An intelligent personnel training data information management system

Abstract

This invention relates to the field of training data management technology, specifically to an intelligent personnel training data information management system. The system includes a training behavior constraint modeling module that acquires training task definition information, training equipment status information, and training process rule information; constructs a training behavior constraint model; and generates a training behavior state network. A training data authenticity deduction module receives multi-source training data and maps it to the training behavior state network, generating behavioral reachable sequences and training behavior trajectory sets by tracing state transition paths. A training behavior gap deduction module performs path completion deduction for discontinuous state transitions based on behavioral dependencies, generating a completed training trajectory set and a training behavior completion relationship set. This invention achieves structured organization, continuous restoration, and gap repair of training data, thereby improving the integrity, interpretability, and traceability of training data management.

Description

Technical Field

[0001] This invention relates to the field of training data management technology, and in particular to an intelligent personnel training data information management system. Background Technology

[0002] With the development of information technology and data management technology, the personnel training process is gradually shifting from traditional manual recording methods to data-driven and systematic management. In actual training, training tasks typically involve multiple training steps, equipment operations, and process rules, generating a large amount of training data during execution. This training data comes from diverse sources, including training operation records, equipment operating status information, and training result data. To effectively manage and analyze the training process, a data management system is needed to uniformly collect, store, and organize training data, thereby achieving the recording and management of training activities and improving the level of data-driven management of the training process.

[0003] However, most existing training data management systems focus on simple storage, retrieval, or statistical analysis of training data, lacking a systematic description of the logical relationships and state evolution processes between training behaviors. In actual training, due to asynchronous data acquisition from multiple sources, inconsistent recording granularity, or missing data, discontinuities in training behavior records can easily occur, making it difficult to fully reconstruct the training behavior evolution path. Furthermore, existing technologies often struggle to identify unrecorded intermediate behavioral states during training and to systematically deduce the relationships between training data, thus affecting the completeness of training process analysis and data utilization efficiency. Summary of the Invention

[0004] This invention provides an intelligent personnel training data information management system. By constructing a training behavior constraint model and generating a training behavior state network, various training behaviors during the training process are expressed in a structured manner. Based on this, multi-source training data is mapped to the training behavior state network and its state transition paths are tracked to generate behavior reachable sequences and training behavior trajectory sets. Furthermore, path completion derivation is performed on discontinuous state transitions in the behavior reachable sequences to identify unrecorded intermediate training behavior states, thereby forming a complete training trajectory set and a training behavior completion relationship set, so as to achieve a complete expression of the training behavior process and accurate management of training data relationships.

[0005] An intelligent personnel training data information management system includes a training behavior constraint modeling module, a training data authenticity deduction module, and a training behavior gap deduction module, wherein; The training behavior constraint modeling module acquires training task definition information, training device status information, and training process rule information, and constructs a training behavior constraint model based on the operation sequence, resource call relationship, and result generation conditions of the training task. The training behavior constraint model is used to describe the allowed sequential relationship, dependency relationship, and state change conditions between various training behaviors during the training process, thereby generating a training behavior state network that represents the reachable path of training behavior. The training data authenticity inference module receives multi-source training data generated during the training process and maps the multi-source training data to the training behavior state network. By tracking the state transition path of the multi-source training data in the training behavior state network, it generates the corresponding behavior reachable sequence and the training behavior trajectory set representing the training behavior evolution trajectory. The training behavior gap derivation module receives the behavior reachable sequence and the training behavior trajectory set. Based on the behavior dependencies defined in the training behavior state network, it performs path completion derivation on the discontinuous state transitions that occur in the behavior reachable sequence to identify intermediate training behavior states that are not recorded in the training behavior trajectory. This generates a complete training trajectory set including the derivation of behavior states and a training behavior completion relation set describing the relationship between the derivation of behavior and multi-source training data.

[0006] Optionally, the training behavior constraint modeling module includes: Construction of training behavior constraint information: Obtain training task definition information, training device status information, and training process rule information. Analyze the task step sequence, resource call relationship, and result generation conditions in the training task definition information. Combined with the device availability status and operation triggering conditions reflected in the training device status information, the training process rule information is expressed in a unified structure to form a set of training behavior constraints describing the triggering conditions, execution order, and resource dependencies of each training behavior, thereby generating the training behavior constraint model. Training Behavior State Network Generation: Based on the training behavior triggering conditions, behavior dependencies, and state change rules defined in the training behavior constraint model, the reachable paths between each training behavior are structured and organized. The training behavior is abstracted into state nodes, and the state transition relationship between training behaviors is abstracted into path edges, thereby constructing a training behavior state network that represents the reachability relationship between training behaviors at different execution stages.

[0007] Optionally, the construction of the training behavior constraint information includes: Multi-source constraint information acquisition and standardized parsing: Acquire training task definition information, training device status information, and training process rule information, and perform field extraction, format standardization, and time sequence alignment processing respectively to form a basic constraint information set. Specifically, the training task definition information is extracted to obtain the task step sequence, step input items, step output items, and step completion identifier; the training device status information is extracted to obtain the device availability status, device functional status, and operation trigger status; and the training process rule information is extracted to obtain the process preconditions, process transition conditions, and process termination conditions, thereby forming the corresponding structured information of the task steps. Structured information on equipment status and structured information of process rules ; Training behavior triggering and dependency generation: Based on the structured information of task steps, device status, and process rules, the triggering conditions, execution order, and resource dependencies of each training behavior are identified. The order of task steps in the training task definition information is used as the basis for behavior sequence, the available device status and operation triggering conditions in the training device status information are used as behavior triggering constraints, and the process preconditions and transition conditions in the training process rules information are used as behavior dependency constraints, thereby forming a set of training behavior constraints describing the relationship between each training behavior. Training behavior constraint model construction: Based on the set of training behavior constraints, each training behavior is abstracted into a behavior node, and the behavior connection relationships that satisfy the triggering conditions and have dependencies are abstracted into behavior transition relationships. The sequential relationships, dependencies, and state change conditions among each behavior node are structured and organized to construct a training behavior constraint model that describes the execution constraint relationships of each training behavior during the training process. .

[0008] Optionally, the generation of training behavior trigger relationships and dependency relationships includes: Trigger condition determination: For any training action The trigger condition judgment value for each training behavior is calculated based on the task step sequence, device status information, and process rule information. When satisfied At that time, determine the training behavior The triggering conditions are met, among which, For the first The trigger threshold for each training action; Dependency determination: Calculate any two training actions and Behavioral dependency strength To characterize the dependencies between training behaviors; Construction of the training behavior constraint set: Based on the determination results of the training behavior trigger conditions and the dependency strength between training behaviors, a training behavior constraint set is constructed. , is represented as: .

[0009] Optionally, the training behavior state network generation includes: Behavior node mapping: Based on the set of training behaviors in the training behavior constraint model, each training behavior is mapped to a corresponding state node, and each state node is associated with the corresponding training behavior triggering condition and state attribute. Path edge generation: Based on the training behavior triggering conditions, behavior dependencies and state change rules defined in the training behavior constraint model, determine whether there is a state transition relationship between different state nodes, and abstract the state transition relationship that meets the triggering conditions into a path edge. Training Behavior State Network Construction: Based on the set of state nodes and the set of path edges, the reachable paths between each training behavior are structured to form a training behavior state network that represents the reachability relationships between training behaviors at different execution stages. .

[0010] Optionally, the training data authenticity inference module includes: Multi-source training data reception and state node mapping: This process receives multi-source training data generated during training and, based on the training behavior identifier, time identifier, device operation identifier, and result identifier within the multi-source training data, maps each piece of multi-source training data to the corresponding state node in the training behavior state network, forming a sequence of observation nodes corresponding to the multi-source training data. ; State transition path tracing and behavior reachability sequence generation: Based on the connectivity of the observation node sequences in the trained behavior-state network, the state transition paths between adjacent observation nodes are traced, and state transition sequences that satisfy the network reachability relationship are extracted to generate the corresponding behavior reachability sequences. ; Construction of training behavior evolution trajectory: Temporal association and path organization are performed on the state nodes and their corresponding state transition relationships in the behavior reachable sequence, ensuring continuous state transitions (when training behavior...). Trigger condition judgment value Meets the trigger threshold And training behavior With training behavior Behavioral dependence strength Not less than the strength of behavioral dependence When determining the state node With state nodes The combination of state nodes (where state transition conditions are met, thus establishing a state transition relationship) forms the training behavior trajectory. The trajectory intensity is calculated based on the path edge weights between state nodes in the training behavior trajectory. This results in a set of training behavior trajectories that includes multiple training behavior trajectories and their intensity. It is used to characterize the state evolution path and path strength characteristics of training behavior during the execution process.

[0011] Optionally, the state transition path tracing and behavior reachability sequence generation includes: Reachability determination of adjacent observation nodes: Based on the connection relationship of the observation node sequence in the training behavior state network, determine whether there is a reachable path between adjacent observation nodes; Optimal state transition path calculation: For neighboring observation nodes with reachable paths, calculate their optimal state transition path in the trained behavioral state network. ; Behavioral reachability sequence generation: Based on the optimal state transition path between all adjacent observation nodes, the state nodes in the path are concatenated in chronological order to generate a behavioral reachability sequence. .

[0012] Optionally, the training behavior gap derivation module includes: Discontinuous state transition identification: Receive the action reachable sequence and the training action trajectory set, and compare the adjacent state nodes in the action reachable sequence with the action dependencies in the training action state network to identify discontinuous state transition segments that do not meet the direct state transition conditions. For each discontinuous state transition segment, combine the reachable path information of the corresponding training action trajectory in the training action state network to determine intermediate candidate state nodes that exist between the starting state node and the target state node but are not explicitly recorded by the current action reachable sequence, thereby forming a set of state segments to be completed. Training trajectory completion generation: Receive the set of state fragments to be completed, and based on the candidate reachable paths between the starting state node and the target state node in the training behavior state network, perform path completion deduction on the intermediate candidate state nodes. Insert the deduced intermediate training behavior states into the corresponding behavior reachable sequences and training behavior trajectories to generate a complete training trajectory set including the deduced behavior states. Simultaneously, based on the temporal adjacency, path continuity, and behavior dependency relationships between the inserted deduced behavior states and the original multi-source training data, generate a training behavior completion relationship set describing the association between the deduced behavior and the multi-source training data.

[0013] Optionally, the discontinuous state transition identification includes: Direct transition determination between adjacent state nodes: Receive the reachable sequence of behaviors and the training set of behavior trajectories, and calculate the transition between any adjacent state nodes in the reachable sequence. and Direct transfer decision value When satisfied When determining the state node With state nodes Form candidate discontinuous state transition segments; Discontinuous state transition segment screening: Screen the candidate discontinuous state transition segments by combining the reachable path information in the training behavior trajectory set to determine whether there is an intermediate state path that is reachable but not explicitly recorded between the start state node and the target state node. For the candidate discontinuous state transition segments , calculate the multi-node reachable path determination value . When the following conditions are met and , determine that is a discontinuous state transition segment and form a set of discontinuous state transition segments ; Generation of the set of state segments to be completed: For each discontinuous state transition segment, extract the intermediate candidate state nodes in the reachable path between its start state node and the target state node, and form a set of state segments to be completed .

[0014] Optionally, the generation of the completed training trajectory includes: Screening of candidate reachable paths and determination of intermediate states: Receive the set of state segments to be completed, and for each state segment to be completed, based on the candidate reachable paths between the start state node and the target state node in the training behavior state network, screen out the target reachable paths for path completion derivation and determine the corresponding intermediate training behavior states; Behavior reachable sequence and training behavior trajectory completion: Insert the determined intermediate training behavior states into the corresponding behavior reachable sequences and training behavior trajectories to generate a set of completed training trajectories including the derived behavior states ; Generation of the training behavior completion relationship set: Based on the time adjacency relationship, path succession relationship, and behavior dependence relationship between the inserted derived behavior states and the original multi-source training data, generate a training behavior completion relationship set that describes the association relationship between the derived behavior and the multi-source training data.

[0015] Advantages of the present invention: In the present invention, by constructing a training behavior constraint model and further generating a training behavior state network, the training task definition information, training device state information, and training process rule information are unifiedly modeled, and the originally scattered step sequences, device conditions, and process rules are transformed into training behavior nodes, behavior transfer relationships, and state change conditions that can be structurally expressed, so as to uniformly describe the behavior sequence logic, dependence logic, and state evolution logic in the training process.

[0016] This invention maps multi-source training data to a training behavior state network and generates behavior reachable sequences and training behavior trajectory sets based on state transition path tracking. This can unify multi-source training data with different sources, formats, and recording granularities during the training process into a continuously analyzable behavior evolution path. It can not only reflect the actual progress order of training behavior during execution, but also use trajectory intensity to characterize the path intensity features of the training behavior evolution path.

[0017] This invention identifies discontinuous state transitions in the reachable sequence of behaviors and, in conjunction with the derivation of the execution path completion of the training behavior state network, generates a complete training trajectory set and a training behavior completion relation set. It can identify intermediate training behavior states that were not explicitly collected during training due to missing records, data breaks, or discontinuous cross-source records, and on this basis, reconstruct a more complete training behavior trajectory. At the same time, it establishes the correlation between the derivation of behavior states and the original multi-source training data. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention 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 for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the system functional modules according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the training behavior constraint modeling module in an embodiment of the present invention. Detailed Implementation

[0020] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. Those skilled in the art may employ other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0021] like Figures 1-2 As shown, an intelligent personnel training data information management system includes a training behavior constraint modeling module, a training data authenticity deduction module, and a training behavior gap deduction module, wherein; The training behavior constraint modeling module obtains training task definition information, training device status information, and training process rule information, and constructs a training behavior constraint model based on the operation sequence, resource call relationship, and result generation conditions of the training task. The training behavior constraint model is used to describe the allowed sequence, dependency relationship, and state change conditions between various training behaviors during the training process, thereby generating a training behavior state network that represents the reachable path of training behavior. The training data authenticity inference module receives multi-source training data generated during the training process and maps the multi-source training data to the training behavior state network. By tracking the state transition path of the multi-source training data in the training behavior state network, it generates the corresponding behavior reachable sequence and the training behavior trajectory set representing the training behavior evolution trajectory. The training behavior gap inference module receives the behavior reachable sequence and the training behavior trajectory set. Based on the behavior dependencies defined in the training behavior state network, it performs path completion inference on the discontinuous state transitions that occur in the behavior reachable sequence to identify the intermediate training behavior states that are not recorded in the training behavior trajectory. This generates a complete training trajectory set that includes the inferred behavior states and a training behavior completion relation set that describes the relationship between the inferred behavior and the multi-source training data.

[0022] The training behavior constraint modeling module includes: Construction of training behavior constraint information: Obtain training task definition information, training device status information, and training process rule information. Analyze the task step sequence, resource call relationship, and result generation conditions in the training task definition information. Combine the device availability status and operation triggering conditions reflected in the training device status information to express the training process rule information in a unified structure. This forms a set of training behavior constraints that describes the triggering conditions, execution order, and resource dependencies of each training behavior, thereby generating a training behavior constraint model. Training Behavior State Network Generation: Based on the training behavior triggering conditions, behavior dependencies, and state change rules defined in the training behavior constraint model, the reachable paths between each training behavior are structured and organized. The training behavior is abstracted into state nodes, and the state transition relationship between training behaviors is abstracted into path edges, thereby constructing a training behavior state network that represents the reachability relationship between training behaviors at different execution stages.

[0023] The construction of training behavior constraint information includes: Multi-source constraint information acquisition and standardized parsing: Acquire training task definition information, training device status information, and training process rule information, and perform field extraction, format standardization, and time sequence alignment processing respectively to form a basic constraint information set. Specifically, the training task definition information is extracted to obtain the task step sequence, step input items, step output items, and step completion identifier; the training device status information is extracted to obtain the device availability status, device functional status, and operation trigger status; and the training process rule information is extracted to obtain the process preconditions, process transition conditions, and process termination conditions, thereby forming the corresponding structured information of the task steps. Structured information on equipment status and structured information of process rules Specifically, it includes: (1) Obtain training task definition information Training equipment status information and training process rules information The fields are then extracted to form the corresponding original task step structured information. Original equipment status structured information and original process rules structured information , is represented as: ; ; ; in, This is a function for extracting task step fields, used to extract the task step sequence, step input items, step output items, and step completion identifier. This is a function for extracting device status fields, used to extract device availability status, device functional status, and operation trigger status. This is a function for extracting process rule fields, used to extract process preconditions, process transition conditions, and process termination conditions. (2) Structured information of the original task steps Original equipment status structured information and original process rules structured information The format is standardized and represented as follows: ; in, For the original structured information set, , It is a structured collection of information with a standardized format. , This is a structured information representation of task steps after standardization. This is a structured information representation of device status after standardization. This is structured information for standardized process rules. , These are the mean and standard deviation of the original structured information set, respectively. (3) Perform time-series alignment processing on the structured information after format unification to form a basic constraint information set. ; Training behavior triggering and dependency generation: Based on the structured information of task steps, device status, and process rules, the triggering conditions, execution order, and resource dependencies of each training behavior are identified. The order of task steps in the training task definition information is used as the basis for behavior sequence, the available device status and operation triggering conditions in the training device status information are used as behavior triggering constraints, and the process preconditions and transition conditions in the training process rules information are used as behavior dependency constraints, thereby forming a set of training behavior constraints that describe the relationship between each training behavior. Training Behavior Constraint Model Construction: Based on the set of training behavior constraints, each training behavior is abstracted into a behavior node, and the behavior connection relationships that satisfy triggering conditions and have dependencies are abstracted into behavior transition relationships. The sequential relationships, dependencies, and state change conditions among each behavior node are structured and organized to construct a training behavior constraint model that describes the execution constraint relationships of each training behavior during the training process. , is represented as: ; in, To train the set of behavior nodes, For the set of behavior transfer relations, This is a set of constraint parameters used to store the triggering conditions of each behavior node, the dependency strength of each behavior transition relationship, and the corresponding state change conditions.

[0024] Training behavior trigger relationships and dependency generation include: Trigger condition determination: For any training action The trigger condition judgment value for each training behavior is calculated based on the task step sequence, device status information, and process rule information. When satisfied At that time, determine the training behavior The triggering conditions are met, among which, For the first The trigger threshold for each training action is expressed as: ; in, For the first One training behavior, For the first The trigger condition judgment value for each training behavior. The sequence satisfaction degree, derived from the order of task steps, is used to characterize whether the preceding steps of the current training action have been completed. The device executability, derived from the training device status information, characterizes whether the device supporting the training behavior is currently in a usable state. The rule matching degree, obtained from the training process rule information, is used to characterize whether the training behavior satisfies the preconditions and transition conditions of the process. , , The task step sequence, equipment status, and process rules are respectively in the [number]th [section]. Weight coefficients in the determination of training behavior triggers; ; in, For the first The total number of preceding steps corresponding to each training action. For the first The training action corresponding to the first A completion marker for each preceding step; when a preceding step is completed and the completion order meets the requirements. ,otherwise, , For the first The training action corresponding to the first The importance weight of each preceding step; ; in, To support the first The total number of devices required for each training action For the first The training action corresponding to the first The availability status value of each device is used to characterize whether the device is online, available, and fault-free. For the first The training action corresponding to the first The functional status value of each device is used to characterize whether the device has the functions required to perform the current training action. For the first The training action corresponding to the first The operation trigger status value of each device is used to characterize whether the device meets the operation conditions for triggering the current training behavior. For the first The training action corresponding to the first The importance weight of each device , , These represent the weighting coefficients of device availability status, device functional status, and operation triggering status in the calculation of device executability, respectively. ; ; ; ; in, For the first The degree to which each training behavior satisfies the preconditions of the process. For the first The degree to which each training behavior satisfies the process transition conditions. For the first The degree to which each training behavior satisfies the process termination condition. , , These represent the weighting coefficients of process preconditions, process transition conditions, and process termination conditions in the rule matching degree calculation. For the first The number of precondition sub-rules corresponding to each training action. For the first The training action corresponding to the first If a precondition sub-rule is satisfied, a flag is set; if satisfied, then... Otherwise take , For the first The weight of each precondition sub-rule For the first The number of transition condition sub-rules corresponding to each training behavior. For the first The training action corresponding to the first Each transition condition sub-rule is marked with a value of 1 if satisfied, and 0 otherwise. For the first The weight of each transition condition sub-rule For the first The number of termination condition sub-rules corresponding to each training action. For the first The training action corresponding to the first Each termination condition sub-rule is marked with a value of 1 if it is satisfied, and 0 otherwise. For the first The weight of each termination condition sub-rule; ; in, This represents the maximum number of preceding steps across all training actions. This represents the maximum total number of devices required across all training actions. The total number of rule sub-items in all training behaviors The maximum value, , , These represent the weight coefficients of the complexity of the preceding steps, the complexity of the device dependency, and the complexity of the rule in the calculation of the trigger threshold, respectively. Dependency determination: Calculate any two training actions and Behavioral dependency strength To characterize the dependencies between training behaviors, it is represented as: ; in, The step sequence correlation is used to characterize training behavior. Is this a training behavior? The preceding behavior, Resource call relevance is used to characterize training behavior. With training behavior Does the device, resource, or resource transfer relationship exist? Generate correlations for the results to characterize training behavior. Does the output constitute training behavior? The triggering basis , , These are the weight coefficients of step sequence correlation, resource call correlation, and result generation correlation in dependency calculation; ; in, For training behavior Position number in the sequence of training task steps For training behavior Position number in the sequence of training task steps; ; in, For training behavior The set of resources invoked For training behavior The set of resources invoked For training behavior The set of output resources, For training behavior The set of input resources, For training behavior With training behavior The number of shared resources, For training behavior Output resources and training behavior The amount of overlap between input resources, For training behavior With training behavior The number of resources in the union, This is the enhancement coefficient for resource transfer relationships; ; in, For training behavior With training behavior The results between them generate a correlation. For training behavior The set of output elements, For training behavior The set of triggering condition elements, For training behavior The output results can be used as training behavior. The number of elements that trigger the event. For training behavior The total number of triggering condition elements; Construction of the training behavior constraint set: Based on the determination results of the training behavior trigger conditions and the dependency strength between training behaviors, a training behavior constraint set is constructed. , is represented as: .

[0025] Training the behavioral state network generation includes: Behavior Node Mapping: Based on the set of training behaviors in the training behavior constraint model, each training behavior is mapped to a corresponding state node, and each state node is associated with the corresponding training behavior triggering condition and state attribute, represented as follows: ; ; in, To train the set of state nodes in the behavioral state network, For the first Training behavior The mapped state nodes, , To train the total number of behaviors, State Node Node attributes, For the first The state attribute identifier corresponding to each training behavior is used to characterize the execution stage or state category of the training behavior; Path edge generation: Based on the training behavior triggering conditions, behavior dependencies, and state change rules defined in the training behavior constraint model, determine whether there are state transition relationships between different state nodes, and abstract the state transition relationships that meet the triggering conditions as path edges, represented as follows: ; Among them, Generate a marker for the path edge between state nodes to state node When it indicates that there is a path edge between state node and state node When it indicates that there is no path edge between state node and state node Is the trigger condition judgment value of the training behavior Is the trigger threshold of the training behavior Is the behavior dependence intensity threshold; ; Among them, , Are respectively the average value and standard deviation of the behavior dependence intensity between historical training behavior pairs, Is the threshold adjustment coefficient; ; Among them, Is the set of path edges in the training behavior state network, Is the path edge between state node to state node ; ; Among them, Is the edge weight value of the path edge between state node and state node , , Are respectively the weight coefficients of the behavior dependence intensity and the degree of satisfaction of the trigger condition in the calculation of the path edge weight value; Training behavior state network construction: According to the set of state nodes and the set of path edges, structurally organize the reachable paths between each training behavior to form a training behavior state network representing the reachable relationship between different execution stages of the training behavior , expressed as: ; Among them, Is the set of edge weight values corresponding to each path edge; ; ; Among them, Is the adjacency matrix of the training behavior state network, Is the state node ​​​To the state node The adjacency value, when When, it indicates a state node. To the state node There exists a state transition relationship, and its transition strength is ,when When, it indicates a state node. To the state node There is no state transition relationship between them.

[0026] The training data authenticity deduction module includes: Multi-source training data reception and state node mapping: This process receives multi-source training data generated during training and, based on the training behavior identifier, time identifier, device operation identifier, and result identifier within the multi-source training data, maps each piece of multi-source training data to the corresponding state node in the training behavior state network, forming a sequence of observation nodes corresponding to the multi-source training data. Specifically, it includes: (1) Let the set of multi-source training data generated during the training process be . ,in, For the first Multiple training data sources , This represents the total number of training data entries from multiple sources. (2) For any multi-source training data Its relationship with state nodes The mapping matching degree is expressed as: ; in, For the first Multi-source training data With the Status nodes mapping matching degree, To train the behavior label matching degree, For time identifier matching degree, For equipment operation identifier matching degree, To indicate the degree of matching for the results, , , , These represent the weight coefficients of training behavior identifier, time identifier, device operation identifier, and result identifier in the mapping matching degree calculation, respectively. (3) The first Each set of multi-source training data is mapped to the state node with the highest matching degree, forming an observation node sequence. , represented as: ; ; State transition path tracing and behavior reachability sequence generation: Based on the connectivity of the observation node sequences in the trained behavior-state network, the state transition paths between adjacent observation nodes are traced, and state transition sequences that satisfy the network reachability relationship are extracted to generate the corresponding behavior reachability sequences. ; Construction of training behavior evolution trajectory: Temporal association and path organization are performed on the state nodes and their corresponding state transition relationships in the behavior reachable sequence, ensuring continuous state transitions (when training behavior...). Trigger condition judgment value Meets the trigger threshold And training behavior With training behavior Behavioral dependence strength Not less than the strength of behavioral dependence When determining the state node With state nodes The combination of state nodes (where state transition conditions are met, thus establishing a state transition relationship) forms the training behavior trajectory. The trajectory intensity is calculated based on the path edge weights between state nodes in the training behavior trajectory. This results in a set of training behavior trajectories that includes multiple training behavior trajectories and their intensity. , used to characterize the state evolution path and path strength features of training behavior during execution, is expressed as: ; ; ; in, For the first The first training behavior trajectory One state node, , For the first The number of state nodes in each training behavior trajectory For the first The first training behavior trajectory The state node to the _th The edge weights of the path edges between state nodes. The total number of training behavior trajectories.

[0027] State transition path tracing and behavior reachability sequence generation include: Reachability determination between adjacent observation nodes: Based on the connectivity of the observation node sequence in the training behavior state network, determine whether there is a reachable path between adjacent observation nodes, expressed as: ; in, State Node To the state node The path reachability determination value, when When, it indicates that there exists a transition from the training behavioral state network. arrive The reachable path, when When this condition is met, it indicates that there is no reachable path in the training behavioral state network. Indicates from the state node To the state node State transition path; Optimal state transition path calculation: For neighboring observation nodes with reachable paths, calculate their optimal state transition path in the trained behavioral state network. Specifically, it includes: (1) Let the weight of the path edge be 1. Then from the state node To the state node The path transition cost is defined as: ; in, For the state node To the state node Path transfer cost, For the state node To the state node The set of candidate paths, This is a path edge in the path; (2) By minimizing path transition cost Determine the optimal state transition path ; Behavioral reachability sequence generation: Based on the optimal state transition path between all adjacent observation nodes, the state nodes in the path are concatenated in chronological order to generate a behavioral reachability sequence. , is represented as: ; in, For the first in the action reachable sequence There are several state nodes.

[0028] The training behavior gap derivation module includes: Discontinuous state transition identification: Receive the action reachable sequence and the training action trajectory set, and compare the adjacent state nodes in the action reachable sequence with the action dependencies in the training action state network to identify discontinuous state transition segments that do not meet the direct state transition conditions. For each discontinuous state transition segment, combine the reachable path information of the corresponding training action trajectory in the training action state network to determine intermediate candidate state nodes that exist between the starting state node and the target state node but are not explicitly recorded by the current action reachable sequence, thereby forming a set of state segments to be completed. Training trajectory completion generation: Receive the set of state fragments to be completed, and based on the candidate reachable paths between the starting state node and the target state node in the training behavior state network, perform path completion deduction on the intermediate candidate state nodes. Insert the deduced intermediate training behavior states into the corresponding behavior reachable sequences and training behavior trajectories to generate a complete training trajectory set including the deduced behavior states. At the same time, based on the temporal adjacency, path continuity, and behavior dependency relationships between the inserted deduced behavior states and the original multi-source training data, generate a training behavior completion relationship set describing the association between the deduced behavior and the multi-source training data.

[0029] Discontinuous state transition identification includes: Direct transition determination between adjacent state nodes: Receive the reachable sequence of behaviors and the training set of behavior trajectories, and calculate the transition between any adjacent state nodes in the reachable sequence. and Direct transfer decision value When satisfied When determining the state node With state nodes The candidate discontinuous state transition segment is represented as: ; in, State Node To the state node Candidate path edges, when When, it indicates a state node. With state nodes There is a direct state transition relationship between them, when When, it indicates a state node. With state nodes There is no direct state transition relationship between them; Discontinuous state transition segment selection: Candidate discontinuous state transition segments are selected by combining reachable path information from the training behavior trajectory set to determine whether there is a reachable but unrecorded intermediate state path between the starting state node and the target state node. For candidate discontinuous state transition segments... Calculate the reachable path determination value for multiple nodes. When satisfied and At that time, the judgment These are discontinuous state transition segments, forming a set of discontinuous state transition segments. , is represented as: ; ; in, For the state node To the state node reachable path, The total number of state nodes contained in the reachable path, when When, it indicates that there exists a slave state node in the training behavioral state network. To the state node A reachable path, and that reachable path contains at least one intermediate state node, when When this condition is met, it indicates that there is no reachable path that satisfies the condition. Generation of the set of state fragments to be completed: For each discontinuous state transition fragment, extract the intermediate candidate state nodes in the reachable path between its starting state node and the target state node, and form the set of state fragments to be completed. Specifically, it includes: (1) For any discontinuous state transition segment Let the corresponding optimal reachable path be: ; in, For the state node To the state node The optimal reachable path, For the th in the optimal reachable path There are 10 intermediate candidate state nodes. This indicates the number of intermediate candidate state nodes; (2) Obtain the optimal reachable path by minimizing the path transition cost: ; in, For the state node To the state node Candidate reachable paths, Let be a path edge among the candidate reachable paths. For path edges The edge weights; (3) Extract the intermediate candidate state nodes as: ; in, Discontinuous state transition segment The corresponding set of intermediate candidate state nodes; (4) The final set of state fragments to be completed is formed: .

[0030] Training trajectory completion generation includes: Candidate reachable path selection and intermediate state determination: Receive the set of state fragments to be completed, and for each fragment, based on the candidate reachable paths from the starting state node to the target state node in the trained behavioral state network, select the target reachable path for path completion derivation, and determine the corresponding intermediate training behavioral states, specifically including: (1) For any fragment of state to be completed Construct its candidate reachable path set, denoted as: ; in, For the state node To the state node The set of candidate reachable paths For the first Candidate reachable paths, This represents the total number of candidate reachable paths.

[0031] (2) Calculate the first The path completion score for the candidate reachable paths is: ; in, For the first Path completion score for candidate reachable paths This represents a path edge in the candidate reachable paths. For path edges The edge weights, For the first The number of state nodes contained in each candidate reachable path. The weighting coefficients of path edge weights in path completion scoring. This represents the weighting coefficient of path compactness in the path completion score. (3) Select the candidate reachable path with the highest path completion score as the target reachable path: ; in, Fragment of state to be completed The corresponding reachable path to the target; (4) Extract intermediate training behavior states in the target reachable path: ; in, This is the set of intermediate training behavioral states in the path to the target. For the first An intermediate training behavior state The number of intermediate training behavioral states; Behavioral reachability sequence and training behavior trajectory completion: The determined intermediate training behavior states are inserted into the corresponding behavioral reachability sequence and training behavior trajectory to generate a completed training trajectory set including the deduced behavior states. Specifically, it includes: (1) For the starting state node in the state fragment to be completed and target state node The intermediate training behavior state set Inserting between the two, we obtain the completed reachable sequence, represented as: ; in, For the first The completed behavior can reach the sequence; (2) Construct the completed training behavior trajectory, represented as: ; (3) The complete training trajectory set is composed of all the completed training behavior trajectories, and is represented as: ; Training behavior completion relation set generation: Based on the temporal adjacency, path continuation, and behavior dependency relationships between the inserted derivation behavior states and the original multi-source training data, a training behavior completion relation set describing the association between the derivation behavior and the multi-source training data is generated, specifically including: (1) For any derivation behavior state With any original multi-source training data Calculate the completion association score, expressed as: ; in, To deduce the behavioral state With the original multi-source training data Completion of correlation scores between them The temporal adjacency value is used to characterize the temporal proximity between the derived behavioral state and the original multi-source training data. This is the path continuity value, used to characterize the degree of continuity between the derived behavioral state and the corresponding response node in the original multi-source training data along the state transition path. The behavior dependency value is used to characterize the degree of dependency between the inferred behavior state and the corresponding training behavior in the original multi-source training data. , , These represent the weight coefficients of temporal adjacency, path continuity, and behavioral dependency in the completion association score, respectively. (2) When satisfied At that time, determine the state of the derivation behavior. With the original multi-source training data There is a completion relationship between them, where, Indicates the completion of the correlation threshold; ; in, , These represent the mean and standard deviation of the historical completion correlation scores, respectively. This is the threshold adjustment coefficient; (3) Generate a set of training behavior completion relations: ; in, Complete the relation set for training behavior.

[0032] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0033] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An intelligent personnel training data information management system, characterized in that, It includes a training behavior constraint modeling module, a training data authenticity deduction module, and a training behavior gap deduction module, among which; The training behavior constraint modeling module acquires training task definition information, training device status information, and training process rule information, and constructs a training behavior constraint model based on the operation sequence, resource call relationship, and result generation conditions of the training task. The training behavior constraint model is used to describe the allowed sequential relationship, dependency relationship, and state change conditions between various training behaviors during the training process, thereby generating a training behavior state network that represents the reachable path of training behavior. The training data authenticity inference module receives multi-source training data generated during the training process and maps the multi-source training data to the training behavior state network. By tracking the state transition path of the multi-source training data in the training behavior state network, it generates the corresponding behavior reachable sequence and the training behavior trajectory set representing the training behavior evolution trajectory. The training behavior gap derivation module receives the behavior reachable sequence and the training behavior trajectory set. Based on the behavior dependencies defined in the training behavior state network, it performs path completion derivation on the discontinuous state transitions that occur in the behavior reachable sequence to identify intermediate training behavior states that are not recorded in the training behavior trajectory. This generates a complete training trajectory set including the derivation of behavior states and a training behavior completion relation set describing the relationship between the derivation of behavior and multi-source training data.

2. The intelligent personnel training data information management system according to claim 1, characterized in that, The training behavior constraint modeling module includes: Construction of training behavior constraint information: Obtain training task definition information, training device status information, and training process rule information. Analyze the task step sequence, resource call relationship, and result generation conditions in the training task definition information. Combined with the device availability status and operation triggering conditions reflected in the training device status information, the training process rule information is expressed in a unified structure to form a set of training behavior constraints describing the triggering conditions, execution order, and resource dependencies of each training behavior, thereby generating the training behavior constraint model. Training Behavior State Network Generation: Based on the training behavior triggering conditions, behavior dependencies, and state change rules defined in the training behavior constraint model, the reachable paths between each training behavior are structured and organized. The training behavior is abstracted into state nodes, and the state transition relationship between training behaviors is abstracted into path edges, thereby constructing a training behavior state network that represents the reachability relationship between training behaviors at different execution stages.

3. The intelligent personnel training data information management system according to claim 2, characterized in that, The construction of the training behavior constraint information includes: Multi-source constraint information acquisition and standardized parsing: Acquire training task definition information, training device status information, and training process rule information, and perform field extraction, format standardization, and time sequence alignment processing respectively to form a basic constraint information set. Specifically, the training task definition information is extracted to obtain the task step sequence, step input items, step output items, and step completion identifier; the training device status information is extracted to obtain the device availability status, device functional status, and operation trigger status; and the training process rule information is extracted to obtain the process preconditions, process transition conditions, and process termination conditions, thereby forming the corresponding structured information of the task steps. Structured information on equipment status and structured information of process rules ; Training behavior triggering and dependency generation: Based on the structured information of task steps, device status, and process rules, the triggering conditions, execution order, and resource dependencies of each training behavior are identified. The order of task steps in the training task definition information is used as the basis for behavior sequence, the available device status and operation triggering conditions in the training device status information are used as behavior triggering constraints, and the process preconditions and transition conditions in the training process rules information are used as behavior dependency constraints, thereby forming a set of training behavior constraints describing the relationship between each training behavior. Training behavior constraint model construction: Based on the set of training behavior constraints, each training behavior is abstracted into a behavior node, and the behavior connection relationships that satisfy the triggering conditions and have dependencies are abstracted into behavior transition relationships. The sequential relationships, dependencies, and state change conditions among each behavior node are structured and organized to construct a training behavior constraint model that describes the execution constraint relationships of each training behavior during the training process. .

4. The intelligent personnel training data information management system according to claim 3, characterized in that, The generation of training behavior trigger relationships and dependency relationships includes: Trigger condition determination: For any training action The trigger condition judgment value for each training behavior is calculated based on the task step sequence, device status information, and process rule information. When satisfied At that time, determine the training behavior The triggering conditions are met, among which, For the first The trigger threshold for each training action; Dependency determination: Calculate any two training actions and Behavioral dependency strength To characterize the dependencies between training behaviors; Construction of the training behavior constraint set: Based on the determination results of the training behavior trigger conditions and the dependency strength between training behaviors, a training behavior constraint set is constructed. , is represented as: 。 5. The intelligent personnel training data information management system according to claim 4, characterized in that, The training behavior state network generation includes: Behavior node mapping: Based on the set of training behaviors in the training behavior constraint model, each training behavior is mapped to a corresponding state node, and each state node is associated with the corresponding training behavior triggering condition and state attribute. Path edge generation: Based on the training behavior triggering conditions, behavior dependencies and state change rules defined in the training behavior constraint model, determine whether there is a state transition relationship between different state nodes, and abstract the state transition relationship that meets the triggering conditions into a path edge. Training Behavior State Network Construction: Based on the set of state nodes and the set of path edges, the reachable paths between each training behavior are structured to form a training behavior state network that represents the reachability relationships between training behaviors at different execution stages. .

6. The intelligent personnel training data information management system according to claim 5, characterized in that, The training data authenticity deduction module includes: Multi-source training data reception and state node mapping: This process receives multi-source training data generated during training and, based on the training behavior identifier, time identifier, device operation identifier, and result identifier within the multi-source training data, maps each piece of multi-source training data to the corresponding state node in the training behavior state network, forming a sequence of observation nodes corresponding to the multi-source training data. ; State transition path tracing and behavior reachability sequence generation: Based on the connectivity of the observation node sequences in the trained behavior-state network, the state transition paths between adjacent observation nodes are traced, and state transition sequences that satisfy the network reachability relationship are extracted to generate the corresponding behavior reachability sequences. ; Construction of training behavior evolution trajectory: Temporal association and path organization are performed on the state nodes and their corresponding state transition relationships in the behavior reachable sequence, and state nodes that meet the continuous state transition conditions are combined to form the training behavior trajectory. The trajectory intensity is calculated based on the path edge weights between state nodes in the training behavior trajectory. This results in a set of training behavior trajectories that includes multiple training behavior trajectories and their intensity. It is used to characterize the state evolution path and path strength characteristics of training behavior during the execution process.

7. The intelligent personnel training data information management system according to claim 6, characterized in that, The state transition path tracing and behavior reachability sequence generation include: Reachability determination of adjacent observation nodes: Based on the connection relationship of the observation node sequence in the training behavior state network, determine whether there is a reachable path between adjacent observation nodes; Optimal state transition path calculation: For neighboring observation nodes with reachable paths, calculate their optimal state transition path in the trained behavioral state network. ; Behavioral reachability sequence generation: Based on the optimal state transition path between all adjacent observation nodes, the state nodes in the path are concatenated in chronological order to generate a behavioral reachability sequence. .

8. The intelligent personnel training data information management system according to claim 7, characterized in that, The training behavior gap derivation module includes: Discontinuous state transition identification: Receive the action reachable sequence and the training action trajectory set, and compare the adjacent state nodes in the action reachable sequence with the action dependencies in the training action state network to identify discontinuous state transition segments that do not meet the direct state transition conditions. For each discontinuous state transition segment, combine the reachable path information of the corresponding training action trajectory in the training action state network to determine intermediate candidate state nodes that exist between the starting state node and the target state node but are not explicitly recorded by the current action reachable sequence, thereby forming a set of state segments to be completed. Training trajectory completion generation: Receive the set of state fragments to be completed, and based on the candidate reachable paths between the starting state node and the target state node in the training behavior state network, perform path completion deduction on the intermediate candidate state nodes. Insert the deduced intermediate training behavior states into the corresponding behavior reachable sequences and training behavior trajectories to generate a complete training trajectory set including the deduced behavior states. Simultaneously, based on the temporal adjacency, path continuity, and behavior dependency relationships between the inserted deduced behavior states and the original multi-source training data, generate a training behavior completion relationship set describing the association between the deduced behavior and the multi-source training data.

9. The intelligent personnel training data information management system according to claim 8, characterized in that, The discontinuous state transition identification includes: Direct transition determination between adjacent state nodes: Receive the reachable sequence of behaviors and the training set of behavior trajectories, and calculate the transition between any adjacent state nodes in the reachable sequence. and Direct transfer decision value When satisfied When determining the state node With state nodes Constitute candidate discontinuous state transition segments; Discontinuous state transition segment selection: Candidate discontinuous state transition segments are selected by combining reachable path information from the training behavior trajectory set to determine whether there is a reachable but unrecorded intermediate state path between the starting state node and the target state node. Calculate the reachable path determination value for multiple nodes. When satisfied and At that time, the judgment These are discontinuous state transition segments, forming a set of discontinuous state transition segments. ; Generation of the set of state fragments to be completed: For each discontinuous state transition fragment, extract the intermediate candidate state nodes in the reachable path between its starting state node and the target state node, and form the set of state fragments to be completed. .

10. The intelligent personnel training data information management system according to claim 9, characterized in that, The generation of the completed training trajectory includes: Candidate reachable path selection and intermediate state determination: Receive the set of state fragments to be completed, and for each state fragment to be completed, select the target reachable path for path completion derivation based on the candidate reachable paths between the starting state node and the target state node in the training behavior state network, and determine the corresponding intermediate training behavior state. Behavioral reachability sequence and training behavior trajectory completion: The determined intermediate training behavior states are inserted into the corresponding behavioral reachability sequence and training behavior trajectory to generate a completed training trajectory set including the deduced behavior states. ; Training behavior completion relation set generation: Based on the temporal adjacency, path continuation, and behavior dependency relationships between the inserted derivation behavior states and the original multi-source training data, a training behavior completion relation set describing the relationship between the derivation behavior and the multi-source training data is generated.

Images