A Cross-System Cooperative Control Method and System Based on Digital Threads

Abstract

This invention provides a cross-system collaborative control method and system based on digital threads. By constructing a cross-system interaction architecture, real-time data transmission and command interaction are established between multiple business systems through a distributed node communication protocol. Based on the cross-system interaction architecture, dynamic operational characteristic data of each business system is collected to generate cross-system collaborative control logic. This logic is converted into standardized control commands, distributed to the corresponding business systems, and executed in real-time collaborative control, while collecting state feedback characteristics during the control process. Dynamic deviation analysis is performed on the state feedback characteristics to generate collaborative control strategy adjustment parameters. These parameters are then synchronized to each business system through the cross-system interaction architecture. This invention improves the efficiency of cross-system collaborative control.

Description

Technical Field

[0001] This invention relates to the field of digital control, and more specifically, to a cross-system collaborative control method and system based on digital threads. Background Technology

[0002] With the development of industrial digitalization and system integration technologies, cross-system collaborative control technology has emerged. This technology refers to the ability of multiple independent business systems to collaboratively execute tasks, allocate resources, and optimize operations under a unified goal. It is widely applied in complex industrial scenarios such as intelligent manufacturing, intelligent transportation, and the energy internet. Its core lies in solving the problems of data interaction, task scheduling, and dynamic adaptation between different business systems. Currently, common cross-system collaborative control technologies typically achieve data transmission between systems by establishing fixed communication links. Control commands are generated based on preset rules or the operating status of a single system, and then distributed to each business system for execution through a centralized node. During execution, the operating status is obtained through periodic manual inspections or system log recordings, allowing for offline adjustments to the control strategy. However, this collaborative approach, relying on fixed communication links and preset rules, struggles to adapt to the dynamically changing operating characteristics of different business systems, resulting in a low degree of matching between control commands and actual system requirements. Furthermore, the centralized node command distribution model is susceptible to single-point failures, and the lag in offline status acquisition and strategy adjustment means that resource competition and task conflicts during cross-system collaboration cannot be resolved in real time, thus affecting the overall stability and efficiency of collaborative control. Therefore, improving the dynamic adaptability and real-time response capability of cross-system collaborative control has become a current research hotspot. Summary of the Invention

[0003] This invention provides a cross-system collaborative control method and system based on digital threads.

[0004] In a first aspect, embodiments of the present invention provide a cross-system collaborative control method based on digital threads, comprising: Construct a cross-system interaction architecture to establish real-time data transmission and command interaction between multiple business systems through a distributed node communication protocol; Based on the cross-system interaction architecture, dynamic operational characteristic data of each business system are collected to generate cross-system collaborative control logic. The cross-system collaborative control logic is converted into standardized control commands, distributed to the corresponding business systems and executed in real-time collaborative control, and the status feedback characteristics during the control process are collected. Dynamic deviation analysis is performed on the state feedback characteristics to generate adjustment parameters for the collaborative control strategy; The adjustment parameters of the collaborative control strategy are synchronized to each business system through the cross-system interaction architecture.

[0005] In a second aspect, embodiments of the present invention provide a computer system, comprising: A memory, wherein a computer program is stored; A processor is used to load the computer program to implement the cross-system cooperative control method based on digital threads as described above.

[0006] This invention provides a cross-system collaborative control method based on digital threads. It achieves real-time data transmission and command interaction between multiple business systems by constructing a cross-system interaction architecture. Based on this architecture, it collects dynamic operational characteristic data and uses a multimodal state fusion algorithm to generate cross-system collaborative control logic. This allows the collaborative control logic to comprehensively reflect the operational status and interrelationships of different business systems, avoiding the limitations of single-system control logic and improving the accuracy of cross-system collaborative control. The cross-system collaborative control logic is converted into standardized control commands and distributed to the corresponding business systems for execution. State feedback characteristics during the control process are collected, and the standardized commands ensure that each business system responds to the control logic correctly. Consistent execution is achieved, while dynamic changes in the collaborative control process are captured in real time through state feedback features, providing a reliable basis for subsequent strategy adjustments. A time-series feature migration algorithm is used to perform dynamic deviation analysis on the state feedback features and generate collaborative control strategy adjustment parameters. Combined with historical collaborative control feature patterns, dynamic trend analysis of the current deviation is achieved, improving the comprehensiveness of deviation analysis and the adaptability of adjustment parameters. The collaborative control strategy adjustment parameters are synchronized to various business systems through a cross-system interaction architecture, realizing dynamic optimization of cross-system collaborative control. Through the tight connection of the entire process, efficient cooperation among various business systems in the collaborative control process is promoted, thereby improving the overall effectiveness of cross-system collaborative control. Attached Figure Description

[0007] Figure 1 This is a flowchart of a cross-system collaborative control method based on digital threads provided in an embodiment of the present invention.

[0008] Figure 2 This is a schematic diagram of the composition of a computer system provided in an embodiment of the present invention. Detailed Implementation

[0009] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0010] Please see Figure 1 , Figure 1A flowchart of a cross-system cooperative control method based on digital threads provided in an embodiment of the present invention, which can be executed by a computer system, includes the following steps: Step S100: Construct a cross-system interaction architecture and establish real-time data transmission and command interaction between multiple business systems through a distributed node communication protocol.

[0011] A cross-system interaction architecture is an architectural framework used to enable interconnection and interaction between multiple business systems, providing a unified communication and collaboration platform for each system. A distributed node communication protocol is a protocol used for communication between nodes in a distributed system, ensuring accurate and efficient data transmission between different nodes. Real-time data transmission refers to transmitting data to the target system simultaneously with its generation, ensuring data timeliness and validity. Command interaction refers to the ability for business systems to send and receive control commands to each other, thereby achieving collaborative control.

[0012] When constructing a cross-system interaction architecture, the network topology, data interface types, and communication requirements of each business system are first analyzed. A suitable distributed node communication protocol, such as MQTT (Message Queuing Telemetry Transport), is then selected. Next, corresponding communication nodes are deployed in each business system; these nodes can be hardware devices or software programs. Through these communication nodes, each business system can perform real-time data transmission and command exchange according to the rules of the distributed node communication protocol.

[0013] For example, in an enterprise information environment that includes a production system, a logistics system, and a sales system, the production system needs to transmit production progress data to the logistics and sales systems in real time so that the logistics system can arrange shipments and the sales system can keep abreast of product supply. Simultaneously, the logistics and sales systems can also send production adjustment instructions to the production system. By building a cross-system interaction architecture and adopting the MQTT protocol, efficient and stable real-time data transmission and instruction exchange can be achieved between the systems.

[0014] Step S200: Collect dynamic operation characteristic data of each business system based on the cross-system interaction architecture, and generate cross-system collaborative control logic.

[0015] Dynamic operational characteristic data refers to the various characteristic data exhibited by each business system during operation. This data reflects the system's operational status and performance. Cross-system collaborative control logic is a set of logical rules used to coordinate the work between multiple business systems. It can rationally allocate resources and arrange tasks according to the operational status and needs of each system, thereby achieving cross-system collaborative work.

[0016] When collecting dynamic operational characteristic data from various business systems based on a cross-system interaction architecture, distributed acquisition nodes within the architecture are used to collect data from each business system at preset time intervals. These distributed acquisition nodes can be hardware devices such as sensors and data acquisition cards, or data acquisition programs running in each business system. The collected dynamic operational characteristic data includes system task queue characteristics, resource usage sequence characteristics, and interaction response time characteristics.

[0017] System task queue characteristics refer to features such as the length of task queues, task type distribution, and task waiting time in a business system, reflecting the system's task load. Resource usage sequence characteristics refer to the system's usage of various resources (such as CPU, memory, and network bandwidth) over a period of time, reflecting the system's resource utilization efficiency. Interaction response time characteristics refer to the system's response time when interacting with other systems, reflecting the system's response speed and interaction performance. After collecting dynamic operational characteristic data, this data is processed and analyzed to generate cross-system collaborative control logic. This process requires comprehensive consideration of factors such as the interrelationships between various business systems, resource contention, and task coupling.

[0018] As one implementation method, step S200 involves collecting dynamic operational characteristic data from various business systems based on a cross-system interaction architecture and generating cross-system collaborative control logic. Specifically, this can be implemented as follows: steps S210~S250: Step S210: Collect dynamic operation feature data of each business system at preset time intervals through the distributed collection nodes in the cross-system interaction architecture. The dynamic operation feature data includes system task queue features, resource occupancy sequence features, and interaction response duration features.

[0019] Distributed data collection nodes are nodes used in cross-system interaction architectures to collect data from various business systems. They are distributed across these systems and can collect system operation data in real time and accurately. The preset time interval refers to a pre-defined time interval for data collection.

[0020] The collection of system task queue characteristics can be achieved by monitoring relevant information about task queues in the business system. For example, for a task queue in an operating system, information such as the queue length, task priority, and task wait time can be obtained through system calls. The collection of resource usage sequence characteristics can be achieved through system monitoring tools. For example, in Linux systems, tools such as the `top` and `vmstat` commands can be used to obtain information about CPU, memory, and disk I / O usage. The collection of interaction response time characteristics can be calculated by sending test messages between systems and recording the message sending and receiving times.

[0021] Step S220: Perform spatiotemporal alignment processing on the dynamic running feature data, map the feature data of different business systems to a unified time coordinate system, and generate a spatiotemporal aligned feature set.

[0022] Spatiotemporal alignment refers to processing feature data collected from different business systems at different times to ensure temporal consistency, thereby enabling effective comparison and analysis. A unified time coordinate system is a reference coordinate system used to unify the time of different business systems, ensuring the temporal comparability of feature data from each system. The spatiotemporally aligned feature set is the set of feature data obtained after spatiotemporal alignment processing, containing feature data from each business system under a unified time coordinate system. Since clocks in different business systems may deviate, and data collection time intervals may also differ, spatiotemporal alignment processing is necessary for dynamically running feature data. This process mainly includes steps such as extracting timestamp information, constructing a time reference axis, interpolation processing, dimension concatenation, and normalization.

[0023] As one implementation method, step S220 involves performing spatiotemporal alignment processing on the dynamic running feature data, mapping the feature data of different business systems to a unified time coordinate system, and generating a spatiotemporally aligned feature set. Specifically, this can be implemented as follows: steps S221~S225: Step S221: Extract the timestamp information from the dynamic running feature data to determine the collection time and data sampling period of the feature data of each business system.

[0024] Timestamp information refers to the timestamps contained in dynamically generated feature data, recording the specific time of data collection. The collection time refers to the specific point in time when the data was collected, and the data sampling period refers to the time interval between two consecutive data collections. When collecting dynamically generated feature data, each data point is accompanied by a timestamp. By parsing these timestamps, the collection time of the feature data for each business system can be determined. Simultaneously, by analyzing the timestamp differences between adjacent data points, the data sampling period can be calculated.

[0025] Step S222: Based on the acquisition time and data sampling period, construct a time reference axis. The smallest time unit of the time reference axis is the common divisor of the sampling periods of each business system.

[0026] A time baseline is a reference axis used to unify the time of different business systems. The smallest time unit is the common divisor of the sampling periods of each business system. By constructing a time baseline, the characteristic data of different business systems can be unified over time.

[0027] Calculate the common divisor of the sampling periods of each business system, and use this common divisor as the smallest time unit on the time reference axis. Then, determine the corresponding position on the time reference axis based on the acquisition time of the characteristic data of each business system.

[0028] Step S223: Interpolate the dynamic operation feature data of each business system, supplement the feature data of missing moments on the time reference axis, and generate feature sequences with equal time intervals.

[0029] Interpolation is the process of inserting new data points between known data points to fill in missing data. A feature sequence with equal time intervals refers to a sequence where, after interpolation, the feature data of each business system has the same time interval. Because different business systems have different sampling periods, there may be some missing moments on the timeline. Interpolation can fill in the missing feature data at these moments. Commonly used interpolation methods include linear interpolation and polynomial interpolation.

[0030] Step S224: Concatenate the time interval feature sequences of each business system according to the system identifier to obtain a three-dimensional feature tensor containing system dimension, time dimension and feature dimension, and use the three-dimensional feature tensor as a spatiotemporal aligned feature set.

[0031] A system identifier is an identifier used to uniquely identify each business system, such as system name or system number. Dimensional concatenation refers to concatenating the feature sequences of different business systems along the system dimension to form a higher-dimensional feature data structure. A three-dimensional feature tensor is a tensor containing system, time, and feature dimensions, which can comprehensively represent the feature data of each business system at different times.

[0032] Arrange the time-interval feature sequences of each business system according to the system identifier, and then concatenate them along the system dimension. For example, there are three business systems A, B, and C, whose time-interval feature sequences are FA, FB, and FC, respectively. Concatenate FA, FB, and FC along the system dimension to obtain a three-dimensional feature tensor, where the first dimension represents the system dimension, the second dimension represents the time dimension, and the third dimension represents the feature dimension.

[0033] Step S225: Normalize the three-dimensional feature tensor to eliminate the dimensional differences in feature data from different business systems, and maintain the temporal trend of each feature sequence unchanged during the normalization process.

[0034] Normalization is a process that processes feature data from different business systems to ensure they have the same dimensions and range, enabling effective comparison and analysis. Differences in dimensions refer to the different units and magnitudes of feature data from different business systems, which can affect the results of data analysis and processing. Feasible normalization methods include maximum-minimum normalization and Z-score normalization. Taking maximum-minimum normalization as an example, for each feature dimension in a three-dimensional feature tensor, its maximum and minimum values ​​are calculated. Then, the minimum value is subtracted from each data point, and the result is divided by the difference between the maximum and minimum values ​​to obtain the normalized data. During normalization, it is crucial to ensure that the temporal trend of each feature sequence remains unchanged, i.e., the relative magnitudes of the data remain constant.

[0035] Step S230: Based on the spatiotemporal alignment feature set, extract the correlation and dependency features between various business systems through the attention mechanism. The correlation and dependency features characterize the resource competition relationship and task coupling degree of different business systems in the task execution process.

[0036] Attention mechanisms are used to automatically focus on important information in large amounts of data, assigning different attention weights based on the importance of the data. Association-dependency features refer to the characteristics of the relationships and degrees of dependence between different business systems, reflecting the mutual influence of each system during task execution. Resource competition refers to the competition between different business systems when using shared resources, and task coupling refers to the degree of correlation between tasks of different business systems. Through attention mechanisms, the relationships between feature data of various business systems in a spatiotemporally aligned feature set can be automatically identified, thereby extracting association-dependency features. This process mainly includes steps such as time window segmentation, similarity weight calculation, weighted aggregation, feature sequence concatenation, and filtering.

[0037] As one implementation method, step S230, based on the spatiotemporal alignment feature set, extracts the correlation and dependency features between various business systems through an attention mechanism, which can be specifically implemented as follows: steps S231~S235: Step S231: Divide the spatiotemporal aligned feature set into multiple time window feature subsets according to the time dimension. Each time window feature subset contains feature data of a preset number of consecutive time steps.

[0038] A time window refers to a fixed-length time period defined along the time dimension. A time window feature subset refers to the set of feature data within a time window. The preset quantity refers to the number of consecutive time steps included in each time window. Based on the preset time window size, the spatiotemporally aligned feature set is segmented along the time dimension.

[0039] Step S232: For each feature subset of the time window, calculate the similarity weight between feature data of different business systems through a self-attention mechanism. The similarity weight represents the degree of correlation between feature data of two business systems within the same time window.

[0040] Similarity weight is a numerical value used to represent the degree of correlation between feature data from two business systems; a higher weight indicates a stronger correlation. For each feature subset within a time window, feature data from different business systems are input into a self-attention mechanism for calculation. The self-attention mechanism calculates similarity scores between feature data and then normalizes these scores to obtain the similarity weight. Similarity calculation methods include, for example, cosine similarity and Euclidean distance.

[0041] Step S233: Based on similarity weights, perform weighted aggregation on the feature data of each business system to generate an interaction correlation vector between business systems. The dimension of the interaction correlation vector is consistent with the number of business systems.

[0042] Weighted aggregation refers to the weighted summation of feature data from various business systems based on similarity weights to obtain a comprehensive feature representation. The interaction correlation vector is a vector representing the interaction relationships between business systems, with its dimension matching the number of business systems.

[0043] Step S234: Concatenate the interaction association vectors according to the time window order to obtain the association dependency feature sequence. Each element in the association dependency feature sequence corresponds to the business system association relationship within a time window.

[0044] The interaction and association vectors generated within each time window are arranged in the order of the time windows and then concatenated along the time dimension to obtain a sequence of association dependency features. Each element in this sequence corresponds to a business system association relationship within a time window.

[0045] Step S235: Filter the correlation dependency feature sequence to eliminate the interference of short-term fluctuations on the extraction of correlation. The filtering algorithm is a moving average algorithm, and the window size is a preset proportion of the number of time windows.

[0046] Filtering refers to data processing to eliminate noise and short-term fluctuations, making the data smoother. The moving average algorithm smooths data by calculating the average value over a fixed window. The window size refers to the length of the window used in the moving average algorithm, and the preset ratio refers to the ratio of the pre-set window size to the number of time windows.

[0047] The window size for the moving average algorithm is determined based on a preset proportion of the number of time windows. Then, a moving average is calculated on the correlation dependency feature sequence. For example, assuming the number of time windows is 100 and the preset proportion is 0.1, the window size is 10. For each element in the correlation dependency feature sequence, the average of its 10 preceding and following elements is calculated as the filtered value for that element.

[0048] Step S240: Input the associated dependency features into the pre-trained collaborative logic generation network. The collaborative logic generation network includes a feature encoding layer, a dependency inference layer and a logic output layer. The feature encoding layer converts the associated dependency features into high-dimensional semantic vectors. The dependency inference layer performs node relationship inference on the high-dimensional semantic vectors through a graph neural network. The logic output layer generates the initial framework of cross-system collaborative control logic based on the inference results.

[0049] The pre-trained collaborative logic generation network (CLRG) is a pre-trained neural network model capable of generating cross-system collaborative control logic based on input dependency features. The feature encoding layer, the first layer of the CLRG, transforms the dependency features into high-dimensional semantic vectors for subsequent processing and analysis. The dependency inference layer, the second layer, uses a graph neural network to infer node relationships from the high-dimensional semantic vectors, uncovering the relationships between various business systems. The logic output layer, the final layer, generates the initial framework of the cross-system collaborative control logic based on the inference results from the dependency inference layer.

[0050] The dependency features are input into the feature encoding layer of the collaborative logic generation network. This layer, which can employ fully connected or convolutional layers, maps the dependency features to a high-dimensional space, resulting in a high-dimensional semantic vector. This high-dimensional semantic vector is then input into the dependency inference layer, which uses a graph neural network (GNN), such as a Graph Convolutional Network (GCN), to infer node relationships within the high-dimensional semantic vector. The GNN treats each business system as a node and the relationships between them as edges. By propagating and aggregating information from nodes and edges, the dependencies between the business systems are obtained. Finally, the logic output layer generates an initial framework for cross-system collaborative control logic based on the inference results from the dependency inference layer. This framework includes basic rules for task allocation and resource scheduling among the business systems.

[0051] As one implementation method, the pre-training process of the collaborative logic generation network can be specifically implemented as follows: S2401~S2405: Step S2401: Construct a training dataset containing historical collaborative control cases. Each case contains dynamic operation feature data of the business system and the corresponding optimal collaborative control logic.

[0052] Historical collaborative control cases refer to real-world examples accumulated during past cross-system collaborative control processes. These cases include dynamic operational characteristic data of the business systems and corresponding optimal collaborative control logic. The training dataset is used to train the collaborative logic generation network, and it contains a large number of historical collaborative control cases.

[0053] Collect cross-system collaborative control cases from the past period, and organize and annotate each case. For example, record the dynamic operational characteristics of each business system in each case, such as system task queue characteristics, resource consumption sequence characteristics, and interaction response time characteristics; at the same time, record the corresponding optimal collaborative control logic, such as task allocation schemes and resource scheduling strategies. Combine these cases into a training dataset.

[0054] Step S2402: Divide the training dataset into a training set and a validation set, and use cross-validation to train the model.

[0055] Cross-validation is a method used to evaluate model performance and select optimal model parameters. It involves dividing the dataset into multiple subsets and using different subsets in turn for training and validation. The training set is the dataset used to train the model, and the validation set is the dataset used to validate the model's performance.

[0056] The training dataset is divided into training and validation sets according to a certain ratio, for example, an 8:2 ratio. Then, cross-validation is used to train the model.

[0057] Step S2403: Initialize the weight parameters of the collaborative logic generation network, set the weights of the feature encoding layer and the dependency inference layer, and set the bias of the logic output layer.

[0058] Weight parameters are learnable parameters in a neural network model that determine the model's performance and behavior. The weights of the feature encoding layer and the dependency inference layer refer to the connection weights between neurons in these two layers, while the bias of the logic output layer refers to the bias values ​​of the neurons in that layer.

[0059] Before training the collaborative logic generation network, its weight parameters need to be initialized. Random initialization can be used, for example, by randomly initializing the weight parameters using a Gaussian distribution. Simultaneously, the weights of the feature encoding layer and the dependency inference layer, as well as the bias of the logic output layer, should be set.

[0060] Step S2404: Using the dynamic running feature data in the training set as input, generate predictive collaborative control logic through network forward propagation, and calculate the edit distance loss between the predictive logic and the true optimal logic.

[0061] Forward propagation in a network refers to the process of processing input data through the layers of a neural network to obtain the output. Predictive cooperative control logic refers to the cooperative control logic generated by the cooperative logic generation network based on the dynamic operational characteristics of the input data. Edit distance loss refers to the degree of difference between the predicted cooperative control logic and the true optimal logic; it can be obtained by calculating the edit distance between the two logics.

[0062] The dynamic runtime feature data from the training set is input into the collaborative logic generation network, and the predicted collaborative control logic is obtained through forward propagation of the network. Then, the edit distance loss between the predicted logic and the true optimal logic is calculated. The edit distance can be calculated using methods such as Levenshtein distance.

[0063] Step S2405: Use an optimizer to minimize the edit distance loss, set an initial learning rate, decay the learning rate according to a preset rule, and iterate until the validation set loss no longer decreases and then stop training.

[0064] An optimizer is an algorithm used to adjust the weight parameters of a neural network model, with the goal of minimizing the loss function. The initial learning rate refers to the learning rate used by the optimizer at the start of training, which determines the step size for updating the weight parameters. Preset rules refer to the pre-defined learning rate decay rules, such as fixed-ratio decay or exponential decay.

[0065] Choose an optimizer, such as the Adam optimizer, to minimize the edit distance loss. Set an initial learning rate, for example, 0.001. Then, decay the learning rate according to a preset rule, for example, decaying it to 0.9 times its original value every 10 epochs. During training, iteratively update the model's weight parameters until the validation set loss no longer decreases.

[0066] Step S250: Perform conflict detection on the initial framework to identify resource allocation conflicts and task execution timing conflicts in the collaborative control logic. Based on the conflict detection results, adjust the task priorities and resource allocation ratios in the initial framework to obtain the final cross-system collaborative control logic.

[0067] Conflict detection refers to examining the collaborative control logic to identify resource allocation conflicts and task execution timing conflicts. Resource allocation conflicts occur when multiple business systems demand the same resource more than the available resource is available. Task execution timing conflicts occur when different tasks in the collaborative control logic conflict in their execution times; for example, task A must be completed before task B, but the logic schedules task B to execute first.

[0068] When performing conflict detection on the initial framework, a rule-matching method can be used. For example, a series of resource allocation rules and task execution timing rules can be defined, and the initial framework can be matched against these rules to identify conflicts. For detected resource allocation conflicts, the conflict can be resolved by adjusting the resource allocation ratio, such as reducing the resource allocation to certain business systems. For task execution timing conflicts, the conflict can be resolved by adjusting task priorities, such as increasing the priority of certain tasks to ensure they are executed first.

[0069] For example, in a collaborative control logic involving multiple production tasks, it was found that both Task A and Task B require the use of the same equipment, but the available time of the equipment cannot meet the needs of both tasks, resulting in a resource allocation conflict. This conflict is resolved by adjusting the resource allocation ratio between Task A and Task B, for example, reducing the equipment usage time for Task A and increasing the equipment usage time for Task B. Simultaneously, it was found that Task C must be completed before Task D, but the logic schedules Task D to execute first, resulting in a task execution timing conflict. This conflict is resolved by increasing the priority of Task C, allowing it to execute first. After conflict detection and adjustment, the final cross-system collaborative control logic is obtained.

[0070] Step S300: Convert the cross-system collaborative control logic into standardized control instructions, distribute them to the corresponding business systems and execute real-time collaborative control, and collect the status feedback characteristics during the control process.

[0071] Standardized control commands are control instructions with a unified format and specifications, which can be accurately identified and executed by various business systems. Real-time collaborative control refers to the real-time coordination and control of task execution among various business systems to achieve cross-system collaborative work. Status feedback characteristics refer to the operational status characteristic data fed back by various business systems during the control process, which can reflect the actual operating status of the system.

[0072] When converting cross-system collaborative control logic into standardized control commands, the logic needs to be parsed and processed, transforming rules such as task allocation and resource scheduling into specific control commands. Then, these standardized control commands are distributed to the corresponding business systems through a cross-system interaction architecture. Upon receiving the control commands, each business system executes the corresponding operations and collects real-time status feedback characteristics during the control process.

[0073] As one implementation method, step S300 involves converting the cross-system collaborative control logic into standardized control commands, distributing them to the corresponding business systems, executing real-time collaborative control, and collecting state feedback characteristics during the control process. Specifically, this can be implemented as follows: steps S310~S340: Step S310: Analyze the task execution process and resource allocation rules in the cross-system collaborative control logic, and determine the business system identifier and operation type corresponding to each control node.

[0074] Parsing refers to the analysis and processing of cross-system collaborative control logic to extract key information. Task execution flow refers to the task execution sequence and steps specified in the collaborative control logic. Resource allocation rules refer to the resource allocation scheme specified in the collaborative control logic. A control node is a specific execution point in the collaborative control logic; each control node corresponds to a specific operation. The business system identifier is an identifier used to uniquely identify each business system, and the operation type refers to the type of operation performed by the control node, such as data read / write or task scheduling.

[0075] The cross-system collaborative control logic is parsed to identify the task execution flow and resource allocation rules. For each control node, its corresponding business system identifier and operation type are determined. For example, in a collaborative control logic, task A is specified to be executed by business system A, task B is executed by business system B, and task A requires certain resources. By parsing this logic, the business system identifier and operation type corresponding to each control node are determined. For instance, the control node for task A corresponds to business system A, and the operation type is task execution; the control node for resource allocation corresponds to the resource management system, and the operation type is resource allocation.

[0076] Step S320: Based on the preset instruction conversion template, the task execution process is converted into structured control instructions. The structured control instructions include an instruction header, an instruction body, and a checksum. The instruction header stores the business system identifier and instruction priority, the instruction body contains specific operation parameters and execution time range, and the checksum is used to ensure the integrity of instruction transmission.

[0077] The preset instruction conversion template is a predefined template used to convert task execution flows into structured control instructions, specifying the format and content of the instructions. Structured control instructions are control instructions with a fixed format and structure, capable of being accurately identified and executed by various business systems. The instruction header is part of the structured control instruction and contains information such as the business system identifier and instruction priority. The instruction body is the core part of the structured control instruction, containing information such as specific operation parameters and execution time range. The checksum is a code used to verify the integrity of the instruction transmission and can be calculated from the instruction content.

[0078] Based on a pre-defined instruction conversion template, each control node in the task execution flow is converted into a corresponding structured control instruction. The instruction header stores the business system identifier and instruction priority; for example, the header of a control instruction for business system A stores the identifier of business system A and the priority of that instruction. The instruction body stores the specific operation parameters and execution time range; for example, the operation parameters for task A are task input data, and the execution time range is from 10:00:00 to 10:30:00. A checksum is calculated using a Cyclic Redundancy Check (CRC) algorithm and added to the instruction.

[0079] As one implementation method, step S320, based on a preset instruction conversion template, converts the task execution flow into structured control instructions, which can be implemented as follows: steps S321~S324: Step S321: Call the preset instruction template library and match the corresponding instruction template according to the operation type in the task execution process. The instruction template library contains three basic templates: data read and write, calculation scheduling, and resource allocation.

[0080] The pre-built instruction template library is a pre-established library for storing various instruction templates, including those for different operation types. Operation type refers to the type of operation performed by each control node in the task execution flow, such as data read / write, computation scheduling, and resource allocation. Based on the operation type in the task execution flow, the corresponding instruction template is selected from the instruction template library. For example, for data read / write operations, a data read / write instruction template is selected; for computation scheduling operations, a computation scheduling instruction template is selected; and for resource allocation operations, a resource allocation instruction template is selected.

[0081] Step S322: Parse the parameter constraints in the task execution flow and map the parameter constraints to the variable fields of the instruction template. The parameter constraints include the upper limit of resource usage, the task execution deadline, and the data transmission bandwidth limit.

[0082] Parameter constraints refer to the restrictions imposed on operational parameters during task execution, such as resource usage limits, task execution deadlines, and data transmission bandwidth limits. Variable fields are fields in the instruction template that can be populated according to specific circumstances.

[0083] Parse the parameter constraints in the task execution flow and map them to variable fields in the instruction template. For example, in a resource allocation instruction template, there is a variable field representing the resource allocation amount. Based on the resource usage limit in the task execution flow, fill the variable field in the instruction template with that limit value.

[0084] Step S323: Perform syntax validation on the mapped instruction template to check whether the parameter format conforms to the business system interface specification. If there is a format error, return an error message and terminate the conversion process. If the format is correct, generate a unique instruction identifier. The unique instruction identifier is generated based on the business system identifier, instruction generation timestamp, and random number combination, and is used to track the execution status of the instruction in the cross-system collaborative control process.

[0085] Syntax validation refers to checking the instruction template to ensure that the parameter format conforms to the business system interface specification. The business system interface specification refers to the requirements of the business system for the format and content of input instructions. An instruction unique identifier is an identifier used to uniquely identify each instruction; it helps track the execution status of instructions during cross-system collaborative control.

[0086] The mapped instruction template undergoes syntax validation to check if the parameter formats conform to the business system interface specifications. For example, it checks if the resource allocation is a positive integer and if the task execution deadline conforms to a date and time format. If a format error is found, an error message is returned and the conversion process terminates. If the format is correct, a unique identifier for the instruction is generated based on the business system identifier, the instruction generation timestamp, and a random number combination.

[0087] Step S324: Assemble the instruction unique identifier, business system identifier, operation parameters and check code into a structured control instruction in a set order, wherein the check code is calculated by a cyclic redundancy check algorithm.

[0088] The instruction's unique identifier, business system identifier, operation parameters, and checksum are assembled in a predetermined order to form a structured control instruction. Cyclic Redundancy Check (CRC) is a commonly used verification algorithm that calculates the checksum based on the instruction content.

[0089] For example, the set order is: instruction unique identifier, business system identifier, operation parameter, and check code. The instruction unique identifier "SystemA_20240101100000_123", the business system identifier "SystemA", the operation parameter "{resource_amount:100,task_deadline:'2024-01-0111:00:00'}", and the check code "ABCD" calculated by the cyclic redundancy check algorithm are assembled into the structured control instruction "SystemA_20240101100000_123_SystemA_{resource_amount:100,task_deadline:'2024-01-0111:00:00'}_ABCD".

[0090] Step S330: Through the instruction distribution module of the cross-system interaction architecture, the structured control instructions are routed to the corresponding nodes according to the business system identifier. During the routing process, a load balancing algorithm is used to allocate instruction transmission channels. After receiving the structured control instructions, each business system executes the corresponding operation and collects the status feedback data in real time during the control process. The status feedback data includes task completion progress, actual resource usage, and system response latency.

[0091] The instruction distribution module is used to distribute control instructions in the cross-system interaction architecture. It routes instructions to the corresponding nodes based on the business system identifier. The load balancing algorithm can rationally allocate instructions according to the load of each channel to improve the efficiency and reliability of instruction transmission. Status feedback data refers to the operational status data returned by each business system during the control process, which reflects the actual operating status of the system.

[0092] The instruction distribution module of the cross-system interaction architecture routes structured control instructions to corresponding nodes according to the business system identifier. During the routing process, a load balancing algorithm is used to allocate instruction transmission channels to ensure efficient and reliable instruction transmission. After receiving the structured control instructions, each business system executes the corresponding operation and collects status feedback data in real time during the control process. For example, for a task execution instruction, the business system executes the task and collects status feedback data such as task completion progress, actual resource usage, and system response latency in real time.

[0093] As one implementation method, in step S330, the load balancing algorithm allocates instruction transmission channels, which can be specifically implemented as follows: steps S331~S335: Step S331: Monitor the current load rate of each transmission channel in the cross-system interaction architecture in real time. The load rate is defined as the ratio of the channel's used bandwidth to the total bandwidth.

[0094] Real-time monitoring refers to the real-time monitoring and updating of the load status of each transmission channel in a cross-system interactive architecture. Load rate is the ratio of the channel's used bandwidth to the total bandwidth, reflecting the channel's load level. By deploying monitoring devices or programs in the cross-system interactive architecture, the used bandwidth and total bandwidth of each transmission channel are monitored in real time. The load rate of each channel is calculated; for example, if channel A has a used bandwidth of 10 Mbps and a total bandwidth of 100 Mbps, then the load rate of channel A is 10%.

[0095] Step S332: Classify the structured control instructions to be transmitted into three categories based on the characteristics of instruction data volume: Category 1, Category 2, and Category 3 instructions. The classification is based on the statistical interval of the historical instruction data volume distribution.

[0096] Instruction data volume characteristics refer to the amount of data contained in structured control instructions. The statistical interval of historical instruction data volume distribution refers to different data volume intervals divided based on the distribution of instruction data volume over a past period.

[0097] Statistical analysis of historical instruction data volume is performed to divide it into different data volume ranges. For example, instructions with data volume less than 1KB are classified into Category 1 instructions, instructions with data volume between 1KB and 10KB are classified into Category 2 instructions, and instructions with data volume greater than 10KB are classified into Category 3 instructions. Based on these classification criteria, the structured control instructions to be transmitted are categorized.

[0098] Step S333: Set channel load rate thresholds for different types of instructions. Instructions with larger data volume characteristics in the classification results correspond to lower thresholds to ensure that instructions with larger data volume characteristics use low-load channels first.

[0099] Channel load rate threshold refers to the upper limit of channel load rate set for different types of instructions. Instructions with larger data volume characteristics correspond to lower thresholds to ensure that these instructions can preferentially use low-load channels, thereby improving the efficiency of instruction transmission.

[0100] Step S334: Traverse all transmission channels and select channels with load rates lower than the corresponding thresholds and bandwidths that meet the command transmission requirements as candidate channels.

[0101] Iterate through all transmission channels and check whether the load rate of each channel is lower than the threshold for the corresponding instruction type and whether the bandwidth meets the instruction transmission requirements. If the conditions are met, then the channel is selected as a candidate channel.

[0102] Step S335: If there are multiple candidate channels, calculate the average historical transmission delay of each channel, select the channel with the smallest delay to assign the instruction, and if there are no candidate channels, add the instruction to the waiting queue and retry periodically.

[0103] Historical average transmission delay refers to the average transmission delay of a channel over a period of time, reflecting the channel's transmission performance. A waiting queue is a queue used to store instructions for which a channel cannot be allocated temporarily.

[0104] If multiple candidate channels exist, calculate the average historical transmission delay of each channel and select the channel with the lowest delay to assign the command. If no candidate channel exists, add the command to the waiting queue and retry the channel assignment periodically, for example, once every 1 minute.

[0105] Step S340: Extract features from the state feedback data, decompose the system response delay data using wavelet transform algorithm to obtain frequency domain features, calculate the rate of change of task completion progress through statistical analysis, generate the distribution features of actual resource occupancy through histogram statistics, and fuse the frequency domain features, rate of change features and distribution features to obtain the state feedback features.

[0106] Feature extraction refers to extracting key features that reflect the system's operating state from state feedback data. Wavelet transform algorithms can decompose time-domain signals into frequency-domain signals, thereby obtaining the signal's frequency domain features. Statistical analysis refers to performing statistical analysis on data to obtain its statistical characteristics, such as mean, variance, and rate of change. Histogram statistics refer to grouping data to obtain its distribution characteristics.

[0107] When extracting features from state feedback data, the system response delay data is first decomposed using a wavelet transform algorithm to obtain frequency domain features. For example, the system response delay data can be treated as a time-domain signal, and the wavelet transform algorithm can be used to decompose it into components of different frequencies to extract frequency domain features. Next, the rate of change of task completion progress is calculated through statistical analysis, such as calculating the rate of change of task completion progress over a period of time. Then, the distribution characteristics of actual resource occupancy are generated through histogram statistics, for example, dividing the actual resource occupancy into different intervals and counting the amount of resource occupancy in each interval to obtain the distribution characteristics of actual resource occupancy. Finally, the frequency domain features, rate of change features, and distribution characteristics are fused to obtain the state feedback features.

[0108] As one implementation method, step S340 involves extracting features from the state feedback data and obtaining frequency domain features by decomposing the system response delay data using a wavelet transform algorithm. Specifically, this can be implemented as follows: steps S341~S345: Step S341: Collect system response delay data within a preset time period to obtain a one-dimensional time series. The length of the time series is the product of the sampling frequency and the preset time.

[0109] A preset time period refers to a pre-defined time range for collecting system response delay data, such as 1 minute. The sampling frequency refers to the frequency at which data is collected within the preset time period, such as once per second. A one-dimensional time series refers to a sequence of system response delay data arranged in chronological order.

[0110] Within a preset time period, system response delay data is collected according to the sampling frequency to obtain a one-dimensional time series. For example, if the preset time period is 1 minute and the sampling frequency is once per second, then 60 system response delay data points are collected to form a one-dimensional time series of length 60.

[0111] Step S342: Perform multi-level wavelet decomposition on the one-dimensional time series based on the preset wavelet basis function to obtain an approximate component and multiple detail components. The approximate component represents the trend characteristics of the system response delay, and the detail components represent the fluctuation characteristics of different frequency bands.

[0112] Wavelet basis functions are the basis functions used in wavelet transform algorithms, determining the properties and effects of the wavelet transform. Multi-level wavelet decomposition refers to performing multiple wavelet decompositions on a one-dimensional time series to obtain frequency domain features at different scales. Approximate components are the low-frequency components obtained after wavelet decomposition, characterizing the trend of the system response delay. Detail components are the high-frequency components obtained after wavelet decomposition, characterizing the fluctuation characteristics of the system response delay in different frequency bands.

[0113] By selecting a preset wavelet basis function, such as the Daubechies wavelet basis function, a multi-level wavelet decomposition is performed on the one-dimensional time series. For example, a three-level wavelet decomposition yields one approximate component and three detail components. The approximate component reflects the overall trend of the system response delay, such as whether it shows an upward or downward trend; the detail components reflect the fluctuation of the system response delay in different frequency bands, such as high-frequency fluctuations that may indicate that the system has been subjected to sudden interference.

[0114] Step S343: Perform a Fourier transform on each detail component to convert the time-domain signal into a frequency-domain signal, and calculate the power spectral density of each frequency-domain signal. The power spectral density is used to characterize the energy distribution of different frequency components.

[0115] The Fourier transform is a mathematical transformation method used to convert time-domain signals into frequency-domain signals. It decomposes a signal into sine and cosine components of different frequencies. Power spectral density refers to the power distribution of different frequency components in a frequency-domain signal; it reflects the energy distribution of the signal at different frequencies.

[0116] Perform a Fourier transform on each detail component to convert it from a time-domain signal to a frequency-domain signal. Then, calculate the power spectral density of each frequency-domain signal. For example, use the Fast Fourier Transform (FFT) algorithm to perform a Fourier transform on the detail components to obtain the frequency-domain signal. Then, obtain the power spectral density by calculating the square of the magnitude of the frequency-domain signal.

[0117] Step S344: Extract the peak frequency, center frequency, and bandwidth parameters from the power spectral density, and combine the peak frequency, center frequency, and bandwidth parameters to obtain the frequency domain features. The dimension of the frequency domain features is consistent with the number of decomposition layers.

[0118] Peak frequency refers to the frequency with the highest power in the power spectral density, reflecting the main frequency components of the signal. Center frequency refers to the frequency at the center of the power spectral density, reflecting the average frequency of the signal. Bandwidth parameter refers to the width of the power distribution in the power spectral density, reflecting the frequency range of the signal.

[0119] Peak frequency, center frequency, and bandwidth parameters are extracted from the power spectral density. For example, the peak frequency is obtained by finding the maximum value in the power spectral density; the center frequency is obtained by calculating the weighted average of the power spectral density; and the bandwidth parameter is obtained by calculating the frequency range in the power spectral density where the power exceeds a certain threshold. These parameters are combined to obtain the frequency domain features, and the dimension of the frequency domain features is consistent with the number of decomposition levels. For example, if a 3-level wavelet decomposition is performed, the dimension of the frequency domain features will be 3.

[0120] Step S345: Normalize the frequency domain features by scaling the values ​​of each parameter to a preset range. The normalization formula uses the maximum and minimum value standardization method.

[0121] Normalization refers to processing the parameter values ​​in the frequency domain characteristics to give them the same dimensions and range, thereby enabling effective comparison and analysis. The maximum-minimum standardization method is a commonly used normalization method. It scales the data values ​​to a preset range by subtracting the minimum value from the data value and dividing by the difference between the maximum and minimum values.

[0122] The frequency domain features are normalized using a maximum-minimum standardization method. For example, with a preset numerical range of [0,1], for each parameter in the frequency domain features, its maximum and minimum values ​​are calculated. Then, the minimum value is subtracted from each parameter value, and the result is divided by the difference between the maximum and minimum values ​​to obtain the normalized parameter value.

[0123] Step S400: Perform dynamic deviation analysis on the state feedback characteristics to generate adjustment parameters for the collaborative control strategy.

[0124] Dynamic deviation analysis refers to comparing and analyzing state feedback characteristics with historical baseline characteristics to identify deviations and trends. Cooperative control strategy adjustment parameters are parameters used to adjust the cooperative control strategy based on the results of dynamic deviation analysis, such as resource allocation ratios and task execution priorities.

[0125] The state feedback characteristics are compared and analyzed with the baseline characteristic set from historical collaborative control processes, and the deviation between the two is calculated. By analyzing the deviation, the trend of deviation change is predicted, and based on a preset deviation threshold, the type and magnitude of collaborative control parameters that need to be adjusted are determined.

[0126] As one implementation method, step S400 involves performing dynamic deviation analysis on the state feedback characteristics to generate adjustment parameters for the collaborative control strategy. Specifically, this can be implemented as follows: steps S410~S450: Step S410: Time-align the state feedback features with the baseline feature set in the historical collaborative control process. The baseline feature set is the average value of the state feedback features collected in historical successful cases.

[0127] Timing alignment aligns state feedback features with a baseline feature set in time to ensure temporal comparability. Historical success cases refer to past successful examples in cross-system collaborative control, and the baseline feature set is the average of the state feedback features collected in these cases.

[0128] To align the state feedback features with the baseline feature set in time, the baseline feature set needs to be processed first, including steps such as retrieving historical cases, outlier detection, statistical calculation, and length adjustment. Then, a dynamic time warping algorithm is used to align the adjusted state feedback features with the baseline feature set.

[0129] As one implementation method, step S410, which involves aligning the state feedback features with the baseline feature set from the historical collaborative control process in terms of time sequence, can be specifically implemented as follows: steps S411~S415: Step S411: Retrieve historical cases of the same type as the current collaborative control task from the historical database, extract the state feedback features of each historical case, and obtain the original baseline feature set.

[0130] The historical database stores past cross-system collaborative control cases, containing detailed information for each case, including state feedback characteristics. The original baseline feature set refers to the set of state feedback characteristics from historical cases retrieved from the historical database that are of the same type as the current collaborative control task.

[0131] Based on the type of the current collaborative control task, relevant historical cases are retrieved from the historical database. For example, if the current collaborative control task is production task scheduling, then all historical cases of production task scheduling are retrieved from the historical database. The state feedback features of these cases are extracted to form the original baseline feature set.

[0132] Step S412: Perform outlier detection on the original baseline feature set, and use statistical methods to identify and remove feature samples that deviate from the normal range to obtain the purified baseline feature set.

[0133] Outlier detection refers to identifying data points in a dataset that deviate from the normal range. These data points may be caused by measurement errors, system malfunctions, or other reasons. Statistical methods refer to statistical analysis methods used to detect outliers, such as methods based on standard deviation or quartiles. The cleaned-up baseline feature set refers to the set of baseline features obtained after outlier detection and removal.

[0134] Statistical methods are used to detect outliers in the original baseline feature set. For example, based on standard deviation, the mean and standard deviation of each feature dimension are calculated, and feature samples that deviate from the mean by more than three times the standard deviation are identified as outliers and removed. After outlier detection and removal, a cleaned-up baseline feature set is obtained.

[0135] Step S413: Calculate the mean and standard deviation of each time step in the purified baseline feature set, and generate a statistical baseline feature containing the mean sequence and standard deviation sequence.

[0136] The mean is the sum of a set of data divided by the number of data points; it reflects the central tendency of the data. The standard deviation is a measure of how much a set of data deviates from its mean; it reflects the dispersion of the data. The mean sequence is a sequence composed of the average values ​​at each time step in the purified baseline feature set, and the standard deviation sequence is a sequence composed of the standard deviations at each time step in the purified baseline feature set. Statistical baseline features refer to a feature set that includes both the mean and standard deviation sequences; it can more comprehensively reflect the state feedback characteristics of historical collaborative control processes.

[0137] Calculate the mean and standard deviation of each time step in the purified baseline feature set, generating a mean sequence and a standard deviation sequence. For example, for each time step in the purified baseline feature set, calculate the mean and standard deviation of all feature samples at that time step. Combine the mean sequences and standard deviation sequences to obtain the statistical baseline features.

[0138] Step S414: Adjust the time series length of the current state feedback feature to be consistent with the statistical baseline feature. If the current sequence length is short, perform interpolation to complete it. If the current sequence length is long, perform downsampling.

[0139] Time series length adjustment refers to adjusting the time series length of the current state feedback feature to match the statistical baseline feature, in order to facilitate subsequent time series alignment and comparative analysis. Interpolation completion refers to inserting new data points into the time series to fill in missing data. Downsampling refers to selecting a subset of data points from the time series to reduce the amount of data.

[0140] Compare the time series length of the current state feedback feature with the time series length of the statistical baseline feature. If the current sequence length is short, use interpolation methods to complete it, such as linear interpolation. If the current sequence length is long, use downsampling methods, such as average downsampling.

[0141] Step S415: Align the mean sequence of the adjusted state feedback features with the statistical benchmark features using a dynamic time warping algorithm, minimize the cumulative distance between the two sequences, and complete the nonlinear time alignment.

[0142] Dynamic time warping (VTW) algorithms minimize the cumulative distance between two sequences by finding the optimal matching path. The cumulative distance is the sum of the distances between points on the matching path between the two sequences in VTW. Non-linear time alignment, on the other hand, aligns two sequences in time, allowing for different time scales between them.

[0143] The mean sequences of the adjusted state feedback features and statistical baseline features are input into the dynamic time warping algorithm. By finding the optimal matching path, the cumulative distance between the two sequences is minimized. For example, the dynamic time warping algorithm can be implemented using dynamic programming, calculating the cumulative distance matrix between the two sequences, and then finding the path with the minimum cumulative distance from the matrix to complete the nonlinear time alignment.

[0144] Step S420: Calculate the dimension-wise deviation value between the aligned state feedback features and the baseline feature set to obtain the deviation feature sequence. Each element in the deviation feature sequence represents the degree of feature deviation at the corresponding time step.

[0145] The dimensional deviation value refers to the difference between the aligned state feedback feature and the baseline feature set in each feature dimension. The deviation feature sequence is a sequence composed of dimensional deviation values, which reflects the degree of feature deviation at each time step.

[0146] For the aligned set of state feedback features and the baseline feature set, the difference between them is calculated at each feature dimension to obtain the dimension-wise deviation value. The dimension-wise deviation values ​​at each time step are combined to obtain the deviation feature sequence. For example, for a set of state feedback features and the baseline feature set containing 3 feature dimensions, the deviation value of each feature dimension is calculated at each time step to obtain a deviation feature sequence of length equal to the number of time steps, where each element contains 3 deviation values.

[0147] Step S430: Predict the trend of the deviation feature sequence through a long short-term memory network. The long short-term memory network includes an input layer, a hidden layer and an output layer. The input layer receives the deviation feature sequence, the hidden layer memorizes long-term dependencies through a gating mechanism, and the output layer predicts the deviation trend at a preset time step in the future.

[0148] Long Short-Term Memory (LSTM) networks can handle long-term dependencies in sequential data. The input layer, the first layer of the LSTM network, receives the input data, specifically the bias feature sequence. The hidden layers, the core layers of the LSTM network, memorize long-term dependencies through a gating mechanism. The output layer, the last layer of the LSTM network, predicts the bias trend at a predetermined time step based on the state of the hidden layers.

[0149] The bias feature sequence is input into the input layer of a Long Short-Term Memory (LSTM) network, which then passes the data to the hidden layers. The hidden layers use gating mechanisms, including input gates, forget gates, and output gates, to control the inflow, outflow, and retention of information, thereby memorizing long-term dependencies. The output layer, based on the state of the hidden layers, predicts the bias trend at a predetermined time step in the future.

[0150] As one implementation method, in step S430, the hidden layer memorizes long-term dependencies through a gating mechanism, which can be specifically implemented as follows: steps S431~S435: Step S431: Adjust the input ratio of feature information at the current moment through the input control mechanism, calculate the input control weight based on the deviation features at the current moment and the hidden state at the previous moment, and control the degree of inclusion of new information.

[0151] The input control mechanism refers to the mechanism in the hidden layer used to control the proportion of feature information input at the current time step. The input control weight refers to the weight used to control the degree of new information inclusion, which is calculated based on the deviation features at the current time step and the hidden state at the previous time step.

[0152] Based on the current time step's bias features and the previous time step's hidden state, input control weights are calculated through a fully connected layer. The input control weights are vectors ranging from [0,1], where each element represents the proportion of new information incorporated into the corresponding feature dimension. For example, an input control weight vector of [0.2, 0.5, 0.8] indicates that the proportion of new information incorporated into the first feature dimension is 20%, the proportion into the second feature dimension is 50%, and the proportion into the third feature dimension is 80%.

[0153] Step S432: Adjust the retention ratio of historical memory information through the forgetting control mechanism. Calculate the forgetting control weight based on the deviation characteristics at the current moment and the hidden state at the previous moment to control the degree of retention of historical memory.

[0154] The forgetting control mechanism refers to the mechanism in the hidden layer used to control the proportion of historical memory information retained. The forgetting control weight refers to the weight used to control the degree of historical memory retention, which is calculated based on the bias characteristics at the current moment and the hidden state at the previous moment.

[0155] Based on the current time step's bias features and the previous time step's hidden state, forgetting control weights are calculated through a fully connected layer. The forgetting control weights are vectors ranging from [0,1], where each element represents the percentage of historical memory retention for the corresponding feature dimension. For example, a forgetting control weight vector of [0.9, 0.7, 0.3] indicates that the historical memory retention percentage for the first feature dimension is 90%, for the second feature dimension it is 70%, and for the third feature dimension it is 30%.

[0156] Step S433: Based on the input control weights and forgetting control weights, update the cell state of the hidden layer, and fuse the new information after input control with the historical memory information after forgetting control to obtain the cell state at the current moment.

[0157] Cellular state is an important state variable in the hidden layers of a Long Short-Term Memory (LSTM) network, used to store historical memory information. New information after input control refers to the deviation characteristics of the current moment after processing by the input control mechanism. Historical memory information after forgetting control refers to the cell state information of the previous moment after processing by the forgetting control mechanism.

[0158] Based on the input control weights, the deviation characteristics at the current moment are weighted to obtain the new information after input control. Based on the forgetting control weights, the cell state at the previous moment is weighted to obtain the historical memory information after forgetting control. The new information after input control and the historical memory information after forgetting control are added together to obtain the cell state at the current moment.

[0159] Step S434: Adjust the output ratio of the current memory information through the output control mechanism, calculate the output control weight based on the cell state at the current moment and the hidden state at the previous moment, and control the output degree of the current cell state.

[0160] The output control mechanism refers to the mechanism in the hidden layer used to control the proportion of currently stored information output. The output control weight refers to the weight used to control the degree of output of the current cell state, which is calculated based on the cell state at the current moment and the hidden state at the previous moment.

[0161] Based on the current cell state and the hidden state from the previous time step, output control weights are calculated through a fully connected layer. The output control weights are vectors ranging from [0,1], where each element represents the proportion of the current cell state output for the corresponding feature dimension. For example, an output control weight vector of [0.6, 0.4, 0.8] indicates that the proportion of the current cell state output for the first feature dimension is 60%, for the second feature dimension it is 40%, and for the third feature dimension it is 80%.

[0162] Step S435: Convert the cell state after output control into the hidden state at the current moment, which is used to remember the long-term dependencies in the bias feature sequence.

[0163] The cell state after output control refers to the cell state at the current moment after processing by the output control mechanism. The hidden state at the current moment refers to the hidden state of the Long Short-Term Memory network at the current moment, used to remember long-term dependencies in the biased feature sequence.

[0164] The cell state after output control is transformed through an activation function (such as the tanh function) to obtain the hidden state at the current time step. This hidden state serves as the input for the next time step calculation, participating in subsequent computations and helping the Long Short-Term Memory (LSTM) network remember long-term dependencies in the biased feature sequence. Through this gating mechanism, the LSM network can effectively handle the long-term dependency problem in sequence data, making the prediction of bias trends more accurate and reliable.

[0165] Step S440: Based on the deviation trend and the preset deviation threshold, determine the types of collaborative control parameters that need to be adjusted. The parameter types include resource allocation ratio, task execution priority, and data transmission bandwidth.

[0166] The deviation trend is the change in deviation at a predetermined future time step predicted by a long short-term memory network. It reflects the direction and extent of the deviation between the state feedback characteristics and the baseline characteristics. The preset deviation threshold is a pre-defined boundary value used to determine whether the deviation needs adjustment; different deviation thresholds correspond to different collaborative control parameter adjustment strategies. Collaborative control parameter types refer to the types of parameters that can be adjusted during cross-system collaborative control. For example, resource allocation ratio determines the amount of resources each business system can acquire, task execution priority affects the order of task execution, and data transmission bandwidth relates to the rate and efficiency of data transmission between systems.

[0167] Determining the type of collaborative control parameter that needs adjustment based on the deviation trend and the preset deviation threshold requires comprehensive consideration of multiple factors and a series of analytical steps.

[0168] As one implementation method, step S440, based on the deviation trend and the preset deviation threshold, determines the type of collaborative control parameter that needs to be adjusted, which can be specifically implemented as follows: steps S441~S445: Step S441: Preset multiple deviation threshold ranges, each threshold range corresponding to a different parameter adjustment level.

[0169] Setting multiple deviation threshold ranges allows for more precise determination of different adjustment strategies based on the magnitude of the deviation. Different deviation threshold ranges represent different degrees of deviation severity, and the parameter adjustment level corresponding to each threshold range reflects the adjustment intensity required for that degree of deviation severity. For example, three deviation threshold ranges can be preset: the range below the first threshold is the mild deviation range, corresponding to a lower parameter adjustment level, such as requiring only slight parameter fine-tuning; the range between the first and second thresholds is the moderate deviation range, corresponding to a moderate parameter adjustment level, which may require some parameter adjustment; and the range above the second threshold is the severe deviation range, corresponding to a higher parameter adjustment level, requiring a larger parameter adjustment.

[0170] Step S442: Match each predicted deviation value in the deviation trend with the preset deviation threshold range to determine the parameter adjustment level for each time step.

[0171] For each predicted deviation value in the bias trend obtained from the Long Short-Term Memory (LSTM) network prediction, it is compared with a preset deviation threshold range to determine which threshold range the deviation value belongs to, thereby determining the corresponding parameter adjustment level. For example, if a predicted deviation value is less than the first threshold, the parameter adjustment level for that time step is a mild adjustment level; if the predicted deviation value is between the first and second thresholds, the corresponding parameter adjustment level is a moderate adjustment level; if it is greater than the second threshold, the corresponding parameter adjustment level is a severe adjustment level. In this way, a corresponding parameter adjustment level is determined for each time step for further analysis.

[0172] Step S443: Count the frequency of adjustment levels of each parameter type within the preset time window, and determine the parameter type with the highest frequency as the main adjustment target.

[0173] A preset time window refers to a fixed time range used to statistically analyze the frequency of adjustment levels for each parameter type. The frequency of different adjustment levels for each parameter type within this time window is counted. For example, for the parameter type "resource allocation ratio," the frequency of mild, moderate, and severe adjustment levels is counted within the preset time window, and these frequencies are summed to obtain the total frequency for that parameter type. This statistical analysis is performed on all parameter types (resource allocation ratio, task execution priority, and data transmission bandwidth). Then, the total frequency of each parameter type is compared, and the parameter type with the highest frequency is identified as the primary adjustment target. This is because this parameter type requires adjustment most frequently within the preset time window and is likely the main factor causing deviations; therefore, it is adjusted first.

[0174] Step S444: Analyze the historical adjustment effect data of the main adjustment object. If the deviation reduction rate after parameter adjustment in the historical adjustment effect data is greater than the preset threshold, then retain the parameter type; otherwise, change it to the second most frequent parameter type.

[0175] Historical adjustment effect data refers to the changes in the deviation between the state feedback characteristics and the baseline characteristics after parameter adjustments were made to the main adjustment object in the past. The deviation reduction rate refers to the percentage decrease in deviation after parameter adjustment. For example, if the deviation before adjustment was 10 and the deviation after adjustment is 5, then the deviation reduction rate is (10-5) / 10 = 50%. The preset threshold is a pre-set boundary value used to judge whether the historical adjustment effect is effective. If the historical adjustment effect data of the main adjustment object shows that the deviation reduction rate after parameter adjustment is greater than the preset threshold, it indicates that adjusting this parameter type is effective and can reduce deviation; therefore, this parameter type is retained as the main adjustment object. Conversely, if the deviation reduction rate is less than or equal to the preset threshold, it indicates that the adjustment effect of this parameter type is poor, and it is necessary to change to the next most frequent parameter type as the main adjustment object and try adjusting again.

[0176] Step S445: Perform correlation analysis on the determined parameter types to determine whether there are any related parameters that need to be adjusted in a coordinated manner. If so, add the related parameters to the adjustment list to obtain the final set of coordinated control parameter types.

[0177] Correlation analysis refers to analyzing whether there are interrelationships and influences between identified parameter types and other parameter types. In cross-system collaborative control, some parameter types may be closely related; for example, adjusting resource allocation ratios may affect task execution priority and data transmission bandwidth. By performing correlation analysis on the identified parameter types, it is determined whether there are other related parameter types that also need to be adjusted simultaneously. If such related parameters exist, they are added to the adjustment list. The resulting adjustment list constitutes the final set of collaborative control parameter types, containing all collaborative control parameter types that need adjustment, so that these parameters can be adjusted accordingly in subsequent steps.

[0178] Step S450: For each parameter type that needs to be adjusted, calculate the adjustment magnitude value. The magnitude of the adjustment magnitude value is positively correlated with the slope of the deviation trend. The adjustment direction is determined according to the sign of the deviation value. Combine the parameter type and the corresponding adjustment magnitude value into the collaborative control strategy adjustment parameters.

[0179] The adjustment magnitude refers to the degree of adjustment determined for each type of coordinated control parameter that needs adjustment. The slope of the deviation trend reflects the rate at which the deviation changes over time. The steeper the slope, the more drastic the deviation change, and correspondingly, a larger adjustment magnitude is needed to correct the deviation. Therefore, the magnitude of the adjustment magnitude is positively correlated with the slope of the deviation trend. The adjustment direction is determined by the sign of the deviation value. If the deviation value is positive, it means that the value of the state feedback characteristic is greater than the value of the baseline characteristic, and the adjustment direction should be to decrease the parameter value. If the deviation value is negative, it means that the value of the state feedback characteristic is less than the value of the baseline characteristic, and the adjustment direction should be to increase the parameter value.

[0180] For each parameter type requiring adjustment (such as resource allocation ratio, task execution priority, and data transmission bandwidth), the adjustment magnitude is first calculated based on the slope of the deviation trend. For example, by establishing a linear function relationship, the adjustment magnitude = slope × proportional coefficient (the proportional coefficient is a pre-set constant used to control the range of adjustment magnitude). Then, the adjustment direction is determined based on the sign of the deviation value. Finally, the parameter type and its corresponding adjustment magnitude value are combined to form the collaborative control strategy adjustment parameters. For example, for the parameter type of resource allocation ratio, if the calculated adjustment magnitude is 10%, and the adjustment direction is determined to be an increase based on the deviation value, then the part of the collaborative control strategy adjustment parameters related to resource allocation ratio can be expressed as "Resource allocation ratio: Increase by 10%". By combining all the parameter types requiring adjustment and their adjustment magnitude values ​​in this way, the complete collaborative control strategy adjustment parameters are obtained.

[0181] Step S500: Synchronize the collaborative control strategy adjustment parameters to each business system through the cross-system interaction architecture.

[0182] Cross-system interaction architecture serves as the infrastructure for data transmission and communication between multiple business systems, providing channels and mechanisms for synchronizing collaborative control strategy adjustment parameters. Synchronizing these parameters to each business system ensures that each system adjusts its operational status and behavior promptly based on the new parameters, thereby achieving cross-system collaborative control optimization. This process requires a series of operations, including encoding, grouping, transmitting, and synchronizing the collaborative control strategy adjustment parameters.

[0183] As one implementation method, step S500 synchronizes the collaborative control strategy adjustment parameters to each business system through a cross-system interaction architecture, which can be implemented as follows: steps S510~S540: Step S510: Encode the adjustment parameters of the collaborative control strategy, and use variable length encoding to convert the parameter type and adjustment magnitude value into a binary data stream.

[0184] Variable-length encoding is a coding method that allocates different code lengths based on the characteristics and frequency of the data. It flexibly allocates code length according to the specific parameters and adjustment values, effectively reducing the storage space and transmission bandwidth requirements of the encoded data. The parameter types (such as resource allocation ratios, task execution priorities, and data transmission bandwidth) and adjustment values ​​(such as specific adjustment percentages or values) in the collaborative control strategy adjustment parameters are converted using variable-length encoding to obtain a binary data stream. For example, for the parameter type "resource allocation ratio" and its adjustment value of 10%, it is converted into the corresponding binary code combination according to the variable-length encoding rules. After all the parameter types and their adjustment values ​​that need to be adjusted undergo such encoding conversion, a complete binary data stream is formed.

[0185] Step S520: Through the synchronization control module of the cross-system interaction architecture, the encoded binary data stream is grouped according to the business system. Each group of data contains all the parameter information that the business system needs to adjust.

[0186] The synchronization control module is a dedicated module in the cross-system interaction architecture for implementing data synchronization. It is responsible for processing and grouping the encoded binary data stream. Based on the needs of each business system, the encoded binary data stream is grouped according to the business system. Each business system corresponds to a set of data, which contains information about all the types of collaborative control parameters that the business system needs to adjust, along with their adjustment values. For example, for business system A, which needs to adjust resource allocation ratios and task execution priorities, the encoded data related to resource allocation ratios and task execution priorities for business system A are combined into a single set and sent specifically to business system A. This grouping ensures that each business system receives only the adjustment information relevant to itself, improving the efficiency and specificity of data transmission.

[0187] Step S530: The packet data is sent to the corresponding business system in asynchronous transmission mode. If the receiver does not return an acknowledgment message within a preset time, the data is resent. After receiving the parameter adjustment data, each business system performs decoding processing and updates the local collaborative control policy. During the update process, atomic operations are used to ensure the consistency of policy replacement and avoid intermediate states.

[0188] Asynchronous transmission mode is a data transmission method in which the sender can continue other operations after sending data without waiting for a response from the receiver, thus improving data transmission efficiency. Grouped data is sent to the corresponding business system through a cross-system interaction architecture. Simultaneously, a timer is started with a preset time. If the receiver returns an acknowledgment message containing the correct identifier within the preset time, the data transmission is considered successful. If no acknowledgment message is received within the preset time, it may indicate that data loss or error occurred during transmission. In this case, the data is retransmitted until an acknowledgment message is received.

[0189] After receiving parameter adjustment data, each business system first decodes the binary data stream, converting it into readable information about the collaborative control parameter type and adjustment magnitude. Then, it updates its local collaborative control strategy based on this information. Atomic operations are used during the update process. An atomic operation is an indivisible operation; it either completes entirely or not at all. This ensures that no partial updates occur during the local collaborative control strategy update, guaranteeing the consistency and integrity of the strategy replacement. For example, when updating resource allocation ratios and task execution priorities, either both parameters are successfully updated, or neither is updated, avoiding inconsistencies where one parameter is updated while the other is not.

[0190] As one implementation method, in step S530, the packet data is sent to the corresponding service system using an asynchronous transmission mode, which can be implemented as follows: steps S531~S535: Step S531: Assign a transmission sequence number to each data packet. The sequence number increments in the order of transmission and is used to identify the order in which data is transmitted.

[0191] A transmission sequence number is a unique identifier used to distinguish different data packets and identify their transmission order. Each data packet is assigned an incrementing sequence number according to its transmission order; for example, the first data packet has a sequence number of 1, the second data packet has a sequence number of 2, and so on. This sequence number plays a crucial role in data transmission. The receiver can use the sequence number to determine the integrity and order of the data, ensuring that the received data is in the correct sequence. Furthermore, in the event of data retransmission, the sequence number helps distinguish between newly transmitted data and retransmitted data.

[0192] Step S532: Combine the packet data, transmission sequence number and checksum into a transmission frame and send it to the target business system through the cross-system interaction architecture.

[0193] A checksum is a numerical value used to verify data integrity; it is obtained through specific calculations on packet data. Packet data, transmission sequence numbers, and checksums are combined to form a transmission frame. The transmission frame is the basic unit of data transmission; it contains the data itself as well as relevant information to ensure the accuracy and integrity of data transmission. Transmission frames are sent to the target business system through a cross-system interaction architecture. During transmission, the cross-system interaction architecture selects appropriate transmission channels and protocols based on network status and configuration to ensure that the transmission frame arrives at the target business system efficiently and reliably.

[0194] Step S533: Start the timer. If an acknowledgment frame containing the same transmission sequence number is received from the target service system within the preset timeout period, the data transmission is considered successful. Otherwise, the data packet is marked as pending retransmission.

[0195] The preset timeout period is a pre-defined time limit used to determine if data transmission has timed out. A timer is started to count down, waiting for an acknowledgment frame from the target service system. The acknowledgment frame is a response sent back by the target service system after successfully receiving the transmission frame; it contains the same transmission sequence number as the sent transmission frame. If an acknowledgment frame with the same transmission sequence number is received within the preset timeout period, it indicates that the target service system has successfully received the data packet, and the data transmission is successful. If no such acknowledgment frame is received within the preset timeout period, the data packet is marked for retransmission in order to facilitate subsequent retransmission operations.

[0196] Step S534: For data packets marked as to be retransmitted, the backoff algorithm is used to calculate the retransmission delay time. If the retransmission attempt fails after reaching the preset limit, a link failure alarm is triggered and the system switches to the backup transmission channel.

[0197] A backoff algorithm is an algorithm used to avoid network congestion caused by simultaneous retransmission of multiple data packets. It calculates the retransmission delay time according to certain rules. For data packets marked as awaiting retransmission, the backoff algorithm calculates the delay time for the next retransmission. For example, common backoff algorithms increase the delay time as the number of retransmissions increases, preventing the simultaneous retransmission of a large amount of data and the resulting increased network congestion when network problems occur.

[0198] The preset limit is the maximum number of retransmissions set in advance. If the target service system still does not receive an acknowledgment frame after the preset limit is reached, it indicates that the current transmission channel may be faulty. At this point, a link failure alarm is triggered, notifying the system administrator or relevant monitoring system. Simultaneously, the system switches to a backup transmission channel and re-attempts to send data packets to ensure continued data transmission.

[0199] Step S535: After all packet data transmissions are completed, a synchronization completion instruction is sent to each business system, instructing each system to simultaneously implement the new collaborative control strategy.

[0200] Once all grouped data has been successfully transmitted and confirmed by the target business system, a synchronization completion command is sent to each business system. This command notifies each business system that all parameter adjustment data has been transmitted and the new collaborative control strategy can take effect simultaneously. In this way, it ensures that all business systems begin operating according to the new collaborative control strategy at the same time, achieving cross-system collaborative control synchronization and improving the overall system performance and stability.

[0201] Step S540: Collect system operating status data after parameter adjustment, calculate the collaborative control efficiency index before and after strategy optimization. If the efficiency index improvement is greater than the preset threshold, the dynamic optimization is completed; otherwise, return to the parameter generation step to recalculate the adjustment value.

[0202] System operational status data refers to the actual operational data of each business system after parameter adjustments, such as task completion time, resource utilization, and system response time. Collaborative control efficiency indicators are quantitative metrics used to measure the effectiveness of cross-system collaborative control. They comprehensively consider multiple factors, such as the effective utilization of resources and task execution efficiency. By collecting system operational status data after parameter adjustments, the collaborative control efficiency indicators after strategy optimization are calculated and compared with the indicators before strategy optimization.

[0203] The preset threshold is a pre-defined boundary value used to determine whether the optimization of the collaborative control strategy is effective. If the calculated improvement in the collaborative control efficiency index is greater than the preset threshold, it indicates that the collaborative control effect of the system has been significantly improved by adjusting the collaborative control parameters, and the dynamic optimization process has been completed. If the improvement is less than or equal to the preset threshold, it indicates that the current adjustment of the collaborative control strategy has not achieved the expected effect, and it is necessary to return to the parameter generation step (step S400), re-perform dynamic deviation analysis based on the new system operating status data, calculate the adjustment magnitude value, and try to adjust the collaborative control strategy again until the improvement in the collaborative control efficiency index is greater than the preset threshold. This continuous dynamic adjustment and optimization ensures that cross-system collaborative control always maintains a high level of efficiency.

[0204] It is understood that the various algorithms involved in the above descriptions of the embodiments of the present invention, such as Euclidean distance algorithm, cosine distance algorithm, interpolation algorithm, backoff algorithm, etc., can all be obtained from relevant content in the prior art. To save space, they will not be elaborated on in the embodiments of the present invention. In addition, those skilled in the art can supplement the details based on common knowledge in the art when implementing the solutions of the present invention. For example, they can use normalization to eliminate dimensional conflicts before feature fusion, use interpolation to eliminate dimensional differences, reasonably set thresholds based on historical data, experience or business scenario requirements, train the model based on a general model training method, set the number of layers in the model structure based on actual needs, select activation functions, etc. The present invention will not provide redundant descriptions of overly detailed implementation processes here.

[0205] Please see Figure 2 , Figure 2 This is a schematic diagram of a computer system provided in an embodiment of the present invention. The computer system includes at least a processor 101, a communication interface 102, and a memory 103. The processor 101, communication interface 102, and memory 103 can be connected via a bus or other means. The processor 101 (or Central Processing Unit, CPU) is the computing and control core of the computer system, capable of parsing various instructions and processing various data within the computer system. The communication interface 102 may optionally include a standard wired interface or a wireless interface (such as Wi-Fi, mobile communication interface, etc.), and can be used to send and receive data under the control of the processor 101; the communication interface 102 can also be used for data transmission and interaction within the computer system. The memory 103 is a storage device in the computer system used to store programs and data. It is understood that the memory 103 here can include the computer system's built-in memory, or it can include extended memory supported by the computer system. The memory 103 provides storage space, which stores the computer system's operating system; this invention does not limit this storage space.

[0206] In one embodiment, the processor 101 executes the cross-system cooperative control method based on digital threads provided in the above embodiments of the present invention by running a computer program in the memory 103.

Claims

1. A method for cross-system collaborative control based on digital thread, characterized in that, include: Construct a cross-system interaction architecture to establish real-time data transmission and command interaction between multiple business systems through a distributed node communication protocol; Based on the cross-system interaction architecture, dynamic operational characteristic data of each business system are collected to generate cross-system collaborative control logic. The cross-system collaborative control logic is converted into standardized control commands, distributed to the corresponding business systems and executed in real-time collaborative control, and the status feedback characteristics during the control process are collected. Dynamic deviation analysis is performed on the state feedback characteristics to generate adjustment parameters for the collaborative control strategy; The adjustment parameters of the collaborative control strategy are synchronized to each business system through the cross-system interaction architecture; The step of collecting dynamic operational characteristic data of each business system based on the cross-system interaction architecture and generating cross-system collaborative control logic includes: The distributed acquisition nodes in the cross-system interaction architecture collect dynamic operational characteristic data of each business system at preset time intervals. The dynamic operational feature data is spatiotemporally aligned to map the feature data of different business systems to a unified time coordinate system, generating a spatiotemporally aligned feature set. Based on the spatiotemporal alignment feature set, the correlation and dependency features between various business systems are extracted through an attention mechanism. These correlation and dependency features characterize the resource competition relationship and task coupling degree of different business systems during task execution. The associated dependency features are input into a pre-trained collaborative logic generation network, which includes a feature encoding layer, a dependency inference layer, and a logic output layer. The feature encoding layer converts the associated dependency features into high-dimensional semantic vectors, the dependency inference layer performs node relationship inference on the high-dimensional semantic vectors through a graph neural network, and the logic output layer generates an initial framework for cross-system collaborative control logic based on the inference results. The initial framework is subjected to conflict detection to identify resource allocation conflicts and task execution timing conflicts in the collaborative control logic. Based on the conflict detection results, the task priorities and resource allocation ratios in the initial framework are adjusted to obtain the final cross-system collaborative control logic.

2. The method of claim 1, wherein, The process of performing spatiotemporal alignment processing on the dynamic operational feature data maps the feature data of different business systems to a unified time coordinate system, generating a spatiotemporally aligned feature set, including: Extract the timestamp information from the dynamic operation feature data to determine the collection time and data sampling period of the feature data of each business system; Based on the acquisition time and data sampling period, a time reference axis is constructed, and the smallest time unit of the time reference axis is the common divisor of the sampling periods of each business system. Interpolation processing is performed on the dynamic operation feature data of each business system to supplement the feature data of missing moments on the time reference axis and generate a feature sequence with equal time intervals. The time interval feature sequences of each business system are concatenated according to the system identifier to obtain a three-dimensional feature tensor containing system dimension, time dimension and feature dimension. The three-dimensional feature tensor is used as the spatiotemporal alignment feature set. The extraction of inter-system dependency features based on the spatiotemporal alignment feature set through an attention mechanism includes: The spatiotemporal alignment feature set is divided into multiple time window feature subsets according to the time dimension, and each time window feature subset contains feature data of a preset number of consecutive time steps. For each feature subset of a time window, a similarity weight between feature data from different business systems is calculated using a self-attention mechanism. The similarity weight represents the degree of correlation between feature data from two business systems within the same time window. Based on the similarity weight, the feature data of each business system are weighted and aggregated to generate an interaction association vector between the business systems. The dimension of the interaction association vector is consistent with the number of business systems. The interaction association vectors are concatenated in time window order to obtain the association dependency feature sequence. Each element in the association dependency feature sequence corresponds to the business system association relationship within a time window. The correlation dependency feature sequence is filtered to eliminate the interference of short-term fluctuations on the extraction of correlation relationships.

3. The method of claim 1, wherein, The process of converting the cross-system collaborative control logic into standardized control commands, distributing them to the corresponding business systems and executing real-time collaborative control, and collecting state feedback characteristics during the control process includes: Analyze the task execution process and resource allocation rules in the cross-system collaborative control logic to determine the business system identifier and operation type corresponding to each control node; Based on a preset instruction conversion template, the task execution process is converted into structured control instructions. The structured control instructions include an instruction header, an instruction body, and a checksum. The instruction header stores the business system identifier and instruction priority, the instruction body contains specific operation parameters and execution time range, and the checksum is used to ensure the integrity of instruction transmission. Through the instruction distribution module of the cross-system interaction architecture, the structured control instructions are routed to the corresponding nodes according to the business system identifier. During the routing process, a load balancing algorithm is used to allocate instruction transmission channels. After receiving the structured control instructions, each business system executes the corresponding operation and collects status feedback data in real time during the control process. The status feedback data includes task completion progress, actual resource usage, and system response latency. Feature extraction is performed on the state feedback data. Frequency domain features are obtained by decomposing the system response delay data using a wavelet transform algorithm. The rate of change of task completion progress is calculated by statistical analysis. The distribution features of actual resource occupancy are generated by histogram statistics. The frequency domain features, the rate of change features, and the distribution features are fused to obtain the state feedback features.

4. The method of claim 3, wherein, The method based on a preset instruction conversion template converts the task execution flow into structured control instructions, including: The system calls a preset instruction template library and matches the corresponding instruction template according to the operation type in the task execution process. The instruction template library contains three basic templates: data read / write, computation scheduling, and resource allocation. The parameter constraints in the task execution process are parsed and mapped to variable fields of the instruction template. The parameter constraints include resource usage limits, task execution deadlines, and data transmission bandwidth limits. Syntax validation is performed on the mapped instruction template to check whether the parameter format conforms to the business system interface specification. If there is a format error, an error message is returned and the conversion process is terminated. If the format is correct, a unique instruction identifier is generated. The unique instruction identifier is generated based on the business system identifier, the instruction generation timestamp, and a random number combination. The instruction unique identifier, the business system identifier, the operation parameters, and the check code are assembled into a structured control instruction in a set order.

5. The method as described in claim 3, characterized in that, The step of extracting features from the state feedback data and obtaining frequency domain features by decomposing the system response delay data using a wavelet transform algorithm includes: Collect system response delay data within a preset time period to obtain a one-dimensional time series. The length of the time series is the product of the sampling frequency and the preset time. The one-dimensional time series is decomposed into multiple wavelet components based on a preset wavelet basis function to obtain an approximate component and multiple detail components. The approximate component represents the trend characteristics of the system response delay, and the detail components represent the fluctuation characteristics of different frequency bands. Perform a Fourier transform on each of the detailed components to convert the time-domain signal into a frequency-domain signal, and calculate the power spectral density of each frequency-domain signal. The power spectral density is used to characterize the energy distribution of different frequency components. The peak frequency, center frequency, and bandwidth parameters are extracted from the power spectral density, and the peak frequency, center frequency, and bandwidth parameters are combined to obtain the frequency domain features. The dimension of the frequency domain features is consistent with the number of decomposition layers. The frequency domain features are normalized to scale the parameter values ​​to a preset range.

6. The method as described in claim 1, characterized in that, The step of performing dynamic deviation analysis on the state feedback characteristics to generate collaborative control strategy adjustment parameters includes: The state feedback features are time-aligned with the set of benchmark features from historical collaborative control processes, where the set of benchmark features is the average value of state feedback features collected in historical successful cases. The dimensional deviation values ​​between the aligned state feedback features and the baseline feature set are calculated to obtain the deviation feature sequence. Each element in the deviation feature sequence represents the degree of feature deviation at the corresponding time step. The deviation feature sequence is predicted by a Long Short-Term Memory (LSTM) network. The LSM network includes an input layer, a hidden layer, and an output layer. The input layer receives the deviation feature sequence, the hidden layer memorizes long-term dependencies through a gating mechanism, and the output layer predicts the deviation trend at a preset time step in the future. Based on the deviation trend and the preset deviation threshold, the types of collaborative control parameters that need to be adjusted are determined. The parameter types include resource allocation ratio, task execution priority, and data transmission bandwidth. For each parameter type that needs adjustment, calculate the adjustment magnitude. The magnitude of the adjustment magnitude is positively correlated with the slope of the deviation trend. The adjustment direction is determined based on the sign of the deviation value. Combine the parameter type and the corresponding adjustment magnitude value into the adjustment parameters for the collaborative control strategy.

7. The method as described in claim 6, characterized in that, The step of aligning the state feedback features with the baseline feature set from the historical collaborative control process in a time sequence includes: Retrieve historical cases of the same type as the current collaborative control task from the historical database, extract the state feedback features of each historical case, and obtain the original baseline feature set; Anomaly detection is performed on the original baseline feature set, and statistical methods are used to identify and remove feature samples that deviate from the normal range to obtain a purified baseline feature set. Calculate the mean and standard deviation of each time step in the purified baseline feature set to generate statistical baseline features containing the mean sequence and standard deviation sequence; Adjust the time series length of the current state feedback feature to be consistent with the statistical baseline feature; The mean sequence of the adjusted state feedback feature and the statistical benchmark feature is aligned by using a dynamic time warping algorithm to minimize the cumulative distance between the two sequences and complete the nonlinear time alignment. The determination of the types of collaborative control parameters that need to be adjusted based on the deviation trend and the preset deviation threshold includes: Multiple deviation threshold ranges are preset, and each threshold range corresponds to a different parameter adjustment level; Each predicted deviation value in the deviation trend is matched with a preset deviation threshold range to determine the parameter adjustment level for each time step; Statistically analyze the frequency of adjustment levels for each parameter type within a preset time window, and identify the parameter type with the highest frequency as the primary adjustment target. Analyze the historical adjustment effect data of the main adjustment object. If the deviation reduction rate after parameter adjustment in the historical adjustment effect data is greater than the preset threshold, then retain the parameter type; otherwise, replace it with the second most frequent parameter type. For the identified parameter types, perform correlation analysis to determine whether there are any related parameters that need to be adjusted in a coordinated manner. If so, add the related parameters to the adjustment list to obtain the final set of coordinated control parameter types.

8. The method as described in claim 1, characterized in that, The step of synchronizing the adjustment parameters of the collaborative control strategy to each business system through the cross-system interaction architecture to achieve dynamic optimization of cross-system collaborative control includes: The adjustment parameters of the collaborative control strategy are encoded, and the parameter type and adjustment magnitude value are converted into a binary data stream using a variable-length encoding method; Through the synchronous control module of the cross-system interaction architecture, the encoded binary data stream is grouped according to the business system, and each group of data contains all the parameter information that the business system needs to adjust. Asynchronous transmission mode is used to send packet data to the corresponding business system. If the receiver does not return an acknowledgment message within a preset time, the data is retransmitted. After receiving the parameter adjustment data, each business system performs decoding processing and updates its local collaborative control strategy. Collect system operation status data after parameter adjustment, calculate the collaborative control efficiency index before and after strategy optimization. If the efficiency index improvement is greater than the preset threshold, the dynamic optimization is completed; otherwise, return to the parameter generation step to recalculate the adjustment value.

9. A computer system, characterized in that, include: A memory, wherein a computer program is stored; A processor for loading the computer program to implement the cross-system cooperative control method based on digital threads as described in any one of claims 1-8.

Images