Sensor data preprocessing method for efficient air-conditioning refrigerating machine room

Abstract

The invention discloses a sensor data preprocessing method for a high-efficiency air-conditioning refrigerating machine room. The sensor data preprocessing method comprises the following steps: acquiring sensor data at different detection positions; determining a type of the sensor data based on a data attribute of the sensor data; different abnormal value detection methods and / or missing value filling methods are / is adopted for preprocessing different types of sensor data, and preprocessed sensor data are obtained. According to the sensor data preprocessing method, the problem that the sensor data is abnormal or missing in the energy management process of the efficient air-conditioning refrigerating machine room can be solved, a high-quality data base is provided for efficient operation management and energy efficiency analysis of the efficient air-conditioning refrigerating machine room, and the stability of an optimization control algorithm is enhanced; and the air-conditioning energy efficiency of the efficient air-conditioning refrigerating room is improved.

Description

Technical Field

[0001] This application relates to the technical field of high-efficiency air-conditioning refrigeration machine rooms, and particularly to a method for preprocessing sensor data for a high-efficiency air-conditioning refrigeration machine room. Background Art

[0002] Higher requirements are put forward for the optimized control and energy management of high-efficiency air-conditioning refrigeration machine rooms. The implementation of relevant technical solutions for the optimized control and energy management of high-efficiency air-conditioning refrigeration machine rooms depends on the accumulation of sensor data collected by sensors.

[0003] Currently, due to problems such as hardware aging and damage, communication link signal interference, or acquisition-end program errors, the problems of abnormality or missing of sensor data collected by sensors are widespread. This will lead to unreasonable technical solutions for the optimized control and energy management of high-efficiency air-conditioning refrigeration machine rooms, resulting in low air-conditioning energy efficiency of high-efficiency air-conditioning refrigeration machine rooms. Summary of the Invention

[0004] In view of this, this application aims to solve at least one of the problems in the related art to some extent. For this purpose, the purpose of this application is to provide a method for preprocessing sensor data for a high-efficiency air-conditioning refrigeration machine room.

[0005] This application provides a method for preprocessing sensor data for a high-efficiency air-conditioning refrigeration machine room. The preprocessing method includes: obtaining sensor data at different detection positions; determining the type of the sensor data based on the data attributes of the sensor data; and performing preprocessing on different types of sensor data using different outlier detection methods and / or missing value filling methods to obtain preprocessed sensor data.

[0006] In some embodiments, different types of sensors are installed at different detection positions in the high-efficiency air-conditioning refrigeration machine room. The obtaining of sensor data at different detection positions includes: obtaining data collected by different types of sensors installed at different detection positions in the high-efficiency air-conditioning refrigeration machine room to obtain the sensor data.

[0007] In some embodiments, the types of the sensor data include inertial data, non-inertial data, and enumerated data. The inertial data includes: data whose values do not change suddenly due to system inertia after a control instruction is issued; the non-inertial data includes: data whose values respond immediately to changes after a control instruction is issued; the enumerated data includes: data whose data attributes reflect physical states.

[0008] In some embodiments, different outlier detection methods and / or missing value filling methods are used for preprocessing different types of sensor data, and the preprocessed sensor data obtained includes: when the sensor data is determined to be the inertial data, the sliding window method is used to detect outliers in the inertial data; if there are outliers in the inertial data, the outliers are deleted and marked as null values.

[0009] In some embodiments, when the sensor data is determined to be the inertial data, using the sliding window method to detect outliers in the inertial data includes: dividing the inertial data into a preset number of data sequences according to a time window of a preset length; calculating the first quartile, the third quartile, and the interquartile range of each data sequence; calculating the lower limit value in the inertial data according to the first quartile and the interquartile range; calculating the upper limit value in the inertial data according to the third quartile and the interquartile range; determining the normal value range of the inertial data according to the lower limit value and the upper limit value; calculating a first difference according to the maximum value and the minimum value of the data sequence; calculating a second difference according to the upper limit value and the lower limit value, and if the inertial data exceeds the normal value range and both the first difference and the second difference are greater than a preset isolation limit value, determining the inertial data as the outlier.

[0010] In some embodiments, different outlier detection methods and / or missing value filling methods are used for preprocessing different types of sensor data, and the preprocessed sensor data obtained includes: when the sensor data is determined to be the non-inertial data, if the non-inertial data exceeds a preset value range, determining the non-inertial data as an outlier, and deleting the outlier and marking it as a null value.

[0011] In some embodiments, different outlier detection methods and / or missing value filling methods are used for preprocessing different types of sensor data, and the preprocessed sensor data obtained includes: when the sensor data is determined to be the inertial data or the non-inertial data, identifying whether there are missing values in the sensor data; when the sensor data is determined to be the enumerated data, identifying whether there are missing values in the sensor data; if missing values are identified in the sensor data, filling the missing values based on preset physical rules; if the missing values cannot be filled based on the preset physical rules, no processing is performed on the missing values or they are marked as null values.

[0012] In some embodiments, the enumerated class data includes the operating states of the operating devices in the high-efficiency air-conditioning refrigeration machine room. The operating states of the devices include the operating state and the stopped state. If a missing value is identified in the sensor data, filling the missing value based on a preset physical rule includes: comparing the load rate and power of the operating device with a preset load range and a preset power range to determine whether the operating state of the operating device is the operating state or the stopped state; filling the missing value according to the determination result of the operating state of the device.

[0013] The present application also provides a preprocessing device for sensor data for a high-efficiency air-conditioning refrigeration machine room. The preprocessing device for sensor data includes: an acquisition module, a type determination module, and a preprocessing module. The acquisition module is configured to acquire sensor data at different detection positions; the type determination module is configured to determine the type of the sensor data based on the data attributes of the sensor data; the preprocessing module is configured to perform preprocessing on different types of sensor data using different outlier detection methods and / or missing value filling methods to obtain preprocessed sensor data.

[0014] The present application also provides a non-volatile computer-readable storage medium containing a computer program. When the computer program is executed by one or more processors, the preprocessing method for sensor data in any of the above embodiments is implemented.

[0015] The preprocessing method for sensor data of the present application can solve the problems of abnormal or missing sensor data in the energy management process of a high-efficiency air-conditioning refrigeration machine room, provide a high-quality data foundation for the efficient operation management and energy efficiency analysis of the high-efficiency air-conditioning refrigeration machine room, enhance the stability of the optimization control algorithm, and improve the air-conditioning energy efficiency of the high-efficiency air-conditioning refrigeration machine room.

[0016] Additional aspects and advantages of the present application will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present application. Description of the Drawings

[0017] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of the embodiments in conjunction with the drawings, wherein:

[0018] Figure 1 is a flowchart of the preprocessing method for sensor data in some embodiments of the present application;

[0019] Figure 2 is a structural diagram of the preprocessing device for sensor data in some embodiments of the present application;

[0020] Figure 3It is a schematic flow diagram of a method for preprocessing sensor data according to some embodiments of the present application;

[0021] Figure 4 It is a schematic flow diagram of a method for preprocessing sensor data according to some embodiments of the present application;

[0022] Figure 5 It is a schematic flow diagram of a method for preprocessing sensor data according to some embodiments of the present application;

[0023] Figure 6 It is a schematic flow diagram of a method for preprocessing sensor data according to some embodiments of the present application. Specific Embodiments

[0024] The following describes in detail the embodiments of the present application. Examples of the embodiments are shown in the drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary only for explaining the present application and should not be construed as limiting the present application.

[0025] In the description of the present application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of the described features. In the description of the present application, "a plurality" means two or more unless otherwise specifically defined.

[0026] In the description of the present application, it should be noted that unless otherwise clearly specified and defined, the terms "installation" and "connection" should be understood in a broad sense, which may refer to fixed connection, detachable connection, or integral connection; it may be mechanical connection, electrical connection, or communication with each other; it may be directly connected, or indirectly connected through an intermediate medium, and may be the internal connection of two elements or the interaction relationship between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific circumstances.

[0027] The following disclosure provides many different embodiments or examples for implementing different structures of the present application. To simplify the disclosure of the present application, components and settings of specific examples are described below. Of course, they are only examples and are not intended to limit the present application. Additionally, the present application may repeat reference numerals and / or reference letters in different examples. This repetition is for the purpose of simplification and clarity and does not itself indicate the relationship between the various embodiments and / or settings discussed.

[0028] Embodiments of the present application will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where like or similar reference numerals denote like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary only for explaining the present application and should not be construed as limiting the present application.

[0029] Please refer to Figure 1 , the present application provides a preprocessing method for sensor data, which is used for an efficient air-conditioning refrigeration machine room. The preprocessing method includes:

[0030] 01: Obtain sensor data at different detection positions;

[0031] 02: Determine the type of sensor data based on the data attributes of the sensor data;

[0032] 03: Perform preprocessing on different types of sensor data using different outlier detection methods and / or missing value filling methods to obtain preprocessed sensor data.

[0033] Please refer to Figure 2 , the present application also provides a preprocessing device 10 for sensor data, which is used for an efficient air-conditioning refrigeration machine room. The preprocessing device 10 includes: an acquisition module 11, a type determination module 12, and a preprocessing module 13.

[0034] Step 01 can be implemented by the acquisition module 11, step 02 can be implemented by the type determination module 12, and step 03 can be implemented by the preprocessing module 13. The acquisition module 11 is used to obtain sensor data at different detection positions; the type determination module 12 is used to determine the type of sensor data based on the data attributes of the sensor data; the preprocessing module 13 is used to perform preprocessing on different types of sensor data using different outlier detection methods and / or missing value filling methods to obtain preprocessed sensor data.

[0035] Specifically, obtaining sensor data at different detection positions means that it is possible to obtain sensor data at different positions corresponding to different environments and equipment in an efficient air-conditioning refrigeration machine room, and it is possible to comprehensively obtain sensor data that affects the energy management process of the efficient air-conditioning refrigeration machine room.

[0036] Since the environmental parameters and equipment states monitored by sensors at different detection positions may be different, it is necessary to classify according to the physical meaning corresponding to the sensor data and use different outlier detection methods and missing value filling methods for processing.

[0037] Determine the type of sensor data based on the data attributes of the sensor data.

[0038] For example, the types of sensor data can include three categories. The first category is inertial data, the second category is non-inertial data, and the third category is enumerated data.

[0039] For different types of sensor data, different outlier detection methods and / or missing value filling methods are used for preprocessing to obtain preprocessed sensor data, including the following situations: (1) When the sensor data only includes inertial data and non-inertial data, different outlier detection methods and missing value filling methods can be used for preprocessing different types of sensor data; (2) When the sensor data only includes enumerated data, different missing value filling methods can be used for preprocessing different types of sensor data. It can be understood that enumerated data is usually a predefined finite state set. This means that each state value is reasonable and there is no so-called "outlier". For example, the operating state of a device may only have several states such as "running", "stopped", "faulty", etc., and these states are all reasonable and do not constitute an anomaly. Therefore, for enumerated data, there is no need to preprocess it through an outlier detection method, and only a missing value filling method is required for preprocessing.

[0040] The preprocessing method of the sensor data in this application can configure the attribution category of each sensor data, the data preprocessing algorithm for various types of sensor data, the value of algorithm-related parameters, and the algorithm execution frequency locally or in the cloud in an efficient air-conditioning refrigeration machine room to preprocess different types of sensor data, thereby solving the problems of abnormal or missing sensor data in the energy management process of the efficient air-conditioning refrigeration machine room and realizing the transformation of the original sensor database into a higher-quality sensor database.

[0041] It can be understood that based on the sensor database composed of preprocessed sensor data, energy consumption or energy efficiency analysis and report display, data mining and machine learning modeling, and optimization control of the machine room equipment in the efficient air-conditioning refrigeration machine room can be directly carried out, thereby further improving the air-conditioning energy efficiency of the efficient air-conditioning refrigeration machine room.

[0042] In this way, the preprocessing method of the sensor data in this application can solve the problems of abnormal or missing sensor data in the energy management process of the efficient air-conditioning refrigeration machine room, provide a high-quality data foundation for the efficient operation management and energy efficiency analysis of the efficient air-conditioning refrigeration machine room, enhance the stability of the optimization control algorithm, and improve the air-conditioning energy efficiency of the efficient air-conditioning refrigeration machine room.

[0043] In some embodiments, different types of sensors are installed at different detection positions in the efficient air-conditioning refrigeration machine room. Step 01 includes:

[0044] 011: Obtain the data collected by different types of sensors installed at different detection positions in the efficient air-conditioning refrigeration machine room to obtain sensor data.

[0045] Please combine with Figure 2 , step 011 can be implemented by the acquisition module 11. That is, the acquisition module 11 is used to acquire the data collected by different types of sensors installed at different detection positions in the high-efficiency air-conditioning refrigeration machine room, and obtain the sensor data.

[0046] Specifically, in the high-efficiency air-conditioning refrigeration machine room, in order to comprehensively and accurately monitor and control the operation status of the machine room, various types of sensors need to be installed at different detection positions. These sensors are responsible for collecting environmental parameters such as temperature, humidity, and light in the machine room, as well as data such as the operation status and energy consumption of the equipment.

[0047] In this way, the present application can obtain the data collected by different types of sensors installed at different detection positions in the high-efficiency air-conditioning refrigeration machine room as sensor data, and achieve comprehensive and accurate monitoring and control of the operation status of the machine room.

[0048] In some embodiments, the sensor data includes inertial data, non-inertial data, and enumerated data. The inertial data includes: the data whose numerical value does not suddenly change due to system inertia after the control instruction is issued; the non-inertial data includes: the data whose numerical value immediately responds to the change after the control instruction is issued; the enumerated data includes: the data of the data attribute reflecting the physical state of the sensor data.

[0049] Specifically, the inertial data may include, for example, the numerical values of the supply and return water temperatures of the chilled water, cooling water main pipes and branch pipes.

[0050] The non-inertial data may include, for example, the load rate of the chiller and the numerical values of the active power and frequency of equipment such as the chiller, chilled water pump, cooling water pump, and cooling tower.

[0051] The data of the data attribute reflecting the physical state may refer to the data reflecting the operation status and operation mode. The enumerated data may include, for example, the data reflecting the operation status of equipment such as the chiller, chilled water pump, cooling water pump, and cooling tower, and the data reflecting the operation mode of the chiller. Among them, the operation status of the equipment includes states such as "running", "stopped", and "fault", and the operation status of "stopped", "running", and "fault" can be represented by numerical values of 0, 1, and 2 respectively.

[0052] That is, the preprocessing method of the sensor data of the present application can classify the sensor data according to the data attributes of different sensor data, thereby laying a foundation for subsequent separate outlier detection and missing value filling for different sensor data.

[0053] Please refer to Figure 3 , in some embodiments, step 03 includes:

[0054] 031: When the sensor data is determined to be inertial data, the sliding window method is used to detect outliers in the inertial data;

[0055] 032: If there are outliers in the inertial data, the outliers are deleted and marked as null values.

[0056] Please refer to Figure 2 , steps 031 and 032 can be implemented by the preprocessing module 13. That is, the preprocessing module 13 is used to detect outliers in the inertial data by using the sliding window method when the sensor data is determined to be inertial data; if there are outliers in the inertial data, the outliers are deleted and marked as null values.

[0057] Specifically, the implementation of the sliding window in the sliding window method mainly relies on double pointers (or iterators) to identify the current processed subsequence (i.e., the window). In the processing of arrays or strings, the window moves step by step, and corresponding operations are performed at each position, such as calculating the maximum value, minimum value, sum, or other aggregation functions. In addition, by dynamically adjusting the window size and maintaining relevant state information, the sliding window algorithm can significantly reduce the time complexity and improve the algorithm efficiency.

[0058] That is, the preprocessing method of the sensor data in this application can detect outliers in the inertial data through the sliding window method. When there are outliers in the inertial data, the outliers can be deleted and marked as null values, so that the inertial data in the obtained sensor data can exclude outliers and improve the quality of the inertial data.

[0059] Please refer to Figure 4 , in some embodiments, step 031 includes:

[0060] 0311: Divide the inertial data into a preset number of sequences according to a time window of a preset length;

[0061] 0312: Calculate the first quartile, third quartile, and interquartile range of each sequence;

[0062] 0313: Calculate the lower limit value in the inertial data according to the first quartile and the interquartile range;

[0063] 0314: Calculate the upper limit value in the inertial data according to the third quartile and the interquartile range;

[0064] 0315: Determine the normal value range of the inertial data according to the lower limit value and the upper limit value;

[0065] 0316: Calculate the first difference according to the maximum value and the minimum value of each sequence

[0066] 0317: Calculate a second difference based on the upper limit value and the lower limit value. If the inertial data exceeds the normal value range and both the first difference and the second difference are greater than a preset isolation limit value, determine that the inertial data is an outlier.

[0067] Please combine Figure 2 , step 0311, step 0312, step 0313, step 0314, step 0315, step 0316, and step 0317 can be implemented by the preprocessing module 13. That is, the preprocessing module 13 is used to divide the inertial data into a preset number of data sequences according to a time window with a preset length; calculate the first quartile, the third quartile, and the interquartile range of each data sequence; calculate the lower limit value in the inertial data based on the first quartile and the interquartile range; calculate the upper limit value in the inertial data based on the third quartile and the interquartile range; determine the normal value range of the inertial data according to the lower limit value and the upper limit value; calculate the first difference according to the maximum value and the minimum value of each data sequence; calculate the second difference according to the upper limit value and the lower limit value. If the inertial data exceeds the normal value range and both the first difference and the second difference are greater than the preset isolation limit value, determine that the inertial data is an outlier.

[0068] Specifically, the preset length of the time window is 30 min, and the length setting of the time window can be adjusted according to the requirement of the sensitivity to outlier detection. The preset number of divided data sequences can be determined jointly according to the quantity of the inertial data and the length of the time window.

[0069] For example, the specific implementation process of detecting outliers in inertial data by using the sliding window method can be as follows:

[0070] First, select the inertial data with the data at the current timestamp as the center, and take the data corresponding to half of the time window forward and backward to form a data sequence. For example, assume that the data is recorded once an hour. If the time window is 2 hours and the current timestamp is 12:00, the data between 11:00 and 13:00 can be selected to form a data sequence. Among them, the timestamp is a mark indicating the exact time when an event or data record occurs. It can be a combination of date and time, such as "2023-04-01 12:00:00". In time series data, each timestamp will correspond to one or more data values. For example, if you are monitoring the number of visits to a website, each timestamp (such as every hour) will have a corresponding number of visit values.

[0071] That is, for the data at each timestamp, take the data of the 1 / 2 time windows before and after it to form a sequence of numbers, and calculate the first quartile (Q1, 25%), the third quartile (Q3, 75%), and the interquartile range (IQR) of this sequence of numbers. For the constructed sequence of numbers, calculate the first quartile (Q1, 25%) and the third quartile (Q3, 75%), and then calculate the interquartile range IQR = Q3 - Q1. Then determine the upper and lower limits for outlier identification based on Q1 - 1.5 * IQR and Q3 + 1.5 * IQR.

[0072] Secondly, when the difference range of the sequence of numbers within a time window is not large, the outlier detection sensitivity of the sliding window method is relatively high. At this time, outliers can be excluded through the "isolation limit" indicator. For example, for the supply water temperature of the chilled water main pipe, the isolation limit can be set to 1°C.

[0073] If the data value at this timestamp exceeds the normal value range defined by the upper and lower limits, the range of the sequence of numbers within the window corresponding to the data at this timestamp is greater than the isolation limit, and the difference between the upper and lower limits for outlier identification is also greater than the isolation limit, then the data at this timestamp will be recorded as an outlier. Among them, the range is the difference between the maximum value and the minimum value in the sequence of numbers.

[0074] If the data at a certain timestamp is recorded as an outlier, then delete it and mark it as a null value. The null value can be represented by NA, or Null, or None, for example.

[0075] In this way, the preprocessing method of the sensor data in this application can calculate the upper limit value and the lower limit value in the inertial data through the sliding window method, the first difference between the maximum value and the minimum value of each sequence of numbers, and calculate the second difference between the upper limit value and the lower limit value. Determine the normal value range of the inertial data according to the lower limit value and the upper limit value. If the inertial data exceeds the normal value range and both the first difference and the second difference are greater than the preset isolation limit, then determine that the inertial data is an outlier.

[0076] In some embodiments, step 03 includes:

[0077] 033: When the sensor data is determined to be non-inertial data, if the non-inertial data exceeds the preset value range, determine that the non-inertial data is an outlier, and delete the outlier and mark it as a null value.

[0078] Please combine Figure 2 , step 033 can be implemented by the preprocessing module 13. That is, the preprocessing module 13 is used to determine that the non-inertial data is an outlier when the sensor data is determined to be non-inertial data and the non-inertial data exceeds the preset value range, and delete the outlier and mark it as a null value.

[0079] Specifically, the preset value range can be a reasonable range set according to the normal working data of different non-inertial data.

[0080] For example, for the active power of the device, the lower limit of the preset value range of the active power of the device is 0, and the upper limit of the preset value range is 1.5 times the rated power.

[0081] It can be understood that when the sensor data is non-inertial data and the non-inertial data exceeds the preset value range, it indicates that the non-inertial data is an outlier. At this time, the outlier can be deleted and marked as a null value, which can exclude the influence of the outlier on the application efficiency of all sensor data.

[0082] Please refer to Figure 5 , in some embodiments, step 03 includes:

[0083] 034: When the sensor data is determined to be inertial data or non-inertial data, identify whether there is a missing value in the sensor data; if a missing value is identified in the sensor data, use the linear interpolation method to fill the missing value;

[0084] 035: When the sensor data is determined to be enumeration data, identify whether there is a missing value in the sensor data; if a missing value is identified in the sensor data, fill the missing value based on the preset physical rules; if the missing value cannot be filled based on the preset physical rules, do not process the missing value or mark it as a null value.

[0085] Please combine Figure 2 , step 034 and step 035 can be implemented by the preprocessing module 13. That is, the preprocessing module 13 is used to identify whether there is a missing value in the sensor data when the sensor data is determined to be inertial data or non-inertial data; if a missing value is identified in the sensor data, use the linear interpolation method to fill the missing value; when the sensor data is determined to be enumeration data, identify whether there is a missing value in the sensor data; if a missing value is identified in the sensor data, fill the missing value based on the preset physical rules; if the missing value cannot be filled based on the preset physical rules, do not process the missing value or mark it as a null value.

[0086] Specifically, the process of identifying whether there is a missing value in the sensor data can be that when the sensor data within a certain preset time period cannot be obtained normally, it is determined that there is a missing value in the sensor data within that time period. For example, the preset time can be 30 minutes, 31 minutes, or 32 minutes, which is not limited here.

[0087] Taking the preset time period of 30 minutes as an example, that is to say, within 30 consecutive minutes, the sensor data within that time period may be missing due to reasons such as device offline or outlier recognition.

[0088] At this time, the missing values of the sensor data within this time period can be filled by linear interpolation, and the maximum filling time required for filling can be adjusted according to the requirements for filling the missing data.

[0089] In this way, the preprocessing method of the sensor data of the present application can fill the missing values of inertial data or non-inertial data by linear interpolation, further improving the quality of the obtained inertial data or non-inertial data.

[0090] It can be understood that if the reported value of the sensor for the device operating state is not empty, the reported sensor data is retained. If the reported value of the sensor for the device operating state is empty, that is, the enumerated data is missing, it can be determined whether the device is operating according to the preset physical rules, so as to fill the missing sensor data. If the missing value cannot be filled based on the preset physical rules, the missing value is not processed or marked as a null value.

[0091] In this way, the preprocessing method of the sensor data of the present application can fill the missing values of the enumerated data based on the preset physical rules. If the missing value cannot be filled based on the preset physical rules, the missing value is not processed or marked as a null value, further improving the quality of the obtained enumerated data.

[0092] Please refer to Figure 6 , in some embodiments, the enumerated data includes the device operating state of the operating devices in the high-efficiency air-conditioning refrigeration machine room, the device operating state includes the operating state and the stop state, and step 035 includes:

[0093] 0351: Compare the load rate and power of the operating device with the preset load range and preset power range to determine whether the device operating state of the operating device is the operating state or the stop state;

[0094] 0352: Fill the missing value according to the judgment result of the device operating state.

[0095] Please combine Figure 2 , step 0351 and step 0352 can be implemented by the preprocessing module 13. That is, the preprocessing module 13 is used to compare the load rate and power of the operating device with the preset load range and preset power range to determine whether the device operating state of the operating device is the operating state or the stop state; fill the missing value according to the judgment result of the device operating state.

[0096] Specifically, for the operating state of the chiller, when the load ratio of the chiller is greater than 20% and the power of the chiller is greater than 10 kW, or the load ratio of the chiller is greater than 20% and the power of the chiller is missing, or the load ratio of the chiller is missing and the power of the chiller is greater than 10 kW, it can be determined at this time that the chiller is in the operating state, that is, the missing enumerated data at this time is the data indicating the "operation" of the chiller.

[0097] When the load ratio of the chiller is less than or equal to 20% and the power of the chiller is less than or equal to 10 kW, or the load ratio of the chiller is less than or equal to 20% and the power of the chiller is missing, or the load ratio of the chiller is missing and the power of the chiller is less than or equal to 10 kW, it can be determined at this time that the chiller is in the stopped state, that is, the missing enumerated data at this time is the data indicating the "stop" of the chiller.

[0098] If the load ratio and power of the cooling unit do not meet the above-mentioned preset physical rules, the missing value of the operating state of the cooling unit can be left unprocessed or marked as a null value.

[0099] In this way, when the enumerated data is missing, the preprocessing method of the sensor data in this application can compare the load ratio and power of the operating equipment with the preset load range and preset power range to determine whether the operating state of the operating equipment is the operating state or the stopped state, fill in the missing value according to the judgment result of the equipment operating state, and if the missing value cannot be filled based on the preset physical rules, the missing value is left unprocessed or marked as a null value, providing higher-quality enumerated data, providing a high-quality data foundation for the efficient operation management and energy efficiency analysis of the high-efficiency air-conditioning refrigeration machine room, enhancing the stability of the optimization control algorithm, and improving the air-conditioning energy efficiency of the high-efficiency air-conditioning refrigeration machine room.

[0100] In addition, for different enumerated data, different preset physical rules can be set to fill in the missing value. This application can also compare the frequency and power of the operating equipment with the preset frequency range and preset power range to determine whether the operating state of the operating equipment is the operating state or the stopped state, and fill in the missing value according to the judgment result of the equipment operating state.

[0101] Specifically, for the operating state of the cooling water pump, chilled water pump or cooling tower, when the frequency of the cooling water pump, chilled water pump or cooling tower is greater than 10 Hz and the power of the cooling water pump, chilled water pump or cooling tower is greater than 1 kW, or the frequency of the cooling water pump, chilled water pump or cooling tower is greater than 10 Hz and the power is missing, or the load ratio of the cooling water pump, chilled water pump or cooling tower is missing and the power is greater than 1 kW, it can be determined at this time that the cooling water pump, chilled water pump or cooling tower is in the operating state, that is, the missing enumerated data at this time is the data indicating the "operation" of the chiller.

[0102] When the frequency of the cooling water pump, the chilled water pump or the cooling tower is less than or equal to 10 Hz and the power of the cooling water pump, the chilled water pump or the cooling tower is less than or equal to 1 kW, or the frequency of the cooling water pump, the chilled water pump or the cooling tower is less than or equal to 10 Hz and the power is missing, or the load rate of the cooling water pump, the chilled water pump or the cooling tower is missing and the power of the cooling water pump, the chilled water pump or the cooling tower is less than or equal to 1 kW, it can be determined at this time that the cooling water pump, the chilled water pump or the cooling tower is in a stopped state, that is, the missing enumerated data at this time is the data indicating "stopped" of the cooling water pump, the chilled water pump or the cooling tower.

[0103] If the frequency and power of the cooling water pump, the chilled water pump or the cooling tower do not satisfy the above preset physical rules, the missing value of the operating state of the cooling water pump, the chilled water pump or the cooling tower may not be processed or the missing value may be marked as a null value.

[0104] In this way, the preprocessing method of the sensor data of the present application can, when the enumerated data is missing, also compare the frequency and power of the operating equipment with the preset frequency range and preset power range, determine whether the operating state of the operating equipment is the operating state or the stopped state, fill in the missing value according to the judgment result of the operating state of the equipment, and if the missing value cannot be filled based on the preset physical rules, the missing value is not processed or marked as a null value, providing higher-quality enumerated data, providing a high-quality data base for the efficient operation management and energy efficiency analysis of the high-efficiency air-conditioning refrigeration machine room, enhancing the stability of the optimization control algorithm, and improving the air-conditioning energy efficiency of the high-efficiency air-conditioning refrigeration machine room.

[0105] The present application also provides a high-efficiency air-conditioning refrigeration machine room. The high-efficiency air-conditioning refrigeration machine room is used to execute the preprocessing method of the sensor data described in any one of the above embodiments. The high-efficiency air-conditioning refrigeration machine room includes the preprocessing device of the sensor data described in the above embodiments.

[0106] Applying the preprocessing method of the sensor data in the high-efficiency air-conditioning refrigeration machine room of the present application can solve the problems of abnormal or missing sensor data in the energy management process of the high-efficiency air-conditioning refrigeration machine room, provide a high-quality data base for the efficient operation management and energy efficiency analysis of the high-efficiency air-conditioning refrigeration machine room, enhance the stability of the optimization control algorithm, and improve the air-conditioning energy efficiency of the high-efficiency air-conditioning refrigeration machine room.

[0107] The present application also provides a non-volatile computer-readable storage medium containing a computer program. When the computer program is executed by one or more processors, the preprocessing method of the sensor data described in any one of the above embodiments is implemented.

[0108] The computer-readable storage medium of the present application applying the above-mentioned preprocessing method for sensor data can solve the problems of abnormal or missing sensor data existing in the energy management process of an efficient air-conditioning refrigeration machine room, provide a high-quality data foundation for the efficient operation management and energy efficiency analysis of the efficient air-conditioning refrigeration machine room, enhance the stability of the optimization control algorithm, and improve the air-conditioning energy efficiency of the efficient air-conditioning refrigeration machine room.

[0109] The present application also provides a computer program product. The computer program product includes a computer program, which when executed by one or more processors, implements the preprocessing method for sensor data described in any one of the above embodiments.

[0110] The computer program product of the present application applying the above-mentioned preprocessing method for sensor data can solve the problems of abnormal or missing sensor data existing in the energy management process of an efficient air-conditioning refrigeration machine room, provide a high-quality data foundation for the efficient operation management and energy efficiency analysis of the efficient air-conditioning refrigeration machine room, enhance the stability of the optimization control algorithm, and improve the air-conditioning energy efficiency of the efficient air-conditioning refrigeration machine room.

[0111] The above embodiments only represent several implementation manners of the present application, and their descriptions are relatively specific and detailed, but should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several deformations and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. A sensor data preprocessing method for a high-efficiency air-conditioning refrigeration room, characterized in that: The pretreatment method comprises: Get sensor data at different detection locations; determining a type of the sensor data based on a data attribute of the sensor data; Different outlier detection methods and / or missing value filling methods are used for preprocessing of different types of sensor data to obtain preprocessed sensor data.

2. The sensor data preprocessing method according to claim 1, characterized in that: Different types of sensors are installed at different detection positions in the high-efficiency air-conditioning refrigeration room, and the acquisition of sensor data at different detection positions includes: The data collected by different types of sensors installed at different detection positions in the high-efficiency air-conditioning and refrigeration room are obtained to obtain the sensor data.

3. The pretreatment method according to claim 1, characterized in that: The types of sensor data include inertial data, non-inertial data and enumeration data. The inertia data includes: the sensor data whose value will not change suddenly due to the system inertia value after the control instruction is issued; The non-inertial data includes: data whose value changes immediately after the control instruction is issued; The enumeration data includes: the sensor data is data that reflects the data attributes of the physical state.

4. The pretreatment method according to claim 3, characterized in that: The sensor data after preprocessing is obtained by using different outlier detection methods and / or missing value filling methods for different types of sensor data, and includes: When the sensor data is determined to be the inertial data, a sliding window method is used to detect abnormal values ​​in the inertial data; If the abnormal value exists in the inertial data, the abnormal value is deleted and marked as a null value.

5. The pretreatment method according to claim 4, characterized in that: When the sensor data is determined to be the inertial data, detecting abnormal values ​​in the inertial data using a sliding window method includes: Dividing the inertial data into a preset number of series according to a time window of a preset length; Calculate the first quartile, third quartile, and interquartile range for each series; Calculate the lower limit value of the inertia data according to the first quartile and the interquartile range; Calculate the upper limit value of the inertia data according to the third quartile and the interquartile range; Determine a normal numerical range of the inertial data according to the lower limit value and the upper limit value; Calculate the first difference between the maximum value and the minimum value of each of the number sequences; A second difference is calculated according to the upper limit and the lower limit, and if the inertia data exceeds the normal value range and both the first difference and the second difference are greater than a preset isolation limit, the inertia data is determined to be the abnormal value.

6. The pretreatment method according to claim 3, characterized in that: The sensor data after preprocessing is obtained by using different outlier detection methods and / or missing value filling methods for different types of sensor data, and includes: When the sensor data is determined to be the non-inertial data, if the non-inertial data exceeds a preset value range, the non-inertial data is determined to be an abnormal value, and the abnormal value is deleted and marked as a null value.

7. The pretreatment method according to claim 3, characterized in that: The sensor data after preprocessing is obtained by using different outlier detection methods and / or missing value filling methods for different types of sensor data, and includes: When the sensor data is determined to be the inertial data or the non-inertial data, identifying whether the sensor data has missing values; if it is identified that there are missing values ​​in the sensor data, using linear interpolation to fill the missing values; When the sensor data is determined to be the enumeration data, identify whether there are missing values ​​in the sensor data; if it is identified that there are missing values ​​in the sensor data, fill the missing values ​​based on preset physical rules; if the missing values ​​cannot be filled based on the preset physical rules, do not process the missing values ​​or mark them as null values.

8. The pretreatment method according to claim 7, characterized in that: The enumeration data includes the equipment operation status of the equipment running in the high-efficiency air-conditioning refrigeration room, and the equipment operation status includes the operation status and the stop status. If it is identified that there are missing values ​​in the sensor data, the missing value filling based on the preset physical rules includes: Based on the comparison between the load rate and power of the running device and the preset load range and the preset power range, determining that the running state of the running device is the running state or the stopped state; The missing values ​​are filled in according to the judgment result of the operation status of the equipment.

9. A sensor data preprocessing device for a high-efficiency air-conditioning refrigeration room, characterized in that: The sensor data preprocessing device comprises: An acquisition module, used to acquire sensor data at different detection positions; A type determination module, configured to determine the type of the sensor data based on data attributes of the sensor data; The preprocessing module is used to preprocess different types of sensor data using different outlier detection methods and / or missing value filling methods to obtain preprocessed sensor data.

10. A non-volatile computer-readable storage medium containing a computer program, characterized in that: When the computer program is executed by one or more processors, the sensor data preprocessing method according to any one of claims 1 to 8 is implemented.

Images