Intelligent ship energy efficiency data quality monitoring management method

Abstract

The application discloses a kind of intelligent ship energy efficiency data quality monitoring management methods, specific steps are as follows: S1: collection intelligent ship energy efficiency data, back data to shore end intelligent ship energy efficiency database;S2: data quality evaluation method is established to intelligent ship energy efficiency data, and data quality monitoring is carried out;S3: missing and abnormal data are repaired and corrected by data quality management method;S4: in combination with ship energy consumption model, real-time update data quality evaluation method;S5: real-time data quality monitoring: qualified data is written into the data set for intelligent ship modeling;Unqualified data forms data quality problem feedback sheet, and promotes data acquisition method optimization.Its advantages are: it can effectively eliminate low-quality data, improve machine learning efficiency and precision;Data problems are fed back to data acquisition method in time, and plan-execution-check-treatment cycle management is formed.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of ship energy efficiency data management, and in particular to an intelligent ship energy efficiency data quality monitoring management method. BACKGROUND

[0002] The intelligent ship energy efficiency data has wide collection range, high frequency and large data volume. The current data frequency of ship-shore transmission level is 1 per second. Data problems are caused by computer clock and equipment or line fault. The former can be repaired by data quality management to repair one or several missing data (values), and the latter will generally continue until the operation and maintenance personnel repair it. In terms of a single ship, if the probability of such failure is less than 3%, the probability of missing several data values in a certain voyage is 50%.

[0003] At present, the research on intelligent ships at home and abroad mainly focuses on the mining and analysis of intelligent ship energy efficiency data and the establishment of models to predict energy consumption. There is less research on how to ensure the quality of intelligent ship energy efficiency data to meet the analysis and modeling. Intelligent ship energy efficiency data monitoring is the basis for the development and utilization of intelligent ship energy efficiency data. Intelligent ship energy efficiency data involves sensor collection, signal transmission, computer reading, calculation and storage processes. The whole process has many links, poor communication conditions and complex calculation logic.

[0004] Data quality management has formed some quality standards. By selecting a suitable standard set, specific data can be efficiently and specifically filtered. However, the generation process of intelligent ship energy efficiency data has its own characteristics such as multiple attributes, large data volume, multiple transmission links and poor communication conditions. Therefore, there is a certain mismatch in screening data according to the existing data quality management method. Therefore, how to establish a special data quality monitoring and management method for intelligent ship energy efficiency data has become a problem to be solved in the field of ship energy efficiency data management. SUMMARY

[0005] In order to solve the technical problem that the existing technology in the field cannot provide an effective data quality monitoring and management method for intelligent ship energy efficiency data with multiple attributes, large data volume, multiple transmission links and poor communication conditions, the present application provides an intelligent ship energy efficiency data quality monitoring and management method, and in particular provides a specific and effective evaluation mechanism for integrity, continuity and effectiveness in data quality monitoring, thereby laying a solid foundation for data mining and simulation modeling of intelligent ship energy efficiency data.

[0006] To achieve the above purpose, the present application adopts the following technical solutions:

[0007] An intelligent ship energy efficiency data quality monitoring and management method, comprising the following steps:

[0008] S1: Set up a data collection system to collect intelligent ship energy efficiency data and transmit it to a shore-based intelligent ship energy efficiency database;

[0009] S2: Establish a data quality evaluation method for intelligent ship energy efficiency data and conduct data quality monitoring: the quality evaluation method is to write an integrity, continuity and effectiveness evaluation mechanism instruction to evaluate whether the data meets the integrity, continuity and effectiveness, if it meets the instruction, the data is valid data; if it does not meet, the data is missing data; the integrity includes detecting whether there is a whole data missing through the is continuous() function, detecting whether there is an attribute value missing through the is blank() function, and using the is normal() function to detect whether the attribute value is within the normal range; the continuity includes whether the change of continuous data sequence is abnormal, and the effectiveness includes the proportion of normal data number;

[0010] S3: Data quality management: evaluate whether the missing data in S2 meets the quality management condition, repair the missing data in S2 and correct the abnormal data according to the missing or abnormal condition in the quality management condition;

[0011] S4: Real-time update of data quality evaluation method combined with ship energy consumption model; use the valid data in S2 and the data after the data quality management in S3 as the input set, use the multilayer perceptron neural network to establish the ship energy consumption model, use the get_params function to obtain the ship energy consumption result calculated by the input set, and compare the matching degree of the result with the true value, feed back the matching degree information to S2, update the evaluation mechanism of integrity, continuity and effectiveness, use the updated evaluation mechanism to monitor the quality of the data in S1 again, quality management, and finally obtain the updated input set to input into the ship energy consumption model, until the matching degree of the ship energy consumption result and the true value reaches the matching degree threshold.

[0012] Preferably, the energy efficiency data includes sailing state, rotating speed, wave direction, wind speed, bow draft, stern draft, wind direction, wave height, wave period, flow speed, flow direction, and heading angle.

[0013] Preferably, the method for evaluating data integrity in S2 is:

[0014] A1: Use the is continuous() function to check whether there is a whole data missing according to the UTC time sequence: sort the data by time, the time difference between the previous and next data is 1 second, if the difference is greater than 1 second, there is a whole data missing; within 30 minutes, if there is no more than 30 data missing, the data is valid data;

[0015] A2: using the is_blank() function, check if there is a blank value in each attribute of each data: according to the attribute list, check if there is a blank value in the attribute list in each valid data obtained by A2: if there is a blank value, the data is abnormal; if there is no blank value, the attribute value is normal;

[0016] A3: using the is_normal() function to check whether each attribute in the data is within the normal range: checking whether the attribute value in the valid data obtained is within the given range: if they are all within the range, they are normal, and if there is an out-of-range, they are abnormal.

[0017] Preferably, the method for evaluating data continuity in S2 is: using the is_regular() function, check if the data attribute has abnormal changes in the data sequence of a certain period of time: check if the maximum and minimum value difference of a single attribute value in the valid data obtained by A3 within a certain period of time is out of limit, and if it is out of limit, it is abnormal.

[0018] Preferably, the method for evaluating data validity in S2 is: obtaining the number of original data number and the number of data amount determined to be normal after the evaluation of continuity; wherein, ratio = amount / number;

[0019] When ratio>0.8, it indicates that the data validity is good, and the ship energy efficiency data is written into the data set for intelligent ship modeling;

[0020] When 0.4<ratio<0.8, it indicates that the data validity is better, then the data is imported into the data management method for data filling;

[0021] When ratio<0.4, it indicates that the data validity is poor, and the data is considered invalid and is removed.

[0022] Preferably, for the missing data in S2, a data quality problem feedback form is formed and fed back to the data collection system in S1 for optimization, including: removing the collection of ship energy efficiency data whose influence on ship energy consumption results is lower than a certain threshold.

[0023] Preferably, the method for repairing the missing data in S3 is:

[0024] The missing data is repaired using the fillup() function: based on the is_continuous() function result, if the missing data within 1 minute does not exceed 3, the missing data is filled according to the average value of the attributes of the data before and after the time, if it is continuous missing, it is linearly interpolated according to the time; the data that cannot be repaired is considered as not meeting the data quality management condition, and the data is removed.

[0025] Preferably, the method for correcting abnormal data in S3 is:

[0026] The abnormal data is corrected using a pad() function: for a single attribute, when there are valid attribute values in the 30 data before and after the current data, the result can be obtained by interpolation according to time to fill in; the uncorrectable data is regarded as not meeting the data quality management condition, and the data is removed.

[0027] The present application has the following beneficial effects:

[0028] The present application provides an intelligent ship energy efficiency data quality monitoring and management method, and particularly provides a specific and effective evaluation mechanism for integrity, continuity and effectiveness in data quality monitoring, which lays a solid foundation for data mining and simulation modeling of intelligent ship energy efficiency data. Specifically, in the S2 step, a specific and effective evaluation mechanism for integrity, continuity and effectiveness is set, the integrity includes detecting whether there is a whole data missing through an is_continuous() function, detecting whether there is an attribute value missing through an is_blank() function, and detecting whether the attribute value range is normal through an is_normal() function; the continuity includes whether the change of the continuous data sequence is abnormal, and the effectiveness includes the proportion of the number of normal data; the quality management of the integrity, continuity and effectiveness of the data is realized. The present application lays a certain foundation for the following ship energy consumption model, and the fitting precision of the machine learning fitting ship energy consumption model is improved by about 1.52%.

[0029] Secondly, in the S3 step, the present application obtains the calculation result of the ship energy consumption model, calculates the matching degree with the true value, updates the evaluation mechanism using the matching degree, improves the evaluation method, enables the data problems to be found in the process of intelligent ship energy efficiency data quality monitoring and management to be fed back to the data acquisition method in time, realizes functions such as fault elimination, data calibration, transmission diagnosis and the like in the acquisition process, forms a plan-execution-check-treatment cycle, and returns the data quality problem to the data full life cycle for management. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 The present application constructs a flowchart of an intelligent ship energy efficiency data quality monitoring and management method. DETAILED DESCRIPTION

[0031] The present application will be further described in detail below with reference to the accompanying drawings, so as to be understood by the skilled in the art:

[0032] Referring to Figure 1 , Figure 1 is a flowchart of an intelligent ship energy efficiency data quality monitoring and management method constructed by the present application.

[0033] An intelligent ship energy efficiency data quality monitoring management method, comprising the following steps:

[0034] S1: setting a data acquisition system to collect intelligent ship energy efficiency data and transmit the data to a shore-end intelligent ship energy efficiency database;

[0035] S2: establishing a data quality evaluation method for the intelligent ship energy efficiency data to perform data quality monitoring: the quality evaluation method is to write an integrity, continuity and validity evaluation mechanism instruction to evaluate whether the data meets the integrity, continuity and validity, if it meets the instruction, the data is valid data; if it does not meet, the data is missing data; the integrity includes detecting whether there is missing whole data by using an is_continuous() function, detecting whether there is missing attribute value by using an is_blank() function, and detecting whether the attribute value is within a normal range by using an is_normal() function; the continuity includes whether the change of a continuous data sequence is abnormal, and the validity includes a proportion of normal data number;

[0036] S3: data quality management: evaluating whether the missing data in S2 meets the quality management condition, repairing the missing data in S2 and correcting the abnormal data according to the missing or abnormal condition in the quality management condition;

[0037] S4: combining a ship energy consumption model to update the data quality evaluation method in real time; using the valid data in S2 and the data after the data quality management in S3 as an input set, using a multilayer perceptron neural network to establish a ship energy consumption model, using a get_params function to obtain a ship energy consumption result calculated by the input set, and comparing the result with a true value to obtain a matching degree, feeding back the matching degree information to S2 to update the evaluation mechanism of integrity, continuity and validity, using the updated evaluation mechanism to monitor the data in S1 again to obtain an updated input set, and inputting the input set into the ship energy consumption model until the matching degree of the ship energy consumption result and the true value reaches a matching degree threshold.

[0038] Preferably, the energy efficiency data includes sailing state, rotating speed, wave direction, wind speed, bow draft, stern draft, wind direction, wave height, wave period, flow speed, flow direction and heading angle.

[0039] Preferably, the method for evaluating data integrity in S2 is:

[0040] A1: using an is_continuous() function to check whether there is missing whole data in a UTC time sequence: sorting the data by time, the time difference between the previous and next data is 1 second, if the difference is greater than 1 second, there is missing whole data; if there is no more than 30 missing data in 30 minutes, the data is valid data;

[0041] A2: using the is_blank() function, check whether there is a blank value in each attribute of each data: according to the attribute list, check whether there is a blank value in the attribute list in each valid data obtained by A2: if there is a blank value, the data is abnormal; if there is no blank value, the attribute value is normal;

[0042] A3: using the is_normal() function to check whether each attribute in the data is within the normal range: checking whether the attribute value in the valid data obtained by A3 is within the given range: if they are all within the range, they are normal, and if there is an out-of-range, they are abnormal.

[0043] Preferably, the method for evaluating data continuity in S2 is: using the is_regular() function, checking whether the data attribute has abnormal changes in the data sequence of a certain period of time: checking whether the maximum and minimum value difference of a single attribute value in the valid data obtained by A3 within a continuous period of time is out of limit, and if it is out of limit, it is abnormal.

[0044] Preferably, the method for evaluating data validity in S2 is: obtaining the number of original data number and the number of data amount determined as normal after evaluating the continuity; wherein, ratio = amount / number;

[0045] When ratio>0.8, it indicates that the data validity is good, and the ship energy efficiency data is written into the data set for intelligent ship modeling;

[0046] When 0.4<ratio<0.8, it indicates that the data validity is better, and the data is imported into the data management method for data filling;

[0047] When ratio<0.4, it indicates that the data validity is poor, and the data is regarded as invalid data and is eliminated.

[0048] Preferably, for the missing data in S2, a data quality problem feedback form is formed and fed back to the data collection system S1 for optimization, including: eliminating the collection of ship energy efficiency data whose influence on ship energy consumption results is lower than a certain threshold.

[0049] Preferably, the method for repairing the missing data in S3 is:

[0050] The missing data is repaired using the fillup() function: based on the is_continuous() result, if the missing data within 1 minute does not exceed 3, the missing data is filled according to the average value of the attributes of the data before and after the time, and if it is continuous missing, it is linearly interpolated according to the time; the data that cannot be repaired is regarded as not meeting the data quality management condition and is eliminated.

[0051] Preferably, the method for correcting abnormal data in S3 is:

[0052] The abnormal data is corrected using the pad() function: for a single attribute, if there are valid attribute values in the current and next 30 data, the result can be obtained by time interpolation for filling; the uncorrectable data is considered as not meeting the data quality management conditions, and the data is excluded.

[0053] An example of checking data missing using the is_continuous() function in S2 is listed below, as shown in Table 1:

[0054] Table 1

[0055]

[0056] Taking the above table data as an example:

[0057] 1. The original data has been arranged in chronological order;

[0058] 2. From the front to the back, first compare 2022 / 5 / 10 12:26:23 and 2022 / 5 / 10 12:26:24, the difference is one second, indicating that the data is normal and continuous; when comparing 2022 / 5 / 10 12:26:25 and 2022 / 5 / 10 12:26:27, the difference is two seconds, indicating that the data is not normal, there is missing data in the two data, and the missing number is 1 (the difference minus 1 is the number of missing data, 2-1=1);

[0059] 3. In the code run, a note will be made in the 2022 / 5 / 10 12:26:25 data, and there is 1 missing data after this data;

[0060] 4. After the first round of calculating data missing and making notes, the second round of data is selected according to the notes. Starting from 2022 / 5 / 10 12:26:23, 30 minutes after 2022 / 5 / 10 12:26:23 is 2022 / 5 / 10 12:56:23, the number of missing data in the data within the time range of 2022 / 5 / 10 12:26:23 to 2022 / 5 / 10 12:56:23 is accumulated, if the total is less than or equal to 30, all data in the period is valid, if the total is greater than 30, no operation is performed; after the judgment is completed, the earliest time with missing data in the time range is set as the next start time, and the next judgment is performed until all data is completed;

[0061] 5. The difference value for judging time continuity is defaulted to 1 second, the data validity time judgment range is defaulted to 30 minutes, and the judgment standard is defaulted to 30, but these three standards will be adjusted according to the model training set size and model fitting results.

[0062] Below as shown in Table 2, citing an instance of using the is_blank() function in S2, checking whether each attribute in each piece of data has a missing value, according to the attribute list (as shown in Table 3), checking whether there is a null value in the attribute value in each piece of valid data obtained in step S211: if there is a null value, the data is abnormal; if there is no null value, the attribute value is normal.

[0063] Table 2

[0064]

[0065] 1. The original data has been arranged in chronological order;

[0066] 2. From front to back, first check 2022 / 5 / 10 12:26:23, each value is not null, 2022 / 5 / 10 12:26:23 is valid data;

[0067] 3. When judging the data of 2022 / 5 / 10 12:26:26, it is found that the value of the stern draft attribute is null, so the data of 2022 / 5 / 10 12:26:26 is invalid data;

[0068] Table 3

[0069]

[0070]

[0071] S213: Check whether the attribute values in the valid data obtained in step S212 are within the given range: all within the range are normal, and there is an out-of-range value, which is abnormal.

[0072] Below is an example of using the is_normal() function to check whether each attribute in the data is within the normal range. Take the data in Table 4 as an example to check whether the attribute values are abnormal:

[0073] Table 4

[0074]

[0075] Take the above table data as an example:

[0076] 1. The original data has been arranged in chronological order;

[0077] 2. For different attributes of each ship, configure the corresponding range value. For example, for a single ship in the table, the given bow draft range is 3-10m, the stern draft range is 3.3-11m, the wave height is 0-12m, the wave direction is 0-360°, the wind speed is 0-30m / s, and the wind direction is 0-360°;

[0078] 3. Determine whether each value in the data is in the given range. If it is outside the range, the data is invalid. If there is no data value outside the range, the data is valid.

[0079] 4. Check the data at 2022 / 5 / 10 12:26:29 and find that the stern draft value 19.29 is outside the range of 3.3-11m. Therefore, the data at 2022 / 5 / 10 12:26:29 is invalid.

[0080] Take the data in Table 5 as an example. Use the is_regular() function to check whether the data attribute has abnormal changes in the data sequence of the previous and subsequent time periods.

[0081] Table 5

[0082]

[0083]

[0084] 1. The original data has been arranged in chronological order.

[0085] 2. For the data values of a continuous period of time, calculate the maximum and minimum values, and the ratio of the maximum value to the minimum value. If the ratio exceeds a certain amount, the data is invalid. Otherwise, the ratio is less than or equal to a certain amount, and the data is valid.

[0086] 3. Check 2022 / 5 / 10 12:26:23 to 2022 / 5 / 10 12:26:32. Within 10 seconds, assume the maximum-to-minimum value ratio is 30%. Check the stern draft maximum value 10.31 and the minimum value 5.35. The ratio is (10.31-5.35) / 5.35, which is about 93%, exceeding 30%. Therefore, this segment of data is invalid.

[0087] Take the data in Table 6 as an example. When the is_continuous() function finds missing data and the amount of missing data is small, it can be repaired using the fillup() function.

[0088] Table 6

[0089]

[0090] 1、is_continuous() function found missing two data between 2022 / 5 / 10 12:26:25 and 2022 / 5 / 10 12:26:28, then based on 25 and 28 data to 26, 27 data value difference, such as 2022 / 5 / 10 12:26:26 bow draft value equals 5.25+(26-25) / (28-25)*(5.22-5.25) (where 26, 25, 28 are the simplified values of the timestamp)

[0091] 2、fillup() function repair 2022 / 5 / 10 12:26:26 and 2022 / 5 / 10 12:26:27 data results as shown in table 7:

[0092] Table 7

[0093]

[0094]

[0095] As shown in table 8, when using is_blank() function to find missing data value and the data before and after is complete, fillup() function can be used to repair missing data;

[0096] Table 8

[0097]

[0098] 1、When is_blank() checks that there is missing data value between 2022 / 5 / 10 12:26:24 and 2022 / 5 / 10 12:26:25, and the data before and after is complete, is_blank() function can be used to repair data value.

[0099] 2、For single data value and complete data before and after, such as the wave height and wave direction value marked in orange in 2022 / 5 / 10 12:26:24, the average value of the corresponding values in the data before and after can be directly taken, such as wave height=(2.18+2.2) / 2=2.19, wave direction=(178.3+178.5) / 2=178.4.

[0100] 3、For the case of continuous missing of a certain data value but not more than three (the default value is three, which can be adjusted according to the situation), interpolation method is used to supplement, 2022 / 5 / 10 12:26:24 draft aft=9.31+(24-23) / (26-23)*(9.28-9.31)=9.30, 2022 / 5 / 10 12:26:25 draft aft=9.31+(25-23) / (26-23)*(9.28-9.31)=9.29

[0101] 4. The results are shown in Table 9 below:

[0102] Table 9

[0103]

[0104] The above embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the present application, several modifications and improvements can be made, which are within the protection scope of the present application.

Claims

1. A method for monitoring and managing the quality of energy efficiency data of intelligent ships, characterized in that: The following steps are involved: S1: Set up a data acquisition system to collect smart ship energy efficiency data and transmit it to the shore-side smart ship energy efficiency database; S2: Establish a data quality assessment method for smart ship energy efficiency data and conduct data quality monitoring: the quality assessment method is to write an assessment mechanism instruction for integrity, continuity and validity to assess whether the data meets the integrity, continuity and validity requirements. If the instructions are met, the data is valid; if not, the data is missing. The integrity includes detecting whether the entire data is missing through the is_continuous() function, detecting whether the attribute value is missing through the is_blank() function, and detecting whether the attribute value range is normal using the is_normal() function. The continuity includes whether the change of the continuous data sequence is abnormal, and the validity includes the proportion of normal data. The method for evaluating data integrity is: A1: Use the is_continuous() function to check whether any data is missing in UTC time order. Sort by time, and the time difference between the two data points is normal if it is 1 second; if it is greater than 1 second, the entire data point is missing. If no more than 30 data points are missing within 30 consecutive minutes, the data is valid. A2 uses the is_blank() function to check whether any attributes in each data entry are missing. Based on the attribute list, A2 checks each valid data entry to see if any attribute in the list is empty. If so, the data entry is considered abnormal. There is no null value, the attribute value is normal; A3: Use the is_normal() function to check whether each attribute in the data is within the normal range. This function checks whether the attribute values ​​in the valid data are within the given range. If they are within the range, it is normal; if they are outside the range, it is abnormal. The method for evaluating data continuity is to use the is_regular() function to check whether data attributes have undergone abnormal changes in the data sequence over a period of time. This includes checking whether the difference between the maximum and minimum values ​​of a single attribute in the valid data obtained by A3 over a period of time exceeds a limit. If the difference exceeds the limit, it indicates an anomaly. S3: Data quality management: Evaluate whether the missing data in S2 meets the quality management conditions, and repair the missing data in S2 and correct the abnormal data according to the missing or abnormal conditions in the quality management conditions; S4: Combine the ship energy consumption model and update the data quality assessment method in real time; use the valid data in S2 and the data after the data quality management in S3 as the input set, establish a ship energy consumption model using a multi-layer perceptron neural network, obtain the ship energy consumption result calculated from the input set using the get_params function, compare the matching degree of this result with the true value, feedback the matching degree information to S2, update the assessment mechanisms for integrity, continuity, and validity, use the updated assessment mechanisms to re-monitor the data quality in S1, perform quality management, and finally obtain an updated input set and input it into the ship energy consumption model until the matching degree of the ship energy consumption result and the true value reaches the matching degree threshold.

2. The method for monitoring and managing the quality of energy efficiency data of an intelligent ship according to claim 1, characterized in that: The energy efficiency data includes navigation status, rotational speed, wave direction, wind speed, forward draft, aft draft, wind direction, wave height, wave period, flow velocity, flow direction, and course angle.

3. The method for monitoring and managing the quality of energy efficiency data of an intelligent ship according to claim 1, characterized in that: The method for evaluating the validity of data in S2 is: obtain the number of original data records number and the number of data records amount determined to be normal after the assessment of data continuity; where ratio = amount / number. When ratio > 0.8, it indicates that the data validity is good, and write the ship energy efficiency data into the data set for intelligent ship modeling. When 0.4 < ratio < 0.8, it indicates that the data validity is relatively good, then import the data into the data management method for data补缺 (it seems there is a typo here, maybe "data filling" instead). When ratio < 0.4, it indicates that the data validity is poor, regard the data as invalid data, and eliminate the data.

4. The method for monitoring and managing the quality of energy efficiency data of an intelligent ship according to claim 1, characterized in that: For the missing data in S2, form a data quality problem feedback form and feedback it to the data acquisition system in S1 for optimization, including: eliminating the acquisition of ship energy efficiency data whose influence on the ship energy consumption result is lower than a certain threshold.

5. The method for monitoring and managing the quality of energy efficiency data of an intelligent ship according to claim 1, characterized in that: The method for repairing the missing data in S3 is: Use the fillup() function to repair the missing data: Based on the result of the is_continuous() function, in the case where the number of missing data within 1 minute does not exceed 3, fill the missing data with the average value of the attributes of the data before and after in time sequence. If it is continuous missing, perform linear interpolation according to time; for data that cannot be repaired, regard it as not meeting the data quality management conditions and eliminate the data.

6. The method for monitoring and managing the quality of energy efficiency data of an intelligent ship according to claim 1, characterized in that: The method for correcting abnormal data in S3 is: Use the pad() function to correct the abnormal data: For a single attribute, when there are valid attribute values in the 30 data records before and after, the result can be filled by interpolating according to time; for data that cannot be corrected, regard it as not meeting the data quality management conditions and eliminate the data.

Images