TWI516886B - Intelligent learning energy-saving control system and method thereof - Google Patents

Description

Intelligent learning energy-saving regulation system and method

This case is about an air conditioning control technology, especially an air conditioning control technology that continuously carries out intelligent learning and regulation.

There are many ways to improve the air conditioning system, such as multi-turning compound control or replacement equipment. However, if the control target can be minimized and the smart technology is introduced, the minimum operating range can be achieved under the existing air conditioning system structure. The minimum control project achieves energy savings goals, thereby saving costs and avoiding the risk of disruptive construction.

According to the data analysis results of air-conditioning system, in terms of internal factors, the energy consumption of the ice water host is the main cause, and the temperature difference of ice water is one of the key controllable factors, that is, the larger the temperature difference is, the more energy is consumed, and it will be It affects the three cycles of ice water, cooling water and refrigerant of the air conditioning system. Therefore, in order to minimize the control target, it is the primary task to control the ice pump.

As the flow rate of the ice water pump is controlled by the ice water pump, the higher the frequency of the variable frequency motor of the ice water pump is, the faster the ice water flow rate will be. Therefore, the return water temperature of the ice water will be closer to the set water temperature. The ice water temperature difference will become smaller. This result will reduce the energy consumption of the ice water host, but it will cause The energy consumption of the motor of the ice pump is increased, and the overall energy consumption does not necessarily decrease. On the contrary, the lower the motor frequency of the ice pump is set, the speed and the ice water flow will be slower. However, although this saves the energy consumption of the ice pump, it will cause the energy consumption of the ice water host. increase.

Therefore, how to find the overall energy consumption is the smallest, and to achieve the best balance between the energy consumption of the ice water host and the motor energy consumption of the ice water pump, that is, according to the current real-time situation, the appropriate use can achieve this performance. The best settings are adjusted at the same time.

In view of the current goals of the industry, the main purpose of this case is to provide an air conditioning control technology that can continuously carry out intelligent learning and regulation.

In order to achieve the above purposes and other purposes, the present invention provides an intelligent learning energy-saving control method, including: setting system control data and capturing instant sensing data for the air-conditioning system; according to the set system control data and the instantaneous sensing acquired Data, using empirical inference model modeling techniques for empirical learning and logical derivation, establishing and packaging temperature difference and energy consumption inferential models; and using the temperature difference and energy consumption inferential model, and the instantaneous sensing from the air conditioning system Data, using data exploration technology to infer the recommendations for energy-saving optimization, and then through the knowledge mining engine technology to make continuous adjustment of the air-conditioning system.

Furthermore, the present invention also provides an intelligent learning energy-saving control system, including: a system control data setting module, which is used to set system control data and capture real-time sensing data for the air-conditioning system; experience learning and logic derivation module, According to the set system control data and the acquired real-time sensing data, the model learning technology is used to conduct empirical learning and logic deduction, and the temperature difference and energy consumption inferential model are established and encapsulated; knowledge discovery and system control module The system is used to utilize the temperature difference and energy consumption inferential model, and the instantaneous sensing data extracted from the air conditioning system, and the data exploration technology in the data exploration module to infer the energy saving optimization setting suggestion, and then through the knowledge mining engine Technology to continuously control the air conditioning system.

Compared with the prior art, the intelligent learning energy-saving control system and method in this case can continuously use the instantaneous temperature difference and energy-consumption inference model to optimize setting and adaptability, so it can consume energy in the ice water host. The best balance between the motor energy consumption of the ice pump and the overall energy saving setting is achieved.

1‧‧‧Intelligent Learning Energy Conservation Control System

10‧‧‧System Control Data Setting Module

11‧‧‧Experience Learning and Logic Derivation Module

12‧‧‧Knowledge Discovery and System Control Module

20 to 23 ‧ ‧ time section

a to p‧‧‧ procedures

Figure 1 is a schematic diagram of the system and process of the intelligent learning energy-saving control system and method of the present invention; and Figure 2 is a timing diagram of the continuous regulation according to the present case.

In order to understand the technical features, contents and advantages of the present invention and the effects thereof, the invention disclosed herein will be described with reference to the accompanying drawings, and the expressions of the embodiments are as follows. The subject matter is only for the purpose of illustration and supplementary explanation. It may not be true proportion and precise configuration after the implementation of the case. Therefore, the scope of the attached drawings and the relationship between the configuration should not be disclosed in the actual implementation scope. .

Please refer to Figure 1 for the intelligent learning energy-saving control system and method applied to the air-conditioning system provided in this case. It should be noted that the intelligent learning energy-saving control system 1 of the embodiment includes a system control data setting module 10, an empirical learning and logic derivation module 11, and a knowledge discovery and system control module 12 to perform the intelligence provided by the present invention. Learning the energy-saving control method, however, the system control data setting module 10, the experience learning and logic derivation module 11, and the knowledge discovery and system control module 12 can also selectively merge or separate and physically and virtually according to different needs. The setting, that is, the intelligent learning energy-saving regulation method provided by the present invention can also be implemented differently from the architecture of the intelligent learning energy-saving control system 1 of the present disclosure. It is necessary to first propose that the intelligent learning energy-saving control system and method of this case can be used for immediate or non-instant scheduling control.

Before implementing the case, the air conditioning system in the building can be configured to measure the sensors required for the target data, and the sensors continuously transmit the real-time monitoring data to the back-end servo host or the information system platform via the communication technology. The way of storing data can be database, data storage, or file system, and can be used as a source of data for statistics and analysis in this case. The data content can be time-series and is a history of the on-site measurements. The control platform of the air conditioning system can instantly set and control related equipment.

In the implementation of the case, the system control data setting module 10 can first set the system control data and capture the instant sensing data for the air conditioning system; then, the experience learning and logic derivation module 11 will adjust the data according to the set system and capture the instant. Sensing data, using empirical inference model modeling techniques for empirical learning and logic derivation, and establishing and encapsulating temperature difference and energy dissipative inference models; After that, the knowledge discovery and system control module 12 will use the temperature difference and energy consumption inference model, and the instantaneous sensing data extracted from the air conditioning system, and use the data exploration technology to infer the energy saving optimization setting suggestion in the data exploration module. And then through the knowledge mining engine technology to continuously adjust the air conditioning system. The data mining technology refers to automatically searching for a calculation process hidden in a large amount of data and having a special relevance. For details, refer to related documents in the technical field, and no further details are provided herein.

In an implementation mode, the intelligent learning energy-saving control system 1 can further determine the compliance according to the system control data for continuously adapting the air-conditioning system and the instantaneous sensing data of the instant sensing data from the air-conditioning system. The re-modeling condition, if so, activates the empirical learning and logic derivation module 11 again, and if not, activates the knowledge discovery and system control module 12 again. This is the program p that executes the schema.

Further, the system control data setting module 10 can first set the source of the instant sensing data, the target, and the method, that is, execute the program a of the schema; and then set the scope of the data timing analysis, that is, the program for executing the schema b; setting the effective data sampling condition, that is, executing the program c of the schema; then setting the continuous learning and fitness adjustment procedure, that is, executing the program d of the schema.

The empirical learning and logic derivation module 11 can first extract and filter the relevant modeling data from the set data source, that is, execute the program of the schema e; then load the modeling data for standardization and standardization processing, That is, the program f is executed; the model data is used to establish the temperature difference adaptive inference model through the adaptive inference model modeling technique, and the program g of the schema is executed; Based on the inferential model modeling technique, and based on the inference results of the temperature difference inferential model, the energy consumption appropriate inference model is further established, and the program h of the schema is executed. The second inferiority model is used to compare the previously established temperature difference and consumption. The adaptive inference model and the newly established temperature difference and energy consumption inferential model are used to select the temperature difference and energy consumption inferential inference model with the smallest error, and load and encapsulate the temperature difference and energy consumption inferential model, that is, the execution pattern Program i.

The knowledge discovery and system control module 12 can first load the selected temperature difference and energy consumption inference model to the data exploration module (not shown), that is, execute the pattern program j; reuse the data exploration module撷Take the instant sensing data, that is, execute the program of the schema k; secondly, based on the instantaneous sensing data, the combination of multiple energy-consuming strategies, that is, the program of executing the schema, is deduced by the appropriate inference technique; The technology selects the most energy-saving optimization proposal from the combination of the plurality of energy-consuming strategies as a setting proposal, that is, a program m for executing the pattern; the setting of the air-conditioning system according to the energy-saving optimization setting recommendation, that is, the execution pattern The procedure n; then the continuity control of the air conditioning system, that is, the execution of the program o. In other words, the case can be based on artificial intelligence, based on artificial inference, appropriate regulation and other techniques, and save and transfer specific knowledge to further achieve self-adaptive effect. Furthermore, the optimization technique, also referred to as optimization technology, refers to an calculus process that finds the most suitable solution in an environment that conflicts with many constraints and conditions. For details, refer to the related art. The technical literature is not further described here.

The specific application content of the procedure a to the procedure p disclosed above may be considered as follows Sample content.

In the program a, the target building and the back-end data server or the data management platform to which the air-conditioning system belongs may be selected, and relevant parameters that can be accessed by the data source and the target data set to be accessed may be provided.

In the procedure b, the unit time of each sensing data may be set first, for example, in an hourly or per minute manner, and the average value of the data in the unit time is used for analysis, and the data is used for sampling. , will be grouped according to the unit time set by this; then the range of valid data to be analyzed can be set, each data is the result of the effective data sampling process, and is sorted by time, and the range setting represents a section The experience and results of air-conditioning system control within a specific time are also the space for systematic learning and adaptability. In other words, the subsequent processing will establish an appropriate inference model for the experience during this period and optimize it to obtain energy-saving control knowledge, and then adjust the current situation of the air-conditioning system based on this.

In program c, two methods, expert interviews and data analysis, can be used to determine the limits of regulation of the air conditioning system and the scope and conditions for defining valid data. For example, firstly from the discussion record of the person in charge of the air-conditioning system control or from the air-conditioning system specification and system operation manual, access the setting restrictions and scopes that should be understood when operating the air-conditioning system, as well as the common operating habits and internal conditions of the various situations. Operational regulations, especially in the immediate monitoring of target data. Then, through the data analysis, the conditions for obtaining the valid data are set as the setting, and the analysis is performed through the historical data of the air conditioning system for at least one year, and the effective range of the specific target data and the condition for filtering outliers are determined.

In the program d, the frequency and method of continuous learning and fitness adjustment can be determined, the learning schedule can be set according to the specified time frequency, and at the set time point, it is judged whether to re-learn the data range defined by the program b. Experience, and then learn the logic, plus the current data, through data exploration and knowledge mining technology to re-adjust the air conditioning system. In other words, the subsequent program can periodically perform the relevant steps in accordance with the schedule set here. Whether or not to immediately enter the learning program to re-establish a new adaptive inference model, it needs to be set here and used as the judgment condition of the program p.

In the program e, the connection with the target data source can be established through the settings made in the program a, and each target data can be processed. The value of each data is the average value per unit time set by program b. Then, according to the conditions set by the program c, the target data is filtered and filtered to obtain a range of valid data records defined by the procedure b, for subsequent correlation analysis processing.

In the program f, the adaptive inference model modeling method and the adaptive inference model data processing and loading may be included. For example, the inference model can be constructed by an algorithm, and the adaptive learning is performed through input and output data to train the logical relationship between the two sets of data, which is used to describe the experience and memorize; after the algorithm model is established, According to the learned experience, the input data can be inferred and a reasonable output result can be obtained; the composition of the adaptive inference model can be realized in a variety of ways, including regression analysis algorithm, neuro-like algorithm, fuzzy logic algorithm. And decision tree algorithms. Furthermore, valid data can be based on data or modulo before loading the adaptive inference model. The characteristics of the type algorithm, to standardize and standardize the data, and then provide the model algorithm for training and learning; the data standardization system provides the range conversion of the original data, making it suitable for being standardized and adaptable. The inference model algorithm accepts; data standardization can normalize the different data samples so that they can be analyzed and processed under the same benchmark; similarly, before performing the logical inference through the model after completing the training learning, This process is required; and after the data is processed, it is loaded into the empirical learning and logical inference module 11.

In the program g, the temperature difference adaptive inference model can be the first model established in the empirical learning and logic inference module 11 to appropriately learn the air conditioning setting and operation experience during the specified period to find the end pressure of the ice water. The relationship between the difference, the set frequency of the ice water pump, the set frequency of the cooling water pump, and the temperature difference between the cooling water and the water, and the temperature difference between the water and the water, and an effective and inferential model for its logic.

In the program h, the energy consumption appropriate inference model can be a model that should be established after the empirical learning and logic inference module 11 establishes the temperature difference inferential inference model to appropriately learn the air conditioning setting and operation within a specified period. Experience to find out the differential pressure at the end of the ice water, the set frequency of the ice water pump, the set frequency of the cooling water pump, the temperature difference between the water and the inlet water, and the inferred temperature difference between the water and the inlet water. Influencing relationships and establishing effective inferential models for their logic.

In the program i, the correlation analysis can be performed using the data in the range set by the program b to compare the new inferiority inference model of the program g and the program h, and the error root mean square of the previously established model (RMS, Root). Of Mean Square), and retain the model with smaller root mean square error, and discard the larger root mean square error.

In program j, the selected two adaptive inference models can be loaded into the data exploration module to prepare data for data exploration to explore the best settings or potential knowledge outside the experience.

In the program k, the instantaneous sensing data at the current time point can be taken, such as the end pressure difference of the ice water, the set frequency of the cooling water pump, the cooling water outlet temperature, or the cooling water inlet water temperature.

In program 1, the instantaneous sensing data of the current time point can be used, and the ice water pump frequency range setting made by the program c can be loaded, and then a possible input combination can be generated. Then, the model is inferred by the temperature difference, and the temperature difference of each combination is deduced according to the learned empirical logic. Then, the energy consumption inferential model is put forward, and each combination is deduced according to the previous experience within the defined range. The total energy consumption of the target. The numerical results obtained from the logical inference model can be inversely scaled and denormalized to restore the data. The scaling refers to the calculus process of scale, the rescaling refers to the process of mapping back to the real scale, and the standardization refers to the process of formulating standards and agreeing on them. For details, refer to related literature in the technical field, and no further details are provided herein. After the inference work is completed, the input and output strategy matrix is completed.

In the program m, it is possible to find the set frequency of the ice water pump that can contribute to the minimum sum of the target energy consumption among the various possible input and output strategies from the strategy matrix completed by the program 1. In the program n, the resulting new ice can be obtained. The pump sets the frequency and makes control settings adjustments to the air conditioning system to operate in accordance with the current energy saving conditions. In the program o, scheduling judgment and control can be performed. If the current time point meets the frequency set in the program d, the next program is entered. Otherwise, continue to stand by.

In the program p, it is possible to judge from the settings made by the program d whether or not to immediately enter the learning program for re-establishing a new adaptive inference model. If so, the experience learning and logic derivation module 11 and the knowledge discovery and system control module 12 are restarted, that is, the program e to the program o are executed. If not, the representative can continue to use the old adaptive inference model to brake the knowledge discovery and system control module 12, that is, execute the previous procedure j to the program o. Of course, the program p allows for the addition of other sub-judging programs, and if the new sub-judge program determines that the conditions are met, the re-modeling program or the adaptive inference model using the specific external import can be entered.

Figure 2 is a schematic diagram showing the continuous adaptive control achieved by the continuation of the case. In the figure, the time segments 20 and 21 represent the intelligent learning energy-saving control method for the first execution of the case, that is, the intelligent learning energy-saving control system 1 for starting the case for the first time; the time segments 22, 23, representing the first The second implementation of the smart learning energy-saving regulation method of this case, that is, the intelligent learning energy-saving control system 1 that started the case for the second time, and so on. Furthermore, the time segments 20, 22 may represent the continuous detection, learning and derivation provided by the techniques of the present application, while the time segments 21, 23 may represent the immediate control of the air conditioning system by the techniques of the present invention. It can be seen that the intelligent learning energy-saving control system and method of this case can be in the n+1th continuous detection and learning at the moment when the air conditioning system is immediately adjusted for the right time. In the process of learning and deriving.

In addition, the foregoing regulation of the air conditioning system may refer to regulation of the ice water host side of the air conditioning system, for example, adjusting the frequency of the ice water pump. Setting the system control data may include setting the recording schedule, the ice pump frequency, and the cooling pump frequency. Instant sensing data may include ice water inlet temperature, ice water outlet temperature, cooling water inlet temperature, cooling water outlet temperature, ice water flow rate, ice water end pressure difference, ice water pump power consumption, cooling Pump power consumption, and ice water host power consumption. The energy saving optimization setting suggestions may include recommendations corresponding to the power consumption of the ice water pump, the power consumption of the cooling water pump, and the power consumption of the ice water host.

Compared with the conventional technology, due to the intelligent learning energy-saving control system and method of the present case, it is possible to continuously utilize the instantaneously established temperature difference and energy consumption inference model to perform energy-saving optimization setting and immediate suitability control for the air-conditioning system.遂 求 求 求 求 求 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 冰 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。

After the actual experiment, compared with the same month of the previous two years, the technology of this case can effectively improve the energy saving, refrigeration capacity (RT) and efficiency ratio (EER) of the air conditioner, such as the following related comparison table. Illustrated:

The above embodiments are intended to illustrate the principles of the present invention and its effects, and are not intended to limit the invention. Any of the above-described embodiments may be modified by those skilled in the art without departing from the spirit and scope of the invention. Therefore, the scope of protection of the rights in this case should be listed in the scope of the patent application mentioned later.

1‧‧‧Intelligent Learning Energy Conservation Control System

10‧‧‧System Control Data Setting Module

11‧‧‧Experience Learning and Logic Derivation Module

12‧‧‧Knowledge Discovery and System Control Module

a to p‧‧‧ procedures

Claims (18)

An intelligent learning energy-saving regulation method includes: (1) setting system control data and extracting instant sensing data from the air conditioning system; and (2) inferring according to the system control data set and the instantaneous sensing data acquired; Model modeling technology for empirical learning and logic derivation, 俾 establish and encapsulate temperature difference and energy consumption inferential model; (3) use the temperature difference and energy consumption inferential model, and the instantaneous sensing data extracted from the air conditioning system, The data exploration technology infers the energy saving optimization setting proposal, and then continuously adjusts the air conditioning system through the knowledge discovery engine technology; and (4) the system regulation data according to the continuous adaptive regulation of the air conditioning system, and The instant sensing data captured by the air conditioning system determines whether the remodeling condition is met, and if so, proceeds to step (2), and if not, proceeds to step (3). For example, in the intelligent learning energy-saving regulation method described in claim 1, the step (1) further includes: setting the source of the instantaneous sensing data, capturing the target and method, setting the data timing analysis range, and setting the effective data sampling. Conditions, as well as setting up continuous learning and fitness adjustment procedures. For example, the intelligent learning energy-saving regulation method described in claim 2, wherein the step (2) further comprises the following steps: (2.1) extracting and filtering relevant modeling data from the source of the set instant sensing data; (2.2) loading the modeling data for standardization and standardization processing; (2.3) using the modeling data to establish a temperature difference appropriate inference model through the adaptive inference model modeling technique; (2.4) using the modeling data to infer the fitness inference Model modeling technology, and based on the inference results of the temperature difference inferential model, further establish the energy consumption inferential inference model; and (2.5) use the adaptive inference model comparison technique to compare the previously established temperature difference and energy consumption inferential model with the latest establishment The temperature difference and the energy consumption suitability inference model are used to select the temperature difference and the energy consumption suitability inference model with the smallest error, and load and package. For example, the intelligent learning energy-saving regulation method described in claim 3, wherein the step (2.3) further comprises the following steps: (2.3.1) loading the selected temperature difference and energy consumption inferential model; (2.3.2) Taking the instant sensing data; (2.3.3) deducing a plurality of energy consumption strategy combinations according to the appropriate sensing data by using the appropriate inference technique; (2.3.4) using the appropriate inference and optimization techniques by the plurality of consumption The most energy-saving optimization proposal in the strategy combination can be selected as the energy-saving optimization setting suggestion; (2.3.5) the air-conditioning system is set according to the energy-saving optimization setting suggestion; and (2.3.6) the air-conditioning system is continuous Appropriate regulation. For example, the intelligent learning energy-saving regulation method described in claim 1 of the patent scope, wherein the air conditioning system is continuously adapted and regulated Refers to the regulation of the ice water host side of the air conditioning system. For example, the intelligent learning energy-saving regulation method described in claim 5, wherein the regulation of the ice water host side of the air-conditioning system refers to controlling the frequency of the ice water pump. For example, the intelligent learning energy-saving regulation method described in claim 1 is characterized in that setting system control data includes setting a record time course, an ice pump frequency, and a cooling water pump frequency. For example, the intelligent learning energy-saving regulation method described in claim 1 includes extracting the instantaneous sensing data, including taking the ice water into the water temperature, the ice water outlet temperature, the cooling water entering the water temperature, the cooling water outlet temperature, and the ice. Water flow rate, ice water end pressure difference, ice water pump power consumption, cooling water pump power consumption, and ice water host power consumption. For example, the intelligent learning energy-saving regulation method described in claim 1 is characterized in that the energy-saving optimization setting recommendation includes a power consumption corresponding to the ice water pump, a cooling water pump power consumption, and an ice water host power consumption. An intelligent learning energy-saving control system, comprising: a system control data setting module, which is used for setting system control data and extracting instant sensing data for an air-conditioning system; an empirical learning and logic derivation module is based on the set system control data And the acquired real-time sensing data, empirical learning and logic derivation based on the inferential model modeling technique, and the establishment and encapsulation of the temperature difference and energy consumption inference model; and the knowledge discovery and system control module to utilize the Temperature difference Energy-consumption inference model, and the real-time sensing data extracted from the air-conditioning system, the data exploration technology is used to infer the energy-saving optimization setting suggestion in a data exploration module, and then the air-conditioning system is continued through the knowledge discovery engine technology. The adaptive control of sexuality; among them, the intelligent learning energy-saving regulation system is based on the system regulation data of the continuous adaptive regulation of the air-conditioning system, and the instantaneous sensing data extracted from the air-conditioning system to determine whether the re-modeling is met. The condition, if yes, activates the empirical learning and logic derivation module again, and if not, activates the knowledge discovery and system control module again. For example, the intelligent learning energy-saving control system described in claim 10, wherein the system control data setting module first sets the source of the instant sensing data, the target and method of capturing, and then sets the data timing analysis range; Set the effective data sampling conditions; then set the continuous learning and fitness adjustment procedures. For example, the intelligent learning energy-saving control system described in claim 11 wherein the empirical learning and logic derivation module firstly extracts and filters relevant modeling data from the source of the set instant sensing data; The modeling data is loaded for standardization and standardization processing; the model data is used to establish a temperature difference adaptive inference model through the adaptive inference model modeling technology; and the modeling data is passed through the fitness inference model modeling technique according to the modeling data. The inference result of the temperature difference inferential model further establishes the energy inducibility inference model; the second uses the adaptive inference model comparison technique to compare the previously established temperature difference and energy consumption inferential model. With the newly established temperature difference and energy consumption inferential model, the model with the least error temperature difference and energy consumption inference model is selected for loading and packaging. For example, the intelligent learning energy-saving control system described in claim 12, wherein the knowledge discovery and system control module first loads the selected temperature difference and energy consumption inference model to the data exploration module; The data exploration module captures the real-time sensing data; secondly, based on the instant sensing data, a plurality of energy-consuming strategy combinations are deduced by a suitable inference technique; and the plurality of energy-consuming strategies are combined by a suitable inference and optimization technique. The most energy-saving optimization proposal is selected as the energy-saving optimization setting suggestion; the air-conditioning system is set according to the energy-saving optimization setting suggestion; then the air-conditioning system is continuously adapted. For example, the intelligent learning energy-saving control system described in claim 10, wherein the continuous regulation of the air-conditioning system refers to the regulation of the ice water host side of the air-conditioning system. For example, the intelligent learning energy-saving control system described in claim 14 of the patent scope, wherein the regulation of the ice water host side of the air-conditioning system refers to regulating the frequency of the ice water pump. For example, the intelligent learning energy-saving control system described in claim 10, wherein setting system control data includes setting a record time course, an ice pump frequency, and a cooling water pump frequency. For example, the intelligent learning energy-saving control system described in claim 10, wherein extracting instant sensing data includes extracting ice water into water temperature, ice water outlet temperature, cooling water entering water temperature, cooling water effluent Temperature, ice water flow, ice water end pressure difference, ice pump power consumption, cooling pump power consumption, and ice water host power consumption. For example, the intelligent learning energy-saving control system described in claim 10, wherein the energy-saving optimization setting recommendation corresponds to the power consumption of the ice water pump, the power consumption of the cooling water pump, and the power consumption of the ice water host.