Method and system for determining base station layout configuration scheme in intelligent building

Abstract

The invention discloses a method and system for determining a base station layout configuration scheme in an intelligent building. The method comprises the steps of collecting first / second building data, historical first / second working condition data and historical first / second base station layout configuration data; taking the acquired first / second building data, the historical first / second working condition data and the historical first / second base station layout configuration data as training data, and performing corresponding training by using a deep learning algorithm to generate a building overall layout configuration model or a building single-layer layout configuration model; building data and working condition requirements of a target building are collected, the building overall layout configuration model and the building single-layer layout configuration model are respectively input, and base station layout configuration of the target building is comprehensively obtained; and monitoring the signal quality of each position in the target building in real time, and dynamically adjusting the configuration of the base stations around the position where the real-time signal quality is lower than the preset quality position by using an AI model. According to the invention, the problem of poor high-frequency communication quality in a complex building environment can be solved.

Description

Technical Field

[0001] The present application relates to the technical field of base station layout configuration in intelligent buildings, and in particular to a method and system for determining a base station layout configuration scheme in an intelligent building. Background Art

[0002] With the rise of smart buildings, high-frequency data exchange has become the key to ensuring the efficient operation of various intelligent facilities in buildings. Modern buildings not only require efficient power supply, temperature control systems and safety protection measures, but also must have strong wireless communication capabilities to support the interconnection of various IoT devices. However, achieving high-quality wireless network coverage in existing buildings, especially meeting the needs of high-frequency band communications, faces many challenges. These problems mainly include severe signal attenuation, frequent multipath interference, and the overall impact of complex indoor environments on communication efficiency.

[0003] Traditional solutions usually rely on preset models and empirical values, and determine the number and location of base stations through theoretical calculations or limited field tests. Specifically, common methods include: 1) using radio wave propagation models to estimate the number of base stations required; 2) selecting the best installation location based on historical cases; 3) adding additional antennas in key areas to enhance local signal strength; 4) using high-performance materials to reduce signal loss. Although these methods have improved the quality of wireless networks to a certain extent, they still have obvious limitations.

[0004] The above methods often fail to accurately evaluate the various interference factors that change dynamically in complex building environments, resulting in the network performance after actual deployment being far lower than expected, especially in scenarios with high-frequency data transmission requirements. Therefore, it is necessary to provide a method for determining the base station layout configuration scheme that can adapt to complex building environments, thereby improving the performance of the deployed network and user experience satisfaction. Summary of the invention

[0005] In order to solve the problem of poor high-frequency communication quality in complex building environments, the present application provides a method and system for determining a base station layout configuration solution in an intelligent building.

[0006] In a first aspect, the present application provides a method for determining a base station layout configuration scheme in an intelligent building, comprising: Collect construction data, historical operating data, and historical base station layout configuration data of several buildings; Statistically obtain first building data, historical first operating condition data, and historical first base station layout configuration data belonging to the entire building, and second building data, historical second operating condition data, and historical second base station layout configuration data belonging to a single floor of the building; The first building data, the first historical working condition data and the first historical base station layout configuration data are used as training data, and a deep learning algorithm is used to train and construct a building overall layout configuration model, wherein the input of the building overall layout configuration model is the first building data and the first working condition data, and the output is the number, configuration and vertical distribution of building base stations that meet the first preset requirement; the first preset requirement includes a reward function with signal coverage and base station installation cost as reward factors reaching the maximum value; the second building data, the second historical working condition data and the second historical base station layout configuration data are used as training data, and a deep learning algorithm is used to train and construct a building single-layer layout configuration model, wherein the input of the building single-layer overall layer layout configuration model is the second building data and the second working condition data, and the output is the number, configuration and horizontal distribution of building single-layer base stations that meet the second preset requirement; the second preset requirement includes a multi-objective optimization function with signal coverage, capacity and base station installation cost reaching the optimal solution; Collect the building data and working condition requirements of the target building and input them into the building overall layout configuration model and the building single-layer layout configuration model respectively, and comprehensively obtain the base station layout configuration of the target building; The sensor array deployed in the target building is used to monitor the signal quality of each location in the target building in real time, and the AI ​​model is used to dynamically adjust the configuration of base stations around the location where the real-time signal quality is lower than the preset quality, so that the location where the real-time signal quality is lower than the preset quality reaches the preset signal quality at the next moment. The configuration includes carrier frequency occupancy, antenna transmission power and beam direction.

[0007] By adopting the above solution, a base station layout configuration model with strong environmental adaptability is constructed using a deep learning algorithm. The layout configuration models of the entire building and a single floor are trained and generated respectively. This ensures that the model fully considers factors affecting data exchange such as building structure and working conditions, and automatically adjusts the number and optimal layout of base stations. On this basis, the signal quality is monitored in real time, and the base station configuration is dynamically adjusted using an AI model, further improving the stability of the system and user experience.

[0008] Preferably, the method of using a deep learning algorithm to train and construct a building overall layout configuration model specifically includes: Performing data preprocessing on the input first building data, the historical first operating condition data, and the historical first base station layout configuration data to generate a first training set; A graph neural network (GNN) integrated with a reinforcement learning framework is selected as the overall building layout configuration model architecture, specifically including: establishing a graph neural network (GNN), including: defining nodes, defining edges, and building a graph structure based on the defined nodes and variables; node definition includes: defining each floor as a node; node feature definition includes: floor height, area, shape and material properties, signal strength, and user density; edge definition includes: defining stairs and elevator shafts as edges; then combining the reinforcement learning framework with the GNN, including: defining an intelligent agent in the reinforcement learning framework, using the intelligent agent to observe the building data, working condition data, and historical base station layout of the target building, and generating a base station layout adjustment decision based on the base station layout output by the GNN; updating the state based on the adjustment decision of the intelligent agent, and calculating the reward function, which is designed based on the signal coverage rate and the base station installation cost, until the reward function reaches the maximum value and the corresponding adjustment decision is output; The first training set is used to train the graph neural network GNN of the constructed integrated reinforcement learning framework to obtain a trained building overall layout configuration model.

[0009] By adopting the above solution, taking into account the overall structure and material characteristics of the building, floor height, floor user density and other contents, the reinforcement learning framework and graph neural network are used as the basic framework of the building overall layout configuration model. It can not only automatically adjust the base station configuration under different building structures and material conditions to improve the adaptability of the model, but also ensure the maximum signal coverage while reducing installation costs.

[0010] Preferably, the method of using a deep learning algorithm to train and construct a single-layer building layout configuration model specifically includes: Performing data preprocessing on the input second building data, the historical second operating condition data, and the historical second base station layout configuration data to generate a second training set; A hybrid neural network of CNN and LSTM is selected as the architecture of the single-floor layout configuration model of the building, specifically including: constructing a CNN network model, using the constructed CNN to process the spatial features of the second building data, including: for a two-dimensional grid image generated by converting the single-floor layout information in the second building data, identifying the spatial features of key grids, including: room locations, wall material distribution, and obstacle locations; Construct an LSTM network model, and use LSTM to process the time series data in the second working condition data and the second base station layout configuration data, including: time series characteristics of user density and signal strength; Define a multi-objective optimization function, including: a target optimization function generated by weighted calculation of signal coverage, signal capacity and base station installation cost; The constructed hybrid neural network of CNN and LSTM is trained using the second training set, and the parameters of the model are continuously adjusted during the training process so that the objective optimization function reaches the optimal solution, thereby obtaining a trained single-story building layout configuration model.

[0011] By adopting the above scheme, the CNN network is used to process spatial features such as room location, material distribution, and obstacle location within a single floor of a building, ensuring that the layout of base stations on a single floor of the building fully considers the impact of the physical environment and reduces signal blind spots; and the LSTM network is used to process the changing trends of floor user density and signal strength to dynamically adjust the base station configuration; the introduction of a multi-objective optimization function balances the relationship between signal coverage, capacity, and base station installation cost to achieve effective utilization of resources.

[0012] Preferably, it also includes: Obtain the satisfaction of users in the target building with signal quality feedback in real time, and determine whether the proportion of current user satisfaction exceeding the preset satisfaction number to the total number of user feedback is lower than the preset proportion. If so, expand the range of base stations around the location where the real-time signal quality is lower than the preset quality, and dynamically adjust the configuration of base stations around the location where the real-time signal quality is lower than the preset quality according to the AI ​​model.

[0013] By adopting the above solution, considering the user's feedback on the satisfaction of the base station layout in the target building, the configuration of these base stations is dynamically adjusted based on the user feedback mechanism to improve the signal coverage quality and stability.

[0014] Preferably, it also includes: If after expanding the base station range around the location where the real-time signal quality is lower than the preset quality, and readjusting the base station configuration, the proportion of user satisfaction with signal quality feedback in the target building obtained over a period of time exceeds the preset satisfaction but is still lower than the preset proportion, then the overall building layout configuration model and / or the single-layer building layout configuration model are optimized accordingly.

[0015] By adopting the above solution, considering that there are still limitations in continuously collecting user signal quality feedback and dynamically adjusting the configuration of base stations based on user feedback, it can be preliminarily determined that there are errors in the base station layout determination model. Correspondingly, new simulation data is introduced as the data source for incremental learning, so that the model can be gradually improved, respond to complex changes in building environments more accurately, and optimize the base station layout.

[0016] Preferably, the optimization of the overall building layout configuration model and / or the single-layer building layout configuration model includes: Create a virtual building signal communication scenario corresponding to a typical building structure, perform signal attenuation simulation, generate signal coverage simulation data, and use it as incremental data for the first training set and the second training set to achieve incremental learning of the building overall layout configuration model and / or the building single-layer layout configuration model.

[0017] By adopting the above scheme, a virtual building signal communication scenario corresponding to a typical building structure is created, the complex environment in reality is fully simulated, simulation data is obtained to supplement the training data, and incremental learning of the overall building layout configuration model and / or the single-layer building layout configuration model is achieved, thereby improving the model's adaptability to different types of buildings.

[0018] Preferably, the collecting of the target building's architectural data and operating condition requirements into the building's overall layout configuration model and the building's single-layer layout configuration model respectively also includes: making a preset proportion increase adjustment for the operating condition requirements of the collected target building, and inputting the collected target building's architectural data and the adjusted operating condition requirements into the building's overall layout configuration model and the building's single-layer layout configuration model respectively.

[0019] By adopting the above solution and adjusting the target building's operating requirements by a preset proportion, we can more comprehensively consider the pressure brought by the possible increase in service demand or the growth in the number of users in the future, optimize the base station configuration in advance, and avoid waste of resources and technical difficulties caused by frequent adjustments in the later stage.

[0020] In a second aspect, the present application provides a system for determining a base station layout configuration solution in an intelligent building, comprising: The building training data collection and processing module is used to collect the building data, historical working condition data and historical base station layout configuration data of several buildings; statistically obtain the first building data, historical first working condition data and historical first base station layout configuration data belonging to the entire building, and the second building data, historical second working condition data and historical second base station layout configuration data belonging to a single floor of the building; A building overall layout configuration model acquisition module is used to use the first building data, the historical first working condition data and the historical first base station layout configuration data as training data, and use a deep learning algorithm to train and construct a building overall layout configuration model, wherein the input of the building overall layout configuration model is the first building data and the first working condition data, and the output is the number, configuration and vertical distribution of building base stations that meet the first preset requirement; the first preset requirement includes that a reward function with signal coverage and base station installation cost as reward factors reaches a maximum value; A building single-layer layout configuration model acquisition module is used to use the second building data, the historical second operating condition data and the historical second base station layout configuration data as training data, and use a deep learning algorithm to train and construct a building single-layer layout configuration model, wherein the input of the building single-layer overall layer layout configuration model is the second building data and the second operating condition data, and the output is the number, configuration and horizontal distribution of building single-layer base stations that meet the second preset requirement; the second preset requirement includes achieving the optimal solution with a multi-objective optimization function of signal coverage, capacity and base station installation cost; The target building base station layout configuration acquisition module is used to collect the building data and working condition requirements of the target building and input them into the building overall layout configuration model and the building single-layer layout configuration model respectively, so as to comprehensively acquire the target building base station layout configuration; The target building base station layout configuration adjustment module is used to use the sensor array deployed in the target building to monitor the signal quality of each location in the target building in real time, and use the AI ​​model to dynamically adjust the configuration of the base stations around the location where the real-time signal quality is lower than the preset quality, so that the location where the real-time signal quality is lower than the preset quality reaches the preset signal quality at the next moment. The configuration includes carrier frequency occupancy, antenna transmission power and beam direction.

[0021] By adopting the above solution, the overall building and single-layer layout configuration models trained with deep learning algorithms can more accurately predict and optimize the number and distribution of base stations. At the same time, a real-time monitoring mechanism is introduced to quickly respond to environmental changes, automatically adjust the working status of base stations, and improve the stability and reliability of the system.

[0022] In a third aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method as described above.

[0023] In a fourth aspect, the present application provides a computer device, comprising a memory, a processor, and a program stored and executable on the memory, wherein the program implements the steps of the above method when executed by the processor.

[0024] In summary, this application has the following beneficial effects: 1. The overall building and single-layer layout configuration models are constructed through deep learning algorithms to adapt to different complex building environments, more accurately predict the signal propagation characteristics and optimal base station layout solutions in complex building environments, and significantly improve the data transmission performance of the deployed network; at the same time, a real-time monitoring mechanism is introduced to dynamically adjust the base station configuration to deal with sudden signal quality problems, further enhancing the stability and reliability of data transmission; 2. Integrate user feedback mechanism to dynamically adjust base station configuration or incrementally train building overall and single-layer layout configuration models based on continuous monitoring of user feedback data. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a flow chart of a method for determining a base station layout configuration scheme in an intelligent building described in a specific embodiment; Figure 2 It is a structural diagram of a system for determining a base station layout configuration scheme in an intelligent building described in a specific embodiment. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application 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 application and are not used to limit the present application.

[0027] like Figure 1 As shown, the embodiment of the present application discloses a method for determining a base station layout configuration scheme in an intelligent building, and the specific steps include: S1. Collect architectural data, historical operating data, and historical base station layout configuration data of several buildings.

[0028] Specifically, the building data, historical working condition data and historical base station layout configuration data of the collected buildings are obtained through various channels, such as: architectural design drawings, construction records, past network maintenance logs, etc. In order to ensure the accuracy and completeness of the collected data, automated data collection tools such as drone aerial photography and 3D scanners can be used to assist manual verification and validation.

[0029] The collected building data include but are not limited to the overall building data, such as: the overall shape of the floor, building height, number of floors, floor height, wall materials, stairs, and door and window locations, etc. The single floor building data include but are not limited to: the room distribution of the floor, wall materials, obstacle locations, etc.; Historical operating condition data includes but is not limited to: user density distribution in buildings over a period of time, signal strength fluctuations, business demand data, equipment failure records, etc.; historical base station layout configuration data includes but is not limited to: the number, location, antenna type and transmission power of base stations deployed in buildings over a period of time.

[0030] S2. Perform data preprocessing on the collected building data, historical operating condition data, and historical base station layout configuration data.

[0031] Specifically, in order to comprehensively and accurately train the model to automatically adjust the number and optimal layout of base stations, the collected data is classified and counted in advance to obtain the first building data belonging to the entire building, the historical first operating condition data and the historical first base station layout configuration data, the second building data belonging to a single floor of the building, the historical second operating condition data and the historical second base station layout configuration data, etc.

[0032] Among them, the first building data includes: overall indicators such as the overall building shape, building height, number of floors, floor height, wall material distribution and structural distribution; the first historical working condition data includes: the overall user density distribution of the building, signal strength distribution, signal interference distribution, whether the base station is faulty or not, etc.; the first historical base station layout configuration data includes: the overall number, location, antenna type and transmission power of base stations in the building, etc.

[0033] Among them, the second building data includes: indicators of a single floor such as the height and area of ​​a single floor, room distribution, wall material distribution, obstacle location, etc.; the historical second operating condition data includes: user density distribution, signal strength distribution, signal interference distribution, base station failure status, etc. on a single floor; the historical second base station layout configuration data includes: the number, location, antenna type and transmission power of base stations on a single floor.

[0034] S3. Build and train a building overall layout configuration model.

[0035] Specifically, based on the deep learning algorithm, a suitable algorithm model is selected in advance to build a building overall layout configuration model to obtain the optimal number of base stations and their vertical distribution in the target building to adapt to the signal attenuation caused by different building structures and material properties, thereby achieving deployment network performance optimization.

[0036] While optimizing the performance of base station deployment, it is also necessary to consider the cost of base station deployment and reduce the base station installation cost as much as possible. Therefore, the input of the building overall layout configuration model is further designed to be the first building data and the first operating condition data, and the output is the number, configuration and vertical distribution of building base stations that meet the first preset requirements, as well as the basic output configurations such as the frequency band allocation and power allocation of the base stations; wherein the first preset requirement includes a reward function with signal coverage and base station installation cost as reward factors reaching the maximum value, thereby ensuring that the determined building base station data and vertical distribution meet the building signal coverage requirements and balance the installation costs.

[0037] The first building data, the first historical operating condition data and the first historical base station layout configuration data are used as training data, and divided into a training set and a test set; the training set is used to train the constructed building overall layout configuration model, and the test set is used to test whether the building overall layout configuration model meets the preset requirements.

[0038] S4. Build and train a single-floor building layout configuration model.

[0039] Specifically, a suitable algorithm model is selected in advance based on the deep learning algorithm to construct a single-layer overall floor layout configuration model of the building, and the optimal number and horizontal distribution of base stations on a single floor of the target building are obtained to adapt to the signal attenuation caused by the structure and material characteristics of the single-layer building, thereby achieving deployment network performance optimization.

[0040] While optimizing the performance of base station deployment, it is also necessary to consider the operating conditions in a single-story building and the cost of base station deployment, and consider the base station deployment layout of a single floor as detailed as possible. Therefore, the input of the single-story overall floor layout configuration model of the building is further designed to be the second building data and the second operating condition data, and the output is the number, configuration and horizontal distribution of single-story base stations in the building that meet the second preset requirements, as well as the basic output configurations such as frequency band allocation and power allocation of the base stations; wherein the second preset requirement includes achieving the optimal solution with a multi-objective optimization function of signal coverage, capacity and base station installation cost, thereby ensuring that the determined single-story floor horizontal distribution meets the signal coverage and specific operating conditions of the building and balances the installation cost.

[0041] The second building data, the historical second operating condition data and the historical second base station layout configuration data are used as training data, and divided into a training set and a test set; the constructed single-layer layout configuration model of the building is trained using the training set, and the single-layer layout configuration model of the building is tested using the test set to meet the preset requirements.

[0042] S5. Collect the building data and working condition requirements of the target building, and input the overall building layout configuration model and the single-layer building layout configuration model to obtain the base station layout of the target building.

[0043] Specifically, the building data and operating condition requirements of the target building are collected, including the building data and operating condition requirements of the target building as a whole, and the building data and operating condition requirements of a single floor of the target building; among which, the operating condition requirements can determine the user density range and signal strength distribution requirements based on the historical operating conditions of the target building; or based on the operating condition requirements set by user input.

[0044] The architectural data and working condition requirements of the target building are collected and input into the overall building layout configuration model to obtain the number, configuration and vertical distribution of base stations in the target building. For example, the overall number of base stations in the target building is 20, and the ratio of base stations in a 6-story building is 1:2:3:1:1:2.

[0045] The building data and working condition requirements of the target building are collected and input into the single-story layout configuration model of the building to obtain the number, configuration and horizontal distribution of base stations in the single-story building. For example, the overall number of base stations in a single-story building is 2, and the single-story building is divided into two areas from east to west for base station installation, with a ratio of 1:1.

[0046] The outputs of the overall building layout configuration model and the single-layer building layout configuration model are integrated to obtain the target building base station layout configuration. In order to avoid output contradictions, the priority between the number, configuration and vertical distribution of target building base stations and the number, configuration and horizontal distribution of single-layer building base stations can be determined according to user needs. For example, if the user pays more attention to the overall base station layout configuration of the building, a higher priority is set for the number, configuration and vertical distribution of target building base stations. On the basis of meeting the number, configuration and vertical distribution of target building base stations, the number, configuration and horizontal distribution of single-layer building base stations are further adjusted. Similarly, if the user pays more attention to the base station layout configuration of a single floor, a higher priority is set for the number, configuration and vertical distribution of base stations on a single floor of the target building. On the basis of meeting the number, configuration and horizontal distribution of base stations on a single floor of the target building, the number, configuration and vertical distribution of base stations on a single floor of the target building are further adjusted.

[0047] S6. Conduct real-time monitoring of target buildings and dynamically adjust and optimize base station configuration.

[0048] In order to further improve the performance of the deployed network, the sensor array deployed in the target building is used to monitor the signal quality of each location in the target building in real time. The signal quality includes quality indicator parameters such as signal strength and signal interference index (signal-to-noise ratio). Specifically, a certain number of micro sensors can be arranged on each floor. These sensors can monitor key indicators such as signal strength, noise level, and multipath effect at each location around the clock.

[0049] The AI ​​model is used to dynamically adjust the configuration of base stations around the location where the real-time signal quality is lower than the preset quality, so that the location where the real-time signal quality is lower than the preset quality reaches the preset signal quality at the next moment. The configuration includes carrier frequency occupancy, antenna transmission power and beam direction. The AI ​​model uses a deep learning neural network, and uses the historical configuration of the base station to be adjusted and its preset surrounding base stations and the historical optimal adjustment plan to train and generate an AI model that outputs the base station configuration adjustment plan.

[0050] By adopting the above solution, a deep learning algorithm is used to build a building overall and single-layer layout configuration model to more accurately determine the optimal base station layout configuration solution in a complex building environment.

[0051] In a specific embodiment, in order to more accurately optimize the base station layout according to the signal propagation characteristics in a complex building environment to improve the transmission performance of the deployed network, the characteristic indicators of the overall base station layout of the building are integrated, and a model capable of extracting the overall environmental characteristics of the building, the signal transmission characteristics and optimizing the layout output is adaptively designed. The method uses a deep learning algorithm to train and construct an overall building layout configuration model, which specifically includes: Data preprocessing is performed on the input first building data, the historical first operating condition data and the historical first base station layout configuration data to generate a first training set; wherein the preprocessing includes data cleaning, feature extraction and standardization.

[0052] The graph neural network (GNN) integrated with the reinforcement learning framework is selected as the overall building layout configuration model architecture, which includes: First, a graph neural network (GNN) is established; specifically, it includes: defining nodes, defining edges and building a graph structure based on the defined nodes and variables; wherein, the node definition includes: defining each floor as a node, and also including base station sub-nodes and user sub-nodes in each floor; the node feature definition includes: attribute features such as floor height, area, shape and material properties, signal strength, user density, etc.; the edge definition includes: defining stairs and elevator shafts as edges; based on the above-defined nodes and edges, a graph structure model of the entire building is constructed; the established graph neural network GNN is used to preliminarily obtain the base station layout, including the number of building base stations and their vertical distribution.

[0053] Secondly, the reinforcement learning framework is combined with GNN; specifically, an intelligent agent is defined in the reinforcement learning framework, and the base station layout problem is defined as a reinforcement learning task of the intelligent agent. The intelligent agent is used to observe the building data, working condition data and historical base station layout of the target building, and the base station layout adjustment decision is generated according to the base station layout output by GNN, that is, adding, reducing or moving base stations; the state is updated according to the adjustment decision of the intelligent agent, and the reward function is calculated. The reward function is designed based on the signal coverage rate and the base station installation cost, such as: setting the scoring criteria of the signal coverage rate and the base station installation cost parameters and weighted calculation, until the reward function reaches the maximum value and the corresponding adjustment decision is output.

[0054] The first training set is used to train the constructed graph neural network GNN of the integrated reinforcement learning framework to obtain a trained building overall layout configuration model. In addition, the preprocessed data can be used to set the first test set, and the first training set and the first test set are combined to achieve accuracy testing and optimization training of the building overall layout configuration model.

[0055] In a specific embodiment, in order to more accurately optimize the base station layout according to the signal propagation characteristics in a complex building environment to improve the transmission performance of the deployed network, the characteristic indicators of the base station layout on a single floor are integrated, and a model for adaptively extracting the environmental characteristics and signal transmission characteristics of a single floor and optimizing the layout output is designed. The method uses a deep learning algorithm to train and construct a single-layer building layout configuration model, which specifically includes: Data preprocessing is performed on the input second building data, the historical second operating condition data, and the historical second base station layout configuration data to generate a second training set.

[0056] The hybrid neural network of CNN and LSTM is selected as the single-layer layout configuration model architecture of the building, which includes: First, a CNN network model is constructed, and the constructed CNN is used to process the spatial features of the second building data, including: for a two-dimensional grid image generated by converting the single floor layout information in the second building data, the spatial features of the key grids are identified, including: room location, wall material distribution, and obstacle location.

[0057] Secondly, an LSTM network model is constructed, and LSTM is used to process the time series data in the second operating condition data and the second base station layout configuration data, including: time series characteristics of user density and signal strength.

[0058] Thirdly, based on the design of a hybrid neural network structure by combining CNN and RNN / LSTM, a multi-objective optimization function is defined as the model loss function, such as the objective optimization function generated by weighted calculation of signal coverage, signal capacity and base station installation cost.

[0059] The constructed hybrid neural network of CNN and LSTM is trained using the second training set, and the parameters of the model are continuously adjusted during the training process so that the target optimization function reaches the optimal solution, that is, the loss function is minimized, and the trained single-layer building layout configuration model is obtained. Similarly, the second test set can also be set using the preprocessed data, and the second training set and the second test set are combined to realize the accuracy test and optimization training of the single-layer building layout configuration model.

[0060] In a specific embodiment, a user feedback mechanism is designed, and a closed-loop feedback mechanism is used to not only promptly correct existing layout defects, but also continuously accumulate new experience and knowledge, so that the layout configuration scheme determination model becomes more and more intelligent and better adapts to more complex environmental changes. The method also includes: Obtain the satisfaction of users in the target building with signal quality feedback in real time, and determine whether the proportion of current user satisfaction exceeding the preset satisfaction number to all user feedback is lower than the preset proportion. If it is lower, it indicates that the satisfaction with the current base station layout configuration is relatively high and needs to be further improved. Without making major adjustments to the base station layout, fine-tuning is performed based on the base station configuration output to expand the range of base stations around the locations where the real-time signal quality is lower than the preset quality, and dynamically adjust the configuration of base stations around the locations where the real-time signal quality is lower than the preset quality according to the AI ​​model.

[0061] Among them, the base station range around the position where the real-time signal quality is lower than the preset quality position can be determined according to the area range of a single floor of the intelligent building, such as being set to 1 / 20 of the area range of a single floor; similarly, the base station range for controlling the expansion of the surrounding area can be determined according to the area range of a single floor of the intelligent building or a predetermined surrounding range, such as being set to a half-ratio expansion of 1 / 20 of the area range of a single floor.

[0062] In addition, if the satisfaction level is still low after expanding the surrounding base stations and adjusting the surrounding base station configurations, it is necessary to optimize the model from the perspective of large base station layout, including: If after expanding the base station range around the location where the real-time signal quality is lower than the preset quality, and readjusting the base station configuration, the proportion of user satisfaction with signal quality feedback in the target building obtained over a period of time exceeds the preset satisfaction but is still lower than the preset proportion, then the incremental learning method can be used to optimize the overall building layout configuration model and / or the single-layer building layout configuration model.

[0063] In addition to continuing to acquire and collect architectural data, historical operating condition data, and historical base station layout configuration data of several buildings to increase the first training set and / or the second training set, the optimization of the overall building layout configuration model and / or the single-layer building layout configuration model also includes: Create a virtual building 3D model corresponding to a typical building structure and a signal communication scenario in the virtual building. Based on this, perform signal attenuation simulation to generate several building-related signal coverage simulation data. Divide the corresponding building structure and the signal coverage simulation data as working condition data and use them as incremental data for the first training set and the second training set to achieve incremental learning of the overall building layout configuration model and / or the single-layer building layout configuration model.

[0064] In a specific embodiment, considering that in the actual base station operation process, there may be a base station failure that causes limited communication transmission, or considering the pressure brought by the possible increase in service demand or the increase in the number of users in the future, the base station configuration is optimized in advance to avoid resource waste and technical difficulties caused by frequent adjustments in the later stage. The method further includes: collecting the building data and working condition requirements of the target building and inputting them into the building overall layout configuration model and the building single-layer layout configuration model respectively. A preset proportion of the working condition requirements of the collected target building is adjusted, and the collected architectural data of the target building and the adjusted working condition requirements are respectively input into the building overall layout configuration model and the building single-layer layout configuration model.

[0065] The preset ratio is manually set, and the preset ratio is adjusted in advance for the user density of the entire target building or a single floor, such as adjusting the user density of M1 to a user density of 1.2 M1.

[0066] In addition, in order to more accurately adjust the operating conditions of the target building, a deep learning algorithm can be used to determine the historical operating conditions based on the historical operating data of the target building, predict the operating conditions in the future, and dynamically adjust the operating conditions of the target building based on the predicted operating conditions.

[0067] like Figure 2 As shown, this embodiment discloses a system for determining a base station layout configuration scheme in an intelligent building, specifically, comprising: The building training data collection and processing module 101 is used to collect the building data, historical working condition data and historical base station layout configuration data of several buildings; statistically obtain the first building data, historical first working condition data and historical first base station layout configuration data belonging to the entire building, and the second building data, historical second working condition data and historical second base station layout configuration data belonging to a single floor of the building; The building overall layout configuration model acquisition module 102 is used to use the first building data, the historical first working condition data and the historical first base station layout configuration data as training data, and use the deep learning algorithm to train and construct the building overall layout configuration model, wherein the input of the building overall layout configuration model is the first building data and the first working condition data, and the output is the number, configuration and vertical distribution of building base stations that meet the first preset requirement; the first preset requirement includes that the reward function with the signal coverage rate and the base station installation cost as the reward factors reaches the maximum value; The building single-layer layout configuration model acquisition module 103 is used to use the second building data, the historical second operating condition data and the historical second base station layout configuration data as training data, and use the deep learning algorithm to train and construct the building single-layer layout configuration model, wherein the input of the building single-layer overall layer layout configuration model is the second building data and the second operating condition data, and the output is the number, configuration and horizontal distribution of the building single-layer base stations that meet the second preset requirements; the second preset requirements include achieving the optimal solution with a multi-objective optimization function of signal coverage, capacity and base station installation cost; The target building base station layout configuration acquisition module 104 is used to collect the building data and working condition requirements of the target building and input them into the building overall layout configuration model and the building single-layer layout configuration model respectively, so as to comprehensively acquire the target building base station layout configuration; The target building base station layout configuration adjustment module 105 is used to use the sensor array deployed in the target building to monitor the signal quality of each location in the target building in real time, and use the AI ​​model to dynamically adjust the configuration of the base stations around the location where the real-time signal quality is lower than the preset quality, so that the location where the real-time signal quality is lower than the preset quality reaches the preset signal quality at the next moment. The configuration includes carrier frequency occupancy, antenna transmission power and beam direction.

[0068] The system further comprises: The target building base station layout configuration feedback module 106 is used to obtain the satisfaction of users in the target building for signal quality feedback in real time, and determine whether the proportion of current user satisfaction exceeding the preset satisfaction number to the total number of user feedback is lower than the preset proportion. If it is lower, the range of base stations around the position where the real-time signal quality is lower than the preset quality is expanded, and the configuration of base stations around the position where the real-time signal quality is lower than the preset quality is dynamically adjusted according to the AI ​​model; it is also used to optimize the overall building layout configuration model and / or the single-layer building layout configuration model accordingly if the proportion of users in the target building who are satisfied with the signal quality feedback exceeding the preset satisfaction level obtained for a period of time is still lower than the preset proportion after expanding the range of base stations around the position where the real-time signal quality is lower than the preset quality and readjusting the base station configuration.

[0069] The embodiment of the present application also discloses a computer-readable storage medium.

[0070] Specifically, the computer-readable storage medium stores a computer program that can be loaded by a processor and executes a method for determining a base station layout configuration plan in a smart building as described above. The computer-readable storage medium includes, for example, various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0071] The embodiment of the present application also discloses a computer device.

[0072] Specifically, the computer device includes a memory and a processor, and the memory stores a computer program that can be loaded by the processor and execute the above-mentioned method for determining the layout configuration scheme of base stations in the intelligent building.

[0073] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any feature disclosed in this specification (including the abstract and drawings), unless otherwise stated, can be replaced by other equivalent or alternative features with similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

Claims

1. A method for determining a base station layout configuration scheme in an intelligent building, characterized in that: include: Collect construction data, historical operating data, and historical base station layout configuration data of several buildings; Statistically obtain first building data, historical first operating condition data, and historical first base station layout configuration data belonging to the entire building, and second building data, historical second operating condition data, and historical second base station layout configuration data belonging to a single floor of the building; The first building data, the first historical working condition data and the first historical base station layout configuration data are used as training data, and a deep learning algorithm is used to train and construct a building overall layout configuration model, wherein the input of the building overall layout configuration model is the first building data and the first working condition data, and the output is the number, configuration and vertical distribution of building base stations that meet the first preset requirement; the first preset requirement includes a reward function with signal coverage and base station installation cost as reward factors reaching the maximum value; the second building data, the second historical working condition data and the second historical base station layout configuration data are used as training data, and a deep learning algorithm is used to train and construct a building single-layer layout configuration model, wherein the input of the building single-layer overall layer layout configuration model is the second building data and the second working condition data, and the output is the number, configuration and horizontal distribution of building single-layer base stations that meet the second preset requirement; the second preset requirement includes a multi-objective optimization function with signal coverage, capacity and base station installation cost reaching the optimal solution; Collect the building data and working condition requirements of the target building and input them into the building overall layout configuration model and the building single-layer layout configuration model respectively, and comprehensively obtain the base station layout configuration of the target building; The sensor array deployed in the target building is used to monitor the signal quality of each location in the target building in real time, and the AI ​​model is used to dynamically adjust the configuration of base stations around the location where the real-time signal quality is lower than the preset quality, so that the location where the real-time signal quality is lower than the preset quality reaches the preset signal quality at the next moment. The configuration includes carrier frequency occupancy, antenna transmission power and beam direction.

2. The method for determining a layout configuration scheme of base stations in an intelligent building according to claim 1, characterized in that: The method of using a deep learning algorithm to train and construct a building overall layout configuration model specifically includes: Performing data preprocessing on the input first building data, the historical first operating condition data, and the historical first base station layout configuration data to generate a first training set; A graph neural network (GNN) integrated with a reinforcement learning framework is selected as the overall building layout configuration model architecture, specifically including: establishing a graph neural network (GNN), including: defining nodes, defining edges, and building a graph structure based on the defined nodes and variables; node definition includes: defining each floor as a node; node feature definition includes: floor height, area, shape and material properties, signal strength, and user density; edge definition includes: defining stairs and elevator shafts as edges; then combining the reinforcement learning framework with the GNN, including: defining an intelligent agent in the reinforcement learning framework, using the intelligent agent to observe the building data, working condition data, and historical base station layout of the target building, and generating a base station layout adjustment decision based on the base station layout output by the GNN; updating the state based on the adjustment decision of the intelligent agent, and calculating the reward function, which is designed based on the signal coverage rate and the base station installation cost, until the reward function reaches the maximum value and the corresponding adjustment decision is output; The first training set is used to train the graph neural network GNN of the constructed integrated reinforcement learning framework to obtain a trained building overall layout configuration model.

3. The method for determining a layout configuration scheme of base stations in an intelligent building according to claim 1, characterized in that: The method of using a deep learning algorithm to train and construct a single-layer building layout configuration model specifically includes: Performing data preprocessing on the input second building data, the historical second operating condition data, and the historical second base station layout configuration data to generate a second training set; A hybrid neural network of CNN and LSTM is selected as the architecture of the single-floor layout configuration model of the building, specifically including: constructing a CNN network model, using the constructed CNN to process the spatial features of the second building data, including: for a two-dimensional grid image generated by converting the single-floor layout information in the second building data, identifying the spatial features of key grids, including: room locations, wall material distribution, and obstacle locations; Construct an LSTM network model, and use LSTM to process the time series data in the second working condition data and the second base station layout configuration data, including: time series characteristics of user density and signal strength; Define a multi-objective optimization function, including: a target optimization function generated by weighted calculation of signal coverage, signal capacity and base station installation cost; The constructed hybrid neural network of CNN and LSTM is trained using the second training set, and the parameters of the model are continuously adjusted during the training process so that the objective optimization function reaches the optimal solution, thereby obtaining a trained single-story building layout configuration model.

4. The method for determining a layout configuration scheme of base stations in an intelligent building according to claim 1, characterized in that: Also includes: Obtain the satisfaction of users in the target building with signal quality feedback in real time, and determine whether the proportion of current user satisfaction exceeding the preset satisfaction number to the total number of user feedback is lower than the preset proportion. If so, expand the range of base stations around the location where the real-time signal quality is lower than the preset quality, and dynamically adjust the configuration of base stations around the location where the real-time signal quality is lower than the preset quality according to the AI ​​model.

5. The method for determining the layout configuration scheme of base stations in an intelligent building according to claim 4, characterized in that: Also includes: If after expanding the base station range around the location where the real-time signal quality is lower than the preset quality, and readjusting the base station configuration, the proportion of user satisfaction with signal quality feedback in the target building obtained over a period of time exceeds the preset satisfaction but is still lower than the preset proportion, then the overall building layout configuration model and / or the single-layer building layout configuration model are optimized accordingly.

6. The method for determining the layout configuration scheme of base stations in an intelligent building according to claim 5, characterized in that: The optimization of the overall building layout configuration model and / or the single-layer building layout configuration model includes: Create a virtual building signal communication scenario corresponding to a typical building structure, perform signal attenuation simulation, generate signal coverage simulation data, and use it as incremental data for the first training set and the second training set to achieve incremental learning of the building overall layout configuration model and / or the building single-layer layout configuration model.

7. The method for determining a layout configuration scheme of base stations in an intelligent building according to claim 1, characterized in that: The collecting of the target building's architectural data and operating requirements into the building's overall layout configuration model and the building's single-layer layout configuration model also includes: adjusting the operating requirements of the target building by a preset proportion, and inputting the collected target building's architectural data and the adjusted operating requirements into the building's overall layout configuration model and the building's single-layer layout configuration model.

8. A system for determining a layout configuration scheme of base stations in an intelligent building, characterized in that: include: Building training data collection and processing module, used to collect construction data, historical working condition data and historical base station layout configuration data of several buildings; Statistically obtain first building data, historical first operating condition data, and historical first base station layout configuration data belonging to the entire building, and second building data, historical second operating condition data, and historical second base station layout configuration data belonging to a single floor of the building; A building overall layout configuration model acquisition module is used to use the first building data, the historical first working condition data and the historical first base station layout configuration data as training data, and use a deep learning algorithm to train and construct a building overall layout configuration model, wherein the input of the building overall layout configuration model is the first building data and the first working condition data, and the output is the number, configuration and vertical distribution of building base stations that meet the first preset requirement; the first preset requirement includes that a reward function with signal coverage and base station installation cost as reward factors reaches a maximum value; A building single-layer layout configuration model acquisition module is used to use the second building data, the historical second operating condition data and the historical second base station layout configuration data as training data, and use a deep learning algorithm to train and construct a building single-layer layout configuration model, wherein the input of the building single-layer overall layer layout configuration model is the second building data and the second operating condition data, and the output is the number, configuration and horizontal distribution of building single-layer base stations that meet the second preset requirement; the second preset requirement includes achieving the optimal solution with a multi-objective optimization function of signal coverage, capacity and base station installation cost; The target building base station layout configuration acquisition module is used to collect the building data and working condition requirements of the target building and input them into the building overall layout configuration model and the building single-layer layout configuration model respectively, so as to comprehensively acquire the target building base station layout configuration; The target building base station layout configuration adjustment module is used to use the sensor array deployed in the target building to monitor the signal quality of each location in the target building in real time, and use the AI ​​model to dynamically adjust the configuration of the base stations around the location where the real-time signal quality is lower than the preset quality, so that the location where the real-time signal quality is lower than the preset quality reaches the preset signal quality at the next moment. The configuration includes carrier frequency occupancy, antenna transmission power and beam direction.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 7.

10. A computer device, characterized in that: The computer device comprises a memory, a processor and a program stored and executable on the memory, and the program implements the steps of the method according to any one of claims 1 to 7 when executed by the processor.

Images