A method, device and electronic equipment for ensuring intelligent management of resources

Abstract

This application provides a method, apparatus, and electronic device for intelligent management of logistical resources, relating to the field of data processing. In this method, multi-dimensional feature vectors of logistical resources are acquired and combined with command link perturbation signal sequences and situational disturbance index sequences, input into a multi-task temporal coding network to generate a dynamic allocation intent matrix. This matrix identifies frozen logistical resources awaiting allocation and writes them into a cognitive domain cache. Upon arrival of an allocation request, frozen logistical resources are matched, generating a picking behavior response flow to complete an anonymous picking operation. During transportation, multiple types of disturbance parameters are collected to construct a path disturbance vector, updating the path disturbance density map. At the handover node, fuzzy matching is performed based on natural action flow behavior feedback and a pre-stored behavior expectation map to confirm the handover. Finally, the path disturbance vector is input into an allocation prediction model to iteratively update the logistical resource cognitive state. Implementing the technical solution provided in this application facilitates intelligent management of logistical resources.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically to a method, apparatus, and electronic device for ensuring intelligent management of resources. Background Technology

[0002] Against the backdrop of the ongoing deepening of the joint support system, support resource management is evolving from a centralized, command-driven model to an intelligent, ubiquitous sensing model.

[0003] Currently, existing resource management methods are generally based on explicit task-driven, centralized allocation command triggering, and path tracking control strategies for allocation management. They rely on static material coding, warehousing information systems, and traditional logistics nodes for physical allocation and status confirmation. This type of management model lacks the ability to predict the dynamic adjustability of resource support under normal circumstances, and cannot model future resource demand trends in advance. This results in long allocation response cycles, predictable path behavior, and highly exposed node handover information, which is detrimental to intelligent management.

[0004] Therefore, there is an urgent need for a method, device, and electronic equipment for ensuring intelligent resource management. Summary of the Invention

[0005] This application provides a method, apparatus, and electronic device for intelligent management of security resources, which facilitates intelligent management of security resources.

[0006] The first aspect of this application provides a method for intelligent management of support resources. The method includes: acquiring a multi-dimensional feature vector corresponding to the support resource, inputting the multi-dimensional feature vector into a multi-task temporal coding network, combining a command link perturbation signal sequence and a situational disturbance index sequence, and outputting a dynamic allocation intention matrix for the support resource; identifying support resources with activation exceeding a set threshold in the dynamic allocation intention matrix, marking them as frozen-state wait-to-be-allocated support resources, constructing a support resource freeze mapping vector for the frozen-state wait-to-be-allocated support resources, and writing the support resource freeze mapping vector into a cognitive domain cache, wherein the frozen-state wait-to-be-allocated support resources remain in a static physical state and do not undergo outbound operations; upon receiving an allocation request, matching the allocation request with the support resource freeze mapping vector in the cognitive domain cache, activating the physical outbound path of the corresponding frozen-state wait-to-be-allocated support resource upon successful matching, and generating a picking behavior response stream through a local physical domain instruction mapper. The picking behavior response flow is used to enable the support resources to complete the picking operation without explicitly marking task information. During transportation, vehicle status, material vibration index, temperature and humidity disturbance, and electromagnetic interference intensity are collected to generate a path disturbance vector. This path disturbance vector is written into the path transcription cache and the path disturbance density map in the cognitive domain cache is updated to perform path transcription processing on the transportation process of the support resources. After the support resources arrive at the handover node, behavioral feedback composed of the natural action flow of the handover node is obtained. This behavioral feedback is fuzzily matched with the pre-stored behavioral expectation map in the frozen mapping vector of the support resources. If the match is successful, the handover confirmation of the support resources is completed, but no explicit receipt information is recorded. The path disturbance vector of the support resources during the allocation process is input into the disturbance input channel of the cognitive domain allocation prediction model to update the task response frequency and path disturbance tolerance of the support resources, completing the adaptive iteration of the cognitive state of the support resources.

[0007] Optionally, the step of obtaining the multi-dimensional feature vector corresponding to the support resource and inputting the multi-dimensional feature vector into a multi-task temporal coding network, combining the command link perturbation signal sequence and the situational disturbance index sequence, and outputting the dynamic allocation intention matrix of the support resource, specifically includes: constructing a static structural feature set of the support resource, the static structural feature set including the support resource's combat readiness level identifier, historical allocation record sequence, modular composition structure, physical storage location coordinates, corresponding combat unit configuration relationship, and typical usage task type label; obtaining the allocation behavior trajectory vector of the support resource within a fixed time window, the allocation behavior trajectory vector being indexed by a timestamp, recording the frequency of call and path history of the support resource; collecting... The command link perturbation signal sequence covering the storage node of the support resource is obtained. The command link perturbation signal sequence includes link activation frequency, low-intensity signaling spectrum and command fragment frequency. The situational disturbance index sequence covering the area where the support resource is located is obtained. The situational disturbance index sequence is used to reflect the task load change rate, threat level diffusion value and the aggregation state of the material call frequency of adjacent combat units. The static structural feature set, the allocation behavior trajectory vector, the command link perturbation signal sequence and the situational disturbance index sequence are time-position nested and aligned to obtain a standardized time series input tensor. The standardized time series input tensor is input into the multi-task time series coding network to generate the dynamic allocation intention matrix of the support resource.

[0008] Optionally, the step of identifying guaranteed resources with activation levels exceeding a set threshold in the dynamic allocation intent matrix, marking them as frozen reserve resources, constructing a guaranteed resource freeze mapping vector for the frozen reserve resources, and writing the guaranteed resource freeze mapping vector into the cognitive domain cache specifically includes: standardizing the allocation probability vector of each guaranteed resource in the dynamic allocation intent matrix, extracting the allocation peak probability value, and calculating the allocation activation score by combining the task adaptability vector and the path response sensitivity vector; comparing the allocation activation score with a preset threshold, if it exceeds... The preset threshold is used to mark the corresponding guaranteed resource as the frozen-state waiting-to-be-allocated guaranteed resource; a guaranteed resource freezing mapping vector is constructed for the frozen-state waiting-to-be-allocated guaranteed resource, which includes the physical identification information of the corresponding guaranteed resource, the allocation probability vector, the allocation activation score, the identification timestamp, the regional location index, the link perturbation feature fragment, and the environmental state fingerprint data; the guaranteed resource freezing mapping vector is written into the cognitive domain cache, and a unique index key is generated by static identification and identification timestamp to ensure the uniqueness of the guaranteed resource freezing mapping vector in the cognitive domain cache.

[0009] Optionally, upon receiving a transfer request, the step involves matching the transfer request with the frozen resource mapping vector in the cognitive domain cache. Upon successful matching, the physical outbound path of the corresponding frozen resource is activated, and a picking action response flow is generated through the local physical domain instruction mapper. Specifically, this includes: deconstructing the transfer request to extract the task instruction number, transfer node identifier, required resource category, task initiation timestamp, preset response time limit, region label, and operational phase parameters to form a transfer request feature vector; performing spatial location matching in the cognitive domain cache based on the unique index key and physical identifier, and matching based on the identification timestamp and physical identifier... The task initiation timestamp execution time window is filtered, and the logical semantic alignment is performed based on the required resource category and modular composition structure, and a matching confidence score is calculated. If the matching confidence score is determined to be higher than the set matching threshold, the physical outbound path of the corresponding resource is activated. The physical domain picking path planner is called to generate a picking trajectory based on the physical storage location of the resource, the warehouse density distribution map, the adjacent path impedance function, and the outbound priority table. The picking trajectory and the resource identification information are input to the local physical domain instruction mapper to generate the picking behavior response stream, and the physical outbound operation of the resource is completed in an anonymous operation sequence manner.

[0010] Optionally, the step of collecting vehicle status, material vibration index, temperature and humidity disturbance, and electromagnetic interference intensity during transportation and generating a path disturbance vector, writing the path disturbance vector into a path transcription cache, and updating the path disturbance density map in the cognitive domain cache specifically includes: collecting vehicle acceleration vector, steering angle sequence, and path inertial deviation information through an inertial navigation unit and a Beidou differential positioning unit to construct the vehicle status; obtaining the vibration spectrum map of the support resources through a micro vibration sensor array, and extracting the energy distribution and peak change points of the main vibration frequency band to construct the material vibration index; and obtaining the ambient temperature through a composite climate sensor unit. The temperature and humidity disturbance is constructed by measuring the rate of change of temperature, relative humidity gradient, and amplitude of external climate fluctuations. The electromagnetic interference intensity is constructed by collecting spectral interference density, signal modulation fluctuation, and high-frequency pulse interference response using an onboard radio frequency analyzer. The vehicle status, material vibration index, temperature and humidity disturbance, and electromagnetic interference intensity are time-aligned and vector-stitched to generate the path disturbance vector. The path disturbance vector is written into the path transcription cache and input into the cognitive domain cache to update the path disturbance density map. The path disturbance density map is used to characterize the distribution of transportation disturbance intensity and frequency characteristics of the guaranteed resources.

[0011] Optionally, after the support resource arrives at the handover node, the step of obtaining behavioral feedback composed of the natural motion flow of the handover node, performing fuzzy matching between the behavioral feedback and the pre-stored behavioral expectation map in the frozen mapping vector of the support resource, and completing the handover confirmation of the support resource upon successful matching without recording explicit receipt information, specifically includes: obtaining the full-body motion sequence of the handover personnel through a posture recognition camera and extracting the limb motion vector flow; obtaining the operation rhythm information of the handover personnel through a micro-motion inertial detector, the operation rhythm information including operation frequency changes, instantaneous pause length, and repetitive motion waveforms; and obtaining the contact path of the support resource through a contact pressure matrix. The contact path data includes physical contact path, load transfer process, and center of gravity change time sequence data. The limb movement vector flow, the operation rhythm information, and the contact path data are time-aligned and input into the behavior fusion encoder to generate the behavior feedback. The behavior feedback is fuzzily matched with the behavior expectation map pre-stored in the resource freeze mapping vector. The matching degree is determined based on the interaction feature mapping distance, temporal similarity, and tolerance verification. When the matching degree is higher than a threshold, the handover confirmation is completed. The handover confirmation process records the matching success identifier and timestamp, but does not generate a receipt number, record the identity information of the operation subject, or write the task identifier.

[0012] Optionally, the step of inputting the path disturbance vector of the guaranteed resource during the allocation process into the disturbance input channel of the cognitive domain allocation prediction model to update the task response frequency and path disturbance tolerance of the guaranteed resource, thereby completing the adaptive iteration of the cognitive state of the guaranteed resource, specifically includes: generating a fused disturbance vector based on the path disturbance vector and the behavioral feedback; inputting the fused disturbance vector into the disturbance input channel of the cognitive domain allocation prediction model, and processing it through the disturbance spectrum reconstructor, response mutation function, and tolerance learning submodule to generate a task response frequency update factor and a path disturbance tolerance update factor; writing the task response frequency update factor and the path disturbance tolerance update factor into the state record structure of the guaranteed resource in the cognitive domain cache, and feeding it back to the multi-task temporal coding network to achieve dynamic adaptive correction of the guaranteed resource allocation intention.

[0013] A second aspect of this application provides an intelligent management device for support resources. The device includes an acquisition module and a processing module. The acquisition module acquires multi-dimensional feature vectors corresponding to support resources and inputs these vectors into a multi-task temporal coding network. By combining the command link perturbation signal sequence and the situational disturbance index sequence, it outputs a dynamic allocation intention matrix for the support resources. The processing module identifies support resources in the dynamic allocation intention matrix whose activation exceeds a set threshold, marks them as frozen-state reserve support resources, constructs a freeze mapping vector for the frozen-state reserve support resources, and writes the freeze mapping vector into a cognitive domain cache. The frozen-state reserve support resources remain in a static physical state and do not undergo outbound operations. The processing module further matches an allocation request with the frozen mapping vector in the cognitive domain cache upon receiving it. Upon successful matching, it activates the physical outbound path of the corresponding frozen-state reserve support resource and generates a picking action response through a local physical domain instruction mapper. The picking behavior response flow is used to enable the support resources to complete the picking operation without explicitly marking task information. The processing module is also used to collect vehicle status, material vibration index, temperature and humidity disturbance amount and electromagnetic interference intensity during transportation and generate path disturbance vector. The path disturbance vector is written into the path transcription cache and the path disturbance density map in the cognitive domain cache is updated to perform path transcription processing on the transportation process of the support resources. The processing module is also used to obtain the behavior feedback composed of the natural action flow of the handover node after the support resources arrive at the handover node. The behavior feedback is fuzzily matched with the behavior expectation map pre-stored in the frozen mapping vector of the support resources. If the match is successful, the handover confirmation of the support resources is completed, but no explicit receipt information is recorded. The processing module is also used to input the path disturbance vector of the support resources during the allocation process into the disturbance input channel of the cognitive domain allocation prediction model, update the task response frequency and path disturbance tolerance of the support resources, and complete the adaptive iteration of the cognitive state of the support resources.

[0014] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, and both the user interface and the network interface are used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method described above.

[0015] A fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed, perform the method described above.

[0016] In summary, one or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] 1. By acquiring multi-dimensional feature vectors of support resources and combining them with command link perturbation signal sequences and situational disturbance index sequences, a dynamic allocation intent matrix is ​​generated through a multi-task time-series coding network. This enables high-dimensional prediction of support resource allocation trends before tasks are explicitly issued. This prediction mechanism is independent of specific instructions or task codes, possesses task-insensitive characteristics, and can proactively perceive support resources that may be allocated and mark them as frozen, thus transforming allocation from "passive reception" to "active pre-positioning."

[0018] 2. This method does not use real path coordinates, geographical nodes, or explicit scheduling tags. Instead, it constructs a path disturbance vector by collecting vehicle status, material vibration index, temperature and humidity disturbance, and electromagnetic interference intensity. This vector is then written into the path transcription cache and the cognitive domain path disturbance density map, forming a disturbance spectrum model for ensuring resource transportation. This achieves the purpose of path transcription and behavior desensitization. This mechanism effectively prevents the enemy from predicting allocation patterns through reverse graph modeling of path behavior, improving the resilience and concealment of the allocation process.

[0019] 3. By analyzing the activation degree of the allocation intention matrix, potential high-probability allocation resources are identified. A resource freeze mapping vector is constructed and written into the cognitive domain cache to ensure that resources remain in a static state in the physical domain, without generating any outbound instructions. This freeze-state waiting mechanism constructs a "non-physical state space" for allocation preparation, significantly reducing the frequency of explicit operations in resource management and improving the low visibility and high responsiveness of the allocation preparation state.

[0020] 4. From the anonymous picking behavior response flow of the physical outbound path to the behavior feedback confirmation mechanism based on the natural action flow at the handover node, the entire method does not record task numbers, call permission verification, or generate receipt records at the execution layer. It only performs fuzzy matching confirmation of behavioral features and expectation maps in the cognitive domain, ensuring that the entire allocation process has low exposure, low perception, and high camouflage. This effectively improves the survivability and concealment of resource allocation in security-sensitive scenarios and facilitates intelligent management of resource support. Attached Figure Description

[0021] Figure 1 A flowchart illustrating a method for intelligent management of security resources provided in this application embodiment;

[0022] Figure 2 A schematic diagram of a module for an intelligent resource management device provided in an embodiment of this application;

[0023] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0024] Explanation of reference numerals in the attached figures: 21. Acquisition module; 22. Processing module; 31. Processor; 32. Communication bus; 33. User interface; 34. Network interface; 35. Memory. Detailed Implementation

[0025] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification 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.

[0026] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0027] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0028] To address the aforementioned technical problems, this application provides a method for intelligent management of security resources, referring to... Figure 1 , Figure 1 This is a flowchart illustrating a method for intelligent resource management as provided in an embodiment of this application. The method is applied to a server and includes steps S110 to S160, as follows:

[0029] S110. Obtain the multi-dimensional feature vector corresponding to the support resources, and input the multi-dimensional feature vector into the multi-task time-series coding network. Combine the command link micro-disturbance signal sequence and the situational disturbance index sequence to output the dynamic allocation intention matrix of the support resources.

[0030] Specifically, a server refers to the computing platform that deploys the intelligent management model for logistical support resources. It is responsible for receiving all logistical support resource data, executing the computational process of a multi-task temporal coding network, and outputting allocation prediction results. Essentially, it is the central hub for allocation control logic. Example: A military logistics center deploys an edge computing node for predictive scheduling of logistical support resources. This node acts as a server, running the intelligent management model for logistical support resources in real time. In a military context, logistical support resources specifically refer to materials, equipment, or components used to support combat, training, exercises, and deployment tasks, characterized by task dependence, limited quantity, and allocation sensitivity. Example: The X-14 tactical communication module is a core logistical support resource used by a forward unit to build shortwave radio links, possessing important task allocation relationships. A multi-dimensional feature vector refers to a data set with multiple semantic dimensions used to describe the state, attributes, behavior, and relationships of logistical support resources, which forms the vector input to the model through structured encoding. The content includes: Static attribute dimension: combat readiness level, material type, batch number; Behavioral space dimension: historical allocation frequency, path record, average response time; Environmental mapping dimension: storage node location, theater number, unit affiliation; Functional dependency dimension: modular composition structure, dependent sub-resources, typical uses. Example: The feature vector of the tactical communication module can be encoded as follows: combat readiness level = level 2, number of allocations in the past 30 days = 3 times, recent mission use = company-level tactical exercise, storage node = K-zone warehouse level 3, associated equipment = shortwave digital radio, average allocation response time = 6 hours.

[0031] Multi-task temporal coding networks (MTCs) are neural network structures that integrate multiple input semantic dimensions and possess time-aware capabilities. They are used to process the state evolution and allocation trend prediction of support resources under different task scenarios. Their internal structure includes: a time-location embedding module to handle the impact of time windows on the state of support resources; a task channel separation structure to support independent modeling of the same resource under different task categories; and an attention weight allocation mechanism to mine the relative influence of each input dimension on the allocation trend. For example, after inputting the past allocation time points, allocation background, and allocation frequency of a communication module into the network, the model can predict the probability of the module being called in two types of tasks—"emergency deployment" and "training reorganization"—within the next 48 hours.

[0032] Command link perturbation signal sequences refer to weak signaling behaviors that reflect changes in the state of the high-level command chain but for which no explicit mission instructions have yet been issued. These include increased link activation rates, abnormal changes in instruction frequency, or offsets in link return traffic. These perturbation signals can be used as leading indicators in predictive models to determine whether a mission is about to be launched. Example: A theater command platform has frequently made short-cycle link maintenance requests to a specific area in the past 48 hours, with a 30% increase in return instruction traffic. This indicates that the node where the communication module is located may soon participate in forward deployment, and this signal constitutes a key a priori indicator of an increasing allocation trend. Situational disturbance index sequences refer to quantitative indicators reflecting changes in mission intensity, deployment pressure, risk diffusion, and instruction density within a tactical area. They are macroscopic predictive quantities representing increased resource usage pressure and allocation demand. They are often generated by the fusion of multiple factors, such as the frequency of tactical activities, the allocation intensity of adjacent units, mission volatility, and enemy situation simulation curves. Example: Recently, several military units have increased their communication support capabilities, and the density of radio modules allocated to adjacent nodes has increased. The resulting situational disturbance index has reached 0.83 (range 0-1), indicating that the probability of communication module allocation has increased significantly.

[0033] The dynamic allocation intent matrix is ​​the core output of the model. It is a multi-dimensional structured tensor used to describe the probability of various support resources being allocated under different future time windows, the assigned task type, and path response sensitivity. Its structure is three-dimensional: First dimension: Resource identifier (each support resource); Second dimension: Time window (e.g., future 24h, 48h, 72h); Third dimension: Allocation intent (allocation probability, assigned task type, path disturbance tolerance). Example: In the dynamic allocation intent matrix, the communication module X-14 has an allocation probability of 0.76 within the next 48 hours, a task category of tactical deployment, and a medium path disturbance sensitivity value. Based on this, the system marks it as a frozen-state support resource.

[0034] In one possible implementation, a multi-dimensional feature vector corresponding to the support resources is obtained, and this feature vector is input into a multi-task temporal coding network. Combined with the command link perturbation signal sequence and the situational disturbance index sequence, a dynamic allocation intent matrix of the support resources is output. Specifically, this includes: constructing a static structural feature set of the support resources, which includes the support resources' combat readiness level identifier, historical allocation record sequence, modular composition structure, physical storage location coordinates, corresponding combat unit configuration relationships, and typical usage task type annotations; obtaining the allocation behavior trajectory vector of the support resources within a fixed time window, where the allocation behavior trajectory vector is indexed by a timestamp, recording the frequency of support resource calls and... The process involves: 1) collecting command link perturbation signal sequences covering the storage nodes of support resources. These sequences include link activation frequency, low-intensity signaling spectrum, and command fragment frequency. 2) acquiring a situational disturbance index sequence covering the area where support resources are located. This sequence reflects the task load variation rate, threat level diffusion value, and the aggregation state of material mobilization frequency of adjacent combat units. 3) aligning the static structural feature set, allocation behavior trajectory vector, command link perturbation signal sequences, and situational disturbance index sequences with time and location nesting to obtain a standardized temporal input tensor. This standardized temporal input tensor is then input into a multi-task temporal coding network to generate a dynamic allocation intent matrix for support resources.

[0035] Specifically, when implementing the technical solution, the intelligent management server for support resources first constructs a static structural feature set of support resources. This feature set is extracted from the support resource database and structurally encoded, including: combat readiness level identifiers (e.g., Level 1 / Level 2 / Reserve), historical allocation records in timestamp sequence (including allocation time, task type, and path segment number), the modular composition structure of support resources in the equipment system (e.g., host modules, communication sub-components, power supply units, etc.), physical storage location coordinates (e.g., "Warehouse A - Shelf 12 - Layer 2"), the configuration relationship between support resources and combat units (e.g., belonging to "37th Combined Arms Brigade Communication Company"), and corresponding typical usage task type labels (e.g., "Exercise Command and Dispatch" and "Emergency Communication Support"). All of the above information is uniformly encoded into a static structural feature vector. Where d represents the feature dimension and i represents the guaranteed resource number.

[0036] The system then acquires the necessary resources within a fixed time window [t0, t]. n The transfer behavior trajectory within ] is used to construct the transfer behavior trajectory vector B. i (t), where each time point t k ∈[t0,t n This corresponds to a set of behavioral event codes, including the call frequency f. i (t k ) and path segment identifier p i (tk This vector, indexed by timestamps, forms a first-order time series, used to represent the behavioral trends and path evolution of resources in the allocation history, as shown in the following formula:

[0037]

[0038] The system synchronously accesses the command link communication data stream of the physical storage nodes covering the protection resources and extracts the command link perturbation signal sequence. This sequence consists of three characteristic variables: the link activation frequency λ per unit time. i (t k Signaling spectral energy density σ i (t k ), and the trigger frequency ρ of unstructured instruction fragments. i (t k This constitutes the following signal vector, used to characterize potential task mobilization indications of the command system regarding the associated storage node of the support resource:

[0039]

[0040] Simultaneously, obtain the situational disturbance index sequence of the region where the security resources are located within the same time window, and construct the situational disturbance vector D. i (t), which consists of: task load variation rate μ i (t k Threat level diffusion value θ i (t k ), the frequency aggregation state ξ of adjacent combat units i (t k ), the corresponding vector expression is:

[0041]

[0042] The above four sets of input vectors S i B i C i D i Perform nested alignment of time positions. Specifically, this is based on a unified timestamp t. k Parallel synchronization is performed on all sequences, and a normalized temporal input tensor is formed using a nested tensor builder:

[0043]

[0044] Where ∥ represents the vector concatenation operation, X i ∈R n×h h is the total feature dimension of each time slice after splicing, and n is the length of the time series.

[0045] Finally, the normalized temporal input tensor X iThe input is fed into a multi-task temporal coding network. This network includes: a temporal location embedding layer TE(t) k The first layer is used to express time dependence; the second layer is the multi-channel task decoupling layer MTL(·), which encodes tasks to ensure the response performance of resources under different allocation types; the third layer is the attention mechanism layer Attn(·), which dynamically allocates the influence weights of each input dimension.

[0046] Network output ensures dynamic allocation intent matrix for resources:

[0047]

[0048] Where, π i (t k ) for in t k The probability of adjusting the allocation at any time, τ i (t k ) represents the prediction result for the allocation task type, δ i (t k The path disturbance tolerance index is used. This intent matrix serves as the direct input for subsequent identification of frozen state reserve resources, enabling non-command predictive control of reserve resource allocation behavior.

[0049] S120. Identify the guaranteed resources whose activation exceeds the set threshold in the dynamic allocation intention matrix, mark them as frozen waiting-to-be-allocated guaranteed resources, construct the guaranteed resource freezing mapping vector of the frozen waiting-to-be-allocated guaranteed resources, and write the guaranteed resource freezing mapping vector into the cognitive domain cache. The frozen waiting-to-be-allocated guaranteed resources remain in a static physical state and do not perform outbound operations.

[0050] Specifically, activation level refers to the comprehensive score of allocation trend calculated by combining the allocation probability vector, task suitability vector, and path disturbance tolerance vector. It is a key indicator for determining whether a resource has "high potential for allocation." Frozen-state reserve resources refer to reserve resources that have been identified as having a high allocation trend but have not yet triggered allocation instructions. They remain static in the physical domain and are not released from inventory; in the cognitive domain, they are marked as "pending allocation" pre-response targets. Their characteristics are: no change in inventory status, no generation of allocation numbers, but a state lock in the cognitive domain, allowing for priority matching of subsequent allocation requests. For example, a communication terminal may not have actually moved or been requested for allocation, but it may have been pre-marked as a reserve resource by the system, allowing the system scheduling module to prioritize the retrieval of such resources.

[0051] The resource freeze mapping vector refers to the complete state description structure of frozen-state resource allocation resources in the cognitive domain. It is used to record their static structural identifiers and allocation behavior prediction elements, forming the "mapping body" of the allocation control chain. It includes the following fields: unique resource number (e.g., RFID code), static physical location information (e.g., level 7 of warehouse A), current status flag (e.g., freeze status code 01), activation score, corresponding allocation probability vector, identification timestamp, and behavior expectation map hash value (used for handover verification). Example: In the freeze mapping vector of a communication terminal, its unique number XQ-MC-014 is recorded, located on level 3 of warehouse B, with a freeze status code of 01, an activation level of 0.82, and a freeze time of 08:00 2025 / 07 / 21.

[0052] The cognitive domain cache refers to the task prediction cache structure in the cognitive layer of resource security. It does not store physical state data but records behavioral expectations, allocation trends, and mapping vector results, serving as the information carrier of the resource's "intent layer." After the frozen mapping vector is written to the cognitive domain cache, the resource is identified as being in an "allocation preparation state," but this is not explicitly manifested as a task path or outbound instruction. Example: After the communication terminal XQ-MC-014 is written to the cognitive domain cache, it is identified as a priority candidate by the resource security allocation and scheduling module. If a weak instruction is subsequently sent by the theater command chain, this resource will respond directly without triggering a recalculation of the allocation path. The static physical state without outbound operations means that the security resource does not undergo location changes in the warehousing system, does not trigger any mechanical picking, does not generate transportation path instructions, and does not affect inventory logic measurement, but enters an "allocation frozen state" at the cognitive level. This design ensures that the entire allocation preparation process does not interfere with the actual logistics system, exhibiting high concealment and low disturbance. Example: Ensure that the resources remain in their original positions on the warehouse shelves, but the server treats them as the first response object. Once a task occurs, it only needs to skip the allocation approval chain and directly execute the outbound operation.

[0053] In summary, this step identifies resources with activation levels exceeding a threshold in the allocation intent matrix, pre-constructs an allocation readiness structure within the cognitive domain, and achieves the core capabilities of "pre-modeling, non-command response, and implicit freezing" of resources. This provides a high-security, high-responsiveness, and low-intervention foundation for the subsequent intelligent allocation process of resources.

[0054] In one possible implementation, resources with activation levels exceeding a set threshold are identified in the dynamic allocation intent matrix and marked as frozen reserve resources. A frozen resource mapping vector is constructed for these frozen reserve resources, and the frozen resource mapping vector is written into the cognitive domain cache. Specifically, this includes: standardizing the allocation probability vector of each reserve resource in the dynamic allocation intent matrix, extracting the allocation peak probability value, and calculating the allocation activation score by combining the task adaptability vector and the path response sensitivity vector; comparing the allocation activation score with a preset threshold, and if it exceeds the preset threshold, marking the corresponding reserve resource as a frozen reserve resource; constructing a frozen resource mapping vector for the frozen reserve resources, which includes the physical identification information of the corresponding reserve resource, the allocation probability vector, the allocation activation score, the identification timestamp, the regional location index, the link perturbation feature fragment, and the environmental state fingerprint data; writing the frozen resource mapping vector into the cognitive domain cache, and generating a unique index key through a static identifier and the identification timestamp to ensure the uniqueness of the frozen resource mapping vector in the cognitive domain cache.

[0055] Specifically, this technical solution is used to identify and manage the cognitive domain freeze of high-potential allocation targets in the dynamic allocation intent matrix of guaranteed resources. The key lies in calculating allocation activation scores from three dimensions: allocation probability, task adaptability, and path tolerance. Based on the scoring results, a guaranteed resource freeze mapping vector is constructed and written into the cognitive domain cache. First, the server standardizes the allocation probability vector for each guaranteed resource in the dynamic allocation intent matrix. Let the guaranteed resource number be i, and the time window be t. k ∈[t1,t n The original allocation probability sequence is π. i (t k Its standardized form is defined as:

[0056]

[0057] Wherein, min(π) i ) and max(π i The values ​​) represent the minimum and maximum allocation probabilities of the resource within the predicted time window, respectively. Standardization is used to eliminate scale differences in allocation probability distributions among resources and enhance the comparability of scores. Subsequently, the peak probability value is extracted from the standardized vector. As the main response indicator of allocation behavior trends.

[0058] Secondly, the probability value of the main peak will be adjusted. Task fit vector τ i With path response sensitivity vector δ i The activation score function is input together to generate the activation score value A. iThe scoring function adopts a weighted linear fusion form:

[0059]

[0060] Among them: A i : Guarantee the activation score of resource i; The average value of the task fit vector is used to characterize its task assignment breadth; The mean of the path response sensitivity vector. The higher the sensitivity, the lower the tolerance, so we take its inverse value; w1, w2, w3: the scoring weight parameters set by the system, which satisfy w1+w2+w3=1.

[0061] Next, the obtained activation score A i With the system-set activation threshold θ th Compare, if A i >θ th If so, the guaranteed resource i is marked as a frozen state awaiting adjustment, and no physical operations are performed; only the state of the response to be adjusted in the cognitive domain is constructed. Subsequently, a frozen mapping vector for the guaranteed resource is constructed. This vector is a complete structured description of the current predicted state of the guaranteed resource, and its fields include: physical identification information ID. i : Unique identifier for guaranteed resources; allocation probability vector Standardized allocation intention sequence; allocation activation score A i ; Identify timestamps The system marks the time of the frozen state; the region location index L i : Identifier of the area to which the current guaranteed resources belong; Link perturbation feature segment C i : Link activation frequency, signaling characteristics, etc. associated with it; environmental status fingerprint data E i The resource's history of temperature and humidity disturbances on the storage node, local environmental noise parameters, etc.

[0062] Finally, the resource freeze mapping vector is written to the cognitive domain cache, using a static identifier ID during the write. i With identification timestamp Jointly construct a unique index key κ i :

[0063]

[0064] Ensure the uniqueness of the mapping of the protected resource in the cognitive domain cache. Even if it is repeatedly activated in multiple rounds of prediction, the state update process of the same resource is unique and traceable, avoiding state conflicts and duplicate judgments.

[0065] This processing chain enables proactive identification, state freezing, and cognitive domain mapping of risk states in resource allocation, effectively supporting the preprocessing logic of non-command-type allocation responses and providing pre-state control support for subsequent matching scheduling and path transcribing.

[0066] S130. After receiving the allocation request, the allocation request is matched with the frozen mapping vector of the guaranteed resource in the cognitive domain cache. If the match is successful, the physical outbound path of the corresponding frozen guaranteed resource is activated. The picking behavior response stream is generated through the local physical domain instruction mapper. The picking behavior response stream is used to enable the guaranteed resource to complete the picking operation in the state where the task information is not explicitly marked.

[0067] Specifically, a resource allocation request refers to a resource call instruction request initiated by the scheduling unit during the operation of a joint support mission. However, in this scheme, the allocation request only provides necessary structured identifiers, such as task number, resource type, allocation area, and response time window, without specifying the task content. For example, a combined arms brigade command module requests the pre-deployment of a "communication support kit (including VHF radio + lithium battery power supply)" within 15 minutes. The support resource frozen mapping vector refers to a cognitive structured representation constructed on a per-resource basis, containing its unique physical identifier (e.g., equipment code CBX-201), predicted allocation main path, activation score, location information, and context-aware features, used to determine its suitability when matching potential future allocation requests. Matching refers to comparing the structured features in the allocation request (e.g., required category, region, response window, etc.) with the corresponding fields in the frozen mapping vector in the cognitive domain cache. The matching logic includes location index consistency verification, time window overlap judgment, and modular composition semantic comparison. For example, if the allocation request requires "allocating one set of communication support boxes from the Southeast Theater Command warehouse within 15 minutes", the system will search the cognitive domain cache for communication support boxes that are highly consistent with the request. If a match is found, the box will be marked as a waiting object for release.

[0068] The physical outbound path refers to the specific picking path for resources within the current storage area, from the storage location to the loading port, including accessible passageways, physical distance, picking order, and reserved passageway resources. This solution employs an activation mechanism: once a match is successful, the frozen static physical state of the resource becomes active, triggering the physical domain path scheduler to execute the optimal outbound path planning. For example, CBX-201 moves from the 3rd column, 2nd floor of Warehouse A in the Southeast Theater Command to Loading Port 1, avoiding heavy-load passageways throughout. The local physical domain instruction mapper is an edge device deployed in the warehouse control center. Its responsibility is to map the resource identifiers and path information confirmed by the upper layer into specific picking instructions, driving automated execution units (such as AGVs and rack lifts) to complete the outbound action according to anonymized behavior flow. The mapper generates a set of operation instructions, such as: "Target ID: CBX-201", "Starting point: Warehouse A-3-2", "Target location: Loading Port 1", "Path: R2→L1→D3", forming a structured picking behavior response flow. The picking behavior response flow is an operation sequence automatically generated by the instruction mapper to drive physical picking equipment to complete resource outbound. This response flow does not contain explicit identifiers such as task codes and allocation reasons. It only completes action scheduling by combining the target resource number and path status, thereby improving the concealment of material allocation and the ability to withstand simulation.

[0069] In one possible implementation, upon receiving a transfer request, the request is matched against the frozen resource mapping vector in the cognitive domain cache. If a match is successful, the physical outbound path of the corresponding frozen resource is activated. A picking action response flow is generated through the local physical domain instruction mapper. Specifically, this includes: deconstructing the transfer request to extract the task instruction number, transfer node identifier, required resource category, task initiation timestamp, preset response time limit, region label, and operational phase parameters to form a transfer request feature vector; and performing spatial location matching based on a unique index key and physical identifier in the cognitive domain cache, based on identification... The system filters the execution time window based on the timestamp and task initiation timestamp, aligns the execution logic semantics based on the required resource categories and modular composition structure, and calculates the matching confidence score. If the matching confidence score is higher than the set matching threshold, the physical outbound path of the corresponding resource is activated. The physical domain picking path planner is invoked to generate a picking trajectory based on the physical storage location of the resource, the warehouse density distribution map, the adjacent path impedance function, and the outbound priority table. The picking trajectory and resource identification information are input into the local physical domain instruction mapper to generate a picking behavior response flow, and the physical outbound operation of the resource is completed in an anonymous operation sequence manner.

[0070] Specifically, firstly, the allocation request is deconstructed to extract its structured parameters, forming a feature vector. This feature vector is denoted as:

[0071] Qreq =[ID task ID node C type ,T req ,T resp ,L zone ,P phase ]

[0072] Among them: ID task Indicates the task instruction number; ID node Indicates the allocation node identifier; C type Indicates the category of protected resources; T req Indicates the task initiation timestamp; T resp Indicates the preset response time limit; L zone Indicates the region label; P phase Indicates parameters for the operational phase.

[0073] Taking "theater mobile communications mission" as an example, its allocation request feature vector might be:

[0074] Q req =[OP-EX-2841,NODE-7,Communication Assurance,1723435536,180,Southeast-02,Expand].

[0075] Secondly, multi-dimensional matching is performed on the allocation request feature vector and the guaranteed resource freeze mapping vector in the cognitive domain cache. Spatial location matching is based on the unique index key K. res Physical identifier ID of the security resources res Completed; Time window filtering calculation of time difference ΔT = |T req -T id |, where T id To ensure the identification timestamp for resource freezing, if ΔT≤T resp This satisfies the time window requirement; logical semantic alignment uses the category field and modular composition structure for semantic similarity modeling. The matching confidence score is:

[0076] S match =α·δ pos +β·δ time +γ·δ struct

[0077] Where: δ pos Spatial location matching score; δ time Score for time window adaptation; δ struct The product category semantic matching score; α, β, and γ are hyperparameters, satisfying α + β + γ = 1. When S match ≥θ match When the match is successful, the subsequent steps are triggered.

[0078] The third step is to invoke the physical domain picking path planner, based on the physical storage location P of the successfully matched guaranteed resource. res Storage density distribution map D map Nearest path impedance function R(p) i ,p j ) and the outbound priority table Ψ(C type Construct the optimal picking path:

[0079]

[0080] in: Represents a path node sequence; R(p) i ,p i+1 ) represents the path impedance between nodes; Ψ(C) type ) represents the priority score for the corresponding guaranteed resource release; λ is the priority adjustment weight.

[0081] Next, the optimal picking path and the guaranteed resource identification information are input into the local physical domain command mapper to generate the corresponding picking behavior response flow. Each a i This indicates specific anonymous physical execution commands (such as "lift item #C15", "move to coordinates (3,4)", "turn path #R3", "drop completed"), without including explicit information such as mission number, allocation node, or combat phase.

[0082] Finally, the picking behavior response stream is sent to the physical execution unit (such as an automated warehouse robot, rail transport platform, or AGV system) to complete the physical outbound process of the guaranteed resources in an anonymous control manner, ensuring that the resource allocation behavior remains unpredictable and untraceable to the outside world.

[0083] This solution utilizes a four-layer mechanism—feature deconstruction, cache matching, path optimization, and instruction mapping—to achieve silent activation and implicit release of security resources. This effectively avoids the problems of explicit path exposure and task structure leakage in traditional allocation processes, supporting flexible response and intelligent management of high-level security resources in non-centralized control environments.

[0084] S140. Collect vehicle status, material vibration index, temperature and humidity disturbance amount and electromagnetic interference intensity during transportation and generate path disturbance vector. Write the path disturbance vector into the path transcription cache and update the path disturbance density map in the cognitive domain cache to perform path transcription processing on the transportation process of ensuring resources.

[0085] Specifically, vehicle status refers to the dynamic operating status of transport vehicles used in the transportation of support resources, including acceleration, steering angular velocity, and path drift. This status is jointly acquired through inertial navigation units (IMUs) and BeiDou differential positioning units. For example, during the transportation of high-energy power components, rapid acceleration or braking of the vehicle may cause inertial offset of the equipment. Material vibration index refers to the intensity and spectral characteristics of vibration experienced by support resources during transportation, which can be acquired through a miniature vibration sensor array installed on the support resources. This index includes the dominant frequency band range, the vibration energy concentration area, and the location of peak abrupt changes. For example, frequent bouncing on uneven road sections can cause the dominant frequency band of high-energy power components to concentrate in the 12-18Hz range, with peak values ​​exceeding the threshold. Temperature and humidity disturbance refers to the intensity of dynamic changes in ambient temperature and relative humidity during transportation, mainly acquired in real time through composite climate sensors. For example, the insulation strength of high-energy power components decreases in high-humidity environments; if the humidity gradient abrupt change exceeds 10% / min, it is marked as a severe disturbance event. Electromagnetic interference intensity refers to the strength and trend of radio frequency interference fields present along the transportation path. The data acquisition equipment includes vehicle-mounted spectrum analyzers or electromagnetic interference detectors. Parameters include high-frequency pulse interference amplitude, modulation index fluctuation rate, and background interference density. For example, when passing near a military airport, high-frequency radar reflected pulse interference may be detected. The path disturbance vector is a joint vector composed of the aforementioned sensing data. The path transcription cache is a storage structure used to temporarily store the path disturbance vector, organized by timestamps and path node labels, for subsequent path map updates. The path disturbance density map in the cognitive domain cache records the disturbance intensity distribution of each path segment on the historical transportation path; it is a density mapping result based on the path disturbance vector.

[0086] In one possible implementation, vehicle status, material vibration index, temperature and humidity disturbance, and electromagnetic interference intensity are collected during transportation to generate a path disturbance vector. This path disturbance vector is then written into a path transcription cache and the path disturbance density map in the cognitive domain cache is updated. Specifically, this includes: collecting vehicle acceleration vectors, steering angle sequences, and path inertial deviation information using an inertial navigation unit and a BeiDou differential positioning unit to construct the vehicle status; obtaining the vibration spectrum map of the support resources using a micro-vibration sensor array and extracting the energy distribution and peak abrupt change points of the main vibration frequency band to construct the material vibration index; and using composite gas... The weather sensor unit acquires the rate of change of ambient temperature, relative humidity gradient, and amplitude of external climate fluctuations to construct temperature and humidity disturbance quantities; the on-board radio frequency analyzer collects spectral interference density, signal modulation fluctuations, and high-frequency pulse interference response to construct electromagnetic interference intensity; the vehicle status, material vibration index, temperature and humidity disturbance quantities, and electromagnetic interference intensity are time-aligned and vector-stitched to generate a path disturbance vector; the path disturbance vector is written into the path transcription cache and input into the cognitive domain cache to update the path disturbance density map, which is used to characterize the distribution of transportation disturbance intensity and frequency characteristics of the guaranteed resources.

[0087] Specifically, firstly, the acceleration vector sequence is obtained through an inertial navigation unit installed on the vehicle body. Combined with the steering angle sequence θ(t) and the continuous path coordinate sequence provided by the BeiDou differential positioning unit. Calculate the path inertia deviation information δ(t). The path inertia deviation is defined as the Euclidean distance sequence between the actual path and the desired path, and can be expressed as:

[0088]

[0089] in This is a sequence of reference points for the task path planning. The final vehicle state vector is constructed as follows:

[0090]

[0091] Secondly, the original vibration signal is collected by a miniature vibration sensor array fixed on the structural unit of the protection resource, and a fast Fourier transform is performed to generate a vibration spectrum F(f), from which the energy distribution of the main frequency band E is extracted. band With peak mutation point f peak The material vibration index vector is constructed as follows:

[0092] I vib =[E band ,f peak ]

[0093] Then, through a composite climate sensing unit installed in the transport cabin, the temperature sequence T(t) and humidity sequence H(t) are collected in real time to calculate the rate of temperature change. Humidity gradient and fluctuation range ΔT amp ,ΔH amp The temperature and humidity disturbance vector is obtained:

[0094]

[0095] Then, using an onboard radio frequency analyzer, the spectral signal P(f) is collected within the frequency range f∈[30MHz,3GHz], and the interference density ρ is extracted. em , Adjustment volatility μ mod High-frequency pulse interference intensity A pulse Construct the electromagnetic interference intensity vector:

[0096] E em =[ρ em ,μ mod A pulse ]

[0097] The above four types of disturbance indicators are aligned with a unified timestamp to ensure that all indicators use the same sampling node as a reference. Then, a vector concatenation operation is performed to generate a path disturbance vector.

[0098] D path (t)=[S veh (t),I vib (t),W env (t),E em (t)]

[0099] Subsequently, the path perturbation vector is written into the path transcription cache according to the unique identifier of the allocation task and the geographical index of the path segment to support real-time perturbation simulation at the task level. The path perturbation density map is then updated in the cognitive domain cache in the following form:

[0100]

[0101] Where φ(·) is a weighting function for calculating the contribution of perturbation intensity based on perturbation vector content, location label, and task context, and x,y are path perturbation densities. Figure 2 Dimensional coordinate index.

[0102] Ultimately, by using the above methods to collect, splice, cache, transcribe, and update the cognitive domain of the disturbance vector, the server can achieve dynamic tolerance reasoning and anomaly avoidance judgment of the transportation path based on the disturbance intensity distribution, thereby improving the robustness, concealment, and intelligent path determination capabilities in the process of ensuring resource allocation.

[0103] S150. After the guarantee resources arrive at the handover node, obtain the behavioral feedback formed by the natural action flow of the handover node, perform fuzzy matching between the behavioral feedback and the pre-stored behavioral expectation map in the frozen mapping vector of the guarantee resources, and complete the handover confirmation of the guarantee resources after successful matching, without recording explicit receipt information.

[0104] Specifically, a handover node refers to the receiving location of military support resources after their transportation mission is completed. This location may be a fixed camp, a field communication position, or a temporary command post. For example, after modular combat communication components are allocated and transported, they are delivered to a battlefield command post; this node is the handover node. Natural action flow refers to the sequence of actions generated autonomously by support resource receiving personnel during unloading, inspection, installation, or handling operations without guidance. This action flow uses sensor data to form a time-series structure, such as limb movement trajectories, operational rhythms, and contact pressure paths. Behavioral feedback refers to the feedback vector formed after structuring the natural action flow data, representing the actual expression of the operational behavior. For example, when soldiers continuously perform standardized operations such as unsealing, wiring, and signal testing of communication components during handover, the behavioral feedback vector formed by their temporal actions will include dimensions such as operational rhythm information, contact point order, and physical load changes. The support resource frozen mapping vector is a structured identifier pre-generated by the server for a specific support resource before allocation, used to track its status throughout the allocation lifecycle. It includes physical identifiers, allocation information, and behavioral expectation maps. For example, the frozen mapping vector of this communication component records its current number, encapsulation status, task association label, and expected receiving action model. A behavior expectation graph is a reference graph formed by modeling the action flow that the receiver may perform during the normal handover of resources, including the standard unloading sequence, inspection nodes, and key operation rhythms. For example, the expectation graph of a communication component might be defined as: holding → power off → unsealing → checking indicator lights → connecting test cables → power confirmation.

[0105] Fuzzy matching refers to the server using time-series comparison, dynamic time warping (DTW), or interaction state graph similarity functions to evaluate the similarity between the actual acquired behavioral feedback and the pre-stored behavioral expectation graph, allowing for deviations in duration, path, or frequency. For example, if the receiving soldier slightly changes the unsealing order but maintains the overall operational rhythm, this action can still be identified as a valid handover through fuzzy matching. Handover confirmation means that when the above fuzzy matching is successful, the server considers the support resource to have been successfully transferred to the task recipient, thus completing the task handover process. Not recording explicit receipt information emphasizes that the entire handover process does not use traditional traceable information such as allocation order numbers, signatures, or user authentication, replacing identity verification with anonymous action matching. For example, if the behavioral feedback of the communication component after being unsealed is highly consistent with the graph, the system automatically generates a "delivered" status marker without recording the executor's name or identity, to prevent the enemy from re-enacting the handover process of sensitive materials.

[0106] In one possible implementation, after the support resources arrive at the handover node, behavioral feedback consisting of the natural motion flow of the handover node is obtained. This behavioral feedback is then fuzzily matched with a pre-stored behavioral expectation map in the frozen mapping vector of the support resources. Upon successful matching, the handover of the support resources is confirmed, without recording explicit receipt information. Specifically, this includes: acquiring the full-body motion sequence of the handover personnel through a posture recognition camera and extracting limb motion vector flows; acquiring the operational rhythm information of the handover personnel through a micro-motion inertial detector, including changes in operational frequency, instantaneous pause length, and repetitive motion waveforms; and acquiring the support resources through a contact pressure matrix. The contact path data includes physical contact path, load transfer process, and center of gravity change time sequence data. The limb motion vector flow and operation rhythm information are time-aligned with the contact path data and input into the behavior fusion encoder to generate behavior feedback. The behavior feedback is then fuzzily matched with the behavior expectation map pre-stored in the resource freeze mapping vector. The matching degree is determined based on the interaction feature mapping distance, temporal similarity, and tolerance check. When the matching degree is higher than the threshold, the handover confirmation is completed. The handover confirmation process records the matching success identifier and timestamp, but does not generate a receipt number, record the identity information of the operation subject, or write the task identifier.

[0107] Specifically, in implementing this technical solution, after ensuring resources arrive at the handover node, posture recognition cameras deployed at the handover site continuously acquire full-body image sequences of the handover personnel. A motion recognition network based on skeletal point tracking and temporal modeling is used to extract limb motion vector streams. The key point positions corresponding to each frame are spatially projected to obtain a three-dimensional pose vector, constructing a temporal structure representing the standard limb behavior trajectory during the receiving process. This limb motion vector stream is used to capture key action segments such as unloading, unsealing, and transportation, and its change process is encoded through a temporal convolutional structure.

[0108] Subsequently, the handover process was sampled at high frequency using a micro-motion inertial detector to obtain operational rhythm information including micro-vibration amplitude, inertial waveform, and time interval. This information includes the operational frequency change Δf and the instantaneous pause length τ. s and the repetitive motion waveform P r These indicators form an operational rhythm vector set. The sampling frequency needs to be higher than 100Hz to capture slight pauses and inertial bounce characteristics, and an operational behavior rhythm tensor is constructed through a window sliding strategy to characterize the inherent rhythmic consistency of task execution.

[0109] In parallel, a contact pressure matrix is ​​deployed on the handover platform or material surface to record the physical contact path data between the personnel and the support resources. This contact path data includes the spatial coordinate sequence C of the contact points. p (t), load transfer process L t (t) and the curve of the center of gravity change Gc (t), where:

[0110] C p (t)={(x i ,y i ,z i )|i∈[1,N p ]}: Indicates the position of the contact point at time t;

[0111] This represents the total force change applied by the set of contacts;

[0112] r represents the position of the center of gravity at the current moment. i (t) is the position vector relative to the reference point.

[0113] The time series data constructed from the above three channels are time-aligned under a unified time coordinate to construct a standard behavior fusion input triple {A(t),R(t),P(t)}, where A(t) is the limb motion vector flow, R(t) is the operation rhythm vector, and P(t) is the contact path tensor.

[0114] The aforementioned triples are input to the behavior fusion encoder, which consists of a cross-attention fusion module and a temporal graph structure encoder. This encoder is used to establish a correlation structure between different data channels and output a behavior feedback vector B in a unified format. f(t) This indicates the overall action expression state during the current time period.

[0115] Behavioral feedback vector B f(t) The behavioral expectation map E stored in the freeze mapping vector of the support resources corresponding to the frozen state climate adjustment support resources. p(t) Fuzzy matching is performed using a weighted feature mapping distance D(B) f E p ), combined with the time series similarity index S t With tolerance check function T c Construct a matching degree evaluation function together:

[0116] M score =α·D(B f E p )+β·S t (B f E p )+γ·T c (B f E p )

[0117] Where αβγ are empirical weighting coefficients, D(B f E pUsing Euclidean distance or dynamic time-warped distance in the feature encoding space, S t T represents the degree of similarity between the two in terms of behavioral rhythm and execution sequence. c Determine whether the allowed timing offset and contact behavior error boundaries are exceeded.

[0118] If M score ≥θ m , where θ m If the threshold is met, the handover is deemed valid, and handover confirmation is completed. The system records a successful match identifier and the current system timestamp, but does not generate any task number, write the operator's identity, or task attribute markers, forming an implicit confirmation mechanism to ensure both the concealment of the allocation path and the controllability of the operation. This solution is suitable for handover scenarios of highly sensitive allocation materials such as combat communication equipment, modular fire control terminals, or emergency medical supplies with high combat readiness levels.

[0119] S160. Input the path disturbance vector of the guaranteed resources during the allocation process into the disturbance input channel of the cognitive domain allocation prediction model, update the task response frequency and path disturbance tolerance of the guaranteed resources, and complete the adaptive iteration of the cognitive state of the guaranteed resources.

[0120] Specifically, inputting the path disturbance vector during the allocation process into the disturbance input channel of the cognitive domain allocation prediction model refers to inputting the path disturbance vector as an interference factor into a specific input port of the cognitive domain allocation prediction model to predict the evolution of the behavioral response trend of the support resource. The cognitive domain allocation prediction model is a deep decision-making model for support resource management, possessing path disturbance modeling capabilities and task response state iteration capabilities. It achieves cognitive closed-loop through disturbance perception, memory update, and frequency prediction. Task response frequency is defined as the number of times a support resource is dynamically allocated and effectively executes tasks within a unit task cycle, reflecting the response strength of the support resource in multi-task scenarios. It exhibits time-varying characteristics and task load dependence. For example, if a certain type of tactical communication base station module has been called multiple times in the past 72 hours, with the response frequency gradually increasing, it can be used to predict its future allocation priority.

[0121] Path disturbance tolerance represents the ability of a resource to withstand path disturbances during a transportation mission. It is the upper bound of the permissible disturbance intensity, reflecting the physical structural stability and environmental adaptability of the resource. For example, the path disturbance tolerance of a specialized medical cold chain container may be highly sensitive to vibration frequency energy distribution and temperature gradient fluctuations; exceeding a certain threshold necessitates repackaging or component replacement. Adaptive iteration of the resource's cognitive state refers to using the aforementioned input path disturbance vector to guide the cognitive domain allocation prediction model to adjust the resource's internal cognitive state vector, including its predicted response frequency distribution and path disturbance tolerance range. This allows the resource to be more adaptable to changes in environmental conditions during subsequent allocation missions. For example, satellite positioning components that perform stably in missions with continuous high electromagnetic interference will have their anti-interference capability tags strengthened in their cognitive state, increasing their priority allocation probability under complex missions.

[0122] The mechanism embodies a dynamic cognitive correction strategy based on perturbation learning feedback, achieving a closed loop of experience-driven and environmental feedback in the resource allocation system. It can be widely applied to deployment scenarios at the edge of the theater, complex logistics coordination systems, and remote modular scheduling networks.

[0123] In one possible implementation, the path disturbance vector of the guaranteed resource during the allocation process is input into the disturbance input channel of the cognitive domain allocation prediction model to update the task response frequency and path disturbance tolerance of the guaranteed resource, thereby completing the adaptive iteration of the cognitive state of the guaranteed resource. Specifically, this includes: generating a fused disturbance vector based on the path disturbance vector and behavioral feedback; inputting the fused disturbance vector into the disturbance input channel of the cognitive domain allocation prediction model, which is then processed by a disturbance spectrum reconstructor, a response variogram function, and a tolerance learning submodule to generate a task response frequency update factor and a path disturbance tolerance update factor; writing the task response frequency update factor and the path disturbance tolerance update factor into the state record structure of the guaranteed resource in the cognitive domain cache, and feeding them back to the multi-task temporal coding network to achieve dynamic adaptive correction of the guaranteed resource allocation intention.

[0124] Specifically, in the process of generating a fused perturbation vector based on the path perturbation vector and behavioral feedback, the path perturbation vector is first subjected to temporal window resampling and feature normalization to ensure temporal consistency and dimensional comparability with the behavioral feedback. The behavioral feedback originates from the natural motion flow encoding results generated by the support resources at the handover node, specifically including limb motion vector flow, operational rhythm information, and contact path data, which are then processed by a behavioral fusion encoder to obtain a multimodal compressed vector. The aforementioned path perturbation vector and behavioral feedback vector are then concatenated in both temporal and spatial dimensions to obtain the fused perturbation vector, which serves as a joint description of the environmental perturbation and interactive response state of the support resources during the current allocation task.

[0125] After the fused perturbation vector is input into the perturbation input channel of the cognitive domain allocation prediction model, it first enters the perturbation spectrum reconstructor module. The perturbation spectrum reconstructor performs a frequency domain transformation on the perturbation dimension based on a sliding Fourier window function, capturing the rhythm of perturbation changes and the main perturbation interval. Let the perturbation input sequence be D(t), and its spectrum function is expressed as:

[0126]

[0127] Where S(f) represents the amplitude of the disturbance spectrum at frequency f, reflecting the interference density of the security resources under different disturbance dimensions.

[0128] Subsequently, the response variability function calculates the response variability intensity based on the residual sequence of the perturbation spectrum and the historical mission response frequencies, generating a mission response frequency update factor. Let the original mission response frequency be R0, the target frequency under perturbation conditions be R1, and the update factor be:

[0129]

[0130] Where λ1 is the response frequency adjustment weighting coefficient, and T is the length of the disturbance influence time window.

[0131] The tolerance learning submodule constructs a regression model between disturbance intensity and path response results to calculate the adjustable range of path disturbance tolerance and generate a path disturbance tolerance update factor. Let the current path disturbance tolerance be C0 and the target disturbance tolerance be C1, the update factor is defined as:

[0132] ΔC = λ²·(C1 - C0)

[0133] Where λ2 is the disturbance tolerance learning rate, which is dynamically set according to the type of protected resources and the historical disturbance adaptability.

[0134] Finally, the task response frequency update factor and path perturbation tolerance update factor are written into the state record structure of the guaranteed resource cache in the cognitive domain. This structure includes static physical identifiers, allocation history tags, perturbation tolerance estimation sequences, and response frequency distribution functions. This state update is then fed back to the multi-task temporal coding network previously used to generate allocation intentions, updating its state vector layer and task scoring channel weights to achieve dynamic adaptive correction of the guaranteed resource allocation intentions, forming a complete perturbation perception and cognitive correction closed loop.

[0135] This application also provides a resource protection intelligent management device, referring to... Figure 2 , Figure 2This is a schematic diagram of a module for an intelligent management device for support resources provided in an embodiment of this application. The device is a server, comprising an acquisition module 21 and a processing module 22. The acquisition module 21 acquires multi-dimensional feature vectors corresponding to support resources and inputs these vectors into a multi-task temporal coding network. Combining the command link perturbation signal sequence and the situational disturbance index sequence, it outputs a dynamic allocation intent matrix for support resources. The processing module 22 identifies support resources in the dynamic allocation intent matrix whose activation exceeds a set threshold, marking them as frozen-state reserve support resources. It constructs a freeze mapping vector for these frozen-state reserve support resources and writes it into the cognitive domain cache. The frozen-state reserve support resources remain in a static physical state and do not undergo outbound operations. Upon receiving an allocation request, the processing module 22 matches the allocation request with the frozen mapping vector in the cognitive domain cache. If a match is successful, it activates the physical outbound path for the corresponding frozen-state reserve support resource and generates a picking behavior response stream through a local physical domain instruction mapper. The picking behavior response flow is used to enable the support resources to complete the picking operation without explicitly marking task information; the processing module 22 collects vehicle status, material vibration index, temperature and humidity disturbance amount and electromagnetic interference intensity during transportation and generates path disturbance vector. It writes the path disturbance vector into the path transcription cache and updates the path disturbance density map in the cognitive domain cache to perform path transcription processing on the transportation process of the support resources; after the support resources arrive at the handover node, the processing module 22 obtains the behavior feedback composed of the natural action flow of the handover node, performs fuzzy matching between the behavior feedback and the pre-stored behavior expectation map in the frozen mapping vector of the support resources. After successful matching, the handover confirmation of the support resources is completed, and no explicit receipt information is recorded; the processing module 22 inputs the path disturbance vector of the support resources in the allocation process into the disturbance input channel of the cognitive domain allocation prediction model, updates the task response frequency and path disturbance tolerance of the support resources, and completes the adaptive iteration of the cognitive state of the support resources.

[0136] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0137] This application also provides an electronic device, with reference to... Figure 3 , Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include: at least one processor 31, at least one network interface 34, a user interface 33, a memory 35, and at least one communication bus 32.

[0138] The communication bus 32 is used to enable communication between these components.

[0139] The user interface 33 may include a display screen and a camera. Optionally, the user interface 33 may also include a standard wired interface and a wireless interface.

[0140] The network interface 34 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0141] The processor 31 may include one or more processing cores. The processor 31 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 35, and by calling data stored in the memory 35. Optionally, the processor 31 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 31 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor 31.

[0142] The memory 35 may include random access memory (RAM) or read-only memory. Optionally, the memory 35 may include a non-transitory computer-readable storage medium. The memory 35 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 35 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the various method embodiments described above, etc.; the data storage area may store data involved in the various method embodiments described above, etc. Optionally, the memory 35 may also be at least one storage device located remotely from the aforementioned processor 31. Figure 3 As shown, the memory 35, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for intelligent resource management.

[0143] exist Figure 3 In the electronic device shown, the user interface 33 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 31 can be used to call an application program stored in the memory 35 that provides a method for intelligent management of security resources. When executed by one or more processors, the electronic device executes one or more methods as described in the above embodiments.

[0144] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0145] This application also provides a computer-readable storage medium storing instructions. When executed by one or more processors, these instructions cause an electronic device to perform one or more of the methods described in the above embodiments.

[0146] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0147] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0148] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0149] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0150] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0151] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for intelligent management of resource security, characterized in that, The method includes: Obtain the multi-dimensional feature vector corresponding to the support resources, and input the multi-dimensional feature vector into the multi-task temporal coding network. Combine the command link micro-disturbance signal sequence and the situational disturbance index sequence to output the dynamic allocation intention matrix of the support resources. In the dynamic allocation intent matrix, identify the reserve resources whose activation exceeds a set threshold, mark them as frozen reserve resources, construct the reserve resource freeze mapping vector of the frozen reserve resources, and write the reserve resource freeze mapping vector into the cognitive domain cache. The frozen reserve resources remain in a static physical state and do not perform outbound operations. Upon receiving an allocation request, the allocation request is matched with the frozen mapping vector of the guaranteed resource in the cognitive domain cache. If the match is successful, the physical outbound path of the corresponding frozen guaranteed resource is activated, and a picking behavior response stream is generated through the local physical domain instruction mapper. The picking behavior response stream is used to enable the guaranteed resource to complete the picking operation in the state where no task information is explicitly marked. The system collects vehicle status, material vibration index, temperature and humidity disturbance amount and electromagnetic interference intensity during transportation and generates path disturbance vector. The path disturbance vector is written into the path transcription cache and the path disturbance density map in the cognitive domain cache is updated to perform path transcription processing on the transportation process of the guaranteed resources. After the support resources arrive at the handover node, the behavioral feedback formed by the natural action flow of the handover node is obtained. The behavioral feedback is then fuzzily matched with the pre-stored behavioral expectation map in the frozen mapping vector of the support resources. If the match is successful, the handover confirmation of the support resources is completed, but no explicit receipt information is recorded. The path disturbance vector of the guaranteed resource during the allocation process is input into the disturbance input channel of the cognitive domain allocation prediction model to update the task response frequency and path disturbance tolerance of the guaranteed resource, thereby completing the adaptive iteration of the cognitive state of the guaranteed resource. The process of acquiring the multi-dimensional feature vectors corresponding to the support resources and inputting these vectors into a multi-task temporal coding network, combined with the command link perturbation signal sequence and the situational disturbance index sequence, to output the dynamic allocation intent matrix of the support resources, specifically includes: constructing a static structural feature set of the support resources, which includes the support resources' combat readiness level identifier, historical allocation record sequence, modular composition structure, physical storage location coordinates, corresponding combat unit configuration relationship, and typical usage task type labeling; acquiring the allocation behavior trajectory vector of the support resources within a fixed time window, which uses timestamps as indexes to record the frequency of call and path history of the support resources; and collecting coverage data. The command link perturbation signal sequence of the support resource storage node includes link activation frequency, low-intensity signaling spectrum, and command fragment frequency; the situational disturbance index sequence covering the area where the support resource is located is obtained, the situational disturbance index sequence is used to reflect the task load change rate, threat level diffusion value, and the aggregation state of the material call frequency of adjacent combat units; the static structural feature set, the allocation behavior trajectory vector, the command link perturbation signal sequence, and the situational disturbance index sequence are time-position nested and aligned to obtain a standardized time-series input tensor, and the standardized time-series input tensor is input into the multi-task time-series coding network to generate the dynamic allocation intention matrix of the support resource.

2. The intelligent resource management method according to claim 1, characterized in that, The process of identifying and marking reserve resources with activation levels exceeding a set threshold in the dynamic allocation intent matrix as frozen reserve resources, constructing a reserve resource freeze mapping vector for the frozen reserve resources, and writing the reserve resource freeze mapping vector into the cognitive domain cache specifically includes: The allocation probability vector of each of the guaranteed resources in the dynamic allocation intention matrix is ​​standardized, the allocation peak probability value is extracted, and the allocation activation score is calculated by combining the task adaptability vector and the path response sensitivity vector. The allocation activation score is compared with a preset threshold. If it exceeds the preset threshold, the corresponding guaranteed resource is marked as the frozen reserve resource. Construct a frozen resource mapping vector for the frozen state weather support resources. The frozen resource mapping vector includes the physical identification information of the corresponding support resources, the allocation probability vector, the allocation activation score, the identification timestamp, the regional location index, the link perturbation feature fragments, and the environmental state fingerprint data. The resource freeze mapping vector is written into the cognitive domain cache, and a unique index key is generated by static identifier and identification timestamp to ensure the uniqueness of the resource freeze mapping vector in the cognitive domain cache.

3. The intelligent resource management method according to claim 2, characterized in that, Upon receiving a transfer request, the transfer request is matched with the frozen mapping vector of the guaranteed resource in the cognitive domain cache. If a match is successful, the physical outbound path of the corresponding frozen guaranteed resource is activated, and a picking behavior response flow is generated through the local physical domain instruction mapper. Specifically, this includes: The allocation request is deconstructed to extract the task instruction number, allocation node identifier, required resource category, task initiation timestamp, preset response time limit, region label and combat phase parameters to form an allocation request feature vector. In the cognitive domain cache, spatial location matching is performed based on the unique index key and physical identifier, time window filtering is performed based on the identification timestamp and the task initiation timestamp, logical semantic alignment is performed based on the required resource categories and modular composition structure, and a matching confidence score is calculated. If the matching confidence score is determined to be higher than the set matching threshold, the physical outbound path of the corresponding guaranteed resource is activated; The physical domain picking path planner is invoked to generate a picking trajectory based on the physical storage location of the guaranteed resources, the warehouse density distribution map, the adjacent path impedance function, and the outbound priority table. The picking trajectory and the guarantee resource identification information are input into the local physical domain instruction mapper to generate the picking behavior response stream, and the physical outbound operation of the guarantee resource is completed in an anonymous operation sequence manner.

4. The intelligent resource management method according to claim 1, characterized in that, The process of collecting vehicle status, material vibration index, temperature and humidity disturbance, and electromagnetic interference intensity during transportation and generating a path disturbance vector, writing the path disturbance vector into the path transcription cache, and updating the path disturbance density map in the cognitive domain cache specifically includes: The vehicle's state is constructed by collecting vehicle acceleration vector, steering angle sequence, and path inertial deviation information through an inertial navigation unit and a Beidou differential positioning unit. The vibration spectrum of the protected resources is obtained by using a miniature vibration sensor array, and the energy distribution and peak change points of the main vibration frequency band are extracted to construct the vibration index of the materials. The ambient temperature change rate, relative humidity gradient and external climate fluctuation amplitude are obtained by the composite climate sensing unit to construct the temperature and humidity disturbance quantity. The electromagnetic interference intensity is constructed by collecting spectral interference density, signal modulation fluctuation, and high-frequency pulse interference response using an on-board radio frequency analyzer. The vehicle status, the material vibration index, the temperature and humidity disturbance amount, and the electromagnetic interference intensity are time-aligned and vector-concatenated to generate the path disturbance vector. The path disturbance vector is written into the path transcription cache and input into the cognitive domain cache to update the path disturbance density map. The path disturbance density map is used to characterize the distribution of transportation disturbance intensity and disturbance frequency characteristics of the guaranteed resources.

5. The intelligent resource management method according to claim 1, characterized in that, After the support resources arrive at the handover node, the process involves obtaining behavioral feedback from the natural action flow of the handover node, performing a fuzzy match between the behavioral feedback and the pre-stored behavioral expectation map in the frozen mapping vector of the support resources, and confirming the handover of the support resources upon successful matching without recording explicit receipt information. Specifically, this includes: The system uses a posture recognition camera to capture the full-body motion sequence of the personnel during the handover and extracts the limb motion vector flow. The operation rhythm information of the handover personnel is obtained by a micro-motion inertial detector. The operation rhythm information includes changes in operation frequency, instantaneous pause length, and repetitive action waveform. The contact path data of the protection resource is obtained through the contact pressure matrix. The contact path data includes physical contact path, load transfer process and center of gravity change time sequence data. The limb motion vector stream, the operation rhythm information, and the contact path data are time-aligned and input into the behavior fusion encoder to generate the behavior feedback. The behavioral feedback is fuzzily matched with the pre-stored behavioral expectation map in the resource freeze mapping vector. The matching degree is determined based on the interaction feature mapping distance, temporal similarity and tolerance verification. When the matching degree is higher than the threshold, the handover confirmation is completed. The handover confirmation process records the matching success identifier and timestamp, but does not generate a receipt number, record the identity information of the operation subject, or write the task identifier.

6. The intelligent resource management method according to claim 1, characterized in that, The step of inputting the path perturbation vector of the guaranteed resource during the allocation process into the perturbation input channel of the cognitive domain allocation prediction model, updating the task response frequency and path perturbation tolerance of the guaranteed resource, and completing the adaptive iteration of the cognitive state of the guaranteed resource, specifically includes: A fused perturbation vector is generated based on the path perturbation vector and the behavior feedback; The fused perturbation vector is input into the perturbation input channel of the cognitive domain allocation prediction model. After processing by the perturbation spectrum reconstructor, response variability function and tolerance learning submodule, the task response frequency update factor and path perturbation tolerance update factor are generated. The task response frequency update factor and the path disturbance tolerance update factor are written into the state record structure of the guaranteed resource in the cognitive domain cache and fed back to the multi-task temporal coding network to realize the dynamic adaptive correction of the guaranteed resource allocation intention.

7. A resource management intelligent management device, characterized in that, The device is used to execute the intelligent resource management method as described in any one of claims 1 to 6, the device comprising an acquisition module (21) and a processing module (22), wherein, The acquisition module (21) is used to acquire the multi-dimensional feature vector corresponding to the support resources, and input the multi-dimensional feature vector into the multi-task time-series coding network, and combine the command link micro-disturbance signal sequence and the situation disturbance index sequence to output the dynamic allocation intention matrix of the support resources; The processing module (22) is used to identify the protection resources whose activation degree exceeds a set threshold in the dynamic allocation intention matrix, mark them as frozen waiting protection resources, construct the protection resource freezing mapping vector of the frozen waiting protection resources, and write the protection resource freezing mapping vector into the cognitive domain cache. The frozen waiting protection resources maintain a static physical state and do not perform outbound operation. The processing module (22) is also used to match the allocation request with the frozen mapping vector of the security resource in the cognitive domain cache after receiving the allocation request. After successful matching, the physical outbound path of the corresponding frozen security resource is activated, and a picking behavior response stream is generated through the local physical domain instruction mapper. The picking behavior response stream is used to enable the security resource to complete the picking operation in the state where the task information is not explicitly marked. The processing module (22) is also used to collect vehicle status, material vibration index, temperature and humidity disturbance amount and electromagnetic interference intensity during transportation and generate path disturbance vector, write the path disturbance vector into the path transcription cache and update the path disturbance density map in the cognitive domain cache, so as to perform path transcription processing on the transportation process of the guarantee resources. The processing module (22) is also used to obtain the behavioral feedback formed by the natural action flow of the handover node after the guarantee resource arrives at the handover node, perform fuzzy matching between the behavioral feedback and the pre-stored behavioral expectation map in the frozen mapping vector of the guarantee resource, and complete the handover confirmation of the guarantee resource after successful matching, without recording explicit receipt information. The processing module (22) is also used to input the path disturbance vector of the guaranteed resource during the allocation process into the disturbance input channel of the cognitive domain allocation prediction model, update the task response frequency and path disturbance tolerance of the guaranteed resource, and complete the adaptive iteration of the cognitive state of the guaranteed resource.

8. An electronic device, characterized in that, The electronic device includes a processor (31), a memory (35), a user interface (33), and a network interface (34). The memory (35) is used to store instructions. The user interface (33) and the network interface (34) are both used to communicate with other devices. The processor (31) is used to execute the instructions stored in the memory (35) to cause the electronic device to perform the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1 to 6.

Images