CN112258337B - Self-complement correction base station energy consumption model prediction method - Google Patents
Self-complement correction base station energy consumption model prediction method Download PDFInfo
- Publication number
- CN112258337B CN112258337B CN202010961685.1A CN202010961685A CN112258337B CN 112258337 B CN112258337 B CN 112258337B CN 202010961685 A CN202010961685 A CN 202010961685A CN 112258337 B CN112258337 B CN 112258337B
- Authority
- CN
- China
- Prior art keywords
- base station
- power consumption
- model
- average
- correction
- 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.)
- Active
Links
- 238000012937 correction Methods 0.000 title claims abstract description 37
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000005265 energy consumption Methods 0.000 title claims abstract description 22
- 238000012417 linear regression Methods 0.000 claims abstract description 24
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 22
- 239000013598 vector Substances 0.000 claims abstract description 14
- 238000007621 cluster analysis Methods 0.000 claims abstract description 5
- 230000005611 electricity Effects 0.000 claims description 32
- 230000007812 deficiency Effects 0.000 claims description 14
- 238000007781 pre-processing Methods 0.000 claims description 8
- 230000002159 abnormal effect Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000001502 supplementing effect Effects 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 3
- 230000007547 defect Effects 0.000 abstract description 5
- 230000000295 complement effect Effects 0.000 abstract description 2
- 238000004891 communication Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 4
- 238000007418 data mining Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 2
- 239000000047 product Substances 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 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
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
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
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Biophysics (AREA)
- Development Economics (AREA)
- Biomedical Technology (AREA)
- Game Theory and Decision Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Entrepreneurship & Innovation (AREA)
- Software Systems (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a self-complement correction base station energy consumption model prediction method, which comprises the steps of firstly, carrying out linear regression modeling on base stations with the number of samples of each payment record reaching a preset number by using average daily temperature and average daily power consumption, removing samples with overlarge deviation from a predicted value, carrying out sample complement on the base stations with the number of effective samples of the rest payment records, carrying out ball tree modeling on a processed feature vector set by using a radius-based ball tree algorithm, and carrying out cluster analysis by using a nearest neighbor algorithm with a variable radius distance. Taking the payment records of the base station and the nearest base station as samples to carry out linear regression to generate a temperature energy consumption model, and finally, predicting by using the corrected model; the invention carries out automatic prediction for the electric charge of the current base station, overcomes the defect of manual auditing, improves auditing efficiency and scientific accuracy, and saves labor cost.
Description
Technical Field
The invention relates to the technical field of management and control of energy consumption in the communication industry, relates to the field of data mining of computers, and aims at the technical solution of power consumption prediction of a communication base station, and the technical solution is composed of a series of data processing methods and model algorithms, in particular to a self-complement correction base station energy consumption model prediction method.
Background
The current situation of the base station power consumption prediction of the communication carrier regarding the base station power consumption prediction technology is:
when financial staff of a communication operator make a budget index, the budget index of the electric charges of all base stations in the managed area of each month of the budget year is needed, as the financial staff often does not have scientific experimental knowledge on the power consumption influence factors of the base stations and the principle thereof, a reasonable power consumption prediction model is difficult to construct, and often, according to the conventional average value prediction, special conditions which do not consider different months exist, so that a reasonable cost management and control target is difficult to scientifically and effectively execute.
The current method for predicting the electricity charge of the base station has the defects that:
(1) Mean prediction using the last three months: because the factors influencing the power consumption of the base station are numerous and complex, the most main power consumption equipment is communication main equipment and an air conditioner, and the main factors influencing the power consumption of the air conditioner are temperature, and the temperature changes along with the temperature; the power consumption of the communication main equipment is constant relative to each month; if the average value of the electric charge of the last three months is used as the basis of the electric charge budget of each month of one year in the following year at the end of the year, the electric charge of each month is different due to the power consumption of the air conditioner in the areas with large temperature difference in four seasons. The method is predicted by low temperature in winter, and the power consumption of the air conditioner is not added, so that the predicted result is excessively low.
(2) Prediction of the electricity charge plus the change coefficient in the same period of year is used: the method is feasible under the condition that the power consumption equipment of the base station and the temperature of two years are affected, but if the variation difference is large or the electricity fee in the past year has problems, the correction and the prediction are difficult;
(3) The human brain natural subjectivity and calculation difficulty errors exist in the manual electric charge prediction calculation;
(4) The financial budget of the general cost budget is difficult to observe with the power consumption model scientifically demonstrated by experts and the budget flow specification managed by companies;
disclosure of Invention
The invention provides a self-complement correction base station energy consumption model prediction method for solving the problems, which aims to solve the problems that a reasonable range is calculated for the current electricity charge of a base station by a large number of data mining technologies and a scientific calculation algorithm of the electricity consumption of the base station to serve as a specific basis for the current prediction, and the automatic prediction is carried out on the electricity charge by adopting computer automation, so that the defect of manual auditing is overcome, the auditing efficiency and scientific accuracy are improved, and the labor cost is saved.
In order to achieve the above purpose, the invention adopts the following technical scheme:
the self-complement correction base station energy consumption model prediction method comprises the following steps:
1) Collecting base station power consumption and base station attribute data, wherein the data comprise ammeter readings and electric charge payment sheets;
2) Selecting a base station historical payment sample and base station attribute data, and preprocessing, wherein the preprocessing comprises weighting the average temperature by adopting the power consumption of an air conditioner and a temperature coefficient T ', and T' =1+ (T-14) 1.1 Wherein T is the actual outdoor temperature, and then the average temperature and average power consumption are calculated;
3) Generating a linear regression prediction model of average daily power consumption for a base station with enough historical payment samples to reach a set number, wherein the linear regression prediction model Y=kX+b is established according to the average daily temperature X of the base station and the average daily power consumption Y; self-correcting the prediction model, and removing samples with overlarge deviation from the predicted value;
the method comprises the following steps of: (1) Extracting base station attributes related to power consumption and corrected historical day average electricity fees for the base stations as clustered feature vectors; (2) carrying out deficiency correction and standardized pretreatment on the feature vector; (3) Calculating average distances of points of all the feature vectors, and taking the average distances as a basis for setting the radius; (4) Constructing a spherical tree coding model for the feature vectors of all base stations by using a spherical tree algorithm in the frNN nearest neighbor algorithm; (5) Using the spherical tree coding model to perform cluster analysis on nearest neighbor base stations within the radius distance of the base stations with insufficient corrected samples; supplementing samples to the base stations with the number smaller than the set number by using the payment records of the nearest neighbor base stations provided by the clustering analysis result, and finally establishing a linear regression prediction model;
4) Repeating step 2) and step 3) more than once to self-correct the sample effectiveness of the base station energy consumption model; if the k coefficient in the linear regression prediction model is smaller than 0, the coefficient correction is required to be performed as follows: k=0, b=the average of the power consumption to be predicted;
5) And predicting the power consumption of the base station by using the corrected linear regression prediction model.
In the self-complement correction base station energy consumption model prediction method, collecting base station power consumption and base station attribute data comprises the steps of identifying ammeter reading by using a Mask-FCNN-based target detection image identification algorithm, extracting data by using scanning data of an electric charge bill fixed template, and importing the data by using a related system interface.
In the self-complement correction base station energy consumption model prediction method, preprocessing comprises the steps of removing abnormal data and deficiency data and correcting the deficiency data; the abnormal data removal comprises comparing daily average electric charges of the base stations and the base stations of the same type, removing outliers, and removing abnormally high electric charges according to the sum of power consumption equipment of the base stations and the upper limit of daily power consumption; the deficiency correction of the deficiency data includes deficiency correction based on each correlation and continuous inertia, and the correction is verified by a calculation relation between the confidence interval and the correlation data.
In the self-complement correction base station energy consumption model prediction method, the self-correction method for the prediction model comprises the following steps: and calculating the average value and standard deviation of residual errors of the power consumption predicted value and the real Y value of the model, and removing outlier samples of |real value-predicted value| > residual error average value+residual error standard deviation.
The beneficial effects of the invention are that:
1. according to the invention, a reasonable range is calculated for the current electricity charge of the base station by a large number of data mining technologies and a scientific calculation algorithm of the electricity consumption of the base station to serve as a specific basis for current prediction, and computer automation is adopted to automatically predict the electricity charge, so that the defect of manual auditing is overcome, meanwhile, the auditing efficiency and scientific accuracy are improved, and the labor cost is saved.
2. The related factors are transformed, such as the temperature is weighted according to the average temperature of the weather, and the power consumption of the air conditioner is actually influenced by the weather, namely, the air conditioner is generally set to be automatically started when the room temperature reaches a certain temperature and automatically closed after the temperature falls back, and the power consumption indexes of the air conditioner are different at different temperatures. Therefore, the temperature is treated as follows:
T′＝1+(T-14) 1.1
so that the factor is that the change in value and the change in Y value more closely match the calculated correlation.
3. Because of the problem of low data quality in practice, the corresponding sample with overlarge residual error is removed by adopting a model self-correction mode, so that abnormal data of outliers can be effectively removed, and the fitting degree of a model is improved;
4. for the characteristic selection of the base station, not only the attribute of the base station is considered, but also the average daily electricity charge value of the past payment is added, so that the problem of insufficient data quality is avoided.
5. And clustering the base stations based on a radius nearest neighbor algorithm, taking payment records of similar base stations into a sample for linear fitting, and effectively solving the problem of insufficient modeling samples for single base stations.
6. Not only is the model corrected by removing outlier samples, but also the coefficient of the model is corrected by normalization, for example, the average value method is adopted for processing and correcting the unreasonable condition that the k coefficient of linear regression is smaller than 0.
Drawings
FIG. 1 is a schematic diagram of the steps performed in the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Referring to fig. 1, in order to help communication operators to scientifically and effectively control the operation cost of a base station, a base station cost control personnel needs to make budget indexes for the cost of electricity charge and the like so as to realize that the cost is in a controllable range. The invention aims to establish an automatic electricity fee prediction method based on the combination of a base station power consumption scientific calculation model and enterprise fee management regulations.
The invention discloses a self-complement correction base station energy consumption model prediction method, which comprises the steps of firstly carrying out linear regression modeling on the base station use average temperature and average power consumption of which the number of samples of each payment record reaches a preset number, removing samples with overlarge deviation from a predicted value, and carrying out sample complement on the base stations with insufficient effective number of the remaining payment records. And collecting data such as power consumption, average value of the electric charge and base station attribute of the historical payment related to the electric charge of the base station, and then carrying out preprocessing and standardization processing according to the basic data. And then using a radius-based Ball Tree (Ball-Tree) algorithm to carry out Ball Tree modeling on the processed feature vector set, finding out the nearest base station set of the base station by utilizing a nearest neighbor algorithm with variable radius distance for the base station with the historical payment number lower than a set threshold value, and selecting N clusters (parameter preset) closest to the base station. And carrying out linear regression on the base station and the payment records of N nearest base stations serving as samples to generate a temperature energy consumption model, repeating the step of model correction once again, checking and correcting the coefficient compliance of the model, and finally predicting by using the corrected model.
The main implementation steps of the electric charge prediction algorithm are as follows:
step 1: and collecting the base station power consumption and base station attribute related data from multiple channels such as ammeter reading, electric charge payment bill, machine room lease contract, base station equipment and the like.
Key details of implementation: (1) reading an ammeter: in order to prevent the base station maintainer from virtually reporting the power consumption, the mobile phone is required to photograph the electricity meter degree, and then a Mask-FCNN-based target detection image recognition algorithm is used for recognizing the electricity meter degree and for detecting and correcting the problem of manually inputted electricity meter degree. (2) Data mobile phones such as electric charge payment bill, contract, base station equipment and the like: the method is to collect and summarize through technologies such as data extraction of fixed templates such as form scanning and other relevant system interface importing channels.
Step 2: the method comprises the steps of manually selecting data in historical payment and base station attributes related to base station power consumption and electricity charge, firstly removing abnormality, deficiency, correction and the like, then extracting the data, calculating the average daily temperature and average daily power consumption of each payment record of the base station, and collecting the average daily temperature and average daily power consumption into reference set alternative factors.
Key details of implementation: (1) removing abnormal data: the base station and similar historical paid daily average electric charge are used for comparing and removing outliers, and meanwhile, abnormally high electric charge can be removed according to the sum of power consumption equipment and the upper limit of daily power consumption, if the quality of data such as power and the like is guaranteed. (2) deficiency correction of deficiency data: the correlation of each dimension table and the continuous inertia of various data are needed to be filled up, and the correction is checked and corrected by the calculation relation between the confidence interval and the related data.
Step 3: and establishing and generating a linear regression prediction model of the average daily power consumption for the base stations with enough historical payment samples to reach the set number, and carrying out self-correction on the prediction model.
Key details of implementation: (1) For a base station with the number of the payment record samples reaching the minimum number of samples required by constructing a linear regression model (the set number is generally N > =5), calculating the average daily temperature X and average daily power consumption Y of a payment period according to the end of the period, and constructing a linear regression prediction model Y=kX+b; the self-correction method for the prediction model comprises the following steps: and calculating the average value and standard deviation of residual errors of the power consumption predicted value and the real Y value of the model, and removing outlier samples of |real value-predicted value| > residual error average value+residual error standard deviation.
Step 4: and carrying out cluster analysis on the base stations with the corrected sample numbers smaller than the set number by adopting a frNN algorithm, and complementing the effective samples.
Key details of implementation: (1) Extracting the attribute of the base station related to the power consumption and the index related to payment from the base stations to construct a feature vector; (2) Preprocessing such as feature vector deficiency correction and standardization; (3) Calculating average distances of points of all the feature vectors, and taking the average distances as a basis for setting the radius; (4) Constructing a spherical tree coding model for the feature vectors of all base stations by using a spherical tree algorithm in the frNN nearest neighbor algorithm; (5) And (3) performing cluster analysis on nearest neighbor base stations within the radius distance of the base stations with insufficient corrected samples by using a spherical tree coding model, and then clustering by using the method of the step (5) to find payment records of the base stations with characteristics similar to each characteristic of the base stations as supplementary samples of the base stations.
The principle of the ball tree algorithm is introduced: a ball tree recursively divides the data into nodes defined centered at CCC and having a radius r, such that each of the nodes is within the hypersphere defined by r and CCC. The triangle inequality is utilized to reduce candidate points of one neighbor point search:
∣x+y∣≤∣x∣+∣y∣
in this arrangement, it is sufficient to calculate only the distance between the test point and the centroid to determine the upper and lower bounds of the distances to all points within the node. Because of the spherical geometry of the spherical tree nodes, although the actual performance is highly dependent on the structure of the training data, its performance in high dimensions is superior to KD-tree.
Step 5: and (3) supplementing samples to the base stations with the sample numbers smaller than the set number by using the payment records of the M nearest neighbor base stations provided by the clustering analysis result in the step (4), and then establishing a linear regression prediction model.
Key details of implementation: (1) Aiming at the base stations with less than set number of all the corrected samples, using the corrected reserved effective payment records of M nearest neighbor base stations clustered in the step 3 to supplement the samples, wherein M can be set manually according to the data quantity; (2) And performing linear regression modeling on the sample set of the base station with the original sample number less than the set number after the supplementary expansion.
And then carrying out linear regression on the effective payment sample to construct a temperature energy consumption prediction model.
Step 6: and repeating the steps 2, 3 and 4 once again to perform sample validity of the self-repairing model.
Step 7: and correcting the temperature energy consumption model according to the rationality range of the index value.
Key details of implementation: for k coefficients in the linear regression prediction model that are less than 0 (typically due to data quality or other factors), the coefficient correction needs to be: k=0, b=the average of the power consumption to be predicted.
Step 8: and predicting the future power consumption of the base station by using the corrected temperature energy consumption model.
Key details of implementation: in general, in actual prediction of electricity charge, the electricity consumption needs to be calculated by using the solar temperature of the same date as the year as an independent variable to carry into a corrected model, and then the total electricity quantity is summarized according to regional months and the like, and then the total electricity quantity is multiplied by the local electricity price to obtain the electricity charge to be predicted.
In the technical scheme and logic algorithm flow of the base station electricity charge prediction method, the electricity consumption of the base station electricity charge is influenced by a plurality of uncertain factors in practice, the base station electricity charge prediction method extracts the main influence factor of the largest variable electricity consumption factor, namely the electricity consumption of an air conditioner, to carry out linear regression, models each base station independently, avoids the influence of complex factors such as different base station equipment configurations and heat transfer coefficients of the environment, and can enable the electricity charge prediction method to be scientific, so that the method can be popularized and applied, and finally, the customer management goal of cost reduction and synergy can be realized.
Any well-proven and accepted method generally needs to be implemented on a specific product carrier to be popularized and used in actual economic activities if actual social and economic benefits are to be generated. The invention can be embedded into related products such as an intelligent data analysis platform, and the corresponding auditing function can help the cost management and control personnel of the clients to improve the scientific and reasonable level and the operation efficiency of the budget of the electric charge.
The method overcomes the defects of the common base station electric charge prediction method, carries out conversion treatment on related factors, such as temperature, not only carries out weighting treatment according to the average air temperature of the day of weather, but also carries out automatic opening and automatic closing after the temperature is fallen back according to the actual influence principle of the air conditioner on the power consumption of the air conditioner, wherein the air conditioner is generally arranged at the room temperature and is automatically opened when the temperature is up to a certain temperature, and the indexes of the power consumption of the air conditioner are different at different temperatures. Therefore, the temperature is treated as follows:
T′＝1+(T-14) 1.1 (T-is the actual outdoor temperature, T' -is the power consumption of the air conditioner after treatment-temperature coefficient)
So that the factor is that the change in value and the change in Y value more closely match the calculated correlation.
Because of the problem of low data quality in practice, the corresponding sample with overlarge residual error is removed by adopting a model self-correction mode, so that abnormal data of outliers can be effectively removed, and the fitting degree of a model is improved;
for the characteristic selection of the base station, not only the attribute of the base station is considered, but also the average daily electricity charge value of the past payment is added, so that the problem of insufficient data quality is avoided.
And clustering the base stations based on a radius nearest neighbor algorithm, taking payment records of similar base stations into a sample for linear fitting, and effectively solving the problem of insufficient modeling samples for single base stations.
Not only is the model corrected by removing outlier samples, but also the coefficient of the model is corrected by normalization, for example, the average value method is adopted for processing and correcting the unreasonable condition that the k coefficient of linear regression is smaller than 0.
The foregoing is a further detailed description of the invention in connection with specific embodiments, and is not intended to limit the practice of the invention thereto. Several equivalent substitutions or obvious modifications will occur to those skilled in the art to which this invention pertains without departing from the spirit of the invention, and the same should be considered to be within the scope of this invention as defined in the appended claims.
Claims (4)
1. A self-complement correction base station energy consumption model prediction method is characterized by comprising the following steps:
1) Collecting base station power consumption and base station attribute data, wherein the data comprise ammeter readings and electric charge payment sheets;
2) Selecting a base station historical payment sample and base station attribute data, and preprocessing, wherein the preprocessing comprises weighting the average temperature by adopting the power consumption of an air conditioner and a temperature coefficient T ', and T' =1+ (T-14) 1.1 Wherein T is the actual outdoor temperature, and then the average temperature and average power consumption are calculated;
3) Generating a linear regression prediction model of average daily power consumption for a base station with enough historical payment samples to reach a set number, wherein the linear regression prediction model Y=kX+b is established according to the average daily temperature X of the base station and the average daily power consumption Y; self-correcting the prediction model, and removing samples with overlarge deviation from the predicted value;
the method comprises the following steps of: (1) Extracting base station attributes related to power consumption and corrected historical day average electricity fees for the base stations as clustered feature vectors; (2) carrying out deficiency correction and standardized pretreatment on the feature vector; (3) Calculating average distances of points of all the feature vectors, and taking the average distances as a basis for setting the radius; (4) Constructing a spherical tree coding model for the feature vectors of all base stations by using a spherical tree algorithm in the frNN nearest neighbor algorithm; (5) Using the spherical tree coding model to perform cluster analysis on nearest neighbor base stations within the radius distance of the base stations with insufficient corrected samples; supplementing samples to the base stations with the number smaller than the set number by using the payment records of the nearest neighbor base stations provided by the clustering analysis result, and finally establishing a linear regression prediction model; 4) Repeating step 2) and step 3) more than once to self-correct the sample effectiveness of the base station energy consumption model; if the k coefficient in the linear regression prediction model is smaller than 0, the coefficient correction is required to be performed as follows: k=0, b=the average of the power consumption to be predicted;
5) And predicting the power consumption of the base station by using the corrected linear regression prediction model.
2. The self-complement-correction base station energy consumption model prediction method according to claim 1, characterized by: collecting the power consumption and the attribute data of the base station comprises the steps of identifying the electric meter reading by using a Mask-FCNN-based target detection image identification algorithm, extracting data by using the scanning data of the electric charge payment bill fixed template, and importing the data by using a related system interface.
3. The self-complement-correction base station energy consumption model prediction method according to claim 1, characterized by: the preprocessing comprises the steps of removing abnormal data and supplementing and correcting missing data; the abnormal data removal comprises comparing daily average electric charges of the base stations and the base stations of the same type, removing outliers, and removing abnormally high electric charges according to the sum of power consumption equipment of the base stations and the upper limit of daily power consumption; the deficiency correction of the deficiency data includes deficiency correction based on each correlation and continuous inertia, and the correction is verified by a calculation relation between the confidence interval and the correlation data.
4. The self-complement-correction base station energy consumption model prediction method according to claim 1, characterized by: the self-correction method for the prediction model comprises the following steps: and calculating the average value and standard deviation of residual errors of the power consumption predicted value and the real Y value of the model, and removing outlier samples of |real value-predicted value| > residual error average value+residual error standard deviation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010961685.1A CN112258337B (en) | 2020-09-14 | 2020-09-14 | Self-complement correction base station energy consumption model prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010961685.1A CN112258337B (en) | 2020-09-14 | 2020-09-14 | Self-complement correction base station energy consumption model prediction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112258337A CN112258337A (en) | 2021-01-22 |
| CN112258337B true CN112258337B (en) | 2024-03-12 |
Family
ID=74231264
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010961685.1A Active CN112258337B (en) | 2020-09-14 | 2020-09-14 | Self-complement correction base station energy consumption model prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112258337B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113469189B (en) * | 2021-09-02 | 2021-11-19 | 国网江西省电力有限公司供电服务管理中心 | Method, system and device for filling missing values of power utilization acquisition data |
| CN113824148A (en) * | 2021-10-19 | 2021-12-21 | 阳光电源股份有限公司 | Power generation system and control method thereof |
| CN114049017B (en) * | 2021-11-16 | 2023-07-04 | 重庆瑞盾科技发展有限公司 | Electric charge cost management method and system based on AC/DC intelligent switch ammeter |
| CN115481813B (en) * | 2022-10-09 | 2024-05-31 | 武汉理工大学 | New energy bus energy consumption prediction method based on machine learning |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009204195A (en) * | 2008-02-26 | 2009-09-10 | Mitsubishi Heavy Ind Ltd | Air conditioning system and power consumption estimating device for building air-conditioning equipment |
| CN103024761A (en) * | 2011-09-26 | 2013-04-03 | 艾默生网络能源有限公司 | Establishing method for energy consumption model of base station, and energy consumption predicating method and device |
| CN105574607A (en) * | 2015-12-10 | 2016-05-11 | 四川省电力公司供电服务中心 | Electricity market monthly electricity utilization prediction method |
| CN105959975A (en) * | 2016-06-22 | 2016-09-21 | 湖南大学 | Automatic assessment method for energy saving quantity of large-scale base station energy saving project |
| CN106228282A (en) * | 2016-07-13 | 2016-12-14 | 中国矿业大学(北京) | The computational methods of electricity consumption based on weather |
| CN106408223A (en) * | 2016-11-30 | 2017-02-15 | 华北电力大学（保定） | Short-term load prediction based on meteorological similar day and error correction |
| CN107315884A (en) * | 2017-07-04 | 2017-11-03 | 北京首钢自动化信息技术有限公司 | A kind of building energy consumption modeling method based on linear regression |
| WO2017212880A1 (en) * | 2016-06-09 | 2017-12-14 | 株式会社日立製作所 | Data prediction system and data prediction method |
| CN108510006A (en) * | 2018-04-08 | 2018-09-07 | 重庆邮电大学 | A kind of analysis of business electrical amount and prediction technique based on data mining |
| CN109063366A (en) * | 2018-08-22 | 2018-12-21 | 深圳市建筑科学研究院股份有限公司 | A kind of building performance online data preprocess method based on time and spatial weighting |
| CN111626518A (en) * | 2020-05-29 | 2020-09-04 | 上海交通大学 | Urban daily water demand online prediction method based on deep learning neural network |
-
2020
- 2020-09-14 CN CN202010961685.1A patent/CN112258337B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009204195A (en) * | 2008-02-26 | 2009-09-10 | Mitsubishi Heavy Ind Ltd | Air conditioning system and power consumption estimating device for building air-conditioning equipment |
| CN103024761A (en) * | 2011-09-26 | 2013-04-03 | 艾默生网络能源有限公司 | Establishing method for energy consumption model of base station, and energy consumption predicating method and device |
| CN105574607A (en) * | 2015-12-10 | 2016-05-11 | 四川省电力公司供电服务中心 | Electricity market monthly electricity utilization prediction method |
| WO2017212880A1 (en) * | 2016-06-09 | 2017-12-14 | 株式会社日立製作所 | Data prediction system and data prediction method |
| CN105959975A (en) * | 2016-06-22 | 2016-09-21 | 湖南大学 | Automatic assessment method for energy saving quantity of large-scale base station energy saving project |
| CN106228282A (en) * | 2016-07-13 | 2016-12-14 | 中国矿业大学(北京) | The computational methods of electricity consumption based on weather |
| CN106408223A (en) * | 2016-11-30 | 2017-02-15 | 华北电力大学（保定） | Short-term load prediction based on meteorological similar day and error correction |
| CN107315884A (en) * | 2017-07-04 | 2017-11-03 | 北京首钢自动化信息技术有限公司 | A kind of building energy consumption modeling method based on linear regression |
| CN108510006A (en) * | 2018-04-08 | 2018-09-07 | 重庆邮电大学 | A kind of analysis of business electrical amount and prediction technique based on data mining |
| CN109063366A (en) * | 2018-08-22 | 2018-12-21 | 深圳市建筑科学研究院股份有限公司 | A kind of building performance online data preprocess method based on time and spatial weighting |
| CN111626518A (en) * | 2020-05-29 | 2020-09-04 | 上海交通大学 | Urban daily water demand online prediction method based on deep learning neural network |
Non-Patent Citations (6)
| Title |
|---|
| 基于人工神经网络方法的通信基站能耗标杆建立与分析;杨天剑等;《北京邮电大学学报(社会科学版)》;第17卷(第06期);第58-63页 * |
| 基于历史误差模型和局域临域校正的预测融合算法;王邑等;《仪器仪表学报》;第32卷(第10期);第2283-2289页 * |
| 基于模糊聚类的蜂窝小区业务量数据填补算法;杨苹等;《计算机工程》;第37卷(第13期);第259-261页 * |
| 基于聚类算法的通信基站能耗标杆建立与分析;杨天剑等;《移动通信》;第39卷(第18期);第92-96页 * |
| 考虑气象因素的短期负荷预测模型研究;朱祥和;《数学的实践与认识》;第48卷(第03期);第131-143页 * |
| 面向季节性时空数据的预测式循环网络及其在城市计算中的应用;张建晋等;《计算机学报》;第43卷(第02期);第286-302页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112258337A (en) | 2021-01-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112258337B (en) | Self-complement correction base station energy consumption model prediction method | |
| CN110223196B (en) | Anti-electricity-stealing analysis method based on typical industry feature library and anti-electricity-stealing sample library | |
| CN110097297B (en) | Multi-dimensional electricity stealing situation intelligent sensing method, system, equipment and medium | |
| Guerrero et al. | Non-technical losses reduction by improving the inspections accuracy in a power utility | |
| CN106022509B (en) | Consider the Spatial Load Forecasting For Distribution method of region and load character double differences | |
| CN112699913A (en) | Transformer area household variable relation abnormity diagnosis method and device | |
| CN109740648A (en) | Electric load disorder data recognition method, apparatus and computer equipment | |
| CN108388974A (en) | Top-tier customer Optimum Identification Method and device based on random forest and decision tree | |
| CN110687361A (en) | Method and system for identifying phase sequence of low-voltage distribution network area | |
| CN111667135B (en) | Load structure analysis method based on typical feature extraction | |
| CN107832927A (en) | 10kV circuits line based on gray relative analysis method becomes relation evaluation method | |
| CN116073436B (en) | Capacity optimization control method for photovoltaic new energy power system | |
| CN110210670A (en) | A kind of prediction technique based on power-system short-term load | |
| CN111709668A (en) | Power grid equipment parameter risk identification method and device based on data mining technology | |
| CN115660262B (en) | Engineering intelligent quality inspection method, system and medium based on database application | |
| CN115358437A (en) | Power supply load prediction method based on convolutional neural network | |
| CN116467648A (en) | Early monitoring method for nonlinear platform power failure based on Internet of things table | |
| CN110968703A (en) | Method and system for constructing abnormal metering point knowledge base based on LSTM end-to-end extraction algorithm | |
| Grigoras et al. | Processing of smart meters data for peak load estimation of consumers | |
| CN109977984A (en) | Stealing user's judgment method based on support vector machines | |
| LU503778B1 (en) | The method for wind speed downscaling of wind farms based on super-resolution reconstruction technology | |
| CN112508278A (en) | Multi-connected system load prediction method based on evidence regression multi-model | |
| CN114676931B (en) | Electric quantity prediction system based on data center technology | |
| CN111091223A (en) | Distribution transformer short-term load prediction method based on Internet of things intelligent sensing technology | |
| CN115759395A (en) | Training of photovoltaic detection model, detection method of photovoltaic power generation and related device |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |