CN116307652A - Artificial intelligent resource allocation method for intelligent power grid - Google Patents
Artificial intelligent resource allocation method for intelligent power grid Download PDFInfo
- Publication number
- CN116307652A CN116307652A CN202310593660.4A CN202310593660A CN116307652A CN 116307652 A CN116307652 A CN 116307652A CN 202310593660 A CN202310593660 A CN 202310593660A CN 116307652 A CN116307652 A CN 116307652A
- Authority
- CN
- China
- Prior art keywords
- data
- photovoltaic
- vector
- data set
- calculation
- 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 149
- 238000013468 resource allocation Methods 0.000 title claims abstract description 29
- 239000013598 vector Substances 0.000 claims abstract description 162
- 238000004364 calculation method Methods 0.000 claims abstract description 76
- 230000005611 electricity Effects 0.000 claims abstract description 65
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 56
- 238000005265 energy consumption Methods 0.000 claims abstract description 51
- 238000004458 analytical method Methods 0.000 claims abstract description 47
- 238000010248 power generation Methods 0.000 claims abstract description 35
- 238000010801 machine learning Methods 0.000 claims abstract description 22
- 238000010606 normalization Methods 0.000 claims abstract description 18
- 238000007781 pre-processing Methods 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims description 75
- 238000013528 artificial neural network Methods 0.000 claims description 47
- 230000008569 process Effects 0.000 claims description 47
- 238000013527 convolutional neural network Methods 0.000 claims description 19
- 230000003044 adaptive effect Effects 0.000 claims description 16
- 230000009467 reduction Effects 0.000 claims description 16
- 238000006243 chemical reaction Methods 0.000 claims description 15
- 238000005286 illumination Methods 0.000 claims description 15
- 238000013135 deep learning Methods 0.000 claims description 14
- 238000007689 inspection Methods 0.000 claims description 12
- 238000009499 grossing Methods 0.000 claims description 11
- 125000004122 cyclic group Chemical group 0.000 claims description 10
- 238000013079 data visualisation Methods 0.000 claims description 10
- 238000012800 visualization Methods 0.000 claims description 10
- 230000036760 body temperature Effects 0.000 claims description 8
- 230000015654 memory Effects 0.000 claims description 8
- 210000002569 neuron Anatomy 0.000 claims description 8
- 208000024891 symptom Diseases 0.000 claims description 8
- 238000012937 correction Methods 0.000 claims description 7
- 238000003062 neural network model Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 6
- 230000000306 recurrent effect Effects 0.000 claims description 4
- 238000013473 artificial intelligence Methods 0.000 abstract description 10
- 238000012549 training Methods 0.000 description 20
- 238000005457 optimization Methods 0.000 description 14
- 238000007405 data analysis Methods 0.000 description 9
- 238000009826 distribution Methods 0.000 description 9
- 230000000694 effects Effects 0.000 description 9
- 238000007726 management method Methods 0.000 description 9
- 238000013461 design Methods 0.000 description 7
- 230000007787 long-term memory Effects 0.000 description 7
- 238000004519 manufacturing process Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 6
- 238000012423 maintenance Methods 0.000 description 6
- 238000013519 translation Methods 0.000 description 6
- 239000002699 waste material Substances 0.000 description 6
- 230000008901 benefit Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 5
- 230000007547 defect Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000011176 pooling Methods 0.000 description 3
- 238000000513 principal component analysis Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 2
- 230000001364 causal effect Effects 0.000 description 2
- 210000004027 cell Anatomy 0.000 description 2
- 238000007621 cluster analysis Methods 0.000 description 2
- 230000008034 disappearance Effects 0.000 description 2
- 230000008451 emotion Effects 0.000 description 2
- 238000004134 energy conservation Methods 0.000 description 2
- 238000004880 explosion Methods 0.000 description 2
- 230000002349 favourable effect Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012821 model calculation Methods 0.000 description 2
- 238000003058 natural language processing Methods 0.000 description 2
- 230000010355 oscillation Effects 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 230000002035 prolonged effect Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000001932 seasonal effect Effects 0.000 description 2
- 230000006403 short-term memory Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000009466 transformation Effects 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
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06315—Needs-based resource requirements planning or analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas 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
Abstract
The invention relates to the technical field of smart power grids, in particular to a smart power grid artificial intelligence resource allocation method. The method comprises the following steps: collecting the data of the light energy power generation area equipment in charge of the intelligent power grid system, preprocessing the data, generating vectors by using different data and generating a photovoltaic system data set by using a normalization standardized formula; performing photovoltaic energy consumption calculation on the photovoltaic system data set by using a machine learning algorithm, so as to generate a photovoltaic energy consumption data set; collecting historical electricity consumption, and performing power demand analysis and calculation on the historical electricity consumption by utilizing a data prediction model combining a momentum method and a back propagation network, so as to generate a power demand value; and carrying out load balancing analysis according to the power demand value and the photovoltaic energy consumption data set, thereby generating a dynamic resource allocation adjustment scheme. The invention can improve the efficiency of resource allocation of the intelligent power grid in the new energy field.
Description
Technical Field
The invention relates to the technical field of smart power grids, in particular to a smart power grid artificial intelligence resource allocation method.
Background
With the continuous development of science and technology, the electric power demand is larger and larger, and higher requirements are put on the efficiency of an electric power system. Smart grid concepts have grown. In a smart grid system, resources are reasonably allocated by using artificial intelligence technology so as to improve the efficiency and reliability of the power system. In the traditional power generation fields of the electric power field, such as the thermal power generation field and the hydroelectric power generation field, the resource allocation method is mature, but in the new energy power generation field, such as the photovoltaic power generation field, the adoption of the traditional allocation method can cause the problems of low efficiency and unreasonable resource allocation. The traditional allocation method has a certain limitation in the field of new energy, and cannot fully utilize data, so that resources are not fully utilized, and high cost is increased. Therefore, how to use artificial intelligence technology to allocate resources to new energy fields becomes a problem.
Disclosure of Invention
The invention provides an artificial intelligent resource allocation method for an intelligent power grid to solve at least one technical problem.
In order to achieve the above purpose, the invention provides a method for distributing artificial intelligence resources of a smart grid, which comprises the following steps:
Step S1: collecting the data of the light energy power generation area equipment in charge of the intelligent power grid system, and preprocessing the data to generate a photovoltaic system data set;
step S2: performing photovoltaic energy consumption calculation on the photovoltaic system data set by using a machine learning algorithm, so as to generate a photovoltaic energy consumption data set;
step S3: collecting historical electricity consumption, and performing power demand analysis and calculation on the historical electricity consumption by utilizing a data prediction model so as to generate a power demand value;
step S4: and carrying out load balancing analysis according to the power demand value and the photovoltaic energy consumption data set, thereby generating a dynamic resource allocation adjustment scheme.
According to the method, the collected data of the light energy power generation area equipment are subjected to data preprocessing to generate the data set of the photovoltaic system, so that the influence caused by the problems of noise, missing, outlier and the like of the collected data of the original light energy power generation area equipment is reduced, and the accuracy of the data is improved; according to the invention, the machine learning algorithm is adopted to calculate the photovoltaic energy consumption of the photovoltaic system data so as to generate the photovoltaic energy consumption data set, so that the current running state of the photovoltaic equipment and the performance of the photovoltaic equipment in future time can be accurately predicted. In addition, the machine learning algorithm is applied to the field of photovoltaic power generation, and can help engineers to quickly detect faults in photovoltaic equipment, including damage, defects and design problems, and can also search for links among data in a large scale through data analysis, so that self-adaptive control optimization is realized, and the service life of a photovoltaic module is prolonged; the invention collects the historical electricity consumption and utilizes the data prediction model to carry out the analysis and calculation of the electricity demand, can more accurately predict the future electricity demand, is beneficial to appointing a reasonable electricity distribution plan, and can reveal the past electricity consumption trend, consumption mode and electricity consumption gap among different areas so as to realize the load balance of the production/consumption modes. In addition, the machine learning model can further analyze the historical power consumption, determine energy conservation and reduce power consumption cost, and implement measures such as reactive compensation and load management through a specific time period or a specific user group; according to the invention, load balancing analysis is carried out according to the power demand value and the photovoltaic energy consumption data set, so that the intelligent power grid system can more efficiently distribute power production, thereby optimizing power distribution, avoiding resource waste and reducing energy consumption. In addition, the system can be more stable, and the efficiency of photovoltaic power generation can be improved.
Optionally, the data preprocessing in step S1 includes data quality inspection processing, feature processing calculation and normalization calculation, and specifically includes the following steps:
step S11: performing data quality inspection processing on the equipment data of the light energy power generation area so as to generate perfect equipment data;
step S12: performing feature processing calculation on the perfect equipment data so as to generate a conversion feature vector;
step S13: performing normalization standardized calculation on the conversion characteristic vector so as to generate a photovoltaic equipment data set;
step S14: and denoising the photovoltaic equipment data set to generate a photovoltaic system data set.
The invention performs data quality inspection and processing on the data of the photovoltaic power generation equipment, improves the operation efficiency of the photovoltaic power generation system, improves the equipment data, can help related staff to find various data indexes of the photovoltaic power generation system in time, and avoids potential safety hazards of the photovoltaic power generation system caused by data errors, defects and other problems, thereby ensuring the safety of the staff and the equipment. In addition, the perfect equipment data quality inspection optimizes the maintenance cost of the power generation system and improves the maintenance efficiency of the power generation system; the method has the advantages that the characteristic processing calculation is carried out on the perfect equipment data, the perfect equipment data are converted into characteristic vectors, some non-critical information and noise in the photovoltaic equipment are removed, and important characteristic information is reserved, so that the quality of the data is improved, the complexity of subsequent model calculation is reduced, and the training and prediction efficiency is improved; the conversion feature vectors are normalized, so that comparison among different features is fairer, the influence of certain features on the final training effect due to the fact that the numerical value is larger is avoided, the dimension influence is eliminated, all the features are unified into the same proportion range through normalization, further processing is facilitated, and high-quality photovoltaic equipment data sets are generated. In addition, the optimization speed of algorithms such as gradient descent is also increased.
Preferably, the feature processing calculation is performed on the perfect equipment data to generate a conversion feature vector, which specifically comprises the following steps:
performing first characteristic classification processing on the external environment temperature and the external environment humidity so as to obtain a temperature and humidity vector;
performing second feature classification processing on the illumination intensity so as to obtain an illumination vector;
performing third characteristic classification processing on the body temperature of the equipment, the effective irradiation area of the solar panel and the equipment angle, thereby obtaining an equipment vector;
and collecting the temperature and humidity vector, the illuminance vector and the equipment vector into a conversion characteristic vector.
According to the invention, different characteristics of different data are classified, so that redundant data can be avoided, the complexity of data processing is reduced, and the efficiency of data processing is improved; and a plurality of parameters are collected into the same feature vector, so that data analysis can be conveniently performed. According to the similarity and the difference between the feature vectors, the relation between the data can be more clearly grasped, and more accurate data analysis is realized; the various parameters are integrated to be used as the feature vector, so that the influence of the factors of the environment and the equipment on the result can be more comprehensively considered, and the prediction accuracy is improved. In addition, by integrating various parameters into feature vectors, the method can be used as a basis for intelligent control, and by detecting and learning the feature vectors in real time, the self-adaptive adjustment and intelligent control of data can be realized.
Further, in step S13, a normalization formula is adopted to calculate and obtain a photovoltaic device data set, where the normalization formula specifically includes:
wherein the method comprises the steps of
，/>
；
The invention firstly obtains the values of the temperature and humidity vector, the illuminance vector and the equipment vector and the characteristic values of the temperature and humidity vector, the illuminance vector and the equipment vector, and for each vector, the corresponding characteristic value needs to be found, and the values are usually obtained by using linear algebra, so that the characteristics of the temperature and humidity vector, the illuminance vector and the equipment vector meet the mean value and the unit variance, and the characteristic values of the temperature and humidity vector, the illuminance vector and the equipment vector are obtained by
The average value of the characteristic values is obtained by accumulation and summation, and the difference value is obtained by each vector to obtain the characteristic value +. >
，/>
Is the characteristic mean value of temperature and humidity vectors, and is->
Is the characteristic mean value of the illuminance vector, +.>
Is the characteristic mean value of the device vector; values using temperature and humidity vector, illuminance vector and device vector +.>
Subtracting the eigenvalue of each vector +.>
The square of the difference obtained>
Obtaining the characteristic variance of the temperature and humidity vector, the illuminance vector and the equipment vector>
，/>
Is the characteristic variance of the temperature and humidity vector, < >>
Characteristic variance of illuminance vector, ++>
The characteristic variance of the device vector; summing them to average +.>
The error of a certain vector is influenced by accidental limit data, and a normalized result photovoltaic equipment data set of a temperature and humidity vector, an illuminance vector and an equipment vector is further obtained in a summation average mode.
Optionally, denoising the photovoltaic device data set, thereby generating a photovoltaic system data set, including the steps of:
performing image visualization processing according to the photovoltaic equipment data set, so as to obtain a photovoltaic equipment data image;
carrying out local weighted regression calculation on the photovoltaic equipment data image by utilizing a local smoothing algorithm, thereby realizing denoising treatment and obtaining a photovoltaic image data set;
and performing digital dimension reduction calculation by using the photovoltaic image dataset to obtain a photovoltaic system dataset.
The invention adopts the technologies of computer vision, image processing, machine learning and the like to realize denoising treatment on the photovoltaic equipment data set, and utilizes the visualization treatment on the photovoltaic equipment data set to obtain a form which is easier to understand and analyze, and provides the focused characteristics and visualization to help staff to observe data better; the partial smoothing algorithm is utilized to remove noise and abnormal values of the photovoltaic equipment data image, so that the influence of factors in the real environment is reduced to a certain extent, and the efficiency of subsequent analysis is improved; the digital dimension reduction calculation is carried out based on the image output, so that the complexity of a model can be reduced, the training and prediction efficiency of the model can be improved, and meanwhile, important features in data can be derived by adopting a machine learning algorithm, so that the linear correlation is eliminated, the dimension of input data is reduced, and the advantage of processing a large amount of data is increased. And finally, the generated photovoltaic system data set is used for learning and calculating a subsequent model, and can provide more comprehensive and accurate information for researchers and engineers, so that the system design is optimized and the energy operation is improved, and the efficiency of artificial intelligent resource allocation of the intelligent power grid system is improved.
Optionally, the machine learning algorithm includes a convolutional neural network, a gating-based recurrent neural network and a deep learning network, specifically:
step S21: performing feature extraction on a photovoltaic system data set by using a convolutional neural network so as to generate a first-order photovoltaic symptom set;
step S22: performing data dimension reduction processing on the first-order photovoltaic collection by using a gate-based cyclic neural network, so as to generate a second-order photovoltaic collection;
step S23: and performing deep fitting on the second-order photovoltaic symptom set by using a deep learning neural network, so as to generate a photovoltaic energy consumption data set.
According to the method, the characteristics of the photovoltaic system data set are extracted through the convolutional neural network, the convolutional layer in the convolutional neural network performs local connection on input data, namely only a small input block is subjected to convolution operation to generate output, so that the problem of overfitting caused by full connection is avoided, the spatial structure information of the input data can be reserved, the data dimension and noise can be reduced, the stability and reliability of characteristic representation are improved, and a first-order photovoltaic collection is generated; the first-order photovoltaic collection is subjected to data dimension reduction processing based on the gating circulating neural network, and when the data dimension reduction is performed, information with little influence on a result can be screened out by using the gating mechanism, so that the dimension of the data is reduced, the efficiency and the accuracy of a model are improved, the calculation efficiency is effectively improved, the calculation load is reduced, and the prediction accuracy is improved, so that the second-order photovoltaic collection is generated; the second-order photovoltaic symptom sets are subjected to deep fitting through the deep learning neural network, and the current-voltage curve data and the power-voltage curve data of the real photovoltaic cells can be better approximated by utilizing the strong nonlinear modeling capability of the deep learning neural network, so that higher prediction precision is obtained, the precision and the robustness of the model are further improved, and a photovoltaic energy consumption data set is generated; the convolutional neural network, the gate-based cyclic neural network and the deep learning network respectively provide accurate and robust results for own tasks, and then the photovoltaic energy consumption correlation is more comprehensively expressed by combination. By selecting features, noise and redundancy are reduced, which can help remove unnecessary information and highlight the most important factors, thus making the overall process more efficient and reliable. In addition, the invention improves modeling accuracy and robustness, and can also support real-time data analysis and decision.
Optionally, the forgetting gate and the input gate of the long-term memory neural network control the state update and the input information of the neurons, so that the network can effectively process long-sequence dependency, and the problems of gradient disappearance and gradient explosion are avoided. Meanwhile, the output gate of the long-short-term memory neural network controls the output information flow, so that the neurons only output needed information, and redundant information and overfitting risks are reduced.
In addition, long-term memory neural networks have been widely used in speech recognition, machine translation, text classification, emotion analysis, and other tasks in natural language processing tasks. For example, in a machine translation task, the long-term memory neural network can better capture semantic dependency relations among long sentences, and the accuracy and fluency of translation are improved. In the text classification task, long text sequences can be better processed by adopting the long-short-term memory neural network, and the classification accuracy and robustness are improved. Long and short term memory neural networks are also often used to predict future values, such as stock prices, air temperatures, traffic flows, during time series data processing tasks.
Optionally, collecting the historical electricity consumption and performing power demand analysis calculation on the historical electricity consumption by using a data prediction model, and specifically includes the following steps:
Step S31: performing data visualization processing according to the historical electricity consumption, so as to obtain a historical electricity consumption curve trend chart;
step S32: performing curvature analysis and calculation according to the historical electricity utilization curve trend graph, so as to obtain the historical electricity utilization curvature;
step S33: and carrying out power demand analysis calculation on the historical power utilization curvature by using the data prediction model so as to generate a power demand value.
The invention performs data visualization processing according to the historical electricity consumption so as to better understand and analyze the historical data. And (5) carrying out next analysis on the obtained electricity consumption curve trend graph. In addition, by observing the historical trend graph, the conventional electricity utilization modes such as periodic factors including seasonal and sunday changes, working time and holidays can be identified, and then the implicit requirements related to electricity consumption can be found; performing curvature analysis and calculation according to the historical electricity utilization curve trend graph to obtain fine electricity utilization data characteristics, wherein the fine electricity utilization data characteristics comprise Gao Fenggu values, fluctuation strength, growth speed and historical electricity utilization curvature, so that the trend and change of historical electricity utilization are more comprehensively known; by utilizing the power demand analysis and calculation of the data prediction model, the future power demand can be rapidly and accurately calculated according to the historical power consumption curve trend and curvature information, so as to guide the power company to formulate differentiated power allocation and production plans. In addition, the data prediction model can further learn historical data, and continuously predict and update upcoming power demand changes on line according to a new historical data record.
Further, the data prediction model in the above description is a neural network model based on back propagation, the learning process of the data prediction model adopts the combination of an adaptive learning rate additional momentum method and a back propagation network, and the adaptive adjustment of the learning parameters is performed twice in each learning process, wherein the adaptive adjustment step specifically comprises:
step S301: the method comprises the steps of carrying out iteration for a threshold number of times in a learning process, and utilizing the total error of the current iteration to carry out comparison operation with the total error of the last time to adjust the learning rate and momentum factor so as to generate a first learning parameter adjustment item;
step S302: the method comprises the steps of carrying out iteration twice the threshold number in the learning process, comparing the average value of the maximum error variation in the last two learning processes, adjusting the average value and the accuracy, and generating a mean square value of a convergence limit, thereby obtaining a second learning parameter adjustment item;
step S303: and carrying out learning parameter correction processing according to the first learning parameter adjustment item and the second learning parameter adjustment item, thereby realizing self-adaptive adjustment of the learning process.
The method of the invention which combines the self-adaptive learning rate additional momentum method and the elastic circulation network is described above. The basic idea is to perform adaptive adjustment of the secondary learning parameters in each learning process: the first time is the self-adaptive adjustment process of the learning rate n and the momentum factor, the method is that after the t-th generation is completed, the total error E (t) of the iteration is compared with the total error E (t-1) of the last time, the learning rate n and the momentum factor are adjusted and the positive and negative relations of the difference between E (t) and E (t-1) are utilized, and finally the connection weight and the value are adjusted: the second time is to adaptively adjust the value of the allowable mean square error e by solving the maximum error SMax (N) of the study after the Nth study (refer to all sample study passes) is completed, then calculating the average value of the maximum error variation in the last 2 study processes, comparing the average value with a convergence limit value e determined according to the precision requirement, and adjusting the allowable mean square error e according to the comparison result.
According to the invention, by comparing the total errors of the two iterations and adaptively adjusting the learning rate and the momentum factor, the learning algorithm can be more stable and effective, so that the learning algorithm converges more quickly. Adjusting the momentum factor can reduce oscillations, i.e., avoid algorithm hunting back and forth during optimization. Meanwhile, the learning rate is adjusted to control the updating of parameters in each step, so that the influence on convergence efficiency caused by too small or too large step length is avoided; the method calculates and generates the mean square value of the convergence limit, can finely adjust the learning algorithm, and further improves the algorithm precision and generalization capability. The average value and the precision are adjusted, the performance of the model on the training set can be directly controlled, and convergence conditions are more reasonably set, so that bad states such as overfitting and the like are prevented; according to the invention, the learning parameter correction processing is carried out by combining the first and second learning parameter adjustment items, so that the self-adaptive adjustment of the learning process is realized, and the algorithm can be ensured to generate more accurate learning parameters after correction is carried out according to the latest overall error and mean square difference value, thereby better adapting to new data and enabling the whole process to have global optimization performance.
Optionally, load balancing analysis is performed according to the power demand value and the photovoltaic energy consumption data set, and specifically comprises the following steps:
Step S41: carrying out equalization processing on the power demand value by utilizing power equalization calculation so as to obtain a power equalization score;
step S42: performing weighted average calculation according to the power balance scores as weights and the photovoltaic energy consumption data set, so as to generate a dynamic power demand index;
step S43: and carrying out load balancing analysis on the dynamic power demand index so as to generate a dynamic resource allocation adjustment scheme.
The invention uses the power balance calculation to balance the power demand value, thereby obtaining the power balance score, ensuring the energy consumption efficiency of the system and improving the service life of the equipment. Through power balance, the energy consumption of different areas in the power grid is uniformly distributed, so that the problems of partial area power failure, overload equipment burning and the like caused by excessive energy consumption are avoided; the power balance score is used as a weight to carry out weighted average calculation with the photovoltaic energy consumption data set, so that a dynamic power demand index is generated, the power demand data can more accurately reflect the actual power consumption situation, a better resource allocation strategy is formulated, and the energy utilization rate is effectively improved; and carrying out load balancing analysis on the dynamic power demand indexes so as to generate a dynamic resource allocation adjustment scheme, and optimizing and distributing different power supplies according to the dynamic demand indexes of different time periods by running a distributed load balancing algorithm to realize optimal power dispatching and allocation. The method can avoid waste in resource use, shorten operation and maintenance time and improve response time and performance of the application program. In addition, the intelligent power grid device is automatically adapted to different environments and requirements, so that the optimal balance is made between reducing resource waste and improving energy utilization efficiency, and meanwhile, the intelligent power grid device can also realize mutual cooperation of a plurality of intelligent power grid devices so as to realize stronger resource allocation performance.
The invention provides a method for distributing artificial intelligence of a smart grid, which can utilize a machine learning algorithm to calculate and analyze data by collecting and preprocessing a data set, thereby generating a dynamic resource distribution adjustment scheme and improving the resource utilization rate and management efficiency of the smart grid system; the method refines the specific steps of data preprocessing, including data quality inspection processing, feature processing calculation and normalization standardized calculation, and enhances the accuracy and usability of the data set; the method describes a specific mode for perfecting the equipment data, and comprises a plurality of characteristic classification processes of the equipment, so that a conversion characteristic vector is generated, and the quality of a photovoltaic system data set is improved; the method defines a normalized standard calculation formula, provides an actual operation method for data processing, and is favorable for quickly and accurately generating a photovoltaic equipment data set; the method describes the specific steps of denoising processing, including image visualization, local smoothing algorithm and digital dimension reduction calculation, so that the quality and usability of a data set of the photovoltaic image equipment are effectively improved; the method provides a specific implementation scheme of a machine learning algorithm, and comprises a convolutional neural network, a gate-based cyclic neural network and a deep learning neural network, so that the accuracy and the reliability of a photovoltaic energy consumption data set are improved; the method specifically describes the structure and the implementation principle of the gate-based cyclic neural network, thereby providing an actual operation scheme for the feature extraction of the photovoltaic energy consumption dataset; the method comprises the steps of providing power demand analysis specific steps based on historical power consumption, including data visualization, curvature analysis and prediction model, so as to provide reliable basic data and prediction results for load balancing analysis of the intelligent power grid system; the method provides a specific learning process of the neural network model based on back propagation, and comprises a mode of combining an adaptive learning rate additional momentum method and a back propagation network, so that an actual optimization scheme is provided for a prediction model of power demand; according to the method, by calculating the power balance score and carrying out load balance analysis, balance processing and dynamic resource distribution of power demands can be realized, the load balance of a power grid is improved, the power consumption cost is reduced, and the stability of the power grid is improved; the method can be applied to power grid management and optimization in the field of energy, and improves the energy management efficiency and the sustainability of the power grid.
Drawings
Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of a non-limiting implementation, made with reference to the accompanying drawings in which:
FIG. 1 is a flow chart of steps of a method for distributing artificial intelligent resources of a smart grid according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating steps for collecting device data and preprocessing the data according to an embodiment of the present invention;
FIG. 3 is a flowchart illustrating steps for photovoltaic energy consumption calculation according to an embodiment of the present invention;
FIG. 4 is a flowchart illustrating steps of power demand analysis and calculation according to an embodiment of the present invention;
FIG. 5 is a flow chart of steps for providing adaptive tuning steps for a combination of an adaptive learning rate added momentum method and a counter-propagating network in accordance with an embodiment of the present invention.
FIG. 6 is a flowchart of steps for providing load balancing analysis from a power demand value and a photovoltaic energy consumption dataset according to an embodiment of the present invention.
Detailed Description
The following is a clear and complete description of the technical method of the present patent in conjunction with the accompanying drawings, and it is evident that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.
Furthermore, the drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. The functional entities may be implemented in software or in one or more hardware modules or integrated circuits or in different networks and/or processor methods and/or microcontroller methods.
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
The embodiment of the invention provides an artificial intelligent resource allocation method for an intelligent power grid. The execution subject of the intelligent power grid artificial intelligent resource allocation method comprises, but is not limited to, at least one of a server and an electronic device of which the terminal can be configured to execute the method provided by the embodiment of the invention. In other words, the artificial intelligence resource allocation method may be performed by software or hardware installed in a terminal device or a server device, and the software may be a blockchain platform. The service end includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like. The server can be an independent server, and can also be a cloud server for providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content distribution networks, basic cloud computing services such as big data and artificial intelligent platforms, and the like.
In an embodiment of the present invention, referring to fig. 1, the method for distributing artificial intelligent resources of a smart grid includes the following steps:
step S1: collecting the data of the light energy power generation area equipment in charge of the intelligent power grid system, and preprocessing the data to generate a photovoltaic system data set;
step S2: performing photovoltaic energy consumption calculation on the photovoltaic system data set by using a machine learning algorithm, so as to generate a photovoltaic energy consumption data set;
step S3: collecting historical electricity consumption, and performing power demand analysis and calculation on the historical electricity consumption by utilizing a data prediction model so as to generate a power demand value;
step S4: and carrying out load balancing analysis according to the power demand value and the photovoltaic energy consumption data set, thereby generating a dynamic resource allocation adjustment scheme.
According to the method, the collected data of the light energy power generation area equipment are subjected to data preprocessing to generate the data set of the photovoltaic system, so that the influence caused by the problems of noise, missing, outlier and the like of the collected data of the original light energy power generation area equipment is reduced, and the accuracy of the data is improved; according to the invention, the machine learning algorithm is adopted to calculate the photovoltaic energy consumption of the photovoltaic system data so as to generate the photovoltaic energy consumption data set, so that the current running state of the photovoltaic equipment and the performance of the photovoltaic equipment in future time can be accurately predicted. In addition, the machine learning algorithm is applied to the field of photovoltaic power generation, and can help engineers to quickly detect faults in photovoltaic equipment, including damage, defects and design problems, and can also search for links among data in a large scale through data analysis, so that self-adaptive control optimization is realized, and the service life of a photovoltaic module is prolonged; according to the invention, the historical electricity consumption of the mobile phone is analyzed and calculated by utilizing the data prediction model, so that the future electricity demand can be predicted more accurately, a reasonable electricity distribution plan can be appointed, and meanwhile, the past electricity consumption trend, the electricity consumption mode and the electricity consumption gap between unused areas can be revealed, so that the load balance of the production/consumption modes can be realized. In addition, the machine learning model can further analyze the historical power consumption, determine energy conservation and reduce power consumption cost, and implement measures such as reactive compensation and load management through a specific time period or a specific user group; according to the invention, load balancing analysis is carried out according to the power demand value and the photovoltaic energy consumption data set, so that the intelligent power grid system can more efficiently distribute power production, thereby optimizing power distribution, avoiding resource waste and reducing energy consumption. In addition, the system can be more stable, and the efficiency of photovoltaic power generation can be improved.
In the embodiment of the present invention, referring to fig. 2, the data preprocessing in step S1 includes data quality inspection processing, feature processing calculation and normalization calculation, and specifically includes the following steps:
step S11: performing data quality inspection processing on the equipment data of the light energy power generation area so as to generate perfect equipment data;
in detail, the collected light energy power generation area equipment data comprises the machine body temperature of equipment, the effective irradiation area of a solar panel, the equipment angle, the external environment temperature, the external environment humidity and the illumination intensity. In addition, the method can also comprise equipment names, models, specifications, manufacturers, production dates, installation dates, position information and the like; the collected equipment data are cleaned, the problems of repetition, deletion, errors, inconsistency and the like are removed, and the accuracy and the integrity of the data are ensured; and storing the processed equipment data into a database or a data warehouse for use by a subsequent light energy power generation management system.
Step S12: performing feature processing calculation on the perfect equipment data so as to generate a conversion feature vector;
in detail, the method for generating the conversion feature vector by carrying out feature processing calculation on the perfect equipment data comprises the following specific steps:
Performing first characteristic classification processing on the external environment temperature and the external environment humidity so as to obtain a temperature and humidity vector;
performing second feature classification processing on the illumination intensity so as to obtain an illumination vector;
performing third characteristic classification processing on the body temperature of the equipment, the effective irradiation area of the solar panel and the equipment angle, thereby obtaining an equipment vector;
and collecting the temperature and humidity vector, the illuminance vector and the equipment vector into a conversion characteristic vector.
According to the invention, different characteristics of different data are classified, so that redundant data can be avoided, the complexity of data processing is reduced, and the efficiency of data processing is improved; and a plurality of parameters are collected into the same feature vector, so that data analysis can be conveniently performed. According to the similarity and the difference between the feature vectors, the relation between the data can be more clearly grasped, and more accurate data analysis is realized; the various parameters are integrated to be used as the feature vector, so that the influence of the factors of the environment and the equipment on the result can be more comprehensively considered, and the prediction accuracy is improved. In addition, by integrating various parameters into feature vectors, the method can be used as a basis for intelligent control, and by detecting and learning the feature vectors in real time, the self-adaptive adjustment and intelligent control of data can be realized.
In the embodiment of the invention, the data of the external environment temperature and humidity are generally obtained through a sensor. The ambient temperature and humidity can be classified into a plurality of levels, for example, low, medium and high levels, or can be classified according to actual demands. For each level, a range is set, classifying the temperature and humidity data. For example, when the ambient temperature is 20 degrees celsius or less, it may be classified as a low temperature level; when the humidity is 30% or less, it can be classified as a low humidity level. Finally, a temperature and humidity vector subjected to first feature classification processing is obtained; the illumination intensity is also typically measured by a sensor. According to practical situations, the illumination intensity is divided into a plurality of levels, for example, the illumination intensity is divided into three levels of low, medium and high. For each level, a range is set, and the illumination intensity data is classified. For example, when the light intensity is less than 1000lux, it may be classified as a low light intensity level; when the illumination intensity is equal to or greater than 5000lux, it can be classified as a high illumination intensity level. Finally, obtaining an illuminance vector subjected to second feature classification processing; the body temperature of the equipment, the effective irradiation area of the solar panel and the equipment angle can be obtained through sensors or through measurement and calculation. The body temperature may be divided into a plurality of levels according to a certain range, for example, into three levels of low, medium and high. The effective irradiation area of the solar panel may be preset at the beginning of design. The device angle can generally be measured. The device angle may also be divided into a plurality of levels according to a certain range. And taking the body temperature, the effective irradiation area of the solar panel and the equipment angle as third characteristics in the characteristic vector, and classifying the third characteristics to obtain an equipment vector subjected to the third characteristic classification. The first feature classification process, the second feature classification process and the third feature classification process may use a method of cluster analysis, decision tree and artificial intelligent neural network model construction to perform cluster analysis and prediction.
Step S13: performing normalization standardized calculation on the conversion characteristic vector so as to generate a photovoltaic equipment data set;
step S14: and denoising the photovoltaic equipment data set to generate a photovoltaic system data set.
The invention performs data quality inspection and processing on the data of the photovoltaic power generation equipment, improves the operation efficiency of the photovoltaic power generation system, improves the equipment data, can help related staff to find various data indexes of the photovoltaic power generation system in time, and avoids potential safety hazards of the photovoltaic power generation system caused by data errors, defects and other problems, thereby ensuring the safety of the staff and the equipment. In addition, the perfect equipment data quality inspection optimizes the maintenance cost of the power generation system and improves the maintenance efficiency of the power generation system; the method has the advantages that the characteristic processing calculation is carried out on the perfect equipment data, the perfect equipment data are converted into characteristic vectors, some non-critical information and noise in the photovoltaic equipment are removed, and important characteristic information is reserved, so that the quality of the data is improved, the complexity of subsequent model calculation is reduced, and the training and prediction efficiency is improved; the conversion feature vectors are normalized, so that comparison among different features is fairer, the influence of certain features on the final training effect due to the fact that the numerical value is larger is avoided, the dimension influence is eliminated, all the features are unified into the same proportion range through normalization, further processing is facilitated, and high-quality photovoltaic equipment data sets are generated. In addition, the optimization speed of algorithms such as gradient descent is also increased.
Further, in step S13, a normalization formula is adopted to calculate and obtain a photovoltaic device data set, where the normalization formula specifically includes:
wherein the method comprises the steps of
，/>
；
The invention firstly obtains the values of the temperature and humidity vector, the illuminance vector and the equipment vector and the characteristic values of the temperature and humidity vector, the illuminance vector and the equipment vector, and for each vector, the corresponding characteristic value needs to be found, and the temperature and humidity vector, the illuminance vector and the equipment vector are obtained usually by using linear algebra, so that the characteristics of the temperature and humidity vector, the illuminance vector and the equipment vector meet the average value and the unit varianceBy the characteristic values of the temperature and humidity vector, the illuminance vector and the equipment vector
The average value of the characteristic values is obtained by accumulation and summation, and the difference value is obtained by each vector to obtain the characteristic value +. >
，/>
Is the characteristic mean value of temperature and humidity vectors, and is->
Is the characteristic mean value of the illuminance vector, +.>
Is the characteristic mean value of the device vector; values using temperature and humidity vector, illuminance vector and device vector +.>
Subtracting the eigenvalue of each vector +.>
The square of the difference obtained>
Obtaining the characteristic variance of the temperature and humidity vector, the illuminance vector and the equipment vector>
，/>
Is the characteristic variance of the temperature and humidity vector, < >>
Characteristic variance of illuminance vector, ++>
The characteristic variance of the device vector; summing them to average +.>
The error of a certain vector is influenced by accidental limit data, and a normalized result photovoltaic equipment data set of a temperature and humidity vector, an illuminance vector and an equipment vector is further obtained in a summation average mode.
In detail, denoising the photovoltaic device data set to generate a photovoltaic system data set, including the steps of:
and performing image visualization processing according to the photovoltaic equipment data set, so as to obtain a photovoltaic equipment data image.
Further, converting the photovoltaic device dataset into an image dataset, the time series of photovoltaic device datasets may be regarded as pixel gray values of the image, visualized as a gray image. Each pixel in the image represents a data point, and its gray value represents the size of the data point.
The method can draw data images by using Matplotlib and other libraries of Python, and adjust the data images to obtain clearer image effects.
And carrying out local weighted regression calculation on the photovoltaic equipment data image by using a local smoothing algorithm, thereby realizing denoising treatment and obtaining a photovoltaic image data set.
Further, the image after the visualization processing of the photovoltaic device data set is subjected to denoising processing, and a local smoothing algorithm can be adopted. The local smoothing algorithm is a non-parametric regression method, and performs a weighted average on data points near a target point in a local range to perform smoothing on the target point. The local smoothing algorithm has the disadvantage of large calculation amount, but can greatly improve the denoising effect.
The method can specifically use Scikit-learn of Python and other libraries to realize the calculation of the algorithm, and adjust the algorithm to obtain a better denoising effect.
And performing digital dimension reduction calculation by using the photovoltaic image dataset to obtain a photovoltaic system dataset.
Further, the photovoltaic image dataset may be digitally dimensionalized using a Principal Component Analysis (PCA) algorithm. The PCA algorithm is a common unsupervised learning algorithm that maps high-dimensional data into a low-dimensional space through linear transformation, preserving the maximum variance of the original data.
The method can specifically use Scikit-learn of Python and other libraries to realize the calculation of the algorithm, and adjust the algorithm to obtain a better dimension reduction effect. After the digital dimension reduction is completed, a photovoltaic system data set can be obtained for further analysis and application.
The invention adopts the technologies of computer vision, image processing, machine learning and the like to realize denoising treatment on the photovoltaic equipment data set, and utilizes the visualization treatment on the photovoltaic equipment data set to obtain a form which is easier to understand and analyze, and provides the focused characteristics and visualization to help staff to observe data better; the partial smoothing algorithm is utilized to remove noise and abnormal values of the photovoltaic equipment data image, so that the influence of factors in the real environment is reduced to a certain extent, and the efficiency of subsequent analysis is improved; the digital dimension reduction calculation is carried out based on the image output, so that the complexity of a model can be reduced, the training and prediction efficiency of the model can be improved, and meanwhile, important features in data can be derived by adopting a machine learning algorithm, so that the linear correlation is eliminated, the dimension of input data is reduced, and the advantage of processing a large amount of data is increased. And finally, the generated photovoltaic system data set is used for learning and calculating a subsequent model, and can provide more comprehensive and accurate information for researchers and engineers, so that the system design is optimized and the energy operation is improved, and the efficiency of artificial intelligent resource allocation of the intelligent power grid system is improved.
In the embodiment of the present invention, referring to fig. 3, the machine learning algorithm mentioned in step S2 includes a convolutional neural network, a gate-based recurrent neural network, and a deep learning network, specifically:
step S21: performing feature extraction on a photovoltaic system data set by using a convolutional neural network so as to generate a first-order photovoltaic symptom set;
further, a convolutional neural network model needs to be designed before feature extraction. Convolutional neural networks typically comprise multiple convolutional layers, pooling layers, and fully-connected layers. Wherein the convolution and pooling layers apply convolution and pooling operations to identify features in the image, and the fully connected layer is used to classify or regress the features. The key to design the convolutional neural network model is to select the appropriate number of layers, the number of neurons in the layer, the size of the convolutional kernel, the activation function and other super parameters.
The method can specifically design and construct a convolutional neural network model by using Keras and other libraries of Python.
It needs to be trained. During training, optimization algorithms such as random gradient descent can be used to minimize the loss function, thereby improving the prediction accuracy of the model. During training, the model parameters are continuously updated using the data of the training set.
The method can specifically train and verify the convolutional neural network model by using Keras and other libraries of Python, and adjust the super parameters to optimize the performance of the model.
The invention adopts simple causal convolution to only review the network width of linear scale, and uses causal convolution of multi-scale information fusion to fuse the acceptance domains with exponential size.
Step S22: performing data dimension reduction processing on the first-order photovoltaic collection by using a gate-based cyclic neural network, so as to generate a second-order photovoltaic collection;
optionally, the gate-control-based circulating neural network is specifically a long-short-term memory neural network, and the forgetting gate and the input gate of the long-short-term memory neural network control the state update and the input information of neurons, so that the network can effectively process long-sequence dependency, and the problems of gradient disappearance and gradient explosion are avoided. Meanwhile, the output gate of the long-short-term memory neural network controls the output information flow, so that the neurons only output needed information, and redundant information and overfitting risks are reduced.
In addition, long-term memory neural networks have been widely used in speech recognition, machine translation, text classification, emotion analysis, and other tasks in natural language processing tasks. For example, in a machine translation task, the long-term memory neural network can better capture semantic dependency relations among long sentences, and the accuracy and fluency of translation are improved. In the text classification task, long text sequences can be better processed by adopting the long-short-term memory neural network, and the classification accuracy and robustness are improved. Long and short term memory neural networks are also often used to predict future values, such as stock prices, air temperatures, traffic flows, during time series data processing tasks.
Further, each long-short-period memory neural network comprises three gating structures including a forgetting gate, an input gate and an output gate so as to control the place where data has information. The forget gate is responsible for discarding and retaining the effective information at the last moment, the input gate stores the effective information at the current moment, and the output gate determines the information output by the neuron. Wherein the candidate state vector is a candidate vector for the state at the current time. It is obtained by fusing the state vector of the previous moment and the input vector of the current moment. The output vector of the output gate determines the extent to which the state vector at the current time has an effect on the output at the current time. The state vector of the cyclic unit is determined by the state vector of the last time, the candidate state vector and the update gate. The state vector is controlled by the gating unit and the hyperbolic tangent function, so that the long sequence dependency relationship can be effectively captured. The output layer converts the final state vector into the desired output form. The output layer may be a simple fully connected neural network or a Softmax classifier.
The method can be specifically designed and constructed using Python.
Step S23: and performing deep fitting on the second-order photovoltaic symptom set by using a deep learning neural network, so as to generate a photovoltaic energy consumption data set.
The method can be specifically established by using a TensorFlow, keras or PyTorch framework.
According to the method, the characteristics of the photovoltaic system data set are extracted through the convolutional neural network, the convolutional layer in the convolutional neural network performs local connection on input data, namely only a small input block is subjected to convolution operation to generate output, so that the problem of overfitting caused by full connection is avoided, the spatial structure information of the input data can be reserved, the data dimension and noise can be reduced, the stability and reliability of characteristic representation are improved, and a first-order photovoltaic collection is generated; the first-order photovoltaic collection is subjected to data dimension reduction processing based on the gating circulating neural network, and when the data dimension reduction is performed, information with little influence on a result can be screened out by using the gating mechanism, so that the dimension of the data is reduced, the efficiency and the accuracy of a model are improved, the calculation efficiency is effectively improved, the calculation load is reduced, and the prediction accuracy is improved, so that the second-order photovoltaic collection is generated; the second-order photovoltaic symptom sets are subjected to deep fitting through the deep learning neural network, and the current-voltage curve data and the power-voltage curve data of the real photovoltaic cells can be better approximated by utilizing the strong nonlinear modeling capability of the deep learning neural network, so that higher prediction precision is obtained, the precision and the robustness of the model are further improved, and a photovoltaic energy consumption data set is generated; the convolutional neural network, the gate-based cyclic neural network and the deep learning network respectively provide accurate and robust results for own tasks, and then the photovoltaic energy consumption correlation is more comprehensively expressed by combination. By selecting features, noise and redundancy are reduced, which can help remove unnecessary information and highlight the most important factors, thus making the overall process more efficient and reliable. In addition, the invention improves modeling accuracy and robustness, and can also support real-time data analysis and decision.
In the embodiment of the present invention, referring to fig. 4, step S3 collects historical electricity consumption and uses a data prediction model to perform power demand analysis calculation on the historical electricity consumption, and specifically includes the following steps:
step S31: and carrying out data visualization processing according to the historical electricity consumption, thereby obtaining a historical electricity consumption curve trend graph.
Specifically, the data of the historical electricity consumption is obtained and sequenced according to the time sequence, so that the later data analysis and visualization processing can be performed.
Further, the method can draw a curve trend graph of historical electricity consumption by using data visualization tools such as matplotlib and seacap libraries in Python. The data is segmented to observe the trend of the power consumption in more detail. And analyzing the drawn historical electricity consumption curve trend graph, and observing the conditions of the form, fluctuation degree, periodicity and the like of the electricity consumption curve so as to better understand the change rule of the electricity consumption.
Step S32: and (5) carrying out curvature analysis and calculation according to the historical electricity utilization curve trend graph, so as to obtain the historical electricity utilization curvature.
Specifically, according to the historical electricity consumption curve trend graph realized by data visualization, a fitting algorithm can be adopted to fit a curve. Algorithms such as polynomial regression or spline interpolation are generally adopted so as to better reflect the fluctuation condition of the electricity consumption; using a derivative calculation formula to derive a fitted historical electricity utilization curve, and obtaining the curvature of the historical electricity utilization curve; the historical electricity consumption curve can be calculated respectively by selecting a proper partial differential formula, and the value of the curvature is drawn into a chart so as to better observe the change trend of the electricity consumption.
Step S33: and carrying out power demand analysis calculation on the historical power utilization curvature by using the data prediction model so as to generate a power demand value.
The invention performs data visualization processing according to the historical electricity consumption so as to better understand and analyze the historical data. And (5) carrying out next analysis on the obtained electricity consumption curve trend graph. In addition, by observing the historical trend graph, the conventional electricity utilization modes such as periodic factors including seasonal and sunday changes, working time and holidays can be identified, and then the implicit requirements related to electricity consumption can be found; performing curvature analysis and calculation according to the historical electricity utilization curve trend graph to obtain fine electricity utilization data characteristics, wherein the fine electricity utilization data characteristics comprise Gao Fenggu values, fluctuation strength, growth speed and historical electricity utilization curvature, so that the trend and change of historical electricity utilization are more comprehensively known; by utilizing the power demand analysis and calculation of the data prediction model, the future power demand can be rapidly and accurately calculated according to the historical power consumption curve trend and curvature information, so as to guide the power company to formulate differentiated power allocation and production plans. In addition, the data prediction model can further learn historical data, and continuously predict and update upcoming power demand changes on line according to a new historical data record.
Further, the data prediction model in the above description is a neural network model based on back propagation, and the learning process adopts an adaptive learning rate additional momentum method and a back propagation network to be combined.
In the embodiment of the present invention, referring to fig. 5, two adaptive adjustment steps of learning parameters are performed in each learning process, where the adaptive adjustment steps specifically include:
step S301: the method comprises the steps of carrying out iteration for a threshold number of times in a learning process, and utilizing the total error of the current iteration to carry out comparison operation with the total error of the last time to adjust the learning rate and momentum factor so as to generate a first learning parameter adjustment item;
step S302: the method comprises the steps of carrying out iteration twice the threshold number in the learning process, comparing the average value of the maximum error variation in the last two learning processes, adjusting the average value and the accuracy, and generating a mean square value of a convergence limit, thereby obtaining a second learning parameter adjustment item;
step S303: and carrying out learning parameter correction processing according to the first learning parameter adjustment item and the second learning parameter adjustment item, thereby realizing self-adaptive adjustment of the learning process.
The above-described method of the present invention employs a combination of adaptive learning rate-additional momentum methods and elastic loop networks. The basic idea is to perform adaptive adjustment of the secondary learning parameters in each learning process: the first time is the self-adaptive adjustment process of the learning rate n and the momentum factor, the method is that after the t-th generation is completed, the total error E (t) of the iteration is compared with the total error E (t-1) of the last time, the learning rate n and the momentum factor are adjusted and the positive and negative relations of the difference between E (t) and E (t-1) are utilized, and finally the connection weight and the value are adjusted: the second time is to adaptively adjust the value of the allowable mean square error e by solving the maximum error SMax (N) of the study after the Nth study (refer to all sample study passes) is completed, then calculating the average value of the maximum error variation in the last 2 study processes, comparing the average value with a convergence limit value e determined according to the precision requirement, and adjusting the allowable mean square error e according to the comparison result.
According to the invention, by comparing the total errors of the two iterations and adaptively adjusting the learning rate and the momentum factor, the learning algorithm can be more stable and effective, so that the learning algorithm converges more quickly. Adjusting the momentum factor can reduce oscillations, i.e., avoid algorithm hunting back and forth during optimization. Meanwhile, the learning rate is adjusted to control the updating of parameters in each step, so that the influence on convergence efficiency caused by too small or too large step length is avoided; the method calculates and generates the mean square value of the convergence limit, can finely adjust the learning algorithm, and further improves the algorithm precision and generalization capability. The average value and the precision are adjusted, the performance of the model on the training set can be directly controlled, and convergence conditions are more reasonably set, so that bad states such as overfitting and the like are prevented; according to the invention, the learning parameter correction processing is carried out by combining the first and second learning parameter adjustment items, so that the self-adaptive adjustment of the learning process is realized, and the algorithm can be ensured to generate more accurate learning parameters after correction is carried out according to the latest overall error and mean square difference value, thereby better adapting to new data and enabling the whole process to have global optimization performance.
Specifically, at the beginning of training the neural network, it is first necessary to initialize the values of the learning rate and the momentum factor. Typically, smaller values may be used to initialize the learning rate and momentum factor. In order to ensure the efficiency and effect of the learning process, a threshold value needs to be set, and training is stopped when the training error changes to a certain extent. And iteratively updating the learning rate and the momentum factor, comparing the current total error with the previous total error in each iteration process, adjusting the values of the learning rate and the momentum factor according to the comparison result, and generating a first learning parameter adjustment item. The learning rate and the momentum factor can be generally adjusted in a similar manner to descent, namely, when the error descends, the values of the learning rate and the momentum factor are gradually reduced, and the model training is quickened; when the error change is not large, gradually increasing the values of the learning rate and the momentum factor, and ensuring the stability of the learning process;
In each iteration process, the maximum error variation between the current iteration and the last iteration is recorded, and then the average value of the maximum error variation between two adjacent iterations is calculated. This average value can be used to determine if the model has tended to stabilize. In order to avoid the influence of noise and fluctuation of the error variation on the stability of the model, the sizes of the error variation and the fluctuation need to be adjusted according to the current precision and average value so as to better reflect the stability of the model. And generating a mean square value of the convergence limit, and generating the mean square value of the convergence limit according to the adjusted average value and the precision. This mean square value can be used as a threshold for model convergence, and training is stopped after the model error reaches this threshold.
Further, the first and second learning parameter adjustment items are combined to obtain a combined learning parameter adjustment item. The comprehensive adjustment item can be better suitable for different training data and training processes, and can better promote convergence and optimization of the model. And applying the comprehensive learning parameter adjustment items to the learning process of the neural network, and correcting the learning parameters according to the adjustment items. Therefore, the learning process of the neural network can be better guided, and the training effect and the training precision of the model are improved.
In the embodiment of the present invention, referring to fig. 6, step S4 performs load balancing analysis according to the power demand value and the photovoltaic energy consumption data set, and specifically includes the following steps:
step S41: carrying out equalization processing on the power demand value by utilizing power equalization calculation so as to obtain a power equalization score;
specifically, according to the collected data, a predicted value of the power demand in each time period can be calculated and compared with an actual demand value to obtain a power balance score. The power balance score may be quantified using various metrics such as sum of squares error, absolute average error, root mean square error, etc.
Step S42: and carrying out weighted average calculation according to the power balance scores as weights and the photovoltaic energy consumption data set, so as to generate a dynamic power demand index.
Specifically, the photovoltaic energy consumption data sets are ordered according to a time sequence, the total photovoltaic energy consumption in each time period is calculated, for example, once every hour, and in order to obtain accurate photovoltaic energy consumption data, a plurality of data points are considered to be randomly selected in each time period and averaged. And using the power balance score as a weight, carrying out weighted average calculation on the photovoltaic energy consumption data in each time period, wherein the weight can be automatically adjusted to the dynamic power demand index in each time period according to the demand.
Step S43: and carrying out load balancing analysis on the dynamic power demand index so as to generate a dynamic resource allocation adjustment scheme.
Specifically, analyzing the load condition of the power grid and the energy consumption condition of each electric equipment includes predicting the future electric load condition. And according to the dynamic power demand index and the power load condition, a reasonable load balancing algorithm is used, so that a dynamic resource allocation adjustment scheme is generated, and the dynamic resource allocation adjustment scheme comprises adjustment of photovoltaic power generation output power, start-stop control of user electric equipment and the like. The adjustment scheme can be optimized by adopting a heuristic algorithm, a machine learning algorithm and the like.
The invention uses the power balance calculation to balance the power demand value, thereby obtaining the power balance score, ensuring the energy consumption efficiency of the system and improving the service life of the equipment. Through power balance, the energy consumption of different areas in the power grid is uniformly distributed, so that the problems of partial area power failure, overload equipment burning and the like caused by excessive energy consumption are avoided; the power balance score is used as a weight to carry out weighted average calculation with the photovoltaic energy consumption data set, so that a dynamic power demand index is generated, the power demand data can more accurately reflect the actual power consumption situation, a better resource allocation strategy is formulated, and the energy utilization rate is effectively improved; and carrying out load balancing analysis on the dynamic power demand indexes so as to generate a dynamic resource allocation adjustment scheme, and optimizing and distributing different power supplies according to the dynamic demand indexes of different time periods by running a distributed load balancing algorithm to realize optimal power dispatching and allocation. The method can avoid waste in resource use, shorten operation and maintenance time and improve response time and performance of the application program. In addition, the intelligent power grid device is automatically adapted to different environments and requirements, so that the optimal balance is made between reducing resource waste and improving energy utilization efficiency, and meanwhile, the intelligent power grid device can also realize mutual cooperation of a plurality of intelligent power grid devices so as to realize stronger resource allocation performance.
The invention provides a method for distributing artificial intelligence of a smart grid, which can utilize a machine learning algorithm to calculate and analyze data by collecting and preprocessing a data set, thereby generating a dynamic resource distribution adjustment scheme and improving the resource utilization rate and management efficiency of the smart grid system; the method refines the specific steps of data preprocessing, including data quality inspection processing, feature processing calculation and normalization standardized calculation, and enhances the accuracy and usability of the data set; the method describes a specific mode for perfecting the equipment data, and comprises a plurality of characteristic classification processes of the equipment, so that a conversion characteristic vector is generated, and the quality of a photovoltaic system data set is improved; the method defines a normalized standard calculation formula, provides an actual operation method for data processing, and is favorable for quickly and accurately generating a photovoltaic equipment data set; the method describes the specific steps of denoising processing, including image visualization, local smoothing algorithm and digital dimension reduction calculation, so that the quality and usability of a data set of the photovoltaic image equipment are effectively improved; the method provides a specific implementation scheme of a machine learning algorithm, and comprises a convolutional neural network, a gate-based cyclic neural network and a deep learning neural network, so that the accuracy and the reliability of a photovoltaic energy consumption data set are improved; the method specifically describes the structure and the implementation principle of the gate-based cyclic neural network, thereby providing an actual operation scheme for the feature extraction of the photovoltaic energy consumption dataset; the method comprises the steps of providing power demand analysis specific steps based on historical power consumption, including data visualization, curvature analysis and prediction model, so as to provide reliable basic data and prediction results for load balancing analysis of the intelligent power grid system; the method provides a specific learning process of the neural network model based on back propagation, and comprises a mode of combining an adaptive learning rate additional momentum method and a back propagation network, so that an actual optimization scheme is provided for a prediction model of power demand; according to the method, by calculating the power balance score and carrying out load balance analysis, balance processing and dynamic resource distribution of power demands can be realized, the load balance of a power grid is improved, the power consumption cost is reduced, and the stability of the power grid is improved; the method can be applied to power grid management and optimization in the field of energy, and improves the energy management efficiency and the sustainability of the power grid.
In the several embodiments provided by the present invention, it should be understood that the methods described may be implemented in other ways. The models mentioned above are merely illustrative, e.g. combinations of neural network models, are merely a logic, and other combinations may be implemented in practice.
In addition, each implementation step of the present invention may be implemented in one module, or may be each independent functional module. Some or all of which may be selected according to actual needs. It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof.
The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and any associated drawings which fall within the meaning of the terms are therefore intended to be interpreted as limiting the description.
The embodiment of the invention can acquire and process the related data based on the artificial intelligence technology. Wherein, artificial intelligence is a theory, method, technique and application system that uses a computer or a machine controlled by a computer to simulate, extend and expand the intelligence of a person, sense the environment, acquire knowledge and use the knowledge to obtain the best result.
Finally, it should be noted that the above-mentioned embodiments are merely for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made to the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention.
Claims (10)
1. The artificial intelligent resource allocation method for the intelligent power grid is characterized by comprising the following steps of:
step S1: collecting the data of the light energy power generation area equipment in charge of the intelligent power grid system, and preprocessing the data to generate a photovoltaic system data set;
step S2: performing photovoltaic energy consumption calculation on the photovoltaic system data set by using a machine learning algorithm, so as to generate a photovoltaic energy consumption data set;
step S3: collecting historical electricity consumption, and performing power demand analysis and calculation on the historical electricity consumption by utilizing a data prediction model so as to generate a power demand value;
step S4: and carrying out load balancing analysis according to the power demand value and the photovoltaic energy consumption data set, thereby generating a dynamic resource allocation adjustment scheme.
2. The method according to claim 1, wherein the data preprocessing includes data quality inspection processing, feature processing calculation, and normalization standard calculation, and the step S1 specifically includes the steps of:
Step S11: performing data quality inspection processing on the equipment data of the light energy power generation area so as to generate perfect equipment data;
step S12: performing feature processing calculation on the perfect equipment data so as to generate a conversion feature vector;
step S13: performing normalization standardized calculation on the conversion characteristic vector so as to generate a photovoltaic equipment data set;
step S14: and denoising the photovoltaic equipment data set to generate a photovoltaic system data set.
3. The method according to claim 2, wherein the perfecting the device data includes a body temperature, an external environment humidity, an illumination intensity, an effective illumination area of the solar panel, and a device angle of the device, and step S12 is specifically:
performing first characteristic classification processing on the external environment temperature and the external environment humidity so as to obtain a temperature and humidity vector;
performing second feature classification processing on the illumination intensity so as to obtain an illumination vector;
performing third characteristic classification processing on the body temperature of the equipment, the effective irradiation area of the solar panel and the equipment angle, thereby obtaining an equipment vector;
and collecting the temperature and humidity vector, the illuminance vector and the equipment vector into a conversion characteristic vector.
4. The method according to claim 2, wherein the normalization calculation in step S13 is specifically:
and calculating and obtaining a photovoltaic equipment data set by utilizing a normalization standardized formula according to the temperature and humidity vector, the illuminance vector and the equipment vector, wherein the normalization standardized formula specifically comprises:
wherein the method comprises the steps of
，/>
；
5. The method according to claim 2, wherein step S14 is specifically:
performing image visualization processing according to the photovoltaic equipment data set, so as to obtain a photovoltaic equipment data image;
carrying out local weighted regression calculation on the photovoltaic equipment data image by utilizing a local smoothing algorithm, thereby realizing denoising treatment and obtaining a photovoltaic image data set;
and performing digital dimension reduction calculation by using the photovoltaic image dataset to obtain a photovoltaic system dataset.
6. The method according to claim 1, wherein the machine learning algorithm in step S2 includes a convolutional neural network, a gate-based recurrent neural network, and a deep learning neural network, and step S2 is specifically:
step S21: performing feature extraction on a photovoltaic system data set by using a convolutional neural network so as to generate a first-order photovoltaic symptom set;
step S22: performing data dimension reduction processing on the first-order photovoltaic collection by using a gate-based cyclic neural network, so as to generate a second-order photovoltaic collection;
step S23: and performing deep fitting on the second-order photovoltaic symptom set by using a deep learning neural network, so as to generate a photovoltaic energy consumption data set.
7. The method according to claim 6, wherein the gating-based recurrent neural network is specifically:
the long-period memory neural network is adopted as a circulating neural network, wherein neurons in the circulating neural network comprise a forgetting gate control structure, an input gate control structure and an output gate control structure, the forgetting gate is responsible for discarding and retaining the effective information at the last moment, the input gate stores the effective information at the current moment, and the output gate determines the neurons to output the effective information.
8. The method according to claim 1, wherein step S3 is specifically:
Step S31: performing data visualization processing according to the historical electricity consumption, so as to obtain a historical electricity consumption curve trend chart;
step S32: performing curvature analysis and calculation according to the historical electricity utilization curve trend graph, so as to obtain the historical electricity utilization curvature;
step S33: and carrying out power demand analysis calculation on the historical power utilization curvature by using the data prediction model so as to generate a power demand value.
9. The method according to claim 1, wherein the data prediction model in step S3 is a neural network model based on back propagation, and the learning process uses an adaptive learning rate additional momentum method and a back propagation network in combination, and the adaptive adjustment of the learning parameters is performed twice in each learning process, and the adaptive adjustment step specifically includes:
step S301: the method comprises the steps of carrying out iteration for a threshold number of times in a learning process, and utilizing the total error of the current iteration to carry out comparison operation with the total error of the last time to adjust the learning rate and momentum factor so as to generate a first learning parameter adjustment item;
step S302: the method comprises the steps of carrying out iteration twice the threshold number in the learning process, comparing the average value of the maximum error variation in the last two learning processes, adjusting the average value and the accuracy, and generating a mean square value of a convergence limit, thereby obtaining a second learning parameter adjustment item;
Step S303: and carrying out learning parameter correction processing according to the first learning parameter adjustment item and the second learning parameter adjustment item, thereby realizing self-adaptive adjustment of the learning process.
10. The method according to claim 1, wherein step S4 is specifically:
step S41: carrying out equalization processing on the power demand value by utilizing power equalization calculation so as to obtain a power equalization score;
step S42: performing weighted average calculation according to the power balance scores as weights and the photovoltaic energy consumption data set, so as to generate a dynamic power demand index;
step S43: and carrying out load balancing analysis on the dynamic power demand index so as to generate a dynamic resource allocation adjustment scheme.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310593660.4A CN116307652A (en) | 2023-05-25 | 2023-05-25 | Artificial intelligent resource allocation method for intelligent power grid |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310593660.4A CN116307652A (en) | 2023-05-25 | 2023-05-25 | Artificial intelligent resource allocation method for intelligent power grid |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116307652A true CN116307652A (en) | 2023-06-23 |
Family
ID=86834532
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310593660.4A Pending CN116307652A (en) | 2023-05-25 | 2023-05-25 | Artificial intelligent resource allocation method for intelligent power grid |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116307652A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116542882A (en) * | 2023-07-06 | 2023-08-04 | 浙江大学 | Photovoltaic power generation smoothing method, system and storage medium |
| CN117094478A (en) * | 2023-10-19 | 2023-11-21 | 深圳库博能源股份有限公司 | Energy scheduling management method, device, equipment and storage medium |
| CN117114364A (en) * | 2023-10-20 | 2023-11-24 | 华北电力大学 | Multi-market collaborative management and control method and system for electric power transaction |
| CN117171551A (en) * | 2023-11-02 | 2023-12-05 | 山东港口科技集团烟台有限公司 | Large-scale industrial equipment data analysis and intelligent management method |
| CN117175556A (en) * | 2023-09-01 | 2023-12-05 | 滁州苏埠新能源科技有限公司 | Photovoltaic power supply management system |
| CN117236527A (en) * | 2023-11-13 | 2023-12-15 | 宁德市天铭新能源汽车配件有限公司 | Automobile part demand prediction method and system based on ensemble learning |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013123446A1 (en) * | 2012-02-16 | 2013-08-22 | Lazaris Spyros | A system and a method for generation. and transmission of electrical energy from renewable resources. |
| WO2019225834A1 (en) * | 2018-05-23 | 2019-11-28 | 주식회사 광명전기 | Power supply control system and method using energy storage device and photovoltaic power generation |
| US20210281077A1 (en) * | 2020-03-03 | 2021-09-09 | Caterpillar Inc. | Micro-grid site predictive control for multi-power resource management with machine learning |
| KR20220075078A (en) * | 2020-11-27 | 2022-06-07 | 주식회사 비온시이노베이터 | Monitoring and control of energy management for economic operation of building-integrated photovoltaic power generation facilities |
| KR20220095313A (en) * | 2020-12-29 | 2022-07-07 | 주식회사 그리드위즈 | Digital Twin Based Distributed Energy Resource and Power Line Management Plan System and Method thereof |
| US20220407310A1 (en) * | 2019-04-26 | 2022-12-22 | Energy Toolbase Software, Inc. | Intelligent energy management system for distributed energy resources and energy storage systems using machine learning |
-
2023
- 2023-05-25 CN CN202310593660.4A patent/CN116307652A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013123446A1 (en) * | 2012-02-16 | 2013-08-22 | Lazaris Spyros | A system and a method for generation. and transmission of electrical energy from renewable resources. |
| WO2019225834A1 (en) * | 2018-05-23 | 2019-11-28 | 주식회사 광명전기 | Power supply control system and method using energy storage device and photovoltaic power generation |
| US20220407310A1 (en) * | 2019-04-26 | 2022-12-22 | Energy Toolbase Software, Inc. | Intelligent energy management system for distributed energy resources and energy storage systems using machine learning |
| US20210281077A1 (en) * | 2020-03-03 | 2021-09-09 | Caterpillar Inc. | Micro-grid site predictive control for multi-power resource management with machine learning |
| KR20220075078A (en) * | 2020-11-27 | 2022-06-07 | 주식회사 비온시이노베이터 | Monitoring and control of energy management for economic operation of building-integrated photovoltaic power generation facilities |
| KR20220095313A (en) * | 2020-12-29 | 2022-07-07 | 주식회사 그리드위즈 | Digital Twin Based Distributed Energy Resource and Power Line Management Plan System and Method thereof |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116542882A (en) * | 2023-07-06 | 2023-08-04 | 浙江大学 | Photovoltaic power generation smoothing method, system and storage medium |
| CN116542882B (en) * | 2023-07-06 | 2023-09-19 | 浙江大学 | Photovoltaic power generation smoothing method, system and storage medium |
| CN117175556A (en) * | 2023-09-01 | 2023-12-05 | 滁州苏埠新能源科技有限公司 | Photovoltaic power supply management system |
| CN117094478A (en) * | 2023-10-19 | 2023-11-21 | 深圳库博能源股份有限公司 | Energy scheduling management method, device, equipment and storage medium |
| CN117094478B (en) * | 2023-10-19 | 2024-02-06 | 深圳库博能源股份有限公司 | Energy scheduling management method, device, equipment and storage medium |
| CN117114364A (en) * | 2023-10-20 | 2023-11-24 | 华北电力大学 | Multi-market collaborative management and control method and system for electric power transaction |
| CN117171551A (en) * | 2023-11-02 | 2023-12-05 | 山东港口科技集团烟台有限公司 | Large-scale industrial equipment data analysis and intelligent management method |
| CN117171551B (en) * | 2023-11-02 | 2024-01-30 | 山东港口科技集团烟台有限公司 | Large-scale industrial equipment data analysis and intelligent management method |
| CN117236527A (en) * | 2023-11-13 | 2023-12-15 | 宁德市天铭新能源汽车配件有限公司 | Automobile part demand prediction method and system based on ensemble learning |
| CN117236527B (en) * | 2023-11-13 | 2024-02-06 | 宁德市天铭新能源汽车配件有限公司 | Automobile part demand prediction method and system based on ensemble learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116307652A (en) | Artificial intelligent resource allocation method for intelligent power grid | |
| US11409347B2 (en) | Method, system and storage medium for predicting power load probability density based on deep learning | |
| Zhao et al. | A novel bidirectional mechanism based on time series model for wind power forecasting | |
| CN109885907B (en) | Cloud model-based satellite attitude control system health state assessment and prediction method | |
| CN112132351A (en) | Electric vehicle charging load clustering prediction method and device | |
| CN112990556A (en) | User power consumption prediction method based on Prophet-LSTM model | |
| CN111241755A (en) | Power load prediction method | |
| Sengar et al. | Ensemble approach for short term load forecasting in wind energy system using hybrid algorithm | |
| CN110717610A (en) | Wind power prediction method based on data mining | |
| Wang et al. | A novel decomposition-ensemble forecasting system for dynamic dispatching of smart grid with sub-model selection and intelligent optimization | |
| CN113011680A (en) | Power load prediction method and system | |
| CN115983485A (en) | Prediction method and system for carbon emission of power plant | |
| Ding et al. | Serial-parallel dynamic echo state network: A hybrid dynamic model based on a chaotic coyote optimization algorithm for wind speed prediction | |
| CN117713221B (en) | Micro-inversion photovoltaic grid-connected optimization system | |
| CN115099511A (en) | Photovoltaic power probability estimation method and system based on optimized copula | |
| CN111815039A (en) | Weekly scale wind power probability prediction method and system based on weather classification | |
| CN113033898A (en) | Electrical load prediction method and system based on K-means clustering and BI-LSTM neural network | |
| CN117113086A (en) | Energy storage unit load prediction method, system, electronic equipment and medium | |
| CN112330044A (en) | Support vector regression model based on iterative aggregation grid search algorithm | |
| Zhai et al. | Combining PSO-SVR and Random Forest Based Feature Selection for Day-ahead Peak Load Forecasting. | |
| Liu et al. | Embedded Scenario Clustering for Wind and Photovoltaic Power, and Load Based on Multi-Head Self-Attention | |
| Liu et al. | Varying condition scopf optimization based on deep learning and knowledge graph | |
| Ye et al. | Optimal Component IGSCV-SVR Ensemble Model Improved by VMD for Ultra-short-term Wind Speed Forecasting. | |
| Wang et al. | Novel wind-speed prediction system based on dimensionality reduction and nonlinear weighting strategy for point-interval prediction | |
| CN117220318B (en) | Power grid digital driving control method and system |
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 |