Communication frequency prediction method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of communication, in particular to a communication frequency prediction method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining an aggregation code of each initial model in a plurality of model class clusters and the fitness of each initial model in each communication scene; based on the fitness, initial models meeting a preset fitness condition are screened out from all the model class clusters, and a target model set is formed; transforming a basic code in the aggregation code of each initial model to change each initial model, and adding the changed initial model into the target model set; taking all the target models in the added target model set as new initial models, and repeating the steps until a preset iteration stopping condition is met; and performing communication frequency prediction based on the final target model and the fitness of each target model in each communication scene. According to the invention, the communication frequency can be predicted.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and more specifically, to a communication frequency prediction method, apparatus, electronic device, and storage medium. Background Technology

[0002] Communication frequency prediction is the process of modeling and analyzing historical spectrum data to estimate the signal strength, occupancy status, or interference level of specific frequency bands in the future. In the field of wireless communication, spectrum resources are extremely scarce, and accurate frequency prediction can bring many benefits. For example, it can effectively avoid channel conflicts, such as interference between 5G communication and radar bands; improve the efficiency of dynamic spectrum access, such as enabling more flexible spectrum allocation in cognitive radio; reduce base station energy consumption by optimizing resource utilization through on-demand bandwidth allocation; and ensure the reliability of critical communications, such as reserving necessary resources for emergency frequency bands.

[0003] Given these important aspects, the research and application of communication frequency prediction are particularly urgent. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a communication frequency prediction method, apparatus, electronic device and storage medium capable of predicting communication frequencies.

[0005] In a first aspect, embodiments of this application provide a communication frequency prediction method, the method comprising: The aggregation code of each initial model in multiple model clusters and the fitness of each initial model in various communication scenarios are obtained; the model clusters are obtained by clustering all initial models based on the aggregation code of each initial model; the aggregation code is composed of the basic codes of multiple sub-modules in the initial model; Based on the fitness, initial models that meet the preset fitness conditions are selected from each model cluster to form a target model set; The basic encoding in the aggregate encoding of each initial model is transformed to change each initial model, and the changed initial model is added to the target model set; All target models in the added target model set are used as new initial models. Then, the process jumps to obtaining the aggregate encoding of each initial model in each model cluster and the fitness of each initial model in each communication scenario, and continues until the preset iteration stop condition is met. Communication frequency prediction is performed based on the final target model and the fitness of each target model in various communication scenarios.

[0006] In one possible implementation, all initial models are clustered based on the aggregation encoding of each initial model to obtain multiple model clusters through the following steps: Based on the aggregated encoding of each initial model, the similarity between any two initial models is calculated. Based on the similarity, all initial models are clustered to obtain multiple model clusters.

[0007] In one possible implementation, the step of selecting initial models that meet preset fitness conditions from various model clusters based on the fitness, thereby forming a target model set, includes: For each model cluster, a predetermined number of initial models with the highest fitness in the model cluster are determined as the first intermediate models; The target model set is determined based on the fitness of the first intermediate model and each initial model in various communication scenarios.

[0008] In one possible implementation, determining the target model set based on the fitness of the first intermediate model and each initial model in each communication scenario includes: According to a preset ratio, select the initial models with the highest fitness from all initial models to obtain the second intermediate model; Merge all first intermediate models and all second intermediate models to obtain the target model set.

[0009] In one possible implementation, the basic encoding in the aggregate encoding of any initial model is transformed by the following steps to change the initial model: For each submodule in each initial model, the mutation intensity corresponding to the submodule in the initial model is calculated based on the mean gradient and learning rate of the submodule in the initial model. And based on the mutation intensity corresponding to the sub-module in the initial model and the maximum mutation intensity corresponding to all sub-modules in the initial model, the mutation probability corresponding to the sub-module in the initial model is calculated; Based on the mutation intensity and mutation probability of each sub-module in the initial model, the basic encoding of each sub-module in the aggregate encoding of the initial model is mutated to change the initial model.

[0010] In one possible implementation, the communication frequency prediction based on the final target models and their fitness in various communication scenarios includes: Obtain the communication frequency of the communication scenario to be predicted at the current moment; Based on the fitness of each target model in the communication scenario to be predicted, a prediction model is selected from all target models. The current communication frequency is input into the prediction model to obtain the communication frequency at the next moment.

[0011] In one possible implementation, selecting a prediction model from all target models based on the fitness of each target model in the communication scenario to be predicted includes: Based on the fitness of each target model in the communication scenario to be predicted, the initial selection probability of each target model is calculated. For each target model, the initial selection probabilities of all target models preceding the target model are summed to obtain the target selection probability of the target model; the target models preceding the target model refer to models with preset numbers less than the preset number of the target model. If the target model's probability of being selected is greater than the probability of being randomly selected, and the probability of being randomly selected is greater than the probability of being selected by the previous target model corresponding to the target model, then the target model is determined as the prediction model.

[0012] Secondly, embodiments of this application also provide a communication frequency prediction device, the device comprising: The acquisition module is used to acquire the aggregated encoding of each initial model in multiple model clusters and the fitness of each initial model in various communication scenarios; the model clusters are obtained by clustering all initial models based on the aggregated encoding of each initial model; the aggregated encoding is composed of the basic encoding of multiple sub-modules in the initial model; The filtering module is used to filter out initial models that meet the preset fitness conditions from each model cluster based on the fitness, and form a target model set; The transformation module is used to transform the basic encoding in the aggregate encoding of each initial model to change each initial model, and add the changed initial model to the target model set. The jump module is used to take all the target models in the added target model set as new initial models, jump to the step of obtaining the aggregate encoding of each initial model in each model cluster and the fitness of each initial model in each communication scenario, so as to continue execution until the preset iteration stop condition is met. The prediction module is used to predict communication frequencies based on the final target model and the fitness of each target model in various communication scenarios.

[0013] Thirdly, embodiments of this application also provide an electronic device, including: a processor, a storage medium, and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to perform the steps of the communication frequency prediction method as described in any of the first aspects.

[0014] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the communication frequency prediction method as described in any of the first aspects.

[0015] This application provides a communication frequency prediction method, apparatus, electronic device, and storage medium. The method includes: acquiring the aggregated encoding of each initial model in multiple model clusters and the fitness of each initial model in various communication scenarios; selecting initial models that meet preset fitness conditions from each model cluster based on fitness to form a target model set; transforming the basic encoding in the aggregated encoding of each initial model to change each initial model, and adding the changed initial models to the target model set; using all target models in the added target model set as new initial models, repeating the above steps until a preset iteration stop condition is met; and predicting communication frequencies based on the final target models and the fitness of each target model in various communication scenarios. This application enables the prediction of communication frequencies. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart of a communication frequency prediction method provided in an embodiment of this application is shown; Figure 2 This illustration shows a breakdown diagram of the initial model provided in an embodiment of this application; Figure 3 This illustration shows a schematic diagram of the disassembly of the model body provided in an embodiment of this application; Figure 4 This illustration shows a variation diagram of the initial model provided in the embodiments of this application; Figure 5 A cross diagram of the initial model provided in the embodiments of this application is shown; Figure 6 A flowchart illustrating the model iteration process provided in an embodiment of this application is shown; Figure 7 A flowchart illustrating the prediction process for communication frequencies provided in an embodiment of this application is shown. Figure 8 This paper shows a schematic diagram of the structure of a communication frequency prediction device provided in an embodiment of this application; Figure 9A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0019] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0020] To enable those skilled in the art to utilize the content of this application, and in conjunction with the specific application scenario of "communication technology," the following embodiments are provided. For those skilled in the art, the general principles defined herein can be applied to other embodiments and application scenarios without departing from the spirit and scope of this application. Although this application is primarily described within the "communication technology field," it should be understood that this is merely an exemplary embodiment.

[0021] It should be noted that the term "comprising" will be used in the embodiments of this application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.

[0022] The following is a detailed description of a communication frequency prediction method provided by an embodiment of this application.

[0023] Reference Figure 1 The diagram shown is a flowchart illustrating a communication frequency prediction method provided in an embodiment of this application. The exemplary steps of this embodiment are described below: S101. Obtain the aggregated encoding of each initial model in multiple model clusters and the adaptability of each initial model in various communication scenarios.

[0024] In this embodiment, model clusters are obtained by clustering all initial models based on their aggregate codes; initial models within the same model cluster have similar functional characteristics. The aggregate code consists of the base codes of all sub-modules within the initial model. The fitness of the initial model in the communication scenario is used to evaluate the model's performance, covering the accuracy of the initial model in predicting communication frequencies, resource consumption, and competition with other initial models in that scenario. The base code of a sub-module includes a preset identifier string corresponding to the sub-module and a preset sequence number corresponding to the initial model. Sub-modules with the same network structure have different preset identifier strings, while sub-modules with the same network structure have the same preset identifier string.

[0025] Here, before obtaining the aggregated encoding of each initial model in multiple model clusters and the fitness of each initial model in various communication scenarios, it is necessary to pre-build an initial model library. Specifically, this includes: selecting at least one initial model from the time series prediction models of each preset model category and adding it to the initial model library; the preset model categories refer to the model categories that can be used for communication frequency prediction, including regression mobility models, time series decomposition models, and neural network models. For example, regression mobility models include ARIMA, SARIMAX, VAR, VARMAX, ARMD, Holt-Winters, etc.; time series decomposition models include Prophet, SEATS, STL, TSR, X11 decomposition, etc.; and neural network models include TCN, LSTM, Timer-XL, Autoformer, Transformer and its variants, etc.

[0026] Furthermore, due to the significant differences in model types and structures among different initial models, it is not conducive to subsequent iterative evolution operations. Therefore, this application designs a model gene encoding strategy to uniformly modularize (block) the models in the model library. Specifically, the gene encoding of each initial model in the initial model library is performed through the following steps to obtain the aggregate encoding of each initial model: Step 1: Divide the initial model into a preprocessing module, the main model body, and a postprocessing module.

[0027] In the embodiments of this application, reference is made to Figure 2The diagram illustrates the breakdown of the initial model provided in this embodiment. The preprocessing module and postprocessing module do not participate in the training of the initial model and are each treated as an independent block within the initial model. Specifically, the preprocessing module is the initial part of the initial model without trainable model parameters, used for operations such as differencing, channel expansion, and loop stacking of the frequency sequence. The postprocessing module is the later part of the initial model without trainable model parameters, used to restore the output of the initial model to a frequency sequence. The main body of the model is the portion between the preprocessing and postprocessing modules, containing the trainable model parameters.

[0028] Step 2: First, decompose the main body of the initial model into the smallest recurring functional units to obtain multiple smallest functional modules; then, take the non-recurring key units in the decomposed main body of the model as a module to obtain non-recurring key modules.

[0029] Typical examples of recurring minimum functional units include: residual blocks in ResNet, Inception blocks in GoogLeNet, and Transformer coding blocks.

[0030] In the embodiments of this application, reference is made to Figure 3 The diagram shown is a breakdown of the main body of the model provided in this embodiment. The non-repeating key modules are generally the input processing module and the output head. Therefore, after splitting, it sequentially includes non-repeating key module 1, minimum functional module 1, minimum functional module 2, ..., minimum functional module n, and non-repeating key module 2.

[0031] For example, if the main body of the model includes an input processing block, nine consecutive residual blocks, and an output head, then the sub-modules after the main body of the model are split into the following: input processing block, residual block 1, residual block 2, residual block 3, residual block 4, residual block 5, residual block 6, residual block 7, residual block 8, residual block 9, and output head.

[0032] Step 3: The preprocessing module, postprocessing module, minimum functional module and non-repeating key module obtained after the split are all used as sub-modules of the initial model.

[0033] Step 4: For each submodule, use the preset identifier string corresponding to the submodule as the prefix of the preset sequence number corresponding to the initial model to obtain the basic code corresponding to the submodule.

[0034] For example, suppose the preprocessing module uses a preset identifier string 'y' as a prefix, and the postprocessing module uses a preset identifier string 'h' as ​​a prefix. Then, the basic encoding corresponding to the preprocessing module of the initial model with preset sequence number 01 is y01, and the basic encoding corresponding to the postprocessing module with preset sequence number 02 is h02.

[0035] Step 5: Arrange the basic codes of each sub-module according to their position in the initial model to obtain the aggregate code of the initial model.

[0036] In the embodiments of this application, through the aforementioned gene encoding processing, all initial models in the initial model library can be represented by block units. For example, an initial model including a preprocessing module + model body (input processing module + residual block + Inception block + output header) + postprocessing module can be represented by the aggregate encoding y01i01r01in01o01h01, where y01 is the preprocessing module of the initial model with preset sequence number 01, i01 is the input processing module of the initial model with preset sequence number 01, r01 is the residual block of the initial model with preset sequence number 01, in01 is the Inception block of the initial model with preset sequence number 01, o01 is the output header of the initial model with preset sequence number 01, and h01 is the postprocessing module of the initial model with preset sequence number 01.

[0037] Furthermore, based on the aggregation encoding of each initial model, all initial models are clustered to obtain multiple model clusters through the following steps: Step 1: Calculate the similarity between any two initial models based on the aggregated encoding of each initial model.

[0038] In this embodiment, the similarity between two initial models is obtained by substituting the aggregated encoding of each pair of initial models into the following formula: ; in, For each pair of initial models, the aggregate encoding is used for the first initial model. For the aggregation encoding of the second initial model out of every two initial models. The similarity between the two initial models; for and The number of identical preset identifier strings; for and The number of strings (twice the total number of submodules in the two initial models, including all preset identifier strings and all preset serial numbers of the initial models). for X The number of strings (twice the number of submodules in the first initial model). for Y The number of strings (twice the number of submodules in the second initial model).

[0039] Step 2: Cluster all initial models based on the similarity to obtain multiple model clusters.

[0040] In this embodiment, two initial models with a similarity greater than a preset similarity threshold (e.g., 80%) can be placed in the same model cluster. Alternatively, based on the similarity between any two initial models, DBSCAN can be used to cluster all models, thereby forming multiple model clusters.

[0041] Here, mimicking examples of natural evolution, initial models within the same model cluster occupy the same niche level. Therefore, it can be assumed that there is strong competition between initial models within the same model cluster, while the competition between models from different model clusters is weaker. The distribution of model clusters will affect the subsequent model iteration process.

[0042] Furthermore, the initial models in the initial model library are subjected to communication frequency prediction tasks in various different communication environments, and the fitness of each initial model in each communication environment is evaluated based on the prediction results. Specifically, the fitness of any initial model in any communication scenario is calculated using the following formula: ; in, The fitness of the initial model in this communication scenario, As the first preset weight, This represents the accuracy of the initial model in predicting communication frequencies in this communication scenario. Preset a precise baseline value (e.g., 0.95). As the second preset weight, The resource consumption of the initial model when predicting communication frequency in this communication scenario (in this embodiment, it is represented by "the time required to execute one communication frequency task"). The maximum acceptable resource overhead is preset (e.g., 200ms). As the third preset weight, This represents the number of initial models in the model cluster to which the initial model belongs (i.e., the number of initial models that compete with this initial model). The maximum number of competing models is preset.

[0043] S102. Based on fitness, select initial models that meet the preset fitness conditions from each model cluster to form the target model set.

[0044] In this embodiment of the application, in order to avoid the loss of specific types of model structures and high-performing models during the iteration process, the initial model that meets the preset fitness conditions is forcibly selected from each model cluster based on fitness during each iteration. This can retain the most comprehensive and high-quality target model to the greatest extent and improve the accuracy of subsequent communication frequency prediction.

[0045] Specifically, based on fitness, initial models that meet preset fitness conditions are selected from various model clusters to form a target model set, including: Step 1: For each model cluster, determine the preset number of initial models with the highest fitness in that model cluster as the first intermediate models.

[0046] In this embodiment of the application, in order to avoid the loss of a specific type of model structure during the iteration process, a preset number of initial models with the highest fitness are selected from each model cluster and determined as the first intermediate models.

[0047] For example, there are model clusters A and B, with a preset number of 3. Then, the 3 initial models with the highest fitness from model cluster A are selected as the first intermediate models; and the 3 initial models with the highest fitness from model cluster B are selected as the first intermediate models.

[0048] Continuing the previous example, suppose model cluster A contains initial model 1, initial model 2, initial model 3, initial model 4, and initial model 5. Their fitness ranking is: initial model 1 > initial model 3 > initial model 5 > initial model 2 > initial model 4. Therefore, initial model 1, initial model 3, and initial model 5 in model cluster A are determined as the first intermediate model.

[0049] Step 2: Determine the target model set based on the fitness of the first intermediate model and each initial model in each communication scenario.

[0050] In this embodiment of the application, in order to avoid losing high-performing models during the iteration process, it is also necessary to select high-performing models from all initial models. Specifically, according to a preset ratio, multiple initial models with the highest fitness are selected from all initial models to obtain second intermediate models; all first intermediate models and all second intermediate models are merged to obtain the target model set.

[0051] For example, if the preset ratio is 10%, then the initial models with the top 10% fitness will be selected from all the initial models and used as the second intermediate models.

[0052] Optionally, to avoid duplicate model structures between the first and second intermediate models, it is necessary to deduplicate the target models in the target model set to obtain the final target model set.

[0053] S103. Transform the basic encoding in the aggregation encoding of each initial model to change each initial model, and add the changed initial model to the target model set.

[0054] In this embodiment of the application, in order to reduce the amount of computation and improve the efficiency of model iteration, the basic encoding in the aggregate encoding of the initial model with the highest fitness can be transformed according to a preset transformation ratio to change each initial model.

[0055] For example, the base encoding in the aggregate encoding of the initial model with fitness in the top 70% is transformed, and the remaining 30% of the initial models are discarded.

[0056] Specifically, the basic encoding in the aggregate encoding of any initial model is transformed through the following steps to change the initial model, including: Step 1: For each submodule in each initial model, calculate the mutation intensity corresponding to that submodule in the initial model based on the mean gradient and learning rate of that submodule.

[0057] In this embodiment, mutation intensity refers to the degree of mutation performed on the submodule. Appropriately setting the mutation intensity is crucial to the performance of the final target model. If the mutation intensity is too high, it may cause premature exit from local optima, thus failing to obtain the optimal target model; if the mutation intensity is too low, it may lead to getting trapped in local optima, preventing the exploration of a better target model. Therefore, this embodiment dynamically calculates the mutation intensity corresponding to the submodule in the initial model based on the average gradient and learning rate of the submodule.

[0058] Specifically, by substituting the mean gradient and learning rate of the submodule in the initial model into the following formula, the mutation intensity corresponding to the submodule in the initial model is obtained: ; in, This represents the mutation intensity corresponding to this submodule in the initial model. This represents the learning rate for this submodule in the initial model. This represents the mean gradient of this submodule in the initial model. The mean is 0 and the standard deviation is The normal distribution, normal distribution Used to introduce randomness in the strength of variation.

[0059] Here, the gradient mean is an important concept in deep learning. It is typically used to describe the statistical properties of gradients during model training, representing the average level of gradients throughout the training process. Specifically, the gradient mean of a submodule refers to the average gradient of all parameters within that submodule in a single iteration. The learning rate of a submodule refers to the learning rate set for a specific submodule (or subnetwork) within the model during training. The learning rate is a hyperparameter that determines the step size for updating model parameters in each iteration.

[0060] Step 2: Calculate the mutation probability of the sub-module in the initial model based on the mutation intensity of the sub-module and the maximum mutation intensity of all sub-modules in the initial model.

[0061] In this embodiment, the maximum mutation intensity corresponding to all sub-modules in the initial model refers to the maximum value among all mutation intensities corresponding to all sub-modules. The mutation probability corresponding to a sub-module is used to characterize the likelihood of the sub-module mutating. If the mutation probability is too high, it may cause premature exit from a local optimum, thus failing to obtain the optimal target model; if the mutation probability is too low, it may cause getting trapped in a local optimum, failing to explore a better target model. Therefore, this embodiment dynamically calculates the mutation probability corresponding to the sub-module in the initial model based on the mutation intensity corresponding to the sub-module in the initial model and the maximum mutation intensity corresponding to all sub-modules in the initial model.

[0062] Specifically, by substituting the mutation intensity corresponding to the submodule in the initial model and the maximum mutation intensity corresponding to all submodules in the initial model into the following formula, the mutation probability corresponding to the submodule in the initial model is obtained: ; in, This represents the mutation probability corresponding to this submodule in the initial model. This is a preset upper limit for the mutation probability. This represents the lower bound of the mutation probability. This represents the mutation intensity corresponding to this submodule in the initial model. This represents the maximum mutation intensity corresponding to all submodules in the initial model.

[0063] Step 3: Based on the mutation intensity and mutation probability of each sub-module in the initial model, mutate the basic encoding of each sub-module in the aggregate encoding of the initial model.

[0064] In the embodiments of this application, mutations include the addition, deletion, and replacement of submodules. These three mutation operations are independent of each other and may occur simultaneously. (See also...) Figure 4 The diagram shown illustrates the variations of the initial model provided in this embodiment. In the initial model structure, unvariated blocks are labeled A, and variated blocks are labeled B. Therefore, Figure 4The initial model consists of nine blocks, A1 to A9. In mutation 1, a replacement occurred, specifically A1 was replaced with B1 and A9 was replaced with B9. In mutation 2, an addition occurred, with B8_9 added between A8 and A9. In mutation 3, both replacement and deletion occurred, with A2 and A7 being replaced with B2 and B7, and A8 being deleted.

[0065] Here, by using the mutation intensity and mutation probability of the gradient-guided submodule in this application, the initial model can be guided to mutate the parts that have a greater impact on the results first, thereby accelerating the exploration of the model structure and enriching the model system.

[0066] In addition, transforming the basic encoding in the aggregate encoding of any initial model to change the initial model further includes: crossing the basic encoding in the aggregate encoding of every two initial models to change the initial model.

[0067] In the embodiments of this application, reference is made to Figure 5 The diagram shows a crossover of the initial models provided in this embodiment. The first initial model undergoing the crossover consists of nine blocks, A1 to A9, and the second initial model consists of nine blocks, B1 to B9. In crossover 1, the two initial models swap their front and back portions; specifically, portions A1-A5 of the first initial model are swapped with portions B1-B5 of the second initial model. In crossover 2, the two initial models swap their middle portions; portions A3-A7 of the first initial model are swapped with portions B3-B7 of the second initial model.

[0068] Here, in each iteration, 15% to 25% of the initial total number of models can be randomly selected for crossover operations, and the crossover position is also determined randomly. No guidance is given here, and diversity is enhanced by completely randomizing the process.

[0069] S104. Take all target models in the added target model set as new initial models, jump to obtain the aggregate encoding of each initial model in each model cluster and the fitness of each initial model in each communication scenario, and continue execution until the preset iteration stop condition is met.

[0070] In the embodiments of this application, the preset iteration stopping condition can be reaching the maximum number of iterations, obtaining a target model that is greater than the fitness threshold, or the rate of increase of the maximum fitness being less than the preset increase rate threshold, etc.

[0071] Here, in S101 to S104, this application treats the initial model library as a population, each initial model in the library as an individual, and communication frequency prediction as an ecological environment, simulating the natural selection and genetic mechanism of Darwin's theory of biological evolution, iterating the model population repeatedly, and constructing the optimal target model set.

[0072] Furthermore, in the integration and optimization of deep learning models, data format mismatches between sub-modules are frequently encountered. For example, one sub-module outputs a size of 256x256, while the next sub-module requires a size of 128x128. This difference hinders direct combination between sub-modules. To overcome this challenge, this method introduces AI programming software, such as RooCode, to assist in executing the iterative process. Specifically, before S105 predicts communication frequency based on the fitness of each target model in the final target model set under various communication scenarios, the method also includes: for any target model in the target model set, generating transition blocks adapted to the data formats between adjacent sub-modules in that target model through large model encoding, such as upsampling or downsampling modules. This method can effectively eliminate the problem of data format mismatches between sub-modules. In this way, not only can seamless collaboration between different parts of the model be ensured, but the model access iteration process can also be accelerated, realizing a rapid model system integration and evolution closed loop, thereby improving the adaptability and flexibility of the model and accelerating the development and iteration process.

[0073] In summary, referring to Figure 6 The diagram shown is a flowchart of the model iteration provided in an embodiment of this application.

[0074] S105. Based on the final target models and the fitness of each target model in various communication scenarios, predict the communication frequency.

[0075] In this embodiment, the communication frequency of the communication scenario to be predicted at the current time is obtained; based on the fitness of each target model in the communication scenario to be predicted, a prediction model is selected from all target models; the communication frequency at the current time is input into the prediction model to obtain the communication frequency at the next time.

[0076] Furthermore, based on the fitness of each target model in the communication scenario to be predicted, a prediction model is selected from all target models, including: Step 1: Calculate the initial selection probability of each target model based on its fitness in the communication scenario to be predicted.

[0077] In this embodiment, the fitness of each target model in the communication scenario to be predicted is substituted into the following formula to obtain the initial selection probability of any target model: ; in, Let i be the initial selection probability of the target model with the preset index i. Let i be the fitness of the target model with a preset index i in the communication scenario to be predicted. The number of target models, The fitness of the target model with a preset index k in the communication scenario to be predicted.

[0078] Step 2: For each target model, sum the initial selection probabilities of all target models preceding it to obtain the target selection probability of the target model.

[0079] In this embodiment, the target models preceding the target model refer to models with preset numbers lower than the target model's preset number. Specifically, the target selection probability of the target model is obtained by summing the initial selection probabilities of all target models preceding the target model using the following formula: ; in, Let i be the probability of the target model with the preset index i being selected. The initial selection probability of the target model with the preset index k.

[0080] Step 3: If the probability of the target model being selected is greater than the probability of random selection, and the probability of random selection is greater than the probability of the target model being selected of the previous target model corresponding to this target model, then the target model is determined as the prediction model.

[0081] In this embodiment of the application, a random number r within the interval [0,1) is generated as the probability of random selection. If, If so, the target model with the preset index i is selected as the prediction model for prediction.

[0082] Here, a roulette wheel selection method is used when selecting models. The probability of each model being selected is generated based on the fitness ratio. High-fitness models occupy a larger area of ​​the roulette wheel and have a higher probability of being selected. Low-fitness models still have a small probability of being selected, maintaining diversity and not being limited to selecting the model with the highest fitness, which is more in line with the idea of ​​model ensemble evolution.

[0083] Reference Figure 7 The diagram shown is a flowchart of the communication frequency prediction process provided in this application embodiment. The entire process starts from the current time T. First, the signal data is processed using time-frequency analysis to extract the signal frequency. Then, the data is input into the process flow, and the "selector" module selects a prediction model from all target models in the model library. Subsequently, the data flows into the core integrated model library, from which the selected prediction model is called to perform prediction, and the predicted value of the communication frequency at time T+1 is output, thus completing a single prediction.

[0084] At time T+1, the output prediction results are compared with the actual data and input into the evaluator to generate evaluation results (such as accuracy, resource consumption, etc.). The generated evaluation results serve as feedback signals to drive a new round of iterative evolution, forming a "prediction-evaluation-optimization" closed loop, continuously enriching the model system in the model library to improve the ability to adapt to non-stationary environments.

[0085] In summary, this application integrates multiple models to construct a model library and employs a heuristic evolutionary learning strategy to enhance adaptability to non-stationary environments by enriching the model system. By integrating multiple models to compensate for the shortcomings of a single model and continuously iterating to enrich the diversity of the model system, the key core lies in the integration, construction, and iterative evolution mechanism of the model system within the model library.

[0086] Based on the same inventive concept, this application also provides a communication frequency prediction device corresponding to the communication frequency prediction method. Since the principle of the device in this application is similar to the communication frequency prediction method described above in this application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0087] Reference Figure 8 The diagram shown is a schematic of a communication frequency prediction device provided in an embodiment of this application. The device includes: The acquisition module 801 is used to acquire the aggregated encoding of each initial model in multiple model clusters and the fitness of each initial model in various communication scenarios; the model clusters are obtained by clustering all initial models based on the aggregated encoding of each initial model; the aggregated encoding is composed of the basic encoding of multiple sub-modules in the initial model; The filtering module 802 is used to filter out initial models that meet the preset fitness conditions from each model cluster based on the fitness, and form a target model set. The transformation module 803 is used to transform the basic encoding in the aggregate encoding of each initial model to change each initial model and add the changed initial model to the target model set. The jump module 804 is used to take all the target models in the added target model set as new initial models, jump to the step of obtaining the aggregate encoding of each initial model in each model cluster and the fitness of each initial model in each communication scenario, so as to continue execution until the preset iteration stop condition is met. The prediction module 805 is used to predict communication frequencies based on the final target model and the fitness of each target model in various communication scenarios.

[0088] The device provided in this application continuously enriches the diversity of the model system through iteration. The key core is the integration, construction, and iterative evolution mechanism of the model system in the model library, which can predict communication frequencies.

[0089] like Figure 9 As shown in the embodiment of this application, an electronic device 900 includes a processor 901, a memory 902, and a bus. The memory 902 stores machine-readable instructions executable by the processor 901. When the electronic device is running, the processor 901 communicates with the memory 902 via the bus, and the processor 901 executes the machine-readable instructions to perform the steps of the communication frequency prediction method described above.

[0090] Specifically, the memory 902 and processor 901 mentioned above can be general-purpose memory and processor, without any specific limitations. When the processor 901 runs the computer program stored in the memory 902, it can execute the above-mentioned communication frequency prediction method.

[0091] Corresponding to the above-described communication frequency prediction method, this application embodiment also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the above-described communication frequency prediction method.

[0092] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the method embodiments, and will not be repeated here. In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection can be through some communication interfaces; the indirect coupling or communication connection of devices or modules can be electrical, mechanical, or other forms.

[0093] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0094] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0095] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the information processing methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0096] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method of predicting a communication frequency, characterized by, The method comprises: obtaining the aggregate encoding of each initial model in each model cluster and the fitness of each initial model in each communication scenario; the model cluster is obtained by clustering all initial models based on the aggregate encoding of each initial model; the aggregate encoding is composed of the base encoding of multiple sub-modules in the initial model; selecting the initial model that meets the preset fitness condition from each model cluster based on the fitness to form a target model set; transforming the base encoding in the aggregate encoding of each initial model to modify each initial model, and adding the modified initial model to the target model set; taking all target models in the target model set after addition as new initial models, and jumping to the step of obtaining the aggregate encoding of each initial model in each model cluster and the fitness of each initial model in each communication scenario to continue execution until a preset iteration stop condition is met; performing communication frequency prediction based on the final target model and the fitness of each target model in each communication scenario.

2. The communication frequency prediction method according to claim 1, characterized by, The method comprises the following steps: calculating the similarity between each two initial models based on the aggregate encoding of each initial model; clustering all initial models based on the similarity to obtain multiple model clusters.

3. The communication frequency prediction method according to claim 1, characterized by, The method comprises the following steps: for each model cluster, determining the preset number of initial models with the maximum fitness in the model cluster as first intermediate models; determining the target model set based on the first intermediate models and the fitness of each initial model in each communication scenario.

4. The communication frequency prediction method according to claim 3, characterized by, The method comprises the following steps: selecting multiple initial models with the maximum fitness from all initial models according to a preset proportion to obtain second intermediate models; merging all first intermediate models and all second intermediate models to obtain the target model set.

5. The communication frequency prediction method according to claim 1, characterized by, The method comprises the following steps: for each sub-module in each initial model, calculating the mutation intensity corresponding to the sub-module in the initial model according to the gradient mean of the sub-module in the initial model and the learning rate; and calculating the mutation probability corresponding to the sub-module in the initial model according to the mutation intensity corresponding to the sub-module in the initial model and the maximum mutation intensity corresponding to all sub-modules in the initial model; based on the mutation intensity and the mutation probability corresponding to each sub-module in the initial model, mutating the base encoding of each sub-module in the aggregate encoding of the initial model to modify the initial model.

6. The communication frequency prediction method according to claim 1, characterized by, The method comprises the following steps: obtaining the communication frequency of the to-be-predicted communication scenario at the current time; selecting a prediction model from all the target models based on fitness of each target model under the communication scenario to be predicted; inputting the current communication frequency into the prediction model to obtain the communication frequency at the next time point.

7. The communication frequency prediction method according to claim 6, characterized by, The selecting a prediction model from all the target models based on fitness of each target model under the communication scenario to be predicted comprises: calculating an initial selection probability of each target model based on fitness of each target model under the communication scenario to be predicted; for each target model, accumulating the initial selection probabilities of all the target models before the target model to obtain a target selection probability of the target model; the target models before the target model refer to the models with preset serial numbers less than that of the target model; if the target selection probability of the target model is greater than a random selection probability, and the random selection probability is greater than a target selection probability of a last target model corresponding to the target model, the target model is determined as the prediction model.

8. A communication frequency prediction apparatus characterized by comprising: The apparatus comprises: an acquisition module configured to acquire an aggregated encoding of each initial model in a plurality of model clusters and fitness of each initial model under each communication scenario; the model clusters are obtained by clustering all the initial models based on the aggregated encoding of each initial model; the aggregated encoding is composed of base encodings of a plurality of sub-modules in the initial model; a screening module configured to screen out initial models satisfying a preset fitness condition from each model cluster based on the fitness to form a target model set; a transformation module configured to transform the base encodings in the aggregated encoding of each initial model to change each initial model and add the changed initial model to the target model set; a jump module configured to take all the target models in the added target model set as new initial models, jump to the acquisition of the aggregated encoding of each initial model in each model cluster and the fitness of each initial model under each communication scenario to continue execution until a preset iteration stop condition is satisfied; a prediction module configured to perform communication frequency prediction based on the final target model and the fitness of each target model under each communication scenario.

9. An electronic device, comprising: comprise: a processor, a storage medium and a bus, the storage medium stores machine readable instructions executable by the processor, when the electronic device is running, the processor and the storage medium communicate through the bus, the processor executes the machine readable instructions to perform the steps of the communication frequency prediction method according to any one of claims 1 to 7.

10. A computer readable storage medium characterized by, The computer readable storage medium stores a computer program, when the computer program is run by the processor, the steps of the communication frequency prediction method according to any one of claims 1 to 7 are executed.

Images