendation device 200 for CAD software includes: at least one processor 201; and a memory 202 communicatively connected to at least one processor 201. The memory 202 stores instructions executable by at least one processor 201, which, when executed, enable the at least one processor 201 to: acquire the user's design state and selected elements in a cloud-based CAD design environment; wherein the design state includes at least one of sketch mode and non-sketch mode; based on the design state and selected elements, fuse the probability values ​​of the next-stage operation commands corresponding to a preset general probability transition matrix and a preset personal probability transition matrix, respectively, to obtain the final probability value of the predicted next-stage operation commands; construct a recommended command list suitable for the current user based on the predicted final probability values ​​of different next-stage operation commands; and dynamically update the preset personal probability transition matrix based on the probability value of the operation command actually selected by the current user and the historical selection probability value corresponding to the selected operation command, so as to predict the next recommended command list based on the updated preset personal probability transition matrix, thereby achieving dynamic optimization of interactive command recommendation.

[0068] This application provides a non-volatile computer storage medium storing computer-executable instructions. These instructions are configured to: acquire the user's design state and selected elements in a cloud-based CAD design environment; wherein the design state includes at least one of sketch mode and non-sketch mode; based on the design state and selected elements, fuse the probability values ​​of the next-stage operation commands corresponding to a preset general probability transition matrix and a preset personal probability transition matrix used to predict the next-stage operation commands to obtain the final probability value of the predicted next-stage operation commands; construct a recommended command list suitable for the current user based on the predicted final probability values ​​of different next-stage operation commands; and dynamically update the preset personal probability transition matrix based on the probability value of the operation commands actually selected by the current user and the historical selection probability values ​​corresponding to the selected operation commands, so as to predict the next recommended command list based on the updated preset personal probability transition matrix, thereby achieving dynamic optimization of interactive command recommendations.

[0069] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0070] The above descriptions are merely embodiments of this application and are not intended to limit the scope of this application. For those skilled in the art, various modifications and variations can be made to the embodiments of this application. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions in the embodiments of this application.

Claims

1. An interactive command recommendation method for CAD software, characterized in that, The method includes: Obtain the user's design status and selected elements in the cloud-based CAD design environment; wherein, the design status includes at least one of sketch mode and non-sketching mode; Based on the design state and the selected element, the probability values ​​of the next stage operation command corresponding to the preset general probability transition matrix and the preset personal probability transition matrix used to predict the next stage operation command are fused to obtain the final probability value of the predicted next stage operation command. Based on the predicted final probability values ​​of different operation commands in the next stage, a list of recommended commands suitable for the current user is constructed. Based on the probability value of the operation command currently selected by the user and the historical selection probability value corresponding to the selected operation command, the preset personal probability transition matrix is ​​dynamically updated. The next recommended command list is predicted based on the updated preset personal probability transition matrix, so as to realize dynamic optimization of interactive command recommendation.

2. The interactive command recommendation method for CAD software according to claim 1, characterized in that, Before fusing the probability values ​​of the next-stage operation commands corresponding to the preset general probability transition matrix and the preset individual probability transition matrix used to predict the next-stage operation command, the method further includes: The historical operation command sequence of a group of users is obtained from the cloud-based CAD platform; wherein the historical operation command sequence includes at least one of the following: timestamp, design status, selected element, and the last executed command; Based on the historical operation command sequence of group users, the general next-stage operation command corresponding to different design states and selected element combinations is determined. Based on the frequency of the general next-stage operation commands corresponding to different combinations, the general probability values ​​of the next-stage operation commands corresponding to different combinations are determined. Based on the combination information and the general probability values ​​of the next-stage operation commands corresponding to the combination information, the preset general probability transition matrix is ​​constructed.

3. The interactive command recommendation method for CAD software according to claim 2, characterized in that, After constructing the preset general probability transition matrix based on the combined information and the general probability value of the next-stage operation command corresponding to the combined information, the method further includes: The general probability value is compared with the preset probability threshold; Delete general probability values ​​that are less than the preset probability threshold; The preset probability threshold is used to avoid the general probability value being 0, so as to ensure that all commands have a basic probability of being recommended.

4. The interactive command recommendation method for CAD software according to claim 1, characterized in that, Before fusing the probability values ​​of the next-stage operation commands corresponding to the preset general probability transition matrix and the preset individual probability transition matrix used to predict the next-stage operation command, the method further includes: Obtain the current user's historical command sequence from the cloud-based CAD platform; Based on the current user's historical operation command sequence, different design states and selected element combinations are determined, and the corresponding next stage operation commands for each user are identified. Based on the frequency of the next-stage operation commands for each individual corresponding to different combinations, the individual probability value of the next-stage operation commands for each individual corresponding to different combinations is determined, and the preset individual probability transition matrix is ​​constructed based on the individual probability value. The elements in the preset personal probability transition matrix represent the probability that the current personal user will execute different operation commands given a design state and selected elements.

5. The interactive command recommendation method for CAD software according to claim 1, characterized in that, Based on the design state and the selected element, the probability values ​​of the next-stage operation command corresponding to the preset general probability transition matrix and the preset individual probability transition matrix used to predict the next-stage operation command are fused to obtain the final probability value of the predicted next-stage operation command. Specifically, this includes: Based on the historical operation command data of the current user and the historical operation command data of the group users, a weighting coefficient is determined to balance the group experience and individual habits. In the preset general probability transition matrix, obtain the first probability value of the next stage operation command corresponding to the current design state and the selected element; wherein, the current design state and the selected element correspond to one or more next stage operation commands; In the preset personal probability transition matrix, obtain the second probability value of the next stage operation command corresponding to the current design state and the selected element; Based on the weighting coefficients, the first probability value and the second probability value corresponding to the same next-stage operation command are weighted and fused to obtain the final probability value of the next-stage prediction.

6. The interactive command recommendation method for CAD software according to claim 1, characterized in that, Based on the predicted final probability values ​​of different operation commands in the next stage, a recommended command list suitable for the current user is constructed, specifically including: Based on the final probability values ​​corresponding to different operation commands in the next stage, the predicted different operation commands in the next stage are sorted to construct an initial recommended command list; Based on the current operation data and historical operation data of the current user, a parametric association graph is constructed; wherein, the parametric association graph includes at least one of the following: parent-child relationship, size-driven relationship, and geometric constraint relationship between different operation commands; The parametric association diagram is compared with a preset CAD domain knowledge rule base. Based on the comparison results, the design conflicts that may occur after executing the recommended command are predicted. In the historical case database, retrieve resolved cases that have a similarity to the current design conflict scenario greater than a preset similarity threshold, and determine the operation commands used in the resolved cases as compensatory operation commands; The compensatory operation command is bound to the recommended command, and the initial recommended command list is optimized based on the bound recommended command to obtain the final recommended command list.

7. The interactive command recommendation method for CAD software according to claim 1, characterized in that, After constructing the recommended command list suitable for the current user, the method further includes: The sequence of operation commands generated by the current user for the current task is compared with a preset parameterized template; wherein, the preset parameterized template consists of historical operation command sequences that meet high-frequency conditions, and also includes parameter value rules corresponding to the historical operation command sequences; When the comparison result meets the triggering condition of the preset parameterized template, a reference command sequence is determined from the preset parameterized template; Using the current design state and the real-time geometric properties corresponding to the selected element as input, parameter values ​​are dynamically filled into each command in the reference command sequence to generate a command sequence instance that is adapted to the current task. The command sequence instance is simulated and executed. If the simulation execution result meets the preset execution conditions, the current user is provided with the option to execute the instantiated command sequence with one click. When the user triggers the option to execute the instantiated command sequence, the instantiated sequence is executed in sequence, and a parameter confirmation interface is provided to the user before each operation command is executed.

8. The interactive command recommendation method for CAD software according to claim 1, characterized in that, The dynamic updating of the preset personal probability transition matrix based on the probability value of the currently selected operation command and the historical selection probability value corresponding to the selected operation command specifically includes: Based on the chronological order of the historical operation commands executed by the current user, a preset number of the historical operation commands are selected, and based on the preset number of the historical operation commands, the probability value of executing different next-stage operation commands under a given design state and selected element is determined. Based on a preset decay factor, the recent operation probability value is weighted and fused with the corresponding probability value in the preset personal probability transition matrix to determine the updated probability value. When there is no corresponding historical operation command for a given design state and selected element, the calculated recent operation frequency is used as the updated probability value. The data in the preset personal probability transition matrix is ​​replaced based on the updated probability values ​​to dynamically update the preset personal probability transition matrix.

9. An interactive command recommendation device for CAD software, characterized in that, The device includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to perform the method described in any one of claims 1-8.

10. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions are capable of performing the method described in any one of claims 1-8.

Images