Power regulation method, device and equipment of indoor distribution system and medium

Abstract

The invention relates to a power regulation method, device and equipment of an indoor distribution system and a medium. The power regulation method of the indoor distribution system comprises the following steps: collecting real-time data of an indoor area; the real-time data are preprocessed; constructing a deep learning model; optimizing the deep learning model by using an energy valley optimization algorithm; inputting the preprocessed real-time data into the optimized deep learning model, and outputting a power regulation scheme by the optimized deep learning model based on the preprocessed real-time data; and performing power regulation on the indoor distribution system based on the power regulation scheme. By introducing the deep learning model, the real-time environment data, the real-time use data of the indoor distribution system and the real-time user behavior data can be analyzed and processed in real time, so that an accurate power regulation scheme is generated. Meanwhile, the deep learning model is optimized by utilizing an energy valley optimization algorithm, so that the prediction precision and the response speed of the deep learning model can be further improved.

Description

Technical Field

[0001] The present invention relates to the field of communications, and in particular to a power regulation method, device, equipment and medium for a room distribution system. Background Art

[0002] Indoor distributed systems (indoor distribution systems) play an important role in modern communication networks. Their main function is to enhance indoor signal coverage. With the popularity of mobile devices and the increasing demand for network quality from users, power regulation of indoor distributed systems has become particularly important. However, most existing power regulation methods for indoor distributed systems rely on fixed parameters or simple rules, which cannot be adjusted accurately and in real time according to changes in the environment and user needs, making it difficult to meet user needs. Summary of the invention

[0003] Based on this, it is necessary to provide a power regulation method, device, equipment and medium for a room distributed system to address the above technical problems.

[0004] In a first aspect, the present invention provides a power regulation method for a distributed indoor system, the power regulation method for the distributed indoor system comprising:

[0005] Collecting real-time data of indoor areas, including real-time environmental data of indoor areas, real-time usage data of indoor distribution systems, and real-time user behavior data of indoor areas;

[0006] Preprocessing the real-time data;

[0007] Build deep learning models;

[0008] Use the energy valley optimization algorithm to optimize the deep learning model;

[0009] The preprocessed real-time data is input into the optimized deep learning model, and the optimized deep learning model outputs a power regulation scheme based on the preprocessed real-time data;

[0010] The power of the indoor distributed system is regulated based on the power regulation scheme.

[0011] Optionally, the preprocessing of the real-time data includes:

[0012] Performing denoising processing on the real-time data;

[0013] Normalize the denoised data;

[0014] The normalized data is used for feature extraction.

[0015] Optionally, constructing a deep learning model includes:

[0016] Build an initial deep learning model;

[0017] Acquire historical data of the indoor area, the historical data including: historical environmental data of the indoor area, historical usage data of the indoor distribution system, and historical user behavior data of the indoor area;

[0018] Training the initial deep learning model based on the historical data;

[0019] The trained initial deep learning model is evaluated, and the trained initial deep learning module is optimized based on the evaluation result to obtain the deep learning model.

[0020] Optionally, the optimizing the deep learning model using an energy valley optimization algorithm includes:

[0021] Randomly generate a parameter set of the deep learning model, the parameter set including a plurality of parameters, each of the parameters being a weight or a bias of the deep learning model;

[0022] Obtaining energy of the deep learning model based on the parameter set;

[0023] Performing iterative optimization based on the parameter set and the energy to obtain an optimal parameter set;

[0024] The optimal parameter set is updated to the deep learning model to optimize the deep learning model.

[0025] Optionally, obtaining the energy of the deep learning model based on the parameter set includes: obtaining the energy of the deep learning model based on the following loss function:

[0026]

[0027] Among them, y i is the true value of the i-th record, is the predicted value output by the deep learning model, N is the number of samples in the deep learning model, and i is an integer greater than or equal to 1 and less than or equal to N.

[0028] Optionally, the iterative optimization based on the parameter set and the energy to obtain an optimal parameter set includes:

[0029] Randomly generate a new parameter set in a neighborhood of the parameter set so that at least one parameter in the parameter set increases by a random value;

[0030] Obtaining new energy for the deep learning model based on the new parameter set;

[0031] If the new energy is less than the current energy, the new parameter set is updated to the current parameter set; if the new energy is greater than or equal to the current energy, the new parameter set is updated to the current parameter set based on the probability obtained by the following probability formula; the probability formula is: Wherein, P is the probability, T is the current temperature of the indoor area, E(θ) is the current energy, and E(θ') is the new energy;

[0032] Repeat the above steps until the iteration ends or the energy converges, and take the current parameter set at this time as the optimal parameter set.

[0033] Optionally, after the power of the indoor distributed system is adjusted based on the power adjustment scheme, the method further includes: continuously repeating the following steps:

[0034] Collecting the current data of the indoor area in real time, the current data includes the current environment data of the indoor area, the current usage data of the indoor distribution system and the current user behavior data of the indoor area;

[0035] Preprocessing the current data;

[0036] Use the energy valley optimization algorithm to optimize the current deep learning model;

[0037] Inputting the preprocessed current data into the optimized current deep learning model, and the optimized current deep learning model outputs a new power regulation scheme based on the preprocessed current data;

[0038] The power of the indoor distributed system is regulated based on a new power regulation scheme.

[0039] In a second aspect, the present invention further provides a power regulation device for a room-divided system, the power regulation device for the room-divided system comprising:

[0040] A collection module is used to collect real-time data of the indoor area, wherein the real-time data includes real-time environmental data of the indoor area, real-time usage data of the indoor distribution system and real-time user behavior data of the indoor area;

[0041] A preprocessing module, used for preprocessing the real-time data;

[0042] Model building module, used to build deep learning models;

[0043] A model optimization module is used to optimize the deep learning model using an energy valley optimization algorithm to obtain an optimized deep learning model. The optimized power regulation module outputs a power regulation solution based on the preprocessed real-time data;

[0044] A power regulation module is used to regulate the power of the indoor distributed system based on the power regulation scheme.

[0045] In a third aspect, the present invention further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of any one of the methods in the first aspect when executing the computer program.

[0046] In a fourth aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of any one of the methods in the first aspect when executed by a processor.

[0047] The power regulation method, device, equipment and medium of the indoor distributed system of the present invention can analyze and process real-time environmental data, real-time usage data of the indoor distributed system and real-time user behavior data in real time by introducing a deep learning model, thereby generating an accurate power regulation scheme. At the same time, the deep learning model is optimized using an energy valley optimization algorithm to further improve the prediction accuracy and response speed of the deep learning model. The present invention can not only meet the needs of different environments and users, but also effectively reduce system energy consumption, improve overall energy efficiency, and enhance user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A schematic flow chart of a power regulation method for a room distribution system provided in one embodiment of the present invention;

[0049] Figure 2 A structural block diagram of a power regulating device for a room distribution system is provided in another embodiment of the present invention;

[0050] Figure 3 This is a diagram of the internal structure of a computer device provided in yet another embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0052] In one embodiment, see Figure 1 The present invention provides a power regulation method for a distributed indoor system. The power regulation method for a distributed indoor system may include the following steps: S10 to S60.

[0053] S10: Collecting real-time data of the indoor area, wherein the real-time data includes real-time environmental data of the indoor area, real-time usage data of the indoor distribution system, and real-time user behavior data of the indoor area.

[0054] S20: Preprocessing the real-time data.

[0055] S30: Build a deep learning model.

[0056] S40: Optimize deep learning models using the energy valley optimization algorithm.

[0057] S50: Input the preprocessed real-time data into the optimized deep learning model, and the optimized deep learning model outputs a power adjustment scheme based on the preprocessed real-time data.

[0058] S60: Perform power regulation on the indoor distributed system based on the power regulation scheme.

[0059] In the power regulation method of the indoor distributed system of the present invention, by introducing a deep learning model, it is possible to analyze and process real-time environmental data, real-time usage data of the indoor distributed system, and real-time user behavior data in real time, thereby generating an accurate power regulation scheme. At the same time, the deep learning model is optimized using an energy valley optimization algorithm to further improve the prediction accuracy and response speed of the deep learning model. The present invention can not only meet the needs of different environments and users, but also effectively reduce system energy consumption, improve overall energy efficiency, and enhance user experience.

[0060] In step S10, refer to Figure 1 In step S10, real-time data of the indoor area is collected, and the real-time data includes real-time environmental data of the indoor area, real-time usage data of the indoor distribution system, and real-time user behavior data of the indoor area.

[0061] Specifically, the real-time data of the indoor area includes the real-time data of the indoor area where the indoor distribution system is located.

[0062] As an example, real-time environmental data can be collected in real time by sensors installed in indoor areas (e.g., temperature sensors, humidity sensors, etc.); these sensors can provide detailed information about the environment of the indoor area, such as temperature changes and humidity levels, etc.

[0063] As an example, the operating status of each device in the distributed indoor system can be detected and the usage data of each device can be collected to obtain the real-time usage data of the distributed indoor system; the real-time usage data of the distributed indoor system can specifically include information such as the real-time power consumption, real-time strength and real-time number of connected users of each device in the distributed indoor system. Specifically, the distributed indoor system can include base stations, distributed antennas, power regulation modules and control centers, etc.; the power output and signal quality of each base station and distributed antenna can be recorded in real time through the power monitoring module; for example, the power monitoring module can, but is not limited to, record the power consumption in sequence every minute, and send the monitored data to the control center for analysis. Data transmission should have high reliability and low latency to ensure that the control center can obtain the latest data in a timely manner.

[0064] As an example, the real-time user behavior data may be collected through a user terminal (e.g., a smart phone or tablet computer, etc.), and the real-time user behavior data may include the user's location information, movement trajectory, connection duration, etc. The real-time user behavior data may be, but is not limited to, interacting with the user device through an application program interface (API) to obtain the user's location information, etc. in real time. For example, when a user moves indoors, the user's current location may be obtained through GPS or WiFi positioning technology, and the data may be transmitted to the control center.

[0065] As an example, the collected real-time data can be transmitted to the control center via a wireless network or a wired network, and the control center will then centrally process and store the real-time data. During the data transmission process, encryption technology can be used to ensure the security and privacy of the data to prevent the data from being tampered with or leaked during the transmission process.

[0066] In step S20, refer to Figure 1 In step S20, the real-time data is preprocessed.

[0067] As an example, step S20 may include the following steps: S201 to S203.

[0068] S201: performing denoising processing on the real-time data.

[0069] S202: normalize the denoised data.

[0070] S203: Extract features from the normalized data.

[0071] As an example, in step S201, a statistical method (eg, mean filtering) may be used to perform denoising on the real-time data.

[0072] As an example, in step S202, the denoised data is normalized and scaled to a range of 0 to 1 to facilitate the training of a deep learning model.

[0073] As an example, in step S203, feature extraction can be performed through methods such as principal component analysis (PCA) to reduce data dimensions and improve the training effect of subsequent deep learning models.

[0074] In step S30, refer to Figure 1 In step S30, a deep learning model is constructed.

[0075] As an example, step S30 may include the following steps: S301 to S305.

[0076] S301: Build an initial deep learning model.

[0077] S302: Acquire historical data of the indoor area, where the historical data includes: historical environmental data of the indoor area, historical usage data of the indoor distribution system, and historical user behavior data of the indoor area.

[0078] S303: Training the initial deep learning model based on the historical data.

[0079] S304: Evaluate the trained initial deep learning model, and optimize the trained initial deep learning module based on the evaluation result to obtain the deep learning model.

[0080] As an example, in step S301, the constructed initial deep learning module may include a neural network model, such as at least one of a convolutional neural network model (CNN) and a recurrent neural network model (RNN). CNN is suitable for processing spatial features and can effectively extract local features in environmental data. The input layer of CNN receives preprocessed environmental data, and extracts high-level feature representations after processing by multiple convolutional layers and pooling layers. RNN is suitable for processing time series data and can capture time dependencies in user behavior data and device usage data. The input layer of RNN receives user behavior data and device usage data, and generates feature representations of time series after processing by multiple recurrent layers.

[0081] As an example, in step S302, the method for obtaining historical data and the method for obtaining real-time data may be the same. For details, please refer to the method for obtaining real-time data, which will not be described again here.

[0082] As an example, in step S304, evaluating the trained initial deep learning model may be specifically: in the process of training the initial deep learning model, a cross entropy loss function (for classification problems) or a mean square error loss function (for regression problems) is used to evaluate the performance of the initial deep learning model. The mathematical expression of the loss function may be:

[0083]

[0084] Among them, y i is the true value of the i-th record, is the predicted value output by the initial deep learning model, and N is the number of samples in the initial deep learning model.

[0085] As an example, in step S304, the parameters of the initial deep learning model can be updated by a back-propagation algorithm and an optimization algorithm (such as Adam or SGD).

[0086] As an example, in step S304, after the initial deep learning model training is completed, the validation set can also be used to evaluate the initial deep learning model to calculate indicators such as accuracy and recall to judge the performance of the initial deep learning model; and then the architecture or hyperparameters of the initial deep learning module can be further adjusted based on the evaluation results to obtain the final deep learning model.

[0087] In step S40, refer to Figure 1 In step S40, the energy valley optimization algorithm is used to optimize the deep learning model.

[0088] As an example, step S40 may include the following steps: S401 to S404.

[0089] S401: Randomly generate a parameter set of the deep learning model, where the parameter set includes multiple parameters, each of which is a weight or a bias of the deep learning model.

[0090] S402: Obtaining energy of the deep learning model based on the parameter set.

[0091] S403: Perform iterative optimization based on the parameter set and the energy to obtain an optimal parameter set.

[0092] S404: Update the optimal parameter set to the deep learning model to optimize the deep learning model.

[0093] As an example, in step S401, a parameter set θ={θ1, θ2, ...θ n}, the number of parameters in the parameter set θ can be n, where n is an integer greater than 2. Each parameter θ in the parameter set θ i Represents a weight or a bias of the deep learning model in the current state. The Education Department is correct. When generating the parameter set θ, a uniform distribution or a normal distribution can be used to generate parameter values ​​to ensure the diversity of parameters in the parameter set θ.

[0094] As an example, in step S402, obtaining the energy of the deep learning model based on the parameter set may include: obtaining the energy of the deep learning model based on the following loss function:

[0095]

[0096] Among them, y i is the true value of the i-th record, is the predicted value output by the deep learning model, N is the number of samples in the deep learning model, and i is an integer greater than or equal to 1 and less than or equal to N.

[0097] It should be noted that the energy E(θ) is the energy of the deep learning model in the current state, y i is the true value of the i-th record, is the predicted value output by the deep learning model in the current state, and N is the number of samples in the deep learning model in the current state.

[0098] As an example, in step S403, performing iterative optimization based on the parameter set and the energy to obtain an optimal parameter set may include the following steps: S4031 to S4034.

[0099] S4031: Randomly generate a new parameter set in the neighborhood of the parameter set, so that at least one parameter in the parameter set increases by a random value.

[0100] S4032: Obtain new energy of the deep learning model based on the new parameter set.

[0101] S4033: If the new energy is less than the current energy, the new parameter set is updated to the current parameter set; if the new energy is greater than or equal to the current energy, the new parameter set is updated to the current parameter set based on the probability obtained by the following probability formula; the probability formula is: Wherein, P is the probability, T is the current temperature of the indoor area, E(θ) is the current energy, and E(θ') is the new energy.

[0102] S4034: Repeat the above steps until the iteration ends or the energy converges, and take the current parameter set at this time as the optimal parameter set.

[0103] As an example, in step S4031, the definition of the neighborhood can be achieved by adding a small random perturbation so that at least one parameter in the parameter set θ is increased by a random value to obtain a new parameter set θ', that is, at least one parameter in the parameter set θ is added with a random value on the basis of the current value to update it to a new parameter.

[0104] It should be noted that since the parameters of the new parameter set θ' are changed compared to the parameter set θ, the state of the deep learning model corresponding to the new parameter set θ' will also change, and will also be updated from the original current state to another different new state.

[0105] As an example, in step S4032, the specific method for obtaining the new energy of the deep learning model based on the new parameter set can refer to the formula of step S402, except that in this step, y∧ i is the predicted value output by the deep learning model of the new state corresponding to the new parameter set, and N is the number of samples in the deep learning model of the new state corresponding to the new parameter set.

[0106] As an example, between step S4033 and step S4044, the following step may also be included: gradually updating the current temperature of the indoor area to reduce the probability of accepting a poor solution. Specifically, the current temperature of the indoor area may be updated based on the following formula: T = αT'; wherein T is the current temperature of the indoor area after update, T' is the current temperature of the indoor area before update, α is the cooling rate, and the value of α is between 0 and 1, that is, α can be greater than 0 and less than 1.

[0107] As an example, in step S4044, energy convergence may be that the energy is less than a set threshold, and the size of the set threshold may be set according to actual conditions and is not limited here.

[0108] Specifically, the parameter set randomly generated in step S401 is θ={0.5, -0.3, 0.8}, and the energy of the deep learning model at this time is calculated to be: E(θ)=0.02. The new parameter set generated in one iteration can be θ'={0.6, -0.2, 0.7}, and the new energy is calculated to be E(θ')=0.015. Since the new energy E(θ') is less than the energy E(θ), the new parameter set is used as the new one to continue iterating the current parameter set. After multiple iterations, the optimal parameter set can be obtained as θ * ={0.75, -0.1, 0.65}; Applying the optimal parameter set to the deep learning network can achieve higher prediction accuracy and faster response speed.

[0109] In step S50, refer to Figure 1 In step S50, the preprocessed real-time data is input into the optimized deep learning model, and the optimized deep learning model outputs a power adjustment scheme based on the preprocessed real-time data.

[0110] As an example, the power adjustment scheme may include power output recommendations for different areas and time periods in the indoor area. The optimal power output for each area in the indoor area can be calculated through the forward propagation of the optimized deep learning model. The output power adjustment scheme should take into account multiple factors such as user needs, environmental changes, and device status.

[0111] As an example, steps S20 to S50 may all be executed by a control center in the indoor distributed system.

[0112] In step S60, refer to Figure 1 In step S60, power is adjusted for the indoor distributed system based on the power adjustment scheme.

[0113] As an example, the power regulation module in the indoor distributed system may regulate the power of the base station and / or antenna in the indoor distributed system based on the power modulation scheme to adjust the output power of each area in the indoor area.

[0114] As an example, after step S60, that is, after the power regulation of the indoor distributed system based on the power regulation scheme, it also includes: continuously repeating the following steps: real-time collection of current data of the indoor area, the current data including current environmental data of the indoor area, current usage data of the indoor distributed system and current user behavior data of the indoor area; preprocessing the current data; optimizing the current deep learning model using the energy valley optimization algorithm; inputting the preprocessed current data into the optimized current deep learning model, and the optimized current deep learning model outputs a new power regulation scheme based on the preprocessed current data; and power regulating the indoor distributed system based on the new power regulation scheme. By continuously repeating the above steps, a closed-loop feedback mechanism can be formed in the process of power regulation of the indoor distributed system, so that the deep learning model can be dynamically adjusted in real time to ensure that the optimal power regulation scheme can be obtained.

[0115] The power regulation method of the indoor distributed system of the present invention optimizes the deep learning model through the energy valley optimization algorithm. The energy valley optimization algorithm can effectively jump out of the local optimal solution and find the global optimal solution, thereby improving the prediction accuracy of the deep learning model; the energy valley optimization algorithm can receive poor solutions in the early stage, thereby accelerating the convergence speed and reducing the iteration time; through the energy valley optimization algorithm, the deep learning model can better adapt to different data distributions and enhance the robustness of the deep learning model; the energy valley optimization algorithm can consider resource usage during the optimization process, thereby achieving the optimal configuration of resources; the energy valley optimization algorithm can optimize multiple objective functions at the same time, thereby solving complex multi-objective optimization problems; the energy valley optimization algorithm can enable the deep learning model to better avoid overfitting and improve generalization ability; the energy valley optimization can continuously optimize the neural network model in the real-time data stream to adapt it to the dynamically changing environment.

[0116] In another embodiment, see Figure 2The present invention also provides a power regulation device for a distributed indoor system, which may include: an acquisition module 10, a preprocessing module 20, a model building module 30, a model optimization module 40 and a power regulation module 50; wherein the acquisition module 10 is used to collect real-time data of the indoor area, and the real-time data includes real-time environmental data of the indoor area, real-time usage data of the distributed indoor system and real-time user behavior data of the indoor area; the preprocessing module 20 is used to preprocess the real-time data; the model building module 30 is used to construct a deep learning model; the model optimization module 40 is used to optimize the deep learning model using an energy valley optimization algorithm to obtain an optimized deep learning model, and the optimized power regulation module outputs a power regulation scheme based on the preprocessed real-time data; the power regulation module 50 is used to power regulate the distributed indoor system based on the power regulation scheme.

[0117] As an example, the power regulation device of the room distribution system can be used to perform the following Figure 1 The power regulation method of the indoor distributed system described in its embodiment.

[0118] As an example, the indoor distributed system may include a base station (not shown), a distributed antenna (not shown), the power adjustment module 50 and a control center (not shown), etc. The acquisition module 10 may include various sensors, such as a temperature sensor and a humidity sensor, etc. The preprocessing module 20, the model building module 30 and the model optimization module 40 may all be located inside the control center.

[0119] For the specific definition of the power regulation device of the distributed indoor system, please refer to the definition of the power regulation method of the distributed indoor system above, which will not be repeated here. The various modules in the power regulation device of the distributed indoor system can be implemented in whole or in part through software, hardware and their combination. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0120] In another embodiment, the present invention further provides a computer device, which may be a server, and its internal structure diagram may be as follows: Figure 3As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data such as real-time data, historical data and power regulation schemes. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it is implemented as follows Figure 1 The steps of the power regulation method of the indoor distributed system and its embodiment.

[0121] In another embodiment, the present invention further provides a computer device, which may be a terminal, and its internal structure diagram may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it is implemented as follows Figure 1 The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0122] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0123] In another embodiment, the present invention further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following is achieved: Figure 1 The steps of the power regulation method of the indoor distributed system and its embodiment.

[0124] In one embodiment, a computer readable storage medium is provided on which a computer program is stored. When the computer program is executed by a processor, Figure 1 The steps of the power regulation method of the indoor distributed system and its embodiment.

[0125] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0126] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0127] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A power regulation method for a room distribution system, It is characterized in that include: Collecting real-time data of indoor areas, including real-time environmental data of indoor areas, real-time usage data of indoor distribution systems, and real-time user behavior data of indoor areas; Preprocessing the real-time data; Build deep learning models; Use the energy valley optimization algorithm to optimize the deep learning model; The preprocessed real-time data is input into the optimized deep learning model, and the optimized deep learning model outputs a power regulation scheme based on the preprocessed real-time data; The power of the indoor distributed system is regulated based on the power regulation scheme.

2. The power regulation method of the indoor distributed system according to claim 1, It is characterized in that The preprocessing of the real-time data comprises: Performing denoising processing on the real-time data; Normalize the denoised data; The normalized data is used for feature extraction.

3. The power regulation method of the indoor distributed system according to claim 1, It is characterized in that The deep learning model is constructed, comprising: Build an initial deep learning model; Acquire historical data of the indoor area, the historical data including: historical environmental data of the indoor area, historical usage data of the indoor distribution system, and historical user behavior data of the indoor area; Training the initial deep learning model based on the historical data; The trained initial deep learning model is evaluated, and the trained initial deep learning module is optimized based on the evaluation result to obtain the deep learning model.

4. The power regulation method of the indoor distributed system according to claim 1, It is characterized in that The method of optimizing the deep learning model using the energy valley optimization algorithm includes: Randomly generate a parameter set of the deep learning model, the parameter set including a plurality of parameters, each of the parameters being a weight or a bias of the deep learning model; Obtaining energy of the deep learning model based on the parameter set; Performing iterative optimization based on the parameter set and the energy to obtain an optimal parameter set; The optimal parameter set is updated to the deep learning model to optimize the deep learning model.

5. The power regulation method of the indoor distributed system according to claim 4, It is characterized in that The obtaining the energy of the deep learning model based on the parameter set includes: obtaining the energy of the deep learning model based on the following loss function: Among them, y i is the true value of the i-th record, is the predicted value output by the deep learning model, N is the number of samples in the deep learning model, and i is an integer greater than or equal to 1 and less than or equal to N.

6. The power regulation method of the indoor distributed system according to claim 4, It is characterized in that The iterative optimization based on the parameter set and the energy to obtain an optimal parameter set includes: Randomly generate a new parameter set in a neighborhood of the parameter set so that at least one parameter in the parameter set increases by a random value; Obtaining new energy for the deep learning model based on the new parameter set; If the new energy is less than the current energy, the new parameter set is updated to the current parameter set; if the new energy is greater than or equal to the current energy, the new parameter set is updated to the current parameter set based on the probability obtained by the following probability formula; the probability formula is: Wherein, P is the probability, T is the current temperature of the indoor area, E(θ) is the current energy, and E(θ') is the new energy; Repeat the above steps until the iteration ends or the energy converges, and take the current parameter set at this time as the optimal parameter set.

7. The power regulation method of the indoor distributed system according to any one of claims 1 to 6, It is characterized in that After the power of the indoor distributed system is adjusted based on the power adjustment scheme, the method further includes: continuously repeating the following steps: Collecting the current data of the indoor area in real time, the current data includes the current environment data of the indoor area, the current usage data of the indoor distribution system and the current user behavior data of the indoor area; Preprocessing the current data; Use the energy valley optimization algorithm to optimize the current deep learning model; Inputting the preprocessed current data into the optimized current deep learning model, and the optimized current deep learning model outputs a new power regulation scheme based on the preprocessed current data; The power of the indoor distributed system is regulated based on a new power regulation scheme.

8. A power regulation device for a room distribution system, It is characterized in that The power regulation device of the indoor distributed system comprises: A collection module is used to collect real-time data of the indoor area, wherein the real-time data includes real-time environmental data of the indoor area, real-time usage data of the indoor distribution system and real-time user behavior data of the indoor area; A preprocessing module, used for preprocessing the real-time data; Model building module, used to build deep learning models; A model optimization module is used to optimize the deep learning model using an energy valley optimization algorithm to obtain an optimized deep learning model. The optimized power regulation module outputs a power regulation solution based on the preprocessed real-time data; A power regulation module is used to regulate the power of the indoor distributed system based on the power regulation scheme.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program. It is characterized in that When the processor executes the computer program, the steps of the power regulation method for the indoor distributed system described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the power regulation method for the indoor distributed system described in any one of claims 1 to 7 are implemented.

Images