CN112396294A - Power grid planning intelligent auxiliary analysis method based on big data - Google Patents
Power grid planning intelligent auxiliary analysis method based on big data Download PDFInfo
- Publication number
- CN112396294A CN112396294A CN202011181993.9A CN202011181993A CN112396294A CN 112396294 A CN112396294 A CN 112396294A CN 202011181993 A CN202011181993 A CN 202011181993A CN 112396294 A CN112396294 A CN 112396294A
- Authority
- CN
- China
- Prior art keywords
- power grid
- planning
- data
- script
- math
- 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
- 238000004458 analytical method Methods 0.000 title claims abstract description 62
- 238000004364 calculation method Methods 0.000 claims abstract description 49
- 238000003860 storage Methods 0.000 claims abstract description 32
- 238000007405 data analysis Methods 0.000 claims abstract description 20
- 238000009826 distribution Methods 0.000 claims abstract description 17
- 238000003062 neural network model Methods 0.000 claims abstract description 13
- 238000005065 mining Methods 0.000 claims abstract description 7
- 238000000034 method Methods 0.000 claims description 21
- 230000008569 process Effects 0.000 claims description 18
- 238000004422 calculation algorithm Methods 0.000 claims description 17
- 238000013528 artificial neural network Methods 0.000 claims description 14
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 238000010801 machine learning Methods 0.000 claims description 5
- 210000002569 neuron Anatomy 0.000 description 12
- 238000012549 training Methods 0.000 description 12
- 230000000694 effects Effects 0.000 description 10
- 230000008859 change Effects 0.000 description 9
- 238000010276 construction Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000004913 activation Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000011478 gradient descent method Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012098 association analyses Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
-
- 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
-
- 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
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Educational Administration (AREA)
- General Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Databases & Information Systems (AREA)
- Quality & Reliability (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Operations Research (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Game Theory and Decision Science (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of power grid informatization and dispatching automation. The utility model discloses a big data-based intelligent auxiliary analysis method for power grid planning, which comprises the following steps: step 1, selecting a multi-dimensional planning index, analyzing power grid planning characteristics, accessing real-time data of a power grid operation mode and acquiring a data analysis model; step 2, establishing a neural network model, extracting planning characteristic elements, and establishing a mixed storage mode of a relational database and a non-relational database; step 3, batch calculation, flow calculation, memory calculation, query calculation and distributed calculation are adopted to meet the calculation requirements of power grid planning services of different timeliness and different scenes; step 4, based on the power grid planning service model data, through big data analysis modeling and model analysis, the support of big data distributed computation is increased, and the analysis and mining requirements of real-time and off-line application of the power grid planning service are met; and 5, checking the statistical chart and displaying the index distribution.
Description
Technical Field
The invention relates to the technical field of power grid informatization and dispatching automation, in particular to a power grid planning intelligent auxiliary analysis method based on big data.
Background
With the comprehensive construction of a smart power grid and the rapid development and application of new generation IT technologies such as big data, the big data of electric power is increased rapidly and forms a certain scale. The key of the application of the large electric power data is not 'big' and 'data', and the core value of the large electric power data is that the data is regarded as the core assets of the electric power enterprise like people and properties, so that the assets create value. The new acquisition, storage and processing technology is required to be adopted to realize cross-business, multi-type, real-time, rapid and flexible customized data association analysis, and meet the requirements of management promotion and business innovation of power grid companies in the aspects of power grid production, operation management, high-quality service and the like.
The large planning is used as an important component of a three-in-five system of a national grid company, and the power grid planning is a primary link of development and construction of the power industry. With the rapid enlargement of the modern power grid scale, the increasingly complex power grid structure and the large uncertainty factor of the planning work, the related departments are multiple and the fields are wide, so that the traditional power grid planning method is a new form and cannot meet the planning requirement of the modern power grid under the new requirement. Therefore, under the guidance of big data wave, the research of an informatization and intelligentized power grid auxiliary decision planning system needs to be established by fully utilizing key technologies and development concepts of the big data of the electric power, so that reference and exploration directions are provided for the construction of a power grid planning system.
Disclosure of Invention
The invention aims to provide an intelligent auxiliary analysis method for power grid planning based on big data, which can give consideration to economy and reliability during planning and flexibly evaluate the economy and reliability of a planning scheme.
The technical scheme provided by the invention is as follows:
an intelligent auxiliary analysis method for power grid planning based on big data comprises the following steps:
step 2, establishing a neural network model, extracting planning characteristic elements, and constructing a mixed storage mode of a relational database and a non-relational database;
step 3, batch calculation, flow calculation, memory calculation, query calculation and distributed calculation are adopted to meet the calculation requirements of power grid planning services with different timeliness and different scenes;
step 4, based on the power grid planning service model data, through big data analysis modeling and model analysis, the support of big data distributed computation is increased, and the analysis and mining requirements of real-time and off-line application of the power grid planning service are met;
and 5, checking the statistical chart and displaying the index distribution.
Further, the accessing of the power grid operation data and the obtaining of the data analysis model in step 1 specifically includes: and accessing real-time data during power grid operation through the stack type message queue, and extracting full or incremental data from the relational database to obtain power grid model data.
Further, the multi-dimensional planning indexes selected in the step 1 comprise power supply reliability, line N-1 passing rate, line average power supply radius, wiring mode reasonable rate, user average distribution and transformation capacity, line average connection and connection capacity, unit asset sale increasing electric quantity and unit investment supply increasing load.
Further, the power grid planning feature analysis in step 1 includes: grid structure adaptability, power supply capability adaptability, equipment level adaptability, load characteristic adaptability and new element grid connection adaptability.
Further, the process of constructing the hybrid storage mode in step 2 is to introduce a non-relational database into the power grid planning system, utilize the non-relational database to realize the rapid storage and reading of the power grid operation measurement data, and use the traditional relational storage power grid equipment model data to construct and form the hybrid storage mode of the relational database and the non-relational database.
Further, the establishing of the neural network model in step 2 specifically includes: and selecting a BPNN neural network algorithm in machine learning.
Further, the extracting of the planning feature elements in step 2 specifically includes: station room planning elements and net rack planning elements.
Further, in step 1, the key is adopted to extract the full or incremental data from the relational database.
Further, the button is composed of a Spoon unit, a Pan unit, a jobe unit and a Kitchen unit, the Spoon unit allows a graphical interface to be used for achieving a data conversion process, the Pan unit runs the Spoon unit data conversion process in batches, the jobe unit monitors whether the instructions are executed and the execution speed, and the Kitchen unit runs in batches.
Further, the distribution of the indexes in the step 5 is displayed according to planning business classification, time and model dimension, the index details are checked, and classification display is performed according to historical data trend and load prediction.
The beneficial effects brought by one aspect of the invention are as follows: aiming at the construction result of the existing power grid planning simulation basic data platform, the invention utilizes big data technology and method to research the construction of the power grid planning auxiliary analysis system, and improves the capabilities of integration, calculation, analysis, application and the like of system planning service data. By comprehensive data source acquisition and high-performance data processing and analysis, an intelligent means is provided for practical application of power grid planning, the efficiency of power grid planning is improved, a certain auxiliary reference basis is provided for power grid planning departments, and a theoretical support and a new exploration direction are provided for promoting wide application of electric power big data.
The beneficial effects brought by one aspect of the invention are as follows: the invention provides a power grid planning adaptability method based on big data.
Drawings
FIG. 1 is a flow chart of steps of an intelligent auxiliary analysis method for power grid planning based on big data according to the present invention;
FIG. 2 is a flow chart of a BPNN algorithm in the big data-based intelligent auxiliary analysis method for power grid planning;
fig. 3 is a schematic diagram of an embodiment of a BPNN algorithm in an intelligent auxiliary analysis method based on big data power grid planning according to the present invention.
Detailed Description
In order to make the technical problems solved, technical solutions adopted and technical effects achieved by the present invention clearer, the technical solutions of the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, those skilled in the art can obtain the embodiments without creative efforts, and the embodiments belong to the protection scope of the present invention.
Example 1
The purpose of this embodiment is to further illustrate the implementation and attention points of the above technical solution, as shown in fig. 1, specifically as follows:
an intelligent auxiliary analysis method for power grid planning based on big data comprises the following steps:
step 2, establishing a neural network model, extracting planning characteristic elements, and constructing a mixed storage mode of a relational database and a non-relational database;
step 3, batch calculation, flow calculation, memory calculation, query calculation and distributed calculation are adopted to meet the calculation requirements of power grid planning services with different timeliness and different scenes;
step 4, based on the power grid planning service model data, through big data analysis modeling and model analysis, the support of big data distributed computation is increased, and the analysis and mining requirements of real-time and off-line application of the power grid planning service are met;
and 5, checking the statistical chart and displaying the index distribution.
The accessing of the power grid operation data and the obtaining of the data analysis model in the step 1 specifically include: and accessing real-time data during power grid operation through the stack type message queue, and extracting full or incremental data from the relational database to obtain power grid model data.
The multi-dimensional planning indexes selected in the step 1 comprise power supply reliability, line N-1 passing rate, line average power supply radius, wiring mode reasonable rate, user average distribution and transformation capacity, line average mounting capacity, unit asset sales increase electric quantity and unit investment supply increase load.
Wherein, the power grid planning feature analysis of step 1, its feature analysis content includes: grid structure adaptability, power supply capability adaptability, equipment level adaptability, load characteristic adaptability and new element grid connection adaptability.
The process of establishing the hybrid storage mode in the step 2 is to introduce a non-relational database into the power grid planning system, utilize the non-relational database to realize the rapid storage and reading of the power grid operation measurement data, and use the traditional relational storage power grid equipment model data to establish the hybrid storage mode for forming the relational database and the non-relational database.
Example 2
On the basis of embodiment 1, a power grid planning intelligent auxiliary analysis method based on big data is further extended, and an algorithm of a neural network model in the process is further introduced to ensure the timeliness and effectiveness of operation, and the specific process is as follows, as shown in fig. 1:
an intelligent auxiliary analysis method for power grid planning based on big data comprises the following steps:
step 2, establishing a neural network model, extracting planning characteristic elements, and constructing a mixed storage mode of a relational database and a non-relational database;
step 3, batch calculation, flow calculation, memory calculation, query calculation and distributed calculation are adopted to meet the calculation requirements of power grid planning services with different timeliness and different scenes;
step 4, based on the power grid planning service model data, through big data analysis modeling and model analysis, the support of big data distributed computation is increased, and the analysis and mining requirements of real-time and off-line application of the power grid planning service are met;
and 5, checking the statistical chart and displaying the index distribution.
The accessing of the power grid operation data and the obtaining of the data analysis model in the step 1 specifically include: and accessing real-time data during power grid operation through the stack type message queue, and extracting full or incremental data from the relational database to obtain power grid model data.
The multi-dimensional planning indexes selected in the step 1 comprise power supply reliability, line N-1 passing rate, line average power supply radius, wiring mode reasonable rate, user average distribution and transformation capacity, line average mounting capacity, unit asset sales increase electric quantity and unit investment supply increase load.
Wherein, the power grid planning feature analysis of step 1, its feature analysis content includes: grid structure adaptability, power supply capability adaptability, equipment level adaptability, load characteristic adaptability and new element grid connection adaptability.
The process of establishing the hybrid storage mode in the step 2 is to introduce a non-relational database into the power grid planning system, utilize the non-relational database to realize the rapid storage and reading of the power grid operation measurement data, and use the traditional relational storage power grid equipment model data to establish the hybrid storage mode for forming the relational database and the non-relational database.
The establishing of the neural network model in the step 2 specifically comprises the following steps: and selecting a BPNN neural network algorithm in machine learning.
The specific implementation process and the key points of the BPNN neural network algorithm adopted in the embodiment are as follows, as shown in fig. 2:
1. network weights and biases are initialized. The connection weights between different neurons are different, so in the initialization stage, each network connection weight needs to be given a very small random number, namely-1.0 to 1.0 or-0.5 to 0.5, and each neuron has a bias and is initialized to be a random number.
2. Forward propagation is performed. A training sample is input, and then the output of each neuron is obtained through calculation. The calculation method of each neuron is the same, and is obtained by linear combination of its inputs, as shown in fig. 3 for further explanation:
, let < script type = "math/tex" id = "math jax-Element-7" > W ^ l _ { ij } </script > mean that the < script type = "math/tex" id = "math jax-Element-8" >/mit > layer < script type = "math/tex" id = "math jax-Element-9" >/mit i descriptor > node and < script type = "math/tex" id = "math jax-Element-10" >/l +1</script > layer < script type = "math/tex" = "math/type-11" >/j </script > node.
Wherein, the < script type = "math/tex" id = "MathJax-Element-12" >/mit = "math/tex" id = "MathJax-Element-13" >/mit | ", the weight between the L1 and the L2 layer in the figure is < script type =" math/tex "id =" MathJax-Element-14"> W ^1 ^ ij } </script >, the weight between the L2 and the L3 layer is < script type =" math/tex "=" MathJax-Element-15"> W ^2 =/ij }. , let < script type = "mat/tex" id = "math jax-Element-16" > b ^ l _ i </script > represent the offset item of < script type = "math/tex" id = "math jax-Element-17" >/mit l +1</script > layer < script type = "math/tex" id = "math jax-Element-18" >/mit/i </script > nodes. , the < script type = "mat/tex" id = "math jax-Element-19" > s ^ l _ j </script > means the input value of < script type = "mat/tex" id = "math jax-Element-20" > l +1</script > layer < script type = "mat/tex" id = "math jax-Element-21" > j </script > nodes.
When < script type = "mat/tex" id = "math jax-Element-22" > l = 1</script >, the < script type = "mat/tex" id = "math jax-Element-23" > s ^1_ j = \ sum _ { i =1} < m W ^1_ { ij }, x _ i + b ^1_ j </script >. , the < script type = "math/tex" id = "math jax-Element-24" > theta (s ^ l _ j) </script > represents the output value of < script type = "math/tex" id = "math jax-Element-25" > l +1</script > layer < script type = "math/tex" id = "math jax-Element-26" > j </script > nodes after the activation function < script type = "math/tex" id = "math jax-Element-27" > theta (x) </script >.
The following equation can be obtained:
<script type="math/tex; mode=display" id="MathJax-Element-102">
thus, one training is completed, and the output result < script type = "mat/tex" id = "MathJax-Element-103" > h _ { W, b } (x) </script > is obtained.
3. The error is calculated and counter-propagated. The final purpose is to ensure the maximum consistency of the output energy after the algorithm and the true value. When the output value is inconsistent with the true value, an error is inevitably generated, and the smaller the error is, the better the prediction effect of the algorithm is represented. How the connection weight affects the output result is further explained below:
assume that the connection weight from the ith node of the input layer to the jth node of the hidden layer has a small change
< script type = "mat/tex" id = "math-Element-104" > Δ w _ = "{ ij }/script >, then < script type =" mat/tex "id =" math-Element-105 "> Δ w _ [ { ij }/script > will have an effect on < script type =" mat/tex "id =" math jax-Element-106"> s _ j ] </script >, causing < script type =" mat/tex "id =" math jax-Element-106"> s _ j ] </script > to also appear a change < script type =" id = "math/tex" = "id =" math jax-Element-107"> s _ j ] </script >, then the" script type/tex "id" = "math-Element-108" = "(" math _ j "</script" -), "script" = "is output to < script layer" = "(" math/script "=" ("math _ type" -), finally, an error Δ e is generated at all output layers. Therefore, the adjustment of the weight will cause the output result to change.
How to adjust the weights is further explained below:
For a given sample, the correct output and the output of the neural network will both produce an error, and it is clear that the smaller the error, the better the network will work. In general, the error magnitude is measured by minimizing the root mean square difference, and the formula is as follows: < script type = "mat/tex" id = "MathJax-Element-110" > l (e) = \ frac12SSE = \ frac12 Σ { j =1} { ke ^2_ j = \ frac12 Σ { j =1} { k (\\ overload { y _ j } -y _ j) ^2</script >.
In order to minimize the error, a gradient descent method is adopted, that is, the weight of each sample is changed towards the negative gradient direction, that is, the gradient of the error L to the weight W is obtained.
For the weight from the input layer to the hidden layer, < script type = "math/tex" id = "MathJax-Element-111" >/frac { \ partial L } { \ partial W ^1 { } ij = \ frac { } partial L } { \\ partial s ^1_ j } } fr \ ac { } partial s ^1_ j } { \ partial W ^1 { ij } } for the weight from the input layer to the hidden layer, the < script type = "math/tex" id "{ \\\ partial W {/m { } 1 \\\\ p \\ m { < 1 { } m { =1 \ text { } m { } 1 \\\ p \\ partial W { } 2 \\\\\\\\ spatial j } may be converted into the above-primitive { < script { < text } 1 \\\ text { } m { < text { } 1 \\\\\ text { } text { < text } text { < -114 ">/frac { \\ partial L } { \ partial W ^1 ^ ij } } = \ frac { \ partial L } { \ partial s ^1_ j } - } x _ i ^ is then required to require < script type =" math/tex "id =" mathJaxElement-115 ">/frac { \ partial L } { \ partial s ^1_ j } </pt >,
because all < script type = "mat/tex" id = "math jax-Element-116" > s ^1_ j </script > have an effect on < script type = "mat/tex" id = "math jax-Element-117" > s ^2_ i </script >.
< script type = "math/tex" id = "MathJax-Element-118" > s ^2_ i = \ sum _ { j =1} ^ n W ^2_ { ji } - (. s ^1_ j) + b ^2_ i ] } we can convert < script type = "math/tex" id = "MathJax-Element-119" >/frac { \\ partial L } { \ partial s ^1_ j } </script >:
<script type="math/tex" id="MathJax-Element-120">\frac{\partial L}{\partial s^1_j} = ∑_{i=1}^k \frac{\partial L}{\partial s^2_i} ⋅ \frac{\partial s^2_i}{\partial s^1_j}</script> <script type="math/tex" id="MathJax-Element-121"> \frac{\partial s^2_i}{\partial s^1_j} = \frac{\partial s^2_i}{\partial θ(s^1_j) } ⋅ \frac{\partialθ(s^1_j) }{\partial s^1_j} = W^2_{ji} ⋅θ′(s^1_j) </script> <script type="math/tex" id="MathJax-Element-122"> \frac{\partial L}{\partial s^1_j} =θ′(s^1_j) ∑_{i=1}^k \frac{\partial L}{\partial s^2_i} ⋅W^2_{ji} </script>
<script type="math/tex" id="MathJax-Element-123"> \delta ^l_i = \frac{\partial L}{\partial s^l_i} </script>。
one rule is seen according to the above operation: the weight gradient of each layer is equal to the input of the previous layer to which the weight of this layer is connected multiplied by the inverted output of the connected next layer.
For example, the weight gradient connected between the input layer and the hidden layer is equal to the input < script type = "math/tex" id = "math jax-Element-135" > x _ i </script > of the output layer multiplied by the reverse output < script type = "math/tex" id = "math jax-Element-136" >/delta ^1_ j </script > of the hidden layer, that is, < script type = "math/tex" id = "math jax-Element-137" >/delta ^1_ j > < x _ i </script >.
4, the weights of the network and the bias of the neural network elements are adjusted, and the weights are updated after the weight gradient is formed.
<script type="math/tex" id="MathJax-Element-138"> W^l_{ij} = W^l_{ij} - \alpha \frac{\partial L}{\partial W^l_{ij}}</script>。
5, the judgment is finished. And for each sample, judging whether the error is smaller than a set threshold value or the iteration number is reached, finishing the training, and returning to the second step to continue the training if the error is not smaller than the set threshold value or the iteration number is reached.
Example 3
On the basis of embodiment 2, a power grid planning intelligent auxiliary analysis method based on big data is further extended and explained, and the specific process is as follows, as shown in fig. 1:
an intelligent auxiliary analysis method for power grid planning based on big data comprises the following steps:
step 2, establishing a neural network model, extracting planning characteristic elements, and constructing a mixed storage mode of a relational database and a non-relational database;
step 3, batch calculation, flow calculation, memory calculation, query calculation and distributed calculation are adopted to meet the calculation requirements of power grid planning services with different timeliness and different scenes;
step 4, based on the power grid planning service model data, through big data analysis modeling and model analysis, the support of big data distributed computation is increased, and the analysis and mining requirements of real-time and off-line application of the power grid planning service are met;
and 5, checking the statistical chart and displaying the index distribution.
The accessing of the power grid operation data and the obtaining of the data analysis model in the step 1 specifically include: and accessing real-time data during power grid operation through the stack type message queue, and extracting full or incremental data from the relational database to obtain power grid model data.
The multi-dimensional planning indexes selected in the step 1 comprise power supply reliability, line N-1 passing rate, line average power supply radius, wiring mode reasonable rate, user average distribution and transformation capacity, line average mounting capacity, unit asset sales increase electric quantity and unit investment supply increase load.
Wherein, the power grid planning feature analysis of step 1, its feature analysis content includes: grid structure adaptability, power supply capability adaptability, equipment level adaptability, load characteristic adaptability and new element grid connection adaptability.
The process of establishing the hybrid storage mode in the step 2 is to introduce a non-relational database into the power grid planning system, utilize the non-relational database to realize the rapid storage and reading of the power grid operation measurement data, and use the traditional relational storage power grid equipment model data to establish the hybrid storage mode for forming the relational database and the non-relational database.
The establishing of the neural network model in the step 2 specifically comprises the following steps: and selecting a BPNN neural network algorithm in machine learning.
The specific implementation process and the key points of the BPNN neural network algorithm adopted in the embodiment are shown in fig. 2, and are as follows:
1. network weights and biases are initialized. The connection weights between different neurons are different, so in the initialization stage, each network connection weight needs to be given a very small random number, namely-1.0 to 1.0 or-0.5 to 0.5, and each neuron has a bias and is initialized to be a random number.
2. Forward propagation is performed. A training sample is input, and then the output of each neuron is obtained through calculation. The calculation method of each neuron is the same, and is obtained by linear combination of its inputs, as shown in fig. 3 for further explanation:
, let < script type = "math/tex" id = "math jax-Element-7" > W ^ l _ { ij } </script > mean that the < script type = "math/tex" id = "math jax-Element-8" >/mit > layer < script type = "math/tex" id = "math jax-Element-9" >/mit i descriptor > node and < script type = "math/tex" id = "math jax-Element-10" >/l +1</script > layer < script type = "math/tex" = "math/type-11" >/j </script > node.
Wherein, the < script type = "math/tex" id = "MathJax-Element-12" >/mit = "math/tex" id = "MathJax-Element-13" >/mit | ", the weight between the L1 and the L2 layer in the figure is < script type =" math/tex "id =" MathJax-Element-14"> W ^1 ^ ij } </script >, the weight between the L2 and the L3 layer is < script type =" math/tex "=" MathJax-Element-15"> W ^2 =/ij }. , let < script type = "mat/tex" id = "math jax-Element-16" > b ^ l _ i </script > represent the offset item of < script type = "math/tex" id = "math jax-Element-17" >/mit l +1</script > layer < script type = "math/tex" id = "math jax-Element-18" >/mit/i </script > nodes. , the < script type = "mat/tex" id = "math jax-Element-19" > s ^ l _ j </script > means the input value of < script type = "mat/tex" id = "math jax-Element-20" > l +1</script > layer < script type = "mat/tex" id = "math jax-Element-21" > j </script > nodes.
When < script type = "mat/tex" id = "math jax-Element-22" > l = 1</script >, the < script type = "mat/tex" id = "math jax-Element-23" > s ^1_ j = \ sum _ { i =1} < m W ^1_ { ij }, x _ i + b ^1_ j </script >. , the < script type = "math/tex" id = "math jax-Element-24" > theta (s ^ l _ j) </script > represents the output value of < script type = "math/tex" id = "math jax-Element-25" > l +1</script > layer < script type = "math/tex" id = "math jax-Element-26" > j </script > nodes after the activation function < script type = "math/tex" id = "math jax-Element-27" > theta (x) </script >.
The following equation can be obtained:
<script type="math/tex; mode=display" id="MathJax-Element-102">
thus, one training is completed, and the output result < script type = "mat/tex" id = "MathJax-Element-103" > h _ { W, b } (x) </script > is obtained.
3. The error is calculated and counter-propagated. The final purpose is to ensure the maximum consistency of the output energy after the algorithm and the true value. When the output value is inconsistent with the true value, an error is inevitably generated, and the smaller the error is, the better the prediction effect of the algorithm is represented. How the connection weight affects the output result is further explained below:
assume that the connection weight from the ith node of the input layer to the jth node of the hidden layer has a small change
< script type = "mat/tex" id = "math-Element-104" > Δ w _ = "{ ij }/script >, then < script type =" mat/tex "id =" math-Element-105 "> Δ w _ [ { ij }/script > will have an effect on < script type =" mat/tex "id =" math jax-Element-106"> s _ j ] </script >, causing < script type =" mat/tex "id =" math jax-Element-106"> s _ j ] </script > to also appear a change < script type =" id = "math/tex" = "id =" math jax-Element-107"> s _ j ] </script >, then the" script type/tex "id" = "math-Element-108" = "(" math _ j "</script" -), "script" = "is output to < script layer" = "(" math/script "=" ("math _ type" -), finally, an error Δ e is generated at all output layers. Therefore, the adjustment of the weight will cause the output result to change.
How to adjust the weights is further explained below:
For a given sample, the correct output and the output of the neural network will both produce an error, and it is clear that the smaller the error, the better the network will work. In general, the error magnitude is measured by minimizing the root mean square difference, and the formula is as follows:
<script type="math/tex" id="MathJax-Element-110">L(e)=\frac12SSE=\frac12∑_{j=1}^ke^2_j=\frac12∑_{j=1}^k(\overline{y_j}−y_j)^2</script>。
in order to minimize the error, a gradient descent method is adopted, that is, the weight of each sample is changed towards the negative gradient direction, that is, the gradient of the error L to the weight W is obtained.
For the weight from the input layer to the hidden layer, < script type = "math/tex" id = "MathJax-Element-111" >/frac { \ partial L } { \ partial W ^1 { } ij = \ frac { } partial L } { \\ partial s ^1_ j } } fr \ ac { } partial s ^1_ j } { \ partial W ^1 { ij } } for the weight from the input layer to the hidden layer, the < script type = "math/tex" id "{ \\\ partial W {/m { } 1 \\\\ p \\ m { < 1 { } m { =1 \ text { } m { } 1 \\\ p \\ partial W { } 2 \\\\\\\\ spatial j } may be converted into the above-primitive { < script { < text } 1 \\\ text { } m { < text { } 1 \\\\\ text { } text { < text } text { < -114 ">/frac { \\ partial L } { \ partial W ^1 ^ ij } } = \ frac { \ partial L } { \ partial s ^1_ j } - } x _ i ^ is then required to require < script type =" math/tex "id =" mathJaxElement-115 ">/frac { \ partial L } { \ partial s ^1_ j } </pt >,
because all < script type = "mat/tex" id = "math jax-Element-116" > s ^1_ j </script > have an effect on < script type = "mat/tex" id = "math jax-Element-117" > s ^2_ i </script >:
< script type = "math/tex" id = "MathJax-Element-118" > s ^2_ i = \ sum _ { j =1} ^ n W ^2_ { ji } - (. s ^1_ j) + b ^2_ i ] } we can convert < script type = "math/tex" id = "MathJax-Element-119" >/frac { \\ partial L } { \ partial s ^1_ j } </script >:
<script type="math/tex" id="MathJax-Element-120">\frac{\partial L}{\partial s^1_j} = ∑_{i=1}^k \frac{\partial L}{\partial s^2_i} ⋅ \frac{\partial s^2_i}{\partial s^1_j}</script> <script type="math/tex" id="MathJax-Element-121"> \frac{\partial s^2_i}{\partial s^1_j} = \frac{\partial s^2_i}{\partial θ(s^1_j) } ⋅ \frac{\partialθ(s^1_j) }{\partial s^1_j} = W^2_{ji} ⋅θ′(s^1_j) </script> <script type="math/tex" id="MathJax-Element-122"> \frac{\partial L}{\partial s^1_j} =θ′(s^1_j) ∑_{i=1}^k \frac{\partial L}{\partial s^2_i} ⋅W^2_{ji} </script>
<script type="math/tex" id="MathJax-Element-123"> \delta ^l_i = \frac{\partial L}{\partial s^l_i} </script>。
one rule is seen according to the above operation: the weight gradient of each layer is equal to the input of the previous layer to which the weight of this layer is connected multiplied by the inverted output of the connected next layer.
For example, the weight gradient connected between the input layer and the hidden layer is equal to the input < script type = "math/tex" id = "math jax-Element-135" > x _ i </script > of the output layer multiplied by the reverse output < script type = "math/tex" id = "math jax-Element-136" >/delta ^1_ j </script > of the hidden layer, that is, < script type = "math/tex" id = "math jax-Element-137" >/delta ^1_ j > < x _ i </script >.
4 network weights and neural network element bias adjustments. With the weight gradient, the weights can be updated.
<script type="math/tex" id="MathJax-Element-138"> W^l_{ij} = W^l_{ij} - \alpha \frac{\partial L}{\partial W^l_{ij}}</script>。
5, the judgment is finished. For each sample, we judge if its error is less than the threshold we set or the number of iterations has been reached. We finish training, otherwise continue back to the second step to continue training.
The extracting of the planning feature elements in the step 2 specifically includes: station room planning elements and net rack planning elements.
And 5, displaying the distribution of the indexes according to the planning service classification, time and model dimension, checking the index detail, and performing classification display according to the historical data trend and load prediction.
Example 4
On the basis of embodiment 3, a power grid planning intelligent auxiliary analysis method based on big data is further extended and explained, and the specific process is as follows, as shown in fig. 1:
an intelligent auxiliary analysis method for power grid planning based on big data comprises the following steps:
step 2, establishing a neural network model, extracting planning characteristic elements, and constructing a mixed storage mode of a relational database and a non-relational database;
step 3, batch calculation, flow calculation, memory calculation, query calculation and distributed calculation are adopted to meet the calculation requirements of power grid planning services with different timeliness and different scenes;
step 4, based on the power grid planning service model data, through big data analysis modeling and model analysis, the support of big data distributed computation is increased, and the analysis and mining requirements of real-time and off-line application of the power grid planning service are met;
and 5, checking the statistical chart and displaying the index distribution.
The accessing of the power grid operation data and the obtaining of the data analysis model in the step 1 specifically include: and accessing real-time data during power grid operation through the stack type message queue, and extracting full or incremental data from the relational database to obtain power grid model data.
The multi-dimensional planning indexes selected in the step 1 comprise power supply reliability, line N-1 passing rate, line average power supply radius, wiring mode reasonable rate, user average distribution and transformation capacity, line average mounting capacity, unit asset sales increase electric quantity and unit investment supply increase load.
Wherein, the power grid planning feature analysis of step 1, its feature analysis content includes: grid structure adaptability, power supply capability adaptability, equipment level adaptability, load characteristic adaptability and new element grid connection adaptability.
The process of establishing the hybrid storage mode in the step 2 is to introduce a non-relational database into the power grid planning system, utilize the non-relational database to realize the rapid storage and reading of the power grid operation measurement data, and use the traditional relational storage power grid equipment model data to establish the hybrid storage mode for forming the relational database and the non-relational database.
The establishing of the neural network model in the step 2 specifically comprises the following steps: and selecting a BPNN neural network algorithm in machine learning.
The specific implementation process and the key points of the BPNN neural network algorithm adopted in the embodiment are as follows, as shown in fig. 2:
1. network weights and biases are initialized. The connection weights between different neurons are different, so in the initialization stage, each network connection weight needs to be given a very small random number, namely-1.0 to 1.0 or-0.5 to 0.5, and each neuron has a bias and is initialized to be a random number.
2. Forward propagation is performed. A training sample is input, and then the output of each neuron is obtained through calculation. The calculation method of each neuron is the same, and is obtained by linear combination of its inputs, as shown in fig. 3 for further explanation:
, let < script type = "math/tex" id = "math jax-Element-7" > W ^ l _ { ij } </script > mean that the < script type = "math/tex" id = "math jax-Element-8" >/mit > layer < script type = "math/tex" id = "math jax-Element-9" >/mit i descriptor > node and < script type = "math/tex" id = "math jax-Element-10" >/l +1</script > layer < script type = "math/tex" = "math/type-11" >/j </script > node.
Wherein, the < script type = "math/tex" id = "MathJax-Element-12" >/mit = "math/tex" id = "MathJax-Element-13" >/mit | ", the weight between the L1 and the L2 layer in the figure is < script type =" math/tex "id =" MathJax-Element-14"> W ^1 ^ ij } </script >, the weight between the L2 and the L3 layer is < script type =" math/tex "=" MathJax-Element-15"> W ^2 =/ij }. , let < script type = "mat/tex" id = "math jax-Element-16" > b ^ l _ i </script > represent the offset item of < script type = "math/tex" id = "math jax-Element-17" >/mit l +1</script > layer < script type = "math/tex" id = "math jax-Element-18" >/mit/i </script > nodes. , the < script type = "mat/tex" id = "math jax-Element-19" > s ^ l _ j </script > means the input value of < script type = "mat/tex" id = "math jax-Element-20" > l +1</script > layer < script type = "mat/tex" id = "math jax-Element-21" > j </script > nodes.
When < script type = "mat/tex" id = "math jax-Element-22" > l = 1</script >, the < script type = "mat/tex" id = "math jax-Element-23" > s ^1_ j = \ sum _ { i =1} < m W ^1_ { ij }, x _ i + b ^1_ j </script >. , the < script type = "math/tex" id = "math jax-Element-24" > theta (s ^ l _ j) </script > represents the output value of < script type = "math/tex" id = "math jax-Element-25" > l +1</script > layer < script type = "math/tex" id = "math jax-Element-26" > j </script > nodes after the activation function < script type = "math/tex" id = "math jax-Element-27" > theta (x) </script >.
The following equation can be obtained:
<script type="math/tex; mode=display" id="MathJax-Element-102">
thus, one training is completed, and the output result < script type = "mat/tex" id = "MathJax-Element-103" > h _ { W, b } (x) </script > is obtained.
3. The error is calculated and counter-propagated. The final purpose is to ensure the maximum consistency of the output energy after the algorithm and the true value. When the output value is inconsistent with the true value, an error is inevitably generated, and the smaller the error is, the better the prediction effect of the algorithm is represented. How the connection weight affects the output result is further explained below:
assume that the connection weight from the ith node of the input layer to the jth node of the hidden layer has a small change
< script type = "mat/tex" id = "math-Element-104" > Δ w _ = "{ ij }/script >, then < script type =" mat/tex "id =" math-Element-105 "> Δ w _ [ { ij }/script > will have an effect on < script type =" mat/tex "id =" math jax-Element-106"> s _ j ] </script >, causing < script type =" mat/tex "id =" math jax-Element-106"> s _ j ] </script > to also appear a change < script type =" id = "math/tex" = "id =" math jax-Element-107"> s _ j ] </script >, then the" script type/tex "id" = "math-Element-108" = "(" math _ j "</script" -), "script" = "is output to < script layer" = "(" math/script "=" ("math _ type" -), finally, an error Δ e is generated at all output layers. Therefore, the adjustment of the weight will cause the output result to change.
How to adjust the weights is further explained below:
For a given sample, the correct output and the output of the neural network will both produce an error, and it is clear that the smaller the error, the better the network will work. In general, the error magnitude is measured by minimizing the root mean square difference, and the formula is as follows: < script type = "mat/tex" id = "MathJax-Element-110" > l (e) = \ frac12SSE = \ frac12 Σ { j =1} { ke ^2_ j = \ frac12 Σ { j =1} { k (\\ overload { y _ j } -y _ j) ^2</script >.
In order to minimize the error, a gradient descent method is adopted, that is, the weight of each sample is changed towards the negative gradient direction, that is, the gradient of the error L to the weight W is obtained.
For the weight from the input layer to the hidden layer, < script type = "math/tex" id = "MathJax-Element-111" >/frac { \ partial L } { \ partial W ^1 { } ij = \ frac { } partial L } { \\ partial s ^1_ j } } fr \ ac { } partial s ^1_ j } { \ partial W ^1 { ij } } for the weight from the input layer to the hidden layer, the < script type = "math/tex" id "{ \\\ partial W {/m { } 1 \\\\ p \\ m { < 1 { } m { =1 \ text { } m { } 1 \\\ p \\ partial W { } 2 \\\\\\\\ spatial j } may be converted into the above-primitive { < script { < text } 1 \\\ text { } m { < text { } 1 \\\\\ text { } text { < text } text { < -114 ">/frac { \\ partial L } { \ partial W ^1 ^ ij } } = \ frac { \ partial L } { \ partial s ^1_ j } - } x _ i ^ is then required to require < script type =" math/tex "id =" mathJaxElement-115 ">/frac { \ partial L } { \ partial s ^1_ j } </pt >,
because all < script type = "mat/tex" id = "math jax-Element-116" > s ^1_ j </script > have an effect on < script type = "mat/tex" id = "math jax-Element-117" > s ^2_ i </script >:
< script type = "math/tex" id = "MathJax-Element-118" > s ^2_ i = \ sum _ { j =1} ^ n W ^2_ { ji } - (. s ^1_ j) + b ^2_ i ] } we can convert < script type = "math/tex" id = "MathJax-Element-119" >/frac { \\ partial L } { \ partial s ^1_ j } </script >:
<script type="math/tex" id="MathJax-Element-120">\frac{\partial L}{\partial s^1_j} = ∑_{i=1}^k \frac{\partial L}{\partial s^2_i} ⋅ \frac{\partial s^2_i}{\partial s^1_j}</script> <script type="math/tex" id="MathJax-Element-121"> \frac{\partial s^2_i}{\partial s^1_j} = \frac{\partial s^2_i}{\partial θ(s^1_j) } ⋅ \frac{\partialθ(s^1_j) }{\partial s^1_j} = W^2_{ji} ⋅θ′(s^1_j) </script> <script type="math/tex" id="MathJax-Element-122"> \frac{\partial L}{\partial s^1_j} =θ′(s^1_j) ∑_{i=1}^k \frac{\partial L}{\partial s^2_i} ⋅W^2_{ji} </script>
<script type="math/tex" id="MathJax-Element-123"> \delta ^l_i = \frac{\partial L}{\partial s^l_i} </script>。
one rule is seen according to the above operation: the weight gradient of each layer is equal to the input of the previous layer to which the weight of this layer is connected multiplied by the inverted output of the connected next layer.
For example, the weight gradient connected between the input layer and the hidden layer is equal to the input < script type = "math/tex" id = "math jax-Element-135" > x _ i </script > of the output layer multiplied by the reverse output < script type = "math/tex" id = "math jax-Element-136" >/delta ^1_ j </script > of the hidden layer, that is, < script type = "math/tex" id = "math jax-Element-137" >/delta ^1_ j > < x _ i </script >.
4 network weights and neural network element bias adjustments. With the weight gradient, the weights can be updated.
<script type="math/tex" id="MathJax-Element-138"> W^l_{ij} = W^l_{ij} - \alpha \frac{\partial L}{\partial W^l_{ij}}</script>。
5, the judgment is finished. For each sample, we judge if its error is less than the threshold we set or the number of iterations has been reached. We finish training, otherwise continue back to the second step to continue training.
The extracting of the planning feature elements in the step 2 specifically includes: station room planning elements and net rack planning elements.
And 5, displaying the distribution of the indexes according to the planning service classification, time and model dimension, checking the index detail, and performing classification display according to the historical data trend and load prediction.
In step 1, extracting full or incremental data from the relational database by using a button. The button comprises a Spoon unit, a Pan unit, a job unit and a Kitchen unit, wherein the Spoon unit allows a graphical interface to be used for realizing a data conversion process, the Pan unit runs the Spoon unit data conversion process in batches, the job unit monitors whether instructions are executed and the execution speed, and the Kitchen unit runs in batches.
Claims (10)
1. An intelligent auxiliary analysis method for power grid planning based on big data is characterized by comprising the following steps:
step 1, selecting a multi-dimensional planning index, analyzing power grid planning characteristics, accessing real-time data of a power grid operation mode and acquiring a data analysis model; step 2, establishing a neural network model, extracting planning characteristic elements, and establishing a mixed storage mode of a relational database and a non-relational database; step 3, batch calculation, flow calculation, memory calculation, query calculation and distributed calculation are adopted to meet the calculation requirements of power grid planning services of different timeliness and different scenes; step 4, based on the power grid planning service model data, through big data analysis modeling and model analysis, the support of big data distributed computation is increased, and the analysis and mining requirements of real-time off-line application of the power grid planning service are met; and 5, checking the statistical chart and displaying the index distribution.
2. The big data-based power grid planning intelligent auxiliary analysis method according to claim 1, wherein the big data-based power grid planning intelligent auxiliary analysis method comprises the following steps: accessing power grid operation data and acquiring a data analysis model in the step 1, specifically, extracting full or incremental data from a relational database to acquire power grid model data through real-time data during power grid operation accessed by a stack type message queue.
3. The big data-based power grid planning intelligent auxiliary analysis method according to claim 1, wherein the big data-based power grid planning intelligent auxiliary analysis method comprises the following steps: and (2) selecting the multidimensional planning indexes in the step (1), wherein the indexes comprise power supply reliability, line N-1 passing rate, line average power supply radius, wiring mode reasonable rate, user average distribution and transformation capacity, line average mounting capacity, unit asset sale increasing electric quantity and unit investment supply increasing load.
4. The big data-based power grid planning intelligent auxiliary analysis method according to claim 1, wherein the big data-based power grid planning intelligent auxiliary analysis method comprises the following steps: the power grid planning feature analysis in the step 1 includes the following feature analysis contents: grid structure adaptability, power supply capability adaptability, equipment level adaptability, load characteristic adaptability and new element grid connection adaptability.
5. The big data-based power grid planning intelligent auxiliary analysis method according to claim 1, wherein the big data-based power grid planning intelligent auxiliary analysis method comprises the following steps: and 2, establishing a hybrid storage mode, wherein a non-relational database is introduced into the power grid planning system, the non-relational database is used for realizing the rapid storage and reading of the power grid operation measurement data, and the traditional relational storage power grid equipment model data is used for establishing the hybrid storage mode for forming the relational database and the non-relational database.
6. The big data-based power grid planning intelligent auxiliary analysis method according to claim 1, wherein the big data-based power grid planning intelligent auxiliary analysis method comprises the following steps: the establishing of the neural network model in the step 2 specifically comprises the following steps: and selecting a BPNN neural network algorithm in machine learning.
7. The big data-based power grid planning intelligent auxiliary analysis method according to claim 1, wherein the big data-based power grid planning intelligent auxiliary analysis method comprises the following steps: the extracting of the planning feature elements in the step 2 specifically includes: station room planning elements and net rack planning elements.
8. The big data-based power grid planning intelligent auxiliary analysis method according to claim 1, wherein the big data-based power grid planning intelligent auxiliary analysis method comprises the following steps: in the step 1, the key is adopted to extract the full or incremental data from the relational database.
9. The big data-based power grid planning intelligent auxiliary analysis method according to claim 8, wherein: the button comprises a Spoon unit, a Pan unit, a job unit and a Kitchen unit, wherein the Spoon unit allows a graphical interface to be used for realizing a data conversion process, the Pan unit runs the Spoon unit data conversion process in batches, the job unit monitors whether instructions are executed and the execution speed, and the Kitchen unit runs in batches.
10. The big data-based power grid planning intelligent auxiliary analysis method according to claim 1, wherein the big data-based power grid planning intelligent auxiliary analysis method comprises the following steps: and 5, displaying the distribution of the indexes in the step 5 according to the planning service classification, time and model dimension, checking the index detail, and performing classification display according to the historical data trend and load prediction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011181993.9A CN112396294A (en) | 2020-10-29 | 2020-10-29 | Power grid planning intelligent auxiliary analysis method based on big data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011181993.9A CN112396294A (en) | 2020-10-29 | 2020-10-29 | Power grid planning intelligent auxiliary analysis method based on big data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112396294A true CN112396294A (en) | 2021-02-23 |
Family
ID=74597394
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011181993.9A Pending CN112396294A (en) | 2020-10-29 | 2020-10-29 | Power grid planning intelligent auxiliary analysis method based on big data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112396294A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108108517A (en) * | 2017-11-23 | 2018-06-01 | 江苏瑞中数据股份有限公司 | A kind of Electric Power Network Planning intelligence aided analysis method based on big data |
| US20190087529A1 (en) * | 2014-03-24 | 2019-03-21 | Imagars Llc | Decisions with Big Data |
| CN109657959A (en) * | 2018-12-12 | 2019-04-19 | 国家电网有限公司 | A kind of distribution network planning calculation and analysis methods containing multivariate data |
| CN111582626A (en) * | 2020-03-17 | 2020-08-25 | 上海博英信息科技有限公司 | Power grid planning adaptability method based on big data |
-
2020
- 2020-10-29 CN CN202011181993.9A patent/CN112396294A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190087529A1 (en) * | 2014-03-24 | 2019-03-21 | Imagars Llc | Decisions with Big Data |
| CN108108517A (en) * | 2017-11-23 | 2018-06-01 | 江苏瑞中数据股份有限公司 | A kind of Electric Power Network Planning intelligence aided analysis method based on big data |
| CN109657959A (en) * | 2018-12-12 | 2019-04-19 | 国家电网有限公司 | A kind of distribution network planning calculation and analysis methods containing multivariate data |
| CN111582626A (en) * | 2020-03-17 | 2020-08-25 | 上海博英信息科技有限公司 | Power grid planning adaptability method based on big data |
Non-Patent Citations (1)
| Title |
|---|
| 姜囡: "《语音信号识别技术与实践》", 30 December 2019 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Innovation efficiency of semiconductor industry in China: A new framework based on generalized three-stage DEA analysis | |
| Hilorme et al. | Smart grid concept as a perspective for the development of Ukrainian energy platform | |
| CN110738435A (en) | distribution network project investment decision evaluation method | |
| CN110689162B (en) | Bus load prediction method, device and system based on user side classification | |
| CN109325880A (en) | A kind of Mid-long term load forecasting method based on Verhulst-SVM | |
| Liu et al. | A systematic procurement supply chain optimization technique based on industrial internet of things and application | |
| Panda et al. | Long term electrical load forecasting: an empirical study across techniques and domains | |
| CN107220758A (en) | A kind of Electric Power Network Planning accessory system | |
| Zhang et al. | Load Prediction Based on Hybrid Model of VMD‐mRMR‐BPNN‐LSSVM | |
| Song et al. | Cloud edge collaborative service composition optimization for intelligent manufacturing | |
| Ding et al. | [Retracted] Low Carbon Economy Assessment in China Using the Super‐SBM Model | |
| Ren et al. | Analysis of the Effect of the Line‐Seru Conversion on the Waiting Time with Batch Arrival | |
| CN106875057A (en) | A kind of electric power meter short term need Forecasting Methodology based on conditional probability adjustment | |
| CN112396294A (en) | Power grid planning intelligent auxiliary analysis method based on big data | |
| Cano-Martínez et al. | Dynamic energy prices for residential users based on Deep Learning prediction models of consumption and renewable generation | |
| Wu et al. | Research on cost forecasting based on the BIM and neural network | |
| Chen et al. | [Retracted] Tracking Control of the Dynamic Input‐Output Economic System Based on Data Fusion | |
| Xiao et al. | Research on application of data mining technology in financial decision support system | |
| Wang et al. | Data analytics enabled power marketing analysis and decision-making supporting system | |
| Song et al. | Application of Operational Research and Cybernetics in Intelligent Management System of New Energy Electronic and Electrical Industry | |
| Zhao et al. | [Retracted] Design and Implementation of Energy‐Saving Logistics Management System for Route Optimization | |
| Luo et al. | Calculation method and system of energy efficiency evaluation based on integrated power grid | |
| Zhang et al. | Research on the Allocation Method of Regional Science and Technology Resources from the Perspective of Rationality | |
| Cheng et al. | Research on Sales Forecasting Method for Transmission Parts of Customized Production | |
| Dou et al. | Analysis of Power Load Components Based on Neural Network |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210223 |
|
| RJ01 | Rejection of invention patent application after publication |