A pon network ai dynamic regulation method and system of a large model fusion

Abstract

The embodiment of the application relates to the technical field of PON network, and specifically discloses a PON network AI dynamic regulation method and system fusing a large model. The embodiment of the application performs multi-source data collection, carries out data fusion and feature extraction on the multi-source collected data, processes structured feature data and historical traffic data based on a fusion large model, generates a plurality of candidate regulation strategies through strategy planning and constraint embedding, selects a target regulation strategy, and executes the strategy; the target strategy benefit is calculated, and the strategy is feedback optimized. The multi-source data collection can be carried out, the future bandwidth demand can be predicted through the fusion large model, the strategy generation and selection are carried out, the optimal target regulation strategy is selected according to the current network state and is executed, the target strategy benefit is calculated, the strategy is feedback optimized, the bandwidth distribution is reasonable, the service quality is ensured, the bandwidth resource waste is avoided, and the regulation efficiency of the PON network is effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of PON network technology, and in particular relates to a PON network AI dynamic control method and system that integrates a large model. Background Technology

[0002] PON networks are fiber optic access technologies based on point-to-multipoint (P2MP) topology. Their core feature is the use of passive optical splitters to distribute signals from a single fiber to multiple user terminals without the need for active electronic devices (such as amplifiers and switches). They are the mainstream solution for current fiber optic access networks (FTTx) and are widely used in scenarios such as home broadband, enterprise leased lines, and mobile base station backhaul.

[0003] Currently, traditional PON network control mainly adopts a static bandwidth allocation method, allocating a fixed bandwidth to each optical network unit. This method cannot predict and dynamically adjust future bandwidth requirements based on actual service needs. Insufficient bandwidth allocation can easily affect service quality and may also lead to some bandwidth being idle, resulting in a waste of bandwidth resources. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for dynamic AI control of PON networks that integrates large models, aiming to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0006] A dynamic AI control method for PON networks integrating large models, the method specifically includes the following steps:

[0007] Multi-source data acquisition is performed to obtain multi-source acquired data, and data fusion and feature extraction are performed on the multi-source acquired data to obtain structured feature data;

[0008] Historical traffic data is acquired, and based on a preset fusion model, the structured feature data and the historical traffic data are processed to obtain future bandwidth requirements and anomaly detection data.

[0009] Based on the future bandwidth requirements and the anomaly detection data, strategy planning and constraint embedding are performed to generate multiple candidate control strategies.

[0010] Obtain the current network status, select a target control strategy from multiple candidate control strategies based on the current network status, and issue and execute the strategy.

[0011] Obtain the expected and actual effects of the strategy, calculate the target strategy benefit, and optimize the strategy based on the target strategy benefit.

[0012] As a further limitation of the technical solution of this embodiment of the invention, the step of performing multi-source data acquisition, obtaining multi-source acquired data, and performing data fusion and feature extraction on the multi-source acquired data to obtain structured feature data specifically includes the following steps:

[0013] Perform multi-source data collection and acquire multi-source collected data;

[0014] The multi-source collected data is effectively extracted and standardized to generate multi-source standard data;

[0015] The multi-source standard data is spatiotemporally aligned and correlated and fused to generate fused standard data;

[0016] The fused standard data is analyzed and extracted for temporal features, spatial features, and contextual features to obtain structured feature data.

[0017] As a further limitation of the technical solution of this embodiment of the invention, the multi-source collected data is composed of network layer data, service layer data and environment layer data. Among them, network layer data includes bandwidth utilization, latency and packet loss rate; service layer data includes service type and QoS requirements; and environment layer data includes fiber optic link loss, temperature and equipment power consumption.

[0018] As a further limitation of the technical solution of this embodiment of the invention, the step of obtaining historical traffic data, based on a preset fusion model, and processing the structured feature data and the historical traffic data to obtain future bandwidth requirements and anomaly detection data specifically includes the following steps:

[0019] Obtain historical traffic data;

[0020] By combining the historical traffic data and the structured feature data, a pre-defined fusion model is used to predict future bandwidth requirements.

[0021] Get real-time traffic;

[0022] Based on the real-time traffic, perform anomaly deviation analysis on the future bandwidth demand and record the anomaly detection data.

[0023] As a further limitation of the technical solution of this invention, the step of generating multiple candidate control strategies by performing strategy planning and constraint embedding based on the future bandwidth requirements and the anomaly detection data specifically includes the following steps:

[0024] Based on the future bandwidth requirements and the anomaly detection data, strategy planning is performed to generate multiple planning and control strategies.

[0025] Constraint analysis is performed on multiple planning and control strategies to determine multiple constraints.

[0026] Multiple constraints are embedded into multiple corresponding planning and control strategies to generate multiple candidate control strategies.

[0027] As a further limitation of the technical solution of this embodiment of the invention, the step of obtaining the current network status, selecting a target control strategy from multiple candidate control strategies based on the current network status, and issuing and executing the strategy specifically includes the following steps:

[0028] Get the current network status;

[0029] Based on the current network state, the optimal target control strategy is dynamically selected from multiple candidate control strategies.

[0030] Based on the target control strategy, generate target control instructions;

[0031] The target control command is issued and executed.

[0032] As a further limitation of the technical solution of this invention embodiment, the steps of obtaining the expected effect and actual effect of the strategy, calculating the target strategy benefit, and performing strategy feedback optimization based on the target strategy benefit specifically include the following steps:

[0033] Based on the target control strategy, determine the expected effect of the corresponding strategy;

[0034] Conduct actual monitoring to obtain the actual effects of the strategy;

[0035] Compare the actual effect of the strategy with the expected effect of the strategy, and calculate the target strategy return;

[0036] Based on the target strategy benefit, plan the strategy optimization parameters, and perform strategy feedback optimization according to the strategy optimization parameters.

[0037] As a further limitation of the technical solution of this embodiment of the invention, the formula for calculating the target strategy benefit is as follows:

[0038] ;

[0039] in, For the target strategy benefit, For actual service quality, To achieve the expected service quality, For expected energy consumption, Actual energy consumption and This is the preset influence coefficient.

[0040] A PON network AI dynamic control system integrating a large model, the system comprising a multi-source data acquisition unit, a demand prediction and anomaly detection unit, a control strategy generation unit, a strategy selection and execution unit, and a strategy feedback optimization unit, wherein:

[0041] A multi-source data acquisition unit is used to acquire multi-source data, obtain multi-source acquired data, and perform data fusion and feature extraction on the multi-source acquired data to obtain structured feature data.

[0042] The demand forecasting and anomaly detection unit is used to acquire historical traffic data, and based on a preset fusion model, process the structured feature data and the historical traffic data to obtain future bandwidth demand and anomaly detection data.

[0043] The regulation strategy generation unit is used to perform strategy planning and constraint embedding based on the future bandwidth demand and the anomaly detection data, and generate multiple candidate regulation strategies.

[0044] The strategy selection and execution unit is used to obtain the current network status, select the target control strategy from multiple candidate control strategies based on the current network status, and issue and execute the strategy.

[0045] The strategy feedback optimization unit is used to obtain the expected effect and actual effect of the strategy, calculate the target strategy benefit, and perform strategy feedback optimization based on the target strategy benefit.

[0046] As a further limitation of the technical solution of this embodiment of the invention, the multi-source data acquisition unit specifically includes:

[0047] The multi-source data acquisition module is used to acquire multi-source data.

[0048] The data preprocessing module is used to effectively extract and standardize the multi-source collected data to generate multi-source standard data;

[0049] The data fusion module is used to perform spatiotemporal alignment and correlation fusion on the multi-source standard data to generate fused standard data;

[0050] The feature extraction module is used to analyze and extract temporal features, spatial features, and contextual features from the fused standard data to obtain structured feature data.

[0051] Compared with the prior art, the beneficial effects of the present invention are:

[0052] This invention, through multi-source data acquisition, performs data fusion and feature extraction on the acquired data; based on a fusion model, it processes structured feature data and historical traffic data; it performs strategy planning and constraint embedding to generate multiple candidate control strategies; it selects a target control strategy and executes it; it calculates the target strategy's benefit and performs strategy feedback optimization. This allows for multi-source data acquisition, prediction of future bandwidth demand through a fusion model, strategy generation and selection, and, based on the current network state, selection and execution of the optimal target control strategy. Furthermore, it calculates the target strategy's benefit and performs strategy feedback optimization, ensuring reasonable bandwidth allocation, guaranteeing service quality, avoiding bandwidth resource waste, and effectively improving the control efficiency of the PON network. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention.

[0054] Figure 1 A flowchart of the method provided by an embodiment of the present invention is shown.

[0055] Figure 2 A flowchart of multi-source data acquisition and processing in the method provided by an embodiment of the present invention is shown.

[0056] Figure 3 The flowchart illustrating the method for obtaining future bandwidth requirements and anomaly detection data provided in an embodiment of the present invention is shown.

[0057] Figure 4 A flowchart of strategy planning and constraint embedding in the method provided by an embodiment of the present invention is shown.

[0058] Figure 5 A flowchart illustrating the selection of a target control strategy in the method provided by an embodiment of the present invention is shown.

[0059] Figure 6 A flowchart illustrating strategy feedback optimization in the method provided by an embodiment of the present invention is shown.

[0060] Figure 7 An application architecture diagram of the system provided in an embodiment of the present invention is shown.

[0061] Figure 8 A structural block diagram of the multi-source data acquisition unit in the system provided by an embodiment of the present invention is shown. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0063] Understandably, current traditional PON network control mainly adopts a static bandwidth allocation method, allocating a fixed bandwidth to each optical network unit. This method cannot predict and dynamically adjust future bandwidth requirements based on actual service needs. Insufficient bandwidth allocation can easily affect service quality and may also lead to some bandwidth being idle, resulting in a waste of bandwidth resources.

[0064] To address the aforementioned issues, this invention employs multi-source data acquisition, fusing and extracting features from the acquired data to obtain structured feature data. Historical traffic data is acquired, and based on a pre-defined fusion model, the structured feature data and historical traffic data are processed to obtain future bandwidth requirements and anomaly detection data. Based on these data, strategy planning and constraint embedding are performed to generate multiple candidate control strategies. The current network state is obtained, and a target control strategy is selected from the candidate strategies and executed. The expected and actual effects of the strategy are obtained, the target strategy's benefit is calculated, and strategy feedback optimization is performed based on this benefit. This approach enables multi-source data acquisition, prediction of future bandwidth requirements through a fusion model, strategy generation and selection, selection of the optimal target control strategy based on the current network state, execution of the target strategy, calculation of its benefit, and strategy feedback optimization. This ensures reasonable bandwidth allocation, guarantees service quality, avoids bandwidth resource waste, and effectively improves the control efficiency of the PON network.

[0065] Figure 1 A flowchart of the method provided by an embodiment of the present invention is shown.

[0066] Specifically, a dynamic AI control method for PON networks integrating large models includes the following steps:

[0067] Step S101: Perform multi-source data acquisition, obtain multi-source acquired data, and perform data fusion and feature extraction on the multi-source acquired data to obtain structured feature data.

[0068] In this embodiment of the invention, multi-source data, including bandwidth utilization, latency, packet loss rate, service type, QoS requirements, fiber link loss, temperature, and device power consumption, are collected from the network layer, service layer, and environment layer. The multi-source data is then effectively identified, and effective data is extracted from it. According to preset standardization requirements, the data is standardized to generate multi-source standard data. Subsequently, the multi-source standard data is spatiotemporally aligned and correlated to generate fused standard data. Finally, the fused standard data is analyzed and extracted for temporal features, spatial features, and contextual features to obtain structured feature data.

[0069] It is understandable that multi-source collected data consists of network layer data, service layer data, and environment layer data. Among them, network layer data includes bandwidth utilization, latency, and packet loss rate; service layer data includes service type and QoS requirements; and environment layer data includes fiber optic link loss, temperature, and equipment power consumption.

[0070] Understandably, temporal characteristics refer to the short-term fluctuation patterns and periodic trends of traffic; spatial characteristics refer to the traffic correlation between regions; and contextual characteristics refer to the business priority.

[0071] Specifically, Figure 2 A flowchart of multi-source data acquisition and processing in the method provided by an embodiment of the present invention is shown.

[0072] In a preferred embodiment of the present invention, the step of performing multi-source data acquisition, obtaining multi-source acquired data, and performing data fusion and feature extraction on the multi-source acquired data to obtain structured feature data specifically includes the following steps:

[0073] Step S1011: Perform multi-source data acquisition to obtain multi-source acquired data;

[0074] Step S1012: Effectively extract and standardize the multi-source collected data to generate multi-source standard data;

[0075] Step S1013: Perform spatiotemporal alignment and correlation fusion on the multi-source standard data to generate fused standard data;

[0076] Step S1014: Analyze and extract temporal features, spatial features and contextual features from the fused standard data to obtain structured feature data.

[0077] Furthermore, the PON network AI dynamic control method based on the fusion of large models also includes the following steps:

[0078] Step S102: Obtain historical traffic data. Based on a preset fusion model, process the structured feature data and the historical traffic data to obtain future bandwidth requirements and anomaly detection data.

[0079] In this embodiment of the invention, historical traffic data is acquired, combined with structured feature data, and then processed using a preset fusion model to predict demand for a preset future time period to obtain future bandwidth demand. Real-time traffic is also acquired, and based on the real-time traffic, anomaly deviation analysis is performed on the future bandwidth demand. The deviation between the real-time traffic and the predicted value corresponding to the future bandwidth demand is compared with a threshold, and anomaly detection data is recorded.

[0080] It is understood that the fusion model in the embodiments of the present invention is a large model that fuses LSTM and graph neural networks.

[0081] Specifically, Figure 3 The flowchart illustrating the method for obtaining future bandwidth requirements and anomaly detection data provided in an embodiment of the present invention is shown.

[0082] In a preferred embodiment of the present invention, the step of acquiring historical traffic data, based on a preset fusion model, and processing the structured feature data and the historical traffic data to obtain future bandwidth requirements and anomaly detection data specifically includes the following steps:

[0083] Step S1021: Obtain historical traffic data;

[0084] Step S1022: Combining the historical traffic data and the structured feature data, predict future bandwidth demand using a preset fusion model;

[0085] Step S1023: Obtain real-time traffic;

[0086] Step S1024: Based on the real-time traffic, perform anomaly deviation analysis on the future bandwidth demand and record the anomaly detection data.

[0087] Furthermore, the PON network AI dynamic control method based on the fusion of large models also includes the following steps:

[0088] Step S103: Based on the future bandwidth requirements and the anomaly detection data, perform strategy planning and constraint embedding to generate multiple candidate control strategies.

[0089] In this embodiment of the invention, strategy planning is performed based on future bandwidth requirements and anomaly detection data to generate multiple planning and control strategies, including dynamic wavelength allocation, bandwidth allocation, and route switching. Constraint analysis is performed on the multiple planning and control strategies to determine multiple constraints, including QoS guarantees and energy consumption limits. These constraints are then embedded into the multiple corresponding planning and control strategies to generate multiple candidate control strategies.

[0090] Specifically, Figure 4 A flowchart of strategy planning and constraint embedding in the method provided by an embodiment of the present invention is shown.

[0091] In a preferred embodiment of the present invention, the step of generating multiple candidate control strategies by performing strategy planning and constraint embedding based on the future bandwidth requirements and the anomaly detection data specifically includes the following steps:

[0092] Step S1031: Based on the future bandwidth requirements and the anomaly detection data, perform strategy planning to generate multiple planning and control strategies;

[0093] Step S1032: Perform constraint analysis on the multiple planning and control strategies to determine multiple constraint conditions;

[0094] Step S1033: Embed the multiple constraints into multiple corresponding planning and control strategies to generate multiple candidate control strategies.

[0095] Furthermore, the PON network AI dynamic control method based on the fusion of large models also includes the following steps:

[0096] Step S104: Obtain the current network status, select the target control strategy from multiple candidate control strategies based on the current network status, and issue and execute the strategy.

[0097] In this embodiment of the invention, the current network status is obtained, including real-time traffic, device load, etc., and then the current network status and multiple candidate control strategies are analyzed based on a preset real-time decision model. The optimal target control strategy is dynamically selected from the multiple candidate control strategies, and then a corresponding target control instruction is generated according to the target control strategy. The target control instruction is then sent to the OLT / ONU device for strategy execution.

[0098] Specifically, Figure 5 A flowchart illustrating the selection of a target control strategy in the method provided by an embodiment of the present invention is shown.

[0099] In a preferred embodiment of the present invention, the step of obtaining the current network status, selecting a target control strategy from a plurality of candidate control strategies based on the current network status, and issuing and executing the strategy specifically includes the following steps:

[0100] Step S1041: Obtain the current network status;

[0101] Step S1042: Based on the current network state, dynamically select the optimal target control strategy from multiple candidate control strategies;

[0102] Step S1043: Generate target control instructions according to the target control strategy;

[0103] Step S1044: The target control command is issued and executed.

[0104] Furthermore, the PON network AI dynamic control method based on the fusion of large models also includes the following steps:

[0105] Step S105: Obtain the expected effect and actual effect of the strategy, calculate the target strategy benefit, and perform strategy feedback optimization based on the target strategy benefit.

[0106] In this embodiment of the invention, the expected effect of the target control strategy is determined according to the target control strategy. During the execution of the target control strategy, actual monitoring is performed on changes in latency reduction rate, packet loss rate, and energy consumption to obtain the actual effect of the strategy. The actual effect of the strategy is compared with the expected effect to calculate the target strategy benefit. Based on the target strategy benefit, strategy optimization parameters are planned, and then strategy feedback optimization is performed according to the optimization parameters. Specifically, the formula for calculating the target strategy benefit is:

[0107] ;

[0108] in, For the target strategy benefit, For actual service quality, To achieve the expected service quality, For expected energy consumption, Actual energy consumption and This is the preset influence coefficient.

[0109] Specifically, Figure 6 A flowchart illustrating strategy feedback optimization in the method provided by an embodiment of the present invention is shown.

[0110] In a preferred embodiment of the present invention, the steps of obtaining the expected effect and actual effect of the strategy, calculating the target strategy benefit, and performing strategy feedback optimization based on the target strategy benefit specifically include the following steps:

[0111] Step S1051: Determine the expected effect of the corresponding strategy based on the target control strategy;

[0112] Step S1052: Conduct actual monitoring to obtain the actual effect of the strategy;

[0113] Step S1053: Compare the actual effect of the strategy with the expected effect of the strategy, and calculate the target strategy return;

[0114] Step S1054: Based on the target strategy benefit, plan the strategy optimization parameters, and perform strategy feedback optimization according to the strategy optimization parameters.

[0115] Furthermore, Figure 7 An application architecture diagram of the system provided in an embodiment of the present invention is shown.

[0116] In another preferred embodiment of the present invention, a PON network AI dynamic control system integrating a large model includes:

[0117] The multi-source data acquisition unit 101 is used to acquire multi-source data, obtain multi-source acquired data, and perform data fusion and feature extraction on the multi-source acquired data to obtain structured feature data.

[0118] In this embodiment of the invention, the multi-source data acquisition unit 101 acquires multi-source data, including bandwidth utilization, latency, packet loss rate, service type, QoS requirements, fiber link loss, temperature, and device power consumption, by collecting data from the network layer, service layer, and environment layer. The multi-source data is then effectively identified, and effective data is extracted from it. According to preset standardization requirements, the data is standardized to generate multi-source standard data. Subsequently, the multi-source standard data is spatiotemporally aligned and correlated to generate fused standard data. Finally, the fused standard data is analyzed and extracted for temporal features, spatial features, and contextual features to obtain structured feature data.

[0119] Specifically, Figure 8 A structural block diagram of the multi-source data acquisition unit 101 in the system provided by an embodiment of the present invention is shown.

[0120] In a preferred embodiment provided by the present invention, the multi-source data acquisition unit 101 specifically includes:

[0121] The multi-source data acquisition module 1011 is used to perform multi-source data acquisition and obtain multi-source acquired data.

[0122] The data preprocessing module 1012 is used to effectively extract and standardize the multi-source collected data to generate multi-source standard data;

[0123] Data fusion module 1013 is used to perform spatiotemporal alignment and correlation fusion on the multi-source standard data to generate fused standard data;

[0124] The feature extraction module 1014 is used to analyze and extract temporal features, spatial features and contextual features from the fused standard data to obtain structured feature data.

[0125] Furthermore, the PON network AI dynamic control system that integrates large models also includes:

[0126] The demand forecasting and anomaly detection unit 102 is used to acquire historical traffic data, process the structured feature data and the historical traffic data based on a preset fusion model, and obtain future bandwidth demand and anomaly detection data.

[0127] In this embodiment of the invention, the demand forecasting and anomaly detection unit 102 acquires historical traffic data, combines the historical traffic data with structured feature data, and then processes the data through a preset fusion model to forecast the demand for a preset future time period, thereby obtaining the future bandwidth demand. It also acquires real-time traffic, performs anomaly deviation analysis on the future bandwidth demand based on the real-time traffic, compares the deviation between the real-time traffic and the predicted value corresponding to the future bandwidth demand with a threshold, and records the anomaly detection data.

[0128] The regulation strategy generation unit 103 is used to perform strategy planning and constraint embedding based on the future bandwidth requirements and the anomaly detection data, and generate multiple candidate regulation strategies.

[0129] In this embodiment of the invention, the regulation strategy generation unit 103 performs strategy planning based on future bandwidth requirements and anomaly detection data, generating multiple planning regulation strategies including dynamic wavelength allocation, bandwidth allocation, and route switching. It also performs constraint analysis on the multiple planning regulation strategies to determine multiple constraints including QoS guarantee and energy consumption limit, and then embeds the multiple constraints into the multiple corresponding planning regulation strategies to generate multiple candidate regulation strategies.

[0130] The strategy selection and execution unit 104 is used to obtain the current network status, select the target control strategy from multiple candidate control strategies based on the current network status, and execute the strategy.

[0131] In this embodiment of the invention, the strategy selection execution unit 104 obtains the current network status, including real-time traffic, device load, etc., and then analyzes the current network status and multiple candidate control strategies based on a preset real-time decision model. From the multiple candidate control strategies, it dynamically selects the optimal target control strategy, and then generates a corresponding target control instruction according to the target control strategy. The target control instruction is then sent to the OLT / ONU device for strategy execution.

[0132] The strategy feedback optimization unit 105 is used to obtain the expected effect and actual effect of the strategy, calculate the target strategy benefit, and perform strategy feedback optimization based on the target strategy benefit.

[0133] In this embodiment of the invention, the strategy feedback optimization unit 105 determines the expected effect of the corresponding strategy based on the target control strategy. During the execution of the target control strategy, it monitors actual changes in latency reduction rate, packet loss rate, and energy consumption to obtain the actual effect of the strategy. By comparing the actual effect with the expected effect, it calculates the target strategy benefit. Based on the target strategy benefit, it plans strategy optimization parameters and then performs strategy feedback optimization according to these parameters. Specifically, the formula for calculating the target strategy benefit is as follows:

[0134] ;

[0135] in, For the target strategy benefit, For actual service quality, To achieve the expected service quality, For expected energy consumption, Actual energy consumption and This is the preset influence coefficient.

[0136] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0137] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0138] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0139] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0140] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for AI dynamic regulation of a PON network with a large model, characterized in that, The method specifically comprises the following steps: Multi-source data acquisition is performed to obtain multi-source acquisition data, and data fusion and feature extraction are performed on the multi-source acquisition data to obtain structured feature data; the multi-source acquisition data is composed of network layer data, service layer data and environment layer data; Historical traffic data is obtained, the structured feature data and the historical traffic data are processed based on a preset fusion large model, future bandwidth demand and anomaly detection data are obtained; Strategy planning and constraint embedding are performed according to the future bandwidth demand and the anomaly detection data, and a plurality of candidate regulation strategies are generated; A current network state is obtained, a target regulation strategy is selected from the plurality of candidate regulation strategies according to the current network state, and strategy issuing and execution are performed; Strategy expected effect and strategy actual effect are obtained, target strategy benefit is calculated, and strategy feedback optimization is performed according to the target strategy benefit; The step of obtaining historical traffic data, processing the structured feature data and the historical traffic data based on a preset fusion large model, and obtaining future bandwidth demand and anomaly detection data specifically comprises the following steps: Obtain historical traffic data; Combine the historical traffic data and the structured feature data, and predict future bandwidth demand through a preset fusion large model; Obtain real-time traffic; Based on the real-time traffic, perform anomaly deviation analysis on the future bandwidth demand, and record anomaly detection data.

2. The PON network AI dynamic regulation method of the fusion large model according to claim 1, characterized in that, The step of performing multi-source data acquisition, obtaining multi-source acquisition data, and performing data fusion and feature extraction on the multi-source acquisition data to obtain structured feature data specifically comprises the following steps: Multi-source data acquisition is performed to obtain multi-source acquisition data; Effective extraction and standardization processing are performed on the multi-source acquisition data to generate multi-source standard data; Temporal and spatial alignment and associated fusion are performed on the multi-source standard data to generate fusion standard data; Temporal feature, spatial feature and context feature analysis and extraction are performed on the fusion standard data to obtain structured feature data.

3. The PON network AI dynamic regulation method of the fusion large model according to claim 2, characterized in that, The network layer data includes bandwidth utilization, delay and packet loss rate; the service layer data includes service type and QoS demand; and the environment layer data includes fiber link loss, temperature and device power consumption.

4. The PON network AI dynamic regulation method of the fusion large model according to claim 1, characterized in that, The step of performing strategy planning and constraint embedding according to the future bandwidth demand and the anomaly detection data to generate a plurality of candidate regulation strategies specifically comprises the following steps: Strategy planning is performed according to the future bandwidth demand and the anomaly detection data to generate a plurality of planning regulation strategies; Constraint analysis is performed on the plurality of planning regulation strategies to determine a plurality of constraint conditions; The plurality of constraint conditions are embedded into a plurality of corresponding planning regulation strategies to generate a plurality of candidate regulation strategies.

5. The PON network AI dynamic regulation method of the fusion large model according to claim 1, characterized in that, The step of obtaining a current network state, selecting a target regulation strategy from the plurality of candidate regulation strategies according to the current network state, and performing strategy issuing and execution specifically comprises the following steps: A current network state is obtained; An optimal target regulation strategy is dynamically selected from the plurality of candidate regulation strategies according to the current network state; A target regulation instruction is generated according to the target regulation strategy; The target regulation instruction is issued and executed.

6. The PON network AI dynamic regulation method of the fusion large model according to claim 1, characterized in that, The acquisition strategy expected effect and strategy actual effect, calculate the target strategy benefit, according to the target strategy benefit, strategy feedback optimization specifically includes the following steps: According to the target control strategy, determine the corresponding strategy expected effect; Actual monitoring, obtain the strategy actual effect; Compare the strategy actual effect with the strategy expected effect, calculate the target strategy benefit; According to the target strategy benefit, planning strategy optimization parameters, and according to the strategy optimization parameters, strategy feedback optimization.

7. The PON network AI dynamic regulation method of the fusion large model according to claim 6, characterized in that, The formula for calculating the target strategy benefit is: ； wherein, is a target policy benefit, is an actual quality of service, is an expected quality of service, is an expected energy consumption, is an actual energy consumption, and is a preset influence coefficient.

8. A PON network AI dynamic regulation system of a fusion large model, characterized in that, The system includes multi-source data acquisition unit, demand prediction and anomaly detection unit, control strategy generation unit, strategy selection execution unit and strategy feedback optimization unit, wherein: Multi-source data acquisition unit, for multi-source data acquisition, obtain multi-source acquisition data, and carry out data fusion and feature extraction on the multi-source acquisition data, obtain structured feature data; The multi-source acquisition data is composed of network layer data, business layer data and environment layer data; Demand prediction and anomaly detection unit, for obtaining historical traffic data, based on the preset fusion big model, processing the structured feature data and the historical traffic data, obtaining future bandwidth demand and anomaly detection data; Control strategy generation unit, for strategy planning and constraint embedding according to the future bandwidth demand and the anomaly detection data, generating a plurality of candidate control strategies; Strategy selection execution unit, for obtaining the current network state, selecting the target control strategy from the plurality of candidate control strategies according to the current network state, and performing strategy issuing and execution; Strategy feedback optimization unit, for obtaining strategy expected effect and strategy actual effect, calculating target strategy benefit, according to the target strategy benefit, strategy feedback optimization; The acquisition of historical traffic data, based on the preset fusion big model, processing the structured feature data and the historical traffic data, obtaining future bandwidth demand and anomaly detection data specifically includes the following steps: Obtain historical traffic data; Combine the historical traffic data and the structured feature data, predict future bandwidth demand through the preset fusion big model; Obtain real-time traffic; Based on the real-time traffic, carry out abnormal deviation analysis on the future bandwidth demand, record anomaly detection data.

9. The PON network AI dynamic regulation system of the fusion large model according to claim 8, wherein, The multi-source data acquisition unit specifically includes: Multi-source data acquisition module, for multi-source data acquisition, obtain multi-source acquisition data; Data preprocessing module, for effective extraction and standardization processing of the multi-source acquisition data, generating multi-source standard data; Data fusion module, for time and space alignment and associated fusion of the multi-source standard data, generating fusion standard data; Feature extraction module, for analyzing and extracting time series features, spatial features and context features of the fusion standard data, obtaining structured feature data.

Images