CN112258338A - Automatic base station cost auditing method based on nearest neighbor algorithm - Google Patents
Automatic base station cost auditing method based on nearest neighbor algorithm Download PDFInfo
- Publication number
- CN112258338A CN112258338A CN202010963119.4A CN202010963119A CN112258338A CN 112258338 A CN112258338 A CN 112258338A CN 202010963119 A CN202010963119 A CN 202010963119A CN 112258338 A CN112258338 A CN 112258338A
- Authority
- CN
- China
- Prior art keywords
- base station
- data
- nearest neighbor
- electric charge
- auditing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 50
- 238000000034 method Methods 0.000 title claims abstract description 36
- 239000013598 vector Substances 0.000 claims abstract description 25
- 238000012545 processing Methods 0.000 claims abstract description 16
- 230000002159 abnormal effect Effects 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 230000005611 electricity Effects 0.000 claims description 18
- 238000004364 calculation method Methods 0.000 claims description 10
- 230000001502 supplementing effect Effects 0.000 claims description 2
- 238000007418 data mining Methods 0.000 abstract description 6
- 238000010801 machine learning Methods 0.000 abstract description 2
- 238000012550 audit Methods 0.000 description 24
- 230000007547 defect Effects 0.000 description 6
- 238000007726 management method Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000012417 linear regression Methods 0.000 description 3
- 239000000523 sample Substances 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 239000013074 reference sample Substances 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Water Supply & Treatment (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Public Health (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a base station expense automatic auditing method based on a nearest neighbor algorithm, which collects data such as historical payment and base station attribute related to base station electric charge through means such as data mining, machine learning and the like, and then carries out preprocessing and standardized processing according to basic data; selecting a strong correlation index above a set threshold as an effective characteristic using the information gain ratio; calculating the average distance between all the multi-dimensional characteristic vectors, setting the distance radius according to the average distance, and performing spherical tree modeling on the characteristic vectors of all the reference sets by using a spherical tree algorithm; according to data distribution self-adaptive radius distance determination, finding out the nearest reference set for the payment records to be audited; and calculating the confidence interval as the threshold value for auditing and judging the exceeding, thereby finally judging whether the electric charge reimbursement limit is abnormal or not.
Description
Technical Field
The invention relates to the technical field of energy consumption and expense management and control in the mobile communication industry, relates to technologies such as data mining and machine learning algorithm modeling of a computer, is a technical solution for rationality audit of electric charge amount reimbursed by a communication base station, consists of a series of data processing methods and model algorithms, and particularly relates to a base station expense automatic audit method based on a nearest neighbor algorithm.
Background
At present, regarding the base station electric charge reimbursement auditing management of a communication operator, usually, a charge auditor judges whether the electric charge amount on an electric charge reimbursement bill of the base station conforms to a specified range according to own experience, however, in the face of the electric consumption commonly influenced by various complex factors such as equipment of various base stations, the environment where the equipment is located and the like, the auditor is often difficult to quickly calculate the reasonable range of the electric consumption of each base station in a payment period, and the goal of reasonable cost control is difficult to scientifically and effectively execute.
The defects of the current manual auditing scheme adopted for the base station electric charge reimbursement are summarized as follows:
(1) the artificial failure can accurately account the reasonable power consumption of the current base station in time: because factors influencing the power consumption of the base station are numerous and complex, for example, the indoor temperature of the machine room influences the power consumption of the air conditioner, the heat exchange coefficient of a wall material of the machine room influences the indoor temperature of the machine room, and in addition, different telephone traffic of people in different areas also influences the power of communication equipment all the time;
(2) the people are hard to remember and quickly select the best matched payment and the reasonable acceptable auditing interval from the verified mass historical payment record data as the concrete auditing basis;
(3) the human brain-born subjectivity and the error difficult to calculate exist in the manual examination of the reasonability of the electric charge;
(4) general auditors are difficult to comply with the power consumption model of experts through scientific demonstration and the audit flow specification of company management during auditing;
introduction of related art:
(1) the auditing method based on the linear model prediction class comprises the following steps:
the idea introduction of the auditing algorithm comprises the following steps: it is common to model each base station independently using a linear regression fitting model based on the main varying influences of the power consumption of the base station (communication equipment-traffic, air conditioning-temperature). The predicted value plus a certain artificially set exceeding proportion is used as the threshold value of the audit.
Analysis of the disadvantages of such methods: the method can be seen from a scatter distribution diagram of data such as power consumption of a base station, daily average temperature, telephone traffic and the like, and the two factors have no obvious linear correlation with the power consumption, so that the problem of mathematically strong fitting exists by using linear regression; in addition, generally, the sample data of a single base station is less, and the reliability of linear regression is poorer; and the time sequence trend fitting of the historical payment power consumption of a single base station can only detect the payment problem of the single station with relatively higher history, and can not check the change of base station equipment and the problem that maintenance personnel of the station always cannot perform transverse comparison.
(2) The auditing method based on the simple KNN algorithm class comprises the following steps:
the idea introduction of the auditing algorithm comprises the following steps: the method comprises the steps of classifying Y values based on simple KNN or by using check density estimation before KNN processing, carrying out LMNN processing on a feature vector set on the basis of the classification, finding Y value sets corresponding to K adjacent vectors of feature vectors to be audited by using an artificial K value determination KNN algorithm, and taking the maximum value as a judgment threshold value of audit.
Analysis of the disadvantages of such methods: in order to meet the problem that KNN treats different features with the same weight mathematically, an LMNN method is adopted to make up for modifying feature vectors, but the method is based on a clear classification problem, but for continuous electricity charge Y values, if the method is adopted to carry out forced classification treatment, adjacent points on two sides near each classification boundary are isolated.
Secondly, the simple KNN algorithm needs to calculate and sort the distances between the feature vectors to be audited and all feature vectors in a reference set, the performance is low, K adjacent circles are set artificially, outliers are compared when the K adjacent circles are met, in order to reach the K adjacent points, the very far points are taken as the adjacent similar clusters, and the problem that the audit error is too large is caused.
Disclosure of Invention
The invention aims to solve the problems and provides a base station expense automatic auditing method based on a nearest neighbor algorithm, which calculates a reasonable range for the electric charge paid by the current base station through a large amount of data mining technology and a scientific calculation algorithm of the power consumption of the base station as a specific basis for current auditing, adopts computer automation to automatically audit an electric charge payment and reimbursement bill, overcomes the defect of manual auditing, improves auditing efficiency and scientific accuracy and saves labor cost.
In order to achieve the purpose, the invention adopts the following technical scheme:
the automatic base station cost auditing method based on the nearest neighbor algorithm comprises the following steps:
1) collecting power consumption and base station attribute data of a base station, wherein the power consumption and base station attribute data comprise one or more items of electric meter reading, electric charge payment bill, machine room lease contract and base station equipment;
2) preprocessing and standardizing alternative indexes of the power consumption of the base station and attribute data of the base station and daily average electric charge;
3) calculating the information gain rate of the alternative indexes and the daily average electricity charge, selecting effective characteristic vectors related to the electricity consumption and adding weights;
4) calculating the average distance of the effective characteristic vectors, and performing ball tree modeling or coding on the effective characteristic vectors by using a ball tree algorithm to form a reference set;
5) processing the payment records to be audited in the steps 1) and 2), and finding out a nearest neighbor reference payment set of the payment records to be audited by using a nearest neighbor algorithm with a variable radius distance;
6) and calculating a confidence interval of the payment records to be audited, using the confidence interval as an auditing judgment over-standard threshold, and finally judging whether the electric charge reimbursement limit is over-standard and abnormal according to the threshold.
In the automatic base station cost auditing method based on the nearest neighbor algorithm, the preprocessing of alternative indexes or daily average electricity charge data comprises removing abnormal data and filling up and correcting missing data, and the standardization processing comprises processing by adopting an (X-avg (X)/std (X) calculation formula, wherein X is a sequence group of values of similar indexes.
In the automatic base station fee auditing method based on the nearest neighbor algorithm, the abnormal data removal comprises the steps of removing outliers by comparing the daily average electric charges of the base station and similar historical payment, or removing abnormally high electric charges according to the daily electric charge upper limit of the sum of the powers of the electric consumption equipment; and supplementing and correcting missing data according to the correlation between the candidate indexes of each dimension and the daily average electric charge table and the continuation inertia of each type of candidate indexes and the daily average electric charge data, and checking and correcting through a confidence interval and a calculation relation of the candidate indexes and the daily average electric charge related data.
In the automatic auditing method for base station cost based on nearest neighbor algorithm, the base station power consumption data also comprises air conditioner power consumption-temperature coefficient T ', T' ═ 1+ (T-14) for weighting the average daily temperature1.1Where T is the actual outdoor temperature.
The beneficial effects produced by adopting the invention are as follows:
according to the invention, a reasonable range is calculated for the electric charge paid by the current base station through a large amount of data mining technology and a scientific calculation algorithm of the electric consumption of the base station as a specific basis of current audit, and the automatic audit is carried out on the electric charge payment reimbursement bill by adopting computer automation, so that the defect of manual audit is overcome, the audit efficiency and scientific accuracy are improved, and the labor cost is also saved.
Drawings
FIG. 1 is a schematic diagram of the steps of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, in order to help communication operators to scientifically and effectively manage and control the operation cost of the base station, the invention checks various false and multi-report situations of the electricity charge report, and realizes cost reduction and efficiency improvement. The invention aims to establish a set of automatic electric charge auditing method based on the combination of a base station power consumption scientific calculation model and enterprise charge management regulations.
The invention discloses a base station expense automatic auditing method based on a nearest neighbor algorithm, which collects data such as historical payment and base station attribute related to base station electric charge by means of data mining and the like, and then carries out preprocessing and standardized processing according to basic data; selecting a strong correlation index above a set threshold as an effective characteristic using the information gain ratio; calculating the average distance between all the multi-dimensional characteristic vectors, setting the distance radius according to the average distance, and performing spherical tree modeling on the characteristic vectors of all the reference sets by using a spherical tree algorithm; using a nearest neighbor algorithm with variable radius distance to find out a nearest neighbor reference payment set for payment to be audited; and calculating the confidence interval as the threshold value for auditing and judging the exceeding, thereby finally judging whether the electric charge reimbursement limit is abnormal or not.
The main program implementation steps are as follows:
step 1: and collecting the relevant data of the power consumption of the base station and the attribute of the base station from multiple channels such as the reading of an electric meter, an electric charge payment bill, a machine room lease contract, base station equipment and the like.
Reading an electric meter: in order to prevent the base station maintainers from falsely reporting the power consumption, the intelligent electric meter is required to automatically transmit the power consumption degree for detecting and correcting the problem of the manually input electric meter degree.
Collecting data such as electric charge payment bill, contract, base station equipment and the like: the data is collected and summarized through technologies such as data extraction of the electronic invoice fixed template and other channels such as other related system interface import.
Step 2: data in historical payment and resource attributes related to the electricity charge of the base station are artificially selected, extracted and aggregated into reference set alternative factors, namely alternative indexes and daily average electricity charge, and the alternative indexes or the daily average electricity charge are preprocessed firstly for removing abnormity, filling up, correcting and the like, and then are subjected to standardization processing.
Removing abnormal data: the base station and the daily average electric charge of similar historical payment are compared to remove outliers, and meanwhile, abnormally high electric charges can be removed according to the daily electric charge upper limit of the sum of the power consumption equipment on the premise that the quality of data such as power is guaranteed.
And (3) filling and correcting missing data: the deficiency is compensated according to the correlation between the candidate indexes of each dimension and the daily average electric charge table and the continuation inertia of the candidate indexes and the daily average electric charge data, and the correction is checked through a confidence interval and a calculation relation of the candidate indexes and the daily average electric charge related data.
And (3) standardization treatment: generally, the method adopts (X-avg (X))/std (X) calculation formula, wherein X is the sequence group of the values of the similar index.
And step 3: for the normalized candidate indexes and the daily average power charge, calculating and selecting effective feature vectors with correlation degrees reaching a certain threshold value by using an information gain rate algorithm (the threshold value needs to be determined according to the accumulated correlation degrees of multiple indexes as appropriate), and performing weighting processing on the effective feature vectors according to the correlation coefficient setting (in order to increase the distance between the features with strong correlation degrees and reduce the distance between the features with weak correlation degrees).
And calculating the correlation degree of each alternative index factor and the Y value of the daily average electric charge by using an information gain rate algorithm, and selecting the characteristics according to the correlation degree.
Information gain: the difference between the information entropy of a node and the sum of the information entropies of all its children nodes. The calculation formula (1) of the information entropy is as follows:
where S is a variable in the set and the probability that S corresponds to the set is Pi。
Information gain ratio: a ratio of a node information gain to a node split information metric. The calculation formula (2) of the information gain is as follows:
where V (A) is the value range of attribute A; s is a sample set; sVIs the set of samples in S with a value equal to V on attribute a.
The Gain ratio metric is defined by the Gain metric Gain (S, a) and the split information metric splittinformation (S, a) together, as shown in the following equation (3) for calculating the information Gain ratio:
wherein, the split information metric is defined as (the split information is used to measure the breadth and uniformity of the attribute split data) formula (4):
where S1 through Sc are c sample subsets of c values of attribute a that segment S. Note that the split information is actually the entropy of S with respect to the values of attribute A. This is in contrast to our previous use of entropy, where we only consider the entropy of S with respect to the value of the target property that the learned tree is to predict.
And selecting factors reaching a certain threshold value as effective characteristics according to the information gain rate of each factor relative to the dependent variable, performing cumulative full percentage conversion on the information gain rate as weight, and processing the effective characteristic weight.
And 4, step 4: calculating the average distance of points represented by all the characteristic vectors to form a reference set as a reference of the selected radius distance; and (3) performing ball tree modeling on the effective characteristic vectors of the whole reference set by using a ball tree algorithm in the nearest neighbor algorithm.
The key is to average the distances for all the points represented by the feature vectors and to use the ball tree algorithm in the nearest neighbor algorithm.
Introduction of the principle of the ball tree algorithm: a ball tree recursively divides the data into nodes defined as having a center CCC and a radius rrr, such that each of the nodes is within the hypersphere defined by rrr, CCC. The triangle inequality is used to reduce the candidate points for a neighbor point search:
∣x+y∣≤∣x∣+∣y∣
in this arrangement, it is sufficient to calculate the distance between a test point and the centroid to determine the upper and lower bounds of the distance to all points within the node. Due to the spherical geometry of the nodes of the spherical tree, the performance of the spherical tree is superior to that of the KD-tree in a high dimension, although the actual performance is highly dependent on the structure of the training data.
And (4) encoding the characteristic vector data of the historical payment records by using a ball tree algorithm to form a reference set.
And 5: the payment records to be audited are firstly standardized by the step 1 to form an auditing set; and then the feature weighting processing of the to-be-audited set is carried out by using the weights of the features which are calculated before the step 2.
The weight multiplied by the features of the set to be audited, i.e. the test set, is kept consistent with the features of the previous reference set.
Step 6: and auditing the feature vectors of the test set one by using a nearest neighbor algorithm based on the radius distance.
And using radius _ neighbors algorithm in the ball tree model constructed by the reference set, wherein the initial radius parameter uses the radius of the previously constructed ball tree model, and calculating the nearest neighbor point set within the radius distance of the multidimensional space point represented by the audit payment feature vector. For the nearest neighbor reference set samples with insufficient number in the set current radius, the coverage rate can be improved by setting an increasing rate to sequentially enlarge the radius distance, and the expandable range needs to be specifically determined through multiple tests according to the data distribution condition: when the feature vector points for auditing are encountered and no set lowest adjacent reference point is found in the initial radius range, the radius distance can be automatically and gradually enlarged to search for adjacent points, and when the minimum number is reached, the enlargement to search is stopped; if the minimum number of adjacent reference points is not found when the preset maximum radius distance is enlarged, the judgment is judged to be impossible due to data loss, and manual examination is waited.
The Euclidean Distance formula used in the nearest neighbor algorithm neighbors for calculating the multidimensional space point is as follows:
and 7: and selecting a threshold value for judging the abnormity from the confidence interval and the maximum value of the Y value list corresponding to the nearest neighbor.
Setting a confidence interval range by solving the mean value and standard deviation of the list of the daily average electric charges when the number of the obtained nearest neighbor reference samples is larger than a set threshold, wherein the upper bound value of the confidence interval is the threshold for judging whether the daily average electric charges paid by the user exceed the standard; and for the found nearest neighbor reference sample number smaller than or equal to the set threshold, only using the maximum daily average electricity charge as the threshold for auditing judgment.
And 8: and (5) an application description of an automatic audit result of the electric charge reimbursement bill.
The auditing algorithm can be used for automatically scanning each payment record one by one, auditing is carried out by using the auditing algorithm, and an electric charge payment bill with the electric charge amount exceeding the standard in the month payment record is given out and converted into an abnormal expense bill to be returned to a claimant for processing.
In the reality of the electricity charge reimbursement of the base station, various false and multiple reports exist, the requirements of scientific management such as the cost control regulation standard of an operator and the scientific calculation method of the electricity consumption of the base station need to be met, the calculated reasonable electricity charge range corresponding to the electricity charge reimbursement bill is used as the threshold value of the abnormal audit judgment, so that the electricity charge audit method is reasonable in conditions, can be accepted by users and auditors, can be popularized and applied, and finally can achieve the customer operation target of cost reduction and efficiency improvement. The technical scheme and the logic algorithm flow of the electric charge auditing method can meet the requirements.
Any good method which is scientifically demonstrated and approved generally needs to be implemented on a specific product carrier to be popularized and used in actual economic activities in order to generate actual social and economic benefits. The invention is embedded in related products such as an audit assistant and a related intelligent data analysis platform which are produced by Shanxi dynasty, and the corresponding audit function can help the cost management and control personnel of customers to improve the scientific and reasonable level and the operation efficiency of auditing the reimbursement bill. Therefore, the technical scheme of the invention can be adopted to develop the IT system to make the electricity fee auditing method fall into an automatic auditing tool.
Aiming at the defects of the common expense auditing method, the method disclosed by the invention makes up for the defects and expands more advantages as follows:
(1) the related factors are changed, for example, for the temperature, weighting processing is carried out according to the average temperature of the weather, the day and the average temperature, and the air conditioner is automatically turned on when the room temperature reaches a certain temperature and is automatically turned off after the temperature returns, and indexes of the power consumption of the air conditioner at different temperatures are different. Therefore, the temperature is treated as follows:
T′＝1+(T-14)1.1
so that the factor is a more computationally consistent correlation of the change in value and the change in Y value.
(2) The characteristic selection is carried out by combining human and information gain rate, so that the method is beneficial to using the experience of people and using the verification between data, and the characteristic selection is more effective;
(3) a ball tree coding model is constructed for training set data by using a ball tree algorithm in a nearest neighbor algorithm, so that an adjacent reference set at a certain radius distance can be quickly found in the actual audit under the frame of the ball tree, and the audit performance is improved.
(4) The use of the radius-based nearest neighbor algorithm instead of K-number-based nearest neighbors effectively avoids the problem that K-number-based nearest neighbors may contain very distant points.
(5) Aiming at the problem that the audit coverage rate is low under the condition of small data quantity, the invention changes the commonly adopted nearest neighbor algorithm with fixed radius distance into the method that the radius distance can be automatically and gradually enlarged according to the condition.
According to the invention, a reasonable range is calculated for the electric charge paid by the current base station through a large amount of data mining technology and a scientific calculation algorithm of the electric consumption of the base station as a specific basis of current audit, and the automatic audit is carried out on the electric charge payment reimbursement bill by adopting computer automation, so that the defect of manual audit is overcome, the audit efficiency and scientific accuracy are improved, and the labor cost is also saved.
The foregoing is a more detailed description of the invention that is presented in connection with specific embodiments, which are not intended to limit the invention to the particular embodiments described herein. For a person skilled in the art to which the invention pertains, several equivalent alternatives or obvious modifications, all of which have the same properties or uses, without departing from the inventive concept, should be considered as falling within the scope of the patent protection of the invention, as determined by the claims filed.
Claims (5)
1. A base station cost automatic auditing method based on a nearest neighbor algorithm is characterized by comprising the following steps:
1) collecting power consumption and base station attribute data of a base station, wherein the power consumption and base station attribute data of the base station comprise multi-source multi-dimensional related data such as electric meter reading, electric charge payment bill, machine room lease contract, base station equipment and the like;
2) preprocessing and standardizing alternative indexes of the power consumption of the base station and attribute data of the base station and daily average electric charge;
3) calculating the information gain rate of the alternative indexes and the daily average electricity charge, selecting effective characteristic vectors related to the electricity consumption, and determining the weight of the effective characteristics according to the information gain rate;
4) calculating the average distance of all points represented by the effective characteristic vector, and performing ball tree modeling or coding on the effective characteristic vector by using a ball tree algorithm to form a reference set based on the radius determined by the average distance;
5) processing the payment records to be audited in the steps 1) and 2), and finding out a payment reference set of nearest neighbor reference of the payment records to be audited by using a nearest neighbor algorithm with variable radius distance;
6) and calculating a confidence interval of the payment records to be audited, using the confidence interval as an auditing judgment over-standard threshold, and finally judging whether the electric charge reimbursement limit is over-standard and abnormal according to the threshold.
2. The method for automatically auditing the costs of base stations based on nearest neighbor algorithm according to claim 1, characterized in that: the alternative indexes or the daily average electricity charge data are preprocessed by removing abnormal data and filling up and correcting missing data, and the standardization processing comprises processing by adopting an (X-avg (X))/std (X) calculation formula, wherein X is a sequence group of values of the similar indexes.
3. The method of claim 2, wherein the method for automatically auditing the cost of the base station based on the nearest neighbor algorithm comprises: removing abnormal data comprises removing outliers by comparing daily average electric charges of the base station and similar historical payment, or removing abnormally high electric charges according to the daily electric charge upper limit of the sum of the powers of the electric consumption equipment; and supplementing and correcting missing data according to the correlation between the candidate indexes of each dimension and the daily average electric charge table and the continuation inertia of each type of candidate indexes and the daily average electric charge data, and checking and correcting through a confidence interval and a calculation relation of the candidate indexes and the daily average electric charge related data.
4. The method for automatically auditing the costs of base stations based on nearest neighbor algorithm according to claim 1, characterized in that: and the method also comprises weighting treatment on the daily average air temperature.
5. The method for automatically auditing the costs of base stations based on nearest neighbor algorithm according to claim 1, characterized in that: the base station power consumption data further includes an air conditioner power consumption-temperature coefficient T', T ═ 1+ (T-14) for weighting the average daily temperature1.1Where T is the actual outdoor temperature.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010963119.4A CN112258338A (en) | 2020-09-14 | 2020-09-14 | Automatic base station cost auditing method based on nearest neighbor algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010963119.4A CN112258338A (en) | 2020-09-14 | 2020-09-14 | Automatic base station cost auditing method based on nearest neighbor algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112258338A true CN112258338A (en) | 2021-01-22 |
Family
ID=74232183
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010963119.4A Pending CN112258338A (en) | 2020-09-14 | 2020-09-14 | Automatic base station cost auditing method based on nearest neighbor algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112258338A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113177902A (en) * | 2021-04-22 | 2021-07-27 | 陕西铁道工程勘察有限公司 | Inclination model and laser point cloud fusion method based on grid index and spherical tree |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106228398A (en) * | 2016-07-20 | 2016-12-14 | 武汉斗鱼网络科技有限公司 | Specific user's digging system based on C4.5 decision Tree algorithms and method thereof |
| CN106650763A (en) * | 2016-07-05 | 2017-05-10 | 国网内蒙古东部电力有限公司电力科学研究院 | Calculating method of index selection, weight optimization and channel planning of electric power payment channel analysis |
-
2020
- 2020-09-14 CN CN202010963119.4A patent/CN112258338A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106650763A (en) * | 2016-07-05 | 2017-05-10 | 国网内蒙古东部电力有限公司电力科学研究院 | Calculating method of index selection, weight optimization and channel planning of electric power payment channel analysis |
| CN106228398A (en) * | 2016-07-20 | 2016-12-14 | 武汉斗鱼网络科技有限公司 | Specific user's digging system based on C4.5 decision Tree algorithms and method thereof |
Non-Patent Citations (5)
| Title |
|---|
| 周秀: "一种基于数据关联的电费稽核方案", 信息通信, no. 6, pages 2 * |
| 张戈;: "课程推荐预测模型优化方案及数据离散化算法", 计算机系统应用, no. 04 * |
| 彭显刚;林利祥;刘艺;林幕群;郑伟钦;: "数据挖掘技术在电价执行稽查中的应用研究", 电气应用, no. 11, pages 2 * |
| 梁腾;: "基站电费报账支付集中稽核管理研究与实现", 信息通信, no. 06 * |
| 陈金坦;杨燕;周伟雄;钟川;: "数据挖掘技术在交通规费征稽中的应用", 现代计算机(专业版), no. 10 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113177902A (en) * | 2021-04-22 | 2021-07-27 | 陕西铁道工程勘察有限公司 | Inclination model and laser point cloud fusion method based on grid index and spherical tree |
| CN113177902B (en) * | 2021-04-22 | 2024-01-26 | 陕西铁道工程勘察有限公司 | Inclined model and laser point cloud fusion method based on grid index and ball tree |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2022135265A1 (en) | Failure warning and analysis method for reservoir dispatching rules under effects of climate change | |
| CN108520357B (en) | Method and device for judging line loss abnormality reason and server | |
| CN106022509B (en) | Consider the Spatial Load Forecasting For Distribution method of region and load character double differences | |
| CN104809658B (en) | A kind of rapid analysis method of low-voltage distribution network taiwan area line loss | |
| CN112258337B (en) | Self-complement correction base station energy consumption model prediction method | |
| CN107437135B (en) | Novel energy storage type selection method | |
| CN103065050A (en) | Health level judging method of information system during operation maintenance period | |
| CN113159841A (en) | Method for compiling newly-built substation project estimation | |
| CN111709668A (en) | Power grid equipment parameter risk identification method and device based on data mining technology | |
| CN108989092A (en) | A kind of wireless network predicting method, electronic equipment and storage medium | |
| Dong | Combining unsupervised and supervised learning for asset class failure prediction in power systems | |
| CN107145986B (en) | Charging electric quantity prediction method and device | |
| CN115186964A (en) | Power distribution network line risk prediction system and method considering subjective evaluation data | |
| CN112258338A (en) | Automatic base station cost auditing method based on nearest neighbor algorithm | |
| CN114676749A (en) | Power distribution network operation data abnormity judgment method based on data mining | |
| CN107274025B (en) | System and method for realizing intelligent identification and management of power consumption mode | |
| CN114139931A (en) | Enterprise data evaluation method and device, computer equipment and storage medium | |
| CN112508254B (en) | Method for determining investment prediction data of transformer substation engineering project | |
| CN111784379B (en) | Estimation method and device for electric charge after-payment and screening method and device for abnormal cases | |
| CN109508820A (en) | Campus electricity demand forecasting modeling method based on differentiation modeling | |
| CN112101673A (en) | Power grid development trend prediction method and system based on hidden Markov model | |
| CN116307844A (en) | Low-voltage transformer area line loss evaluation analysis method | |
| CN116151799A (en) | BP neural network-based distribution line multi-working-condition fault rate rapid assessment method | |
| CN115545342A (en) | Risk prediction method and system for enterprise electric charge recovery | |
| CN109255470A (en) | A kind of non-item class forecasting of cost method of transformer equipment based on big data analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |