A method and terminal for studying low voltage performance characteristics

Abstract

This invention discloses a method and terminal for studying the characteristics of low-voltage indicators. The method receives indicators related to low voltage to be analyzed; searches a dataset to find all items containing the indicators, forming an itemset; determines the indicator data corresponding to each indicator in the itemset; and splits the indicator data corresponding to each indicator in the itemset so that the ranges of any two indicator data corresponding to each indicator do not overlap, forming a split itemset; and uses an improved Apriori algorithm to determine indicators strongly correlated with low voltage based on the split itemset. In the improved Apriori algorithm, all candidate itemsets are first determined based on the split itemset, and then support is calculated for all candidate itemsets in a single scan of the split itemset. The accuracy and efficiency of low-voltage indicator correlation analysis are ensured through data splitting and a reduction in the number of adaptive scans.

Description

Technical Field

[0001] This invention relates to the field of power distribution network data analysis, and in particular to a method and terminal for studying low voltage index characteristics. Background Technology

[0002] With the rapid development of science and technology, new-generation information technologies such as cloud computing, big data, the Internet of Things, and artificial intelligence are being promoted and applied more and more widely and deeply. Among them, big data analysis plays an important role in all walks of life. In power distribution networks, big data analysis can serve the planning work of power distribution networks. By analyzing and applying massive amounts of data in power distribution network planning, in-depth applications of planning data can be achieved, thereby improving the level of power grid planning and providing strong support for power distribution network planning and construction.

[0003] In distribution network planning, a common approach is to use data mining analysis algorithms to analyze the characteristics and relationships of different indicators within the distribution network, based on collected and analyzed current distribution network data. This process uncovers the correlations and influence levels between these indicators, providing fundamental support for distribution network planning. However, existing data mining analysis models often lack sufficient granularity, leading to inaccurate results, or they involve large and cumbersome data processing, resulting in low efficiency. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide a method and terminal for studying the characteristics of low voltage indicators, which ensures the accuracy and efficiency of correlation analysis of low voltage indicators.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] A method for studying low-voltage performance characteristics, comprising the following steps:

[0007] S1, Receive the low-voltage related indicators to be analyzed;

[0008] S2. Based on the indicator item, search the dataset for all items containing the indicator item to form an item set;

[0009] S3. Determine the indicator data corresponding to each indicator item in the project set, and split the indicator data corresponding to each indicator item according to the indicator data, so that the ranges of any two indicator data corresponding to each indicator item do not overlap, thus forming the split project set.

[0010] S4. Based on the split item set, the improved Apriori algorithm is used to determine the index items that are strongly associated with low voltage. In the improved Apriori algorithm, all candidate item sets are first determined based on the split item set, and then the support of all candidate item sets is calculated in one scan of the split item set.

[0011] To solve the above-mentioned technical problems, another technical solution adopted by the present invention is as follows:

[0012] A low-voltage performance characteristic research terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it performs the following steps:

[0013] S1, Receive the low-voltage related indicators to be analyzed;

[0014] S2. Based on the indicator item, search the dataset for all items containing the indicator item to form an item set;

[0015] S3. Determine the indicator data corresponding to each indicator item in the project set, and split the indicator data corresponding to each indicator item according to the indicator data, so that the ranges of any two indicator data corresponding to each indicator item do not overlap, thus forming the split project set.

[0016] S4. Based on the split item set, the improved Apriori algorithm is used to determine the index items that are strongly associated with low voltage. In the improved Apriori algorithm, all candidate item sets are first determined based on the split item set, and then the support of all candidate item sets is calculated in one scan of the split item set.

[0017] The beneficial effects of this invention are as follows: When determining the indicators associated with low voltage, since these indicators are numerical data with a range, the indicator data corresponding to each indicator is first split, ensuring that the ranges of any two indicator data corresponding to each indicator do not overlap. This refines the granularity of data analysis and improves the accuracy of the determined indicators associated with low voltage. Since splitting the indicator data leads to an increase in the number of indicator data items, the Apriori algorithm is adaptively improved. First, all candidate option sets are determined based on the split item sets. Then, support is calculated for all candidate option sets in a single scan of the split item sets. That is, support is calculated for all candidate option sets during a single scan, reducing the number of scans and thus compensating for the increased computational complexity caused by the increase in indicator data items. This improves the efficiency of data analysis. Through data splitting and the adaptive reduction in the number of scans, the accuracy and efficiency of low voltage indicator association analysis are guaranteed. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the steps of a method for studying low-voltage performance characteristics according to an embodiment of the present invention.

[0019] Figure 2 This is a schematic diagram of the structure of a low-voltage performance research terminal according to an embodiment of the present invention. Detailed Implementation

[0020] To explain in detail the technical content, objectives, and effects of the present invention, the following description is provided in conjunction with the embodiments and accompanying drawings.

[0021] Please refer to Figure 1 A method for studying low-voltage performance characteristics, comprising the following steps:

[0022] S1, Receive the low-voltage related indicators to be analyzed;

[0023] S2. Based on the indicator item, search the dataset for all items containing the indicator item to form an item set;

[0024] S3. Determine the indicator data corresponding to each indicator item in the project set, and split the indicator data corresponding to each indicator item according to the indicator data, so that the ranges of any two indicator data corresponding to each indicator item do not overlap, thus forming the split project set.

[0025] S4. Based on the split item set, the improved Apriori algorithm is used to determine the index items that are strongly associated with low voltage. In the improved Apriori algorithm, all candidate item sets are first determined based on the split item set, and then the support of all candidate item sets is calculated in one scan of the split item set.

[0026] As described above, the beneficial effects of this invention are as follows: When determining the indicators associated with low voltage, since the indicators associated with low voltage are numerical data with a range, the indicator data corresponding to each indicator is first split, so that the ranges of any two indicator data corresponding to each indicator do not overlap. This refines the granularity of data analysis and improves the accuracy of the determined indicators associated with low voltage. Since splitting the indicator data leads to an increase in the number of indicator data items, the Apriori algorithm is adaptively improved. First, all candidate option sets are determined based on the split item sets. Then, the support of all candidate option sets is calculated in one scan of the split item sets. That is, the support of all candidate option sets is calculated in one scan. The support of all possible candidate option sets is calculated while scanning, reducing the number of scans. This compensates for the increased data computation complexity caused by the increase in the number of indicator data items and improves the efficiency of data analysis. Through data splitting and adaptive reduction of the number of scans, the accuracy and efficiency of low voltage indicator association analysis are guaranteed.

[0027] Furthermore, the step of splitting the indicator data corresponding to each indicator item based on the indicator data includes:

[0028] Determine the numerical range of the indicator data corresponding to each indicator item based on the indicator data;

[0029] Sort the index data corresponding to each index item in ascending order according to the left boundary of the numerical range;

[0030] For index data with overlapping numerical ranges, split the overlapping parts into independent index data.

[0031] As described above, when splitting the indicator data corresponding to each indicator item, the data is first sorted in ascending order according to the left boundary of the numerical range of each indicator data. Then, the indicator data with overlapping numerical ranges are split, which improves the speed of data splitting and ensures the comprehensiveness of data splitting.

[0032] Furthermore, the splitting of indicator data with overlapping numerical ranges includes:

[0033] The system determines the proportion of overlapping numerical ranges to the index data in which they are located. If the proportion is greater than a preset threshold, the index data with overlapping numerical ranges are split; otherwise, they are not split.

[0034] As described above, only index data whose overlapping numerical range accounts for more than a preset threshold are split, thus avoiding an excessive number of index data resulting from the split, ensuring the rationality of the number of index data after splitting, improving the effectiveness of the splitting and improving data processing efficiency.

[0035] Further, S4 includes:

[0036] Based on the split project set, determine all candidate project sets and add all candidate project sets to the candidate project set set;

[0037] The split item set is scanned. During the scan, the support of each candidate item set in the candidate item set is determined. If the determined current support is less than the minimum support, the first target candidate item set corresponding to the current support and all second target candidate item sets containing the first target candidate item set are deleted from the candidate item set.

[0038] The confidence level of the association rule corresponding to each candidate option set is determined based on the support level of each candidate option set in the determined candidate option set.

[0039] Association rules with a confidence level greater than or equal to the minimum confidence level are defined as strong association rules;

[0040] The index items that are strongly correlated with low voltage are determined based on the strong correlation rules.

[0041] As described above, the improved Apriori algorithm determines all candidate item sets based on the split item sets and adds them to the candidate item set set. During the scanning of the split item sets, the algorithm simultaneously calculates the support of candidate item sets based on a preset minimum support, and synchronously deletes candidate item sets that do not meet the minimum support requirement, as well as supersets containing such candidate item sets, from the candidate item set set. Finally, the candidate item sets remaining in the candidate item set set all meet the minimum support requirement. When determining strong association rules, the confidence score is directly calculated based on the association rule corresponding to each candidate item in the candidate item set set. This eliminates the computational complexity problem caused by the existing Apriori algorithm requiring multiple scans and multiple deletions of candidate item sets that do not meet the requirements, thus reducing computational complexity and improving computational speed.

[0042] Furthermore, the step of determining the index items strongly correlated with low voltage according to the strong correlation rule includes:

[0043] The index items strongly associated with low voltage are determined based on the strong association rule with the highest confidence level and the conclusion of low voltage.

[0044] As can be seen from the above description, when determining the indicators that are strongly correlated with low voltage, directly selecting the strong correlation rule with the highest confidence level and based on low voltage can quickly and accurately identify the indicators that are strongly correlated with low voltage.

[0045] Please refer to Figure 2 A low-voltage performance characteristic research terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it performs the following steps:

[0046] S1, Receive the low-voltage related indicators to be analyzed;

[0047] S2. Based on the indicator item, search the dataset for all items containing the indicator item to form an item set;

[0048] S3. Determine the indicator data corresponding to each indicator item in the project set, and split the indicator data corresponding to each indicator item according to the indicator data, so that the ranges of any two indicator data corresponding to each indicator item do not overlap, thus forming the split project set.

[0049] S4. Based on the split item set, the improved Apriori algorithm is used to determine the index items that are strongly associated with low voltage. In the improved Apriori algorithm, all candidate item sets are first determined based on the split item set, and then the support of all candidate item sets is calculated in one scan of the split item set.

[0050] As described above, the beneficial effects of this invention are as follows: When determining the indicators associated with low voltage, since the indicators associated with low voltage are numerical data with a range, the indicator data corresponding to each indicator is first split, so that the ranges of any two indicator data corresponding to each indicator do not overlap. This refines the granularity of data analysis and improves the accuracy of the determined indicators associated with low voltage. Since splitting the indicator data leads to an increase in the number of indicator data items, the Apriori algorithm is adaptively improved. First, all candidate option sets are determined based on the split item sets. Then, the support of all candidate option sets is calculated in one scan of the split item sets. That is, the support of all candidate option sets is calculated in one scan. The support of all possible candidate option sets is calculated while scanning, reducing the number of scans. This compensates for the increased data computation complexity caused by the increase in the number of indicator data items and improves the efficiency of data analysis. Through data splitting and adaptive reduction of the number of scans, the accuracy and efficiency of low voltage indicator association analysis are guaranteed.

[0051] Furthermore, the step of splitting the indicator data corresponding to each indicator item based on the indicator data includes:

[0052] Determine the numerical range of the indicator data corresponding to each indicator item based on the indicator data;

[0053] Sort the index data corresponding to each index item in ascending order according to the left boundary of the numerical range;

[0054] For index data with overlapping numerical ranges, split the overlapping parts into independent index data.

[0055] As described above, when splitting the indicator data corresponding to each indicator item, the data is first sorted in ascending order according to the left boundary of the numerical range of each indicator data. Then, the indicator data with overlapping numerical ranges are split, which improves the speed of data splitting and ensures the comprehensiveness of data splitting.

[0056] Furthermore, the splitting of indicator data with overlapping numerical ranges includes:

[0057] The system determines the proportion of overlapping numerical ranges to the index data in which they are located. If the proportion is greater than a preset threshold, the index data with overlapping numerical ranges are split; otherwise, they are not split.

[0058] As described above, only index data whose overlapping numerical range accounts for more than a preset threshold are split, thus avoiding an excessive number of index data resulting from the split, ensuring the rationality of the number of index data after splitting, improving the effectiveness of the splitting and improving data processing efficiency.

[0059] Further, S4 includes:

[0060] Based on the split project set, determine all candidate project sets and add all candidate project sets to the candidate project set set;

[0061] The split item set is scanned. During the scan, the support of each candidate item set in the candidate item set is determined. If the determined current support is less than the minimum support, the first target candidate item set corresponding to the current support and all second target candidate item sets containing the first target candidate item set are deleted from the candidate item set.

[0062] The confidence level of the association rule corresponding to each candidate option set is determined based on the support level of each candidate option set in the determined candidate option set.

[0063] Association rules with a confidence level greater than or equal to the minimum confidence level are defined as strong association rules;

[0064] The index items that are strongly correlated with low voltage are determined based on the strong correlation rules.

[0065] As described above, the improved Apriori algorithm determines all candidate item sets based on the split item sets and adds them to the candidate item set set. During the scanning of the split item sets, the algorithm simultaneously calculates the support of candidate item sets based on a preset minimum support, and synchronously deletes candidate item sets that do not meet the minimum support requirement, as well as supersets containing such candidate item sets, from the candidate item set set. Finally, the candidate item sets remaining in the candidate item set set all meet the minimum support requirement. When determining strong association rules, the confidence score is directly calculated based on the association rule corresponding to each candidate item in the candidate item set set. This eliminates the computational complexity problem caused by the existing Apriori algorithm requiring multiple scans and multiple deletions of candidate item sets that do not meet the requirements, thus reducing computational complexity and improving computational speed.

[0066] Furthermore, the step of determining the index items strongly correlated with low voltage according to the strong correlation rule includes:

[0067] The index items strongly associated with low voltage are determined based on the strong association rule with the highest confidence level and the conclusion of low voltage.

[0068] As can be seen from the above description, when determining the indicators that are strongly correlated with low voltage, directly selecting the strong correlation rule with the highest confidence level and based on low voltage can quickly and accurately identify the indicators that are strongly correlated with low voltage.

[0069] The low-voltage index characteristic research method and terminal of the present invention can be applied to the determination of indicators that are strongly correlated with low voltage in distribution networks. The following is a description of specific implementation methods:

[0070] Example 1

[0071] Please refer to Figure 1 A method for studying low-voltage performance characteristics, comprising the following steps:

[0072] S1, Receive the low-voltage related indicators to be analyzed;

[0073] Among them, the indicators related to low voltage can be set according to the specific application scenario. In this embodiment, the indicators used for correlation analysis include: average load rate of distribution transformer, annual maximum load rate, average distribution transformer capacity per household, and number of power supply users. If in another application scenario, the number of power supply users is not very relevant to low voltage, then the indicators used for correlation analysis can be: average load rate of distribution transformer, annual maximum load rate, and average distribution transformer capacity per household.

[0074] S2. Based on the indicator item, search the dataset for all items containing the indicator item to form an item set;

[0075] In this embodiment, the dataset consists of 10kV distribution transformers and low-voltage information. When determining the item set, the 10kV distribution transformers and low-voltage information are searched based on the low-voltage item and the four received indicators mentioned above. Items containing the low-voltage item and the four received indicators are identified to form the item set. For example, assuming I5 represents the average load factor of the distribution transformer, I15 represents the average transformer capacity per household, I23 represents low voltage, I10 represents the maximum annual load factor, and I20 represents the number of power supply users, the search will yield the item set shown in Table 1 below:

[0076] Table 1

[0077] TID Project Collection 1 I5, I15, I23 2 I10,I23 3 I20, I23 4 I5, I10, I23 5 I5, I20 6 I20, I23 7 I5, I20 8 I5, I15, I20, I23 9 I5, I20, I23

[0078] S3. Determine the indicator data corresponding to each indicator item in the project set, and split the indicator data corresponding to each indicator item according to the indicator data, so that the ranges of any two indicator data corresponding to each indicator item do not overlap, thus forming the split project set.

[0079] For example, in Table 1, the indicator data corresponding to I5 with number 1 is [0.7, 0.8], and the indicator data corresponding to I5 with number 4 is [0.75, 0.89]. There is an overlapping range of values ​​between them [0.75, 0.8]. In this case, the indicator data corresponding to I5 with number 1 can be split into [0.7, 0.75] and [0.75, 0.8], and the indicator data corresponding to I5 with number 4 can be split into [0.75, 0.8] and [0.8, 0.89]. This ensures that the range of any two indicator data corresponding to each indicator item does not overlap.

[0080] S4. Based on the split item set, the improved Apriori algorithm is used to determine the index items that are strongly associated with low voltage. In the improved Apriori algorithm, all candidate item sets are first determined based on the split item set, and then the support of all candidate item sets is calculated in one scan of the split item set.

[0081] In other words, the improved Apriori algorithm used in this embodiment does not require multiple scans like the existing Apriori algorithm, which scans the item set every time it calculates the support of a candidate item set. Furthermore, unlike the existing Apriori algorithm where candidate item sets are determined based on permutations of defined indicator items, this embodiment directly calculates the support of all possible candidate item sets determined based on the split item sets during a single scan of the split item sets. For example, if the item set is {I5, I15, I23}, then when scanning I5, the support of candidate item set {I5} is incremented by 1; when scanning I15, the support of candidate item sets {I15} and {I5, I15} is incremented by 1; and when scanning I23, the support of candidate item sets {I23}, {I5, I23}, {I15, I23}, and {I5, I15, I23} is incremented by 1.

[0082] Since the candidate set is generated based on the itemset, in an optional implementation, it is not necessary to determine all candidate sets in advance based on the split itemset. Instead, during the scanning process, the corresponding candidate items are determined synchronously, and the support of the corresponding candidate items is calculated synchronously, thereby further reducing computational complexity. Specifically, for each item:

[0083] For each data item in a project that has been scanned, the currently scanned data items in that project are identified. Based on the scanned data items, all possible candidate item sets are generated, forming a candidate item set set. For each candidate item set in the candidate item set set, it is determined whether it has been saved. If so, the support of the corresponding saved candidate item set is updated by incrementing its support by one. If not, the candidate item set is saved, and its support is calculated by incrementing its support by one. In this way, candidate item sets are generated and support is calculated simultaneously while scanning. This eliminates the need for the existing Apriori algorithm to scan the item set and candidate item set set multiple times. It also avoids the situation in the existing technology where candidate item sets determined directly based on the permutation and combination of indicator items may be invalid candidate item sets. That is, the determined candidate item set does not exist in the item set, its support is 0, and thus it leads to an invalid scan.

[0084] Example 2

[0085] This embodiment further specifies how to split the indicator data corresponding to each indicator item, so that the ranges of any two indicator data corresponding to each indicator item do not overlap. Specifically:

[0086] Determine the numerical range of the indicator data corresponding to each indicator item based on the indicator data;

[0087] Sort the index data corresponding to each index item in ascending order according to the left boundary of the numerical range;

[0088] For index data with overlapping numerical ranges, split the overlapping parts into independent index data.

[0089] For example, if the average load factor of the distribution transformer has the following indicators: [0.7, 0.9], [0.9, 1], [0.8, 1], then they can be sorted in ascending order according to the size of their left boundary as follows:

[0090] [0.7,0.9], [0.8,1], [0.9,1];

[0091] Then, the overlapping parts are split:

[0092] [0.7,0.9] is split into [0.7,0.8] and [0.8,0.9];

[0093] [0.8,1] is split into [0.8,0.9] and [0.9,1];

[0094] In an optional implementation, the splitting of indicator data with overlapping numerical ranges includes:

[0095] Determine the proportion of overlapping numerical ranges to the index data in which they are located. If the proportion is greater than a preset threshold, then the index data with overlapping numerical ranges are split; otherwise, they are not split.

[0096] For example, if a preset threshold of 20% is set, then for the indicator data [0.7,0.9] and [0.8,1], since the overlapping part is [0.8,0.9], which accounts for more than 20% of both [0.7,0.9] and [0.8,1] respectively, it needs to be split. However, if the indicator data is [0.7,0.9] and [0.89,1], since the overlapping part is [0.89,0.9], which accounts for less than 20%, it does not need to be split.

[0097] Example 3

[0098] This embodiment further defines how to improve the existing Apriori algorithm. Specifically, S4 includes:

[0099] Based on the split project set, determine all candidate project sets and add all candidate project sets to the candidate project set set;

[0100] The split item set is scanned. During the scan, the support of each candidate item set in the candidate item set is determined. If the determined current support is less than the minimum support, the first target candidate item set corresponding to the current support and all second target candidate item sets containing the first target candidate item set are deleted from the candidate item set.

[0101] The confidence level of the association rule corresponding to each candidate option set is determined based on the support level of each candidate option set in the determined candidate option set.

[0102] Association rules with a confidence level greater than or equal to the minimum confidence level are defined as strong association rules;

[0103] The index items strongly correlated with low voltage are determined based on the strong correlation rules.

[0104] In an optional implementation, determining the index items strongly correlated with low voltage according to the strong correlation rule includes:

[0105] The index items strongly correlated with low voltage are determined based on the strong correlation rule with the highest confidence level and the conclusion of low voltage.

[0106] The following is an example to illustrate this:

[0107] Taking the item set described in Table 1 as an example, the value range of I5 in item 1 is [0.7, 1], and the value range of I5 in the remaining items is [0.8, 1]. Therefore, I5 in item 1 needs to be split to obtain [0.7, 0.8] and [0.8, 1]. The item corresponding to [0.7, 0.8] can be labeled as I5', and the item corresponding to [0.8, 1] can be labeled as I5. Then, Table 1 is transformed into Table 2 as shown below:

[0108] Table 2

[0109] TID Project Collection 1 I5',I5,I15,I23 2 I10,I23 3 I20, I23 4 I5, I10, I23 5 I5, I20 6 I20, I23 7 I5, I20 8 I5, I15, I20, I23 9 I5, I20, I23

[0110] In implementing the Apriori algorithm, one can first generate all possible candidate sets based on each item in the itemset of Table 2, and then perform the scan; alternatively, one can generate candidate sets while scanning the itemset of Table 2. In this embodiment, candidate sets are generated while scanning the itemset of Table 2. Specifically:

[0111] First, scan the item labeled 1 {I5',I5,I15,I23}. When I5' is scanned, add {I5'} to the candidate item set and calculate its corresponding support count as 1. When I5 is scanned, determine that both the candidate item sets {I5} and {I5,I5'} are newly generated candidate item sets, add them to the candidate item set, and calculate their respective support counts as 1. When I15 is scanned, determine the candidate item sets {I15}, {I15,I5'}, {I15,I5}, and {I15,I5',I5}, and add them to the candidate item set respectively. The candidate item set is generated, and the corresponding support count is calculated to be 1 for each item. This process is repeated for each data item, with candidate item sets generated and support counts calculated simultaneously. If a candidate item already exists in the candidate item set, the support count is updated directly. For example, when the data item I5 with the number 4 is scanned, since it has already been generated and has a support count of 1 when the item with the number 1 is scanned, it can be updated to 2. After the scanning is completed, the final candidate item set and its corresponding support count are shown in Table 3.

[0112] Table 3

[0113]

[0114]

[0115] Setting the minimum support count to 2 means that candidate itemsets with a minimum support count greater than or equal to 2 are defined as frequent itemsets. Candidate itemsets with a support count less than the minimum support count are then removed from the candidate itemset set. According to the rule that if a set is not a frequent itemset, then all its supersets are also not frequent itemsets, when deleting a candidate itemset that is not a frequent itemset, all its corresponding supersets can be deleted simultaneously without needing to compare them one by one. For example, if {I5,I10} is a non-frequent itemset, then the superset {I23,I5,I10} containing {I5,I10} can be deleted; if {I15,I20} is a non-frequent set, then the supersets {I5,I15,I20}, {I23,I15,I20}, and {I23,I5,I15,I20} containing {I15,I20} can be deleted.

[0116] After removing all infrequent itemsets from the candidate itemset set, all candidate itemsets in the set will then be frequent itemsets. Next, strong association rules with a minimum confidence level greater than or equal to the set value are constructed based on all frequent itemsets. The steps for generating strong association rules are as follows:

[0117] 1) For each frequent itemset l, generate all its non-empty proper subsets;

[0118] 2) For each non-empty proper subset s, if

[0119]

[0120] Then output Where minconfidence is the minimum confidence threshold, and support_count represents the support count;

[0121] For example, given the frequent set {I5, I15, I23}, the non-empty proper subsets of this frequent set are {I5, I15}, {I5, I23}, {I15, I23}, {I5}, {I23}, and {I15}. According to the confidence formula:

[0122]

[0123] The corresponding confidence levels are as follows:

[0124]

[0125]

[0126]

[0127]

[0128]

[0129]

[0130] In this embodiment, setting minconfidence = 70% results in the following strong rules: Where I5 ​​represents the average load rate of the distribution transformer is 80% to 100%, I15 represents the average distribution transformer capacity per household [2,3] (unit is kVA, kilovolt-ampere), and I23 represents low voltage;

[0131] Since the conclusion is that a low voltage occurred in the transformer area, the strong correlation rule with I23 (low voltage occurred in the transformer area) as the final conclusion is: Right now:

[0132] The average load rate of the distribution transformer is 80% to 100% && the average distribution transformer capacity per household [2,3) => low voltage in the distribution area (when the minimum support count is 2 and the minimum confidence threshold is 70%).

[0133] Therefore, the indicators strongly correlated with low voltage can be determined as the average load rate of distribution transformers being 80% to 100% and the average distribution transformer capacity per household [2,3);

[0134] Since the conclusion has been established that low voltage occurred in the transformer area, in an optional implementation, when selecting frequent itemsets for association rule calculation, the frequent itemset containing I23 with the largest number of items can be selected from the candidate itemset set. Furthermore, when calculating the confidence level, the case where the conclusion is I23 can be selected for calculation. For example, in Table 3, only {I5,I15,I23} and {I5,I20,I23} can be selected for calculation, and only the following calculations are performed:

[0135] I5&&I15=>I23 Confidence=2 / 2=100%

[0136] I5&&I20=>I23 Confidence=2 / 4=50%

[0137] Therefore, the strong correlation can be derived as: I5 && I15 => I23.

[0138] Example 4

[0139] Please refer to Figure 2 A low-voltage performance characteristic research terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements each step of the low-voltage performance characteristic research method described in any one of Embodiments 1 to 3.

[0140] In summary, the low-voltage performance characteristic research method and terminal provided by this invention, when determining the performance indicators associated with low voltage, first splits the performance data corresponding to each performance indicator, given that these indicators have numerical ranges. This ensures that the ranges of any two performance data points corresponding to each performance indicator do not overlap, refining the granularity of data analysis and improving the accuracy of the determined performance indicators associated with low voltage. Since splitting the performance data leads to an increase in the number of performance data points, the Apriori algorithm is adaptively improved, performing scanning and analysis simultaneously during the scanning process. For each data item in the split project set, a corresponding candidate option set is determined, and the support of the corresponding candidate option set is calculated simultaneously. That is, all candidate option sets are generated and the support is calculated in a single scan. By generating all possible candidate option sets while scanning and calculating the support of all possible candidate option sets, the number of scans is reduced and the computational complexity is lowered. This compensates for the problem of increased data computational complexity caused by the increase in the number of indicator data items, and improves the efficiency of data analysis. Through data splitting and adaptive reduction of the number of scans, the accuracy and efficiency of low voltage indicator correlation analysis are ensured.

[0141] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent modifications made based on the content of the present invention specification and drawings, or direct or indirect applications in related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for studying low-voltage performance characteristics, characterized in that, Including the following steps: S1, Receive the low-voltage related indicators to be analyzed; S2. Based on the indicator item, search the dataset for all items containing the indicator item to form an item set; S3. Determine the indicator data corresponding to each indicator item in the project set, and split the indicator data corresponding to each indicator item according to the indicator data, so that the ranges of any two indicator data corresponding to each indicator item do not overlap, thus forming the split project set. S4. Based on the split item set, the improved Apriori algorithm is used to determine the index items that are strongly associated with low voltage. In the improved Apriori algorithm, all candidate item sets are first determined based on the split item set, and then the support of all candidate item sets is calculated in one scan of the split item set. The step of splitting the indicator data corresponding to each indicator item based on the indicator data includes: Determine the numerical range of the indicator data corresponding to each indicator item based on the indicator data; Sort the index data corresponding to each index item in ascending order according to the left boundary of the numerical range; For index data with overlapping numerical ranges, split the overlapping parts into independent index data. The splitting of indicator data with overlapping numerical ranges includes: Determine the proportion of overlapping numerical ranges to the index data in which they are located. If the proportion is greater than a preset threshold, then the index data with overlapping numerical ranges are split; otherwise, they are not split. S4 includes: Based on the split project set, determine all candidate project sets and add all candidate project sets to the candidate project set set; The split item set is scanned. During the scan, the support of each candidate item set in the candidate item set is determined. If the determined current support is less than the minimum support, the first target candidate item set corresponding to the current support and all second target candidate item sets containing the first target candidate item set are deleted from the candidate item set. The confidence level of the association rule corresponding to each candidate option set is determined based on the support level of each candidate option set in the determined candidate option set. Association rules with a confidence level greater than or equal to the minimum confidence level are defined as strong association rules; The index items that are strongly correlated with low voltage are determined based on the strong correlation rules.

2. The method for studying low-voltage performance characteristics according to claim 1, characterized in that, The indicators that are strongly correlated with low voltage, as determined according to the strong correlation rule, include: The index items strongly associated with low voltage are determined based on the strong association rule with the highest confidence level and the conclusion of low voltage.

3. A low-voltage performance characteristic research terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it performs the following steps: S1, Receive the indicators related to low voltage to be analyzed; S2. Based on the indicator item, search the dataset for all items containing the indicator item to form an item set; S3. Determine the indicator data corresponding to each indicator item in the project set, and split the indicator data corresponding to each indicator item according to the indicator data, so that the ranges of any two indicator data corresponding to each indicator item do not overlap, thus forming the split project set. S4. Based on the split item set, the improved Apriori algorithm is used to determine the index items that are strongly associated with low voltage. In the improved Apriori algorithm, all candidate item sets are first determined based on the split item set, and then the support of all candidate item sets is calculated in one scan of the split item set. The step of splitting the indicator data corresponding to each indicator item based on the indicator data includes: Determine the numerical range of the indicator data corresponding to each indicator item based on the indicator data; Sort the index data corresponding to each index item in ascending order according to the left boundary of the numerical range; For index data with overlapping numerical ranges, split the overlapping parts into independent index data. The splitting of indicator data with overlapping numerical ranges includes: Determine the proportion of overlapping numerical ranges to the index data in which they are located. If the proportion is greater than a preset threshold, then the index data with overlapping numerical ranges are split; otherwise, they are not split. S4 includes: Based on the split project set, determine all candidate project sets and add all candidate project sets to the candidate project set set; The split item set is scanned. During the scan, the support of each candidate item set in the candidate item set is determined. If the determined current support is less than the minimum support, the first target candidate item set corresponding to the current support and all second target candidate item sets containing the first target candidate item set are deleted from the candidate item set. The confidence level of the association rule corresponding to each candidate option set is determined based on the support level of each candidate option set in the determined candidate option set. Association rules with a confidence level greater than or equal to the minimum confidence level are defined as strong association rules; The index items that are strongly correlated with low voltage are determined based on the strong correlation rules.

4. A low-voltage performance characteristic research terminal according to claim 3, characterized in that, The indicators that are strongly correlated with low voltage, as determined according to the strong correlation rule, include: The index items strongly associated with low voltage are determined based on the strong association rule with the highest confidence level and the conclusion of low voltage.

Images