A data cleaning method for building cold station sub-metering data

Abstract

The application relates to a data cleaning method for cold station machine room sub-metering data, characterized in that the method comprises the following steps: step 1), obtaining building cold station related data; step 2), cleaning total energy consumption of cold machines, total energy consumption of chilled water pumps, total energy consumption of cooling water pumps and total energy consumption of cooling tower fans; step 3), cleaning energy consumption of single cold machine, energy consumption of single chilled water pump, energy consumption of single cooling water pump and energy consumption of single cooling tower fan; step 4), cleaning start-stop state of single cold machine, frequency of single chilled water pump, frequency of single cooling water pump and frequency of single cooling tower fan; step 5), cleaning dry pipe chilled water flow and dry pipe cooling water flow; and step 6), cleaning dry pipe chilled water supply water temperature, return water temperature, dry pipe cooling water supply water temperature, return water temperature and cold load. Compared with the prior art, the cleaned data can objectively reflect the physical law and control logic of the cold station system, and has the advantages of being convenient for relevant personnel to analyze energy-saving potential, simple and easy to use and the like.

Description

Technical Field

[0001] This invention relates to the field of data cleaning, and in particular to a data cleaning method for sub-item metering data of building cooling plants. Background Technology

[0002] Because the data quality varies significantly from different energy consumption monitoring platforms—with numerous missing values, zero values, outliers, or misaligned data labels and variables—the raw data struggles to objectively reflect the operational status of a building's HVAC system. This makes it difficult to analyze a building's energy-saving potential in a short period. Furthermore, the diverse data quality issues from different platforms result in inefficient data cleaning. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a simple and easy-to-use data cleaning method for sub-metering data of building cooling plants, thereby improving data cleaning efficiency and facilitating energy-saving potential analysis.

[0004] The objective of this invention can be achieved through the following technical solutions:

[0005] A data cleaning method for sub-item metering data of building cooling plants includes the following steps:

[0006] Step 1) Obtain relevant data for the building chiller plant. All data is time-series data, including the total energy consumption of the chiller group. Total energy consumption of chilled water pumps Total energy consumption of cooling water pump Total energy consumption of cooling tower fans Energy consumption per chiller, energy consumption per chilled water pump, energy consumption per cooling water pump, energy consumption per cooling tower fan, start / stop status of per chiller, frequency of per chilled water pump, frequency of per cooling water pump, frequency of per cooling tower fan, cooling load. chilled water flow rate in main pipe Main pipe cooling water flow rate Chilled water supply temperature of main pipe Chilled water return temperature in main pipe Main cooling water supply temperature Main pipe cooling water return temperature , among which, the Taiwan refrigeration unit energy consumption is , No. The energy consumption of the chilled water pump is , No. The energy consumption of the cooling water pump is , No. The energy consumption of the cooling tower fan is , No. Taiwan Refrigerator Start-up and Stop Status , No. The frequency of the chilled water pump is , No. The frequency of the cooling water pump is , No. The frequency of the cooling tower fan is The first The start / stop status of the chiller is a dimensionless categorical variable, where 0 represents the chiller being off and 1 represents the chiller being running.

[0007] Step 2) Clean the total energy consumption of the chiller group Total energy consumption of chilled water pumps Total energy consumption of cooling water pump Total energy consumption of cooling tower fans The data time series was cleaned by removing missing values, zero values, and outliers. The cleaned data consisted of all variables corresponding to the time of the removed value. The cleaned data represented the total energy consumption of the chiller group. Total energy consumption of chilled water pumps Total energy consumption of cooling water pump Total energy consumption of cooling tower fans ;

[0008] Step 3) Clean the energy consumption data for a single chiller, a single chilled water pump, a single cooling water pump, and a single cooling tower fan. Remove missing values, zero values, and outliers from the data time series. The data to be removed consists of all variables corresponding to the time at which the removed value occurs. Outliers are determined according to Kirchhoff's laws. After cleaning, the [number]th [item]... Taiwan refrigeration unit energy consumption is , No. The energy consumption of the chilled water pump is , No. The energy consumption of the cooling water pump is , No. The energy consumption of the cooling tower fan is ;

[0009] Step 4) Clean the start / stop status of a single chiller, the frequency of a single chilled water pump, the frequency of a single cooling water pump, and the frequency of a single cooling tower fan. First, remove missing values ​​and zero values ​​from the data time series. The data to be removed consists of all variables corresponding to the time of the removed value. Then, process the data for the next step... Taiwan Refrigerator Start-up and Stop Status The consistency principle is used to correct outliers, for the first... Taiwan chilled water pump frequency , No. Taiwan cooling water pump frequency , No. Cooling tower fan frequency The correspondence between data labels and data is corrected using the similarity criterion law. After cleaning, the first... Taiwan chiller start / stop status No. The frequency of the chilled water pump is , No. The frequency of the cooling water pump is , No. The frequency of the cooling tower fan is ;

[0010] Step 5) Clean the chilled water flow rate of the main pipe and main cooling water flow rate Missing values, zero values, and outliers are removed from the data time series. The removed missing and zero values ​​are the data of all variables corresponding to the time at which the removed value occurs. Outliers are determined by comparing the dry pipe chilled water flow rate with the dry pipe cooling water flow rate calculated using a similarity criterion. After cleaning, the dry pipe chilled water flow rate is... The main cooling water flow rate is ;

[0011] Step 6) Clean the chilled water supply temperature of the main pipe Return water temperature Main cooling water supply temperature Return water temperature Cooling load First, missing values ​​and zero values ​​are removed from the data time series. The removed data consists of all variables corresponding to the time when the removed value is located. Then, outliers are identified based on expert experience, and the law of conservation of energy is used to remove outliers and correct the data.

[0012] The timestamps of all variables in the data correspond one-to-one, with data granularity at every 15 minutes or hour, including the total energy consumption of the chiller cluster. Total energy consumption of chilled water pumps Total energy consumption of cooling water pump Total energy consumption of cooling tower fans , No. Taiwan Refrigeration Unit Energy Consumption , No. Energy consumption of chilled water pump , No. Energy consumption of cooling water pump , No. Cooling tower fan energy consumption , No. Taiwan Refrigerator Start-up and Stop Status , No. Taiwan chilled water pump frequency , No. Cooling water pump frequency , No. Cooling tower fan frequency Cooling load chilled water flow rate in main pipe Main pipe cooling water flow rate Chilled water supply temperature of main pipe Chilled water return temperature in main pipe Main cooling water supply temperature Main pipe cooling water return temperature .

[0013] The outlier detection method in step 2) is as follows:

[0014] like Then it is considered that the first Moment abnormal;

[0015] like Then it is considered that the first Moment abnormal;

[0016] like Then it is considered that the first Moment abnormal;

[0017] like Then it is considered that the first Moment abnormal;

[0018] in, For the first The nominal power of the chiller For the first Nominal power of chilled water pump For the first The nominal power of the cooling pump For the first The nominal power of the cooling tower fan.

[0019] The outlier detection method in step 3) is as follows:

[0020] like Then it is considered that the first Moment abnormal;

[0021] like Then it is considered that the first Moment abnormal;

[0022] like Then it is considered that the first Moment abnormal;

[0023] like Then it is considered that the first Moment abnormal.

[0024] The first step in step 4) Taiwan Refrigerator Start-up and Stop Status The outlier correction method is as follows:

[0025] when , Then it is considered that the first time Error, corrected to ;

[0026] when , Then it is considered that the first time Error, corrected to ;

[0027] in, For the first Nominal power of the chiller.

[0028] The first step in step 4) Taiwan chilled water pump frequency , No. Taiwan cooling water pump frequency , No. Cooling tower fan frequency The outlier correction method is as follows:

[0029] Will any and By combining them, one or more sets of similarity criteria can be found. and Fit a cubic curve The applicable range of this curve Based on expert experience, the group with the most effectively fitted data points will be selected. and Treating these as the same frequency and energy consumption data for a single refrigeration pump, and correcting its data label, will represent the first... The subscript of the equipment Corrected to the same number;

[0030] Will any and By combining them, according to the similarity criterion, there will always be one or more combinations. and Fit a cubic curve The applicable range of this curve Based on expert experience, the group with the most effectively fitted data points will be selected. and Treating these as the same cooling pump's frequency and energy consumption data, and correcting its data label, will represent the first... The subscript of the equipment Corrected to the same number;

[0031] Will any and By combining them, we can find more than one set of similarity criteria. and Fit a cubic curve The applicable range of this curve Based on expert experience, the group with the most effectively fitted data points will be selected. and Treating these as the same frequency and energy consumption data of the cooling tower water pump, and correcting its data label, will represent the first... The subscript of the equipment Corrected to the same number;

[0032] in, For the first Nominal power of chilled water pump For the first The nominal power of the cooling pump For the first The nominal power of the cooling tower fan.

[0033] The outlier detection method in step 5) is as follows:

[0034] like Then it is considered that the first Moment abnormal;

[0035] like Then it is considered that the first Moment abnormal;

[0036] No. Moment and Anomalies cannot coexist simultaneously;

[0037] in, For the first The nominal flow rate of the chilled water pump For the first The nominal flow rate of the cooling water pump.

[0038] The outlier detection method in step 6) is as follows:

[0039] like Then it is considered that the first Moment and abnormal;

[0040] like Then it is considered that the first Moment and abnormal;

[0041] If the first Moment , , , , and If two or more anomalies exist, they are considered as all anomalies.

[0042] The outlier removal and data correction method in step 6) is as follows:

[0043] When the There are no outlier data at any time, if Then it is considered as the first All time-based data that is abnormal is removed, and the remaining time-based data is corrected. ;

[0044] When the If only one data point is abnormal at any given time, then according to the law of conservation of energy... Correct the abnormal data;

[0045] in, The density of water, This is the specific heat capacity of water at constant pressure.

[0046] The total number of chillers, chilled water pumps, cooling water pumps, and cooling tower fans are as follows: ,in, , , , , This refers to the number of standby water pumps.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] (1) The data after cleaning can reflect the general laws of actual air conditioning system operation: the frequency and power consumption of the same water pump in the operating state meet the similarity criterion law; there is a one-to-one start-stop relationship between the chiller and the water pump under actual manual no-load conditions; the energy transfer between the refrigeration side and the cooling side follows the law of conservation of energy.

[0049] (2) The data cleaning method of the present invention is simple and easy to learn, and easy to promote.

[0050] (3) The data cleaning method of the present invention has high cleaning efficiency, which is beneficial to analyze the energy-saving potential of buildings in a short time. Attached Figure Description

[0051] Figure 1 This is a flowchart of the present invention;

[0052] Figure 2 The data cleaning comparison charts are for the total energy consumption of the chiller group and the total energy consumption of the chilled water pump group. Among them, (a) is the original total energy consumption histogram of the chiller group, (b) is the total energy consumption histogram of the chiller group after cleaning according to step 2), (c) is the original total energy consumption histogram of the chilled water pump group, and (d) is the total energy consumption histogram of the chilled water pump group after cleaning according to step 2).

[0053] Figure 3 The data cleaning comparison charts for the energy consumption of a single chilled water pump are shown below. (a) shows the relationship between the original sum of energy consumption of a single chilled water pump and the total energy consumption of the chilled water pump group after cleaning. (b) shows the relationship between the sum of energy consumption of a single chilled water pump after removing zero and null values ​​and the total energy consumption of the chilled water pump group after cleaning. (c) shows the relationship between the sum of energy consumption of a single chilled water pump after removing outliers according to Kirchhoff's laws and the total energy consumption of the chilled water pump group after cleaning.

[0054] Figure 4 The relationship between frequency and energy consumption of a single chilled water pump after removing zero and null values ​​is shown in the figure. The subgraph enclosed by the dashed line represents the frequency and energy consumption groups that satisfy the similarity criterion.

[0055] Figure 5 To correct the relationship between the frequency and energy consumption of a single chilled water pump after data labeling according to the similarity criterion, the subgraph enclosed by the dashed line represents the frequency and energy consumption group that meets the similarity criterion;

[0056] Figure 6 The graphs show the comparison between the monitored and calculated values ​​of the chilled water flow rate and the cooling water flow rate of the main pipe. (a) is a scatter plot of the monitored chilled water flow rate and the chilled water flow rate calculated using the similarity criterion, and (b) is a scatter plot of the monitored cooling water flow rate and the cooling water flow rate calculated using the similarity criterion.

[0057] Figure 7 The data cleaning results for main pipe temperature, main pipe flow rate, and cooling load are as follows: (a) is a scatter plot of monitored main pipe chilled water flow rate and cooling water flow rate after removing data with abnormal chilled water flow rate and cooling water flow rate; (b) is a scatter plot of corrected main pipe chilled water flow rate and cooling water flow rate; (c) is the relationship between main pipe cooling water temperature difference and chilled water temperature difference after removing null values, zero values, and outliers; and (d) is the relationship between measured main pipe chilled water flow rate after removing outliers and corrected main pipe cooling water flow rate. Detailed Implementation

[0058] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0059] A data cleaning method for sub-item metering data of building cooling plants includes the following steps:

[0060] Step 1) Obtain relevant data for the building chiller plant. All data is time-series data, including the total energy consumption of the chiller group. Total energy consumption of chilled water pumps Total energy consumption of cooling water pump Total energy consumption of cooling tower fans Energy consumption per chiller, energy consumption per chilled water pump, energy consumption per cooling water pump, energy consumption per cooling tower fan, start / stop status of per chiller, frequency of per chilled water pump, frequency of per cooling water pump, frequency of per cooling tower fan, cooling load. chilled water flow rate in main pipe Main pipe cooling water flow rate Chilled water supply temperature of main pipe Chilled water return temperature in main pipe Main cooling water supply temperature Main pipe cooling water return temperature , among which, the Taiwan refrigeration unit energy consumption is , No. The energy consumption of the chilled water pump is , No. The energy consumption of the cooling water pump is , No. The energy consumption of the cooling tower fan is , No. Taiwan Refrigerator Start-up and Stop Status , No. The frequency of the chilled water pump is , No. The frequency of the cooling water pump is , No. The frequency of the cooling tower fan is The first The start / stop status of the chiller is a dimensionless categorical variable, where 0 represents the chiller being off and 1 represents the chiller being running.

[0061] The timestamps of all variables in the data correspond one-to-one, with data granularity at every 15 minutes or hour, including the total energy consumption of the chiller cluster. Total energy consumption of chilled water pumps Total energy consumption of cooling water pump Total energy consumption of cooling tower fans , No. Taiwan Refrigeration Unit Energy Consumption , No. Energy consumption of chilled water pump , No. Energy consumption of cooling water pump , No. Cooling tower fan energy consumption , No. Taiwan Refrigerator Start-up and Stop Status , No. Taiwan chilled water pump frequency , No. Taiwan cooling water pump frequency , No. Cooling tower fan frequency Cooling load chilled water flow rate in main pipe Main pipe cooling water flow rate Chilled water supply temperature of main pipe Chilled water return temperature in main pipe Main cooling water supply temperature Main pipe cooling water return temperature .

[0062] The basic equipment information of the building in this embodiment is shown in the table below.

[0063]

[0064] Step 2) Clean the total energy consumption of the chiller group Total energy consumption of chilled water pumps Total energy consumption of cooling water pump Total energy consumption of cooling tower fans The data time series was cleaned by removing missing values, zero values, and outliers. The cleaned data consisted of all variables corresponding to the time of the removed value. The cleaned data represented the total energy consumption of the chiller group. Total energy consumption of chilled water pumps Total energy consumption of cooling water pump Total energy consumption of cooling tower fans ;

[0065] The outlier detection method in step 2) is as follows:

[0066] like Then it is considered that the first Moment abnormal;

[0067] like Then it is considered that the first Moment abnormal;

[0068] like Then it is considered that the first Moment abnormal;

[0069] like Then it is considered that the first Moment abnormal;

[0070] in, For the first The nominal power of the chiller For the first Nominal power of chilled water pump For the first The nominal power of the cooling pump For the first The nominal power of the cooling tower fan.

[0071] Step 2) Data cleaning results are as follows Figure 2 As shown.

[0072] Step 3) Clean the energy consumption data for a single chiller, a single chilled water pump, a single cooling water pump, and a single cooling tower fan. Remove missing values, zero values, and outliers from the data time series. The data to be removed consists of all variables corresponding to the time at which the removed value occurs. Outliers are determined according to Kirchhoff's laws. After cleaning, the [number]th [item]... Taiwan refrigeration unit energy consumption is , No. The energy consumption of the chilled water pump is , No. The energy consumption of the cooling water pump is , No. The energy consumption of the cooling tower fan is ;

[0073] The outlier detection method in step 3) is as follows:

[0074] like Then it is considered that the first Moment abnormal;

[0075] like Then it is considered that the first Moment abnormal;

[0076] like Then it is considered that the first Moment abnormal;

[0077] like Then it is considered that the first Moment abnormal.

[0078] Step 3) Data cleaning results are as follows Figure 3 As shown.

[0079] Step 4) Clean the start / stop status of a single chiller, the frequency of a single chilled water pump, the frequency of a single cooling water pump, and the frequency of a single cooling tower fan. First, remove missing values ​​and zero values ​​from the data time series. The data to be removed consists of all variables corresponding to the time of the removed value. Then, process the data for the next step... Taiwan Refrigerator Start-up and Stop Status Outliers are corrected using the consistency principle. Taiwan chilled water pump frequency , No. Cooling water pump frequency , No. Cooling tower fan frequency The correspondence between data labels and data is corrected using the similarity criterion law. After cleaning, the first... Taiwan chiller start / stop status No. The frequency of the chilled water pump is , No. The frequency of the cooling water pump is , No. The frequency of the cooling tower fan is ;

[0080] The first step in step 4) Taiwan Refrigerator Start-up and Stop Status The outlier correction method is as follows:

[0081] when , Then it is considered that the first time Error, corrected to ;

[0082] when , Then it is considered that the first time Error, corrected to ;

[0083] in, For the first Nominal power of the chiller.

[0084] The first step in step 4) Taiwan chilled water pump frequency , No. Cooling water pump frequency , No. Cooling tower fan frequency The outlier correction method is as follows:

[0085] Will any and By combining them, one or more sets of similarity criteria can be found. and Fit a cubic curve The applicable range of this curve Based on expert experience, the group with the most effectively fitted data points will be selected. and Treating these as the same frequency and energy consumption data for a single refrigeration pump, and correcting its data label, will represent the first... The subscript of the equipment Corrected to the same number;

[0086] Will any and By combining them, according to the similarity criterion, there will always be one or more combinations. and Fit a cubic curve The applicable range of this curve Based on expert experience, the group with the most effectively fitted data points will be selected. and Treating these as the same cooling pump's frequency and energy consumption data, and correcting its data label, will represent the first... The subscript of the equipment Corrected to the same number;

[0087] Will any and By combining them, we can find more than one set of similarity criteria. and Fit a cubic curve The applicable range of this curve Based on expert experience, the group with the most effectively fitted data points will be selected. and Treating these as the same frequency and energy consumption data of the cooling tower water pump, and correcting its data label, will represent the first... The subscript of the equipment Corrected to the same number;

[0088] in, For the first Nominal power of chilled water pump For the first The nominal power of the cooling pump For the first The nominal power of the cooling tower fan.

[0089] Step 4) Data cleaning results are as follows Figure 4 and Figure 5 As shown.

[0090] Step 5) Clean the chilled water flow rate of the main pipe and main cooling water flow rate Missing values, zero values, and outliers are removed from the data time series. The removed missing and zero values ​​are the data of all variables corresponding to the time at which the removed value occurs. Outliers are determined by comparing the dry pipe chilled water flow rate with the dry pipe cooling water flow rate calculated using a similarity criterion. After cleaning, the dry pipe chilled water flow rate is... The main cooling water flow rate is ;

[0091] The outlier detection method in step 5) is as follows:

[0092] like Then it is considered that the first Moment abnormal;

[0093] like Then it is considered that the first Moment abnormal;

[0094] No. Moment and Anomalies cannot coexist simultaneously;

[0095] in, For the first The nominal flow rate of the chilled water pump For the first The nominal flow rate of the cooling water pump.

[0096] Step 5) Data cleaning process as follows Figure 6 As shown, Figure 6 (a) Shows abnormal chilled water flow rates in the main pipe after November 2020, excluding all abnormal data after November; Figure 6 (b) If all the cooling water flow rates in the main pipe are abnormal, remove all data.

[0097] Step 6) Clean the chilled water supply temperature of the main pipe Chilled water return temperature in main pipe Main cooling water supply temperature Main pipe cooling water return temperature Cooling load First, missing values ​​and zero values ​​are removed from the data time series. The removed data consists of all variables corresponding to the time when the removed value is located. Then, outliers are identified based on expert experience, and the law of conservation of energy is used to remove outliers and correct the data.

[0098] The outlier detection method in step 6) is as follows:

[0099] like Then it is considered that the first Moment and abnormal;

[0100] like Then it is considered that the first Moment and abnormal;

[0101] If the first Moment , , , , and If two or more anomalies exist, they are considered as all anomalies.

[0102] The outlier removal and data correction method in step 6) is as follows:

[0103] When the There are no outlier data at any time, if Then it is considered as the first All time-based data that is abnormal is removed, and the remaining time-based data is corrected. ;

[0104] When the If only one data point is abnormal at any given time, then according to the law of conservation of energy... Correct the abnormal data;

[0105] in, The density of water, This is the specific heat capacity of water at constant pressure.

[0106] Step 6) Data cleaning results are as follows Figure 7 As shown,

[0107] The total number of chillers, chilled water pumps, cooling water pumps, and cooling tower fans are as follows: ,in, , , , , This refers to the number of standby water pumps.

Claims

1. A data cleaning method for building cold station sub-metering data, characterized in that, The method comprises the following steps: Step 1) obtaining building cold station related data, the data are all time series data, including total energy consumption of cold machine group total energy consumption of refrigerated water pump total energy consumption of cooling water pump total energy consumption of cooling tower fan energy consumption of single cold machine, energy consumption of single refrigerated water pump, energy consumption of single cooling water pump, energy consumption of single cooling tower fan, start-stop state of single cold machine, frequency of single refrigerated water pump, frequency of single cooling water pump, frequency of single cooling tower fan, cooling load dry pipe refrigerated water flow dry pipe cooling water flow dry pipe refrigerated water supply temperature dry pipe refrigerated water return temperature dry pipe cooling water supply temperature dry pipe cooling water return temperature , wherein the energy consumption of the first cold machine is , the energy consumption of the first refrigerated water pump is , the energy consumption of the first cooling water pump is , the energy consumption of the first cooling tower fan is , the start-stop state of the first cold machine is , the frequency of the first refrigerated water pump is , the frequency of the first cooling water pump is , the frequency of the first cooling tower fan is , and the start-stop state of the first cold machine is a dimensionless classification variable, wherein 0 is cold machine off and 1 is cold machine running. Step 2) total energy consumption of the chilled water pump total energy consumption of the chilled water pump total energy consumption of the chilled water pump total energy consumption of the chilled water pump , the missing values, zero values and abnormal values in the data time series are removed, the data corresponding to all variables at the time of the removed values is removed, and the cleaned data is the total energy consumption of the chilled water pump total energy consumption of the chilled water pump total energy consumption of the chilled water pump total energy consumption of the chilled water pump ; Step 3) washing the energy consumption of a single chiller, the energy consumption of a single chilled water pump, the energy consumption of a single cooling water pump, the energy consumption of a single cooling tower fan, removing missing values, zero values and abnormal values in the data time series, removing data corresponding to all variables at the time of the removed values, the abnormal values are judged according to Kirchhoff's law, and the data after washing is The energy consumption of a single chiller is , the energy consumption of a single chilled water pump is , the energy consumption of a single cooling water pump is , and the energy consumption of a single cooling tower fan is , and the energy consumption of a single cooling tower fan is , and the energy consumption of a single cooling tower fan is , and the energy consumption of a single cooling tower fan is ; Step 4) washing single chiller start-stop state, single chilled water pump frequency, single cooling water pump frequency, single cooling tower fan frequency, first eliminating the missing value, zero value in the data time series, eliminating data is the elimination value corresponding to all variables at the moment, then the first Chiller start-stop state The abnormal value is corrected by adopting the consistency principle, and the first Chilled water pump frequency , the first Cooling water pump frequency , the first Cooling tower fan frequency The data label and the corresponding relationship of the data are corrected by adopting the similarity criterion law, and the first Chiller start-stop state is The first Chilled water pump frequency is , the first Cooling water pump frequency is , the first Cooling tower fan frequency is ; Step 5) cleaning the dry pipe chilled water flow and the dry pipe cooling water flow , eliminating missing values, zero values and abnormal values in the data time series, wherein the eliminated missing values and zero values are the data of all variables corresponding to the time of the eliminated values, and the abnormal values are determined according to the comparison result of the dry pipe chilled water flow comparison value and the dry pipe cooling water flow comparison value calculated by using the similarity criterion, the dry pipe chilled water flow after cleaning is , and the dry pipe cooling water flow is ; Step 6) Clean the dry pipe chilled water supply temperature , return water temperature , dry pipe cooling water supply temperature , return water temperature , cooling load , first eliminate the missing value, zero value in the data time series, wherein the data is the data of all variables corresponding to the elimination value at the elimination time, and then judge the abnormal value according to the expert experience, eliminate the abnormal value by using the law of conservation of energy and correct the data; The step 4) in the first Chiller water pump frequency , the first Chiller water pump frequency , the first Chiller tower fan frequency The abnormal value correction method is: Any of and are combined, find one or more sets of and , fit a cubic curve whose range of applicability is the largest number of data points and fit by that set, consider that set to be the frequency and power consumption data for the same chiller pump, correct the data labels to represent the first unit of equipment, correct the subscript to the same number; Any of and are combined, according to the similarity criterion law, one or more groups of and are fitted with cubic curves , the applicable range of the curve , according to the expert experience, the group with the most effective fitting data points and , are considered as the same cooling pump frequency and energy consumption data, correct their data labels, the subscripts representing the device are corrected to the same number ; Any of and are combined, find a group of more than one and , fit a cubic curve , the scope of the curve , according to the expert experience of the most effective fitting data points and , it is regarded as the same cooling tower pump frequency and energy consumption data, correct its data label, will represent the first device subscript corrected to the same number; wherein, is the number of cooling towers, is the nominal power of the cooling tower fan, is the number of chillers, is the nominal power of the chiller, is the number of cooling pumps, is the nominal power of the cooling pump.

2. The data cleaning method for building cooling sub-metering data according to claim 1, wherein, The timestamps of the data all variables correspond one by one, the data granularity is every 15 minutes or every hour, wherein the total energy consumption of the cold machine group , the total energy consumption of the refrigeration water pump , the total energy consumption of the cooling water pump , the total energy consumption of the cooling tower fan , the energy consumption of the first cold machine , the energy consumption of the first refrigeration water pump , the energy consumption of the first cooling water pump , the energy consumption of the first cooling tower fan , the start-stop state of the first cold machine , the frequency of the first refrigeration water pump , the frequency of the first cooling water pump , the frequency of the first cooling tower fan , the cooling load , the dry pipe refrigeration water flow , the dry pipe cooling water flow , the dry pipe refrigeration water supply water temperature , the dry pipe refrigeration water return water temperature , the dry pipe cooling water supply water temperature , the dry pipe cooling water return water temperature .

3. The data cleaning method for building cooling sub-metering data according to claim 1, wherein, The abnormal value judgment method in the step 2) is: If , the first time point is considered abnormal; If , the first time of the day is considered abnormal; If , the first time point is considered abnormal; If , the first abnormality is considered to have occurred at the time ; wherein, is the number of the first nominal power of the air cooling machine, is the number of the first nominal power of the chilled water pump, is the number of the first nominal power of the cooling pump, is the number of the first nominal power of the cooling tower fan.

4. The data cleaning method for building cooling sub-metering data according to claim 1, wherein, The abnormal value judgment method in the step 3) is: If , the first time point is considered abnormal; If , the first time point is considered abnormal; If , the first time is considered anomaly; If , the first time is considered anomaly.

5. The data cleaning method for building cooling sub-metering data according to claim 1, wherein, The first Start-stop state of the accelerator The abnormal value correction method is: When , , the first moment is considered abnormal and is corrected to ; When , , the first moment is considered abnormal and is corrected to ; wherein, is the first the nominal power of the accelerator.

6. The data cleaning method for building cooling sub-metering data according to claim 1, wherein, The abnormal value judgment method in the step 5) is: If , the first time is considered abnormal; If , the first time point is considered abnormal; No. Moment and Anomalies cannot coexist simultaneously; wherein, is the nominal flow of the i-th cooling water pump. is the nominal flow of the i-th cooling water pump.

7. The data cleaning method for building cooling sub-metering data according to claim 1, wherein, The abnormal value judgment method in the step 6) is: If , the first time point is considered and an anomaly; If , the first time point is considered and an anomaly; If there are two or more abnormalities at the same time , , , , , and , all of them are regarded as abnormalities.

8. The data cleaning method for building cold station sub-metering data according to claim 7, wherein, The abnormal value judgment method in the step 6) is: The abnormal value elimination and data correction method in the step 6) is: When the 1st moment has no abnormal value data, if , it is considered that the 1st moment data is abnormal in whole and is eliminated, and the of the rest moments is revised ; When the first moment only one data exists exception, then according to the law of conservation of energy The abnormal data is corrected; wherein is the density of water, is the specific heat capacity of water at constant pressure.

9. The data cleaning method for building cooling sub-metering data according to claim 1, wherein, The total number of the cold water pumps, the chilled water pumps, the cooling water pumps and the cooling tower fan stations is respectively wherein, , , , , is the number of standby water pumps.

Images