Cold station power control method, apparatus, and electronic device

Abstract

The application relates to the technical field of cold station regulation and control, and discloses a cold station power control method, a cold station power control device and electronic equipment. The method comprises the following steps: obtaining target control power of a cold station; predicting future overall control parameters of the cold station based on environmental parameters, the target control power and cold station operation data; obtaining power prediction data and comfort prediction data of the cold station based on the overall control parameters of the cold station; obtaining control parameter basic data of the cold station according to the power prediction data, the comfort prediction data and the overall control parameters of the cold station; and controlling the operation of the cold station based on the control parameter basic data. Through the above method, the power of the cold station operation can be effectively controlled while the comfort is taken into account, and the utilization efficiency of electric power resources is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cold station regulation, and in particular to a cold station power control method and device and electronic equipment. BACKGROUND

[0002] In recent years, with the rapid development of the power system, demand-side response and flexible load regulation have gradually become the focus of attention. During the peak load period, by appropriately reducing the power consumption of part of the load, the power response is carried out according to the set target, which has significant application value. This adjustment method not only effectively alleviates the pressure on the power grid, but also improves the utilization efficiency of power resources, and has broad market prospects and great development potential.

[0003] The central air conditioning cold station, as a system commonly operated in summer, plays an important role in load-side power regulation. Aggregating a large number of cold stations and other adjustable load resources can generate a large amount of load adjustment space, effectively addressing the contradiction between power supply and demand. However, the related power control technology has many limitations: on the one hand, related means often aim to prioritize meeting the overall building performance, resulting in a significant sacrifice in human comfort, making it difficult to optimize comfort during the adjustment process; on the other hand, the existing technology has low efficiency and poor effect in adjusting the power of the cold station, making it difficult to accurately balance the relationship between power control and comfort, often resulting in adjustment lag and insensitive response. In addition, the lack of systematic control strategies results in insufficient flexibility and adaptability of cold station power regulation, which cannot effectively cope with complex and changing power demand and environmental conditions.

[0004] Therefore, how to minimize the impact on human comfort while achieving efficient and accurate regulation of cold station power remains a key problem to be solved. The breakthrough of this problem will significantly improve the overall performance of cold station power regulation, providing important support for the stable operation of the power system and the optimal allocation of resources. SUMMARY

[0005] Embodiments of the present application aim to at least partially solve one of the technical problems in the related art. To this end, the embodiments of the present application propose a cold station power control method, device, electronic equipment, computer product and medium.

[0006] The embodiments of the present application provide a cold station power control method, which comprises: obtaining a target control power of a cold station; predicting future overall control parameters of the cold station based on environmental parameters, the target control power and cold station operation data; obtaining power prediction data and comfort prediction data of the cold station based on the overall control parameters of the cold station; obtaining control parameter basic data of the cold station according to the power prediction data, the comfort prediction data and the overall control parameters of the cold station; and controlling the operation of the cold station based on the control parameter basic data.

[0007] In some embodiments, the control parameter basic data of the cold station is obtained according to the power prediction data, the comfort prediction data and the overall control parameter of the cold station, including: when the power prediction data is within the range of the target control power, and the comfort prediction data is within the reasonable comfort range, the overall control parameter of the cold station is determined as the control parameter basic data of the cold station; when the power prediction data is not within the range of the target control power, or the comfort prediction data is not within the reasonable comfort range, the control parameter basic data of the cold station is re-predicted based on the power prediction data and the comfort prediction data.

[0008] In some embodiments, the control parameter basic data includes a plurality of control parameter basic values corresponding to a plurality of time periods; and the cold station is controlled to operate based on the control parameter basic data, including: obtaining an overall control parameter corresponding to a current time period based on a control parameter basic value corresponding to the current time period; obtaining an overall control parameter corresponding to a next time period according to an actual power and a current environment parameter in a case that the cold station is controlled to operate based on the overall control parameter corresponding to the current time period; and controlling the cold station to operate based on the overall control parameter corresponding to the next time period in a case that the next time period arrives.

[0009] In some embodiments, the overall control parameter corresponding to the next time period is obtained according to the actual power and the current environment parameter in the case that the cold station is controlled to operate based on the overall control parameter corresponding to the current time period, including: when the actual power is within the range of the target control power, and the current environment parameter is within the preset environment threshold range, a control parameter basic value corresponding to the next time period is determined as the overall control parameter corresponding to the next time period; and when the actual power is not within the range of the target control power, or the current environment parameter is not within the preset environment threshold range, the overall control parameter corresponding to the next time period is re-predicted based on target data from a reference time period to the current time period, wherein the target data includes at least one of the control parameter basic data, operation data of the cold station, environment data and comfort data.

[0010] In some embodiments, the power prediction data of the cold station is obtained by a power prediction model, the power prediction model includes a plurality of long short-term memory network layers and a full connection layer, and the power prediction model is obtained by training in the following manner: obtaining power sample data and power labels corresponding to the power sample data, wherein the power sample data includes outdoor temperature, outdoor humidity and overall control parameter; processing the power sample data by using the plurality of long short-term memory network layers and the full connection layer to obtain a power prediction result; determining an error between the power prediction result and the power label by using a root mean square error method, and adjusting model parameters of the power prediction model based on the error to obtain a trained power prediction model.

[0011] In some embodiments, the comfort prediction data of the cold station is obtained by a comfort prediction model, the power prediction model comprises a plurality of long short-term memory network layers and a full connection layer, and the comfort prediction model is trained in the following manner: comfort sample data and comfort labels corresponding to the comfort sample data are obtained, wherein the comfort sample data comprises at least one of outdoor temperature, outdoor humidity, overall control parameters, and artificial operation weighted times, wherein the artificial operation weighted times comprise a weighted average of the number of temperature increasing operations, the number of temperature decreasing operations, and the number of air speed adjusting operations; the comfort sample data is processed by using the plurality of long short-term memory network layers and the full connection layer to obtain comfort prediction results; the error between the comfort prediction results and the comfort labels is determined by using a root mean square error method, and the model parameters of the comfort prediction model are adjusted based on the error to obtain a trained comfort prediction model.

[0012] In some embodiments, the control parameter basis data of the cold station is obtained by a reinforcement learning model, and the reinforcement learning model is trained in the following manner: cold station historical data is obtained, wherein the cold station historical data comprises at least one of historical control parameter basis data, historical operation data, historical outdoor temperature, historical outdoor humidity, and historical comfort data; and the reinforcement learning model is trained based on the cold station historical data.

[0013] In some embodiments, the reinforcement learning model is trained based on the cold station historical data, comprising: training the reinforcement learning model according to a reward function based on the cold station historical data, wherein the reward function is associated with at least one of a cold station efficiency reward function, a comfort sample data penalty function, a constraint penalty function, and a final reward function; wherein the cold station efficiency reward function is associated with real-time refrigeration capacity, real-time power, and a reward coefficient of the cold station; wherein the comfort sample data penalty function is associated with artificial operation weighted times and an operation weighted times penalty coefficient; and wherein the constraint penalty function is associated with demand power, real-time power, and a penalty coefficient of the cold station.

[0014] Embodiments of the present application provide a power control device of a cold station, the device comprising: an acquisition module configured to acquire target control power of the cold station; a prediction module configured to predict future overall control parameters of the cold station based on environmental parameters, the target control power, and cold station operation data, and to obtain power prediction data and comfort prediction data of the cold station based on the overall control parameters of the cold station; an inference module configured to obtain control parameter basis data of the cold station based on the power prediction data, the comfort prediction data, and the overall control parameters of the cold station; and a control module configured to control operation of the cold station based on the control parameter basis data of the cold station.

[0015] Embodiments of the present application provide an electronic device, comprising: a memory, and one or more processors connected in communication with the memory; the memory has stored instructions executable by the one or more processors, and the instructions are executed by the one or more processors to cause the one or more processors to implement the steps of the method of any of the above embodiments.

[0016] Embodiments of the present application provide a computer-readable storage medium having stored thereon a computer program, the computer program being executed by a processor to implement the steps of the method of any of the above embodiments.

[0017] Embodiments of the present application provide a computer program product comprising a computer program, the computer program being executed by a processor to implement the steps of the method according to any of the above embodiments.

[0018] The cold station power control method provided by the embodiments of the present application can realize high-precision adjustment of the power of the cold station by obtaining the power regulation curve in advance, and ensure that the output power meets the target control power range. At the same time, based on the environmental parameters and the cold station operation data, the future overall control parameters of the cold station are dynamically predicted, which significantly improves the system response speed and prediction accuracy. Through the comfort prediction model, the method effectively maintains the indoor environment within the reasonable range of human comfort while meeting the target control power range, and optimizes the overall control parameters of the cold station through multiple reasoning when the power or comfort prediction data does not meet the requirements, so as to realize the dynamic balance of power and comfort. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 A flowchart of a cold station power control method provided by an embodiment of the present application is shown in the figure.

[0020] Figure 2 A flowchart of the implementation process of the cold station power control method provided by an embodiment of the present application is shown in the figure.

[0021] Figure 3 A schematic diagram of a power control device of a cold station provided by an embodiment of the present application is shown in the figure.

[0022] Figure 4 A block diagram of an electronic device provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0023] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference signs represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary and are intended to explain the present application, and cannot be understood as limiting the present application.

[0024] In recent years, with the rapid development of power systems, demand-side response and flexible regulation of load have gradually become the focus of attention. During peak load periods, by appropriately reducing the power consumption of part of the load, the power response is carried out according to the set target, which has significant application value. This adjustment method not only effectively relieves the pressure on the power grid, but also improves the utilization efficiency of power resources, and has broad market prospects and great development potential.

[0025] The central air conditioning cold station, as a system commonly operated in summer, plays an important role in load-side power regulation. Aggregating a large number of cold station and other adjustable load resources can generate a large amount of load adjustment space, effectively addressing the contradiction between power supply and demand. However, there are many limitations in related power flexible regulation technology: on the one hand, related means often aim to prioritize meeting the overall building performance, resulting in a significant sacrifice of human comfort, making it difficult to optimize comfort during the adjustment process; on the other hand, the existing technology has low efficiency and poor effect in adjusting the power of the cold station, making it difficult to accurately balance the relationship between power control and comfort, often resulting in adjustment lag, unresponsive and other problems. In addition, the lack of systematic control strategies results in insufficient flexibility and adaptability of cold station power regulation, which cannot effectively respond to complex and changing power demand and environmental conditions.

[0026] In summary, there are still many technical problems in the current field in terms of cold station power flexible regulation, including the accuracy of power prediction, the balance between comfort and cold station power, and the optimization of cold station time response characteristics. Therefore, it is of great practical significance and application value to study a systematic and efficient method for flexible regulation of cold station power.

[0027] Therefore, the purpose of the present application is to propose a cold station power control method that can systematically control the flexibility of cold station power to solve the problems of existing technology in the power regulation process, such as large comfort sacrifice, low regulation efficiency, and response lag. Based on multi-dimensional data such as building energy load and cold station power prediction, comfort prediction, etc., the method constructs an accurate power regulation model to achieve rapid and accurate response to power limits. At the same time, by optimizing the control strategy, the impact of power regulation on human comfort is minimized while ensuring the overall performance of the building, significantly improving the accuracy and adaptability of cold station power regulation.

[0028] Figure 1 A flowchart of a cold station power control method provided by an embodiment of the present application is shown.

[0029] As shown in Figure 1 The present application provides a cold station power control method 100, which comprises:

[0030] Step 110, obtaining the target control power of the cold station.

[0031] Exemplarily, the target control power of the cold station can be read in advance by the demand reading module. For example, the target control power can be data in a power regulation curve issued by the power grid dispatching received and parsed by the module, for example, data in the cold station power regulation curve for the next 24 hours (00:00-24:00) can be obtained and parsed within a time window of 22:00±30 minutes every day. The data in the power regulation curve can be 96 target control powers and other cold station operation data with an interval of 15 minutes, and the power regulation curve can also include an upper threshold of the cold station power.

[0032] Step 120, based on the environmental parameters, the target control power and the cold station operation data, predicting the future cold station overall control parameters.

[0033] Exemplarily, the environmental parameters can be the outdoor temperature and humidity predicted by the weather forecast for the next 24 hours, and the 96 cold station overall control parameters for the next 24 hours can be predicted within a time window of 22:00±30 minutes every day by an reinforcement learning algorithm based on the 96 target control powers and the cold station operation data in the power regulation curve. The time nodes of each cold station control parameter correspond to the time nodes of the 96 target control powers.

[0034] Step 130, based on the cold station overall control parameters, obtaining the power prediction data and the comfort prediction data of the cold station.

[0035] Exemplarily, the power prediction data and the comfort prediction data of the cold station can be predicted by the cold station prediction model and the comfort prediction model, for example, based on the 96 cold station overall control parameters and the environmental parameters, so the number of the power prediction data and the comfort prediction data is also 96, and the time nodes correspond to the time nodes of the 96 cold station overall control parameters.

[0036] Step 140, according to the power prediction data, the comfort prediction data and the cold station overall control parameters, obtaining the control parameter basic data of the cold station.

[0037] Exemplarily, in step 130, when the power prediction data and the comfort prediction data of the 96 cold stations are predicted by the cold station prediction model and the comfort prediction model, the 96 cold station overall control parameters are used, when it is judged that the 96 predicted power prediction data are all in the range of the target control power obtained in step 110, and the 96 comfort prediction data are all in the reasonable range of human comfort, all the cold station overall control parameters can be determined as the control parameter basic data of the cold station. Otherwise, when any data point of the 96 power prediction data and the 96 comfort prediction data does not satisfy the above range, the method of step 120 is needed to re-reason the data point that does not satisfy the above range, and the re-reasoning is performed at most n times, and n can be set to 3 or other values. Taking n = 3 as an example, the data point with the least number of times of not satisfying the range is selected together with other data points satisfying the range as the 96 control parameter basic data of the cold station in the future 24 hours.

[0038] In step 150, the cold station is controlled to run based on the control parameter basic data.

[0039] Exemplarily, the 96 control parameter basic data of the cold station in the future 24 hours are sent to a feedback control module, which can control the cold station to run based on the control parameter basic data, and can also monitor the change of the power output and the comfort of the cold station in real time through the sensor and the data acquisition system during the running of the cold station.

[0040] The cold station power control method provided by the embodiment of the application can realize high-precision adjustment of the cold station power by obtaining the power regulation curve in advance, and ensure that the output power meets the target control power range. Meanwhile, based on the environmental parameters and the cold station running data, the cold station overall control parameters in the future 24 hours are dynamically predicted, which significantly improves the system response speed and the prediction accuracy. Through the comfort prediction model, the method can effectively maintain the indoor environment in the reasonable range of human comfort while meeting the target control power range, and when the power or comfort prediction data does not meet the requirements, the cold station overall control parameters are optimized through multiple reasoning, so as to realize the dynamic balance of the power and the comfort.

[0041] In another embodiment of the application, the control parameter basic data of the cold station is obtained according to the power prediction data, the comfort prediction data and the cold station overall control parameter, including: when the power prediction data is in the range of the target control power, and the comfort prediction data is in the reasonable comfort range, the cold station overall control parameter is determined as the control parameter basic data of the cold station; when the power prediction data is not in the range of the target control power, or the comfort prediction data is not in the reasonable comfort range, the control parameter basic data of the cold station is re-predicted based on the power prediction data and the comfort prediction data.

[0042] Exemplarily, when the power prediction data and the comfort prediction data of the 96 cold stations are predicted, the 96 overall control parameters of the cold stations are used, when it is judged that the 96 predicted power prediction data are all within the range of the target control power obtained in step 110, and the 96 comfort prediction data are all within the reasonable range of the human comfort, all the overall control parameters of the cold stations can be determined as the control parameter basic data of the cold stations. Otherwise, when any one of the 96 power prediction data and the 96 comfort prediction data does not satisfy the above range, the method of step 120 is needed to re-reason the node that does not satisfy the above range, at most 3 times, and select the one with the least number of nodes that do not satisfy the range once, and the other nodes that satisfy the range as the 96 control parameter basic data of the cold stations in the future 24 hours.

[0043] In the embodiment of the present application, the power and the comfort of the 96 time nodes are verified twice to ensure that the prediction data is within the target control power range and the reasonable range of human comfort, thereby generating reliable control parameter basic data; when the prediction data does not meet the requirements, the optimal result is selected through at most 3 times of reasoning optimization, and combined with other nodes that satisfy the conditions, to finally form the control parameter basic data in the future 24 hours. This method realizes the dual goals of power control and comfort guarantee, and improves the accuracy and reliability of the control parameters through the adaptive optimization mechanism, thereby providing technical support for the efficient and stable operation of the cold station.

[0044] In another embodiment of the present application, the control parameter basic data includes a plurality of control parameter basic values corresponding to a plurality of time periods; the cold station is controlled to operate based on the control parameter basic data, including: obtaining an overall control parameter corresponding to a current time period based on a control parameter basic value corresponding to the current time period; obtaining an overall control parameter corresponding to a next time period according to an actual power and a current environment parameter in a case that the cold station is controlled to operate based on the overall control parameter corresponding to the current time period; and controlling the cold station to operate based on the overall control parameter corresponding to the next time period in a case that the next time period arrives.

[0045] In another example, in the case of controlling the operation of the cold station based on the overall control parameter corresponding to the current time period, the overall control parameter corresponding to the next time period is obtained according to the actual power of the operation and the current environmental parameter, including: when the actual power is within the range of the target control power, and the current environmental parameter is within the preset environmental threshold range, determining the control parameter basic value corresponding to the next time period as the overall control parameter corresponding to the next time period; when the actual power is not within the range of the target control power, or the current environmental parameter is not within the preset environmental threshold range, re-predicting the overall control parameter corresponding to the next time period based on the target data from the reference time period to the current time period, wherein the target data includes at least one of the control parameter basic data, the operation data of the cold station, the environmental data, and the comfort data.

[0046] Exemplarily, the control parameter basic data can include 96 control parameter basic values at intervals of 15 minutes. For example, when the cold station operates based on the third control parameter basic value corresponding to the third 15-minute time period, the actual power and the current environmental parameter can be monitored in real time at the same time, such as the humidity and temperature data of the current environment when the cold station operates. If the actual power of the cold station monitored in real time deviates from the target control power by +-5%, or the outdoor temperature in the current environmental parameter deviates from the preset environmental threshold by +-3℃, and the humidity deviation deviates from the preset environmental threshold by +-3RH (Relative Humidity, relative humidity). The fourth control parameter basic value corresponding to the fourth 15-minute time period needs to be re-predicted. For example, the control parameter basic value can be re-predicted for 3 times, among which the power prediction data and the comfort prediction data based on the predicted control parameter basic value are selected, both of which have the smallest deviation from the target control power and the preset environmental threshold. The re-predicted control parameter basic value is taken as the fourth control parameter basic value corresponding to the 4 15-minute time period.

[0047] In the embodiments of the present application, by monitoring the actual power and the environmental parameter of the cold station in real time, when the deviation exceeds the preset threshold (such as power deviation +-5%, temperature deviation +-3℃, humidity deviation +-3RH), the control parameter basic data is dynamically optimized, and the control parameter basic value that makes the actual power of the cold station deviate from the target control power as small as possible is re-predicted and selected. Through the embodiments, adaptive adjustment in the operation process of the cold station can be realized, the real-time matching of the power control precision and the environmental parameter change is ensured, the stability and energy efficiency performance of the operation of the cold station are significantly improved, and reliable technical support is provided for precise control in complex environments.

[0048] In another embodiment of the present application, the power prediction data of the cold station is predicted by a power prediction model, the power prediction model comprising a plurality of long short-term memory network layers and a fully connected layer, and the power prediction model is trained by: obtaining power sample data and power labels corresponding to the power sample data, wherein the power sample data comprises outdoor temperature, outdoor humidity, and overall control parameters; processing the power sample data by the plurality of long short-term memory network layers and the fully connected layer to obtain power prediction results; determining the error between the power prediction results and the power labels by a root mean square error method, and adjusting the model parameters of the power prediction model based on the error to obtain a trained power prediction model.

[0049] Exemplarily, the power prediction data of the cold station is predicted by a power prediction model, the power prediction model adopting a deep learning network architecture, comprising an input layer, a long short-term memory network layer (LSTM for short), a fully connected layer, and an output layer. The input layer can be used to receive power sample data, such as outdoor temperature, outdoor humidity, and overall control parameters of the cold station. The long short-term memory network layer is configured, for example, with 3 layers, and the number of neurons in each layer is, for example, 128, 64, and 32, respectively, for extracting time series features. The fully connected layer is configured, for example, with 2 layers, and the number of neurons in each layer is, for example, 16 and 1, respectively, for generating power prediction data. The output layer is used to output the power prediction results.

[0050] Exemplarily, in the training phase of the power prediction model, the overall control parameters of the cold station, the cold station operation data, the outdoor temperature, the outdoor humidity, and the comfort data of the cold station in the recent period (e.g., the previous 365 days) can be obtained from the operation log of the cold station as basic training data, the power prediction model is trained, the basic training data is input into the long short-term memory network layer and the fully connected layer of the power prediction model, and the power prediction results in the training phase are generated. The root mean square error (RMSE for short) can also be used to calculate the error between the power prediction results and the power labels. For example, the total time cost of model training is not more than 2 hours.

[0051] Exemplarily, in the optimization phase of the power prediction model, the power sample data and the corresponding power labels can be obtained from the operation log of the cold station at 22 o'clock at the end of each month, and the resolution of the data time can be 15 minutes. The power prediction model is optimized and evaluated using the obtained power sample data and the corresponding power labels, and the entire process takes no more than 2 hours. The cold station operation data of the previous 365 days, including the power of the cold station, the outdoor temperature, the outdoor humidity, the overall control parameters of the cold station and the comfort data, can be further read, and the power prediction model is optimized again. When the optimized power prediction model is evaluated using RMSE, the model parameters of the power prediction model can be adjusted based on the error, and if the RMSE of the new model is lower than that of the current model, the current model can be replaced with the new model.

[0052] In an embodiment of the present application, by constructing a power prediction model based on a deep learning network architecture, the model is trained using historical operation data of the cold station (such as power, outdoor temperature, humidity, overall control parameters, etc. in the previous 365 days), and the prediction accuracy is evaluated by RMSE; at 22 o'clock at the end of each month, the model is dynamically optimized, retrained and evaluated using the latest data, and if the accuracy of the new model is improved, the current model is replaced, and the training and optimization process is completed within 2 hours. This technical means can effectively improve the accuracy of the power prediction of the cold station and dynamically optimize it, which can adapt to the dynamic changes of different external variables. The embodiments provided in the present application can enhance the adaptability of the model to different external variables by unified coding and integration of multiple external variables. At the same time, by means of efficient data processing and model updating mechanism, and by introducing multiple external variables and improving the model structure, the embodiments provided in the present application can more accurately predict the power of the cold station. Compared with the traditional method, the prediction error is reduced by more than 20% in a typical scenario.

[0053] In another embodiment of the present application, the comfort prediction data of the cold station is obtained by a comfort prediction model, the comfort prediction model includes a plurality of long short-term memory network layers and a full connection layer, and the comfort prediction model is trained in the following manner: obtaining comfort sample data and comfort labels corresponding to the comfort sample data, wherein the comfort sample data includes at least one of outdoor temperature, outdoor humidity, overall control parameters, and weighted number of manual operations, wherein the weighted number of manual operations includes the weighted average number of the number of temperature increasing operations, the number of temperature decreasing operations, and the number of air speed adjusting operations; the comfort sample data is processed by using a plurality of long short-term memory network layers and a full connection layer to obtain a comfort prediction result; the error between the comfort prediction result and the comfort label is determined by the root mean square error method, and the model parameters of the comfort prediction model are adjusted based on the error to obtain a trained comfort prediction model.

[0054] Exemplarily, the structure of the comfort degree prediction model and the process of training and optimization are similar to the structure of the power prediction model and the process of training and optimization described above, and are not repeated here. Different from the process of training and optimization of the comfort degree prediction model, the data used in the process includes not only the outdoor temperature, the humidity, and the overall control parameter of the cold station, but also the historical data of the weighted number of terminal manual operations (historical comfort degree data), and the weighted number of terminal manual operations includes the weighted average number of times of temperature increasing operation, temperature decreasing operation, and air speed adjusting operation.

[0055] In the embodiment of the present application, by constructing a comfort degree prediction model based on LSTM and a full connection layer, training and optimizing the model based on historical data, and evaluating the prediction accuracy by RMSE, the model is dynamically updated at the end of each month to improve performance. While significantly improving the power control accuracy of the cold station, the impact of the comfort degree in the building is effectively minimized, achieving the dual goals of flexible power control and comfort degree guarantee in the building.

[0056] In another embodiment of the present application, the control parameter basic data of the cold station is obtained by a reinforcement learning model, and the reinforcement learning model is trained in the following manner: obtaining cold station historical data, wherein the cold station historical data includes at least one of historical control parameter basic data, historical running data, historical outdoor temperature, historical outdoor humidity, and historical comfort degree data; and training the reinforcement learning model based on the cold station historical data.

[0057] The embodiment of the present application adopts a reinforcement learning model to automatically optimize the overall control parameter based on the outdoor temperature, the outdoor humidity, and the target control power of the cold station, so as to maximize the efficiency of the cold station and minimize the weighted number of terminal manual operations under the condition of meeting the power demand of the cold station.

[0058] In another embodiment of the present application, the reinforcement learning model is trained based on cold station historical data, including: training the reinforcement learning model according to a reward function based on the cold station historical data, wherein the reward function is associated with at least one of a cold station efficiency reward function, a comfort degree sample data penalty function, a constraint penalty function, and a final reward function; wherein the cold station efficiency reward function is associated with the real-time refrigerating capacity, the real-time power, and the reward coefficient of the cold station; wherein the comfort degree sample data penalty function is associated with the weighted number of manual operations and the operation weighted number penalty coefficient; and wherein the constraint penalty function is associated with the demand power, the real-time power, and the penalty coefficient of the cold station.

[0059] Exemplarily, the design of the reinforcement learning model also includes a state space, an action space, a reward function, and the like. The state space describes the running state of the current cold station system, including: chilled water supply temperature, chilled water return temperature, cooling water supply temperature, cooling water return temperature, cooling water flow, chilled water flow, outdoor temperature, outdoor humidity, end manual operation weighted times, cold station power demand, cold station real-time power, and cold station real-time refrigeration capacity. The action space describes the controllable control parameters, i.e., the overall control parameters.

[0060] The reward function is used to guide the optimization goal of the reinforcement learning model, and is designed as follows:

[0061] The cold station efficiency reward function is:

[0062] R cop =C real / P real *W c

[0063] Wherein, C real is the cold station real-time refrigeration capacity, P real is the cold station real-time power, and W c is the reward coefficient of the cold station efficiency, and the goal is to maximize it.

[0064] The comfort sample data penalty function is:

[0065] R op =-O*W o

[0066] Wherein, C is the end manual operation weighted times, and W o is the penalty coefficient of the operation weighted times, and the goal is to minimize it.

[0067] The constraint penalty function is:

[0068] R v =-P set / P real *W v

[0069] Wherein, P set / P real is the degree of unmet cold station power demand, P set is the demand power, P real is the cold station real-time power, and W v is the penalty coefficient.

[0070] The final reward function is:

[0071] R=R cop +R op +R V

[0072] The embodiment of the present application can also provide a feedback control model. The model receives the overall control parameters output by the reinforcement learning model, monitors the power output and comfort change of the cold station in real time, adjusts the start-stop and frequency of the cold station, cooling pump, refrigeration pump, cooling tower and other equipment, and the refrigerated water inlet and outlet water, cooling water inlet and outlet water parameter setting, adjusts the operation state of the cold station system equipment based on the monitored data through the feedback control algorithm, so that the deviation of the predicted power from the target control power is adjusted, and the actual power of the cold station is ensured to meet the range of the target control power. During the adjustment process of the cold station system equipment, the time response characteristics of different equipment can also be given, and the actual power of the cold station, comfort data and equipment energy consumption are comprehensively considered to optimize and select multiple output targets.

[0073] Illustratively, the reinforcement learning model can also be used to adjust the control parameter base value of the next time period. For example, after obtaining the control parameter base value of the next day, the feedback control model controls the cold station equipment (chiller, cooling pump, refrigeration pump, cooling tower, etc.) based on the control parameter base value. During operation, the actual power deviation of the cold station from the power control target and the outdoor temperature and humidity data deviation from the environmental parameters are checked every 15 minutes. When the power deviation is +-5% or the outdoor temperature deviation is +-3 degrees or the humidity deviation is +-3 RH, the overall control parameters, operation data, outdoor temperature and humidity, and the number of weighted times of terminal manual operation are used to infer the overall control parameters of the next 15 minutes. Similarly, the overall control parameters are inferred at most 3 times, and the overall control parameters with the minimum deviation between the predicted power data and the target control power and the minimum number of weighted times of terminal manual operation are selected as the next 15-minute control parameters and issued to the feedback control model; if the external data change amplitude does not exceed the control threshold, the overall control parameters at this time point are directly issued to the feedback control model.

[0074] In the embodiment of the present application, the outdoor temperature, humidity and target control power are combined to automatically optimize the overall control parameters, so that the cold station power demand is met, the cold station efficiency is maximized, and the number of weighted times of terminal manual operation is minimized. By designing the state space, action space and reward function, the real-time operation state of the cold station, the refrigeration efficiency, the frequency of manual operation and the power demand constraint are considered to construct a multi-objective optimization mechanism. This method significantly improves the operation efficiency of the cold station, reduces the frequency of manual intervention, and ensures the accuracy of power control.

[0075] Figure 2 The flowchart of the implementation process of the cold station power control method provided by an embodiment of the present application is shown.

[0076] The present application provides an example, and the content is described in combination with Figure 2 to better illustrate the content of the technical solutions provided by the present application:

[0077] A centralized cooling plant in an office park consists of n chillers, chilled water pumps, cooling pumps, and cooling towers. The original control system was set to a fixed chilled water outlet temperature and adjusted the frequency of the chilled and cooling water pumps based on pressure. The number of cooling towers was generally controlled according to the number of chillers. This system could only perform basic cooling control. Power was adjusted by increasing or decreasing the number of chillers in operation, with power adjustments in 1 / n increments. It could not flexibly adjust power, nor could it pursue optimal system efficiency or maintain comfort when adjusting power.

[0078] After adding the flexible power control method based on this invention, the cooling plant can achieve a precision adjustment of 1kW through predictive optimization adjustment based on weather and comfort feedback, while ensuring the operating efficiency of the cooling plant and the comfort of the terminal space.

[0079] During the model training phase, the historical operation records or personnel records of the original system (such as environmental parameters, real-time power of the chiller plant, chilled water supply temperature, chilled water return temperature, chilled water flow rate, cooling water supply temperature, cooling water return temperature, chiller set temperature, chiller on / off status, chilled pump frequency, cooling pump frequency, chiller tower on / off status, chiller tower frequency, etc.) are first entered into the system. After the data is entered, the power prediction model, comfort prediction model, and reinforcement learning model are automatically trained. Once the training is complete, flexible power control can begin.

[0080] like Figure 2 As shown, in the stage of flexible power control of the power plant using a model, for example, the demand input module can obtain the target control power of the chiller in advance. Then, the predictive control module can output the overall control parameters of the chiller based on environmental parameters, target control power, and chiller operation data, and predict changes in power and comfort. The predictive control module can include a power prediction model, a comfort prediction model, and a reinforcement learning model. When the predictive control module determines, based on the power prediction model and the comfort prediction model, that the predicted power data is within the range of the target control power and the predicted comfort data is also within the reasonable range of human comfort, the power prediction data, comfort prediction data, and the overall control parameters of the chiller can be used to obtain the basic control parameter data of the chiller, which is then determined as the overall control parameters. The chiller system is then controlled to operate based on these overall control parameters.

[0081] During the operation of the cold station system, the prediction control module can also check the actual power and the current environmental parameters of the cold station system every interval preset time period (such as 15 minutes), and when it is found that the actual power is not within the range of the target control power or the environmental parameters are not within the preset environmental threshold range, the corresponding overall control parameters of the next time period are re-predicted. At this time, the power prediction model and the comfort prediction model based on the re-predicted overall control parameters are used to predict the power prediction data and the comfort prediction data, and compared with the previous data, there are power variables and comfort variables.

[0082] During the operation of the cold station system, the feedback control module can also be used to monitor the power output and the comfort change of the cold station system in real time, and the start-stop and frequency of the cold machine, the cooling pump, the refrigeration pump, the cooling tower and the like and the refrigeration water inlet and outlet water parameter setting and the like are adjusted to feedback and fine-tune the deviation of the actual power of the cold station system from the target control power. During the adjustment process, the time response characteristics of different devices can be classified, the adjustable parameters of the cold station are obtained, and the power of the cold station is optimized and selected for multiple output targets. Through the cooperative work of the above modules, the flexible adjustment of the power of the cold station can be realized, the power demand of the power system can be met, the comfort in the building can be ensured, the stable operation of the power system and energy saving and emission reduction can be contributed.

[0083] The power prediction model, the comfort prediction model and the reinforcement learning model can be continuously automatically trained and optimized in the operation according to the method of the embodiment of the application; the overall control parameters (including the refrigeration water set temperature, the number of cold machine start-ups, the refrigeration water pump frequency, the cooling pump frequency, the number of cooling towers, the cooling tower frequency, etc.) output by the reinforcement learning model are transmitted to the reinforcement learning model, the reinforcement learning model dynamically adjusts the control parameters according to the actual data change and outputs them to the feedback control model, and the feedback control model is transmitted to the controlled devices (the cold machine, the refrigeration water pump, the cooling pump and the cooling tower).

[0084] Figure 3 A schematic diagram of a power control device of a cold station provided by the embodiment of the application.

[0085] The embodiment of the application also provides a power control device 300 of a cold station, characterized in that the device 300 comprises:

[0086] The acquisition module 310 is configured to acquire the target control power of the cold station.

[0087] The prediction module 320 is configured to predict the overall control parameters of the cold station in the future based on the environmental parameters, the target control power and the cold station operation data, and to obtain the power prediction data and the comfort prediction data of the cold station based on the overall control parameters of the cold station.

[0088] The reasoning module 330 is configured to obtain the control parameter basic data of the cold station according to the cold station power prediction data, the comfort prediction data, and the cold station overall control parameter.

[0089] The control module 340 is configured to control the cold station to operate based on the cold station control parameter basic data.

[0090] It can be understood that the specific description of the power control device 300 of the cold station can refer to the description of the power control method 100 of the cold station in the foregoing.

[0091] In some embodiments, the reasoning module 330 is further configured to: when the power prediction data is within the range of the target control power, and the comfort prediction data is within the reasonable comfort range, determine the cold station overall control parameter as the control parameter basic data of the cold station; and when the power prediction data is not within the range of the target control power, or the comfort prediction data is not within the reasonable comfort range, re-predict the control parameter basic data of the cold station based on the power prediction data and the comfort prediction data.

[0092] In some embodiments, the control parameter basic data includes a plurality of control parameter basic values corresponding to a plurality of time periods; and the control module 340 is further configured to: obtain the overall control parameter corresponding to the current time period based on the control parameter basic value corresponding to the current time period; obtain the overall control parameter corresponding to the next time period according to the actual power of the operation and the current environment parameter in a case that the cold station is controlled to operate based on the overall control parameter corresponding to the current time period; and control the cold station to operate based on the overall control parameter corresponding to the next time period in a case that the next time period arrives.

[0093] In some embodiments, in a case that the cold station is controlled to operate based on the overall control parameter corresponding to the current time period, the overall control parameter corresponding to the next time period is obtained according to the actual power of the operation and the current environment parameter, including: when the actual power is within the range of the target control power, and the current environment parameter is within the preset environment threshold range, determining the control parameter basic value corresponding to the next time period as the overall control parameter corresponding to the next time period; and when the actual power is not within the range of the target control power, or the current environment parameter is not within the preset environment threshold range, re-predicting the overall control parameter corresponding to the next time period based on target data from the reference time period to the current time period, wherein the target data includes at least one of the control parameter basic data, the operation data of the cold station, the environment data, and the comfort data.

[0094] In some embodiments, the power prediction data of the cold station is predicted by a power prediction model including a plurality of long short-term memory network layers and a fully connected layer, and the device 300 further includes: a first acquisition module configured to acquire power sample data and power labels corresponding to the power sample data, wherein the power sample data includes outdoor temperature, outdoor humidity, and overall control parameters; a first prediction module configured to process the power sample data by using the plurality of long short-term memory network layers and the fully connected layer to obtain power prediction results; and a first adjustment module configured to determine errors between the power prediction results and the power labels by using a root mean square error method, and adjust model parameters of the power prediction model based on the errors to obtain a trained power prediction model.

[0095] In some embodiments, the comfort prediction data of the cold station is obtained by a comfort prediction model including a plurality of long short-term memory network layers and a fully connected layer, and the device 300 further includes: a second acquisition module configured to acquire comfort sample data and comfort labels corresponding to the comfort sample data, wherein the comfort sample data includes at least one of outdoor temperature, outdoor humidity, overall control parameters, and weighted number of manual operations, wherein the weighted number of manual operations includes a weighted average number of times of heating operation, times of cooling operation, and times of adjusting air speed operation; a second prediction module configured to process the comfort sample data by using the plurality of long short-term memory network layers and the fully connected layer to obtain comfort prediction results; and a second adjustment module configured to determine errors between the comfort prediction results and the comfort labels by using a root mean square error method, and adjust model parameters of the comfort prediction model based on the errors to obtain a trained comfort prediction model.

[0096] In some embodiments, the control parameter basic data of the cold station is obtained by a reinforcement learning model, and the device 300 further includes: a third acquisition module configured to acquire cold station historical data, wherein the cold station historical data includes at least one of historical control parameter basic data, historical operation data, historical outdoor temperature, historical outdoor humidity, and historical comfort data; and a third training module configured to train the reinforcement learning model based on the cold station historical data.

[0097] In some embodiments, the third training module is further configured to train the reinforcement learning model according to a reward function based on the cold station historical data, wherein the reward function is associated with at least one of a cold station efficiency reward function, a comfort sample data penalty function, a constraint penalty function, and a final reward function; wherein the cold station efficiency reward function is associated with real-time refrigeration capacity, real-time power, and a reward coefficient of the cold station; wherein the comfort sample data penalty function is associated with the weighted number of manual operations and an operation weighted number penalty coefficient; and wherein the constraint penalty function is associated with demand power, real-time power, and a penalty coefficient of the cold station.

[0098] Embodiments of the present application provide an electronic device, comprising a memory and one or more processors connected with the memory; the memory stores instructions executable by the one or more processors, and the instructions are executed by the one or more processors to enable the one or more processors to implement the steps of the method of any of the above embodiments.

[0099] Embodiments of the present application provide a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the method of any of the above embodiments.

[0100] Embodiments of the present application provide a computer program product, which comprises instructions, and the instructions are executed by a processor of a computer device to enable the computer device to perform the steps of the method of any of the above embodiments.

[0101] Figure 3 A block diagram of an electronic device according to an embodiment of the present application is provided.

[0102] Embodiments of the present application provide an electronic device, comprising a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the method of any of the above embodiments.

[0103] As shown in Figure 4 , for ease of understanding, embodiments of the present application show a specific electronic device 400.

[0104] The electronic device 400 is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit the implementations of the present disclosure described and / or claimed in this document.

[0105] As shown in Figure 4 , the device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 402 or a computer program loaded from a storage unit 408 into a random access memory (RAM) 403. In the RAM 403, various programs and data required for the operation of the electronic device 400 can also be stored. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other through a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0106] A plurality of components in the electronic device 400, including an input unit 406, e.g., a keyboard, a mouse, etc., an output unit 407, e.g., various types of displays, speakers, etc., a storage unit 408, e.g., a magnetic disk, an optical disk, etc., and a communication unit 409, e.g., a network card, a modem, a wireless communication transceiver, etc., are connected to the I / O interface 405. The communication unit 409 allows the electronic device 400 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0107] The computing unit 401 can be various general and / or special purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 401 performs various methods described above. For example, in some embodiments, any one or more of the above methods can be implemented as a computer software program, which is tangibly embodied in a machine-readable medium, such as the storage unit 408. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 400 via the ROM 402 and / or the communication unit 409. When the computer program is loaded onto the RAM 403 and executed by the computing unit 401, one or more steps of any one or more of the above methods can be performed. Alternatively, in other embodiments, the computing unit 401 can be configured to perform any one or more of the above methods by other any appropriate means, e.g., by means of firmware.

[0108] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequential list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this application, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0109] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0110] In the description of this application, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this application, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0111] In the description of the present application, it needs to be understood that the orientation or positional relationship indicated by the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like is based on the orientation or positional relationship shown in the drawings, and is only for the purpose of facilitating the description of the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application.

[0112] In addition, the terms "first", "second", and the like used in the embodiments of the present application are only for the purpose of description, and cannot be understood as indicating or implying relative importance, or implicitly indicating the number of technical features referred to in the embodiments. Therefore, the features defined with the terms "first", "second" and the like in the embodiments of the present application can be explicitly or implicitly indicated to include at least one of the features. In the description of the present application, the meaning of the word "plurality" is at least two or two or more, such as two, three, four, etc., unless otherwise specifically limited in the embodiments.

[0113] In the present application, unless otherwise specifically defined or limited in the embodiments, the terms "mounting", "connecting", "connecting" and "fixing" and the like appearing in the embodiments should be broadly understood, for example, the connection can be a fixed connection, or a detachable connection, or integrated, which can be understood, or can be a mechanical connection, an electrical connection, etc. Of course, it can also be directly connected, or indirectly connected through an intermediate medium, or it can be the internal communication of two elements, or the interaction relationship between two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific implementation situation.

[0114] In the present application, unless otherwise specifically defined or limited, the first feature "on" or "under" the second feature can be direct contact between the first and second features, or indirect contact between the first and second features through an intermediate medium. Moreover, the first feature "above", "above" and "above" the second feature can be directly above or obliquely above the first feature, or only indicate that the horizontal height of the first feature is higher than that of the second feature. The first feature "below", "below" and "below" the second feature can be directly below or obliquely below the first feature, or only indicate that the horizontal height of the first feature is less than that of the second feature.

Claims

1. A power control method for a chiller plant, characterized in that, The method includes: Obtain the target control power of the cooling station; Based on environmental parameters, the target control power, and chiller plant operation data, predict the future overall control parameters of the chiller plant; Based on the overall control parameters of the cooling plant, the power prediction data and comfort prediction data of the cooling plant are obtained; Based on the power prediction data, the comfort prediction data, and the overall control parameters of the cooling station, the basic control parameter data of the cooling station is obtained; wherein, obtaining the basic control parameter data of the cooling station based on the power prediction data, the comfort prediction data, and the overall control parameters of the cooling station includes: when the power prediction data is within the range of the target control power and the comfort prediction data is within the reasonable comfort range, the overall control parameters of the cooling station are determined as the basic control parameter data; The operation of the cooling plant is controlled based on the aforementioned control parameter data. The basic control parameter data includes multiple basic control parameter values ​​corresponding to multiple time periods. Controlling the operation of the chiller plant based on the basic control parameter data includes: obtaining the overall control parameters of the chiller plant for the current time period based on the basic control parameter values ​​for the current time period; when controlling the operation of the chiller plant based on the overall control parameters for the current time period, obtaining the overall control parameters of the chiller plant for the next time period based on the actual operating power and current environmental parameters; and when the next time period arrives, controlling the operation of the chiller plant based on the overall control parameters for the next time period. Specifically, when the operation of the cooling station is controlled based on the overall control parameters of the cooling station corresponding to the current time period, the overall control parameters of the cooling station corresponding to the next time period are obtained according to the actual operating power and the current environmental parameters. This includes: when the actual power is within the range of the target control power and the current environmental parameters are within the range of the preset environmental threshold, the basic value of the control parameters corresponding to the next time period is determined as the overall control parameters of the cooling station corresponding to the next time period.

2. The method according to claim 1, characterized in that, The step of obtaining the basic control parameter data of the cooling station based on the power prediction data, the comfort prediction data, and the overall control parameters of the cooling station further includes: When the power prediction data is not within the range of the target control power, or the comfort prediction data is not within the range of reasonable comfort, the basic data of the control parameters are re-predicted based on the power prediction data and the comfort prediction data.

3. The method according to claim 1, characterized in that, When controlling the operation of the cooling plant based on the overall control parameters of the cooling plant corresponding to the current time period, the method for obtaining the overall control parameters of the cooling plant corresponding to the next time period based on the actual operating power and current environmental parameters further includes: When the actual power is not within the range of the target control power, or when the current environmental parameters are not within the range of the preset environmental threshold, the overall control parameters of the cooling station corresponding to the next time period are re-predicted based on the target data from the reference time period to the current time period. The target data includes at least one of the basic control parameter data, the cooling station's operating data, environmental data, and comfort data.

4. The method according to any one of claims 1-3, characterized in that, The power prediction data of the cooling plant is obtained through a power prediction model, which includes multiple long short-term memory network layers and fully connected layers. The power prediction model is trained in the following way: Obtain power sample data and the corresponding power tags, wherein the power sample data includes outdoor temperature, outdoor humidity, and overall control parameters; The power sample data is processed using multiple long short-term memory network layers and fully connected layers to obtain power prediction results; The error between the power prediction result and the power label is determined by the root mean square error method, and the model parameters of the power prediction model are adjusted based on the error to obtain a trained power prediction model.

5. The method according to any one of claims 1-3, characterized in that, The comfort prediction data for the cooling station is obtained through a comfort prediction model, which includes multiple long short-term memory network layers and fully connected layers. The comfort prediction model is trained in the following way: Obtain comfort sample data and corresponding comfort labels for the comfort sample data. The comfort sample data includes at least one of outdoor temperature, outdoor humidity, overall control parameters, and weighted number of manual operations. The weighted number of manual operations includes a weighted average of the number of heating operations, the number of cooling operations, and the number of wind speed adjustment operations. The comfort sample data is processed using multiple long short-term memory network layers and fully connected layers to obtain comfort prediction results; The error between the comfort prediction result and the comfort label is determined by the root mean square error method, and the model parameters of the comfort prediction model are adjusted based on the error to obtain a trained comfort prediction model.

6. The method according to any one of claims 1-3, characterized in that, The basic data of the control parameters of the cooling plant are obtained through a reinforcement learning model, which is trained in the following way: Acquire historical data of the cooling plant, wherein the historical data of the cooling plant includes at least one of historical control parameter basic data, historical operating data, historical outdoor temperature, historical outdoor humidity and historical comfort data; The reinforcement learning model is trained based on the historical data of the cold storage station.

7. The method according to claim 6, characterized in that, Based on the historical data of the cold storage station, the reinforcement learning model is trained, including: Based on the historical data of the cooling station, the reinforcement learning model is trained according to the reward function, wherein the reward function is associated with at least one of the following: cooling station efficiency reward function, comfort sample data penalty function, constraint penalty function, and final reward function; The efficiency reward function of the cooling plant is related to the real-time cooling capacity, real-time power, and reward coefficient of the cooling plant. The comfort sample data penalty function is associated with the weighted number of manual operations and the penalty coefficient for the weighted number of operations. The constraint penalty function is associated with the cooling station's required power, real-time power, and penalty coefficient.

8. A power control device for a chiller plant, characterized in that, The device includes: The acquisition module is used to acquire the target control power of the cooling plant; The prediction module is used to predict the future overall control parameters of the cooling station based on environmental parameters, the target control power, and the cooling station operation data. It is also used to obtain the power prediction data and comfort prediction data of the cooling station based on the overall control parameters of the cooling station. The inference module is used to obtain basic control parameter data of the cooling station based on the power prediction data, the comfort prediction data, and the overall control parameters of the cooling station; wherein, obtaining the basic control parameter data of the cooling station based on the power prediction data, the comfort prediction data, and the overall control parameters of the cooling station includes: when the power prediction data is within the range of the target control power and the comfort prediction data is within the reasonable comfort range, the overall control parameters of the cooling station are determined as the basic control parameter data; The control module is used to control the operation of the cooling plant based on the control parameter basic data; The basic control parameter data includes multiple basic control parameter values ​​corresponding to multiple time periods. Controlling the operation of the chiller plant based on the basic control parameter data includes: obtaining the overall control parameters of the chiller plant for the current time period based on the basic control parameter values ​​for the current time period; when controlling the operation of the chiller plant based on the overall control parameters for the current time period, obtaining the overall control parameters of the chiller plant for the next time period based on the actual operating power and current environmental parameters; and when the next time period arrives, controlling the operation of the chiller plant based on the overall control parameters for the next time period. Specifically, when the operation of the cooling station is controlled based on the overall control parameters of the cooling station corresponding to the current time period, the overall control parameters of the cooling station corresponding to the next time period are obtained according to the actual operating power and the current environmental parameters. This includes: when the actual power is within the range of the target control power and the current environmental parameters are within the range of the preset environmental threshold, the basic value of the control parameters corresponding to the next time period is determined as the overall control parameters of the cooling station corresponding to the next time period.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1-7.

Images