Energy conservation and emission reduction potential evaluation method based on incomplete energy consumption data

By dividing the time period and filling probability of energy consumption data, combined with multiple filling methods, the evaluation deviation caused by incomplete energy consumption data is solved, the accuracy and reliability of energy conservation and emission reduction potential assessment is improved, and scientificity is provided for enterprises. energy conservation and emission reduction strategies support.

CN119090133AInactive Publication Date: 2024-12-06LUZHOU VOCATIONAL & TECHN COLLEGE +1
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411121088.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-15
Publication Date
2024-12-06
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

In the prior art, due to incomplete energy consumption data, there are deviations in the evaluation results of energy conservation and emission reduction potential, and there is a lack of effective data integrity processing methods.

Method used

By dividing the energy usage data by time period, the data integrity is judged, and whether to fill is determined based on the filling probability of missing energy usage attribute values. The filling probability threshold is used as the judgment standard, combining the incomplete matrix calculation method and multiple filling methods (mean filling) Methods, thermal card filling method, KNN filling method, regression prediction method) perform data filling to ensure the completeness and accuracy of the evaluation data.

Benefits of technology

It improves the accuracy and reliability of the assessment of energy-saving and emission reduction potential, avoids distortion of evaluation results caused by data loss, ensures the quality and integrity of input data, and supports enterprises to formulate scientific energy-saving and emission reduction strategies.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119090133A_ABST
    Figure CN119090133A_ABST
Patent Text Reader

Abstract

The invention provides an energy conservation and emission reduction potential evaluation method based on incomplete energy consumption data, and relates to the technical field of energy conservation and emission reduction, and the method comprises the following steps: obtaining an energy consumption data set, and constructing an energy conservation and emission reduction potential evaluation model; dividing the energy consumption data set into energy consumption data subsets according to time periods; judging whether the energy consumption data is complete or not to obtain to-be-evaluated energy consumption data; traversing all energy consumption data in the energy consumption data subset to obtain a to-be-evaluated energy consumption data subset; judging whether the to-be-evaluated energy consumption data subset is complete or not to obtain the to-be-evaluated energy consumption data subset; traversing all the to-be-evaluated energy consumption data in the to-be-evaluated energy consumption data subset to obtain a complete to-be-evaluated energy consumption data subset; and inputting the complete to-be-evaluated energy consumption data subset into the energy conservation and emission reduction potential evaluation model according to time periods to obtain an evaluation result. The problem of inaccuracy of energy conservation and emission reduction potential evaluation when energy consumption data is incomplete is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of energy conservation and emission reduction, and in particular to a method for evaluating energy conservation and emission reduction potential based on incomplete energy consumption data. Background Art

[0002] In the current global context of energy conservation and carbon reduction, energy conservation and carbon reduction potential analysis plays a very important role. Through in-depth analysis and evaluation of energy consumption data, it can help high-energy-consuming enterprises and government agencies fully understand the potential for energy conservation and carbon reduction, identify potential energy waste and carbon emission improvement space, and formulate effective energy conservation and emission reduction strategies.

[0003] Due to differences in corporate management systems and statistical systems, some companies have problems such as inaccurate energy consumption data collected and discontinuous data records. Energy conservation and emission reduction potential assessment based on incomplete energy consumption data is likely to lead to biased assessment results. Summary of the invention

[0004] The purpose of the present invention is to provide a method for evaluating energy conservation and emission reduction potential based on incomplete energy consumption data. The technical problem to be solved is to ensure the accuracy of energy conservation and emission reduction potential evaluation when the energy consumption data is incomplete.

[0005] The present invention is achieved through the following technical solutions:

[0006] A method for evaluating energy conservation and emission reduction potential based on incomplete energy consumption data comprises the following steps:

[0007] S100, obtaining an energy consumption data set of a target enterprise; constructing an energy conservation and emission reduction potential assessment model based on the energy consumption data set of the target enterprise;

[0008] S200, dividing the energy consumption data set into a plurality of energy consumption data subsets according to time periods;

[0009] S300, determining whether each energy consumption data in the energy consumption data subset is complete;

[0010] S310, if complete, taking the energy consumption data as the energy consumption data to be evaluated;

[0011] S320: If incomplete, determine the probability of filling the missing energy attribute values ​​in the energy consumption data;

[0012] Calling the filling probability threshold, if the filling probability of the missing energy attribute value is greater than or equal to the filling probability threshold, then filling the missing energy attribute value;

[0013] If the probability of filling the missing energy attribute value is less than the filling probability threshold, the missing energy attribute value is ignored;

[0014] After traversing all energy usage attributes in the above energy usage data, the energy usage data to be evaluated is obtained;

[0015] S400, after traversing all the energy consumption data in the energy consumption data subset, obtaining the energy consumption data subset to be evaluated;

[0016] Determine whether the number of energy consumption data to be evaluated in the above-mentioned energy consumption data subset to be evaluated is complete;

[0017] S410: If complete, the energy consumption data subset to be evaluated is used as the complete energy consumption data subset to be evaluated;

[0018] S420, if incomplete, determining the probability of filling the missing energy consumption data to be evaluated in the above energy consumption data subset to be evaluated;

[0019] Calling the filling probability threshold, if the filling probability of the missing energy consumption data to be evaluated is greater than or equal to the filling probability threshold, then filling the missing energy consumption data to be evaluated;

[0020] If the probability of filling the missing energy consumption data to be evaluated is less than the filling probability threshold, the missing energy consumption data to be evaluated is ignored to obtain a subset of energy consumption data to be evaluated;

[0021] After traversing all the energy consumption data to be evaluated in the above energy consumption data subset to be evaluated, a complete energy consumption data subset to be evaluated is obtained;

[0022] S500, inputting the complete subset of energy consumption data to be evaluated into the energy conservation and emission reduction potential evaluation model in sequence according to the above time periods to obtain the energy conservation and emission reduction potential evaluation results of the target enterprise.

[0023] Dividing the energy consumption data set by time period helps to identify the energy consumption characteristics and potential differences in different time periods. The energy consumption data subsets are checked and processed by time period. For incomplete energy consumption data, it is not simply discarded, but the filling probability of the missing energy consumption attribute value is used to decide whether to fill it, reducing the evaluation bias caused by blind filling or deletion of data; the introduction of the filling probability threshold provides a clear judgment standard for whether to fill the missing value. When the filling probability is high, it is considered that the missing energy consumption attribute value has a high probability of being accurately estimated, so it is filled; otherwise, it is chosen to be ignored to avoid introducing unnecessary errors; on the basis of considering the integrity of energy consumption data, the integrity of the energy consumption data subset level is further considered, which is not only the processing of a single data point, but also the maintenance of the structure of the entire energy consumption data set; through a similar filling probability judgment mechanism, the missing energy consumption data at the energy consumption data subset level is processed to ensure that the energy consumption data subset used for the final evaluation is as complete and accurate as possible; it helps to avoid the distortion of the evaluation results caused by the missing energy consumption data at individual time nodes, and improves the stability and reliability of the evaluation; the processed complete energy consumption data subset to be evaluated is input into the energy conservation and emission reduction potential evaluation model for the final evaluation. Since the data was checked and processed in the early stage for integrity, the quality and integrity of the input data were ensured, thereby improving the accuracy of the evaluation results.

[0024] Furthermore, the incomplete matrix calculation method is used to determine the probability of filling the missing energy attribute values ​​in the above energy consumption data, including the following steps:

[0025] S321, obtaining all energy usage data and energy usage attributes in the energy usage data subset, using a value of 1 to represent a complete energy usage attribute value in an incomplete matrix, and a value of 0 to represent a missing energy usage attribute value;

[0026] S322, counting the number of complete energy usage attribute values ​​and the number of missing energy usage attribute values ​​under the same energy usage attribute;

[0027] S323. Calculate the missing rate of the corresponding energy attribute in the energy usage data subset by using the number of complete energy usage attribute values ​​and the number of missing energy usage attribute values ​​under the same energy usage attribute in the energy usage data subset. The formula is as follows:

[0028]

[0029] Among them, p i represents the missing rate of the i-th energy-using attribute; m i represents the number of missing energy attribute values ​​under the i-th energy attribute; n i represents the number of complete energy usage attribute values ​​under the i-th energy usage attribute;

[0030] S324. Calculate the missing entropy of each energy-using attribute in turn using the following formula:

[0031] H(i)=-p i log2p i -(1-p i )log2(1-p i );

[0032] Among them, H(i) represents the missing entropy of the i-th energy-using attribute;

[0033] S325. For any of the above missing energy usage attribute values, the probability of filling the missing energy usage value in the above energy usage data is calculated using the following formula:

[0034] P i (fill) = 1 - H(i);

[0035] Among them, P i (Fill) represents the probability of filling the i-th missing energy attribute.

[0036] All energy consumption data and energy consumption attributes in the above energy consumption data subset are converted into an incomplete matrix. The completeness of the energy consumption data is intuitively displayed in matrix form, the degree of missingness of each energy consumption attribute is quantified, and the uncertainty of the energy consumption attribute is measured by missing entropy. The larger the missing entropy, the higher the uncertainty of the energy consumption attribute.

[0037] Furthermore, the incomplete matrix calculation method is used to determine the probability of filling the missing energy consumption data to be evaluated in the above energy consumption data subset to be evaluated:

[0038] S421, obtaining the energy consumption data to be evaluated at each time node in the above-mentioned subset of energy consumption data to be evaluated, and using a value of 1 in the incomplete matrix to represent complete energy consumption data to be evaluated, and a value of 0 to represent missing energy consumption data to be evaluated;

[0039] S422, counting the number of complete energy consumption data to be evaluated and the number of missing energy consumption data to be evaluated in the above-mentioned subset of energy consumption data to be evaluated;

[0040] S423. Calculate the missing rate of the energy consumption data to be evaluated in the subset of energy consumption data to be evaluated by the number of complete energy consumption data to be evaluated and the number of missing energy consumption data to be evaluated in the subset of energy consumption data to be evaluated. The formula is as follows:

[0041]

[0042] Among them, f T M represents the missing rate of energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T; T Indicates the number of missing energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T; N TIndicates the number of complete energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T;

[0043] S424, using the following formulas to calculate the missing entropy of the energy consumption data to be evaluated in the above-mentioned energy consumption data subset to be evaluated:

[0044] H(T)=-f T log2f T -(1-f T )log2(1-f T );

[0045] Wherein, H(T) represents the missing entropy of the energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T;

[0046] S425. For any of the above missing energy consumption data to be evaluated, the probability of filling the missing energy consumption data to be evaluated in the above energy consumption data subset to be evaluated is calculated using the following formula:

[0047] F T (fill) = 1-H(T);

[0048] Among them, F T (Filling) represents the probability of filling in the missing energy consumption data to be evaluated in the subset of energy consumption data to be evaluated in time period T.

[0049] Each time node and the energy consumption data to be evaluated in the above-mentioned energy consumption data subset to be evaluated are converted into an incomplete matrix. The integrity of the energy consumption data to be evaluated is intuitively displayed in the form of a matrix. The degree of missingness of each energy consumption data to be evaluated in the energy consumption data subset to be evaluated is quantified. The missing entropy is used to measure the uncertainty of the energy consumption data to be evaluated at each time node. The larger the missing entropy, the higher the uncertainty of the energy consumption data to be evaluated at the time node.

[0050] Furthermore, the specific steps of filling the missing energy attribute values ​​include:

[0051] S326, obtaining complete energy usage data from the energy usage data subset, and using the complete energy usage data as training data;

[0052] S327. Randomly select an energy usage attribute value from the above training data as the missing energy usage attribute value, and use the mean filling method, hot card filling method, KNN filling method and regression prediction method to fill the missing energy usage attribute value, and obtain the mean filling value, hot card filling value, KNN filling value and regression filling value;

[0053] S328, respectively calculating the absolute differences between the mean filling value, the heat card filling value, the KNN filling value, the regression filling value and the missing energy attribute value, and obtaining the mean filling error, the heat card filling error, the KNN filling error, and the regression filling error;

[0054] S329, by comparing the above mean filling error, hot card filling error, KNN filling error, and regression filling error, the maximum error and the minimum error are obtained;

[0055] S330, respectively calculating the weights of the mean filling method, the hot card filling method, the KNN filling method and the regression prediction method through the above mean filling error, the hot card filling error, the KNN filling error, the regression filling error and the maximum error;

[0056] S331, using the above-mentioned mean filling method, hot card filling method, KNN filling method and regression prediction method to fill in the missing energy attribute values ​​in the energy consumption data, and obtain the mean filling estimate value, the hot card filling estimate value, the KNN filling estimate value and the regression filling estimate value;

[0057] S332. Determine the filling data for missing energy attribute values ​​in the energy consumption data through the weights of the above-mentioned mean filling method, hot card filling method, KNN filling method and regression prediction method as well as the mean filling estimate value, hot card filling estimate value, KNN filling estimate value and regression filling estimate value.

[0058] In the above training data, the situation of missing energy attribute values ​​is simulated, and an energy attribute value is selected as the "missing energy attribute value". Then, four filling methods are used to fill it, and the effects of different filling methods are compared. A weight is assigned to each filling method (the method with smaller error should get a higher weight). The four filling methods are used to fill the actual missing energy attribute values, and four different filling estimates are obtained. According to the weights and filling estimates of the four filling methods, the final filling data is determined by weighted averaging, which combines the advantages of different filling methods to make the obtained filling data more accurate.

[0059] Furthermore, the absolute differences between the above mean filling value, heat card filling value, KNN filling value, regression filling value and the missing energy attribute value are calculated respectively to obtain the mean filling error, heat card filling error, KNN filling error and regression filling error. The specific formulas are as follows:

[0060]

[0061] Among them, d 均值 ,d 热卡 ,d KNN and d 回归 Respectively represent mean filling error, hot card filling error, KNN filling error, and regression filling error; a' 均值 、a' 热卡 、a' KNN and a' 回归They represent mean filling value, hot card filling value, KNN filling value and regression filling value respectively; A' represents the missing energy attribute value selected in the training data.

[0062] Furthermore, the weights of the mean filling method, hot card filling method, KNN filling method and regression prediction method are calculated respectively through the above mean filling error, hot card filling error, KNN filling error, regression filling error and maximum error. The formula is as follows:

[0063]

[0064] Among them, w 均值 、w 热卡 、w KNN and w 回归 Represent the weights of mean filling method, hot card filling method, KNN filling method and regression prediction method respectively; d max Indicates the maximum error.

[0065] Furthermore, the filling data of the missing energy attribute values ​​in the energy consumption data are determined by the weights of the above-mentioned mean filling method, hot card filling method, KNN filling method and regression prediction method, as well as the mean filling estimate value, hot card filling estimate value, KNN filling estimate value and regression filling estimate value. The formula is as follows:

[0066] A=a 均值 w 均值 +a 热卡 w 热卡 +a KNN w KNN +a 回归 w 回归 ;

[0067] Among them, A represents the filling data of missing energy attribute values ​​in energy consumption data; a 均值 represents the mean filling estimate; a 热卡 Indicates the estimated value of hot card filling; a KNN Indicates KNN fills in the estimated value; a 回归 Represents the regression fill estimate.

[0068] Furthermore, the specific steps to fill the missing energy consumption data to be evaluated include:

[0069] S426, obtaining the energy consumption data to be evaluated at adjacent time nodes where the energy consumption data to be evaluated is missing from the above-mentioned subset of energy consumption data to be evaluated;

[0070] S427: when the energy consumption data to be evaluated at the adjacent time nodes include the energy consumption data to be evaluated at the previous time node and the energy consumption data to be evaluated at the next time node, taking the average value of the energy consumption data to be evaluated at the previous time node and the energy consumption data to be evaluated at the next time node as filling data for the missing energy consumption data to be evaluated;

[0071] S428. When the energy consumption data to be evaluated at the adjacent time nodes is only the energy consumption data to be evaluated at the previous time node, the energy consumption data to be evaluated at the previous time node is used as filling data for the missing energy consumption data to be evaluated;

[0072] S429: When the energy consumption data to be evaluated of the adjacent time nodes only include the energy consumption data to be evaluated of the next time node, the energy consumption data to be evaluated of the next time node is used as filling data for the missing energy consumption data to be evaluated.

[0073] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0074] Dividing the energy consumption data set by time period helps to identify the energy consumption characteristics and potential differences in different time periods. The energy consumption data subsets are checked and processed by time period. For incomplete energy consumption data, it is not simply discarded, but the filling probability of the missing energy consumption attribute value is used to decide whether to fill it, reducing the evaluation bias caused by blind filling or deletion of data; the introduction of the filling probability threshold provides a clear judgment standard for whether to fill the missing value. When the filling probability is high, it is considered that the missing energy consumption attribute value has a high probability of being accurately estimated, so it is filled; otherwise, it is chosen to be ignored to avoid introducing unnecessary errors; on the basis of considering the integrity of energy consumption data, the integrity of the energy consumption data subset level is further considered, which is not only the processing of a single data point, but also the maintenance of the structure of the entire energy consumption data set; through a similar filling probability judgment mechanism, the missing energy consumption data at the energy consumption data subset level is processed to ensure that the energy consumption data subset used for the final evaluation is as complete and accurate as possible; it helps to avoid the distortion of the evaluation results caused by the missing energy consumption data at individual time nodes, and improves the stability and reliability of the evaluation; the processed complete energy consumption data subset to be evaluated is input into the energy conservation and emission reduction potential evaluation model for the final evaluation. Since the data was checked and processed in the early stage for integrity, the quality and integrity of the input data were ensured, thereby improving the accuracy of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:

[0076] Figure 1 For the flow chart. DETAILED DESCRIPTION

[0077] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0078] Example 1

[0079] This embodiment 1 provides a method for evaluating energy conservation and emission reduction potential based on incomplete energy consumption data, including the following steps:

[0080] S100, obtaining an energy consumption data set from energy consumption monitoring data collected from the target enterprise; constructing an energy conservation and emission reduction potential assessment model based on the energy consumption data set of the target enterprise;

[0081] S200, dividing the energy consumption data set into a plurality of energy consumption data subsets according to time periods; the time period is a continuous time series, which may be a day, a week, a month or a year, and a time period includes a plurality of time nodes, and the intervals between adjacent time nodes are equal in length;

[0082] S300, determining whether each energy consumption data in the energy consumption data subset is complete;

[0083] S310, if complete, taking the energy consumption data as the energy consumption data to be evaluated;

[0084] S320, if incomplete, determine the probability of filling the missing energy attribute value in the energy usage data; the energy attribute is the type of energy consumed, including electricity, natural gas, diesel, gasoline and nitrogen; the energy attribute value is the consumption of the corresponding energy type;

[0085] Calling the filling probability threshold, if the filling probability of the missing energy attribute value is greater than or equal to the filling probability threshold, then filling the missing energy attribute value;

[0086] If the probability of filling the missing energy attribute value is less than the filling probability threshold, the missing energy attribute value is ignored;

[0087] After traversing all energy usage attributes in the above energy usage data, the energy usage data to be evaluated is obtained;

[0088] S400, after traversing all the energy consumption data in the energy consumption data subset, obtaining the energy consumption data subset to be evaluated;

[0089] Determine whether the number of energy consumption data to be evaluated in the above-mentioned energy consumption data subset to be evaluated is complete;

[0090] S410: If complete, the energy consumption data subset to be evaluated is used as the complete energy consumption data subset to be evaluated;

[0091] S420, if incomplete, determining the probability of filling the missing energy consumption data to be evaluated in the above energy consumption data subset to be evaluated;

[0092] Calling the filling probability threshold, if the filling probability of the missing energy consumption data to be evaluated is greater than or equal to the filling probability threshold, then filling the missing energy consumption data to be evaluated;

[0093] If the probability of filling the missing energy consumption data to be evaluated is less than the filling probability threshold, the missing energy consumption data to be evaluated is ignored to obtain a subset of energy consumption data to be evaluated;

[0094] After traversing all the energy consumption data to be evaluated in the above energy consumption data subset to be evaluated, a complete energy consumption data subset to be evaluated is obtained;

[0095] S500, inputting the complete subset of energy consumption data to be evaluated into the energy conservation and emission reduction potential evaluation model in sequence according to the above time periods to obtain the energy conservation and emission reduction potential evaluation results of the target enterprise.

[0096] Dividing the energy consumption data set by time period helps to identify the energy consumption characteristics and potential differences in different time periods. The energy consumption data subsets are checked and processed by time period. For incomplete energy consumption data, it is not simply discarded, but the filling probability of the missing energy consumption attribute value is used to decide whether to fill it, reducing the evaluation bias caused by blind filling or deletion of data; the introduction of the filling probability threshold provides a clear judgment standard for whether to fill the missing value. When the filling probability is high, it is considered that the missing energy consumption attribute value has a high probability of being accurately estimated, so it is filled; otherwise, it is chosen to be ignored to avoid introducing unnecessary errors; on the basis of considering the integrity of energy consumption data, the integrity of the energy consumption data subset level is further considered, which is not only the processing of a single data point, but also the maintenance of the structure of the entire energy consumption data set; through a similar filling probability judgment mechanism, the missing energy consumption data at the energy consumption data subset level is processed to ensure that the energy consumption data subset used for the final evaluation is as complete and accurate as possible; it helps to avoid the distortion of the evaluation results caused by the missing energy consumption data at individual time nodes, and improves the stability and reliability of the evaluation; the processed complete energy consumption data subset to be evaluated is input into the energy conservation and emission reduction potential evaluation model for the final evaluation. Since the data was checked and processed in the early stage for integrity, the quality and integrity of the input data were ensured, thereby improving the accuracy of the evaluation results.

[0097] In addition, the complete subset of energy consumption data to be evaluated is input in sequence by time period, and the impact of time factors on energy consumption and energy conservation and emission reduction potential is also taken into account, making the evaluation results more in line with actual conditions.

[0098] In summary, this method effectively solves the problem of incomplete energy consumption data through data preprocessing, probability-based missing value processing mechanism and subset-level integrity assessment, and significantly improves the accuracy and reliability of energy conservation and emission reduction potential assessment, which is of great significance for enterprises to formulate scientific energy conservation and emission reduction strategies and optimize energy management.

[0099] Example 2

[0100] On the basis of Example 1, the incomplete matrix calculation method is used to determine the probability of filling the missing energy attribute values ​​in the above energy consumption data, including the following steps:

[0101] S321, obtaining all energy usage data and energy usage attributes in the energy usage data subset, using a value of 1 to represent a complete energy usage attribute value in an incomplete matrix, and a value of 0 to represent a missing energy usage attribute value;

[0102] S322, counting the number of complete energy usage attribute values ​​and the number of missing energy usage attribute values ​​under the same energy usage attribute;

[0103] S323. Calculate the missing rate of the corresponding energy attribute in the energy usage data subset by using the number of complete energy usage attribute values ​​and the number of missing energy usage attribute values ​​under the same energy usage attribute in the energy usage data subset. The formula is as follows:

[0104]

[0105] Among them, p i represents the missing rate of the i-th energy-using attribute; m i represents the number of missing energy attribute values ​​under the i-th energy attribute; n i represents the number of complete energy usage attribute values ​​under the i-th energy usage attribute;

[0106] S324. Calculate the missing entropy of each energy-using attribute in turn using the following formula:

[0107] H(i)=-p i log2p i -(1-p i )log2(1-p i );

[0108] Among them, H(i) represents the missing entropy of the i-th energy-using attribute;

[0109] S325. For any of the above missing energy usage attribute values, the probability of filling the missing energy usage value in the above energy usage data is calculated using the following formula:

[0110] P i (fill) = 1 - H(i);

[0111] Among them, P i (Fill) represents the probability of filling the i-th missing energy attribute.

[0112] All energy consumption data and energy consumption attributes in the above energy consumption data subset are converted into an incomplete matrix. The completeness of the energy consumption data is intuitively displayed in matrix form, the degree of missingness of each energy consumption attribute is quantified, and the uncertainty of the energy consumption attribute is measured by missing entropy. The larger the missing entropy, the higher the uncertainty of the energy consumption attribute.

[0113] Example 1: The target enterprise produces aluminum materials. During the production of aluminum materials, energy consumption data is monitored. The energy sources include electricity, natural gas, diesel, gasoline, and nitrogen. The monitored energy data are collected as energy consumption data sets. The energy consumption data sets are divided into several energy consumption data subsets by week. All energy consumption data and energy consumption attributes in the energy consumption data subsets of time period T are converted into a matrix, as shown in Table 1.

[0114] Table 1

[0115] electricity natural gas diesel fuel gasoline Nitrogen Energy consumption data1 1 1 1 1 1 Energy consumption data 2 1 0 1 1 1 Energy consumption data 3 1 1 0 1 1 Energy consumption data 4 1 1 1 1 0 Energy consumption data 5 1 0 1 1 1 Energy consumption data6 1 1 1 1 0

[0116] According to Table 1, energy consumption data 1 is complete energy consumption data. The following explanation is given by taking the probability of filling in the missing energy consumption attribute (natural gas) value in energy consumption data 2 as an example:

[0117] use Calculate the missing rate of natural gas in the energy consumption data subset, which is use Calculate the missing entropy of natural gas; use P 天然气 (Filling)=1-H(Natural Gas) calculates the filling probability of natural gas in energy consumption data 2.

[0118] Example 3

[0119] On the basis of any of the above embodiments, the probability of filling the missing energy consumption data to be evaluated in the above energy consumption data subset to be evaluated is determined by using the incomplete matrix calculation method:

[0120] S421, obtaining the energy consumption data to be evaluated at each time node in the above-mentioned subset of energy consumption data to be evaluated, and using a value of 1 in the incomplete matrix to represent complete energy consumption data to be evaluated, and a value of 0 to represent missing energy consumption data to be evaluated;

[0121] S422, counting the number of complete energy consumption data to be evaluated and the number of missing energy consumption data to be evaluated in the above-mentioned subset of energy consumption data to be evaluated;

[0122] S423. Calculate the missing rate of the energy consumption data to be evaluated in the subset of energy consumption data to be evaluated by the number of complete energy consumption data to be evaluated and the number of missing energy consumption data to be evaluated in the subset of energy consumption data to be evaluated. The formula is as follows:

[0123]

[0124] Among them, f T M represents the missing rate of energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T; T Indicates the number of missing energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T; N T Indicates the number of complete energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T;

[0125] S424, using the following formulas to calculate the missing entropy of the energy consumption data to be evaluated in the above-mentioned energy consumption data subset to be evaluated:

[0126] H(T)=-f T log2f T -(1-f T )log2(1-f T );

[0127] Wherein, H(T) represents the missing entropy of the energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T;

[0128] S425. For any of the above missing energy consumption data to be evaluated, the probability of filling the missing energy consumption data to be evaluated in the above energy consumption data subset to be evaluated is calculated using the following formula:

[0129] F T (fill) = 1-H(T);

[0130] Among them, F T (Filling) represents the probability of filling in the missing energy consumption data to be evaluated in the subset of energy consumption data to be evaluated in time period T.

[0131] Each time node and the energy consumption data to be evaluated in the above-mentioned energy consumption data subset to be evaluated are converted into an incomplete matrix. The integrity of the energy consumption data to be evaluated is intuitively displayed in the form of a matrix. The degree of missingness of each energy consumption data to be evaluated in the energy consumption data subset to be evaluated is quantified. The missing entropy is used to measure the uncertainty of the energy consumption data to be evaluated at each time node. The larger the missing entropy, the higher the uncertainty of the energy consumption data to be evaluated at the time node.

[0132] Example 2: The energy consumption data to be evaluated at each time node (the time node is in days) in the energy consumption data subset to be evaluated in time period T is converted into a matrix, as shown in Table 2;

[0133] Table 2

[0134] on Monday Tuesday Wednesday Thursday Friday Saturday Saturday Energy consumption data to be evaluated 1 1 0 1 1 0 1

[0135] use Calculate the missing rate of energy consumption data to be evaluated in the energy consumption data subset to be evaluated. The missing rate is use Calculate the missing entropy of the energy consumption data to be evaluated in the energy consumption data subset to be evaluated; use F T (Filling) = 1-H(T) calculates the probability of filling the missing energy consumption data to be evaluated on Wednesday in the energy consumption data subset to be evaluated.

[0136] Example 4

[0137] Based on any of the above embodiments, the specific steps of filling the missing energy attribute value include:

[0138] S326, obtaining complete energy usage data from the energy usage data subset, and using the complete energy usage data as training data;

[0139] S327. Randomly select an energy usage attribute value from the above training data as the missing energy usage attribute value, and use the mean filling method, hot card filling method, KNN filling method and regression prediction method to fill the missing energy usage attribute value, and obtain the mean filling value, hot card filling value, KNN filling value and regression filling value;

[0140] S328, respectively calculating the absolute differences between the mean filling value, the heat card filling value, the KNN filling value, the regression filling value and the missing energy attribute value, and obtaining the mean filling error, the heat card filling error, the KNN filling error, and the regression filling error;

[0141] S329, by comparing the above mean filling error, hot card filling error, KNN filling error, and regression filling error, the maximum error and the minimum error are obtained;

[0142] S330, respectively calculating the weights of the mean filling method, the hot card filling method, the KNN filling method and the regression prediction method through the above mean filling error, the hot card filling error, the KNN filling error, the regression filling error and the maximum error;

[0143] S331, using the above-mentioned mean filling method, hot card filling method, KNN filling method and regression prediction method to fill in the missing energy attribute values ​​in the energy consumption data, and obtain the mean filling estimate value, the hot card filling estimate value, the KNN filling estimate value and the regression filling estimate value;

[0144] S332. Determine the filling data for missing energy attribute values ​​in the energy consumption data through the weights of the above-mentioned mean filling method, hot card filling method, KNN filling method and regression prediction method as well as the mean filling estimate value, hot card filling estimate value, KNN filling estimate value and regression filling estimate value.

[0145] In the above training data, the situation of missing energy attribute values ​​is simulated, and an energy attribute value is selected as the "missing energy attribute value". Then, four filling methods are used to fill it, and the effects of different filling methods are compared. A weight is assigned to each filling method (the method with smaller error should get a higher weight). The four filling methods are used to fill the actual missing energy attribute values, and four different filling estimates are obtained. According to the weights and filling estimates of the four filling methods, the final filling data is determined by weighted averaging, which combines the advantages of different filling methods to make the obtained filling data more accurate.

[0146] In a specific embodiment, the absolute differences between the mean filling value, the heat filling value, the KNN filling value, the regression filling value and the missing energy attribute value are calculated respectively to obtain the mean filling error, the heat filling error, the KNN filling error and the regression filling error. The specific formulas are as follows:

[0147]

[0148] Among them, d 均值 ,d 热卡 ,d KNN and d 回归 Respectively represent mean filling error, hot card filling error, KNN filling error, and regression filling error; a' 均值 、a' 热卡 、a' KNN and a' 回归 They represent mean filling value, hot card filling value, KNN filling value and regression filling value respectively; A' represents the missing energy attribute value selected in the training data.

[0149] In a specific embodiment, the weights of the mean filling method, the hot card filling method, the KNN filling method and the regression prediction method are calculated respectively by using the above-mentioned mean filling error, hot card filling error, KNN filling error, regression filling error and maximum error. The formula is as follows:

[0150]

[0151] Among them, w 均值 、w 热卡 、w KNN and w 回归 Represent the weights of mean filling method, hot card filling method, KNN filling method and regression prediction method respectively; d max Indicates the maximum error.

[0152] In a specific embodiment, the filling data of missing energy attribute values ​​in the energy consumption data are determined by using the weights of the above-mentioned mean filling method, hot card filling method, KNN filling method and regression prediction method and the mean filling estimate value, hot card filling estimate value, KNN filling estimate value and regression filling estimate value. The formula is as follows:

[0153] A=a 均值 w 均值 +a 热卡 w 热卡 +a KNN w KNN +a 回归 w 回归 ;

[0154] Among them, A represents the filling data of missing energy attribute values ​​in energy consumption data; a 均值 represents the mean filling estimate; a 热卡 Indicates the estimated value of hot card filling; a KNN Indicates KNN fills in the estimated value; a 回归 Represents the regression fill estimate.

[0155] By combining the advantages of different imputation methods and balancing their impacts according to their respective weights, a more accurate and reliable imputation data is obtained.

[0156] In a specific embodiment, the specific steps of filling the above missing energy consumption data to be evaluated include:

[0157] S426, obtaining the energy consumption data to be evaluated at adjacent time nodes where the energy consumption data to be evaluated is missing from the above-mentioned subset of energy consumption data to be evaluated;

[0158] S427, based on the smoothness of the energy consumption data to be evaluated in the time series, the energy consumption data to be evaluated at adjacent time nodes will not change too much; when the energy consumption data to be evaluated at the adjacent time nodes include the energy consumption data to be evaluated at the previous time node and the energy consumption data to be evaluated at the next time node, the average value of the energy consumption data to be evaluated at the previous time node and the energy consumption data to be evaluated at the next time node is taken as the filling data for the missing energy consumption data to be evaluated; (for example: the missing energy consumption data to be evaluated is in the middle of the time series);

[0159] S428. Based on the continuity of the data in the time series, it is considered that the energy consumption data to be evaluated does not change much in a short period of time; when the energy consumption data to be evaluated at the adjacent time nodes is only the energy consumption data to be evaluated at the previous time node, the energy consumption data to be evaluated at the previous time node is used as the filling data for the missing energy consumption data to be evaluated; (for example, the missing energy consumption data to be evaluated is at the end of the time series);

[0160] S429. When the energy consumption data to be evaluated of the adjacent time nodes mentioned above only include the energy consumption data to be evaluated of the next time node, the energy consumption data to be evaluated of the next time node is used as filling data for the missing energy consumption data to be evaluated; (for example, the missing energy consumption data to be evaluated is at the beginning of the time series).

[0161] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for evaluating energy conservation and emission reduction potential based on incomplete energy consumption data. It is characterized in that The following steps are involved: S100, obtaining an energy consumption data set of a target enterprise; constructing an energy conservation and emission reduction potential assessment model based on the energy consumption data set of the target enterprise; S200, dividing the energy consumption data set into a plurality of energy consumption data subsets according to time periods; S300, determining whether each energy consumption data in the energy consumption data subset is complete; S310: If complete, use the energy consumption data as energy consumption data to be evaluated; S320: If incomplete, determine the probability of filling the missing energy usage attribute values ​​in the energy usage data; Calling a filling probability threshold, if the filling probability of the missing energy attribute value is greater than or equal to the filling probability threshold, filling the missing energy attribute value; If the probability of filling the missing energy attribute value is less than the filling probability threshold, the missing energy attribute value is ignored; After traversing all energy usage attributes in the energy usage data, the energy usage data to be evaluated is obtained; S400, after traversing all energy consumption data in the energy consumption data subset, obtaining an energy consumption data subset to be evaluated; Determining whether the number of energy consumption data to be evaluated in the energy consumption data subset to be evaluated is complete; S410: If complete, taking the energy consumption data subset to be evaluated as the complete energy consumption data subset to be evaluated; S420: If incomplete, determine the probability of filling the missing energy consumption data to be evaluated in the energy consumption data subset to be evaluated; Calling a filling probability threshold, if the filling probability of the missing energy consumption data to be evaluated is greater than or equal to the filling probability threshold, filling the missing energy consumption data to be evaluated; If the probability of filling the missing energy consumption data to be evaluated is less than the filling probability threshold, the missing energy consumption data to be evaluated is ignored to obtain a subset of energy consumption data to be evaluated; After traversing all the energy consumption data to be evaluated in the energy consumption data subset to be evaluated, a complete energy consumption data subset to be evaluated is obtained; S500, inputting the complete subset of energy consumption data to be evaluated into the energy conservation and emission reduction potential evaluation model in sequence according to the time period to obtain the energy conservation and emission reduction potential evaluation result of the target enterprise.

2. According to claim 1, a method for evaluating energy conservation and emission reduction potential based on incomplete energy consumption data, It is characterized in that Determining the probability of filling missing energy attribute values ​​in the energy consumption data by using an incomplete matrix calculation method includes the following steps: S321, obtaining all energy usage data and energy usage attributes in the energy usage data subset, using a value of 1 to represent a complete energy usage attribute value in an incomplete matrix, and a value of 0 to represent a missing energy usage attribute value; S322, counting the number of complete energy usage attribute values ​​and the number of missing energy usage attribute values ​​under the same energy usage attribute; S323. Calculate the missing rate of the corresponding energy attribute in the energy usage data subset by the number of complete energy usage attribute values ​​and the number of missing energy usage attribute values ​​under the same energy usage attribute in the energy usage data subset. The formula is as follows: Among them, p i represents the missing rate of the i-th energy-using attribute; m i represents the number of missing energy attribute values ​​under the i-th energy attribute; n i represents the number of complete energy usage attribute values ​​under the i-th energy usage attribute; S324. Calculate the missing entropy of each energy-using attribute in turn using the following formula: H(i)=-p i log 2 p i -(1-p i )log 2 (1-p i ); Among them, H(i) represents the missing entropy of the i-th energy-using attribute; S325. For any missing energy usage attribute value, the probability of filling the missing energy usage value in the energy usage data is calculated using the following formula: P i (fill) = 1 - H(i); Among them, P i (Fill) represents the probability of filling the i-th missing energy attribute.

3. According to claim 1, a method for evaluating energy conservation and emission reduction potential based on incomplete energy consumption data, It is characterized in that The incomplete matrix calculation method is used to determine the probability of filling the missing energy consumption data to be evaluated in the energy consumption data subset to be evaluated: S421, obtaining the energy consumption data to be evaluated at each time node in the subset of energy consumption data to be evaluated, and using a value of 1 in the incomplete matrix to represent complete energy consumption data to be evaluated, and a value of 0 to represent missing energy consumption data to be evaluated; S422, counting the number of complete energy consumption data to be evaluated and the number of missing energy consumption data to be evaluated in the subset of energy consumption data to be evaluated; S423, calculating the missing rate of the energy consumption data to be evaluated in the subset of energy consumption data to be evaluated by the number of complete energy consumption data to be evaluated and the number of missing energy consumption data to be evaluated in the subset of energy consumption data to be evaluated, and the formula is as follows: Among them, f T M represents the missing rate of energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T; T Indicates the number of missing energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T; N T Indicates the number of complete energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T; S424, using the following formulas to calculate the missing entropy of the energy consumption data to be evaluated in the energy consumption data subset to be evaluated: H(T)=-f T log 2 f T -(1-f T )log 2 (1-f T ); Wherein, H(T) represents the missing entropy of the energy consumption data to be evaluated in the energy consumption data subset to be evaluated in time period T; S425. For any missing energy consumption data to be evaluated, the probability of filling the missing energy consumption data to be evaluated in the energy consumption data subset to be evaluated is calculated using the following formula: F T (fill) = 1-H(T); Among them, F T (Filling) represents the probability of filling in the missing energy consumption data to be evaluated in the subset of energy consumption data to be evaluated in time period T.

4. According to claim 1, a method for evaluating energy conservation and emission reduction potential based on incomplete energy consumption data, It is characterized in that The specific steps of filling the missing energy attribute value include: S326, acquiring complete energy usage data from the energy usage data subset, and using the complete energy usage data as training data; S327, arbitrarily selecting an energy usage attribute value in the training data as the missing energy usage attribute value, and respectively using the mean filling method, the hot card filling method, the KNN filling method and the regression prediction method to fill the missing energy usage attribute value, to obtain the mean filling value, the hot card filling value, the KNN filling value and the regression filling value; S328, respectively calculating the absolute differences between the mean filling value, the heat filling value, the KNN filling value, the regression filling value and the missing energy attribute value, and obtaining the mean filling error, the heat filling error, the KNN filling error, and the regression filling error; S329, obtaining a maximum error and a minimum error by comparing the mean filling error, the hot card filling error, the KNN filling error, and the regression filling error; S330, respectively calculating the weights of the mean filling method, the hot card filling method, the KNN filling method and the regression prediction method through the mean filling error, the hot card filling error, the KNN filling error, the regression filling error and the maximum error; S331, using the mean filling method, hot card filling method, KNN filling method and regression prediction method to fill in the missing energy attribute values ​​in the energy consumption data, and obtain the mean filling estimate value, the hot card filling estimate value, the KNN filling estimate value and the regression filling estimate value; S332. Determine the filling data for missing energy attribute values ​​in the energy usage data by using the weights of the mean filling method, hot card filling method, KNN filling method and regression prediction method as well as the mean filling estimate value, hot card filling estimate value, KNN filling estimate value and regression filling estimate value.

5. A method for evaluating energy conservation and emission reduction potential based on incomplete energy consumption data according to claim 4, It is characterized in that The absolute differences between the mean filling value, the heat filling value, the KNN filling value, the regression filling value and the missing energy attribute value are calculated respectively to obtain the mean filling error, the heat filling error, the KNN filling error and the regression filling error. The specific formulas are as follows: Among them, d 均值 d 热卡 d KNN and d 回归 Respectively represent mean filling error, hot card filling error, KNN filling error, and regression filling error; a' 均值 、a' 热卡 、a' KNN and a' 回归 They represent mean filling value, hot card filling value, KNN filling value and regression filling value respectively; A' represents the missing energy attribute value selected in the training data.

6. A method for evaluating energy conservation and emission reduction potential based on incomplete energy consumption data according to claim 5, It is characterized in that The weights of the mean filling method, the hot card filling method, the KNN filling method and the regression prediction method are calculated respectively by the mean filling error, the hot card filling error, the KNN filling error, the regression filling error and the maximum error. The formulas are as follows: Among them, w 均值 、w 热卡 、w KNN and w 回归 Represent the weights of mean filling method, hot card filling method, KNN filling method and regression prediction method respectively; d max Indicates the maximum error.

7. A method for evaluating energy conservation and emission reduction potential based on incomplete energy consumption data according to claim 6, It is characterized in that The filling data of missing energy attribute values ​​in the energy consumption data are determined by using the weights of the mean filling method, hot card filling method, KNN filling method and regression prediction method and the mean filling estimate value, hot card filling estimate value, KNN filling estimate value and regression filling estimate value. The formula is as follows: A=a 均值 w 均值 +a 热卡 w 热卡 +a KNN w KNN +a 回归 w 回归 ; Among them, A represents the filling data of missing energy attribute values ​​in energy consumption data; a 均值 represents the mean filling estimate; a 热卡 Indicates the estimated value of hot card filling; a KNN Indicates KNN fills in the estimated value; a 回归 Represents the regression fill estimate.

8. The energy saving and emission reduction potential assessment method based on incomplete energy consumption data according to claim 1, It is characterized in that The specific steps to fill the missing energy consumption data to be evaluated include: S426, acquiring the energy consumption data to be evaluated at adjacent time nodes where the energy consumption data to be evaluated is missing from the subset of energy consumption data to be evaluated; S427: when the energy consumption data to be evaluated at the adjacent time nodes include the energy consumption data to be evaluated at the previous time node and the energy consumption data to be evaluated at the next time node, taking the average value of the energy consumption data to be evaluated at the previous time node and the energy consumption data to be evaluated at the next time node as filling data for the missing energy consumption data to be evaluated; S428. When the energy consumption data to be evaluated at the adjacent time nodes only include the energy consumption data to be evaluated at the previous time node, the energy consumption data to be evaluated at the previous time node is used as filling data for the missing energy consumption data to be evaluated; S429: When the energy consumption data to be evaluated at the adjacent time nodes only include the energy consumption data to be evaluated at the next time node, the energy consumption data to be evaluated at the next time node is used as filling data for the missing energy consumption data to be evaluated.

Citation Information

Patent Citations

  • High-dimension incomplete data feature selection method

    CN107220346A

  • Dam safety monitoring data complementation method based on optimal mixed estimated value

    CN113139570A

  • Electric quantity data missing value filling method and system

    CN114519379A

  • Electrification potential evaluation method, device and equipment compatible with incomplete energy consumption data

    CN117350595A

  • Energy data processing evaluation management system and method

    CN118364337A