CN117522207A - Layered identification method for energy storage installation potential of industrial enterprise under user side view angle - Google Patents
Layered identification method for energy storage installation potential of industrial enterprise under user side view angle Download PDFInfo
- Publication number
- CN117522207A CN117522207A CN202311506635.4A CN202311506635A CN117522207A CN 117522207 A CN117522207 A CN 117522207A CN 202311506635 A CN202311506635 A CN 202311506635A CN 117522207 A CN117522207 A CN 117522207A
- Authority
- CN
- China
- Prior art keywords
- load
- user
- data
- index
- energy storage
- 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
- 238000000034 method Methods 0.000 title claims abstract description 134
- 238000004146 energy storage Methods 0.000 title claims abstract description 96
- 238000009434 installation Methods 0.000 title claims abstract description 19
- 238000004458 analytical method Methods 0.000 claims abstract description 89
- 238000012545 processing Methods 0.000 claims abstract description 22
- 238000004140 cleaning Methods 0.000 claims abstract description 11
- 230000002159 abnormal effect Effects 0.000 claims abstract description 4
- 238000010219 correlation analysis Methods 0.000 claims abstract description 4
- 238000002360 preparation method Methods 0.000 claims abstract description 4
- 230000005611 electricity Effects 0.000 claims description 46
- 238000004422 calculation algorithm Methods 0.000 claims description 33
- 238000011156 evaluation Methods 0.000 claims description 20
- 238000010606 normalization Methods 0.000 claims description 18
- 238000010276 construction Methods 0.000 claims description 11
- 238000007405 data analysis Methods 0.000 claims description 11
- 238000009826 distribution Methods 0.000 claims description 9
- 230000010354 integration Effects 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000000556 factor analysis Methods 0.000 claims description 7
- 230000005484 gravity Effects 0.000 claims description 6
- 238000003064 k means clustering Methods 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 238000005065 mining Methods 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000013507 mapping Methods 0.000 claims description 3
- 238000011425 standardization method Methods 0.000 claims description 3
- 238000002759 z-score normalization Methods 0.000 claims description 3
- 230000008030 elimination Effects 0.000 claims description 2
- 238000003379 elimination reaction Methods 0.000 claims description 2
- 238000007418 data mining Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 7
- 230000002354 daily effect Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 6
- 238000013210 evaluation model Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 230000003247 decreasing effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 239000006185 dispersion Substances 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 238000009960 carding Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000004138 cluster model Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000000265 homogenisation Methods 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000005381 potential energy Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
- 238000005303 weighing 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Tourism & Hospitality (AREA)
- Evolutionary Computation (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Quality & Reliability (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Operations Research (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Game Theory and Decision Science (AREA)
- General Health & Medical Sciences (AREA)
Abstract
The invention relates to a layered identification method for energy storage installation potential of an industrial enterprise at a user side view angle, which comprises the following steps: 1. data preparation and pretreatment: collecting required data, including user profile information and load information of an industrial enterprise; the method comprises the steps of completing the cleaning work of collected data, including processing missing values, abnormal values and noise data, and carrying out standardized processing on the data; 2. modeling analysis: carrying out industrial load characteristic analysis, user load adjustable potential analysis and user load level analysis, and constructing indexes; and calculating the weight of each index by a correlation analysis method in combination with the index information of the user, calculating the comprehensive score of the user according to the method that a plurality of indexes are multiplied by the weights and summed, and layering the user according to the comprehensive score.
Description
Technical Field
The invention relates to the technical field of large data analysis of power grids, in particular to a layered identification method for energy storage installation potential of an industrial enterprise at a user side view angle.
Background
The industrial enterprise is an extremely important component part on the user side, at present, for the scheme of the energy storage mode on the user side of the industrial enterprise, such as the scheme of responding to time-of-use electricity price participation peak valley arbitrage, responding to two electricity price manufacturing participation maximum demand management, or combining two types to construct an energy storage configuration model combining maximum demand management and peak clipping and valley filling, the scheme of responding to the energy storage participation demand on the user side of the industrial enterprise is less, and in addition, along with the continuous enrichment and accumulation of data, the condition of carrying out the energy storage potential on the user side of the industrial enterprise is mature gradually, so a method for exploring the layered identification and evaluation of the energy storage potential from the view angle of the user side of the industrial enterprise is needed.
The techniques for identifying and evaluating the energy storage potential at the user side are mainly two types at present: one type is data analysis and the other type is data mining.
(1) Data analysis
The method is based on the interpretation, sorting, inspection and analysis processes of existing datasets to gain insight into the data. The method uses statistical methods and related technologies to reveal modes, trends, associations and anomalies in the data, and interprets and summarizes the data by means of statistical inference, visualization and the like so as to support decision making and promote business development.
For example, it is proposed to perform user-side energy storage assessment by "building a distributed energy storage convergence potential assessment model", which is specifically: firstly, determining basic elements to be evaluated, namely, taking sensitive factors which influence distributed energy storage convergence as index layer parameters; secondly, 6 representative criterion factors are selected, wherein the criterion factors comprise dynamic response capability factors, capacity support capability factors, power support capability factors, effective convergence time ratio, system stability factors and system reliability factors; and finally, calculating the convergence potential index of the distributed energy storage system by adopting an analytic hierarchy process, and taking the convergence potential index as an evaluation method of the convergence providing output sequence of the distributed energy storage participation nodes.
The literature also proposes that identification and evaluation of the energy storage potential of the user side are carried out by a data-driven user side energy storage installation potential layering evaluation method, and the specific method is as follows: firstly, carrying out priority assessment, namely carrying out sudden accidents and grid breakdown on a power system, wherein some important parties are greatly influenced, and the user side energy storage installation potential of the important parties is maximum and the priority assessment is carried out; and then carrying out fine evaluation and constructing an evaluation model, wherein the fine evaluation starts from the element which can attract the user most, the element evaluates from five indexes of peak-valley difference, voltage fluctuation, distribution capacity, enterprise electricity charge and reliability, and finally, an F value is obtained by a weighted summation method, the F value represents the potential of the user for installing energy storage, and the larger the F value is, the larger the potential of the user for installing energy storage is.
(2) Data mining
The data mining is a process of finding hidden patterns, associations, trends, anomalies and knowledge from a large-scale data set, automatically finding rules and patterns in the data set by using statistical, machine learning, artificial intelligence and other technical methods, helping to find potential hole finding and unknown knowledge, and providing decision support. For example, the method comprises the steps of extracting a user load characteristic curve, clustering an original load curve, calculating corresponding load density according to basic characteristics of characteristic index analysis curves, extracting power utilization change characteristics of users before and after additional energy storage, analyzing the power utilization potential of the users participating in energy storage, and utilizing a data mining mode to mine potential energy storage users in a regional power grid.
Chinese patent (application number: 20211139939. X; application date 2021.11.19) discloses a method for acquiring and distributing power values in an industrial production area based on a power distribution network, but no proposal is made for energy storage potential. At present, the application of energy storage on the user side is actively explored, corresponding energy storage is established according to the characteristics of the energy storage on the user side, the utilization efficiency of the energy source on the user side is improved, the stability of the power grid load is ensured, the energy utilization quality on the user side is improved, the peak pressure of the power grid is relieved, and the method has important engineering practical value and development prospect.
Disclosure of Invention
The invention aims to provide a layered identification method for the energy storage installation potential of an industrial enterprise at a user side view angle, aiming at the problems in the background technology, corresponding energy storage is established according to the characteristics of the energy storage at the user side, the utilization efficiency of the energy at the user side is improved, the stability of the power grid load is ensured, and the energy utilization quality at the user side is improved.
In order to achieve the above purpose, the invention adopts the following technical scheme: a hierarchical identification method for energy storage installation potential of an industrial enterprise under a user side view angle comprises the following steps:
1. Data preparation and pretreatment: collecting required data, including user profile information and load information of an industrial enterprise; the method comprises the steps of completing the cleaning work of collected data, including processing missing values, abnormal values and noise data, and carrying out standardized processing on the data;
2. modeling analysis: carrying out industrial load characteristic analysis, user load adjustable potential analysis and user load level analysis, and constructing indexes; calculating the weight of each index by a correlation analysis method in combination with index information of the user, calculating the comprehensive score of the user according to a method that a plurality of indexes are multiplied by the weights and summed, and layering the user according to the comprehensive score; the method comprises the following steps:
2.1 Industrial load profile analysis)
Clustering the historical load curves of the users according to industry and user load characteristic data by a K-Means, DBscan, FCM clustering algorithm to obtain cluster-like numbers and a clustering center, and completing user load characteristic analysis;
2.2 User load adjustable potential analysis)
Evaluating the user load adjustable potential value of the industrial enterprise;
2.3 A), user load level analysis;
2.4 Model construction for identifying and evaluating energy storage potential of industrial enterprise user side
Building an industrial enterprise user side energy storage potential identification and assessment model: according to the layering identification and evaluation requirements of the energy storage potential of the industrial enterprise user side, the weights of all the indexes are calculated by a related weight measuring method including a hierarchical analysis method, an entropy method and a factor analysis method in combination with index information of the users, then comprehensive weights of the users are calculated in a way of multiplying and summing the indexes and the weights, and finally the users are layered according to the comprehensive weights, so that labels are given;
2.5 Construction of energy storage strategy method library)
Constructing an energy storage strategy method library: and externally providing a user energy storage strategy and internally providing a service optimization strategy aiming at different user layering grades to form an energy storage strategy method library.
Further, the data required in the first step includes archive information, load information and electricity fee information of the industrial enterprise users.
Still further, the data preprocessing in the first step includes the steps of:
1.1 Data cleaning)
The method for processing the missing value comprises the following steps: deleting, filling and not processing; deleting a sample row or a characteristic column with a missing value to obtain a complete data set; filling is to fill null values with set values to complete the data set, and filling a missing value according to the distribution condition of the values of other objects in the initial data set;
Filling the missing data by using a KNN algorithm;
1.2 Data conversion)
And (3) carrying out data normalization processing: the normalization method used in clustering was zero-mean normalization (Z-score normalization); mapping the original data to a distribution with a mean value of 0 and a standard deviation of 1; specifically, assuming that the mean value of the original feature is μ and the standard deviation is σ, the normalization formula is defined as:
1.3 Data integration)
Performing integrated splicing and redundancy elimination before data modeling analysis; the method comprises the steps of integrating clustering result data with industry archive data, and establishing a data analysis mining broad table.
Further, the industry and user load characteristic data in step 2.1) includes performing industry load characteristic analysis, and summarizing industry load characteristic and rule data; the multi-source data is loaded based on electricity user profile information and daily frequency.
Further, in the step 2.1), the industrial load characteristic analysis is to summarize the load characteristic and rule of each sub-industry through the analysis of the electric loads of all the sub-industries; analyzing whether the electricity load characteristics of different users in the same industry are different or not by analyzing the electricity load characteristics of different users in the same industry; by analyzing the power load characteristics among different industries, whether the power load characteristics among different industries have similarity or not;
The user load characteristic analysis is to cluster the user history load curve through a K-Means, DBscan, FCM clustering algorithm to obtain cluster-like numbers and cluster centers, and then analyze and describe the load characteristics of each class to complete the user load characteristic analysis.
Still further, a K-Means clustering algorithm is selected; the algorithm clustering process is as follows: (1) setting a K value and determining a cluster number; (2) Calculating the distance from each record to the class center, and dividing the distance into K classes; (3) The K-class center is used as a new center, and the distance is recalculated; (4) Iterating until all the data cannot be updated to other data sets;
the method for determining the cluster K value comprises the following steps: davison burg index; the davison baudiner index calculates the Davies-Bouldin score, automatically calculates the K value, and the formula is:
wherein DBI is an index value,for the average Euclidean distance of the i-th class sample to its class center, < >>For the average Euclidean distance of the j-th class sample to the center of the class, the I wi-wi I2 is the Euclidean distance of the class centers of the i-th and j-th classes.
Further, in step 2.2), based on the load data of the user at each time point and the clustering result of the industrial load characteristic analysis part, performing quantitative analysis on the load adjustable potential by using the related index;
The index selection comprises the following steps: analyzing load adjustable potential through two indexes of the most probable adjustment quantity and the most probable adjustment quantity, wherein the most probable adjustment quantity and the most probable adjustment quantity are load adjustable quantities in a time period appointed by an operation day, the most probable adjustment quantity is the difference value between a base line load and the lowest load in an electricity utilization class to which the base line load belongs, and the most probable adjustment quantity is the difference value between the base line load and the average load of the electricity utilization class of the minimum load, wherein the base line load is calculated by the load average value of the first thirty days of the operation day;
ΔH t ＝x t -min(x c )
ΔP t ＝x t -mean(x e )
where xt is the base line load, n is set to 30, xc is the cluster load in the electricity usage class to which the base line load belongs, Δht is the most likely adjustment amount, xe is the minimum load electricity usage cluster load, and Δpt is the maximum adjustable amount.
Further, in step 2.3), the maximum load and the average load of the user are calculated based on the historical daily frequency load data, and the maximum load and the average load are load values in a time period appointed by the operation day; the maximum load is the maximum load of the user, and the average load is the average load;
H max ＝max(x ij )
H mean ＝mean(∑x ij )
wherein x is ij Is the load at a certain point of the day.
Further, the index information in the step 2.4) comprises the most possible adjustment amount, the maximum adjustable amount, the maximum load, the average load, the peak-valley fluctuation degree, the maximum load of unit capacity and the electric charge of unit capacity;
The peak-to-valley waviness:
if the peak-valley difference of the load of the user reaches the set limit and exceeds the average load of the user, the user has potential for energy storage, and the calculation mode of the minimum load is the same as that of the maximum load and is the minimum load; the specific formula is as follows:
wherein H is max Indicating maximum load, H min Represents the minimum load,Representing the average load;
the maximum load per unit capacity:
if the peak load of the user is higher than the running capacity of the user, the user has potential for energy storage, and the specific formula is as follows:
l＝H max /Q
wherein H is max Representing maximum load, Q representing user operating capacity;
the unit capacity electricity fee:
the larger the ratio of the electricity charge of the user to the running capacity is, the more the user has the potential of energy storage, and the specific formula is as follows:
c＝p t /Q
wherein p is t The electric charge is represented, and Q represents the running capacity of a user;
the weight of each index is calculated by an analytic hierarchy process, an entropy method or a factor analysis method, and finally, the user comprehensive score is calculated by multiplying and summing each index value and the respective weight, wherein the formula is as follows:
wherein w is i Representing weights, x i Indicating an index.
Still further, the method for calculating the index weight selects an entropy method, and the algorithm comprises the following steps:
index normalization
Firstly, carrying out standardization processing on measurement units of each index, namely converting absolute values of the indexes into relative values, and carrying out data standardization processing on positive and negative indexes by using different algorithms; seven indexes related to the model are all forward indexes, and the standardization method is as follows:
wherein x is ij A j index value (i=1, 2,) representing an i-th user; j=1, 2., m).
Calculating the specific gravity of the index
Calculating normalized x's' ij The specific gravity of the value in the current column of index data;
calculating the entropy value of the j index
Through the above mentioned x' ij Calculating the entropy value of the j index according to the proportion of the value in the index data of the current column;
wherein k=1/ln (n) >0;
calculating information entropy redundancy
d j ＝1-e j
Calculating the weight of each index
The invention has the technical effects that: according to the layered identification method for the energy storage installation potential of the industrial enterprise at the view angle of the user side, corresponding energy storage is established according to the characteristics of the energy storage of the user side, the utilization efficiency of the energy source of the user side is improved, the stability of the power grid load is ensured, and the energy utilization quality of the user side is improved.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a flow chart of overall modeling of big data according to the present invention;
FIG. 3 is a flow chart of the user-side energy storage potential identification and assessment model construction of the present invention;
FIG. 4 is a graph of a model index system of the present invention.
Detailed Description
The invention relies on the high-frequency characteristic of the power load data, adopts a multi-source data fusion technology and a big data analysis technology, and deeply analyzes data such as user load, electric charge and the like to finish industrial load characteristic analysis, develop user load adjustable potential analysis, load level analysis and the like, builds a user side energy storage potential identification and assessment model based on the power big data based on the analysis result, finally proposes an energy storage strategy, and excavates and efficiently implements energy for the potential of user side response.
Firstly, collecting public data, and developing layering identification and current state service carding of evaluation of energy storage potential of a user side of an industrial enterprise; secondly, analyzing the layered identification and evaluation analysis of the energy storage potential of the users of various existing industrial enterprises, and comparing and analyzing the types of support technology; thirdly, searching data sources based on service requirements, and analyzing data supporting conditions.
Data preparation and pretreatment: and collecting required data, including industrial enterprise user archive information and load information. And (3) finishing the cleaning work of the collected data, wherein the cleaning work comprises the steps of processing missing values, abnormal values and noise data and carrying out standardized processing on the data.
Modeling analysis: and carrying out industrial load characteristic analysis, user load adjustable potential analysis and user load level analysis, and constructing indexes including 'most possible adjustment quantity', 'maximum adjustable quantity', 'maximum load'. And calculating the weight of each index by a correlation analysis method in combination with the index information of the user, calculating the comprehensive score of the user according to the method that a plurality of indexes are multiplied by the weights and summed, and layering the user according to the comprehensive score.
The method mainly comprises the following steps:
2.1 Industrial load profile analysis)
According to industry and user load characteristics: firstly, carrying out industrial industry load characteristic analysis and summarizing industrial load characteristic and rule data; and secondly, clustering historical load curves of the users by a clustering algorithm such as K-Means, DBscan, FCM based on multi-source data such as electricity user profile information and daily frequency load to obtain cluster-like numbers and a clustering center, and completing user load characteristic analysis.
2.2 User load adjustable potential analysis)
Evaluating the user load adjustable potential value of the industrial enterprise: the quantification of the user demand response capability is realized through analysis of the user load adjustable potential, the user load adjustable potential can reflect the capability and the potential of the user for adjusting the power load under the set condition, and the partial analysis specifically measures the magnitude of the user load adjustable potential through two indexes, namely 'the most possible adjustment amount' and 'the maximum adjustable amount'.
2.3 User load level analysis)
Assessing the industrial enterprise user load level: the user load level refers to the electricity demand condition of the user in a period of time, the load level of the user is represented by indexes such as maximum load, average load and the like, the maximum load represents the load peak value of the user in a period of time, namely the highest point of the electricity demand, the average load refers to the average value of the daily load of the user in a period of time, and the overall electricity demand level of the user can be reflected.
2.4 Model construction for identifying and evaluating energy storage potential of industrial enterprise user side
Building an industrial enterprise user side energy storage potential identification and assessment model: according to the layering identification and evaluation requirements of the energy storage potential of the industrial enterprise user side, the index information such as 'most possible adjustment quantity', 'maximum adjustable quantity', 'maximum load', 'average load' of the user is combined, the weight of each index is calculated through the related weight weighing method such as a hierarchical analysis method, an entropy value method and a factor analysis method, the comprehensive weight of the user is calculated according to the mode that the index is multiplied by the weight and summed, and finally the user is layered according to the comprehensive weight, and the labels such as 'high attention, moderate attention, common attention' and the like are given.
2.5 Construction of energy storage strategy method library)
Constructing an energy storage strategy method library: and providing a user energy storage strategy outwards and a service optimization strategy inwards according to different user layering grades to form a targeted and differentiated energy storage strategy method library.
The method comprises the following steps: based on big data analysis, a corresponding energy storage strategy and a service optimization strategy are provided for a user.
The data required by the invention comprises archive information, load information and electricity charge information of industrial enterprise users, and the data is derived from a marketing business application system and an electricity consumption information acquisition system.
1. Data preprocessing
The quality of data is one of key factors influencing the effect of a big data analysis model, and before data modeling analysis, the data quality is required to be explored, diagnosed and processed. In the data preprocessing process, scattered big data resources are reconstructed according to the preset analysis service through the steps of data cleaning, data conversion, data integration and the like, so that reliable basic data support is provided for subsequent analysis and mining.
1.1 Data cleaning)
The data cleaning is an important ring in the data mining engineering, and the algorithm model applied by the invention has high requirements on the data integrity, and the missing values are required to be processed in the data cleaning process. Specific:
The method for processing the missing value mainly comprises the following three steps: delete, fill, do not process. Deleting, that is, deleting the sample (row) or feature (column) with the missing value, thereby obtaining a complete data set; filling, namely filling null values by using set values so as to integrate a data set, wherein filling a missing value according to the distribution condition of values of other objects in an initial data set generally based on a statistical principle, and filling special values, average filling, hot card filling, KNN filling, regression model filling, multiple interpolation, lagrange interpolation and the like are commonly used in data mining; without processing, the filling processing only fills in unknown values with subjective estimated values, and does not necessarily completely conform to objective facts, the original information system is changed more or less while incomplete information is filled in, and new noise is often introduced into data when filling in empty values incorrectly, so that data mining produces erroneous results.
When user load clustering is carried out, a K-Means clustering model needs to be built, but the used load data has missing values, the similarity between data points is calculated by a K-Means clustering algorithm based on distance measurement, and the missing data influences the calculation of the distance, so that the missing values need to be processed. The invention fills up the missing data by using KNN algorithm filling, and the reason is that: the algorithm can keep similarity among samples, is a machine learning algorithm without parameters, considers the relation of a plurality of characteristics and can be flexibly applied to different data sets.
1.2 Data conversion)
The different characteristic values are different in size, the situation can affect the data analysis result, reduce the stability of a model, algorithm performance and the like, if the load value interval at some time is 3-50kVA, the load value interval at some time is 5000-8000kVA, the difference between the values is too large, and in order to eliminate the influence of the too large difference between the values, data normalization processing is needed. The normalization method used in the clustering of the invention is zero-mean normalization (Z-score normalization). It will map the raw data to a distribution with a mean of 0 and standard deviation of 1. Specifically, assuming that the mean value of the original feature is μ and the standard deviation is σ, the normalization formula is defined as:
1.3 Data integration)
The data integration is a process of data integration and splicing, and the data integration with different structures and different attributes is integrated by fusing all data sources, namely the data integration. As naming rules are different when different data sources define attributes, the stored data formats, value-taking modes and units are different. Thus even though the two values represent the same business meaning, it does not represent that the values present in the database are the same. Therefore, before data modeling analysis, integrated splicing is needed, redundancy is removed, and data quality is ensured.
In the invention, for example, clustering result data and industry archive data are integrated together, a data analysis mining broad table is established, and subsequent mining analysis is supported.
2. Big data modeling analysis
And carrying out industrial load characteristic analysis, user load adjustable potential analysis and user load level analysis based on large data such as files and loads, setting up a large data related algorithm range such as machine learning, deep learning and the like, carrying out quality comparison analysis on the effects, efficiency, stability and the like among algorithms through Python training, completing training and construction of an industrial enterprise user side energy storage potential identification and evaluation model based on large electric power data, so as to mine user load adjustable potential, capture user load level, realize identification and evaluation of user energy storage potential, and finally form a targeted and differentiated energy storage strategy method library.
2.1 Industrial load profile analysis)
Firstly, carrying out industry load characteristic analysis, and expanding the analysis from three aspects: summarizing the load characteristics and rules of each industry sub-industry by analyzing the power consumption load of all the industry sub-industries; analyzing whether the electricity load characteristics of different users in the same industry are different or not by analyzing the electricity load characteristics of different users in the same industry; by analyzing the electricity load characteristics among different industries, whether the electricity load characteristics among different industries have similarity or not is judged.
Secondly, user load characteristic analysis is carried out, a user history load curve is clustered through a clustering algorithm such as K-Means, DBscan, FCM (the baseline load curve is obtained by calculating a load average value of the first N days of operation days, N can be dynamically adjusted, N is set reasonably, the condition that the user load level cannot be truly reflected due to too short time is possibly caused, the condition that the user recent load cannot be reflected due to too long time is likely to be caused), cluster numbers and cluster centers are obtained, then the load characteristics of each class are analyzed and described, and user load characteristic analysis is completed.
Algorithm selection
The clustering algorithm is a K-Means clustering algorithm, and various clustering algorithms such as hierarchical clustering, DBscan clustering, FCM clustering, K-Means clustering and the like are adopted; the method is a common and well-known clustering algorithm, is widely applied to the fields of data mining and category identification, has the advantages of simple and efficient idea, simple principle, convenience in implementation, high convergence speed, better clustering effect and stronger model interpretability, and can divide data into a plurality of clusters with similar characteristics. The algorithm clustering process is as follows: (1) setting a K value and determining a cluster number; (2) Calculating the distance (Euclidean distance) from each record to the class center, and dividing the distance into K classes; (3) The K-class center is used as a new center, and the distance is recalculated; (4) The iteration is continued until all the data cannot be updated to other data sets.
Method for determining K value of cluster number
The clustering K value determining method comprises the following steps: davison burg index; the davison burg index was framed as the cluster K value method by comparison experiments on three methods, elbowMethod, interval statistic (gapstatic), davison burg Ding Zhishu (Davies-Bouldin). The reason is that the judgment of the elbow rule is not automatic enough and intelligent calculation cannot be realized, which leads to the fact that the subsequent model construction cannot be continued; the interval statistics can be automated, the goal of the method is to find the maximum Gap value (Gap value is the difference between the loss of random samples and the loss of actual samples), but the method is not applicable to partial data sets, and has the disadvantage that the target Gap value pursued by the method can be monotonically increased or monotonically decreased, that is, the continuous change of the K value can cause the continuous change of the Gap value; the davison baudiner index calculates a Davies-Bouldin score, can automatically calculate a K value (the number of clusters) and does not have the problem of interval statistics. I.e. the larger the inter-class distance, the smaller the intra-class distance (the more concentrated the data), the better the formula:
Wherein DBI is an index value,for the average Euclidean distance of the i-th class sample to its class center, < >>For the average Euclidean distance of the j-th class sample to the center of the class, the I wi-wi I2 is the Euclidean distance of the class centers of the i-th and j-th classes.
Analysis process and results
Step1: and acquiring daily load data and acquiring latest archive data.
Step2: data preprocessing, including data type conversion, deletion of relevant fields, field renaming, field mapping, missing value filling, normalization, etc.
Step3: and carrying out industry load characteristic analysis.
Step4: the working steps mainly comprise the steps of selecting a method for determining the number of clusters, constructing a K-Means cluster model, analyzing the cluster result and the like.
2.2 User load adjustable potential analysis)
Analysis idea
Based on the load data of the user at each time point and the clustering result of the industrial load characteristic analysis part, the load adjustable potential is quantitatively analyzed by using the related indexes, and the load adjustable degree of the system in the current running state is determined.
In index selection, a mature user load adjustable potential measurement method is used as a reference, load adjustable potential is analyzed through two indexes of 'the most probable adjustment amount' and 'the maximum adjustable amount', and the larger the 'the most probable adjustment amount' and the 'the maximum adjustable amount', the larger the potential of the user for installing energy storage is indicated.
Index rules
The most probable adjustment amount and the most probable adjustment amount are load adjustment amounts in a time period appointed by the operation day, the most probable adjustment amount is the difference value of the base line load and the lowest load in the electricity utilization category to which the base line load belongs, the most probable adjustment amount is the difference value of the base line load and the average load of the electricity utilization category of the minimum load, and the base line load is calculated by the load average value of the first thirty days of the operation day.
△H t ＝x t -min(x c )
ΔP t ＝x t -mean(x e )
Where xt is the base line load, n is set to 30 (adjustable), xc is the cluster load in the electricity category to which the base line load belongs, Δht is the most likely adjustment amount, xe is the minimum load electricity cluster load, and Δpt is the maximum adjustable amount.
Analysis process and results
Step1: locking the electricity utilization type of the base line load;
step2: taking out the baseline load data of the appointed period, and calculating a baseline load average value;
step3: taking out load data of a specified period of electricity consumption class to which the baseline load belongs, calculating average load of each day, and then sequencing the average load of each day according to ascending order, and locking the lowest load in the class to which the baseline load belongs;
step4: calculating a 'most probable adjustment' according to a formula;
step5: locking the minimum load power utilization type of the user, namely the power utilization type with the minimum corresponding average load value;
Step6: and (3) taking out load data of a specified period belonging to the minimum load electricity utilization category, calculating average load of each day, and then averaging the average load of each day to finally obtain the minimum load electricity utilization category average load.
Step7: the "maximum adjustable amount" is calculated according to the formula.
2.3 User load level analysis)
User load level analysis is a method for evaluating and analyzing user loads in an electric power system, and aims to know load characteristics and requirements of users in depth, so that important support is provided for energy storage potential identification and evaluation of the users.
In order to fully show the historical load condition of the user, based on the deep analysis of the big data of the user load, the optimal index is extracted, and the comprehensive condition of the historical load of the user can be shown in an all-round and multi-angle mode to the greatest extent. In the present invention, two key indicators, namely "maximum load", "average load", are extracted to describe the load level of the user. Through analysis of two indexes of maximum load and average load, the load condition of a user can be better known, and an important reference basis is provided for constructing an industrial enterprise user side energy storage potential identification and evaluation model.
Index rules
And calculating the maximum load and the average load of the user based on the daily frequency historical load big data, wherein the maximum load and the average load are load values in a time period appointed by the operation day. The maximum load is the maximum load of the user, and the average load is the average load.
H max ＝max(x ij )
H mean ＝mean(Σx ij )
Wherein x is ij Is the load at a certain point of the day.
Analysis process and results
Step1: load data of a specified period is taken out;
step2: locking the maximum load to obtain a maximum load;
step3: calculating the average load corresponding to the user every day;
step4: based on the average load, the average load is further averaged, and finally the average load is obtained.
2.4 User-side energy storage potential identification and assessment model construction)
According to the layering identification and evaluation requirements of the energy storage potential of the industrial enterprise user side, combining index information of the most probable adjustment quantity, the maximum adjustable quantity and the maximum load of the user, constructing an all-dimensional and multi-dimensional index system, calculating the weight of each index through a related weight measuring method of a hierarchical analysis method, an entropy value method and a factor analysis method, namely, each index has a corresponding weight, calculating a user comprehensive score according to a method that a plurality of indexes are multiplied by the weights and summed, layering the user according to the 'tertile digit' in statistics, and giving a label of 'high attention, medium attention and common attention'.
Index system
In order to forge a better model, three new indexes are added on the basis of basic indexes, wherein the three indexes are respectively "peak-valley fluctuation degree", "unit capacity maximum load", "unit capacity electric charge", and further an index system with five view angles is formed, and the five view angles (index system) are respectively "load management view angles" (including "most possible adjustment amount", "maximum adjustment amount" index), a "load characteristic view angle" (including "maximum load", "average load" index), a "safety view angle" (including "unit capacity maximum load" index), a "production schedule view angle" (including "peak-valley fluctuation degree" index) and "economic view angle" (including "unit capacity electric charge" index).
The "peak-valley fluctuation", "unit capacity maximum load", "unit capacity electric charge" indexes are as follows:
degree of fluctuation of peak and valley
If the peak-valley difference of the load of the user reaches the set limit and exceeds the average load of the user, the user may have potential for energy storage, and the minimum load is calculated in the same manner as the maximum load (which is the minimum load). The specific formula is as follows:
wherein H is max Indicating maximum load, H min Represents the minimum load, Representing the average load.
Maximum load per unit capacity
If the peak load of the user is higher than the running capacity of the user, the user may have potential for energy storage, and the specific formula is:
l＝H max /Q
wherein H is max Representing maximum load, Q represents user operating capacity.
Electric charge per unit capacity
The larger the ratio of the electricity charge to the running capacity of the user is, the more the user may have potential for energy storage, and the specific formula is:
c＝p t /Q
wherein p is t The electric charge is represented, and Q represents the user operation capacity.
The analysis process comprises the following steps: the maximum load, the minimum load and the average load are described above to explain the analysis process, and the repeated description is omitted, so that the electricity charge and the running capacity can be directly obtained.
Model setting
The method comprises the steps of integrating 'most possible adjustment quantity', 'maximum adjustable quantity', 'maximum load', 'average load', 'peak-to-valley fluctuation', 'maximum load per unit capacity', 'electricity charge per unit capacity', calculating the weight of each index by a hierarchical analysis method, an entropy method or a factor analysis method, and finally calculating a user comprehensive score by multiplying and summing each index value with the respective weight, wherein the formula is as follows:
wherein w is i Representing weights, x i Indicating an index.
Method for calculating index weight
In the method for calculating the index weight, an entropy method is selected, because the entropy method can calculate the weight according to the difference and the information quantity between the indexes, instead of simply assuming that the weights of the indexes are equal, the entropy method can more accurately reflect the contribution degree of the indexes to the decision result and better reflect the reality; the calculation process of the entropy method does not depend on subjective judgment of a decision maker, is based on the statistical characteristics of index data, and calculates practically, so that the entropy method has higher objectivity and consistency; the entropy method is suitable for various index types and data types, and can perform effective weight calculation no matter qualitative indexes or quantitative indexes; the method has no strict requirements on the scale of index data and the form of data distribution, and has higher flexibility and adaptability.
In information theory, entropy is a measure of uncertainty. The larger the information amount is, the smaller the uncertainty is, and the smaller the entropy is; the smaller the amount of information, the greater the uncertainty and the greater the entropy. According to the characteristics of entropy, the randomness and disorder degree of an event can be judged by calculating an entropy value, or the degree of dispersion of a certain index can be judged by using the entropy value, and the larger the degree of dispersion of the index is, the larger the influence (weight) of the index on comprehensive evaluation is, and the smaller the entropy value is. The method comprises the following algorithm steps of calculating index weights such as 'most possible adjustment quantity', 'maximum adjustable quantity', 'maximum load' and the like through an entropy method:
index normalization
Because the measurement units of the indexes are not uniform, before the comprehensive indexes are calculated by the measurement units, the measurement units are subjected to standardization treatment, namely the absolute values of the indexes are converted into relative values, so that the homogenization problem of various different quality index values is solved. Further, since the positive index and the negative index have different meanings (the higher the positive index value is, the better the negative index value is, the lower the negative index value is), the data normalization process is performed for the positive and negative indexes using different algorithms. Seven indexes related to the model are all forward indexes, and the standardization method is as follows:
Wherein x is ij A j index value (i=1, 2,) representing an i-th user; j=1, 2., m).
Calculating the specific gravity of the index
Calculating normalized x's' ij The specific gravity of the value in the current column index data.
Calculating the entropy value of the j index
Through the above mentioned x' ij And calculating the entropy value of the jth index according to the proportion of the value in the index data of the current column.
/>
Where k=1/ln (n) >0.
Calculating information entropy redundancy
d j ＝1-e j
Calculating the weight of each index
Analysis process and results
Step1: normalizing the original data, and scaling each index value of the 'most probable adjustment quantity', 'maximum adjustable quantity', 'maximum load', 'average load', 'peak-to-valley fluctuation degree', 'maximum load per unit capacity', 'electric charge per unit capacity' to be within the interval range of [0,1] by using a 'maximum-minimum normalization' method;
step2: determining the weight of each index according to an entropy method (the sum of the weights of the indexes is 1);
step3: multiplying the weight of each index by the normalized index value and then summing to obtain the comprehensive score of each user, wherein the range of the comprehensive score is between 0 and 1;
step4: layering the users according to the comprehensive scores, and giving a label of 'high attention, moderate attention and general attention', wherein the layering rule takes a theory of 'three digits' in statistics as a support, and the specific contents are as follows: the numbers in a sequence are sequentially ordered from small to large, then the sequence is divided into three equal parts, and finally the three parts are divided into a low value part, a median value part and a high value part.
In the implementation of the present invention, the rule for layering the user by using the "tertile" is: the users with the comprehensive score being more than or equal to 0.66 are high attention objects, the users with the comprehensive score being more than or equal to 0.33 and less than 0.66 are medium attention objects, the users with the comprehensive score being less than 0.33 are common attention objects, the value range of the comprehensive score of each user is between 0 and 1, and the layering threshold of the users can be dynamically adjusted according to the actual effect of the model.
The invention provides a layered identification and evaluation method for energy storage potential under the view angle of an industrial enterprise user side, which firstly provides multiple indexes such as 'most possible adjustment quantity', 'maximum adjustable quantity', 'maximum load', and the like to participate in the identification and evaluation of the energy storage potential. The indexes such as the most probable adjustment quantity, the maximum adjustable quantity, the maximum load, the average load, the peak-valley fluctuation degree, the maximum load of unit capacity, the electric charge of unit capacity and the like jointly form an index system for the layered identification and evaluation of the energy storage potential of the energy storage user side in view angles, in the system, the energy storage potential of the user is analyzed from multiple view angles, such as the index of the most probable adjustment quantity, the index of the maximum adjustable quantity, the index of the maximum load, the index of the average load, the index of the maximum load of unit capacity, the index of the peak-valley fluctuation degree, the index of the maximum load of unit capacity, the index of the electric charge of unit capacity, the effect and the effect of a model are greatly improved when the energy storage problem is seen from an all-around and multi-view angle, and the layered identification and evaluation of the energy storage potential of the user side are more scientific, rigorous, comprehensive and efficient. Secondly, davison fort Ding Zhishu (Davies-Bouldin) was used for user load profiling. In the K-Means algorithm K value determination method, the davidienberg index is framed as a clustering K value method by comparison experiments of three methods, namely an elbow rule (ElbowMethod), an interval statistic (gapstatic) and davidienberg Ding Zhishu (Davies-Bouldin). The reason is that the judgment of the elbow rule is not automatic enough and intelligent calculation cannot be realized, which leads to the fact that the subsequent model construction cannot be continued or the subsequent manual intervention cannot be continued; the interval statistics can be automated, the goal of the method is to find the maximum Gap value (Gap value is the difference between the loss of random samples and the loss of actual samples), but the method is not applicable to partial data sets because the target Gap value pursued by the method is monotonically increasing or monotonically decreasing, that is, the continuous change of the K value can cause the continuous change of the Gap value, and the method also has the defect; the davison baudiner index calculates the Davies-Bouldin score, the K value (the number of clusters) can be automatically calculated, the problem of interval statistics can not occur, and more importantly, the method comprehensively considers the similarity of samples in the class and the difference of samples between the classes.
In summary, the invention aims at the problems of high electricity cost, serious contradiction between power supply and demand, insufficient power supply of a power grid and the like, and relies on a big data modeling analysis technology to carry out deep analysis from the contents of user load adjustable potential analysis, user load level analysis and the like, so as to construct a set of big data analysis model for layered identification and evaluation of energy storage potential under the view angle of the user side of an industrial enterprise, and the invention builds a user energy storage strategy and a service optimization strategy by assistance, thereby on one hand, further meeting the requirements of users on electricity utilization just, improving the electric energy quality and saving the expenditure of electric charge, and on the other hand, smoothing the power grid load, relieving the power grid pressure and improving the service level.
Claims (10)
1. The hierarchical identification method for the energy storage installation potential of the industrial enterprise under the view angle of the user side is characterized by comprising the following steps:
1. data preparation and pretreatment: collecting required data, including user profile information and load information of an industrial enterprise; the method comprises the steps of completing the cleaning work of collected data, including processing missing values, abnormal values and noise data, and carrying out standardized processing on the data;
2. modeling analysis: carrying out industrial load characteristic analysis, user load adjustable potential analysis and user load level analysis, and constructing indexes; calculating the weight of each index by a correlation analysis method in combination with index information of the user, calculating the comprehensive score of the user according to a method that a plurality of indexes are multiplied by the weights and summed, and layering the user according to the comprehensive score; the method comprises the following steps:
2.1 Industrial load profile analysis)
Clustering the historical load curves of the users according to industry and user load characteristic data by a K-Means, DBscan, FCM clustering algorithm to obtain cluster-like numbers and a clustering center, and completing user load characteristic analysis;
2.2 User load adjustable potential analysis)
Evaluating the user load adjustable potential value of the industrial enterprise;
2.3 A), user load level analysis;
2.4 Model construction for identifying and evaluating energy storage potential of industrial enterprise user side
Building an industrial enterprise user side energy storage potential identification and assessment model: according to the layering identification and evaluation requirements of the energy storage potential of the industrial enterprise user side, the weights of all the indexes are calculated by a related weight measuring method including a hierarchical analysis method, an entropy method and a factor analysis method in combination with index information of the users, then comprehensive weights of the users are calculated in a way of multiplying and summing the indexes and the weights, and finally the users are layered according to the comprehensive weights, so that labels are given;
2.5 Construction of energy storage strategy method library)
Constructing an energy storage strategy method library: and externally providing a user energy storage strategy and internally providing a service optimization strategy aiming at different user layering grades to form an energy storage strategy method library.
2. The hierarchical identification method for energy storage installation potential of industrial enterprise under the view angle of user side according to claim 1, wherein the data required in the step one comprises archive information, load information and electricity charge information of the industrial enterprise user.
3. The hierarchical identification method for energy storage installation potential of industrial enterprises under the view angle of the user side according to claim 1 or 2, wherein the data preprocessing in the step one comprises the following steps:
1.1 Data cleaning)
The method for processing the missing value comprises the following steps: deleting, filling and not processing; deleting a sample row or a characteristic column with a missing value to obtain a complete data set; filling is to fill null values with set values to complete the data set, and filling a missing value according to the distribution condition of the values of other objects in the initial data set;
filling the missing data by using a KNN algorithm;
1.2 Data conversion)
And (3) carrying out data normalization processing: the normalization method used in clustering was zero-mean normalization (Z-score normalization); mapping the original data to a distribution with a mean value of 0 and a standard deviation of 1; specifically, assuming that the mean value of the original feature is μ and the standard deviation is σ, the normalization formula is defined as:
1.3 Data integration)
Performing integrated splicing and redundancy elimination before data modeling analysis; the method comprises the steps of integrating clustering result data with industry archive data, and establishing a data analysis mining broad table.
4. The hierarchical identification method for energy storage installation potential of industrial enterprises under the view angle of the user side according to claim 1, wherein the industry and user load characteristic data in the step 2.1) comprises carrying out industrial industry load characteristic analysis and summarizing load characteristic and rule data of the industry; the multi-source data is loaded based on electricity user profile information and daily frequency.
5. The hierarchical identification method for energy storage installation potential of industrial enterprises under the view angle of the user side according to claim 1, wherein the industrial load characteristic analysis in the step 2.1) is to summarize the load characteristic and rule of each sub-industry by analyzing the electric loads of all the sub-industries of the industry; analyzing whether the electricity load characteristics of different users in the same industry are different or not by analyzing the electricity load characteristics of different users in the same industry; by analyzing the power load characteristics among different industries, whether the power load characteristics among different industries have similarity or not;
The user load characteristic analysis is to cluster the user history load curve through a K-Means, DBscan, FCM clustering algorithm to obtain cluster-like numbers and cluster centers, and then analyze and describe the load characteristics of each class to complete the user load characteristic analysis.
6. The hierarchical identification method for energy storage installation potential of an industrial enterprise under the view angle of a user side according to claim 5, wherein a K-Means clustering algorithm is selected; the algorithm clustering process is as follows: (1) setting a K value and determining a cluster number; (2) Calculating the distance from each record to the class center, and dividing the distance into K classes; (3) The K-class center is used as a new center, and the distance is recalculated; (4) Iterating until all the data cannot be updated to other data sets;
the method for determining the cluster K value comprises the following steps: davison burg index; the davison baudiner index calculates the Davies-Bouldin score, automatically calculates the K value, and the formula is:
wherein DBI is an index value,for the average Euclidean distance of the i-th class sample to its class center, < >>For the average Euclidean distance of the j-th class sample to the center of the class, the I wi-wi I2 is the Euclidean distance of the class centers of the i-th and j-th classes.
7. The hierarchical recognition method for energy storage installation potential of industrial enterprises under the view angle of the user side according to claim 1, wherein in the step 2.2), based on the load data of the users at each time point and the clustering result of the industrial load characteristic analysis part, the load adjustable potential is quantitatively analyzed by using the related indexes;
The index selection comprises the following steps: analyzing load adjustable potential through two indexes of the most probable adjustment quantity and the most probable adjustment quantity, wherein the most probable adjustment quantity and the most probable adjustment quantity are load adjustable quantities in a time period appointed by an operation day, the most probable adjustment quantity is the difference value between a base line load and the lowest load in an electricity utilization class to which the base line load belongs, and the most probable adjustment quantity is the difference value between the base line load and the average load of the electricity utilization class of the minimum load, wherein the base line load is calculated by the load average value of the first thirty days of the operation day;
ΔH t ＝x t -min(x c )
ΔP t ＝x t -mean(x e )
where xt is the base line load, n is set to 30, xc is the cluster load in the electricity usage class to which the base line load belongs, Δht is the most likely adjustment amount, xe is the minimum load electricity usage cluster load, and Δpt is the maximum adjustable amount.
8. The hierarchical recognition method for energy storage installation potential of an industrial enterprise under the view angle of a user side according to claim 1, wherein in the step 2.3), the maximum load and the average load of the industrial enterprise are calculated based on the historical load big data of the daily frequency of the user, and the maximum load and the average load are load values in a time period appointed by an operation day; the maximum load is the maximum load of the user, and the average load is the average load;
H max ＝max(x ij )
H mean ＝mean(∑x ij )
Wherein x is ij Is the load at a certain point of the day.
9. The hierarchical recognition method for energy storage installation potential of an industrial enterprise under a user side view angle according to claim 1, wherein the index information in the step 2.4) comprises the most possible adjustment amount, the maximum adjustable amount, the maximum load, the average load, the peak-to-valley fluctuation degree, the maximum load per unit capacity and the electric charge per unit capacity;
the peak-to-valley waviness:
if the peak-valley difference of the load of the user reaches the set limit and exceeds the average load of the user, the user has potential for energy storage, and the calculation mode of the minimum load is the same as that of the maximum load and is the minimum load; the specific formula is as follows:
wherein H is max Indicating maximum load, H min Represents the minimum load,Representing the average load;
the maximum load per unit capacity:
if the peak load of the user is higher than the running capacity of the user, the user has potential for energy storage, and the specific formula is as follows:
l＝H max /Q
wherein H is max Representing maximum load, Q representing user operating capacity;
the unit capacity electricity fee:
the larger the ratio of the electricity charge of the user to the running capacity is, the more the user has the potential of energy storage, and the specific formula is as follows:
c＝p t /Q
wherein p is t The electric charge is represented, and Q represents the running capacity of a user;
The weight of each index is calculated by an analytic hierarchy process, an entropy method or a factor analysis method, and finally, the user comprehensive score is calculated by multiplying and summing each index value and the respective weight, wherein the formula is as follows:
wherein w is i Representing weights, x i Indicating an index.
10. The hierarchical recognition method for energy storage installation potential of industrial enterprises under the view angle of the user side according to claim 9, wherein the method for calculating the index weight selects an entropy method, and the algorithm steps are as follows:
index normalization
Firstly, carrying out standardization processing on measurement units of each index, namely converting absolute values of the indexes into relative values, and carrying out data standardization processing on positive and negative indexes by using different algorithms; seven indexes related to the model are all forward indexes, and the standardization method is as follows:
wherein x is ij The j index value (i=1, 2,) representing the i-th user, n; j=1, 2 …, m).
Calculating the specific gravity of the index
Calculating normalized x's' ij The specific gravity of the value in the current column of index data;
calculating the entropy value of the j index
Through the above mentioned x' ii Calculating the entropy value of the j index according to the proportion of the value in the index data of the current column;
wherein k=1/ln (n) > 0;
Calculating information entropy redundancy
d j ＝1-e j
Calculating the weight of each index
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311506635.4A CN117522207A (en) | 2023-11-13 | 2023-11-13 | Layered identification method for energy storage installation potential of industrial enterprise under user side view angle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311506635.4A CN117522207A (en) | 2023-11-13 | 2023-11-13 | Layered identification method for energy storage installation potential of industrial enterprise under user side view angle |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117522207A true CN117522207A (en) | 2024-02-06 |
Family
ID=89762053
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311506635.4A Pending CN117522207A (en) | 2023-11-13 | 2023-11-13 | Layered identification method for energy storage installation potential of industrial enterprise under user side view angle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117522207A (en) |
-
2023
- 2023-11-13 CN CN202311506635.4A patent/CN117522207A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Rajabi et al. | A comparative study of clustering techniques for electrical load pattern segmentation | |
| US11043808B2 (en) | Method for identifying pattern of load cycle | |
| US20210056647A1 (en) | Method for multi-dimensional identification of flexible load demand response effect | |
| CN112561156A (en) | Short-term power load prediction method based on user load mode classification | |
| CN111324642A (en) | Model algorithm type selection and evaluation method for power grid big data analysis | |
| CN111624931B (en) | Industrial park electricity utilization internet intelligent operation and maintenance management and control system and method | |
| CN111724278A (en) | Fine classification method and system for power multi-load users | |
| CN110503256A (en) | Short-term load forecasting method and system based on big data technology | |
| CN111680764B (en) | Industry reworking and production-resuming degree monitoring method | |
| CN111291822B (en) | Equipment running state judging method based on fuzzy clustering optimal k value selection algorithm | |
| CN112819299A (en) | Differential K-means load clustering method based on center optimization | |
| CN105868887A (en) | Building comprehensive energy efficiency analysis method based on subentry measure | |
| Damayanti et al. | Electrical load profile analysis using clustering techniques | |
| CN115907822A (en) | Load characteristic index relevance mining method considering region and economic influence | |
| CN112288157A (en) | Wind power plant power prediction method based on fuzzy clustering and deep reinforcement learning | |
| CN115660170A (en) | Multidimensional index weight collaborative optimization data asset management effect differentiation evaluation method and system | |
| Fontanini et al. | A data-driven BIRCH clustering method for extracting typical load profiles for big data | |
| CN113094448B (en) | Analysis method and analysis device for residence empty state and electronic equipment | |
| CN114266457A (en) | Method for detecting different loss inducement of distribution line | |
| CN115905319B (en) | Automatic identification method and system for abnormal electricity fees of massive users | |
| CN117522207A (en) | Layered identification method for energy storage installation potential of industrial enterprise under user side view angle | |
| CN114676931B (en) | Electric quantity prediction system based on data center technology | |
| CN114722098A (en) | Typical load curve identification method based on normal cloud model and density clustering algorithm | |
| Shen et al. | A Novel AI-based Method for EV Charging Load Profile Clustering | |
| CN113298148A (en) | Ecological environment evaluation-oriented unbalanced data resampling method |
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 |