An ai-based communication base station current apportionment method and system

Abstract

The application provides an AI-based communication base station current allocation method and system, relates to the technical field of image data processing, and comprises the following steps: constructing an AI model, training the AI model according to historical electric meter image data and corresponding electric quantity data labels, and obtaining the AI model after training; collecting real-time electric meter image data corresponding to each operator in a communication base station, identifying the real-time electric meter image data by using the AI model after training, and obtaining real-time electric quantity data corresponding to each operator; and finally obtaining the current allocation proportion of each operator in the communication base station according to the real-time electric quantity data corresponding to each operator, so that the allocation of the current of the communication base station can be automatically completed, the current allocation efficiency can be improved, and errors caused by manual allocation can be avoided.

Description

Technical Field

[0001] This application relates to the field of image data processing technology, and more specifically, to an AI-based communication base station current sharing method and system. Background Technology

[0002] With the popularization of 5G technology and the explosive growth of communication services, the number of communication base stations and energy consumption have increased dramatically. To save construction costs and improve resource utilization efficiency, multiple operators sharing the same base station site, equipment room, power supply, and other infrastructure has become the mainstream model in the industry, known as the "co-construction and sharing" model. In co-construction and sharing base stations, a single main power system typically supplies power to all operators' communication equipment (such as baseband processing units and radio frequency remote units). Currently, this is usually achieved by manually taking photos of electricity meters on-site and manually reviewing each image, manually entering the current readings and base station information from the photos into an electronic ledger. Since the same operator may correspond to multiple photos, the current readings in these photos need to be manually summed and calculated to ultimately achieve current allocation. This allocation process is tedious and prone to errors. Summary of the Invention

[0003] This application aims to provide an AI-based method and system for current allocation in communication base stations, which addresses the problems of error-proneness and time-consuming and labor-intensive processes in the existing technology of manually collecting data and allocating current.

[0004] The first aspect of this application provides an AI-based method for current sharing in communication base stations, comprising:

[0005] Acquire pre-stored historical electricity meter image data or data input by staff through human-computer interaction, along with their corresponding electricity data tags;

[0006] An AI model is constructed and trained based on the historical electricity meter image data and its corresponding electricity data labels to obtain the trained AI model.

[0007] Real-time electricity meter image data corresponding to each operator in the communication base station is collected, and the trained AI model is used to identify the real-time electricity meter image data to obtain the real-time electricity data corresponding to each operator.

[0008] Based on the real-time power data of each operator, the current sharing ratio of each operator in the communication base station is obtained, and the current sharing of the communication base station based on AI is completed.

[0009] In one possible implementation, the AI ​​model is trained based on the historical electricity meter image data and its corresponding electricity data labels to obtain the trained AI model, including:

[0010] Based on the model parameters of the AI ​​model, multiple different individual parameters are obtained;

[0011] Using historical electricity meter image data as input and the electricity data labels corresponding to the historical electricity meter image data as the expected output, the fitness of each individual parameter is obtained.

[0012] The individual with the highest fitness is identified as the optimal individual.

[0013] Based on the optimal parameter individual, a spiral cooperative method is used to perform cooperative search on the parameter individual to determine the parameter individual after the cooperative search;

[0014] Based on the fitness of the individual parameters after the cooperative search, the global search probability and the local search probability are obtained by a mutually coupled dual-probability acquisition method determined by fitness.

[0015] Based on the local search probability, an adaptive local search is performed on the parameter individuals after the cooperative search around the optimal parameter individual to determine the parameter individuals after the adaptive local search.

[0016] Based on the global search probability, a global mutation search is performed on the parameter individuals after the cooperative search to determine the parameter individuals after the global mutation search.

[0017] Based on the parameter individuals after the adaptive local search and the parameter individuals after the global mutation search, determine the parameter individuals after the second search.

[0018] Determine whether the training termination condition is met. If so, redetermine the optimal parameter individual based on the parameter individuals after the second search to obtain the target parameter individual. Otherwise, return to the step of obtaining the fitness of the parameter individuals.

[0019] The model parameters in the target parameter individuals are used as the final parameters of the AI ​​model to obtain the trained AI model.

[0020] In one possible implementation, the acquisition of multiple different individual parameters based on the model parameters of the AI ​​model includes:

[0021] For any model parameter of the AI ​​model, it is randomly initialized between the upper limit and the lower limit of the model parameter, and the initialized model parameters are combined into a vector to obtain the individual parameter.

[0022] The model parameters are initialized multiple times and combined into vectors to obtain multiple different individual parameters.

[0023] In one possible implementation, the step of obtaining the fitness of each individual parameter by taking historical electricity meter image data as input and the electricity data label corresponding to the historical electricity meter image data as the expected output includes:

[0024] For any individual parameter, the model parameters in the individual parameter are applied to the AI ​​model, and historical electricity meter image data is used as input, and the electricity data label corresponding to the historical electricity meter image data is used as the expected output to obtain the root mean square loss function value corresponding to the individual parameter.

[0025] The fitness of an individual parameter is obtained by adding the root mean square loss function value to a preset constant and taking the reciprocal.

[0026] Iterate through all individual parameters and determine the fitness of each individual parameter.

[0027] In one possible implementation, the step of performing a cooperative search on the parameter individuals using a spiral cooperative method based on the optimal parameter individuals, and determining the parameter individuals after the cooperative search, includes:

[0028] Based on the fitness of the individual parameters, arrange the individual parameters in ascending order of fitness to obtain the arranged individual parameters.

[0029] Based on the sorted parameter individuals and the optimal parameter individual, a cooperative search is performed on the parameter individuals to determine the parameter individuals after the cooperative search:

[0030]

[0031]

[0032]

[0033]

[0034] in, Indicates the first t During the training process, the first n Individual parameters after permutation, Indicates the first n The number of parameter individuals after a collaborative search, n=1,2,…,N, where N represents the total number of parameter individuals. Indicates the optimal search factor. Indicates collaborative search factors. Represents the individual with optimal parameters. Indicates the first t During the training process, the first n +1 permuted parameter individuals, and in n for N hour, Set as a randomly selected individual parameter. Indicates intermediate parameters. Let π represent the mathematical constant pi, and cos represent the cosine function. Indicates the maximum number of training iterations. Represents the first random number between (0,1). Let represent the sine function, and e represent the natural constant.

[0035] In one possible implementation, the step of obtaining the global search probability and the local search probability using a fitness-determined, mutually coupled dual-probability acquisition method based on the fitness of the individual parameters after the cooperative search includes:

[0036] Based on the fitness of the individuals after the cooperative search, the fitness cooperation factor is determined as follows:

[0037]

[0038] in, Indicates fitness cooperation factor. Indicates the first m The fitness of an individual after a collaborative search. This represents the average fitness of the individual parameters after collaborative search;

[0039] Based on the fitness cooperation factor, the mutually coupled global search control factor and local search control factor are determined as follows:

[0040]

[0041]

[0042] in, This represents the local search control factor. This represents the global search control factor. This represents an exponential function with the natural constant e as the base, and L represents a constant scaling factor;

[0043] Based on the global search control factor, the global search probability is obtained as follows:

[0044]

[0045]

[0046]

[0047]

[0048] in, Indicates the first mThe global search probability of an individual after a collaborative search. Indicates the first m The global search convergence factor of individual parameters after a collaborative search. Indicates the first m The global search relevance parameters corresponding to the individual parameters after a collaborative search. Indicates the first m The local search relevance parameters corresponding to the individual parameters after a collaborative search. Indicates the first s During the first global search process m The change in fitness of an individual after a collaborative search; S represents the change in fitness of the first individual. m The parameters of each individual after the collaborative search are the total number of times the global search is performed based on the global search probability during the first t training iterations; max represents the maximum value function. Indicates the first m The fitness of an individual after the s-th global search, following the parameters of the collaborative search. Indicates the first m The fitness of an individual after a collaborative search and the (s-1)th global search.

[0049] Based on the local search control factor, the local search probability is obtained as follows:

[0050]

[0051]

[0052]

[0053]

[0054] in, Indicates the first m The local search probability of an individual after a collaborative search. Indicates the first m The local search convergence factor of individual parameters after a collaborative search. Indicates the first k During the first local search process m The change in fitness of an individual after a collaborative search; K Indicates the first m The parameters of the individual after the collaborative search are in front. t The total number of local searches performed based on the local search probability during each training process; Indicates the first m The parameter individuals after the collaborative search are in the 1st... k Fitness after each local search. Indicates the first m The parameter individuals after the collaborative search are in the 1st... k -1 fitness after performing a local search.

[0055] In one possible implementation, the step of performing an adaptive local search on parameter individuals after the cooperative search, based on the local search probability and focusing on the optimal parameter individual, to determine the parameter individuals after the adaptive local search includes:

[0056] Generate local search decision factors between (0,1);

[0057] When the local search probability is greater than or equal to the local search decision factor, an adaptive local search is performed on the parameter individuals after the cooperative search, and the parameter individuals after the adaptive local search are determined as follows:

[0058]

[0059]

[0060] in, Indicates the first m Individual parameters after an adaptive local search This represents the second random number between (0,1). This represents the adaptive local search factor. This represents the individual parameters following a random collaborative search. This represents the maximum value of the adaptive local search factor. Indicates the first t During the training process, the first m Individual parameters after a collaborative search. This represents the fitness of an individual with optimal parameters. Indicates the first m The fitness of an individual after a collaborative search. The fitness of the worst-performing individual is represented by , T represents the preset maximum number of training iterations, and lg represents the logarithmic function.

[0061] When the local search probability is less than the local search decision factor, the parameter individuals after the cooperative search are directly used as the parameter individuals after the adaptive local search.

[0062] In one possible implementation, the step of performing a global mutation search on the parameter individuals after the cooperative search based on the global search probability, and determining the parameter individuals after the global mutation search, includes:

[0063] Generate global search decision factors between (0,1);

[0064] When the global search probability is greater than or equal to the global search decision factor, a global mutation search is performed on the parameter individuals after the cooperative search to determine the parameter individuals after the global mutation search as follows:

[0065]

[0066]

[0067] in, Indicates the first m The parameter of the individual after the global mutation search. d 3D model parameters, d =1,2,…,D, where D represents the total dimension of the model parameters for each individual parameter. Indicates the first t During the training process, the first m The parameter of the individual after the first collaborative search d 3D model parameters, Indicates the global variable factor. Indicates the variation impact factor. Indicates the total number of mutations. i Indicates the number of mutations. The first individual with optimal parameters represents the first... d 3D model parameters, This represents the first value of the mean individual corresponding to all individual parameters. d The dimensional model parameters, its first d The dimensional model parameters are the parameters of all individuals after collaborative search in the 1st dimension. d The mean of the model parameters in the dimension;

[0068] When the global search probability is less than the global search decision factor, the parameter individuals after the cooperative search are directly used as the parameter individuals after the global mutation search.

[0069] In one possible implementation, determining whether the training termination condition is met includes: determining whether the current number of training iterations is greater than or equal to the preset maximum number of training iterations; if so, then the training termination condition is met; otherwise, the training termination condition is not met.

[0070] The second aspect of this application provides an AI-based communication base station current sharing system, including: a historical data acquisition module, an AI training module, a current data recognition module, and a current sharing module;

[0071] The historical data acquisition module is used to acquire pre-stored historical electricity meter image data or data input by staff through human-computer interaction, along with their corresponding electricity data tags.

[0072] The AI ​​training module is used to build an AI model and train the AI ​​model based on the historical electricity meter image data and its corresponding electricity data labels to obtain the trained AI model.

[0073] The current data recognition module is used to collect real-time electricity meter image data corresponding to each operator in the communication base station, and to use the trained AI model to recognize the real-time electricity meter image data to obtain the real-time electricity data corresponding to each operator.

[0074] The current sharing module is used to obtain the current sharing ratio of each operator in the communication base station based on the real-time power data of each operator, and complete the AI-based current sharing of the communication base station.

[0075] Beneficial effects:

[0076] This application provides an AI-based method and system for current allocation in communication base stations. It constructs an AI model and trains it using historical electricity meter image data and corresponding electricity data labels to obtain a trained AI model. Then, it collects real-time electricity meter image data corresponding to each operator in the communication base station and uses the trained AI model to identify the real-time electricity data for each operator. Finally, based on the real-time electricity data for each operator, it obtains the current allocation ratio for each operator in the communication base station. This method can automatically complete the current allocation of communication base stations, improving efficiency and avoiding errors caused by manual allocation. Attached Figure Description

[0077] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0078] Figure 1 This is a flowchart of an AI-based communication base station current sharing method proposed in one embodiment of this application;

[0079] Figure 2 This is a schematic diagram of the structure of an AI-based communication base station current sharing system proposed in one embodiment of this application;

[0080] Figure labeling: 201-Historical data acquisition module, 202-AI training module, 203-Current data recognition module, 204-Current allocation module. Detailed Implementation

[0081] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0082] like Figure 1 As shown in the figure, this application provides an AI-based communication base station current sharing method, including:

[0083] S101. Obtain historical meter image data and its corresponding electricity data tags that are pre-stored or input by staff through human-computer interaction.

[0084] To help those skilled in the art understand the communication base station current sharing method described in the embodiments of this application, a technical term is first introduced: communication base station current sharing refers to the technical means of accurately measuring and reasonably allocating the power consumption of each operator's equipment in a scenario where multiple operators share a base station. Its core objective is to achieve fair sharing of electricity costs and optimization of energy consumption management.

[0085] Historical meter image data and their corresponding electricity data labels usually indicate a learnable data relationship. This data relationship can be learned and recognized by AI (Artificial Intelligence) models, thereby enabling the acquisition of real-time electricity data.

[0086] S102. Construct an AI model and train the AI ​​model based on the historical electricity meter image data and its corresponding electricity data labels to obtain the trained AI model.

[0087] A convolutional neural network can be used to build an AI model, and then the AI ​​model can be trained based on the historical electricity meter image data and its corresponding electricity data labels, so that the trained AI model can have the ability to recognize electricity meter image data, thereby realizing the recognition of electricity data.

[0088] S103. Collect real-time electricity meter image data corresponding to each operator in the communication base station, and use the trained AI model to identify the real-time electricity meter image data to obtain the real-time electricity data corresponding to each operator.

[0089] The system collects real-time electricity meter image data from various operators in communication base stations, including real-time electricity meter image data uploaded by staff or collected by image data acquisition equipment at fixed locations. Regardless of whether it is real-time electricity meter image data or historical electricity meter image data, the electricity meter display area can be cropped and its size standardized before being input into the AI ​​model. The electricity meter display area obtained after cropping and size standardization is used as input, thereby improving the accuracy of data recognition.

[0090] S104. Based on the real-time power data of each operator, obtain the current sharing ratio of each operator in the communication base station, and complete the AI-based communication base station current sharing.

[0091] After determining the real-time power consumption data for each operator, the power consumption between the two measurements can be determined based on the historical power consumption data from the previous measurement. Then, the power consumption ratio for each operator can be determined, thus obtaining the current allocation ratio for each operator in the communication base station. If an operator has multiple meters, the power consumption data from all meters can be added together before determining the power consumption ratio, thereby improving current allocation efficiency and avoiding errors caused by manual operation.

[0092] In one possible implementation, the AI ​​model is trained based on the historical electricity meter image data and its corresponding electricity data labels to obtain the trained AI model, including:

[0093] Based on the model parameters of the AI ​​model, multiple different individual parameters are obtained;

[0094] Using historical electricity meter image data as input and the electricity data labels corresponding to the historical electricity meter image data as the expected output, the fitness of each individual parameter is obtained.

[0095] The individual with the highest fitness is identified as the optimal individual.

[0096] Based on the optimal parameter individual, a spiral cooperative method is used to perform cooperative search on the parameter individual to determine the parameter individual after the cooperative search;

[0097] Based on the fitness of the individual parameters after the cooperative search, the global search probability and the local search probability are obtained by a mutually coupled dual-probability acquisition method determined by fitness.

[0098] Based on the local search probability, an adaptive local search is performed on the parameter individuals after the cooperative search around the optimal parameter individual to determine the parameter individuals after the adaptive local search.

[0099] Based on the global search probability, a global mutation search is performed on the parameter individuals after the cooperative search to determine the parameter individuals after the global mutation search.

[0100] Based on the parameter individuals after the adaptive local search and the parameter individuals after the global mutation search, the parameter individuals after the second search are determined; that is, the parameter individuals with higher fitness among the parameter individuals after the adaptive local search and the parameter individuals after the global mutation search are taken as the parameter individuals after the second search.

[0101] Determine whether the training termination condition is met. If so, redetermine the optimal parameter individual based on the parameter individuals after the second search to obtain the target parameter individual. Otherwise, return to the step of obtaining the fitness of the parameter individuals.

[0102] The model parameters in the target parameter individuals are used as the final parameters of the AI ​​model to obtain the trained AI model.

[0103] Optionally, after each individual parameter changes, out-of-bounds handling of the individual parameter and maintenance of the optimal individual parameter for that training process can be performed to ensure the effectiveness of the training.

[0104] In existing technologies, conventional particle swarm optimization or genetic algorithms are typically used to train AI models (such as training the connection weights between neural network layers in the AI ​​model). These methods often get stuck in local optima during training, resulting in the trained AI model being unable to accurately identify electricity data. Therefore, this application's embodiments employ the aforementioned training method to improve the global search capability and convergence ability of the AI ​​model during training, thereby improving the accuracy of the AI ​​model's data identification.

[0105] In one possible implementation, the acquisition of multiple different individual parameters based on the model parameters of the AI ​​model includes:

[0106] For any model parameter of the AI ​​model, it is randomly initialized between the upper limit and the lower limit of the model parameter, and the initialized model parameters are combined into a vector to obtain the individual parameter.

[0107] The model parameters are initialized multiple times and combined into vectors to obtain multiple different individual parameters.

[0108] In one possible implementation, the step of obtaining the fitness of each individual parameter by taking historical electricity meter image data as input and the electricity data label corresponding to the historical electricity meter image data as the expected output includes:

[0109] For any individual parameter, the model parameters in the individual parameter are applied to the AI ​​model, and historical electricity meter image data is used as input, and the electricity data label corresponding to the historical electricity meter image data is used as the expected output to obtain the root mean square loss function value corresponding to the individual parameter.

[0110] The fitness of an individual parameter is obtained by adding the root mean square loss function value to a preset constant and taking the reciprocal.

[0111] Iterate through all individual parameters and determine the fitness of each individual parameter.

[0112] In one possible implementation, the step of performing a cooperative search on the parameter individuals using a spiral cooperative method based on the optimal parameter individuals, and determining the parameter individuals after the cooperative search, includes:

[0113] Based on the fitness of the individual parameters, arrange the individual parameters in ascending order of fitness to obtain the arranged individual parameters.

[0114] Based on the sorted parameter individuals and the optimal parameter individual, a cooperative search is performed on the parameter individuals to determine the parameter individuals after the cooperative search:

[0115]

[0116]

[0117]

[0118]

[0119] in, Indicates the first t During the training process, the first n Individual parameters after permutation, Indicates the first n The number of parameter individuals after a collaborative search, n=1,2,…,N, where N represents the total number of parameter individuals. Indicates the optimal search factor. Indicates collaborative search factors. Represents the individual with optimal parameters. Indicates the first t During the training process, the first n +1 permuted parameter individuals, and in n for N hour, Set as a randomly selected individual parameter. Indicates intermediate parameters. Let π represent the mathematical constant pi, and cos represent the cosine function. Indicates the maximum number of training iterations. Represents the first random number between (0,1). Let represent the sine function, and e represent the natural constant.

[0120] Employing a spiral cooperative approach to perform cooperative searches on the individual parameters allows the algorithm to prioritize cooperative searches in the early stages, increasing the solution space exploration rate and thus improving its global search capability. In the later stages, the algorithm focuses more on learning information from the optimal individual parameters, improving the precision of its convergence. As the algorithm progresses, the optimal search factor fluctuates, thereby enhancing the algorithm's ability to escape local optima in the mid-stages.

[0121] In one possible implementation, the step of obtaining the global search probability and the local search probability using a fitness-determined, mutually coupled dual-probability acquisition method based on the fitness of the individual parameters after the cooperative search includes:

[0122] Based on the fitness of the individuals after the cooperative search, the fitness cooperation factor is determined as follows:

[0123]

[0124] in, Indicates fitness cooperation factor. Indicates the first m The fitness of an individual after a collaborative search. This represents the average fitness of the individual parameters after collaborative search;

[0125] Based on the fitness cooperation factor, the mutually coupled global search control factor and local search control factor are determined as follows:

[0126]

[0127]

[0128] in, This represents the local search control factor. Indicates the global search control factor. This represents an exponential function with the natural constant e as the base, and L represents a constant scaling factor, which can be set to 60;

[0129] Based on the global search control factor, the global search probability is obtained as follows:

[0130]

[0131]

[0132]

[0133]

[0134] in, Indicates the first m The global search probability of an individual after a collaborative search. Indicates the first m The global search convergence factor of individual parameters after a collaborative search. Indicates the first m The global search relevance parameters corresponding to the individual parameters after a collaborative search. Indicates the first m The local search relevance parameters corresponding to the individual parameters after a collaborative search. Indicates the first s During the first global search process m The change in fitness of an individual after a collaborative search; S represents the change in fitness of the first individual. m The parameters of each individual after the collaborative search are the total number of times the global search is performed based on the global search probability during the first t training iterations; max represents the maximum value function. Indicates the first m The fitness of an individual after the s-th global search, following the parameters of the collaborative search. Indicates the first m The fitness of an individual after a collaborative search and the (s-1)th global search.

[0135] Based on the local search control factor, the local search probability is obtained as follows:

[0136]

[0137]

[0138]

[0139]

[0140] in, Indicates the first m The local search probability of an individual after a collaborative search. Indicates the first m The local search convergence factor of individual parameters after a collaborative search. Indicates the first k During the first local search process m The change in fitness of an individual after a collaborative search; K Indicates the first m The parameters of the individual after the collaborative search are in front. tThe total number of local searches performed based on the local search probability during each training process; Indicates the first m The parameter individuals after the collaborative search are in the 1st... k Fitness after each local search. Indicates the first m The parameter individuals after the collaborative search are in the 1st... k -1 fitness after performing a local search.

[0141] By employing a mutually coupled dual-probability acquisition method based on fitness decisions to obtain global and local search probabilities, two mutually influential probabilities can be obtained, thereby enabling the adaptive adjustment of the algorithm's global and local search capabilities. This allows the algorithm to possess greater global search capability in the early and mid-stages, while gradually using the parameter vector obtained after local development as the parameter vector after balancing the search in the later stages indicates that the algorithm gradually shifts to local search but retains a certain degree of global search capability.

[0142] In one possible implementation, the step of performing an adaptive local search on parameter individuals after the cooperative search, based on the local search probability and focusing on the optimal parameter individual, to determine the parameter individuals after the adaptive local search includes:

[0143] Generate local search decision factors between (0,1);

[0144] When the local search probability is greater than or equal to the local search decision factor, an adaptive local search is performed on the parameter individuals after the cooperative search, and the parameter individuals after the adaptive local search are determined as follows:

[0145]

[0146]

[0147] in, Indicates the first m Individual parameters after an adaptive local search This represents the second random number between (0,1). This represents the adaptive local search factor. This represents the individual parameters following a random collaborative search. This represents the maximum value of the adaptive local search factor, which can be set to 0.85; Indicates the first t During the training process, the first m Individual parameters after a collaborative search. This represents the fitness of an individual with optimal parameters. Indicates the first m The fitness of an individual after a collaborative search. The fitness of the worst-performing individual is represented by , T represents the preset maximum number of training iterations, and lg represents the logarithmic function.

[0148] When the local search probability is less than the local search decision factor, the parameter individuals after the cooperative search are directly used as the parameter individuals after the adaptive local search.

[0149] Adaptive local search on individual parameters after collaborative search can effectively improve the local convergence ability of the algorithm. Furthermore, adaptive search based on fitness during the local search process can accelerate the training speed in the early and middle stages of the algorithm and improve the fine search capability in the later stages.

[0150] In one possible implementation, the step of performing a global mutation search on the parameter individuals after the cooperative search based on the global search probability, and determining the parameter individuals after the global mutation search, includes:

[0151] Generate global search decision factors between (0,1);

[0152] When the global search probability is greater than or equal to the global search decision factor, a global mutation search is performed on the parameter individuals after the cooperative search to determine the parameter individuals after the global mutation search as follows:

[0153]

[0154]

[0155] in, Indicates the first m The parameter of the individual after the global mutation search. d 3D model parameters, d =1,2,…,D, where D represents the total dimension of the model parameters for each individual parameter. Indicates the first t During the training process, the first m The parameter of the individual after the first collaborative search d 3D model parameters, Represents the global variable factor. This represents the variation impact factor, which can be set to 0.4; This represents the total number of mutations and can be set to 4. i Indicates the number of mutations. The first individual with optimal parameters represents the... d 3D model parameters, This represents the first value of the mean individual corresponding to all individual parameters. d The dimensional model parameters, its first d The dimensional model parameters are the parameters of all individuals after collaborative search in the 1st dimension. dThe mean of the model parameters in dimension.

[0156] When the global search probability is less than the global search decision factor, the parameter individuals after the cooperative search are directly used as the parameter individuals after the global mutation search.

[0157] Performing a global mutation search on the individual parameters after the collaborative search provides the algorithm with powerful global search capabilities. Furthermore, a greedy strategy can be employed during the global mutation search process to control the algorithm's training speed.

[0158] In one possible implementation, determining whether the training termination condition is met includes: determining whether the current number of training iterations is greater than or equal to the preset maximum number of training iterations; if so, then the training termination condition is met; otherwise, the training termination condition is not met.

[0159] This application provides an AI-based method for current allocation in communication base stations. It involves constructing an AI model and training it using historical electricity meter image data and corresponding electricity data labels to obtain a trained AI model. Then, it collects real-time electricity meter image data corresponding to each operator within the communication base station and uses the trained AI model to identify the real-time electricity data for each operator. Finally, based on the real-time electricity data for each operator, it obtains the current allocation ratio for each operator within the communication base station. This method can automatically complete the current allocation of communication base stations, improving efficiency and avoiding errors caused by manual allocation.

[0160] like Figure 2 As shown, this application provides an AI-based communication base station current sharing system, including: a historical data acquisition module 201, an AI training module 202, a current data recognition module 203, and a current sharing module 204.

[0161] The historical data acquisition module 201 is used to acquire pre-stored historical electricity meter image data or input by staff through human-computer interaction and their corresponding electricity data tags;

[0162] AI training module 202 is used to build an AI model and train the AI ​​model based on the historical electricity meter image data and its corresponding electricity data labels to obtain the trained AI model.

[0163] The current data recognition module 203 is used to collect real-time electricity meter image data corresponding to each operator in the communication base station, and use the trained AI model to recognize the real-time electricity meter image data to obtain the real-time electricity data corresponding to each operator.

[0164] The current sharing module 204 is used to obtain the current sharing ratio of each operator in the communication base station based on the real-time power data of each operator, and complete the AI-based current sharing of the communication base station.

[0165] The AI-based communication base station current sharing system provided in this application embodiment can execute the above-mentioned method and technical solution. Its principle and beneficial effects are similar, and will not be repeated here.

[0166] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0167] This application describes embodiments with reference to flowchart illustrations and / or block diagrams of methods, apparatuses, electronic devices, and computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0168] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0169] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0170] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0171] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0172] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. An AI-based communication base station current apportioning method, characterized by, The method comprises the following steps: acquiring historical electric meter image data and corresponding electric quantity data labels pre-stored or input by a worker through human-computer interaction; constructing an AI model and training the AI model according to the historical electric meter image data and the corresponding electric quantity data labels to obtain the AI model after training, comprising the following steps: acquiring a plurality of different parameter individuals based on model parameters of the AI model; taking the historical electric meter image data as input and the corresponding electric quantity data labels as expected output to acquire the fitness of each parameter individual; determining the parameter individual with the maximum fitness as the optimal parameter individual; performing collaborative search on the parameter individual after the collaborative search according to the optimal parameter individual by using a spiral collaboration method; acquiring global search probability and local search probability by using a mutual coupling double-probability acquisition method based on the fitness of the parameter individual after the collaborative search; performing adaptive local search on the parameter individual after the collaborative search based on the local search probability to determine the parameter individual after the adaptive local search; performing global mutation search on the parameter individual after the collaborative search based on the global search probability to determine the parameter individual after the global mutation search; determining the parameter individual after the secondary search according to the parameter individual after the adaptive local search and the parameter individual after the global mutation search; determining the optimal parameter individual again according to the parameter individual after the secondary search to obtain a target parameter individual if the training end condition is met, otherwise returning to the step of acquiring the fitness of the parameter individual; taking the model parameters in the target parameter individual as the final parameters of the AI model to obtain the AI model after training; acquiring real-time electric meter image data corresponding to each operator in a communication base station and identifying the real-time electric meter image data by using the AI model after training to obtain real-time electric quantity data corresponding to each operator; acquiring current allocation ratios of each operator in the communication base station based on the real-time electric quantity data corresponding to each operator to complete AI-based current allocation of the communication base station. 2.The AI-based communication base station current apportioning method according to claim 1, wherein, The step of acquiring a plurality of different parameter individuals based on model parameters of the AI model comprises the following steps: randomly initializing the model parameters between the upper limit and the lower limit of the model parameters for any one model parameter of the AI model, and composing a vector with the initialized model parameters to obtain a parameter individual; composing a vector with the initialized model parameters to obtain a plurality of different parameter individuals. 3.The AI-based communication base station current apportioning method according to claim 1, wherein, The step of taking the historical electric meter image data as input and the corresponding electric quantity data labels as expected output to acquire the fitness of each parameter individual comprises the following steps: applying the model parameters in the parameter individual to the AI model, taking the historical electric meter image data as input and the corresponding electric quantity data labels as expected output to acquire the root mean square loss function value corresponding to the parameter individual; adding the root mean square loss function value to a preset constant and taking the reciprocal to obtain the fitness of the parameter individual. Traverse all parameter individuals, determine the fitness of each parameter individual. 4.The AI-based communication base station current apportioning method according to claim 1, wherein, According to the optimal parameter individual, the spiral cooperation method is used to cooperatively search the parameter individuals, and parameter individuals after cooperative search are determined. According to the fitness of the parameter individuals, the parameter individuals are arranged in order of fitness from small to large, and parameter individuals after arrangement are obtained. Based on the parameter individuals after arrangement and the optimal parameter individual, the parameter individuals are cooperatively searched, and parameter individuals after cooperative search are determined. wherein, denotes the parameter individual after the t th permutation in the n th training process, denotes the parameter individual after the n th cooperative search, n = 1, 2, …, N, N denotes the total number of parameter individuals, denotes the optimal direction search factor, denotes the cooperative search factor, denotes the optimal parameter individual, denotes the parameter individual after the t th permutation in the n th training process, and in n is N , is set as a randomly selected parameter individual, denotes the intermediate parameter, denotes the constant pi, and cos denotes the cosine function, denotes the maximum number of training, denotes the first random number between (0, 1), denotes the sine function, and e denotes the natural constant. 5.The AI-based communication base station current apportioning method according to claim 4, wherein, According to the fitness of the parameter individuals after cooperative search, the mutual coupling double probability acquisition method is used to acquire global search probability and local search probability according to the fitness decision, including: According to the fitness of the parameter individuals after cooperative search, the fitness cooperation factor is determined. wherein, denotes a fitness cooperation factor, denotes the fitness of the parameter individual after the m cooperative search, denotes the average fitness of the parameter individuals after the cooperative search; Based on the fitness cooperation factor, the mutual coupling global search control factor and the local search control factor are determined. wherein denotes a local search control factor, denotes a global search control factor, denotes an exponential function with the natural constant e as base, L denotes a constant scaling factor; According to the global search control factor, the global search probability is obtained. in, Indicates the first m The global search probability of an individual after a collaborative search. Indicates the first m The global search convergence factor of individual parameters after a collaborative search. Indicates the first m The global search relevance parameters corresponding to the individual parameters after a collaborative search. Indicates the first m The local search relevance parameters corresponding to the individual parameters after a collaborative search. Indicates the first s During the first global search process m The change in fitness of an individual after a collaborative search; S represents the change in fitness of the first individual. m The parameters of each individual after the collaborative search are the total number of times the global search is performed based on the global search probability during the first t training iterations; max represents the maximum value function. Indicates the first m The fitness of an individual after the s-th global search, following the parameters of the collaborative search. Indicates the first m The fitness of an individual after a collaborative search and the (s-1)th global search. According to the local search control factor, the local search probability is obtained. wherein, denotes the local search probability of the parameter individual after the m th cooperative search, denotes the local search convergence factor of the parameter individual after the m th cooperative search, denotes the fitness change amount of the parameter individual after the k th cooperative search in the m th local search process; K denotes the total number of times of performing local search according to the local search probability of the parameter individual after the m th cooperative search in the previous t th training process; denotes the fitness of the parameter individual after the m th cooperative search in the k th local search process, denotes the fitness of the parameter individual after the m th cooperative search in the k -1th local search process. 6.The AI-based communication base station current apportioning method according to claim 5, wherein, The adaptive local search of the parameter individuals after cooperative search is carried out around the optimal parameter individual based on the local search probability, and the parameter individuals after adaptive local search are determined, including: A local search decision factor between (0, 1) is generated. When the local search probability is greater than or equal to the local search decision factor, the adaptive local search of the parameter individuals after cooperative search is carried out, and the parameter individuals after adaptive local search are determined. wherein, denotes the parameter individual after the m th adaptive local search, denotes a second random number between (0, 1), denotes an adaptive local search factor, denotes the parameter individual after the random cooperative search, denotes a maximum value of the adaptive local search factor, denotes the parameter individual after the t th cooperative search in the m th training process, denotes the fitness of the optimal parameter individual, denotes the fitness of the parameter individual after the m th cooperative search, denotes the fitness of the worst parameter individual, T denotes a preset maximum training number, and lg denotes a logarithmic function. When the local search probability is less than the local search decision factor, the parameter individuals after cooperative search are directly used as the parameter individuals after adaptive local search. 7.The AI-based communication base station current allocation method according to claim 6, wherein, The global mutation search of the parameter individuals after cooperative search is carried out based on the global search probability, and the parameter individuals after global mutation search are determined, including: A global search decision factor between (0, 1) is generated. When the global search probability is greater than or equal to the global search decision factor, the global mutation search of the parameter individuals after cooperative search is carried out, and the parameter individuals after global mutation search are determined. in, Indicates the first m The parameter of the individual after the global mutation search. d 3D model parameters, d =1,2,…,D, where D represents the total dimension of the model parameters for each individual parameter. Indicates the first t During the training process, the first m The parameter of the individual after the first collaborative search d 3D model parameters, Indicates the global variable factor. Indicates the variation impact factor. Indicates the total number of mutations. i Indicates the number of mutations. The first individual with optimal parameters represents the first... d 3D model parameters, This represents the first value of the mean individual corresponding to all individual parameters. d The dimensional model parameters, its first d The dimensional model parameters are the parameters of all individuals after collaborative search in the 1st dimension. d The mean of the model parameters in the dimension; When the global search probability is less than the global search decision factor, the parameter individuals after cooperative search are directly used as the parameter individuals after global mutation search. 8.The AI-based communication base station current apportioning method according to claim 1, wherein, The judgment whether the training end condition is met includes: judging whether the current training frequency is greater than or equal to the preset maximum training frequency, if yes, determining that the training end condition is met, otherwise determining that the training end condition is not met.

9. An AI-based communication base station current apportioning system capable of performing the AI-based communication base station current apportioning method of any one of claims 1-8, characterized by Including: The historical data acquisition module, the AI training module, the current data identification module and the current allocation module; The historical data acquisition module is used for acquiring the historical electric meter image data and the corresponding electric quantity data label pre-stored or input by the staff through human-computer interaction; The AI training module is used for constructing an AI model, and training the AI model according to the historical electric meter image data and the corresponding electric quantity data label, and acquiring the AI model after training, including: Based on the model parameters of the AI model, a plurality of different parameter individuals are obtained; The historical electric meter image data is taken as input, and the electric quantity data corresponding to the historical electric meter image data is taken as expected output to obtain the fitness of each parameter individual; The parameter individual with the maximum fitness is determined as the optimal parameter individual; According to the optimal parameter individual, a spiral cooperation method is used to perform cooperative search on the parameter individual to determine the parameter individual after cooperative search; According to the fitness of the parameter individual after cooperative search, a mutual coupling double probability acquisition method determined by fitness is used to acquire a global search probability and a local search probability; Based on the local search probability, adaptive local search is performed on the parameter individual after cooperative search around the optimal parameter individual to determine the parameter individual after adaptive local search; Based on the global search probability, global mutation search is performed on the parameter individual after cooperative search to determine the parameter individual after global mutation search; According to the parameter individual after adaptive local search and the parameter individual after global mutation search, a parameter individual after secondary search is determined; It is judged whether the training end condition is met, if yes, the optimal parameter individual is re-determined according to the parameter individual after secondary search to obtain a target parameter individual, otherwise, the step of acquiring the fitness of the parameter individual is returned; The model parameters in the target parameter individual are taken as the final parameters of the AI model to obtain the AI model after training; The current data recognition module is configured to collect real-time electric meter image data corresponding to each operator in the communication base station, and recognize the real-time electric meter image data by using the AI model after training to obtain real-time electric quantity data corresponding to each operator; The current allocation module is configured to acquire current allocation proportions of each operator in the communication base station according to the real-time electric quantity data corresponding to each operator, and complete AI-based current allocation of the communication base station.

Images