Air conditioner intelligent control system and method based on dynamic load response

Abstract

The invention belongs to the technical field of intelligent control, and provides an air conditioner intelligent control system and method based on dynamic load response, and the system comprises a collection subsystem which is used for collecting indoor and outdoor multi-dimensional measured data and air conditioning unit multi-dimensional measured data; and the processing subsystem is used for obtaining a load prediction result in a future set time period according to the multi-dimensional actually measured data by utilizing a pre-constructed load prediction model, determining a target working mode of the air conditioning unit according to the outdoor actually measured temperature and the load prediction result in the multi-dimensional actually measured data, and controlling the air conditioning unit to work in the target working mode. According to the multi-dimensional actual measurement data and the load prediction result, a dynamic regulation and control instruction for the air conditioning unit is generated and sent out; and the execution subsystem is used for driving at least part of execution components in the air conditioning unit to cooperatively operate according to the dynamic regulation and control instruction. According to the scheme, stable operation of equipment can be guaranteed, meanwhile, the requirement for indoor environment comfort can be better met, and the accuracy, reliability and stability of a control link are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and in particular to an intelligent control system and method for air conditioning based on dynamic load response. Background Technology

[0002] In building environment control in cold regions, air conditioning units need to simultaneously cope with extreme weather conditions such as extremely low temperatures and strong winds, as well as the dynamic changes in indoor temperature, humidity, and air quality. Traditional air conditioning control systems are no longer able to meet the precise operation requirements under complex scenarios.

[0003] Specifically, traditional air conditioning control systems often rely on single environmental parameters to trigger control actions, lacking comprehensive perception of multi-dimensional indoor and outdoor parameters. This leads to a disconnect between equipment operation and actual needs, easily resulting in problems such as insufficient heating and unbalanced fresh air volume regulation. Furthermore, traditional air conditioning control systems struggle to capture the nonlinear characteristics of load data in cold regions and the abrupt changes under extreme operating conditions, causing delayed mode switching. This wastes energy and affects the stability of the indoor environment, especially during transitional seasons when frequent equipment start-ups and shutdowns are common, making it difficult to meet the dual demands of reliability and comfort for air conditioning systems in cold regions.

[0004] It is not difficult to see that traditional air conditioning control systems suffer from technical problems such as low control precision, poor reliability, and insufficient stability. Summary of the Invention

[0005] This invention provides an intelligent air conditioning control system and method based on dynamic load response, which solves the defects of traditional air conditioning control systems such as low control accuracy, poor reliability and insufficient stability.

[0006] On one hand, the present invention provides an intelligent air conditioning control system based on dynamic load response, comprising: The data acquisition subsystem is used to collect multi-dimensional measured data from indoor and outdoor environments and air conditioning units. The processing subsystem is used to obtain the load forecast result for a future set time period based on the multidimensional measured data and a pre-built load forecast model; determine the target operating mode of the air conditioning unit based on the measured outdoor temperature in the multidimensional measured data and the load forecast result; and generate and issue dynamic control instructions for the air conditioning unit based on the multidimensional measured data and the load forecast result under the target operating mode. An execution subsystem is used to drive at least some of the execution components in the air conditioning unit to operate collaboratively according to the dynamic control instructions.

[0007] According to the intelligent air conditioning control system based on dynamic load response provided by the present invention, the data acquisition subsystem includes: Outdoor environment data acquisition components are used to collect measured outdoor temperature and measured outdoor wind speed. Indoor environmental data acquisition components are used to collect measured indoor temperature, measured indoor humidity, and measured indoor carbon dioxide concentration. The unit operation data acquisition component is used to collect the measured temperature of key parts of the air conditioning unit, the measured operating current of the motor, and the measured opening degree of key valves.

[0008] According to the intelligent air conditioning control system based on dynamic load response provided by the present invention, the load prediction model includes: The CNN module is used to convert the multidimensional measured data into a two-dimensional feature matrix, extract the hidden features of the data in the two-dimensional feature matrix, and output key feature vectors. The LSTM module is used to extract the long-term temporal dependencies of the data in the key feature vectors and output preliminary prediction results. The XGBoost module is used to correct the preliminary prediction results based on the key environmental parameters in the multidimensional measured data through ensemble learning, and output the load prediction results.

[0009] According to the intelligent air conditioning control system based on dynamic load response provided by the present invention, before obtaining the load forecast result for a future set time period based on the multidimensional measured data and using a pre-built load forecast model, the processing subsystem is further configured to: Identify and remove abnormal data from the multidimensional measured data caused by sensor failure or extreme weather interference to obtain valid measured data; The effective measured data are standardized to obtain preprocessed multidimensional measured data.

[0010] According to the intelligent air conditioning control system based on dynamic load response provided by the present invention, the processing subsystem determines the target operating mode of the air conditioning unit based on the measured outdoor temperature in the multi-dimensional measured data and the load prediction results, including: Determine the target temperature range to which the measured outdoor temperature belongs in the multidimensional measured data; Determine the load type corresponding to the load forecast results; The target operating mode of the air conditioning unit is determined based on the target temperature range and the load type.

[0011] According to the intelligent air conditioning control system based on dynamic load response provided by the present invention, the target operating mode of the air conditioning unit is determined based on the target temperature range and the load type, including: When the target temperature range is below the first set temperature and the load type is heat load, the target operating mode of the air conditioning unit is determined to be the heating mode; When the target temperature range is from the first set temperature to the second set temperature and the load type is primary cooling load, the target operating mode of the air conditioning unit is determined to be natural cooling mode; When the target temperature range is above the second set temperature and the load type is medium cooling load, the target operating mode of the air conditioning unit is determined to be cooling mode; Wherein, the first set temperature is lower than the second set temperature.

[0012] According to the intelligent air conditioning control system based on dynamic load response provided by the present invention, the multi-dimensional measured data includes: indoor measured carbon dioxide concentration, outdoor measured wind speed, indoor measured temperature, measured temperature of key parts of the air conditioning unit, measured opening degree of key valves, and indoor measured humidity. The processing subsystem generates dynamic control commands for the air conditioning unit based on the multidimensional measured data and the load forecast results, including: Based on the measured indoor carbon dioxide concentration and the measured outdoor wind speed, the fresh air adjustment amount is determined, and a fresh air control instruction is generated according to the fresh air adjustment amount. Based on the load forecast results, the measured indoor temperature, and the measured opening of the key valves, the adjustment amount of the hot and cold water valve opening is determined, and a hot and cold water valve opening control command is generated according to the adjustment amount of the hot and cold water valve opening. Based on the measured indoor humidity and the measured temperature of key parts of the air conditioning unit, the dehumidification parameter adjustment amount is determined, and a dehumidification control command is generated based on the dehumidification parameter adjustment amount. The fresh air control command, the hot and cold water valve opening control command, and the dehumidification control command are used as dynamic control commands for the air conditioning unit.

[0013] According to the intelligent air conditioning control system based on dynamic load response provided by the present invention, the multi-dimensional measured data further includes: measured operating current of the motor; The processing subsystem is also used to determine the motor's operating state based on the measured operating current of the motor, and when the motor's operating state is detected to be overloaded, control the motor's operating frequency to decrease from the current frequency value to the target frequency value according to the set frequency adjustment step size.

[0014] The intelligent air conditioning control system based on dynamic load response provided by the present invention further includes: a management subsystem; The management subsystem is used to acquire key execution data and execution environment data of the execution subsystem, establish a feedback dataset based on the key execution data and execution environment data, and optimize and update the load prediction model based on the feedback dataset.

[0015] On the other hand, the present invention also provides an intelligent air conditioning control method based on dynamic load response. Based on any of the above-described intelligent air conditioning control systems based on dynamic load response, the method includes: The data acquisition subsystem collects multi-dimensional measured data from indoor and outdoor environments and air conditioning units. The processing subsystem obtains the load forecast results for a future set time period based on the multidimensional measured data and the pre-built load forecast model. Based on the outdoor measured temperature in the multidimensional measured data and the load forecast results, the target operating mode of the air conditioning unit is determined. Under the target operating mode, dynamic control commands for the air conditioning unit are generated and issued based on the multidimensional measured data and the load forecast results. By executing the subsystem according to the dynamic control instructions, at least some of the execution components in the air conditioning unit are driven to operate in coordination.

[0016] The intelligent air conditioning control system and method based on dynamic load response provided by this invention comprehensively acquires multi-dimensional measured data of indoor and outdoor environments and air conditioning units through the acquisition subsystem, providing data support for precise control; the processing subsystem uses a load prediction model to predict the load in advance, combines the measured outdoor temperature to determine the target operating mode and generate dynamic control commands, achieving forward-looking and adaptable control; the execution subsystem drives the components to work together to ensure the commands are implemented. Overall, it can effectively improve the operating accuracy of air conditioning units under complex operating conditions, reduce energy waste, ensure stable equipment operation, and better meet the comfort requirements of the indoor environment, thereby improving the accuracy, reliability and stability of the control process. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the structure of an intelligent air conditioning control system based on dynamic load response provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the acquisition subsystem in an embodiment of the present invention; Figure 3 This is a schematic diagram of the load forecasting model in an embodiment of the present invention; Figure 4 This is a flowchart illustrating the intelligent air conditioning control method based on dynamic load response provided in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0020] The following is combined Figures 1 to 4 This invention describes the detailed scheme of an intelligent air conditioning control system and method based on dynamic load response provided in embodiments of the present invention.

[0021] like Figure 1 As shown, the intelligent air conditioning control system based on dynamic load response provided in this embodiment of the invention mainly includes: The data acquisition subsystem 110 is used to collect multi-dimensional measured data from indoor and outdoor environments and air conditioning units.

[0022] The processing subsystem 120 is used to obtain the load forecast results for a future set period based on multidimensional measured data and a pre-built load forecast model. Based on the outdoor measured temperature and load forecast results in the multidimensional measured data, it determines the target operating mode of the air conditioning unit. Under the target operating mode, it generates and issues dynamic control commands to the air conditioning unit based on the multidimensional measured data and load forecast results.

[0023] The execution subsystem 130 is used to drive at least some of the execution components in the air conditioning unit to operate in coordination according to dynamic control commands.

[0024] In one embodiment, such as Figure 2 As shown, the acquisition subsystem 110 specifically includes: The outdoor environment acquisition component 210 is used to collect the measured outdoor temperature and measured outdoor wind speed.

[0025] In practical applications, low-temperature temperature sensors and wind speed sensors adapted to extremely low temperatures of -30℃ to -10℃ can be deployed at the fresh air inlet of the air conditioning unit and on the building roof to collect real-time outdoor measured temperature and wind speed; at the same time, atmospheric pressure sensors and rain and snow monitoring sensors can be added to detect outdoor air pressure and extreme wind and rain conditions.

[0026] The indoor environment acquisition component 220 is used to collect the measured indoor temperature, measured indoor humidity, and measured indoor carbon dioxide concentration.

[0027] In practical applications, carbon dioxide concentration sensors and indoor temperature and humidity sensors can be deployed in various functional areas of the building to collect the measured indoor temperature, indoor humidity, and indoor carbon dioxide concentration. The sampling frequency can be set to once per minute to ensure the timeliness of the data.

[0028] The unit operation data acquisition component 230 is used to collect the measured temperature of key parts of the air conditioning unit, the measured operating current of the motor, and the measured opening degree of key valves.

[0029] In practical applications, dedicated sensors can be installed inside the air conditioning unit and at the low-temperature motor to collect measured temperatures of key components such as motor winding temperature, surface cooler inlet water temperature, and hot and cold water supply and return temperatures. They can also collect measured openings of key valves such as hot and cold water valves and fresh air valves, as well as key data such as measured motor operating current. This allows for monitoring of the operating status of the core components of the air conditioning unit and provides data support for subsequent control.

[0030] In one embodiment, such as Figure 3 As shown, the load forecasting model specifically includes: The CNN module 310 is used to convert multidimensional measured data into a two-dimensional feature matrix, extract the hidden features of the data in the two-dimensional feature matrix, and output key feature vectors.

[0031] In this embodiment, the CNN (Convolutional Neural Network) module adopts a three-level structure of input reshaping, convolution extraction, and pooling dimensionality reduction to adapt to the multidimensional data characteristics of air conditioning systems in cold regions. Structurally, the CNN module first reshapes the preprocessed multidimensional measured data into a two-dimensional feature matrix through a feature reshaping layer. Specifically, it adopts a matrix-based approach using feature groups and time steps, dividing outdoor environmental parameters, indoor environmental parameters, and unit operating parameters into three feature groups. Analogous to the RGB channels of an image, a 3×24×M three-dimensional tensor is constructed, where M is the number of sampling points per hour. Then, it is converted into a two-dimensional matrix format through dimensionality compression.

[0032] The core processing layer consists of 2-3 alternating convolutional layers and max pooling layers. The convolutional layers use 3×3 kernels that slide along the time step with a stride of 1, capturing hidden relationships between parameters through non-linear activation functions, such as the coupling between -25℃ low temperature and a sudden increase in motor current, and the constraint relationship between strong wind and the opening of the fresh air valve. The pooling layers downsample the convolution results with a 2×2 window, preserving key features while reducing data dimensionality and subsequent computation. Finally, a flattening layer maps multidimensional features into one-dimensional key feature vectors, achieving precise extraction of core features from raw data.

[0033] The LSTM module 320 is used to extract the long-term temporal dependencies of data in key feature vectors and output preliminary prediction results.

[0034] LSTM (Long Short Term Memory) modules employ a stacked hidden layer and gating mechanism structure to specifically address the long sequence dependencies in workload data from cold regions. Structurally, it consists of an input layer, 2-3 stacked LSTM hidden layers, and a fully connected output layer. Each hidden layer contains 64-128 neurons, and selective information transmission is achieved through gating units (i.e., forget gate, input gate, and output gate).

[0035] During data processing, the key feature vectors output by the CNN module are first input into the first LSTM hidden layer in chronological order. The forget gate uses a sigmoid activation function to determine whether to retain historical features, such as the low-temperature load features of the same period the previous day. The input gate filters key features of the current moment, such as the impact of real-time wind speed on the load, and combines them with candidate memory states to update the cell state. The output gate controls the intensity of information passed from the cell state to subsequent layers. Stacked hidden layers further enhance the ability to extract temporal features. The upper layer captures short-term fluctuation features, such as the impact of hourly temperature changes on the load, while the lower layer mines long-term dependency patterns, such as diurnal temperature differences and cyclical load change trends. Finally, a fully connected layer maps the temporal features output by the hidden layer to the heating and cooling loads and fresh air demand loads for the next hour, forming preliminary prediction results. This effectively avoids the gradient vanishing problem of traditional RNN models.

[0036] The XGBoost module 330 is used to revise the preliminary prediction results based on key environmental parameters in multidimensional measured data through ensemble learning, and output the load prediction results.

[0037] In this embodiment, the XGBoost module employs a multi-tree CART (Classification and Regression Tree) ensemble and gradient optimization structure, focusing on correcting prediction biases under extreme conditions in cold regions. Structurally, it uses classification and regression trees as the base model, constructing an ensemble model of 100-200 trees through additive training. Each tree contains 8-12 layers of decision nodes, and node splitting is determined based on the principle of maximizing information gain.

[0038] During data processing, the input consists of two parts: the initial prediction results output by the LSTM module and key environmental parameters from multi-dimensional measured data, such as extreme outdoor low temperatures, average outdoor wind speed over 10 minutes, and rain / snow condition indicators. During training, the mean squared error is used as the objective function, and decision trees are added one by one using gradient descent. Each new tree fits the prediction residuals of the previous model, while regularization terms (such as tree complexity penalties and leaf node weight constraints) are introduced to avoid overfitting. Under extreme conditions, such as temperatures as low as -30℃ or winds exceeding 12 m / s, the XGBoost module can correct biases through collaborative decision-making across multiple trees. For example, if the initial predicted load underestimates heating demand due to extreme low temperatures, the XGBoost module can adjust the predicted value by 10%-15% based on historical data of similar low-temperature conditions through a multi-tree voting mechanism, ultimately outputting a corrected and accurate load prediction result, ensuring lower prediction errors under extreme environments.

[0039] In one embodiment, before obtaining the load forecast results for a future set time period based on multidimensional measured data and a pre-built load forecasting model, the processing subsystem may also be used to: First, identify and remove abnormal data from the multidimensional measured data caused by sensor failure or extreme weather interference to obtain valid measured data.

[0040] In practical applications, the 3σ criterion can be used to identify and eliminate abnormal data caused by sensor malfunctions or extreme weather interference, such as invalid values ​​from temperature and humidity sensors that drop sharply in low-temperature environments, to ensure data quality. Meanwhile, for missing data in multidimensional measured data, linear interpolation or the nearest-time average method can be used to supplement the missing data to ensure data integrity.

[0041] Then, the valid measured data are standardized to obtain preprocessed multidimensional measured data.

[0042] In practical applications, valid measured data can be mapped to the interval [0,1] to eliminate the dimensional differences between different types of data. Specifically, the minimax method can be used to standardize the valid measured data.

[0043] In one embodiment, the processing subsystem determines the target operating mode of the air conditioning unit based on the measured outdoor temperature and load prediction results from the multi-dimensional measured data, specifically including: First, determine the target temperature range to which the outdoor measured temperature belongs in the multidimensional measured data.

[0044] Understandably, multiple temperature ranges can be pre-defined to meet control requirements under different temperature environments.

[0045] At the same time, determine the load type corresponding to the load forecast results.

[0046] Understandably, the load type mainly reflects whether the subsequent control stage is cooling or heating, providing the main direction for control.

[0047] Finally, the target operating mode of the air conditioning unit is determined based on the target temperature range and load type.

[0048] In a specific implementation, the target operating mode of the air conditioning unit is determined based on the target temperature range and load type, specifically including: In one scenario, when the target temperature range is below the first set temperature and the load type is heat load, the target operating mode of the air conditioning unit is determined to be heating mode.

[0049] In this embodiment, the first set temperature can be -10℃. When the target working mode is determined to be the heating mode, the heating mode can be started immediately. At the same time, the low temperature motor is linked to enter the low temperature enhanced operation state and the natural cooling circuit is closed to prevent the unit from freezing inside.

[0050] In another scenario, when the target temperature range is from the first set temperature to the second set temperature and the load type is primary cooling load, the target operating mode of the air conditioning unit is determined to be natural cooling mode.

[0051] In this embodiment, the second set temperature can be 15°C. When the target working mode is determined to be the natural cooling mode, the natural cooling mode can be started immediately. At this time, the fresh air valve needs to be opened to the maximum extent. The upper limit of the opening needs to be determined in combination with the strength of the wind-resistant structure. For example, when the outdoor wind speed is 10m / s, the opening of the fresh air valve is limited to no more than 60%. At this time, the outdoor low temperature air can be used to achieve free cooling and reduce energy consumption.

[0052] In another scenario, when the target temperature range is above the second set temperature and the load type is medium cooling load, the target operating mode of the air conditioning unit is determined to be cooling mode.

[0053] The first set temperature is lower than the second set temperature.

[0054] When the target operating mode is determined to be cooling mode, the cooling mode can be started immediately. At this time, the cold water valve needs to be opened. At the same time, the inlet water temperature of the surface cooler can be monitored to ensure that it is maintained below 7°C to ensure the cooling effect.

[0055] In one embodiment, the multidimensional measured data specifically includes: indoor measured carbon dioxide concentration, outdoor measured wind speed, indoor measured temperature, measured temperature of key parts of the air conditioning unit, measured opening degree of key valves, and indoor measured humidity.

[0056] Furthermore, the processing subsystem generates dynamic control commands for the air conditioning unit based on multi-dimensional measured data and load forecast results, specifically including: On the one hand, based on the measured indoor carbon dioxide concentration and the measured outdoor wind speed, the fresh air regulation amount is determined, and fresh air control instructions are generated according to the fresh air regulation amount.

[0057] In this case, the processing subsystem first extracts the measured carbon dioxide concentration and measured outdoor wind speed from the multi-dimensional measured data of each functional area indoors, and determines the fresh air regulation amount with the dual core of ensuring indoor air quality and adapting to strong wind protection in cold regions.

[0058] First, a baseline fresh air volume is set, which is based on the building's occupancy density, space volume, and minimum fresh air volume standard. Then, the baseline fresh air volume is dynamically adjusted according to the measured indoor carbon dioxide concentration: when the measured indoor carbon dioxide concentration exceeds the set threshold (e.g., 600ppm) by 100ppm, the baseline fresh air volume is increased by 10%, while the upper limit of the baseline fresh air volume is locked at 1.2 times to avoid excessive introduction of low-temperature fresh air and increased heating load.

[0059] Next, adjustments are made based on the measured outdoor wind speed. If the measured outdoor wind speed is below 8 m / s, the adjusted fresh air volume can be used directly as the fresh air regulation volume. If the measured outdoor wind speed is between 8 m / s and 10 m / s, the fresh air volume is reduced by 20% to prevent strong winds from impacting the unit's duct support system. If the measured outdoor wind speed is above 10 m / s, the fresh air volume is reduced by 60% to prevent strong winds from impacting the unit's duct support system. At the same time, the windproof sealing valve on the unit's outer casing is pre-activated.

[0060] Finally, the processing subsystem converts the final fresh air adjustment amount into the fresh air valve opening percentage, generates and issues a fresh air control command, which simultaneously includes an emergency valve shut-off trigger condition when the wind speed exceeds 12m / s, to ensure equipment safety under extreme strong winds.

[0061] On the other hand, based on the load forecast results, the measured indoor temperature, and the measured opening of key valves, the adjustment amount of the hot and cold water valve opening is determined, and the hot and cold water valve opening control command is generated according to the adjustment amount of the hot and cold water valve opening.

[0062] In this embodiment, the processing subsystem uses the load prediction results, the measured indoor temperature, and the measured opening of the hot and cold water valves among the key valves as core parameters, and determines the adjustment amount of the hot and cold water valve opening in three steps and generates the hot and cold water valve opening control command.

[0063] The first step is to determine the temperature adjustment priority based on the load type and magnitude of the load forecast results: if the load type is heat load, the goal is to make the measured indoor temperature approach the heating set temperature, and the temperature deviation is calculated; if the load type is cooling load, the goal is to make the temperature deviation approach the cooling set temperature.

[0064] The second step involves calculating the basic adjustment amount by combining the proportional coefficient and the measured opening degree of the hot and cold water valves. Under heating conditions, the proportional coefficient is set to 1.0-1.2 to enhance heating response, and the basic opening degree adjustment amount is the product of the temperature deviation and the proportional coefficient. Under cooling conditions, the proportional coefficient is set to 0.8-1.0 to avoid over-cooling, and the basic opening degree adjustment amount is also the product of the temperature deviation and the proportional coefficient. Subsequently, load forecast results are used for correction. If the absolute value of the load forecast result is higher than 70% of the unit's rated load, the basic opening degree adjustment amount is increased by 15% to accelerate load response. If the absolute value of the load forecast result is lower than 30% of the rated load, the basic opening degree adjustment amount is decreased by 10% to prevent frequent valve opening and stopping.

[0065] The third step is to compare the baseline adjustment with the current valve opening to ensure the final opening is within the safe range of 0%-100%. If the final opening exceeds 100% after adjustment, 100% is taken as the final opening; if the final opening is below 0%, 0% is taken as the final opening. Otherwise, the currently determined baseline adjustment is directly used as the adjustment amount for the hot and cold water valve opening. The processing subsystem converts the hot and cold water valve opening adjustment amount into an electrical signal command, specifying the valve adjustment rate, such as adjusting the opening by 5% per minute, to avoid sudden changes in opening causing water temperature fluctuations.

[0066] On the other hand, based on the measured indoor humidity and the measured temperature of key parts of the air conditioning unit, the dehumidification parameter adjustment amount is determined, and a dehumidification control command is generated based on the dehumidification parameter adjustment amount.

[0067] In this embodiment, the processing subsystem generates dehumidification control commands based on the measured indoor humidity and the measured temperature of key parts of the air conditioning unit, with the principles of precise humidity control and prevention of low-temperature freezing.

[0068] First, the dehumidification start-up conditions are determined. When the measured indoor humidity is higher than 60%, the dehumidification process is triggered. If the measured indoor humidity is lower than 60%, a dehumidification device shutdown command is generated.

[0069] After the dehumidification process is triggered, the first step is to determine the dehumidification priority. If the measured indoor humidity is higher than 70%, the high humidity condition is met, with the goal of rapid dehumidification. If the measured indoor humidity is between 60% and 70%, the medium humidity condition is met, with the goal of controlling humidity without affecting temperature control.

[0070] The second step involves determining the dehumidification parameter adjustment based on the measured temperatures of key components. For high humidity conditions, priority is given to adjusting the evaporator inlet water temperature, setting the target value at 5-6℃ (below the dew point). The difference between the current measured evaporator inlet water temperature and the target value is calculated to generate an adjustment command for the evaporator chilled water valve opening. Simultaneously, the independent dehumidification device is activated, and the dehumidification fan speed is set to 80% of its rated speed. For medium humidity conditions, the operating power of the independent dehumidification device is reduced to 50% of its rated power, and the measured evaporator inlet water temperature is controlled at 7℃ to prevent evaporator frost from affecting heating.

[0071] The third step is to monitor the changes in indoor humidity in real time during the dehumidification process: when the indoor humidity drops to 55%, a dehumidification parameter callback instruction is generated, such as reducing the opening of the surface cooler chilled water valve by 10% to avoid excessive dehumidification and resulting in indoor dryness; if the measured temperature of the surface cooler inlet water is lower than 4°C during the dehumidification process, a surface cooler water temperature increase instruction is immediately generated to prevent the surface cooler pipes from freezing due to excessive cold, thus ensuring the safe operation of the unit in cold regions.

[0072] Finally, the fresh air control command, the hot and cold water valve opening control command, and the dehumidification control command are used as dynamic control commands for the air conditioning unit.

[0073] In one embodiment, the multidimensional measured data may further include: the measured operating current of the motor.

[0074] The processing subsystem is also used to determine the motor's operating status based on the measured operating current of the motor, and when the motor's operating status is detected to be overloaded, control the motor's operating frequency to decrease from the current frequency value to the target frequency value according to the set frequency adjustment step size.

[0075] In this embodiment, the processing subsystem can simultaneously extract the measured operating current of the motor from the multi-dimensional measured data to determine the motor's operating status and implement dynamic control, with the goal of ensuring the long-term stable operation of the low-temperature motor.

[0076] First, the processing subsystem needs to call the motor rated parameter library to obtain the rated current of the currently running motor and set the overload judgment threshold. When the measured operating current of the motor exceeds 10% of the rated current, the motor is judged to have entered an overload state. At this time, the overload protection control should be activated first to avoid damage to the motor windings due to overheating or accelerated aging of the insulation layer due to high temperature.

[0077] Next, the processing subsystem determines the target frequency value by combining the current motor operating frequency with the load forecast results: if the load forecast results indicate that the current load type is a heat load and the load forecast results show that the load value is decreasing, then the target frequency value is set to 0.8 times the current motor operating frequency, thereby reducing the motor operating frequency while meeting the load demand; if the forecast results show that the current load is still relatively high, such as stable heating demand under extreme low temperatures, then the target frequency value is set to 0.9 times the current motor operating frequency to balance the load demand and motor protection, while ensuring that the target frequency value is not lower than the minimum stable operating frequency of the motor, which is usually 50% of the rated frequency, to prevent the motor from stalling.

[0078] Finally, the processing subsystem adjusts the frequency step size according to the set frequency, such as reducing it by 5Hz every 10 seconds, to avoid sudden frequency drops causing unit load fluctuations. It generates motor frequency adjustment commands and monitors the changes in the measured operating current of the motor in real time during the adjustment process: when it drops to less than 1.05 times the rated current, it pauses the frequency reduction and maintains the current frequency operation; if the measured operating current of the motor is still higher than the set current threshold after adjusting to the target frequency value, it further feeds the signal of the measured operating current exceeding the threshold to the load prediction correction link. Then, it can reduce the overall load of the unit by fine-tuning parameters such as the opening degree of the hot and cold water valves and the fresh air volume, forming a closed-loop control process of motor protection and load adaptation, ensuring that the motor in cold regions operates under safe conditions.

[0079] In one embodiment, the above-mentioned intelligent air conditioning control system based on dynamic load response may further include: a management subsystem; The management subsystem is used to acquire key execution data and execution environment data from the execution subsystem, establish a feedback dataset based on the key execution data and execution environment data, and optimize and update the load prediction model based on the feedback dataset.

[0080] In this embodiment, the management subsystem aims to improve the adaptability and accuracy of the load forecasting model by forming a closed-loop optimization through data collection, dataset construction, and model iteration.

[0081] First, the management subsystem and the execution subsystem establish a real-time data interaction channel to continuously acquire two types of key data: one is key execution data, including the actual opening degree of the hot and cold water valves driven by the execution subsystem, the actual adjustment amount of the fresh air valve, the operating power of the dehumidifier, the final operating frequency of the motor, and other action data of the execution components, as well as the actual changes in indoor temperature, humidity, and carbon dioxide concentration after execution, to ensure coverage of key parameters throughout the entire process of implementing control commands; the other is execution environmental data, covering external environmental data such as real-time outdoor temperature, wind speed, and rain / snow conditions during the execution process, as well as equipment operating environment data such as the inlet water temperature of the air conditioning unit's surface cooler, the motor winding temperature, and the casing wind pressure, to comprehensively capture environmental influencing factors in the execution process.

[0082] Next, the management subsystem preprocesses the two types of data, removes outliers caused by momentary sensor failures, supplements occasional missing data using linear interpolation, standardizes data of different dimensions to the same interval, and integrates them according to the structure of execution timestamp, execution key data, execution environment data, and actual load deviation. The actual load deviation is obtained by comparing the actual load calculated by indoor and outdoor parameters after execution with the predicted load output by the original load prediction model. Finally, a feedback dataset containing time series, multi-dimensional parameters, and deviation annotations is constructed.

[0083] Subsequently, the management subsystem initiates the model optimization process according to a set cycle (e.g., every 7 days), dividing the feedback dataset into incremental training and validation sets proportionally and inputting them into the original load prediction model. First, the hidden features in the feedback data are re-extracted through the CNN module, and the feature weights under extreme conditions are updated. Then, the LSTM module is used to learn the temporal variation pattern of the execution data and correct the fitting bias of the long-term and short-term dependencies. Finally, the XGBoost module adjusts the decision nodes and weights of the ensemble tree based on the actual load deviation to optimize the prediction bias correction logic under extreme conditions.

[0084] During the optimization process, the prediction error of the validation set is used as the evaluation index. If the error is reduced by more than 5% compared with the original model, the optimized model parameters are saved and the original load prediction model is replaced. If the error does not meet the standard, the abnormal operating conditions in the feedback data are analyzed back, such as the extreme low temperature scenario that is not covered. Data on such operating conditions is collected and the model is retrained until the model accuracy meets the requirements. This enables the load prediction model to be continuously optimized as the system operates, thereby improving its adaptability to complex operating conditions.

[0085] Based on the same general inventive concept, this invention also protects an intelligent air conditioning control method based on dynamic load response. The intelligent air conditioning control method based on dynamic load response provided by this invention will be described below. The intelligent air conditioning control method based on dynamic load response described below can be referred to in correspondence with the intelligent air conditioning control system based on dynamic load response described above.

[0086] like Figure 4 As shown, the intelligent air conditioning control method based on dynamic load response provided in this embodiment of the invention is implemented based on the intelligent air conditioning control system based on dynamic load response provided in the above embodiments. The method specifically includes the following steps: Step 410: Collect multi-dimensional measured data of indoor and outdoor spaces and air conditioning units through the data acquisition subsystem.

[0087] Step 420: The processing subsystem obtains the load forecast results for the future set time period based on multi-dimensional measured data and a pre-built load forecast model. Based on the outdoor measured temperature and load forecast results in the multi-dimensional measured data, the target operating mode of the air conditioning unit is determined. Under the target operating mode, dynamic control commands for the air conditioning unit are generated and issued based on the multi-dimensional measured data and load forecast results.

[0088] Step 430: The execution subsystem drives at least some of the execution components in the air conditioning unit to operate in coordination according to the dynamic control instructions.

[0089] The specific implementation of each step in the methods described in the above embodiments has been described in detail in the embodiments of the relevant systems, and will not be elaborated further here.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent air conditioning control system based on dynamic load response, characterized in that, include: The data acquisition subsystem is used to collect multi-dimensional measured data from indoor and outdoor environments and air conditioning units. The processing subsystem is used to obtain the load forecast result for a future set time period based on the multidimensional measured data and a pre-built load forecast model; determine the target operating mode of the air conditioning unit based on the measured outdoor temperature in the multidimensional measured data and the load forecast result; and generate and issue dynamic control instructions for the air conditioning unit based on the multidimensional measured data and the load forecast result under the target operating mode. An execution subsystem is used to drive at least some of the execution components in the air conditioning unit to operate collaboratively according to the dynamic control instructions.

2. The intelligent air conditioning control system based on dynamic load response according to claim 1, characterized in that, The acquisition subsystem includes: Outdoor environmental data acquisition components are used to collect measured outdoor temperature and measured outdoor wind speed. Indoor environmental data acquisition components are used to collect measured indoor temperature, measured indoor humidity, and measured indoor carbon dioxide concentration. The unit operation data acquisition component is used to collect the measured temperature of key parts of the air conditioning unit, the measured operating current of the motor, and the measured opening degree of key valves.

3. The intelligent air conditioning control system based on dynamic load response according to claim 1, characterized in that, The load forecasting model includes: The CNN module is used to convert the multidimensional measured data into a two-dimensional feature matrix, extract the hidden features of the data in the two-dimensional feature matrix, and output key feature vectors. The LSTM module is used to extract the long-term temporal dependencies of the data in the key feature vectors and output preliminary prediction results. The XGBoost module is used to correct the preliminary prediction results based on the key environmental parameters in the multidimensional measured data through ensemble learning, and output the load prediction results.

4. The intelligent air conditioning control system based on dynamic load response according to claim 1, characterized in that, Before obtaining the load forecast results for a future set time period based on the multidimensional measured data and using a pre-built load forecasting model, the processing subsystem is further configured to: Identify and remove abnormal data from the multidimensional measured data caused by sensor failure or extreme weather interference to obtain valid measured data; The effective measured data are standardized to obtain preprocessed multidimensional measured data.

5. The intelligent air conditioning control system based on dynamic load response according to claim 1, characterized in that, The processing subsystem determines the target operating mode of the air conditioning unit based on the measured outdoor temperature from the multidimensional measured data and the load prediction results, including: Determine the target temperature range to which the measured outdoor temperature belongs in the multidimensional measured data; Determine the load type corresponding to the load forecast results; The target operating mode of the air conditioning unit is determined based on the target temperature range and the load type.

6. The intelligent air conditioning control system based on dynamic load response according to claim 5, characterized in that, Based on the target temperature range and the load type, the target operating mode of the air conditioning unit is determined, including: When the target temperature range is below the first set temperature and the load type is heat load, the target operating mode of the air conditioning unit is determined to be the heating mode; When the target temperature range is from the first set temperature to the second set temperature and the load type is primary cooling load, the target operating mode of the air conditioning unit is determined to be natural cooling mode; When the target temperature range is above the second set temperature and the load type is medium cooling load, the target operating mode of the air conditioning unit is determined to be cooling mode; Wherein, the first set temperature is lower than the second set temperature.

7. The intelligent air conditioning control system based on dynamic load response according to claim 1, characterized in that, The multidimensional measured data includes: indoor measured carbon dioxide concentration, outdoor measured wind speed, indoor measured temperature, measured temperature of key parts of the air conditioning unit, measured opening degree of key valves, and indoor measured humidity. The processing subsystem generates dynamic control commands for the air conditioning unit based on the multidimensional measured data and the load forecast results, including: Based on the measured indoor carbon dioxide concentration and the measured outdoor wind speed, the fresh air adjustment amount is determined, and a fresh air control instruction is generated according to the fresh air adjustment amount. Based on the load forecast results, the measured indoor temperature, and the measured opening of the key valves, the adjustment amount of the hot and cold water valve opening is determined, and a hot and cold water valve opening control command is generated according to the adjustment amount of the hot and cold water valve opening. Based on the measured indoor humidity and the measured temperature of key parts of the air conditioning unit, the dehumidification parameter adjustment amount is determined, and a dehumidification control command is generated based on the dehumidification parameter adjustment amount. The fresh air control command, the hot and cold water valve opening control command, and the dehumidification control command are used as dynamic control commands for the air conditioning unit.

8. The intelligent air conditioning control system based on dynamic load response according to claim 7, characterized in that, The multidimensional measured data also includes: the measured operating current of the motor; The processing subsystem is also used to determine the motor's operating state based on the measured operating current of the motor, and when the motor's operating state is detected to be overloaded, control the motor's operating frequency to decrease from the current frequency value to the target frequency value according to the set frequency adjustment step size.

9. The intelligent air conditioning control system based on dynamic load response according to claim 1, characterized in that, The system also includes: a management subsystem; The management subsystem is used to acquire key execution data and execution environment data of the execution subsystem, establish a feedback dataset based on the key execution data and execution environment data, and optimize and update the load prediction model based on the feedback dataset.

10. An intelligent air conditioning control method based on dynamic load response, characterized in that, Based on the intelligent air conditioning control system based on dynamic load response as described in any one of claims 1 to 9, the method includes: The data acquisition subsystem collects multi-dimensional measured data from indoor and outdoor environments and air conditioning units. The processing subsystem obtains the load forecast results for a future set time period based on the multidimensional measured data and the pre-built load forecast model. Based on the outdoor measured temperature in the multidimensional measured data and the load forecast results, the target operating mode of the air conditioning unit is determined. Under the target operating mode, dynamic control commands for the air conditioning unit are generated and issued based on the multidimensional measured data and the load forecast results. By executing the subsystem according to the dynamic control instructions, at least some of the execution components in the air conditioning unit are driven to operate in coordination.

Images