Cold station temperature control method and device, electronic equipment and readable storage medium

Abstract

The application relates to the technical field of computer models, and provides a cold station temperature control method and device, electronic equipment and a readable storage medium. The method comprises the following steps: acquiring operation parameters of a cold station device and corresponding environment parameters; inputting the environment parameters and the operation parameters into a cold quantity demand prediction model for cold quantity demand prediction, wherein the cold quantity demand prediction model comprises a plurality of layers of prediction modules, each layer of prediction module is obtained through cross-validation training of at least one initial prediction model of each layer of prediction module, at least one initial prediction model of each layer of prediction module is used for generating at least one prediction result, and at least one prediction result of each layer is used as input of a next layer of prediction module; and a cold quantity demand prediction value is acquired to control the cold station temperature. The application can significantly improve the accuracy of cold quantity demand prediction, and then accurately control the cold station temperature by using the cold quantity demand prediction, so that the purpose of reducing energy waste is achieved.

Description

Technical Field

[0001] This application relates to the field of computer modeling technology, and in particular to a cooling plant temperature control method, apparatus, electronic device, and readable storage medium. Background Technology

[0002] With rapid urbanization and booming commercial activities, energy consumption in public places such as shopping malls has become an increasingly important concern. In these places, chiller systems, as key facilities for regulating indoor temperature, account for a significant portion of the mall's total energy consumption, significantly impacting operating costs and the environment. Traditional chiller systems typically employ temperature regulation methods based on human experience and simple feedback control. Such methods often rely on fixed parameter settings and manual adjustments, lacking adaptability to real-time environmental changes. In actual mall operations, indoor temperatures fluctuate frequently due to various factors such as weather changes and indoor-outdoor temperature differences. However, traditional chiller systems often fail to respond promptly and accurately to changes in environmental factors, leading to inaccurate temperature control and consequently, energy waste. Summary of the Invention

[0003] In view of this, embodiments of this application provide a cooling plant temperature control method, apparatus, electronic device, and readable storage medium to solve the problem in the prior art that it cannot respond to changes in environmental factors in a timely and accurate manner, resulting in inaccurate temperature control and thus energy waste.

[0004] A first aspect of this application provides a cooling plant temperature control method, the method comprising:

[0005] The system acquires the operating parameters of the chiller plant equipment and the corresponding environmental parameters. These parameters are then input into a trained cooling demand prediction model. The model predicts the cooling demand of the chiller plant using a multi-layer prediction module, which is trained through cross-validation of at least one initial prediction model from each layer. Each initial prediction model generates at least one prediction result, which is then used as input to the next layer for further prediction. The system then acquires the predicted cooling demand values ​​output by the model and controls the chiller plant temperature based on these values.

[0006] A second aspect of this application provides a cooling plant temperature control device, the method comprising:

[0007] The acquisition module is configured to acquire the operating parameters of the chiller plant equipment and the environmental parameters corresponding to the operation of the chiller plant equipment. The input module is configured to input the environmental parameters and operating parameters into a trained cooling demand prediction model, and use the cooling demand prediction model to predict the cooling demand of the chiller plant. The cooling demand prediction model contains multiple prediction modules, which are obtained by cross-validation training of at least one initial prediction model of each prediction module. At least one initial prediction model of each prediction module is used to generate at least one prediction result, and at least one prediction result of each layer is used as the input of the next prediction module for further prediction. The control module is configured to acquire the cooling demand prediction value output by the cooling demand prediction model and control the temperature of the chiller plant based on the cooling demand prediction value.

[0008] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0009] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method.

[0010] The beneficial effects of the embodiments in this application compared with the prior art are:

[0011] The system acquires the operating parameters of the chiller plant equipment and the corresponding environmental parameters. These parameters are then input into a trained cooling demand prediction model. The model predicts the cooling demand of the chiller plant using a multi-layer prediction module, which is trained through cross-validation of at least one initial prediction model from each layer. Each initial prediction model generates at least one prediction result, which is then used as input to the next layer for further prediction. The system then acquires the predicted cooling demand values ​​output by the model and controls the chiller plant temperature based on these values. This application, through a multi-layered prediction module, can fully explore the nonlinear relationship between operating parameters, environmental parameters, and cooling demand, thereby significantly enhancing the accuracy and stability of cooling demand prediction. Furthermore, by combining at least one initial prediction model in each prediction module and employing cross-validation training, the robustness, generalization, and adaptability of the cooling demand prediction model can be further improved, enhancing the accuracy of cooling demand prediction and obtaining more precise cooling demand prediction values. These predicted values ​​can then be used to precisely control the temperature of the cooling plant, thereby reducing energy waste and lowering operating costs. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic flowchart of a cooling plant temperature control method provided in an embodiment of this application;

[0014] Figure 2 This is a schematic diagram of the structure of a cooling plant temperature control device provided in an embodiment of this application;

[0015] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0016] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0017] The following detailed description, in conjunction with the accompanying drawings, describes a cooling plant temperature control method, apparatus, electronic device, and readable storage medium according to embodiments of this application.

[0018] Figure 1 This is a schematic flowchart of a cooling plant temperature control method provided in an embodiment of this application. Figure 1 As shown, the method includes:

[0019] S101, obtain the operating parameters of the cooling plant equipment, and obtain the environmental parameters corresponding to the operation of the cooling plant equipment;

[0020] S102, input the environmental parameters and operating parameters into the trained cooling demand prediction model, and use the cooling demand prediction model to predict the cooling demand of the cooling plant. The cooling demand prediction model includes a multi-layer prediction module. The multi-layer prediction module is obtained by cross-validation training of at least one initial prediction model of each prediction module. At least one initial prediction model of each prediction module is used to generate at least one prediction result, and at least one prediction result of each layer is used as the input of the next layer prediction module for further prediction.

[0021] S103: Obtain the predicted cooling demand value output by the cooling demand forecasting model, and control the temperature of the cooling station based on the predicted cooling demand value.

[0022] Specifically, the operating parameters of the chiller plant equipment include parameters reflecting the operating status and performance of the chiller plant equipment, such as chiller power, water pump frequency, chilled water supply and return water temperature, chilled water supply and return water pressure, chilled water supply and return water flow rate, and cooling tower operating status. The environmental parameters corresponding to the operation of the chiller plant equipment include parameters that affect indoor comfort and cooling load demand, such as indoor and outdoor temperature, humidity, pedestrian traffic, wind force and speed, and weather. The parameters can be obtained by corresponding sensors, such as temperature sensors, humidity sensors, pressure sensors, and flow sensors.

[0023] It is understandable that the cooling demand forecasting model is an artificial intelligence model containing multiple prediction modules. These modules, by stacking multiple prediction layers, can extract and combine features from the acquired parameters layer by layer, thereby more accurately capturing the complex relationships within the parameters. At least one initial prediction model in each prediction module can be further optimized and adjusted based on at least one prediction result from the previous layer as input, further improving the accuracy and reliability of cooling demand forecasting. Furthermore, at least one initial prediction model in each prediction module undergoes cross-validation training; that is, each initial prediction model is trained and tested on different subsets of data to improve its predictive performance. In this way, the cooling demand forecasting model can predict the cooling demand of a cooling plant based on the operating parameters of the cooling plant equipment and environmental parameters, generating a predicted cooling demand value. The cooling plant then adjusts its equipment parameters according to this predicted value to achieve precise temperature control.

[0024] It should be noted that the specific number of layers in the multi-layer prediction module can be flexibly configured according to actual needs and application scenarios. Furthermore, the number and type of initial prediction models for each layer can also be selected and adjusted. For example, different types of machine learning algorithms can be selected as initial prediction models, such as linear regression, decision trees, neural networks, etc., and combined and optimized according to different parameter characteristics and prediction requirements. This embodiment does not impose specific limitations on this.

[0025] According to the technical solution provided in this application, the operating parameters of the chiller plant equipment and the corresponding environmental parameters are obtained. The environmental parameters and operating parameters are input into a trained cooling demand prediction model. The cooling demand of the chiller plant is predicted using the cooling demand prediction model, which includes multiple prediction modules. Each prediction module is trained by cross-validation of at least one initial prediction model from each prediction module. Each initial prediction model generates at least one prediction result, and each prediction result serves as the input for the next prediction module. The predicted cooling demand value output by the cooling demand prediction model is obtained, and the chiller plant temperature is controlled based on the predicted cooling demand value. By employing multi-layered prediction modules, the nonlinear relationship between operating and environmental parameters and cooling demand can be fully explored, thereby significantly enhancing the accuracy and stability of cooling demand prediction. Furthermore, through the diversity of combinations of at least one initial prediction model in each prediction module and cross-validation training, the robustness, generalization, and adaptability of the cooling demand prediction model can be further improved, enhancing the accuracy of cooling demand prediction and obtaining more precise cooling demand prediction values. These predicted values ​​can then be used to precisely control the temperature of the cooling plant, thereby reducing energy waste and lowering operating costs.

[0026] In some embodiments, before inputting environmental parameters and operating parameters into the trained cooling demand prediction model, the method further includes:

[0027] Obtain a training sample set, and construct an initial prediction model for each layer of the multi-layer prediction module based on the training sample set and a preset machine learning algorithm; train the initial prediction model for each layer of the multi-layer prediction module based on the training sample set; obtain the prediction result output by the last layer prediction module; construct a loss function based on the prediction result output by the last layer prediction module and the corresponding true label; if the loss value of the loss function is less than a preset value, the multi-layer prediction module is trained, and a cooling demand prediction model is obtained based on the trained multi-layer prediction module.

[0028] Specifically, to obtain a high-quality cooling demand prediction model, it is necessary to train and optimize its multi-layer prediction modules. During the training process, a training sample set containing rich data can be obtained first. This training sample set should include environmental and operational parameters under various conditions, along with corresponding real labels for cooling demand. Furthermore, the training sample set should cover data from different seasons and climatic conditions to fully consider the impact of environmental parameters on cooling demand. Next, based on the training sample set and a pre-defined machine learning algorithm, an initial prediction model is constructed for each layer of the multi-layer prediction module. It is understood that the pre-defined machine learning algorithm can include methods such as linear regression, decision trees, neural networks, and gradient boosting machines. Specifically, an algorithm suitable for the task characteristics and data features can be selected for model construction; this implementation does not impose specific limitations on this. Then, the initial prediction model for each layer of the prediction module is trained using the training sample set. By iteratively optimizing the model's parameters, it can better fit the training data and predict cooling demand.

[0029] Furthermore, after training, the prediction results output by the last prediction module are obtained and compared with the corresponding true labels. A loss function is constructed to measure the difference between the prediction results and the true labels. By minimizing the loss value of the loss function, the model parameters are further adjusted and optimized. When the loss value of the loss function is less than a preset value, the model can be considered to have been successfully trained and has good prediction performance. The trained multi-layer prediction module is obtained, and a cooling demand prediction model is built based on this module.

[0030] This allows for the creation of a well-trained and high-performance cooling demand prediction model. This model comprehensively considers the impact of environmental and operational parameters on cooling demand and performs accurate predictions based on multi-layered prediction modules. In application, real-time acquired environmental and operational parameters can be input into the trained cooling demand prediction model to obtain accurate cooling demand prediction results.

[0031] It is important to note that when building and training a cooling demand prediction model, the diversity and representativeness of the training sample set must be considered. The training sample set should include environmental and operational parameters under various conditions to ensure the model can adapt to different application scenarios and needs. For example, when building the training sample set, data can be collected under different seasons and weather conditions, including environmental parameters such as temperature, humidity, and wind speed, as well as operational parameters such as equipment uptime, load, and energy efficiency rating. Simultaneously, it is necessary to ensure that the true cooling demand label for each sample is accurate. By constructing a rich, diverse, and representative training sample set, the generalization ability and accuracy of the cooling demand prediction model can be improved.

[0032] According to the technical solution of this embodiment, a training sample set is obtained. Based on the training sample set and a preset machine learning algorithm, an initial prediction model for each layer of the multi-layer prediction module is constructed. The initial prediction model for each layer of the multi-layer prediction module is then trained based on the training sample set. The prediction result output by the last layer prediction module is obtained. A loss function is constructed based on the prediction result output by the last layer prediction module and the corresponding true label. If the loss value of the loss function is less than a preset value, the multi-layer prediction module is considered trained. Based on the trained multi-layer prediction module, a cooling demand prediction model is obtained. By constructing an initial prediction model based on the training sample set and continuously optimizing the model parameters during training, it can be ensured that the model can better fit the training data and adapt to various complex environmental conditions and changes in operating parameters. Furthermore, the multi-layer prediction module structure can fully utilize the prediction results from different levels, improving the accuracy and reliability of the prediction results.

[0033] In some embodiments, an initial prediction model for each layer of the multi-layer prediction module is constructed based on a training sample set and a preset machine learning algorithm, including:

[0034] The training sample set is divided into a training set and a validation set. A preset machine learning algorithm is trained on the training set and the validation set to obtain multiple intermediate prediction models. The loss value between the prediction result and the true value of each sample in the validation set is calculated. Based on the loss value, at least one of the mean absolute loss, mean square loss and root mean square loss of each intermediate prediction model is calculated. The evaluation result of the multiple intermediate prediction models is determined according to at least one of the mean absolute loss, mean square loss and root mean square loss. Based on the evaluation result of the multiple intermediate prediction models, a preset number of intermediate prediction models are selected from the multiple intermediate prediction models as the initial prediction model of each prediction module in the multi-layer prediction module.

[0035] Specifically, when constructing the initial prediction model for a multi-layer prediction module, to improve the model's stability and generalization ability, a pre-defined machine learning algorithm can be trained using a training sample set to obtain multiple intermediate prediction models. For example, the training sample set can be divided into a training set and a validation set, where the training set is used for model training and the validation set is used to evaluate the model's performance. By training the pre-defined machine learning algorithm based on the training and validation sets, multiple intermediate prediction models can be obtained. These intermediate prediction models are obtained based on the same training data and parameter settings, but because each intermediate prediction model has a different structure, they have different predictive capabilities and performance characteristics.

[0036] Understandably, to evaluate the performance of these intermediate prediction models, the loss value between the predicted result and the true value for each sample in the validation set can be calculated. This loss value reflects the degree of difference between the intermediate prediction model's prediction and the true value, and can be used to measure the model's prediction accuracy and precision. By calculating evaluation metrics such as mean absolute loss, mean squared loss, and root mean square loss, the performance of multiple intermediate prediction models can be comprehensively evaluated. Based on these evaluation metrics, the evaluation results of multiple intermediate prediction models can be determined. The evaluation results can include information on each model's prediction accuracy, stability, robustness, etc. Based on these evaluation results, a predetermined number of models can be selected from multiple intermediate prediction models as the initial prediction models for each layer of the multi-layer prediction module. In this way, selecting the model with the best prediction performance from multiple intermediate prediction models as the initial prediction model improves the stability and generalization ability of the multi-layer prediction module. At the same time, it can also effectively avoid problems such as overfitting and underfitting, improving the prediction accuracy and reliability of the cooling demand prediction model.

[0037] It's worth noting that the diversity and complementarity of models should also be considered when selecting initial prediction models. Since each layer of a multi-layer prediction module consists of at least one initial prediction model, the models in different layers should have a certain degree of difference and complementarity to improve the overall prediction performance of the module. For example, different types of machine learning algorithms can be selected as initial prediction models for different layers, or the parameters and structures of different layer models can be adjusted to give them different prediction characteristics and advantages. Furthermore, in constructing the initial prediction model for each layer, the construction should ideally have more models at the bottom and fewer models at the top to improve model stability and generalization ability.

[0038] According to the technical solution of this embodiment, the training sample set is divided into a training set and a validation set, and a preset machine learning algorithm is trained based on the training set and the validation set to obtain multiple intermediate prediction models; the loss value between the prediction result and the true value of each sample in the validation set is calculated; at least one of the mean absolute loss, mean square loss and root mean square loss of each intermediate prediction model is calculated based on the loss value; the evaluation result of the multiple intermediate prediction models is determined based on at least one of the mean absolute loss, mean square loss and root mean square loss; based on the evaluation result of the multiple intermediate prediction models, a preset number of intermediate prediction models are selected from the multiple intermediate prediction models as the initial prediction model of each prediction module in the multi-layer prediction module. By using a cross-validation method based on the training set and the validation set, a model with excellent performance is selected as the initial prediction model of each prediction module in the multi-layer prediction module. This can construct a stable and accurate cold demand prediction model, thereby providing more accurate and reliable prediction results.

[0039] In some embodiments, training is performed on each layer of the multi-layer prediction module based on a training sample set, including:

[0040] The preprocessed training sample set is divided into K subsets, each containing the operating parameters of the chiller equipment within different time periods, as well as the environmental parameters corresponding to the operation of the chiller equipment. For each initial prediction model of the first-layer prediction module, K-1 subsets are used as training data and the remaining subset is used as validation data, and K training and validation cycles are performed to obtain the initial prediction result corresponding to each initial prediction model. For the remaining prediction modules other than the first-layer prediction module, the initial prediction result obtained from the previous layer prediction module is used as the input of the next layer prediction module, and K cross-validation training is repeated until the training of the last layer prediction module is completed.

[0041] Specifically, during the training of the multi-layer prediction module, to fully utilize the information in the training sample set and improve the model's generalization ability, the preprocessed training sample set can be divided into K subsets. Each subset contains the operating parameters of the cooling station equipment and the corresponding environmental parameters for different time periods. For the first layer of the multi-layer prediction module, K-1 subsets are selected as training data, and the remaining subset is used as validation data. This combination is performed K times, and each training iteration yields an initial prediction result corresponding to an initial prediction model. This ensures that each subset has at least one opportunity to serve as validation data, thus fully utilizing the information in the training sample set. For the prediction modules other than the first layer, the initial prediction result obtained from the previous layer is used as the input to the next layer. Then, the cross-validation training process is repeated K times until the last layer prediction module is trained. By fully utilizing the output information of the previous layer prediction module and using it as the input to the next layer prediction module, the prediction defects of the previous layer model can be corrected, thereby constructing a more accurate and stable prediction model.

[0042] It should be noted that for the prediction modules other than the first-layer prediction module, in addition to using the multiple initial prediction results obtained from the previous layer prediction module as the input of the next layer prediction module, the training sample set can also be used as the input of each layer module. The next layer prediction module can be cross-validated by combining the multiple initial prediction results obtained from the previous layer prediction module and the training sample set, so as to make full use of the existing training sample set, further explore the potential information in the data, and enable each layer prediction module to obtain more comprehensive and accurate training, thereby further improving the prediction accuracy and stability of the model.

[0043] According to the technical solution of this application embodiment, the preprocessed training sample set is divided into K subsets, each subset containing the operating parameters of the cooling station equipment within different time periods, as well as the environmental parameters corresponding to the operation of the cooling station equipment. For each initial prediction model of the first-layer prediction module, K-1 subsets are used as training data, and the remaining subset is used as validation data, for K training and validation iterations, to obtain the initial prediction result corresponding to each initial prediction model. For the remaining prediction modules other than the first-layer prediction module, the initial prediction result obtained from the previous layer prediction module is used as the input of the next layer prediction module, and K cross-validation training is repeated until the last layer prediction module is trained. By employing the K-times cross-validation training method, the multi-layer prediction modules can be fully trained and validated, ensuring that the model maintains stable performance under different time periods and data distributions. Simultaneously, it can effectively avoid overfitting and underfitting problems, improving the prediction accuracy and reliability of the model.

[0044] In some embodiments, obtaining the training sample set includes:

[0045] The equipment operation time of the cooling plant is determined from the historical operating parameters of the cooling plant equipment. When the equipment operation time exceeds the preset time threshold, the operating parameters of the cooling plant equipment with the equipment operation time exceeding the preset time threshold and the corresponding environmental parameters are used as the training sample set.

[0046] Specifically, in acquiring the training sample set, to ensure the quality and validity of the samples, data with longer operating times can be selected from the historical operating parameters of the cooling station equipment. That is, the operating time of the equipment for each run can be determined, which can be achieved by analyzing historical operating records or logs; this implementation does not impose specific limitations on this. Only when the equipment operating time exceeds a preset time threshold are the corresponding operating parameters and environmental parameters included in the training sample set. In this way, data with excessively short or unstable operating times can be excluded, thus ensuring the quality of the training sample set. Simultaneously, this also helps improve the stability and generalization ability of the model, because long-term operating data usually better reflects the actual operating conditions and performance characteristics of the equipment.

[0047] According to the technical solution provided in this application, the equipment operating time of the cooling plant is determined from the historical operating parameters of the cooling plant equipment. When the equipment operating time exceeds a preset time threshold, the operating parameters of the cooling plant equipment with operating times exceeding the preset time threshold and the corresponding environmental parameters are used as a training sample set. By selecting data with longer equipment operating times as the training sample set, the quality and effectiveness of the samples are ensured.

[0048] In some embodiments, before obtaining the training sample set, the method further includes:

[0049] The method involves at least one of the following: using the operating parameters of the chiller equipment in the training sample set and the corresponding environmental parameters during the operation of the chiller equipment, constructing periodic operating characteristics, truncating negative values ​​of cooling capacity, calculating the cumulative cooling capacity within a preset time, performing indoor temperature and humidity analysis and statistics, calculating wet-bulb temperature, constructing indoor and outdoor temperature difference characteristics and cold source system temperature difference characteristics, performing feature processing on tilted features, and performing one-heat vector encoding on categorical variables.

[0050] Specifically, to further improve the prediction accuracy and performance of the model, a series of feature processing operations can be performed on the training sample set. These operations include, but are not limited to, constructing periodic operating features, truncating negative values ​​of cooling capacity, calculating the cumulative cooling capacity within a preset time period, performing indoor temperature and humidity analysis and statistics, constructing indoor and outdoor temperature difference features and cooling source system temperature difference features, processing tilt features, and performing one-heat vector encoding on categorical variables. By constructing periodic operational characteristics, the operating features and patterns of chiller equipment in different seasons and times can be reflected; by truncating negative cooling values, the interference of negative values ​​on model training can be avoided; by calculating the cumulative cooling capacity within a preset time period (e.g., 1 hour), the cumulative effect of cooling capacity can be captured; through indoor temperature and humidity analysis and statistics, the condition of the indoor environment can be understood and abnormal sensor data can be filtered out; by constructing indoor-outdoor temperature difference characteristics and chiller system temperature difference characteristics, the impact of environmental temperature changes on cooling demand can be reflected; wet-bulb temperature can be calculated as an additional feature based on the collected indoor temperature and humidity data to more comprehensively reflect the state of the indoor environment; processing skewed features can eliminate correlations or redundancy between features, specifically including transformation processing, normalization truncation, etc.; one-heat vector encoding of categorical variables can convert categorical data into a numerical form that the model can process. These feature processing operations can be selected and combined according to actual conditions and needs to improve model performance.

[0051] According to the technical solution provided in this application, before acquiring the training sample set, a series of feature processing operations are performed on the running parameters and environmental parameters, which can effectively improve the prediction accuracy and stability of the model. These feature processing operations can extract key features that have an important impact on the prediction results, eliminate noise and redundant information in the data, and make the model easier to understand and learn the inherent laws of the data. By constructing periodic operating features, the operating characteristics of the chiller equipment in different time periods can be captured, thereby more accurately predicting its future operating status. Truncation of negative cooling load values ​​can avoid interference from negative values ​​on model training, improving the stability and reliability of the model. Calculating the cumulative cooling load within a preset time period can reflect the cumulative effect of cooling load, further improving the accuracy of prediction. Indoor temperature and humidity analysis and statistics can help understand the indoor environment and help the model better adapt to different environmental conditions. Constructing indoor and outdoor temperature difference features and chiller system temperature difference features can reflect the impact of environmental temperature changes on cooling load demand, improving the model's sensitivity to environmental changes. Processing skew features can eliminate the correlation or redundancy between features, improving the model's generalization ability. One-hot vector encoding of categorical variables can transform categorical data into a numerical form that the model can process, enabling the model to make full use of all input information.

[0052] In some embodiments, after controlling the temperature of the cooling station based on the predicted cooling demand, the method further includes:

[0053] The predicted cooling demand, along with the corresponding environmental and operational parameters, is stored in time series format. The predicted cooling demand, along with the corresponding operational parameters and the latest preset number of days of data selected from the environmental parameters, are used as a new training sample set to train the cooling demand prediction model, resulting in an updated cooling demand prediction model. Cooling demand is then predicted based on the updated cooling demand prediction model.

[0054] Specifically, after controlling the chiller plant temperature based on the predicted cooling demand, the model can be periodically updated and optimized to further improve its predictive performance. Specifically, the predicted cooling demand, along with corresponding environmental and operational parameters, can be stored in time-series format for later use. Then, the latest data for a preset number of days is selected from the stored data as a new training sample set. This preset number of days can be set according to actual needs, for example, selecting data from the most recent week or two. Next, the cooling demand prediction model is trained using the new training sample set to obtain an updated cooling demand prediction model. By continuously updating and optimizing the model, it can better adapt to changes in the environment and equipment, improving the accuracy and reliability of predictions. Finally, cooling demand is predicted based on the updated cooling demand prediction model to achieve more accurate and efficient chiller plant temperature control. This periodic model update and optimization process can be repeated to ensure the model's continuous effectiveness and performance improvement.

[0055] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the process of the embodiments of this application.

[0056] All of the above-mentioned optional technical solutions can be combined in any way to form the optional embodiments of this application, and will not be described in detail here.

[0057] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0058] Figure 2 This is a schematic diagram of the structure of a cooling plant temperature control device provided in an embodiment of this application. Figure 2 As shown, the device includes:

[0059] The acquisition module 201 is configured to acquire the operating parameters of the cooling plant equipment and the environmental parameters corresponding to the operation of the cooling plant equipment.

[0060] Input module 202 is configured to input environmental parameters and operating parameters into a trained cooling demand prediction model, and use the cooling demand prediction model to predict the cooling demand of the cooling plant. The cooling demand prediction model includes a multi-layer prediction module, which is obtained by cross-validation training of at least one initial prediction model of each prediction module. At least one initial prediction model of each prediction module is used to generate at least one prediction result, and at least one prediction result of each layer is used as the input of the next layer prediction module for further prediction.

[0061] The control module 203 is configured to acquire the predicted cooling demand value output by the cooling demand prediction model and control the temperature of the cooling station based on the predicted cooling demand value.

[0062] In some embodiments, the cooling plant temperature control device further includes a training module 204, which is configured to: acquire a training sample set; construct an initial prediction model for each layer of the multi-layer prediction module based on the training sample set and a preset machine learning algorithm; train the initial prediction model for each layer of the multi-layer prediction module based on the training sample set; acquire the prediction result output by the last layer prediction module; construct a loss function based on the prediction result output by the last layer prediction module and the corresponding real label; obtain the trained multi-layer prediction module when the loss value of the loss function is less than a preset value; and obtain a cooling demand prediction model based on the trained multi-layer prediction module.

[0063] In some embodiments, the training module 204 is further configured to divide the training sample set into a training set and a validation set, and train a preset machine learning algorithm based on the training set and the validation set to obtain multiple intermediate prediction models; calculate the loss value between the prediction result and the true value of each sample in the validation set; calculate at least one of the mean absolute loss, mean square loss and root mean square loss of each intermediate prediction model based on the loss value; determine the evaluation result of the multiple intermediate prediction models based on at least one of the mean absolute loss, mean square loss and root mean square loss; and select a preset number of intermediate prediction models from the multiple intermediate prediction models as the initial prediction model for each layer prediction module in the multi-layer prediction module based on the evaluation result of the multiple intermediate prediction models.

[0064] In some embodiments, the training module 204 is further configured to divide the training sample set into K subsets, wherein each subset contains the operating parameters of the chiller equipment in different time periods, as well as the environmental parameters corresponding to the operation of the chiller equipment; for each initial prediction model of the first-layer prediction module, K-1 subsets are used as training data in sequence, and the remaining subset is used as validation data, and K training and validation are performed to obtain the initial prediction result corresponding to each initial prediction model; for the remaining prediction modules other than the first-layer prediction module, the initial prediction result obtained by the previous-layer prediction module is used as the input of the next-layer prediction module, and K cross-validation training is repeatedly performed until the last-layer prediction module is trained.

[0065] In some embodiments, the training module 204 is further configured to determine the equipment operating time of the chiller for each time from the historical operating parameters of the chiller equipment; when the equipment operating time is greater than a preset time threshold, the operating parameters of the chiller equipment with the equipment operating time greater than the preset time threshold and the corresponding environmental parameters are used as the training sample set.

[0066] In some embodiments, the training module 204 is further configured to perform at least one of the following: constructing periodic operating characteristics using the operating parameters of the chiller equipment in the training sample set and the corresponding environmental parameters during the operation of the chiller equipment; truncating negative values ​​of cooling capacity; calculating the cumulative cooling capacity within a preset time period; performing indoor temperature and humidity analysis and statistics; calculating wet-bulb temperature; constructing indoor and outdoor temperature difference characteristics and cold source system temperature difference characteristics; performing feature processing on tilted features; and performing one-hot vector encoding on categorical variables.

[0067] In some embodiments, the control module 203 is further configured to store the predicted cooling demand value and the corresponding environmental parameters and operating parameters in the form of a time series; select the latest preset number of days of data from the stored predicted cooling demand value and the corresponding environmental parameters and operating parameters as a new training sample set to train the cooling demand prediction model, obtain the updated cooling demand prediction model, and perform cooling demand prediction based on the updated cooling demand prediction model.

[0068] Figure 3 This is a schematic diagram of the electronic device 3 provided in an embodiment of this 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, it implements the steps in the various method embodiments described above. Alternatively, when the processor 301 executes the computer program 303, it implements the functions of each module / unit in the various device embodiments described above.

[0069] Electronic device 3 can be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 3 may include, but is not limited to, processor 301 and memory 302. Those skilled in the art will understand that... Figure 3 This is merely an example of electronic device 3 and does not constitute a limitation on electronic device 3. It may include more or fewer components than shown, or different components.

[0070] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0071] The memory 302 can be an internal storage unit of the electronic device 3, such as a hard disk or RAM of the electronic device 3. The memory 302 can also be an external storage device of the electronic device 3, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc., equipped on the electronic device 3. The memory 302 can also include both internal and external storage units of the electronic device 3. The memory 302 is used to store computer programs and other programs and data required by the electronic device.

[0072] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to 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 embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0073] If integrated modules / units are implemented as software functional units and sold or used as independent products, they can be stored in a readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The readable storage medium may include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0074] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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. Such 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 this application, and should all be included within the protection scope of this application.

Claims

1. A method for controlling temperature at a chiller plant, characterized in that, include: Obtain the operating parameters of the cooling plant equipment, and obtain the environmental parameters corresponding to the operation of the cooling plant equipment; The environmental parameters and the operating parameters are input into the trained cooling demand prediction model, and the cooling demand of the cooling station is predicted using the cooling demand prediction model. The cooling demand prediction model includes a multi-layer prediction module. The multi-layer prediction module is obtained by cross-validation training of at least one initial prediction model of each prediction module. The at least one initial prediction model of each prediction module is used to generate at least one prediction result, and the at least one prediction result of each layer is used as the input of the next layer prediction module for further prediction. Obtain the predicted cooling demand value output by the cooling demand prediction model, and control the temperature of the cooling station based on the predicted cooling demand value. Before inputting the environmental parameters and the operating parameters into the trained cooling demand prediction model, the method further includes: Obtain a training sample set, and construct an initial prediction model for each prediction module in the multi-layer prediction module based on the training sample set and a preset machine learning algorithm; The initial prediction model of each prediction module in the multi-layer prediction module is trained based on the training sample set. Obtain the prediction results output by the last prediction module; Based on the prediction results output by the last layer prediction module and the corresponding real labels, a loss function is constructed. When the loss value of the loss function is less than a preset value, the multi-layer prediction module is trained and completed. Based on the trained multi-layer prediction module, the cooling demand prediction model is obtained. The step of training each layer of the multi-layer prediction module based on the training sample set includes: The training sample set is divided into K subsets, each subset containing the operating parameters of the cooling station equipment in different time periods, as well as the environmental parameters corresponding to the operation of the cooling station equipment. For each initial prediction model in the first-layer prediction module, K-1 subsets are used as training data and the remaining subset is used as validation data. K training and validation cycles are performed to obtain the initial prediction result corresponding to each initial prediction model. For the prediction modules other than the first-layer prediction module, the initial prediction result obtained from the previous-layer prediction module is used as the input of the next-layer prediction module, and cross-validation training is repeated K times until the last-layer prediction module is trained.

2. The method according to claim 1, characterized in that, The construction of the initial prediction model for each layer of the multi-layer prediction module based on the training sample set and a preset machine learning algorithm includes: The training sample set is divided into a training set and a validation set, and a preset machine learning algorithm is trained based on the training set and the validation set to obtain multiple intermediate prediction models. Calculate the loss value between the predicted result and the true value for each sample in the validation set; Calculate at least one of the mean absolute loss, mean squared loss, and root mean squared loss for each intermediate prediction model based on the loss value; The evaluation results of the plurality of intermediate prediction models are determined based on at least one of the mean absolute loss, mean square loss, and root mean square loss. Based on the evaluation results of the multiple intermediate prediction models, a predetermined number of intermediate prediction models are selected from the multiple intermediate prediction models as the initial prediction models for each layer of the multi-layer prediction module.

3. The method according to claim 1, characterized in that, The acquisition of the training sample set includes: The equipment operation time of the cooling plant for each time is determined from the historical operating parameters of the cooling plant equipment; When the equipment running time exceeds a preset time threshold, the operating parameters of the cooling station equipment with the equipment running time exceeding the preset time threshold and the corresponding environmental parameters are used as the training sample set.

4. The method according to claim 1, characterized in that, Before obtaining the training sample set, the process also includes: The system utilizes at least one of the following methods: constructing periodic operating characteristics using the operating parameters of the cold station equipment in the training sample set and the corresponding environmental parameters during the operation of the cold station equipment; truncating negative values ​​of cooling capacity; calculating the cumulative cooling capacity within a preset time period; performing indoor temperature and humidity analysis and statistics; calculating wet-bulb temperature; constructing indoor and outdoor temperature difference characteristics and cold source system temperature difference characteristics; performing feature processing on tilted features; and performing one-hot vector encoding on categorical variables.

5. The method according to claim 1, characterized in that, After controlling the temperature of the cooling station based on the predicted cooling demand, the method further includes: The predicted cooling demand value, along with the corresponding environmental parameters and operating parameters, are stored in time series format. The latest preset number of days of data are selected from the stored predicted cooling demand values ​​and the corresponding environmental parameters and operating parameters as a new training sample set to train the cooling demand prediction model, thereby obtaining an updated cooling demand prediction model, and cooling demand is predicted based on the updated cooling demand prediction model.

6. A temperature control device for a chiller plant, characterized in that, include: The acquisition module is configured to acquire the operating parameters of the cooling plant equipment and the environmental parameters corresponding to the operation of the cooling plant equipment. The input module is configured to input the environmental parameters and the operating parameters into a trained cooling demand prediction model, and use the cooling demand prediction model to predict the cooling demand of the cooling station. The cooling demand prediction model includes a multi-layer prediction module, which is obtained by cross-validation training of at least one initial prediction model of each prediction module. The at least one initial prediction model of each prediction module is used to generate at least one prediction result, and the at least one prediction result of each layer is used as the input of the next layer prediction module for further prediction. The control module is configured to acquire the predicted cooling demand value output by the cooling demand prediction model, and control the temperature of the cooling station based on the predicted cooling demand value. It also includes a training module, used for: acquiring a training sample set, and constructing an initial prediction model for each layer of the multi-layer prediction module based on the training sample set and a preset machine learning algorithm; training the initial prediction model for each layer of the multi-layer prediction module based on the training sample set; acquiring the prediction result output by the last layer prediction module; constructing a loss function based on the prediction result output by the last layer prediction module and the corresponding true label; if the loss value of the loss function is less than a preset value, obtaining the trained multi-layer prediction module; and obtaining the cooling demand prediction model based on the trained multi-layer prediction module; wherein, the step of training the initial prediction model for each layer of the multi-layer prediction module based on the training sample set... The prediction module trains each layer of the prediction module by dividing the training sample set into K subsets, each subset containing the operating parameters of the cooling station equipment in different time periods and the environmental parameters corresponding to the operation of the cooling station equipment; for each initial prediction model of the first layer prediction module, K-1 subsets are used as training data and the remaining subset is used as validation data, and K training and validations are performed to obtain the initial prediction result corresponding to each initial prediction model; for the remaining prediction modules other than the first layer prediction module, the initial prediction result obtained by the previous layer prediction module is used as the input of the next layer prediction module, and K cross-validation training is repeated until the last layer prediction module is trained.

7. 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, it implements the steps of the method as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 5.

Images