Cooling capacity demand prediction method and device, electronic equipment and storage medium

Abstract

The invention provides a cooling capacity demand prediction method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring internal and external environment data and equipment operation data of a target place by using a deployed Internet of Things sensor; the internal and external environment data and the equipment operation data are preprocessed, and preprocessing comprises data cleaning and normalization processing; feature engineering is carried out according to the preprocessed data, and a feature set used for cold capacity demand prediction is constructed; according to the feature sets corresponding to the one or more target places, a machine learning algorithm is utilized to train a cooling capacity demand prediction model, and hyper-parameters of the cooling capacity demand prediction model are optimized; and utilizing the trained cooling capacity demand prediction model to generate a cooling capacity demand prediction result of the target place in a future time period according to the internal and external environment data and similar date characteristics collected in real time. The cold quantity demand prediction precision can be improved, the model generalization ability is improved, and the model updating and maintenance difficulty is reduced.

Description

Technical Field

[0001] The present application relates to the field of intelligent energy management technology, and in particular to a cooling demand prediction method, device, electronic device and storage medium. Background Art

[0002] With the acceleration of urbanization and the increase in commercial activities, energy consumption management in public places such as shopping malls has received increasing attention. As the core energy equipment of large shopping malls and public buildings, the main function of the cooling station system is to adjust the indoor temperature, and its energy consumption accounts for a large proportion of the total energy consumption. Traditional cooling station systems usually rely on manual experience or simple feedback control to adjust the temperature, which is difficult to effectively adapt to complex and changing environmental factors, resulting in low system operation efficiency and energy waste.

[0003] In order to improve the energy efficiency of the cooling station system, some shopping malls have introduced the Internet of Things (IoT) technology, which uses sensors to collect real-time indoor and outdoor environmental data (such as temperature and humidity) and equipment operation data, and predicts cooling demand based on these data. However, the existing cooling demand prediction technology still has the following shortcomings:

[0004] Most existing models are based on the historical data of a single shopping mall, which makes it difficult to fully capture the complex dynamic changes in the indoor and outdoor environments, resulting in large prediction errors, which in turn affects the temperature control effect and energy utilization efficiency.

[0005] The models of existing systems are usually only applicable to specific shopping malls and cannot be effectively promoted to different regions, climatic conditions or newly built shopping malls. Especially when there is less data from new shopping malls, the model effect is significantly reduced.

[0006] The existing model is unable to effectively capture the mutual influence of cooling demand among multiple shopping malls in the same area, resulting in the system's insufficient ability to respond to local environmental changes.

[0007] The existing system usually requires each shopping mall to update the model independently. The update process is time-consuming and complex to manage. Especially when managing multiple projects nationwide, the difficulty of updating and maintaining the model increases significantly. Summary of the invention

[0008] In view of this, the embodiments of the present application provide a cooling demand prediction method, device, electronic device and storage medium to solve the problems of low cooling demand prediction accuracy, large prediction error, poor model generalization ability and difficulty in model updating and maintenance in the prior art.

[0009] According to a first aspect of an embodiment of the present application, a method for predicting cooling demand is provided, comprising: using deployed Internet of Things sensors to collect internal and external environmental data and equipment operation data of a target place; preprocessing the internal and external environmental data and equipment operation data, wherein the preprocessing includes data cleaning and normalization; performing feature engineering based on the preprocessed data to construct a feature set for cooling demand prediction, wherein the feature set includes physical parameters, geographic location information, similar date features, and newly added features and cross-features of the target place; based on the feature sets corresponding to one or more target places, using a machine learning algorithm to train a cooling demand prediction model, and optimizing the hyperparameters of the cooling demand prediction model; using the trained cooling demand prediction model to generate a cooling demand prediction result for the target place in a future time period based on the real-time collected internal and external environmental data and similar date features.

[0010] According to a second aspect of an embodiment of the present application, a cooling demand prediction device is provided, comprising: an acquisition module, configured to use deployed Internet of Things sensors to collect internal and external environmental data and equipment operation data of a target place; a preprocessing module, configured to preprocess the internal and external environmental data and equipment operation data, wherein the preprocessing includes data cleaning and normalization; a construction module, configured to perform feature engineering based on the preprocessed data, and construct a feature set for cooling demand prediction, wherein the feature set includes physical parameters, geographic location information, similar date features, and newly added features and cross-features of the target place; a training module, configured to train a cooling demand prediction model using a machine learning algorithm based on feature sets corresponding to one or more target places, and optimize the hyperparameters of the cooling demand prediction model; a generation module, configured to use the trained cooling demand prediction model to generate a cooling demand prediction result for the target place in a future time period based on real-time collected internal and external environmental data and similar date features.

[0011] According to a third aspect of an embodiment of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the steps of the above method are implemented when the processor executes the computer program.

[0012] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, which stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0013] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects:

[0014] By utilizing deployed IoT sensors to collect internal and external environmental data and equipment operation data of the target place; preprocessing the internal and external environmental data and equipment operation data, wherein the preprocessing includes data cleaning and normalization; performing feature engineering based on the preprocessed data, and constructing a feature set for cold demand prediction, wherein the feature set includes the physical parameters, geographic location information, similar date features, and newly added features and cross-features of the target place; according to the feature sets corresponding to one or more target places, using machine learning algorithms to train the cold demand prediction model, and optimizing the hyperparameters of the cold demand prediction model; using the trained cold demand prediction model, according to the real-time collected internal and external environmental data and similar date features, generate the cold demand prediction results of the target place in the future time period. This application can improve the accuracy of cold demand prediction, improve the generalization ability of the model, and reduce the difficulty of model update and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0016] Figure 1 It is a flow chart of the cooling demand prediction method provided in the embodiment of the present application;

[0017] Figure 2 is a schematic diagram of the structure of a cooling demand prediction device provided in an embodiment of the present application;

[0018] Figure 3 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0019] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0020] It should be understood that the various steps described in the method implementation of the present application can be performed in different orders and / or performed in parallel. In addition, the method implementation may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this respect.

[0021] With the increase of urbanization and commercial activities, energy consumption in public places has become a core issue, especially in large buildings such as shopping malls. The cooling station system is an important facility for regulating the indoor temperature of public places such as shopping malls. Due to its huge energy consumption, optimizing energy consumption management is of great significance to reducing overall energy consumption.

[0022] At present, some shopping malls’ cooling station systems have begun to introduce Internet of Things (IoT) technology, using sensors to collect indoor and outdoor environmental data, such as temperature, humidity, and equipment operating parameters. Based on this data, the system can predict the cooling demand of the mall, thereby optimizing the operation of the cooling station. However, the existing technical solutions still have many shortcomings, especially in terms of the accuracy of cooling demand prediction, the generalization ability of the model, and the management complexity of the system.

[0023] Existing cold station systems mostly rely on IoT sensors to obtain environmental parameters, and use historical data to build models and predict cooling capacity. This approach is usually based on a single shopping mall model. The main technical means include: collecting indoor and outdoor temperature, humidity and other environmental parameters through IoT sensors, and analyzing them with the operation data of the cold station system. Based on the historical data of the shopping mall, a cold demand forecasting model is built to predict the cooling demand in the future. The forecast results are used to adjust the operation of the cold station system to achieve energy saving.

[0024] Although these technical means have alleviated the problem of energy waste to a certain extent, there are still many shortcomings. The existing technology still has the following problems:

[0025] 1. Insufficient accuracy in cooling demand prediction: Since cooling demand prediction models are usually based on historical data from a single shopping mall, it is difficult to effectively capture complex and changing indoor and outdoor environmental factors. For example, the environment of a shopping mall is affected by multiple factors, such as climate change and changes in passenger flow at different time periods. Existing models cannot fully consider these factors, resulting in large cooling demand prediction errors, which in turn affects the temperature control effect and causes energy waste.

[0026] 2. Poor generalization of the model: Existing models are usually designed for a single shopping mall and lack generalization capabilities. The model has poor adaptability under different geographical or climatic conditions, especially for new shopping malls with less data. The existing model has more limited effects and cannot be extended to other scenarios.

[0027] 3. Weak regional trend learning ability: The existing system cannot capture the mutual influence of cooling demand between multiple shopping malls in the same region. For example, multiple shopping malls in the same region may be affected by similar environmental changes, but the existing model does not fully consider these regional factors, resulting in a slow response to local environmental changes and an inability to dynamically adjust cooling demand forecasts.

[0028] 4. Complex system update management: When the existing system is applied in multiple shopping malls, it is very complicated to update and manage the model. Usually, each shopping mall needs to update the model independently. This traditional update method is time-consuming and difficult to manage, especially when it involves multiple projects across the country, the management complexity and maintenance costs increase significantly.

[0029] In summary, although the current cold station system has utilized technical means such as IoT sensors, it still has many deficiencies in cold demand forecasting, model generalization, regional learning, and system management. The existing model cannot effectively capture complex environmental changes, has low prediction accuracy, and poor generalization ability, making it difficult to promote and apply in different scenarios. In addition, the management and update of the system are complex, especially when unified management is required across multiple shopping malls. These problems have led to low energy efficiency of the cold station system and its energy-saving potential cannot be fully utilized.

[0030] In view of the problems existing in the above-mentioned prior art, this application proposes a cooling demand prediction system based on full data training, aiming to solve the problems of poor generalization ability, low prediction accuracy and complex model management in traditional cooling demand prediction systems. By integrating IOT data from shopping malls across the country and conducting comprehensive feature engineering, the system can more accurately predict cooling demand and improve the intelligence and adaptability of the model. The system achieves more accurate and intelligent cooling demand prediction through the following technical solutions:

[0031] Full data integration model: By integrating the IoT data of shopping malls across the country and adding features such as the mall name, province, city, volume, climate zone, etc., a basic model with strong generalization capabilities is established, and the correlation between shopping malls is captured through regional features to improve the prediction effect.

[0032] Feature engineering and cross-features: Through cross-features in the same area, such as outdoor temperature difference and traffic flow change rate, the mutual influence between shopping malls can be captured to improve the regional adaptability of the model.

[0033] Similar date features: Based on the temperature, humidity and other parameters of the weather forecast, find the date in history that is most similar to the current conditions, and input the cooling and environmental differences of the similar date as features into the model to improve the model's ability to respond to short-term changes.

[0034] Simplified model update management: With a fully unified basic model, all shopping malls share one model. When new data or projects are launched, only one national model needs to be updated, which simplifies model management and improves system update efficiency.

[0035] The cooling demand prediction system based on full data training proposed in this application mainly includes the following components:

[0036] 1. Data acquisition module: obtain environmental data and equipment operation status inside and outside the mall through IOT sensors.

[0037] 2. Data preprocessing and feature engineering module: Clean, normalize and construct cross-features of raw data to improve data availability. Use historical similar dates to find the date closest to current weather conditions to enhance the model's ability to respond to short-term changes.

[0038] 3. Full-volume cold volume prediction model training module: A basic model is trained based on the historical data of all shopping malls across the country, integrating multiple features and capturing the correlation between regions.

[0039] 4. Cooling demand prediction module: predict future cooling demand through fusion model and provide control data for the operation of the shopping mall cooling station system.

[0040] 5. Model update and management module: A simplified unified model update mechanism allows demand forecasting for all shopping malls to be achieved by updating only one model, reducing management difficulty.

[0041] The contents of the technical solution of the present application are described in detail below with reference to the accompanying drawings and specific embodiments.

[0042] Figure 1 is a flow chart of the cooling demand prediction method provided in the embodiment of the present application. Figure 1 As shown, the cooling demand prediction method may specifically include:

[0043] S101, using deployed IoT sensors to collect internal and external environmental data and equipment operation data of the target location;

[0044] S102, preprocessing the internal and external environment data and the equipment operation data, wherein the preprocessing includes data cleaning and normalization processing;

[0045] S103, performing feature engineering based on the preprocessed data to construct a feature set for cooling demand prediction, wherein the feature set includes physical parameters of the target location, geographic location information, similar date features, and newly added features and cross features of the target location;

[0046] S104, training a cooling demand prediction model using a machine learning algorithm according to a feature set corresponding to one or more target locations, and optimizing hyperparameters of the cooling demand prediction model;

[0047] S105, using the trained cooling demand prediction model, based on the real-time collected internal and external environmental data and similar date characteristics, generates a cooling demand prediction result for the target location in a future time period.

[0048] In some embodiments, the system collects real-time internal and external environmental data and equipment operation data of the target location through Internet of Things (IoT) sensors deployed at the target location, so as to dynamically grasp the cooling demand. Specifically, IoT sensors can be installed in various areas of the shopping mall to monitor key indoor and outdoor environmental parameters, including but not limited to:

[0049] Temperature: Sensors record real-time temperature changes inside and outside the mall, providing basic data for cooling demand forecasting;

[0050] Humidity: Collecting indoor and outdoor humidity data and combining it with temperature data can more accurately reflect the comfort requirements of the environment;

[0051] Crowd flow: The flow of people in the mall is monitored through technologies such as infrared sensors or cameras. Changes in the flow of people directly affect the mall's demand for cooling. Especially when the flow of people fluctuates greatly, the system needs to dynamically adjust the cooling supply.

[0052] In addition, with regard to the operating status of the cold station system, this application collects the following equipment operating data through IoT devices:

[0053] Chiller power: Real-time monitoring of the chiller power output of the refrigeration system to understand the actual capacity of the current cooling supply;

[0054] Cooling flow: Collect the flow data of cooling water or refrigerant flowing through the cooling station system to reflect the efficiency of cooling transfer and the load of the system.

[0055] In practical applications, in order to ensure the real-time and reliability of data, the system adopts a high-frequency collection strategy. The sensor can collect data according to the set time interval (such as every few minutes) and report in time when sudden environmental changes occur, so that the system can respond quickly and adjust the cooling supply strategy.

[0056] Through the deployment and data collection of the above IoT sensors, the system can fully grasp the environmental status of the mall and the operating status of the cold station system, and provide high-precision input data for subsequent cold demand forecasting. During the data collection process, all data will be transmitted to the cloud or local server in real time for the system to perform data preprocessing and feature extraction, ensuring the dynamic and accurate cold demand forecasting.

[0057] In some embodiments, preprocessing of internal and external environment data and equipment operation data includes:

[0058] Perform noise filtering, missing value filling and abnormal data elimination on internal and external environmental data and equipment operation data to remove invalid data and abnormal values;

[0059] Normalize the cleaned data to unify the data ranges from different sources and different dimensions;

[0060] Convert the normalized data into a unified format to meet the format requirements of feature engineering.

[0061] Specifically, the embodiment of the present application pre-processes the collected internal and external environmental data and equipment operation data to improve the validity and consistency of the data and ensure the accuracy of subsequent cooling demand prediction.

[0062] First, noise filtering is performed on the collected internal and external environmental data and equipment operation data. Since IoT sensors may be subject to external interference during actual use, such as sensor hardware failure, network jitter, etc., some data may contain noise or abnormal fluctuations. To ensure the reliability of the data, the system uses filtering algorithms (such as mean filtering or median filtering) to remove these noises and retain the valid information of the data.

[0063] Furthermore, in the actual data collection process, data may be missing in certain periods, such as temporary sensor failure or network transmission interruption. To ensure the integrity of the data, the system will fill in the missing values. Specifically, an interpolation algorithm can be used to estimate the missing values ​​based on the data information of the adjacent time period, or the missing values ​​can be filled in with data from similar historical scenarios to ensure the continuity of the data in the time series.

[0064] Furthermore, in the data cleaning process, it is also necessary to identify and remove abnormal data. These abnormal data may be caused by sensor failure or emergencies. The system sets a reasonable threshold range (such as the normal fluctuation range of indoor temperature, human flow and other data) to remove abnormal data that exceeds the threshold. Abnormal data removal helps to avoid the impact of extreme data on the cooling capacity prediction model and improve the accuracy of the model.

[0065] Furthermore, the cleaned data is normalized. Since data from different sources (such as temperature, humidity, chiller power, etc.) have different dimensions and value ranges, directly using these data will lead to unbalanced weights of various features during model training. To this end, the system uses normalization methods (such as Min-Max normalization or Z-score standardization) to convert data of different features into the same numerical range, so that each feature can be comparable in the cooling capacity prediction model, improving the training effect of the model.

[0066] Furthermore, the normalized data needs to be further formatted to meet the input format requirements of feature engineering. Specifically, the system converts different types of environmental data and device data into a unified data structure and arranges them in the form of a time series so that the data can be smoothly input into the feature extraction module. This step ensures the consistency of the data format and facilitates the subsequent training and prediction of the machine learning model.

[0067] Through the preprocessing steps of the above-mentioned embodiment, the system can effectively clean and standardize the collected internal and external environmental data and equipment operation data, ensure data quality, and provide high-quality input data for feature engineering and cooling demand prediction models, thereby improving the prediction accuracy and operation efficiency of the entire system.

[0068] In some embodiments, feature engineering is performed based on the preprocessed data, including:

[0069] According to the current weather conditions, the weather variables in the historical data are compared with the weather variables of the current weather conditions by using the distance measurement method to determine the similar historical date closest to the current weather conditions;

[0070] Relevant features are extracted from weather variables corresponding to similar historical dates, and similar date features are constructed based on the extracted relevant features. The similar date features are used as input features for cooling demand prediction.

[0071] Specifically, the embodiment of the present application will also perform feature engineering based on the preprocessed data. The system constructs similar date features to improve the responsiveness of the cooling demand forecasting model to short-term weather changes.

[0072] First, the system looks for the historical date that is most similar to the current weather conditions based on the weather conditions of the day, including but not limited to the maximum temperature, minimum temperature, and humidity. To achieve this goal, the system uses a distance metric, such as Euclidean distance, to quantify the difference between the current weather variables and the historical weather variables. Euclidean distance can take into account multi-dimensional weather data (such as temperature, humidity, etc.) to determine the similar historical date that is closest to the current conditions.

[0073] Furthermore, after determining a similar historical date, the system extracts relevant features from the weather data corresponding to the date. The specific extracted features include but are not limited to cooling capacity, temperature difference, humidity difference, and pedestrian flow difference.

[0074] Cooling capacity: Extract cooling capacity demand data of the same time period on similar historical dates to reflect the actual cooling capacity demand under similar conditions.

[0075] Temperature difference: Calculate and extract the indoor and outdoor temperature difference of the historical date to reflect the impact of ambient temperature changes on cooling demand on that date.

[0076] Humidity difference: Extract the humidity change on that date and combine it with the temperature difference characteristics to further capture the effect of humidity on cooling demand.

[0077] Difference in passenger flow: Extract the difference between the passenger flow changes on similar dates and the current passenger flow to evaluate the impact of passenger flow fluctuations on cooling demand.

[0078] These features reflect the changes in cooling demand that the system may encounter under similar weather conditions, and are provided as new input features to the cooling demand prediction model.

[0079] Furthermore, the extracted data such as cooling capacity, temperature difference, humidity difference and traffic difference are integrated to construct similar date features as one of the important inputs of the prediction model. Similar date features help the model to more accurately reflect the correlation between current weather conditions and historical weather conditions, thereby improving the prediction accuracy of short-term cooling demand fluctuations.

[0080] Through the method of the above-mentioned embodiment of the present application, the system utilizes similar date characteristics to enhance the flexibility and adaptability of cooling demand forecasting, so that when facing short-term weather changes, the model can more accurately predict cooling demand, thereby optimizing the control strategy of the cold station system and avoiding energy waste.

[0081] In some embodiments, the newly added features in the feature set of the embodiments of the present application include the following:

[0082] Different from the single shopping mall model, the embodiment of the present application also adds the following features: mall name, province, city, mall volume, height, climate zone, number of above-ground floors, number of underground floors, floor area, number of imports and exports, window-to-wall ratio, proportion of each business format, etc.

[0083] Among them, the differences in the physical, geographical and operational environments of shopping malls have an important impact on the prediction of cooling demand, and these added features can effectively capture this difference. Features such as climate zone, window-to-wall ratio, and the proportion of various business formats can reflect the external environment and structural characteristics of the shopping mall, which is crucial for accurately predicting energy consumption demand.

[0084] In some embodiments, feature engineering is performed based on the preprocessed data, including:

[0085] According to the outdoor temperature difference and indoor temperature difference between the target place and other places in the same area, the temperature difference related features in the same area are constructed;

[0086] According to the outdoor humidity difference and indoor humidity difference between the target place and other places in the same area, the humidity difference related features in the same area are constructed;

[0087] According to the change rate and difference of the flow of people between the target place and other places in the same area, the relevant characteristics of the flow of people in the same area are constructed;

[0088] According to the difference in weather impact between the target location and other locations in the same area, weather-related features in the same area are constructed;

[0089] According to the regional differences in cooling capacity change rate and cooling capacity demand between the target location and other locations in the same area, the cooling capacity related characteristics in the same area are constructed;

[0090] According to the temperature difference related features in the same area, the humidity difference related features in the same area, the pedestrian flow related features in the same area, the weather related features in the same area and the cooling capacity related features in the same area, cross features are constructed, where the cross features are used to reflect the correlation between the target place and other places.

[0091] Specifically, in order to enhance the generalization ability and regional adaptability of the model and enable it to learn the trend information of adjacent shopping malls in the region, more cross-features can be generated to capture the connection between shopping malls in the same region. The embodiment of the present application also introduces multiple cross-features to capture the correlation of shopping malls in the same region. The following are the cross-features and their calculation methods provided by the embodiment of the present application:

[0092] 1) Characteristics of temperature differences in the same region

[0093] Outdoor temperature difference in the same area:

[0094] ΔT out =T out,商场A -T out,平均区域

[0095] Among them, T out,平均区域 It is the average outdoor temperature of all malls in the area. The difference between the outdoor temperature of each mall and the average value can reflect whether the mall is in abnormal weather conditions.

[0096] Indoor temperature difference in the same area:

[0097] ΔT in =T in,商场A -T in,平均区域

[0098] The calculation method is similar to the outdoor temperature, reflecting the difference in temperature control inside the mall and the surrounding malls.

[0099] 2) Characteristics of humidity differences in the same region

[0100] Outdoor humidity difference in the same area:

[0101] ΔH out =Hout,商场A -H out,平均区域

[0102] Among them, H out,平均区域 It is the average outdoor humidity of the mall in the area, which is used to capture abnormal humidity of the mall's external environment.

[0103] Indoor humidity difference in the same area:

[0104] ΔH in =H in,商场A -H in,平均区域

[0105] 3) Characteristics of passenger flow in the same area

[0106] Change rate of traffic in the same area:

[0107]

[0108] It is used to calculate the rate of change of pedestrian flow in the shopping mall, so as to predict the impact of instantaneous changes on the demand for cooling capacity.

[0109] Differences in traffic volume in the same area:

[0110] ΔF region =F 商场A -F 区域平均

[0111] It can be used to measure the difference between the flow of people in a shopping mall and the flow of people in shopping malls in the region.

[0112] 4) Weather-related characteristics in the same region

[0113] Differences in weather impacts in the same region:

[0114] Use weather forecast data, such as wind speed, precipitation, air pressure, etc., to calculate the differences between each mall and other malls in the same area. For example:

[0115] ΔW 风速 =W 风速,商场A -W 风速,区域平均

[0116] 5) Characteristics of cooling capacity in the same area

[0117] Cooling capacity change rate:

[0118]

[0119] Reflects the instantaneous rate of change of cooling demand, helping to identify periods of drastic fluctuations in cooling demand.

[0120] Regional differences in cooling demand:

[0121] ΔQ region =Q商场A -Q 区域平均

[0122] Used to measure the difference in cooling demand between a shopping mall and other shopping malls in the area.

[0123] In some embodiments, a machine learning algorithm is used to train a cooling demand prediction model, and hyperparameters of the cooling demand prediction model are optimized, including:

[0124] Determine a machine learning algorithm for cooling demand prediction, and build a cooling demand prediction model based on the machine learning algorithm;

[0125] The cooling demand prediction model is trained using the feature set to obtain a preliminary cooling demand prediction model;

[0126] The random search optimization algorithm is used to optimize the hyperparameters in the preliminary cooling demand prediction model, where the hyperparameters include learning rate, tree depth and leaf node number.

[0127] In the process of hyperparameter optimization, the time series cross-validation method is used to verify the cooling demand prediction model;

[0128] The optimized cooling demand prediction model is evaluated using the preset prediction performance evaluation index to obtain the cooling demand prediction model finally used for cooling demand prediction.

[0129] Specifically, the embodiment of the present application uses a machine learning algorithm to train the cooling demand prediction model and optimizes the hyperparameters of the model to improve the prediction accuracy and generalization ability.

[0130] First, determine a machine learning algorithm suitable for cooling demand prediction. In this embodiment, the decision tree-based ensemble learning model LightGBM is selected for cooling demand prediction. The LightGBM model is suitable for processing complex environmental data for cooling demand prediction in shopping malls due to its fast training speed and efficient processing capabilities for large-scale data sets.

[0131] In addition, the system can also select other machine learning models according to specific needs, such as XGBoost and CatBoost, which are also based on the gradient boosting algorithm and are suitable for cold volume prediction. In addition, when it is necessary to process complex time series data, deep learning models such as long short-term memory networks (LSTM) or gated recurrent units (GRU) can also be selected to better capture long-term dependencies in time series.

[0132] Furthermore, the selected cooling demand forecasting model is trained using the feature set after preprocessing and feature engineering. Through feature engineering, the system has constructed multi-dimensional data including physical environment, weather conditions, similar date features, etc. After feature screening, the most important features for cooling demand forecasting are extracted. For example, the geographical location of the mall, indoor and outdoor temperature, humidity, traffic flow, equipment operating parameters, etc.

[0133] Furthermore, during the training process, the system generates a preliminary cooling demand prediction model by inputting these filtered feature data. This preliminary model can provide a prediction of the future cooling demand of the shopping mall, but in order to further improve the prediction accuracy and adaptability of the model, the model's hyperparameters need to be optimized.

[0134] Furthermore, in order to optimize the performance of the cooling demand prediction model, this embodiment uses a randomized search optimization algorithm (RandomizedSearchCV) to tune multiple hyperparameters in the model. The optimization of hyperparameters involves key parameters in model training, including:

[0135] Learning rate: controls the learning speed of the model. Too large a learning rate may lead to unstable model training, while too small a learning rate may lead to too long training time.

[0136] Tree depth: The maximum depth of the decision tree controls the complexity of the model. Too large a depth may lead to overfitting, while too small a depth may lead to underfitting.

[0137] Number of leaf nodes: The maximum number of leaf nodes in each tree, which affects the model's ability to learn details.

[0138] The random search optimization algorithm finds the optimal parameter combination by randomly sampling within the range of these hyperparameters for training and evaluating the performance of the model.

[0139] Furthermore, during the hyperparameter optimization process, the system uses the time series cross-validation method to validate the cooling demand forecasting model. The characteristic of time series cross-validation is that the data is divided into training set and validation set in chronological order, ensuring that the data in the validation set is always later than the training set, thereby avoiding future information leakage problems. In multiple cross-validations, the system continuously adjusts the hyperparameters to ensure that the model performs consistently on data in different time periods and has good generalization capabilities.

[0140] Furthermore, after completing the hyperparameter optimization, the system uses preset prediction performance evaluation indicators to evaluate the optimized cooling demand prediction model. The performance evaluation indicators used in this embodiment include but are not limited to the following indicators:

[0141] Root mean square error (RMSE): It is used to measure the deviation between the model prediction value and the actual value, reflecting the prediction accuracy.

[0142] Mean Absolute Percentage Error (MAPE): It is used to evaluate the percentage of prediction error relative to the actual value and can measure the relative error of the prediction results.

[0143] Through the above evaluation indicators, the system determines whether the optimized cooling demand prediction model has achieved the expected prediction performance, and finally obtains the cooling demand prediction model used for actual cooling demand prediction.

[0144] Through the method of the above-mentioned embodiment of the present application, this embodiment constructs an efficient and accurate refrigeration demand prediction model through the steps of selecting a machine learning algorithm, optimizing hyperparameters, and cross-validating time series, thereby improving the model's accuracy in predicting the refrigeration demand of shopping malls and its adaptability to environmental changes in different time periods.

[0145] In some embodiments, the cooling demand prediction model is validated using a time series cross validation method, including:

[0146] According to the time sequence of the data, the historical data is divided into multiple training sets and validation sets based on the progressive order of the time periods;

[0147] In each cross-validation step, the cooling demand forecasting model is trained using the data of the early time period, and the trained cooling demand forecasting model is validated using the data of the subsequent time period;

[0148] According to the progressive cross-validation method, the training and validation steps are repeatedly performed on the cooling demand prediction model so that the data of each time period are used as the training set and the validation set respectively.

[0149] Specifically, in order to verify the generalization ability and accuracy of the cooling demand forecasting model, the time series cross-validation method was used to verify the model. This method is particularly suitable for cooling demand data with time series characteristics to ensure that the model can maintain good forecasting performance in different time periods and avoid model overfitting problems caused by future data leakage.

[0150] The system first divides the data set into multiple training sets and validation sets according to the time sequence of historical data. The division process adopts a progressive time period division strategy, that is, the time range of the training set is gradually expanded in chronological order, and the subsequent time period data is used as the validation set. For example, assuming there is a continuous period of historical data (such as cooling demand data for the past year), the data of the first 3 months can be used as the first training set, and the data of the subsequent 1 month can be used as the validation set. Subsequently, the training set can be expanded to the first 4 months, and the subsequent 1 month can be used as the validation set again, and so on.

[0151] This time-series-based division method ensures that future information is not included in the validation set, preventing future data from leaking into the training process, thereby ensuring the model's generalization ability over time series.

[0152] Furthermore, in each cross-validation step, the cooling demand forecasting model is trained using the training set data of the early time period, and after the training is completed, the model is validated using the validation set data of the subsequent time period. In this way, the system can evaluate the accuracy of the model in predicting future cooling demand. As the time period progresses, the data of each time period can be used for model training and validation respectively.

[0153] For example, in the first round of cross-validation, the model is trained using data from the first three months, and then validated using data from the fourth month; in the second round, the training set is expanded to the first four months, and the validation set is data from the fifth month. This process is repeated until data from all time periods have been trained and validated.

[0154] Furthermore, the progressive cross-validation method aims to ensure that the model has good generalization ability throughout the entire time series. By continuously incorporating more historical data into the training set and using data from subsequent time periods for verification, the system can ensure that the prediction performance of the cooling demand forecasting model is always stable in different time periods.

[0155] Specifically, in each cross-validation, the system records the performance of the model on the validation set, including key indicators such as prediction error, such as root mean square error (RMSE) and mean absolute percentage error (MAPE). After multiple validations, the system calculates the average error in these validation steps to evaluate the overall prediction performance of the model.

[0156] Furthermore, to ensure the reliability of verification, the system strictly divides the data in chronological order to prevent future data from leaking to the model during the training process. In this way, the model will not "see" the data to be predicted in advance, which effectively prevents the model from overfitting and ensures the effectiveness of the model in actual use.

[0157] In addition, in each round of training and validation, the system adjusts parameters based on the current validation results to ensure that the prediction performance in different time periods meets the expected standards.

[0158] Through the method of the above-mentioned embodiment of the present application, through this time series-based cross-validation method, the system can ensure the applicability and stability of the cooling demand prediction model in different time periods, improve the time generalization ability of the model, avoid the prediction inaccuracy problem caused by time data leakage, and thus improve the actual application effect of the model.

[0159] In some embodiments, the optimized cooling demand forecasting model is evaluated using a preset forecasting performance evaluation index, including:

[0160] According to the results of each verification, the prediction performance evaluation index of the cooling demand prediction model is calculated, and the prediction performance of the cooling demand prediction model is evaluated by the calculated prediction performance evaluation index, wherein the prediction performance evaluation index includes root mean square error, mean absolute percentage error and symmetric mean percentage error.

[0161] Specifically, the embodiment of the present application uses preset prediction performance evaluation indicators to evaluate the optimized cooling demand forecasting model to ensure the prediction accuracy and robustness of the model in practical applications. In order to comprehensively evaluate the performance of the cooling demand forecasting model, the system uses a number of common prediction performance evaluation indicators, including root mean square error (RMSE), mean absolute percentage error (MAPE), and symmetric mean percentage error (SMAPE). These indicators evaluate the prediction effect of the model from different angles, as follows:

[0162] RMSE (root mean square error): used to measure the deviation between the model prediction value and the actual value. RMSE can reflect the absolute value of the model prediction error. The smaller the deviation, the better the prediction performance of the model.

[0163] MAPE (mean absolute percentage error): used to measure the percentage of prediction error relative to the actual value. MAPE expresses the error in percentage form and is suitable for evaluating the relative error of the model in prediction scenarios of different magnitudes. By calculating the error ratio of the predicted value to the actual value, MAPE can provide an effective evaluation of the overall accuracy of the model.

[0164] SMAPE (Symmetric Mean Percentage Error): SMAPE improves the robustness of model evaluation by balancing positive and negative errors. SMAPE can effectively reduce the impact of extreme errors on evaluation results and further improve the adaptability of the model under different data ranges.

[0165] Furthermore, during each cross-validation process, the system calculates the above evaluation indicators based on the actual data of the validation set and the prediction results of the model. First, by comparing the prediction results of all data in the validation set with the true values, the corresponding RMSE, MAPE and SMAPE values ​​are calculated respectively.

[0166] Furthermore, in the verification results of each time period, the system will first calculate RMSE to measure the prediction deviation of the model in that time period; then calculate MAPE to evaluate the percentage of the model prediction error relative to the actual value, which is particularly suitable for scenarios with large changes in cooling demand values; finally, the system uses SMAPE to evaluate the balance between positive and negative errors to ensure that the model performs consistently in different error directions.

[0167] Furthermore, after multiple validation steps are completed, the system will summarize and average the RMSE, MAPE, and SMAPE calculated in each round of validation to generate the overall performance of the model during the entire validation process. Based on the summary results of these evaluation indicators, the system can judge the prediction accuracy and robustness of the model.

[0168] If the RMSE is low, it means that the model performs well in absolute error; if the MAPE is low, it means that the model can maintain high accuracy in relative error; and a decrease in SMAPE indicates that the model has a better balance in different error directions.

[0169] Furthermore, based on the performance evaluation results, if a certain indicator (such as RMSE or MAPE) performs poorly, the system will further tune the model according to the specific situation. The tuning process may include adjusting hyperparameters, optimizing feature engineering, etc., until all evaluation indicators meet the preset performance requirements.

[0170] Through the method of the above-mentioned embodiment of the present application, through this multi-dimensional performance evaluation method, the system can ensure the high accuracy and robustness of the cooling demand prediction model in practical applications, provide reliable cooling demand prediction results for the shopping mall's refrigeration station system, avoid unnecessary energy waste and improve energy efficiency.

[0171] It should be noted that all target places in the country (such as shopping malls) can share a cooling demand forecasting model. The system only needs to update one cooling demand forecasting model, without having to update each shopping mall independently, which simplifies the model management process and reduces the model maintenance cost. In addition, the system can be extended to the cooling demand forecasting of other types of public places (such as office buildings, industrial plants, etc.), and introduce more external features, such as meteorological data, building materials, etc., to further improve the prediction effect of the model.

[0172] According to the technical solution provided by the present application, the technical solution of the present application has at least the following advantages:

[0173] 1. Full cooling capacity prediction model: This application constructs a cooling capacity prediction model with strong generalization ability based on data from shopping malls across the country, which effectively improves the accuracy of cross-regional predictions.

[0174] 2. Cross-analysis of regional features: By introducing cross-features such as temperature difference, humidity difference and traffic change rate in the same area, the model can better learn the correlation between shopping malls and improve regional adaptability.

[0175] 3. Enhanced features of similar dates: Find similar historical dates through weather forecast parameters, and combine their cooling and environmental differences as feature inputs to enhance the accuracy of short-term forecasts.

[0176] 4. Simplified model update management: All shopping malls share one model, and only the entire model needs to be updated during updates, which greatly reduces the complexity of model management and improves the maintenance efficiency of the system.

[0177] Therefore, through the above-mentioned technical solution of the present application, the present application can provide accurate and intelligent cooling demand prediction, significantly improve energy efficiency, and be widely applicable to cooling station systems in different regions and locations.

[0178] The following is an embodiment of the device of the present application, which can be used to execute the embodiment of the method of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method of the present application.

[0179] Figure 2 Schematic diagram of the structure of the cooling demand prediction device provided in the embodiment of the present application. Figure 2 As shown, the cooling demand prediction device comprises:

[0180] The collection module 201 is configured to collect internal and external environment data and equipment operation data of the target location using deployed IoT sensors;

[0181] The preprocessing module 202 is configured to preprocess the internal and external environment data and the equipment operation data, wherein the preprocessing includes data cleaning and normalization processing;

[0182] A construction module 203 is configured to perform feature engineering based on the preprocessed data to construct a feature set for cooling demand prediction, wherein the feature set includes physical parameters of the target location, geographic location information, similar date features, and newly added features and cross features of the target location;

[0183] The training module 204 is configured to train the cooling demand prediction model using a machine learning algorithm according to a feature set corresponding to one or more target locations, and optimize hyperparameters of the cooling demand prediction model;

[0184] The generation module 205 is configured to use the trained cooling demand prediction model to generate cooling demand prediction results for the target location in a future time period based on real-time collected internal and external environmental data and similar date features.

[0185] In some embodiments, Figure 2 The preprocessing module 202 performs noise filtering, missing value filling and abnormal data elimination on the internal and external environmental data and equipment operation data to remove invalid data and abnormal values; normalizes the cleaned data to unify the data ranges from different sources and different dimensions; and converts the normalized data into a unified format to meet the format requirements of feature engineering.

[0186] In some embodiments, Figure 2 The construction module 203 compares the weather variables in the historical data with the weather variables of the current weather conditions according to the current weather conditions by using a distance measurement method, and determines the similar historical date closest to the current weather conditions; extracts relevant features from the weather variables corresponding to the similar historical dates, and constructs similar date features based on the extracted relevant features, and uses the similar date features as input features for cooling demand prediction.

[0187] In some embodiments, Figure 2 The construction module 203 constructs the temperature difference related features in the same area according to the outdoor temperature difference and indoor temperature difference between the target place and other places in the same area; constructs the humidity difference related features in the same area according to the outdoor humidity difference and indoor humidity difference between the target place and other places in the same area; constructs the pedestrian flow related features in the same area according to the pedestrian flow change rate and pedestrian flow difference between the target place and other places in the same area; constructs the weather related features in the same area according to the weather impact difference between the target place and other places in the same area; constructs the cooling capacity related features in the same area according to the cooling capacity change rate and the regional difference in cooling capacity demand between the target place and other places in the same area; constructs the cross-features according to the temperature difference related features in the same area, the humidity difference related features in the same area, the pedestrian flow related features in the same area, the weather related features in the same area and the cooling capacity related features in the same area, wherein the cross-features are used to reflect the correlation between the target place and other places.

[0188] In some embodiments, Figure 2 The training module 204 determines the machine learning algorithm for cooling demand prediction, and constructs a cooling demand prediction model according to the machine learning algorithm; uses the feature set to train the cooling demand prediction model to obtain a preliminary cooling demand prediction model; uses the random search optimization algorithm to optimize the hyperparameters in the preliminary cooling demand prediction model, wherein the hyperparameters include the learning rate, the depth of the tree, and the number of leaf nodes; in the process of optimizing the hyperparameters, uses the time series cross-validation method to verify the cooling demand prediction model; uses the preset prediction performance evaluation index to evaluate the optimized cooling demand prediction model, and obtains the cooling demand prediction model finally used for cooling demand prediction.

[0189] In some embodiments, Figure 2 The training module 204 divides the historical data into multiple training sets and validation sets based on the progressive order of time periods according to the time sequence of the data; in each cross-validation step, the data of the early time period is used to train the cooling demand prediction model, and the data of the subsequent time period is used to verify the trained cooling demand prediction model; according to the progressive cross-validation method, the training and validation steps are repeated on the cooling demand prediction model so that the data of each time period is used as the training set and the validation set respectively.

[0190] In some embodiments, Figure 2 The training module 204 calculates the prediction performance evaluation index of the cooling demand prediction model according to the result of each verification, and uses the calculated prediction performance evaluation index to evaluate the prediction performance of the cooling demand prediction model, wherein the prediction performance evaluation index includes the root mean square error, the mean absolute percentage error and the symmetric mean percentage error.

[0191] Figure 3 Schematic diagram of the structure of the electronic device 3 provided in the embodiment of the present application. Figure 3 As shown, the electronic device 3 of this embodiment includes: a processor 301, a memory 302, and a computer program 303 stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program 303, the steps in the above-mentioned method embodiments are implemented. Alternatively, when the processor 301 executes the computer program 303, the functions of the modules / units in the above-mentioned device embodiments are implemented.

[0192] Exemplarily, the computer program 303 may be divided into one or more modules / units, which are stored in the memory 302 and executed by the processor 301 to complete the present application. The one or more modules / units may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program 303 in the electronic device 3.

[0193] The electronic device 3 may be a desktop computer, a notebook, a PDA, a cloud server, or other electronic device. The electronic device 3 may include, but is not limited to, a processor 301 and a memory 302. Those skilled in the art will appreciate that Figure 3 It is only an example of the electronic device 3 and does not constitute a limitation of the electronic device 3. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0194] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0195] The memory 302 may be an internal storage unit of the electronic device 3, for example, a hard disk or memory of the electronic device 3. The memory 302 may also be an external storage device of the electronic device 3, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 3. Further, the memory 302 may also include both an internal storage unit and an external storage device of the electronic device 3. The memory 302 is used to store computer programs and other programs and data required by the electronic device. The memory 302 may also be used to temporarily store data that has been output or is to be output.

[0196] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0197] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0198] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments applied for herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0199] In the embodiments provided in the present application, it should be understood that the disclosed devices / computer equipment and methods can be implemented in other ways. For example, the device / computer equipment embodiments described above are only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation. Multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0200] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0201] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0202] If the integrated module / unit is implemented in the form of 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 present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. The computer program may include computer program code, which may be in source code form, object code form, executable file or some intermediate form. Computer-readable media may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0203] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for predicting cooling demand, characterized in that: include: Use the deployed IoT sensors to collect internal and external environmental data and equipment operation data of the target location; Preprocessing the internal and external environment data and equipment operation data, wherein the preprocessing includes data cleaning and normalization; Performing feature engineering on the preprocessed data to construct a feature set for cooling demand prediction, wherein the feature set includes physical parameters of the target location, geographic location information, similar date features, and newly added features and cross-features of the target location; According to the feature set corresponding to one or more of the target places, a cooling demand prediction model is trained using a machine learning algorithm, and hyper parameters of the cooling demand prediction model are optimized; The trained cooling demand prediction model is used to generate cooling demand prediction results for the target location in a future time period based on real-time collected internal and external environmental data and the similar date characteristics.

2. The method according to claim 1, characterized in that The preprocessing of the internal and external environment data and the equipment operation data includes: Perform noise filtering, missing value filling and abnormal data elimination on the internal and external environment data and the equipment operation data to remove invalid data and abnormal values; Normalize the cleaned data to unify the data ranges from different sources and different dimensions; The normalized data is converted into a unified format to meet the format requirements of the feature engineering.

3. The method according to claim 1, characterized in that The feature engineering is performed based on the preprocessed data, including: According to the current weather conditions, using a distance measurement method to compare the weather variables in the historical data with the weather variables of the current weather conditions, and determine the similar historical date closest to the current weather conditions; Relevant features are extracted from the weather variables corresponding to the similar historical dates, and the similar date features are constructed according to the extracted relevant features, and the similar date features are used as input features for cooling demand prediction.

4. The method according to claim 1, characterized in that: The feature engineering is performed based on the preprocessed data, including: According to the outdoor temperature difference and indoor temperature difference between the target place and other places in the same area, the temperature difference related features in the same area are constructed; According to the outdoor humidity difference and indoor humidity difference between the target place and other places in the same area, the humidity difference related features in the same area are constructed; According to the change rate and difference of the flow of people between the target place and other places in the same area, the relevant characteristics of the flow of people in the same area are constructed; According to the difference in weather impact between the target location and other locations in the same area, weather-related features in the same area are constructed; According to the regional differences in cooling capacity change rate and cooling capacity demand between the target location and other locations in the same area, the cooling capacity related characteristics in the same area are constructed; The cross-feature is constructed based on the temperature difference related features in the same area, the humidity difference related features in the same area, the pedestrian flow related features in the same area, the weather related features in the same area, and the cooling capacity related features in the same area, wherein the cross-feature is used to reflect the correlation between the target place and other places.

5. The method according to claim 1, characterized in that The method of using a machine learning algorithm to train the cooling demand prediction model and optimizing the hyperparameters of the cooling demand prediction model includes: Determining a machine learning algorithm for cooling demand prediction, and constructing the cooling demand prediction model according to the machine learning algorithm; Using the feature set to train the cooling demand prediction model to obtain a preliminary cooling demand prediction model; Optimizing hyperparameters in the preliminary cooling demand prediction model using a random search optimization algorithm, wherein the hyperparameters include a learning rate, a tree depth, and a number of leaf nodes; In the process of optimizing the hyperparameters, the cooling demand prediction model is verified by using a time series cross-validation method; The optimized cooling demand prediction model is evaluated using the preset prediction performance evaluation index to obtain the cooling demand prediction model finally used for cooling demand prediction.

6. The method according to claim 5, characterized in that The method of using a time series cross validation method to validate the cooling demand prediction model includes: According to the time sequence of the data, the historical data is divided into multiple training sets and validation sets based on the progressive order of the time periods; In each cross-validation step, the cooling demand forecasting model is trained using the data of the early time period, and the trained cooling demand forecasting model is validated using the data of the subsequent time period; According to the progressive cross-validation method, the training and validation steps are repeatedly performed on the cooling demand prediction model so that the data of each time period are used as a training set and a validation set respectively.

7. The method according to claim 6, characterized in that The optimized cooling demand prediction model is evaluated by using preset prediction performance evaluation indicators, including: According to the result of each verification, the prediction performance evaluation index of the cooling demand prediction model is calculated, and the prediction performance of the cooling demand prediction model is evaluated based on the calculated prediction performance evaluation index, wherein the prediction performance evaluation index includes root mean square error, mean absolute percentage error and symmetric mean percentage error.

8. A cooling demand prediction device, characterized in that: include: A collection module, configured to collect internal and external environment data and equipment operation data of a target location using deployed IoT sensors; A preprocessing module is configured to preprocess the internal and external environment data and the equipment operation data, wherein the preprocessing includes data cleaning and normalization processing; A construction module is configured to perform feature engineering based on the preprocessed data to construct a feature set for cooling demand prediction, wherein the feature set includes physical parameters of the target place, geographic location information, similar date features, and newly added features and cross-features of the target place; A training module is configured to train the cooling demand prediction model using a machine learning algorithm according to a feature set corresponding to one or more of the target locations, and optimize hyperparameters of the cooling demand prediction model; The generation module is configured to use the trained cooling demand prediction model to generate a cooling demand prediction result for the target place in a future time period based on the real-time collected internal and external environmental data and the similar date characteristics.

9. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images