Method, system and device for environmental conditioning based on base station air conditioning

Abstract

This invention discloses an environmental control method, system, and device based on base station air conditioning; it relates to the field of environmental control technology; specifically targeting the scenario of substation computer rooms, this invention improves upon the deep limitations exposed by existing computer room environmental control methods in dealing with high-density heat loads, complex internal airflow organization, and temperature and humidity coordinated control in critical infrastructure such as substations, by providing an environmental control method based on base station air conditioning. This method utilizes high-density, gridded sensors for spatial perception, enabling multi-timescale prediction and multi-objective optimization control. It achieves coordinated, predictive, and adaptive management of the temperature and humidity field of base station air conditioning within substation data centers, effectively ensuring the long-term stable operation of equipment in the computer room and significantly improving energy efficiency. This invention provides an environmental control method, system, and device adapted to base station air conditioning in substations.

Description

Technical Field

[0001] This application relates to the field of environmental control technology, and in particular to environmental control methods, systems and devices based on base station air conditioning. Background Technology

[0002] In the unique scenario of substation equipment rooms, and as the core carriers of information communication and critical business systems, the stable control of the internal environment is crucial for ensuring the long-term reliable operation of equipment. Traditional environmental control solutions often employ comfort air conditioning systems. These systems, initially designed for commercial or residential environments, rely solely on a single temperature measurement point for simple start-stop control, lacking precise humidity regulation capabilities. While this solution provided basic cooling in the early stages when equipment integration was low, its inherent limitations have become increasingly apparent as the power consumption density of data center equipment has risen dramatically. Comfort air conditioning systems cannot detect dynamic changes in temperature and humidity distribution within the equipment room. When equipment layouts are complex or external environmental disturbances occur, localized hot and cold spots can easily form, leading to chaotic airflow organization and a significant increase in spatial temperature and humidity unevenness within the equipment room.

[0003] Existing technical solutions lack the ability to perceive the grid-based environment in the spatial dimension and lack a predictive mechanism for future temperature and humidity changes in the temporal dimension. This results in a lagging and crude control process, making it difficult to cope with sudden changes in heat load caused by power load fluctuations. This seriously restricts the refined management of the data center environment and the safety of equipment operation. Summary of the Invention

[0004] In order to at least overcome the above-mentioned deficiencies in the prior art, the purpose of this application is to provide an environmental control method, system and device based on base station air conditioning to solve the above problems.

[0005] Firstly, this application provides an environmental control method based on base station air conditioning, including: Environmental sensors are deployed in a grid pattern to acquire first environmental data for a first time interval; the first environmental data is configured as: first temperature data, first humidity data, first airflow velocity data, and first device power data; Based on the first environmental data and the spatial geometric parameters of the base station equipment room, the temperature and humidity distribution in the base station equipment room is obtained based on the preset first interpolation observation model, which serves as the first temperature and humidity field. Based on the set standard environmental data and the first temperature and humidity field, and based on the preset second time series prediction model, the predicted temperature and humidity change distribution for the second time interval is obtained as the second temperature and humidity field; based on the second temperature and humidity field, and based on the preset third control model, the optimal control parameters are obtained to control the corresponding equipment to execute the corresponding instruction strategy. In response to the corresponding device completing the execution of the corresponding instruction strategy, the third environmental data at this time is obtained to obtain the difference between the third environmental data and the standard environmental data, thereby obtaining the updated third control model; The difference between the third environmental data and the standard environmental data is within the set range.

[0006] In one possible implementation, the grid-like deployment of environmental sensors to acquire first environmental data for a first time interval includes: Inside the base station equipment room, multiple sets of environmental sensors are deployed along the height, width, and depth directions at a preset grid density; Each set of environmental sensors is configured as follows: a temperature sensor for collecting the first temperature data, a humidity sensor for collecting the first humidity data, an airflow speed sensor for collecting the first airflow speed data, and a power sensor for collecting the first equipment power data, which is electrically connected to various electrical devices inside the base station equipment room. Each set of environmental sensors synchronously collects data at a preset sampling frequency. The first temperature data, first humidity data, first airflow velocity data, and first device power data collected within the first time interval are packaged together as the first environmental data for the first time interval.

[0007] In one possible implementation, the step of obtaining the temperature and humidity distribution within the base station equipment room space as a first temperature and humidity field based on the first environmental data and the spatial geometric parameters of the base station equipment room, using a preset first interpolation observation model, includes: Based on the spatial geometric parameters of the base station equipment room, a three-dimensional model of the base station equipment room is obtained. Based on the first environmental data, and using the three-dimensional coordinate system of the three-dimensional model of the base station equipment room, the first environmental data collected by each set of environmental sensors is stored in the form of a three-dimensional coordinate system to obtain discrete measurement point data of the base station equipment room. The preset first interpolation observation model is configured to take discrete measurement point data as input, and output data of any unmeasured point in space based on the Kriging algorithm. Based on the discrete measurement point data of the base station equipment room, and based on the preset first interpolation observation model, the temperature and humidity values ​​of any unmeasured points in the space of the base station equipment room are obtained to obtain the temperature and humidity distribution in the space of the base station equipment room, which serves as the first temperature and humidity field.

[0008] In one possible implementation, the step of obtaining the predicted temperature and humidity change distribution for a second time interval, based on preset standard environmental data and the first temperature and humidity field and a pre-defined second time series prediction model, as the second temperature and humidity field, includes: The set standard environmental data and the first temperature and humidity field are used as inputs to the second time series prediction model. The standard environmental data includes the target temperature range and the target humidity range. The second time series prediction model is configured as a time series prediction model based on a long short-term memory network and an autoregressive moving average model, and external influencing factors are introduced; The second time series prediction model is trained by dividing historical environmental data, power load at the corresponding time, and external weather into training set, validation set, and test set to obtain the predicted temperature and humidity change distribution in the base station equipment room space within the second time interval. The second time interval is divided into multiple time sub-intervals to accommodate power loads at different time scales; The first temperature and humidity field is used as the starting state of the second time interval, and the set standard environmental data is used as the ending state of the second time interval. The data is input into the second time series prediction model, and the predicted temperature and humidity change distribution of the computer room within the second time interval is output as the second temperature and humidity field.

[0009] In one possible implementation, the step of inputting the starting state and the ending state into a second time series prediction model and outputting the temperature and humidity variation distribution of the base station equipment room within the predicted second time interval as a second temperature and humidity field includes: The second time series prediction model is configured to include a long short-term memory network and a Gaussian process regression. Based on the Long Short-Term Memory network, the power grid load forecast for the second time interval is obtained; Set the length of the second time interval, take the first temperature and humidity field as the starting state of the second time interval, and take the set standard environmental data as the ending state of the second time interval. Based on the first temperature and humidity field and load prediction results, the predicted temperature and humidity values ​​for each preset time step at all grid points in the base station equipment room within the second time interval are obtained through Gaussian process regression. Based on the predicted temperature and humidity values ​​at each preset time step on all grid points in the base station equipment room, the temperature and humidity variation distribution in the base station equipment room space within the second time interval is obtained. The temperature and humidity distribution within the base station equipment room is used as the second temperature and humidity field.

[0010] In one possible implementation, obtaining optimal control parameters based on the second temperature and humidity field and a preset third control model to obtain a strategy for controlling the corresponding device to execute corresponding instructions includes: The second temperature and humidity field characterizes the temperature and humidity change behavior at all grid points within the base station equipment room space during the second time interval; The first temperature and humidity field and the second temperature and humidity field are used as inputs to the third control model; The third control model is configured to include a multi-objective optimization algorithm, the objectives of which include at least: maintaining the temperature and humidity of the base station equipment room within the standard range, minimizing the energy consumption of the base station air conditioning, minimizing the spatial non-uniformity of temperature and humidity, and maximizing the reliability of equipment operation. Based on the second temperature and humidity field, and through the third control model, the optimal control parameter sequence is obtained under the constraints of the equipment operating range and the temperature safety range. According to the optimal control parameter sequence, the first control parameter of the sequence is applied to the base station air conditioner as the optimal control parameter; The optimal control parameters are sent to the base station air conditioning controller, which then obtains the corresponding instruction strategy based on the optimal control parameters.

[0011] In one possible implementation, the instruction strategy includes a temperature and humidity control strategy and an airflow control strategy; The temperature and humidity control strategy includes: When the overall temperature of the base station equipment room is higher than the preset high threshold and the humidity is within the normal range or lower than the preset low threshold, the set temperature value and the percentage of the air supply fan speed of the base station air conditioner are increased, while the compressor operating frequency is reduced. When the overall temperature of the base station equipment room is higher than the preset high threshold and the humidity is higher than the preset threshold, the set temperature value of the base station air conditioner is reduced, the compressor operating frequency is increased, and the dehumidifier is started. When the overall temperature of the base station room is at a preset low threshold and the humidity is within the normal range or higher than the preset high threshold, the first set temperature value of the base station air conditioner is increased, the operating frequency of the first compressor is reduced, the speed percentage of the first air supply fan is increased, and the operating power percentage of the first dehumidifier is activated. When the overall temperature of the base station equipment room is within the normal range and the humidity is higher than the preset high threshold, the set temperature value is maintained, but the dehumidifier working power percentage is increased, and the air supply fan speed percentage is adjusted. When the overall temperature of the base station equipment room is within the normal range and the humidity is low, maintain the first set temperature value, but increase the percentage of humidifier working power and adjust the percentage of air supply fan speed. The airflow control strategy includes: Based on the first temperature and humidity field and the third control model, local hot spots and cold spots are obtained; simultaneously, based on the second temperature and humidity field and the third control model, predicted future local hot spots are obtained. The hot spot area includes areas where the heat load is higher than a set threshold. Based on the local hot spot area, cold spot area and future local hot spot area, the angle of the air outlet guide vanes of the base station air conditioner is adjusted to guide the cold airflow to the local hot spot area, while avoiding cold air short-circuiting and local cold spot area.

[0012] In one possible implementation, the step of responding to the corresponding device completing the execution of the corresponding instruction strategy, acquiring the third environmental data at this time, and obtaining the difference between the third environmental data and the standard environmental data to obtain the updated third control model includes: In response to the corresponding device completing the execution of the corresponding instruction strategy, third environmental data is acquired based on the environmental sensor; the third environmental data includes third temperature data, third humidity data, third airflow speed data, and third device power data; Based on the third environmental data and the first interpolation observation model, the third temperature and humidity field is obtained; Based on the third temperature and humidity field, the deviations between the temperature and humidity values ​​of each calculation unit and the standard environmental data are calculated to obtain a difference matrix. Based on the difference matrix, the parameters are corrected as the feedback error of the third control model to obtain the updated third control model.

[0013] Secondly, this application provides an environmental control system based on base station air conditioning, including a data acquisition unit, an environmental observation unit, an environmental prediction unit, a control unit, and an optimization and updating unit connected in sequence. The data acquisition unit is configured to deploy environmental sensors in a grid pattern to acquire first environmental data for a first time interval; The environmental observation unit is configured to: obtain the temperature and humidity distribution in the base station equipment room based on the first environmental data and the spatial geometric parameters of the base station equipment room, and use it as the first temperature and humidity field based on a preset first interpolation observation model; The environmental prediction unit is configured to: obtain the predicted temperature and humidity change distribution for a second time interval based on the set standard environmental data and the first temperature and humidity field, using a preset second time series prediction model as the second temperature and humidity field; The control unit is configured to: obtain optimal control parameters based on the second temperature and humidity field and a preset third control model, so as to control the corresponding device to execute the corresponding instruction strategy; The optimization and update unit is configured to: in response to the corresponding device completing the execution of the corresponding instruction strategy, obtain the third environmental data at this time, obtain the difference between the third environmental data and the standard environmental data, and thus obtain the updated third control model.

[0014] Thirdly, this application provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of any one of the environmental control methods based on base station air conditioning.

[0015] In summary, the beneficial effects that this application can achieve are: This application proposes a grid-deployed environmental sensor and Kriging interpolation modeling method. Through three-dimensional full-coverage sampling and spatial optimal estimation, it solves the problem of local blind spots in traditional sampling, achieving refined perception of the temperature and humidity field in the computer room. The proposed combination of Long Short-Term Memory (LSTM) networks and Gaussian process regression to construct a predictive model addresses the lack of forward-looking control basis through multi-timescale heat load and temperature and humidity prediction, enabling the prediction of temperature and humidity exceeding limits. The proposed multi-objective optimization control and closed-loop feedback mechanism, through coordinated control of temperature, humidity, and airflow and adaptive model updates, solves the problems of poor coordinated control and high energy consumption, achieving multi-objective optimal control. The methods in this application demonstrate significant technological advancements and beneficial effects in improving control accuracy, energy efficiency, and equipment stability. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the method steps in an embodiment of this application; Figure 2 This is a schematic diagram of the method flow of an embodiment of this application; Figure 3 This is a schematic diagram of the system structure according to an embodiment of this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0018] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0019] Step S1: Deploy environmental sensors in a grid pattern to acquire first environmental data for a first time interval; the first environmental data is configured as: first temperature data, first humidity data, first airflow velocity data, and first device power data; Step S2: Based on the first environmental data and the spatial geometric parameters of the base station equipment room, and based on the preset first interpolation observation model, obtain the temperature and humidity distribution in the base station equipment room space as the first temperature and humidity field. Step S3: Based on the set standard environmental data and the first temperature and humidity field, and based on the preset second time series prediction model, obtain the predicted temperature and humidity change distribution for the second time interval, which serves as the second temperature and humidity field. This time series prediction is based on days and years, so that different strategies can be adopted according to different daily and seasonal electricity consumption. If the power grid electricity consumption surges, the computational load of the base station will surge, and it is even more important to ensure temperature and humidity. Step S4: Based on the predicted second temperature and humidity field and the preset third control model, obtain the optimal control parameters to control the corresponding equipment to execute the corresponding instruction strategy. Step S5: In response to the corresponding device completing the execution of the corresponding instruction strategy, obtain the third environmental data at this time, and obtain the difference between the third environmental data and the standard environmental data, thereby obtaining the updated third control model; Step S6, repeat steps S1-S5 until the difference between the third environmental data and the standard environmental data is within the set range.

[0020] In implementing this application, considering the deep limitations of existing data center environmental control methods in addressing the challenges of high-density heat loads, complex internal airflow organization, dynamic external environmental disturbances, and coordinated temperature and humidity control in critical infrastructure such as substations, the method is improved to provide an environmental control method based on base station air conditioning. The method is characterized by using high-density, gridded sensors to construct spatial perception capabilities, enabling multi-timescale prediction and multi-objective optimization control. This achieves refined, coordinated, predictive, and adaptive management of the temperature and humidity field inside the substation data center, effectively ensuring the long-term stable operation of data center equipment and significantly improving energy utilization efficiency.

[0021] In this embodiment, environmental sensors are deployed in a grid pattern to comprehensively collect temperature, humidity, airflow velocity, and equipment power data inside the base station equipment room, forming first environmental data for a first time interval. This data not only reflects the heat source distribution characteristics at different locations within the equipment room but also correlates with the equipment operating load, providing a high-resolution data foundation for subsequent analysis. Then, combined with the spatial geometric parameters of the base station equipment room, a preset first interpolation observation model is used to transform the discrete sensor data into a continuous spatial temperature and humidity distribution, generating a first temperature and humidity field.

[0022] Based on the first temperature and humidity field and the established standard environmental data, a second temperature and humidity field is generated by extrapolating the temperature and humidity variation distribution within the second time interval through a pre-set second time series prediction model. This prediction model uses the first temperature and humidity field as the starting state and the standard environmental data as the ending state, closely coupling the actual heat load dynamics with the target constraints, and can identify potential temperature and humidity drift trends in advance. Therefore, it possesses forward-looking control capabilities, avoiding the lag of traditional reactive control, and exhibits higher adaptability, especially when temperature and humidity changes are caused by power load fluctuations.

[0023] Based on the second temperature and humidity field, the optimal control parameters are generated using a preset third control model. This model comprehensively considers multiple objectives such as temperature and humidity control accuracy, energy consumption minimization, spatial uniformity, and equipment operational reliability. It analyzes the predicted temperature and humidity change distribution and dynamically generates command strategies. For example, by adjusting the angle of the air conditioner's air outlet guide vanes, the cold airflow is directed to the predicted hot spot area, while avoiding cold air short-circuiting and the formation of local cold spots, thereby effectively eliminating local overheating and reducing energy waste. Then, after the equipment executes the command strategy, the third environmental data is acquired and compared with the standard environmental data to calculate the difference matrix. This difference matrix is ​​then used as feedback error input to the third control model to correct the model parameters and achieve adaptive optimization of the system.

[0024] Through the aforementioned closed-loop mechanism, continuous optimization and control can be achieved during long-term operation, ensuring that temperature and humidity remain within safe threshold ranges, thereby improving the reliability of equipment operation.

[0025] This technical solution integrates prediction and closed-loop optimization mechanisms to achieve coordinated, predictive, and adaptive management of temperature and humidity fields by base station air conditioning in substation data centers. This effectively ensures the long-term stable operation of equipment in the data center and significantly improves energy efficiency. It can effectively overcome the problems of temperature and humidity imbalance, local hotspots, and high energy consumption caused by traditional methods that rely on single measurement points and lack spatial dimension analysis and time prediction capabilities. It realizes coordinated temperature and humidity control and dynamic regulation of airflow organization.

[0026] Example 2 Based on Example 1, please refer to the following: Figure 2 The above is a flowchart illustrating the environmental control method based on base station air conditioning provided in an embodiment of the present invention. Further, the environmental control method based on base station air conditioning may specifically include the following contents.

[0027] This invention provides an environmental control method based on base station air conditioning, which aims to overcome the inherent limitations of existing data center environmental control schemes in dealing with high-density heat loads and coordinating temperature and humidity control. By constructing a method of spatial perception, multi-time scale prediction and dynamic multi-objective optimization control, it achieves coordinated and predictive management of the temperature and humidity field of the substation data center, thereby ensuring long-term stable operation of the equipment and significantly improving energy utilization efficiency.

[0028] In step S1, environmental sensors are deployed in a grid pattern to acquire first environmental data for a first time interval; the first environmental data is configured as first temperature data, first humidity data, first airflow velocity data, and first device power data.

[0029] Environmental sensors are deployed in a grid pattern to acquire initial environmental data for a first time interval, including: Inside the base station equipment room, multiple sets of environmental sensors are deployed along the height, width, and depth directions at a preset grid density; Each set of environmental sensors is configured as follows: a temperature sensor for collecting first temperature data, a humidity sensor for collecting first humidity data, an airflow velocity sensor for collecting first airflow velocity data, and a power sensor that is electrically connected to various electrical devices inside the base station equipment room for collecting first device power data. Each set of environmental sensors synchronously collects data at a preset sampling frequency. The first temperature data, first humidity data, first airflow velocity data, and first device power data collected within the first time interval are packaged together as the first environmental data for the first time interval.

[0030] In the implementation of this application embodiment, multiple sets of environmental sensors are deployed inside the base station equipment room at a preset grid density along the height, width, and depth directions. During deployment, the grid density can be flexibly configured according to the actual size of the equipment room, equipment layout, and the required precision of the temperature and humidity field. It is typically set to a density sufficient to capture the temperature and humidity gradient near key heat or cold sources. For example, in this embodiment, a specific equipment room is a standard base station equipment room with a side length of 8 meters, a width of 6 meters, and a height of 3 meters. Based on the actual size of this equipment room, sensor points spaced 1 meter apart can be deployed, dividing its internal space into cubic grid units with a side length of 1.0 meter, forming a total of 9×7×3=189 grid vertices, creating a three-dimensional sensor network. A set of environmental sensors is installed at each vertex, totaling 189 sets, uniformly covering the entire equipment room space.

[0031] Each set of environmental sensors integrates a first temperature sensor for collecting first temperature data, a first humidity sensor for collecting first humidity data, and a first airflow velocity sensor for collecting first airflow velocity data. Power sensors that are electrically connected to various electrical devices inside the base station equipment room, such as servers, network switches, and power modules, are also deployed to collect first device power data in real time.

[0032] Then, data can be collected at a set time period. If synchronous sampling is performed at a period of 10 seconds, the temperature, humidity, airflow speed and associated device power values ​​of all 189 nodes at the same time are packaged into a data frame, forming a snapshot of the environmental state that is updated every 10 seconds but is discrete according to the time interval, thus constituting the first environmental dataset for 24 consecutive hours within the first time interval.

[0033] By deploying sensors in a high-density, gridded manner, this invention can acquire detailed environmental data from different locations within the computer room, effectively overcoming the problem of blind spots in local area perception caused by traditional single-point or a few-point sampling methods, and providing sufficient and reliable raw data support for subsequent temperature and humidity field modeling.

[0034] Step S2: Based on the first environmental data and the spatial geometric parameters of the base station equipment room, and based on the preset first interpolation observation model, obtain the temperature and humidity distribution in the base station equipment room space as the first temperature and humidity field.

[0035] Based on the first environmental data and the spatial geometric parameters of the base station equipment room, and based on a preset first interpolation observation model, the temperature and humidity distribution within the base station equipment room is obtained as the first temperature and humidity field, including: Based on the spatial geometric parameters of the base station equipment room, a three-dimensional model of the base station equipment room is obtained. Based on the first environmental data, and using the three-dimensional coordinate system of the three-dimensional model of the base station equipment room, the first environmental data collected by each set of environmental sensors is stored in the form of a three-dimensional coordinate system to obtain discrete measurement point data of the base station equipment room. The preset first interpolation observation model is configured to take discrete measurement point data as input, and output data of any unmeasured point in space based on the Kriging algorithm; Based on the discrete measurement point data of the base station equipment room, and based on the preset first interpolation observation model, the temperature and humidity values ​​of any unmeasured point in the base station equipment room are obtained to obtain the temperature and humidity distribution in the base station equipment room, which serves as the first temperature and humidity field.

[0036] In the implementation of this application embodiment, after acquiring the first environmental data within the first time interval, the collected data is used to construct the environmental field.

[0037] The system receives the first environmental data collected within the first time interval and simultaneously acquires spatial geometric parameters such as the physical dimensions of the base station equipment room, cabinet layout, and equipment placement. These parameters are stored in a three-dimensional coordinate system, serving as the boundary conditions for spatial interpolation. The origin of the coordinate system is then set at the lower left front corner of the equipment room, where the x-axis of the equipment room is along its width. The y-axis is along the depth direction The z-axis is vertically upward. .

[0038] Then, the actual installation location of each group of environmental sensors is determined by its position in that coordinate system. Coordinate representation, then measurement points Measurement points coordinates and the corresponding first temperature data collected And the first humidity data Binding to this coordinate point forms a set of 189 discrete measurement points, which can be represented as follows: At this time , The discrete set of point data measured by sensors is used as input and substituted into the preset first interpolation observation model. This model uses the Kriging algorithm to achieve optimal unbiased estimation through spatial autocorrelation modeling, and estimates the temperature and humidity values ​​of any unmeasured point in the computer room space to reconstruct the continuous temperature and humidity distribution inside the computer room.

[0039] First, calculate the Euclidean distance between all points, then measure the points. and measurement points The Euclidean distance can be expressed as: The distance is divided into zones based on distance intervals, with each interval being 0.5 meters. In this case, r is 0.5. The mean square of the temperature difference within each zone is calculated to generate an empirical semivariogram value. This embodiment uses a spherical model of the variogram to fit the semivariogram, which can be expressed as: ; in, It's the nugget effect, and then This is the partial sill value, characterizing the spatial correlation strength, where r is the maximum influence distance of spatial autocorrelation. It is the maximum value of the semi-mutation function. Then, by fitting the historical data using the weighted least squares method, the parameters can be directly determined. , , .

[0040] Then, record the data of all estimated arbitrary unmeasured points within any space as... Then, based on the spherical model, the Kriging algorithm is used to establish a set of equations for the weights of the Kriging algorithm, which can be expressed as: ; First, use the known measurement data to determine the weighting coefficients of this system of equations. and Lagrange multipliers The temperature can be calculated directly, and then the estimated temperature values ​​for all points can be expressed as: ; in, Indicates any unmeasured point Its coordinates are The estimated temperature at this location, which corresponds to the center of any computing unit within the computer room. Indicates at known sampling points The temperature measurements at the location are directly derived from real-time data from the first temperature sensor and the first humidity sensor. Representation and measurement value The corresponding interpolation weighting coefficients, the values ​​of which reflect the sampling points For the estimated point The degree of influence, where N represents the number of known sampling points used for estimation, and then this is 189.

[0041] Then, using the same method, the humidity was estimated. All estimated values.

[0042] Kriging interpolation is performed point by point to construct a first temperature and humidity field, which includes a temperature field matrix and a humidity field matrix, based on the estimated humidity and temperature values ​​at each point. This temperature and humidity field is updated based on the collected data.

[0043] In step S3, based on the set standard environmental data and the first temperature and humidity field, and based on the preset second time series prediction model, the predicted temperature and humidity change distribution for the second time interval is obtained as the second temperature and humidity field.

[0044] Based on the established standard environmental data and the first temperature and humidity field, and using a pre-set second time series prediction model, the predicted temperature and humidity change distribution for the second time interval is obtained, serving as the second temperature and humidity field, including: The set standard environmental data and the first temperature and humidity field are used as inputs to the second time series prediction model. The standard environmental data includes the target temperature range and the target humidity range. The second time series forecasting model was configured as a time series forecasting model based on a long short-term memory network and an autoregressive moving average model, and external influencing factors were introduced. The second time series prediction model is trained by dividing historical environmental data, power load at the corresponding time, and external weather into training set, validation set, and test set to obtain the distribution of temperature and humidity changes in the base station equipment room space within the predicted second time interval. The second time interval is divided into multiple time sub-intervals to accommodate power loads at different time scales; The first temperature and humidity field is used as the starting state of the second time interval, and the set standard environmental data is used as the ending state of the second time interval. The data is input into the second time series prediction model, and the predicted temperature and humidity change distribution of the computer room within the second time interval is output as the second temperature and humidity field.

[0045] The starting and ending states are input into the second time series prediction model, which outputs the temperature and humidity variation distribution of the base station equipment room within the predicted second time interval, as the second temperature and humidity field, including: The second time series prediction model was configured to include a long short-term memory network and a Gaussian process regression. Based on the Long Short-Term Memory (LSTM) network, the power grid load forecast for the second time interval is obtained, thereby obtaining the forecast of the total power consumption of the equipment. Since the higher the total power consumption of the equipment, the greater the heat generation at the corresponding location, the LSTM network can make full use of historical equipment power data and power grid load forecast data to identify the correlation between power system load fluctuations and base station computing power, thereby predicting the future heat load of the computer room. Then, based on the first temperature and humidity field and load prediction results, the length of the second time interval is set. The first temperature and humidity field is used as the starting state of the second time interval, and the set standard environmental data is used as the ending state of the second time interval. A regression model is established through Gaussian process regression to predict the changes in temperature and humidity. The predicted temperature and humidity values ​​at each preset time step are output at all grid points in the base station equipment room within the second time interval. Based on the predicted temperature and humidity values ​​at each preset time step on all grid points in the base station equipment room, the temperature and humidity variation distribution in the base station equipment room space within the second time interval is obtained. The temperature and humidity distribution within the base station equipment room is used as a second temperature and humidity field.

[0046] In the implementation of this application embodiment, a second time interval is set by combining the set standard environmental data and the first temperature and humidity field. The first temperature and humidity field is used as the starting state of the second time interval, and the set standard environmental data is used as the ending state of the second time interval. Then, the starting point, the ending point, and the duration are all input into the second time series prediction model, and the predicted temperature and humidity change distribution of the computer room within the second time interval is output as the second temperature and humidity field.

[0047] After obtaining the first temperature and humidity field, a pre-set prediction model is used to predict future environmental trends and obtain a predicted second temperature and humidity field. Then, the temperature and humidity in the actual computer room are adjusted according to the predicted environment so that the actual temperature and humidity in the computer room remain within the set standard environmental data and change according to the predicted second temperature and humidity field.

[0048] The standard environmental data and the first temperature and humidity field are used as inputs. The standard environmental data consists of the target temperature and target humidity.

[0049] The second time series forecasting model comprises a long short-term memory network and an autoregressive moving average model, incorporating external influencing factors. This allows it to comprehensively capture changes in the temperature and humidity field of the computer room. By using historical data, current environmental conditions, and future external environments (including weather forecasts for the next 24 to 72 hours and pre-predicted power grid load data), it forecasts the temperature and humidity distribution of each computing unit within the computer room space within the second time interval, generating a second temperature and humidity field. Furthermore, this second time interval can be flexibly divided into multiple sub-intervals, such as daily forecasts on an hourly basis and weekly or monthly forecasts on a daily basis, to adapt to environmental changes at different time scales and the periodic changes in internal heat load caused by varying power loads over different time periods. The length of the sub-intervals can be adjusted according to specific needs.

[0050] The prediction process of the second time series prediction model can be divided into two levels: the first level predicts the power load, and the second level predicts the dynamic temperature and humidity field inside the computer room.

[0051] In the first level, load forecasting is performed to predict the internal heat load. Because fluctuations in grid load lead to fluctuations in the corresponding total power consumption of equipment, the internal heat load exhibits a periodic pattern. Therefore, a prediction based on a Long Short-Term Memory (LSTM) network is used. The LSM network introduces a gating mechanism with input gates, forget gates, and output gates, enabling it to effectively learn and remember long-term dependencies. The LSM network can be represented as: Input gate ; Forgotten Gate ; Candidate cell state ; Cell state update ; Output gate ; Hidden state ; in, The input vector at the current moment consists of historical equipment power data, power grid load forecast data, and relevant influencing factors such as changes in the external environment. These are the activation vectors for the input gate, forget gate, and output gate, respectively, which control the inflow, outflow, and retention of information in the cell state. The current state of the cell represents the network's long-term memory. This represents the cell state at the previous moment; This represents the current hidden state, which is also the output. This is the hidden state from the previous moment; b and b are the weight matrix and bias vector, respectively, which are the parameters learned by the model during training. The Sigmoid activation function is used to map values ​​between 0 and 1, acting as a gating mechanism. This is the hyperbolic tangent activation function, used to map numerical values ​​to the range of -1 to 1; ⊙ represents element-wise multiplication.

[0052] This long short-term memory network can make full use of historical equipment power data and power grid load forecast data to identify the correlation between power system load fluctuations and base station computing load, thereby predicting the future heat load of the computer room, significantly improving the accuracy and real-time performance of the prediction, and is especially suitable for scenarios where a surge in power system load leads to a sudden change in base station computing load.

[0053] The second level is the prediction of the temperature and humidity field inside the computer room. Taking the first temperature and humidity field as the starting state and the standard environmental data as the ending state, combined with the prediction of the total power consumption of the equipment output by the long short-term memory network, a Gaussian process regression model is established to predict the changes in temperature and humidity.

[0054] Gaussian process regression models learn the input-output relationship from historical data, enabling them to capture complex nonlinear mappings and provide predictions, including the predicted mean. and variance The mathematical expression is as follows: ; ; Where X is the training data input matrix, which includes historical temperature and humidity field distribution, external environmental parameters and equipment power; y is the output vector of the training data, containing the future evolution of the historical temperature and humidity field; Input for the point to be predicted; For the training data covariance matrix, its elements , representing training data points and Similarity between them; The covariance vector generated between the predicted points and the training data; The covariance of the prediction point itself; This is the noise covariance matrix, used to capture measurement errors and model uncertainties; The kernel function is used to measure the similarity between two input points. A squared exponential kernel function is directly used to describe a smooth and infinitely differentiable functional relationship, which is suitable for the continuous characteristics of temperature and humidity fields. Squared exponential kernel function: ; in, It is the signal variance, which reflects the overall magnitude of the function's change; The feature length determines the smoothness of the function's variation; that is, how far apart points in the input space need to be for them to differ significantly in the output space. These hyperparameters... and It is calculated directly from the training data by maximizing the log-likelihood estimate.

[0055] This hybrid prediction model combines long short-term memory networks for power load prediction with Gaussian process regression for modeling the internal temperature and humidity field. This enables prediction of future temperature and humidity distribution, considering not only the heat load and air conditioning system response within the computer room but also the impact of power grid load fluctuations on the computer room's computational load and heat generation. By combining daily and yearly prediction capabilities, this invention provides a forward-looking control basis for base station air conditioning systems, enabling preventative control and effectively preventing temperature and humidity exceedances. The second temperature and humidity field is also a three-dimensional matrix data structure, where each element corresponds to a calculation unit within the computer room and predicts the temperature and humidity for a second future time interval.

[0056] Step S4: Based on the predicted second temperature and humidity field and the preset third control model, obtain the optimal control parameters to control the corresponding equipment to execute the corresponding instruction strategy.

[0057] Based on the predicted second temperature and humidity field and a preset third control model, optimal control parameters are obtained to control the corresponding equipment to execute corresponding command strategies, including: The second temperature and humidity field characterizes the temperature and humidity change behavior of all grid points in the base station equipment room space within the second time interval. The first and second temperature and humidity fields are used as inputs to the third control model; The third control model is configured to include a multi-objective optimization algorithm, the objectives of which include at least: maintaining the temperature and humidity of the base station equipment room within the standard range, minimizing the energy consumption of the base station air conditioning, minimizing the spatial non-uniformity of temperature and humidity, and maximizing the reliability of equipment operation. The behavior predicted by the second time series prediction model over a period of time is the second temperature and humidity field. The optimal control parameter sequence is obtained through the third control model under the constraints of the equipment operating range and the temperature safety range. Based on the optimal control parameter sequence, the first control parameter of the sequence is applied to the base station air conditioner as the optimal control parameter. The optimal control parameters are sent to the base station air conditioning controller, which then obtains the corresponding instruction strategy based on the optimal control parameters.

[0058] The command strategies include temperature and humidity control strategies and airflow control strategies; Temperature and humidity control strategies include: When the overall temperature of the base station equipment room is higher than the preset high threshold and the humidity is within the normal range or lower than the preset low threshold, the set temperature value and the percentage of the air supply fan speed of the base station air conditioner are increased, while the compressor operating frequency is reduced. When the overall temperature of the base station equipment room is higher than the preset high threshold and the humidity is higher than the preset threshold, the set temperature value of the base station air conditioner is reduced, the compressor operating frequency is increased, and the dehumidifier is started. When the overall temperature of the base station equipment room is preset to a low threshold and the humidity is within the normal range or higher than the preset high threshold, the first set temperature value of the base station air conditioner is increased, the operating frequency of the first compressor is reduced, the speed percentage of the first air supply fan is increased, and the operating power percentage of the first dehumidifier is started. When the overall temperature of the base station equipment room is within the normal range and the humidity is higher than the preset high threshold, maintain the set temperature value, but increase the dehumidifier's working power percentage and adjust the air supply fan speed percentage. When the overall temperature of the base station equipment room is within the normal range and the humidity is low, maintain the first set temperature value, but increase the percentage of humidifier working power and adjust the percentage of air supply fan speed. Airflow control strategies include: Based on the first temperature and humidity field and the third control model, the identified local hot spots and cold spots are obtained; at the same time, based on the second temperature and humidity field and the third control model, the predicted future local hot spots are obtained. Hotspot areas are those where the heat load exceeds a set threshold. Based on local hot spots, cold spots, and future local hot spots, adjust the angle of multiple air outlet guide vanes of the base station air conditioner to guide the cold airflow to local hot spots, while avoiding cold air short-circuiting and local cold spots.

[0059] In the implementation of this application embodiment, the first temperature and humidity field and the second temperature and humidity field are used as inputs to the third control model. Then, the third control model finds the gap between the predicted second temperature and humidity field and the real-time first temperature and humidity field, performs multi-objective function calculation, and obtains the optimal control sequence. Only the first control action of the sequence is applied to the base station air conditioner to obtain the control parameter.

[0060] Then, let's denote this objective function as J. It's designed to consider multiple interrelated and potentially conflicting control objectives, including but not limited to: maintaining the temperature and humidity of the computer room within standard ranges, minimizing the energy consumption of the base station air conditioning system, and minimizing spatial non-uniformity of temperature and humidity. The goal is to achieve comprehensive optimal control of the computer room environment, so the objective function is set to find its minimum value. The objective function can be expressed as: ; in, and To predict the start and end times in the time domain, for example, it can be set to the next 1 to 24 hours to provide a sufficient time window for forward optimization; To control the step size in the time domain, for example, it can be set to 15 to 30 minutes in the future, representing the time interval for each generated control sequence to be obtained by performing multi-objective function calculations; and These are the predicted temperature and humidity of each computing unit in the computer room at a prediction time step of k, respectively. These data are directly derived from the second temperature and humidity field. and These are the set standard temperature and standard humidity, respectively; To predict the base station air conditioning energy consumption at a time step k, this can be directly estimated based on parameters such as the actual air conditioning cooling capacity, air volume, and compressor power. It is an indicator of spatial non-uniformity of temperature and humidity. For example, it can be defined as the difference between the highest and lowest temperatures and the difference between the highest and lowest humidity in each computing unit inside the computer room. The squared term of the base station air conditioning control parameters is used to penalize drastic changes in control actions when the control time is k steps, thereby avoiding oscillations during air conditioning adjustment and smoothly adjusting temperature and humidity. These are the weighting coefficients for each objective function. These coefficients are adjustable parameters used to balance the importance of different control objectives. For example, when energy conservation is emphasized, they can be appropriately increased. The weight of temperature and humidity uniformity is increased when the emphasis is on uniformity. The weights are determined by taking into account factors such as the operational priority of the computer room, equipment characteristics, and environmental requirements.

[0061] Control parameters Adjustable parameters include the first set temperature value, the first set humidity value, the percentage of the first air supply fan speed, the first compressor operating frequency, the percentage of the first humidifier operating power, the percentage of the first dehumidifier operating power, and the angle of the first air outlet guide vanes. These parameters directly cover the main operation and airflow regulation capabilities of the air conditioner.

[0062] The first set temperature value is the target temperature of the temperature control sensor inside the base station air conditioner, which directly affects the cooling / heating capacity. The first set humidity value is the target humidity of the humidity control sensor inside the base station air conditioner, which directly affects the humidification / dehumidification capacity. The percentage of the first air supply fan speed directly affects the air supply volume and the overall airflow organization inside the machine room, and is usually adjustable within the range of 0% to 100%. The operating frequency of the first compressor affects the cooling capacity output, and it is usually adjustable in the range of 20Hz to 120Hz or higher. The first percentage of the humidifier's operating power affects the humidification output and is suitable for scenarios with excessively low humidity. The adjustable range is 0% to 100%. The first dehumidifier's operating power percentage affects the dehumidification capacity and is suitable for scenarios with excessively high humidity; the adjustable range is 0% to 100%. The angle of the first air outlet guide vane is usually precisely controlled by a stepper motor, which affects the local airflow direction and coverage area, and is crucial for solving local hot or cold spot problems.

[0063] The third control model must comply with equipment operating range constraints, temperature and humidity safety constraints, and response constraints of the equipment used during the calculation process to ensure the safe and stable operation of the system.

[0064] Equipment operating range constraints include ensuring that all control parameters of the base station air conditioner are within the safe operating range specified by the manufacturer. For example, the fan speed should be limited to 50-90%, and the compressor frequency should be limited to the minimum and maximum frequencies recommended by the manufacturer to avoid equipment damage or inefficiency. Temperature and humidity safety constraints include ensuring that the temperature and humidity of all computing units inside the computer room are always kept within the critical threshold for safe operation of the equipment, even if they may deviate from standard environmental data in the short term. For example, the temperature is limited to 15-40°C and the humidity is limited to 10%-50%RH. This is a fundamental requirement to ensure the reliability of the equipment. Response constraints, including limiting the rate of change of control parameters, should be implemented to avoid excessive stress on the system or frequent start-stops. For example, the fan speed should not change by more than 10% per minute, and the compressor frequency should not change by more than 5 Hz per minute. These limits help extend equipment life and reduce energy consumption.

[0065] The third control model uses numerical optimization algorithms, such as interior-point methods and sequential quadratic programming, to solve the aforementioned multi-objective optimization problem with constraints, and then obtains the optimal control sequence in the control time domain. Only the first control action of the sequence The optimal control parameters are sent to the base station air conditioning controller.

[0066] After receiving the optimal control parameters, the base station air conditioning controller converts them into specific instruction strategies to control the base station air conditioning and its auxiliary equipment to perform corresponding operations.

[0067] The command strategy includes a temperature and humidity coordinated control strategy: When the overall temperature in the computer room is high, while the humidity is normal or low: the command strategy will coordinate to increase the first set temperature value of the base station air conditioner, for example, by slightly adjusting it from 22°C to 23°C. Simultaneously, it will increase the percentage of the first air supply fan speed, for example, from 70% to 85%, to increase airflow and improve the overall heat exchange efficiency of the computer room. In this case, to avoid excessive cooling leading to further humidity reduction, the operating frequency of the first compressor will be appropriately reduced. This strategy aims to remove heat by increasing airflow circulation, rather than simply relying on powerful cooling.

[0068] When the overall temperature and humidity in the server room are both high, this presents a complex situation requiring simultaneous cooling and dehumidification. The command strategy will lower the initial set temperature of the base station air conditioner, for example, from 22°C to 21°C, and significantly increase the operating frequency of the first compressor to enhance cooling capacity. Simultaneously, to prevent condensation and further improve dehumidification efficiency, the operating power percentage of the first dehumidifier will be activated or increased, for example, from 0% to 60%. The fan speed can then be adjusted according to actual conditions, allowing for fine-tuning based on airflow organization requirements.

[0069] When the overall temperature of the computer room is low, and the humidity is normal or high: the command strategy will increase the first set temperature value of the base station air conditioner, for example, from 22°C to 24°C, and reduce the operating frequency of the first compressor or stop it to reduce the cooling capacity. Simultaneously, to increase the overall temperature of the computer room and promote even heat distribution, the percentage of the first air supply fan speed will be increased, for example, from 60% to 75%. If the humidity is high, the percentage of the first dehumidifier's operating power will be activated or increased.

[0070] When the overall temperature of the computer room is normal, but the humidity is high: the command strategy will maintain the first set temperature value, but will increase the percentage of the first dehumidifier's operating power, for example, from 30% to 70%, to effectively remove moisture. At the same time, to further enhance dehumidification, the percentage of the first supply fan speed may be fine-tuned to optimize the airflow speed through the evaporator.

[0071] When the overall temperature of the computer room is normal, but the humidity is low: the command strategy will maintain the first set temperature value, but will increase the percentage of the first humidifier's operating power, for example, from 0% to 50%, to increase the humidity in the computer room. Similarly, the percentage of the first air supply fan speed will be adjusted to promote the uniform diffusion of moisture.

[0072] These instruction strategies achieve precise and coordinated control of temperature and humidity inside the computer room through the coordinated action of multiple control parameters, effectively avoiding the problem of mutual interference and neglect of one over the other in traditional solutions.

[0073] In this embodiment, in order to make the environmental control efficiency of the base station air conditioner higher, not only the temperature is controlled, but also the coordinated airflow is considered to achieve regulation. At this time, the effect of one plus one is greater than two. Therefore, the design command strategy also includes an airflow organization optimization strategy.

[0074] The third control model adjusts the angles of multiple first air outlet guide vanes of the base station air conditioner based on the local hot and cold spots identified in real time by the first temperature and humidity field and the future local hot spots predicted by the second temperature and humidity field. For example, when a hot spot is detected in front of a cabinet, the air outlet vanes above it can be instructed to deflect in that direction to guide the cold airflow to be accurately delivered to the area with higher heat load. At the same time, reverse adjustment avoids the occurrence of cold air short-circuiting, that is, the cold air returns directly to the air conditioner return air vent without fully absorbing heat. It also avoids the formation of local cold spots, i.e., over-cooled areas.

[0075] By finely adjusting the percentage of the first air supply fan speed, the overall airflow pattern inside the computer room is optimized while ensuring heat exchange requirements. For example, the fan speed is reduced at low loads to save energy, and the fan speed is increased at high loads to enhance circulation. At the same time, the return flow area and stagnation area, such as the back of the cabinet or corners, are reduced to improve the utilization efficiency of cooling capacity and ensure that cool air can effectively reach all equipment that needs cooling.

[0076] When local hotspots persist and cannot be effectively eliminated by adjusting the base station air conditioning, including air volume and air outlet angle, the command strategy can further coordinate the auxiliary fans configured in the equipment room. If there are multiple fans, they are classified according to their location. Then, any fan located in a non-independent computing unit is designated as an auxiliary fan. The speed of the auxiliary fan is then started or increased to enhance local airflow circulation, break up local heat accumulation, assist in heat dissipation, and thus maintain the stability of the overall environment.

[0077] Then, the third control model combines real-time measurement data with future prediction information to achieve forward-looking control of the base station air conditioning. Compared with traditional reactive control, the third control model can predict environmental change trends and take control measures in advance to effectively avoid exceeding temperature and humidity limits. At the same time, it finds the optimal balance point among multiple conflicting objectives, such as temperature and humidity accuracy, energy consumption, and uniformity, significantly improving the robustness and energy efficiency of the control.

[0078] In step S5, in response to the corresponding device completing the execution of the corresponding instruction strategy, the third environmental data at this time is obtained to obtain the difference between the third environmental data and the standard environmental data, thereby obtaining the updated third control model.

[0079] In response to the corresponding device completing the execution of the corresponding instruction strategy, the system acquires the third environmental data at this time to obtain the difference between the third environmental data and the standard environmental data, thereby obtaining an updated third control model, including: In response to the corresponding device completing the execution of the corresponding instruction strategy, the environmental sensor acquires third environmental data; the third environmental data includes third temperature data, third humidity data, third airflow velocity data, and third device power data; Based on the third environmental data, and using the first interpolation observation model, it is converted into a third temperature and humidity field; Based on the third temperature and humidity field, the deviations of the temperature and humidity values ​​of each calculation unit from the standard environmental data are calculated to obtain the difference matrix. Based on the difference matrix, which serves as the feedback error of the third control model, parameters are corrected to obtain the updated third control model.

[0080] After the base station air conditioner executes the command strategy, the effect can be evaluated, and then the model can be adjusted to achieve an adaptive closed loop.

[0081] When implementing the embodiments of this application, after the base station air conditioner completes the execution of the instruction strategy, the environmental sensor will collect the third environmental data inside the base station room again at a preset sampling frequency, for example, once per second to once every 10 seconds.

[0082] The third environmental data includes third temperature data, third humidity data, third airflow velocity data, and third equipment power data. Its data structure is the same as that of the first environmental data to ensure data consistency and comparability.

[0083] Then, by receiving the third environmental data and converting it into a third temperature and humidity field using the first interpolation observation model, the deviation between the temperature and humidity values ​​of each calculation unit in the third temperature and humidity field and the set standard environmental data can be calculated. In other words, the deviation of the target temperature and target humidity is calculated, forming a difference matrix. This difference matrix records in detail the difference between the actual temperature and humidity of each spatial point inside the computer room and the ideal target value, providing specific feedback signals for subsequent model correction.

[0084] The difference matrix serves as the feedback error of the third control model, used to update and optimize the internal parameters of the third control model, thereby achieving adaptive learning. Here, an adaptive algorithm is directly used to approximate the target and correct the model parameters.

[0085] In this embodiment, the adaptive algorithm can be a reinforcement learning algorithm. The difference matrix is ​​transformed into a negative reward signal, the magnitude of which is proportional to the absolute value of the deviation. The larger the deviation, the larger the negative reward. Therefore, the third regulation model is regarded as an intelligent agent. Then, based on this reward signal, the policy gradient or other reinforcement learning algorithms are used to adjust its internal control policy parameters.

[0086] For example, by adjusting the weight coefficients in the objective function Alternatively, the parameters of the third control model can be adjusted to maximize the cumulative rewards obtained in the future, that is, to minimize the temperature and humidity deviation of the computer room environment and minimize energy consumption.

[0087] The adaptive algorithm in this embodiment can also be calculated using the recursive least squares method, in which case the difference matrix is ​​used to iteratively update the parameters of the third control model. Least squares is an online parameter estimation algorithm, particularly suitable for handling time-varying systems. It can dynamically adjust model parameters based on the latest observation data to better fit the actual system behavior.

[0088] Using the least squares method, the parameter estimation vector update of the third regulation model can be expressed as: ; The covariance matrix update of the third regulation model can be expressed as: ; in, Let be the model parameter estimation vector at time k, representing the set of parameters of the predictive model within the predictive control MPC at the current time. Let be the regression vector at time k, which contains the input variables used for prediction, such as temperature and humidity fields, air conditioning control variables, external environmental parameters, etc. This represents the actual observed output at time k, i.e., the actual temperature and humidity values ​​in the third temperature and humidity field. Let be the covariance matrix at time k, which reflects the uncertainty in the estimation of the model parameters. It is the forgetting factor, which is usually slightly less than 1, for example, 0.95 to 0.99. Then, because recent data has a greater weight, older data can be gradually forgotten during the iteration process, so that the model can adapt to slow changes in the environment and parameter drift more quickly.

[0089] Through this feedback mechanism, the third control model can adaptively learn and correct parameters based on actual operating results, continuously improving its prediction accuracy and control performance. This continuous self-optimization capability can better adapt to dynamic changes in the computer room environment, equipment aging, sensor drift and other nonlinear factors, thereby maintaining long-term robustness and efficiency.

[0090] This embodiment also includes step S6, which determines whether the difference between the third environmental data and the standard environmental data is within a preset range. If the difference is within the preset range, it is considered that the computer room environment has reached a stable and optimal control state, the current cycle is stopped, and the current base station air conditioning operation strategy is maintained until the next preset control cycle.

[0091] The setting range is the allowable deviation range determined comprehensively based on the operating standards of the base station equipment room, the sensitivity of the internal equipment, and the energy efficiency target. For example, it can be set to a temperature difference within ±5°C and a humidity difference within ±5%RH. The threshold is to ensure that the equipment room environment meets the equipment operation requirements while avoiding excessive control that leads to increased energy consumption.

[0092] If the difference is within the set range, the computer room environment is considered to have reached a stable and optimal control state. The current cycle can be stopped, and the current base station air conditioning operation strategy can be maintained until the next preset control cycle, such as a complete control cycle every hour or every four hours. Alternatively, the cycle can be restarted when a new significant environmental disturbance is detected, such as a sudden change in external temperature or a sharp increase in equipment load.

[0093] If the difference exceeds the set range, it will automatically return to the first step, re-collect the latest first environmental data, and restart the above complete environmental control process. This iterative process ensures that the environmental control system of the base station equipment room can continuously perform self-sensing, self-prediction, self-optimization, and self-correction, thereby maintaining the long-term stability and optimality of the equipment room environment, effectively responding to sudden situations and gradual changes, minimizing operational risks, and improving energy efficiency.

[0094] Setting a loop termination condition ensures that the system can enter a stable operating state after reaching performance goals, while also being able to respond quickly and re-optimize when necessary, demonstrating adaptability.

[0095] Example 3 This is the third embodiment of the present invention. Based on embodiments 1 and 2, please refer to the following references. Figure 3 This embodiment provides an environmental control system based on base station air conditioning, including a data acquisition unit, an environmental observation unit, an environmental prediction unit, a control unit, and an optimization and update unit connected in sequence. The data acquisition unit is configured to deploy environmental sensors in a grid pattern to acquire first environmental data for a first time interval. The environmental observation unit is configured to: obtain the temperature and humidity distribution in the base station equipment room based on the first environmental data and the spatial geometric parameters of the base station equipment room, and use it as the first temperature and humidity field based on the preset first interpolation observation model; The environmental prediction unit is configured to: obtain the predicted temperature and humidity change distribution for a second time interval based on the set standard environmental data and the first temperature and humidity field, and use it as the second temperature and humidity field, according to the preset second time series prediction model; The control unit is configured to: obtain the optimal control parameters based on the predicted second temperature and humidity field and a preset third control model, so as to control the corresponding device to execute the corresponding instruction strategy; The optimization and update unit is configured to: in response to the corresponding device completing the execution of the corresponding instruction strategy, obtain the third environment data at this time, obtain the difference between the third environment data and the standard environment data, and thus obtain the updated third control model.

[0096] In the implementation of this application, by combining the deployment of environmental sensors in a grid pattern with multidimensional data analysis, and introducing an interpolation observation model and a time series prediction model guided by spatial geometric parameters, the system achieves the perception and prediction of temperature and humidity distribution inside the base station equipment room. This results in the elimination of local hotspots, reduction of energy consumption, and improvement of environmental stability. Furthermore, by continuously optimizing the control model through a closed-loop feedback mechanism, the system acquires adaptive capabilities, enabling it to cope with challenges posed by complex heat source distributions and external disturbances.

[0097] In practical applications, deploying environmental sensors in a grid pattern can be understood as arranging sensors in different locations within a base station equipment room according to certain rules, with the aim of comprehensively collecting environmental status information. For example, a uniform grid layout can be used, distributing sensors at fixed intervals along the height, width, and depth of the equipment room. Alternatively, a non-uniform grid layout can be used, increasing sensor density in hotspot areas, based on the heat source distribution characteristics of the equipment within the equipment room. The first environmental data—first temperature data, first humidity data, first airflow velocity data, and first equipment power data—can be collected separately by various types of sensors. For example, temperature sensors collect temperature data, humidity sensors collect humidity data, wind speed sensors collect airflow velocity data, and power sensors calculate equipment power data by monitoring the current and voltage of electrical equipment.

[0098] The first interpolation observation model can then be implemented using various mathematical methods, such as interpolation algorithms based on inverse distance weighting or spline functions. Its main function is to transform discrete sensor data into continuous spatially distributed data. The spatial geometric parameters of the base station equipment room can include information such as the length, width, and height of the room, as well as the location and size of internal obstacles. These parameters are used to guide the interpolation calculation process, thereby improving the accuracy of temperature and humidity field reconstruction.

[0099] The second time series forecasting model can be implemented using various machine learning algorithms, such as support vector machine regression or random forest regression. Its input data includes historical environmental data and corresponding power load data. By learning from historical data, this model can predict temperature and humidity trends over a future period. Standard environmental data can be pre-set by the user according to the data center's operational needs, and includes target temperature and humidity values.

[0100] Then, the third control model can be implemented using various optimization algorithms, such as genetic algorithms or particle swarm optimization algorithms. Its goal is to comprehensively consider multiple constraints and generate optimal equipment control parameters. Optimal control parameters may include the air conditioner's set temperature, compressor operating frequency, and percentage of the blower fan speed, etc. These parameters guide the equipment in executing specific control strategies.

[0101] Through the above technical solutions, the system can continuously optimize control accuracy during long-term operation, ensuring that temperature and humidity remain within safe threshold ranges, thereby improving the reliability of equipment operation. The overall technical solution integrates spatial perception, prediction, and closed-loop optimization mechanisms to construct an environmental control system. This effectively overcomes the problems of temperature and humidity imbalance, local hotspots, and high energy consumption caused by traditional methods that rely on single measuring points and lack spatial dimension analysis and temporal prediction capabilities. It achieves coordinated temperature and humidity control and dynamic optimization of airflow organization.

[0102] Example 4 The fourth embodiment of the present invention differs from the previous embodiments in that: Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0103] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, or may be electrical, mechanical, or other forms of connection.

[0104] The units described as separate components may or may not be physically separate. As will be apparent to those skilled in the art, the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0105] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0106] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or grid device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0107] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An environmental control method based on base station air conditioning, characterized in that, include: Environmental sensors are deployed in a grid pattern to acquire initial environmental data for the first time interval. The first environmental data is configured as: first temperature data, first humidity data, first airflow velocity data, and first device power data; Based on the first environmental data and the spatial geometric parameters of the base station equipment room, the temperature and humidity distribution in the base station equipment room is obtained based on the preset first interpolation observation model, which serves as the first temperature and humidity field. Based on the set standard environmental data and the first temperature and humidity field, and based on the preset second time series prediction model, the predicted temperature and humidity change distribution for the second time interval is obtained as the second temperature and humidity field; based on the second temperature and humidity field, and based on the preset third control model, the optimal control parameters are obtained to control the corresponding equipment to execute the corresponding instruction strategy. In response to the corresponding device completing the execution of the corresponding instruction strategy, the third environmental data at this time is obtained to obtain the difference between the third environmental data and the standard environmental data, thereby obtaining the updated third control model; The difference between the third environmental data and the standard environmental data is within the set range.

2. The environmental control method based on base station air conditioning according to claim 1, characterized in that, The environmental sensors are deployed in a grid pattern to acquire first environmental data for a first time interval, including: Inside the base station equipment room, multiple sets of environmental sensors are deployed along the height, width, and depth directions at a preset grid density; Each set of environmental sensors is configured as follows: a temperature sensor for collecting the first temperature data, a humidity sensor for collecting the first humidity data, an airflow speed sensor for collecting the first airflow speed data, and a power sensor for collecting the first equipment power data, which is electrically connected to various electrical devices inside the base station equipment room. Each set of environmental sensors synchronously collects data at a preset sampling frequency. The first temperature data, first humidity data, first airflow velocity data, and first device power data collected within the first time interval are packaged together as the first environmental data for the first time interval.

3. The environmental control method based on base station air conditioning according to claim 1, characterized in that, The step of obtaining the temperature and humidity distribution within the base station equipment room space, based on the first environmental data and the spatial geometric parameters of the base station equipment room, and using a preset first interpolation observation model, as the first temperature and humidity field, includes: Based on the spatial geometric parameters of the base station equipment room, a three-dimensional model of the base station equipment room is obtained. Based on the first environmental data, and using the three-dimensional coordinate system of the three-dimensional model of the base station equipment room, the first environmental data collected by each set of environmental sensors is stored in the form of a three-dimensional coordinate system to obtain discrete measurement point data of the base station equipment room. The preset first interpolation observation model is configured to take discrete measurement point data as input, and output data of any unmeasured point in space based on the Kriging algorithm. Based on the discrete measurement point data of the base station equipment room, and based on the preset first interpolation observation model, the temperature and humidity values ​​of any unmeasured points in the space of the base station equipment room are obtained to obtain the temperature and humidity distribution in the space of the base station equipment room, which serves as the first temperature and humidity field.

4. The environmental control method based on base station air conditioning according to claim 1, characterized in that, The step of obtaining the predicted temperature and humidity change distribution for a second time interval, based on the set standard environmental data and the first temperature and humidity field and a preset second time series prediction model, as the second temperature and humidity field, includes: The set standard environmental data includes target temperature and target humidity; The second time series prediction model is trained by dividing historical environmental data and corresponding power load into training set, validation set and test set. The second time interval is divided into multiple time sub-intervals to accommodate power loads at different time scales; Set the length of the second time interval, take the first temperature and humidity field as the starting state of the second time interval, and take the set standard environmental data as the ending state of the second time interval. The starting state and the ending state are input into the second time series prediction model, and the temperature and humidity change distribution of the base station equipment room within the predicted second time interval is output as the second temperature and humidity field.

5. The environmental control method based on base station air conditioning according to claim 4, characterized in that, The step of inputting the starting state and the ending state into the second time series prediction model and outputting the temperature and humidity change distribution of the base station equipment room within the predicted second time interval as the second temperature and humidity field includes: The second time series prediction model is configured to include a long short-term memory network and a Gaussian process regression. Based on the Long Short-Term Memory network, the power grid load forecast for the second time interval is obtained; Set the length of the second time interval, take the first temperature and humidity field as the starting state of the second time interval, and take the set standard environmental data as the ending state of the second time interval. Based on the first temperature and humidity field and load prediction results, the predicted temperature and humidity values ​​for each preset time step at all grid points in the base station equipment room within the second time interval are obtained through Gaussian process regression. Based on the predicted temperature and humidity values ​​at each preset time step on all grid points in the base station equipment room, the temperature and humidity variation distribution in the base station equipment room space within the second time interval is obtained. The temperature and humidity distribution within the base station equipment room is used as the second temperature and humidity field.

6. The environmental control method based on base station air conditioning according to claim 1, characterized in that, The step of obtaining optimal control parameters based on the second temperature and humidity field and a preset third control model to obtain a control strategy for the corresponding device to execute corresponding instructions includes: The second temperature and humidity field characterizes the temperature and humidity change behavior at all grid points within the base station equipment room space during the second time interval; The first temperature and humidity field and the second temperature and humidity field are used as inputs to the third control model; The third control model is configured to include a multi-objective optimization algorithm, the objectives of which include at least: maintaining the temperature and humidity of the base station equipment room within the standard range, minimizing the energy consumption of the base station air conditioning, minimizing the spatial non-uniformity of temperature and humidity, and maximizing the reliability of equipment operation. Based on the second temperature and humidity field, and through the third control model, the optimal control parameter sequence is obtained under the constraints of the equipment operating range and the temperature safety range. According to the optimal control parameter sequence, the first control parameter of the sequence is applied to the base station air conditioner as the optimal control parameter; The optimal control parameters are sent to the base station air conditioning controller, which then obtains the corresponding instruction strategy based on the optimal control parameters.

7. The environmental control method based on base station air conditioning according to claim 5, characterized in that, The command strategy includes a temperature and humidity control strategy and an airflow control strategy; The temperature and humidity control strategy includes: When the overall temperature of the base station equipment room is higher than the preset high threshold and the humidity is within the normal range or lower than the preset low threshold, the set temperature value and the percentage of the air supply fan speed of the base station air conditioner are increased, while the compressor operating frequency is reduced. When the overall temperature of the base station equipment room is higher than the preset high threshold and the humidity is higher than the preset threshold, the set temperature value of the base station air conditioner is reduced, the compressor operating frequency is increased, and the dehumidifier is started. When the overall temperature of the base station room is at a preset low threshold and the humidity is within the normal range or higher than the preset high threshold, the first set temperature value of the base station air conditioner is increased, the operating frequency of the first compressor is reduced, the speed percentage of the first air supply fan is increased, and the operating power percentage of the first dehumidifier is activated. When the overall temperature of the base station equipment room is within the normal range and the humidity is higher than the preset high threshold, the set temperature value is maintained, but the dehumidifier working power percentage is increased, and the air supply fan speed percentage is adjusted. When the overall temperature of the base station equipment room is within the normal range and the humidity is low, maintain the first set temperature value, but increase the percentage of humidifier working power and adjust the percentage of air supply fan speed. The airflow control strategy includes: Based on the first temperature and humidity field and the third control model, local hot spots and cold spots are obtained; simultaneously, based on the second temperature and humidity field and the third control model, predicted future local hot spots are obtained. The hotspot area includes areas where the heat load exceeds a set threshold. Based on the local hot spots, cold spots, and future local hot spots, the angles of multiple air outlet guide vanes of the base station air conditioner are adjusted to guide the cold airflow to the local hot spots, while avoiding cold air short-circuiting and local cold spots.

8. The environmental control method based on base station air conditioning according to claim 1, characterized in that, The step of responding to the corresponding device completing the execution of the corresponding instruction strategy, acquiring the third environmental data at this time, and obtaining the difference between the third environmental data and the standard environmental data, thereby obtaining the updated third control model, includes: In response to the corresponding device completing the execution of the corresponding instruction strategy, third environmental data is acquired based on the environmental sensor; the third environmental data includes third temperature data, third humidity data, third airflow speed data, and third device power data; Based on the third environmental data and the first interpolation observation model, the third temperature and humidity field is obtained; Based on the third temperature and humidity field, the deviations between the temperature and humidity values ​​of each calculation unit and the standard environmental data are calculated to obtain a difference matrix. Based on the difference matrix, the parameters are corrected as the feedback error of the third control model to obtain the updated third control model.

9. An environmental control system based on base station air conditioning, characterized in that, It includes a data acquisition unit, an environmental observation unit, an environmental prediction unit, a control unit, and an optimization and update unit, all connected in sequence. The data acquisition unit is configured to deploy environmental sensors in a grid pattern to acquire first environmental data for a first time interval; The environmental observation unit is configured to: obtain the temperature and humidity distribution in the base station equipment room based on the first environmental data and the spatial geometric parameters of the base station equipment room, and use it as the first temperature and humidity field based on a preset first interpolation observation model; The environmental prediction unit is configured to: obtain the predicted temperature and humidity change distribution for a second time interval based on the set standard environmental data and the first temperature and humidity field, using a preset second time series prediction model as the second temperature and humidity field; The control unit is configured to: obtain optimal control parameters based on the second temperature and humidity field and a preset third control model, so as to control the corresponding device to execute the corresponding instruction strategy; The optimization and update unit is configured to: in response to the corresponding device completing the execution of the corresponding instruction strategy, obtain the third environmental data at this time, obtain the difference between the third environmental data and the standard environmental data, and thus obtain the updated third control model.

10. A computer system apparatus, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 8.

Images