CN113469228A - Power load abnormal value identification method based on data flow space-time characteristics - Google Patents
Power load abnormal value identification method based on data flow space-time characteristics Download PDFInfo
- Publication number
- CN113469228A CN113469228A CN202110680396.9A CN202110680396A CN113469228A CN 113469228 A CN113469228 A CN 113469228A CN 202110680396 A CN202110680396 A CN 202110680396A CN 113469228 A CN113469228 A CN 113469228A
- Authority
- CN
- China
- Prior art keywords
- data
- state
- time
- space
- target node
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Water Supply & Treatment (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Public Health (AREA)
- Molecular Biology (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Complex Calculations (AREA)
Abstract
A power load abnormal value identification method based on data flow space-time characteristics belongs to the technical field of equipment operation. Step 1) obtaining a real-time data stream in a time-space sliding window, and mapping the real-time data stream to a corresponding state space; step 2) constructing the state sequence of the target node into a Markov chain form, and then extracting the time characteristic of the target node data stream by calculating a state transition probability matrix; step 3) respectively constructing the state sequences of the target node and the neighbor nodes into a Markov chain form, and then calculating a cross state probability transition matrix to extract spatial features; and 4) incorporating the extracted space-time characteristics into a trained convolutional neural network model, and carrying out anomaly detection and classification identification on the target node data stream. The invention can accurately identify the abnormal value and the normal data in the transmission data of the power system, has lower false detection and missed detection rates, can timely transmit the abnormal data to the data terminal, record the abnormal value and carry out data verification.
Description
Technical Field
A power load abnormal value identification method based on data flow space-time characteristics belongs to the technical field of equipment operation.
Background
With the development of big data and smart grids, the correct transmission of data is the most basic requirement and foundation of big data analysis and smart grids. At present, intelligent data analysis based on a big data platform is a current research hotspot, and the correctness of data is a basic guarantee. Therefore, in the era of mass data transmission and big outbreak, the correctness of power transmission data is realized based on big data, and the fundamental value of system data analysis is guaranteed to have important significance.
The condition that the received data is abnormal due to delayed data uploading and transmission errors or faults exists in the data transmission of the power system, and the data monitoring and analysis of the power system are influenced. Therefore, a large number of methods are currently developed for identification of power transmission data. The load identification method based on the association rule, the confidence interval and other methods has the advantages that the abnormal value can be accurately identified based on certain data distribution, the calculation amount is small, and the calculation is convenient. The defects are that the connection and the data distribution among the data are over-dependent, and the identification precision of the connection among the lacking data is poor. The clustering method based on local matrix reconstruction and spatial density can accurately identify more obvious abnormal loads, and under the condition that cluster division is not obvious, the correctness of data is difficult to effectively distinguish, and missing detection or false detection is easy to cause. In addition, algorithms based on data spatio-temporal features identify fault outliers by configuring confidence levels of adjacent neighbors. However, the data identification method based on the data space-time characteristics is relatively dependent on the connection and correlation among data, and the accuracy effect on the identification of the data is not ideal.
The method aims at the problem that the identification method of the abnormal value of the power load depends on the data distribution and the correlation between the data. Therefore, research work for developing a load abnormal value identification method based on load data transmitted by a power system without depending on data characteristic value distribution and data correlation is urgently needed.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the method overcomes the defects of the prior art, and provides the method for identifying the abnormal power load value based on the time-space characteristics of the data stream, which is used for identifying the abnormal load value based on the load data transmitted by the power system under the condition of not depending on the distribution of the characteristic values of the data and the relevance of the data.
The technical scheme adopted by the invention for solving the technical problems is as follows: the method for identifying the abnormal value of the power load based on the time-space characteristics of the data stream is characterized in that: the method comprises the following steps:
step 1) obtaining a real-time data stream in a time-space sliding window, and mapping the real-time data stream to a corresponding state space;
step 2) constructing the state sequence of the target node into a Markov chain form, and then extracting the time characteristic of the target node data stream by calculating a state transition probability matrix;
step 3) respectively constructing the state sequences of the target node and the neighbor nodes into a Markov chain form, and then calculating a cross state probability transition matrix to extract spatial features;
and 4) incorporating the extracted space-time characteristics into a trained convolutional neural network model, and carrying out anomaly detection and classification identification on the target node data stream.
Preferably, in the data values and the data selection range in the spatio-temporal sliding window in step 1), determining a transmission sequence G of data:
G={G1,L,Gn},
wherein G is1Refer to target node, { G2,L,GnThe method comprises the steps that (1) a neighboring node sequence of a target node is referred, a time-space sliding window is formed by detection data of F moments near the node, the window size is n multiplied by F, and n refers to the transmission moment of the data;
deleting the last end data of the transmission sequence G, adding the latest data, completing the acquisition of real-time data, and determining a new data transmission sequence G':
G′={G1,L,Gn}+Gn+1-G1={G2,L,Gn+1},
wherein G isn+1Representing data newly generated and added to a data transmission sequence;
according to the 3 sigma criterion, the data acquisition sequence is subjected to basic state division:
{Δst-3λ,Δst+3λ},
wherein Δ stλ is the standard deviation of the sequence, which is the sample mean of the transmitted data at time t.
Preferably, the eigenvalue region is divided into nine sub-regions, G ═ a, b, c, d, e, f, G, h, i }, and the transmission data is mapped to the state space according to the size of the difference, and the corresponding mapping function is:
preferably, the first-order Markov chain model in step 2) is T ═ { T ═ T1,T2,L,TtCorresponding to the state space G ═ a, b, c, d, e, f, G, h, i, for any G1、g2、L、gtE.g. G, has
L{Tt=gt|T0=g0,T1=g1,L,Tt-1=gt-1}=L{Tt=gt|Tt-1=gt-1}。
Preferably, the state transition rate K is calculatedi,j:
Wherein, giAnd gjRespectively representing a first-order Markov chainThe state of the model T at time T-1 and T, and Z is the total number of occurrences of the state.
Preferably, the state transition probability κ is calculatedi,jThe state transition probability matrix K constructed in the state space G:
wherein, κi,jIs not less than 0 andgiand gjRepresenting the states of the first-order Markov chain model T at times T-1 and T, respectively.
Preferably, let A be the target node, B be the neighbor node of A, and the Markov chain forms of the state sequences of nodes A and B are respectivelyThe state space is GAAnd GB。
Preferably, the first and second liquid crystal materials are,as a model TAIn the state at the time t-1,as a model TBRecording the cross state transition probability for the state at time t
Where Z is the total number of occurrences of the state.
Preferably, the formed cross-state transition probability matrix K is calculatedAB:
Preferably, the convolutional neural network in step 4) includes an input layer, a max pooling layer, a convolutional layer, a full link layer and an output layer;
the training method of the convolutional neural network comprises the following steps:
step 4001) giving an input vector and an objective function output;
step 4002) obtaining a hidden layer, and outputting the output of each unit of the hidden layer;
step 4003) calculating a target value and an actual output bias value e;
step 4004) judging whether e is within an allowable range, if so, finishing training; if not, execute step 4005;
step 4005) calculating errors of neurons in the network layer;
step 4006) calculating a gradient of the error;
step 4007) calculating errors of neurons in the network layer;
step 4008) performs step 4002.
Compared with the prior art, the invention has the beneficial effects that:
the method for identifying the abnormal value of the power load based on the time-space characteristics of the data stream establishes a time-space sliding window, and determines a transmission sequence of data according to a data value and a data selection range in the sliding window; after new data is generated, deleting the data at the tail end, adding the latest data, and finishing the acquisition of real-time data; according to a 3 sigma criterion, carrying out state division on a data acquisition sequence; mapping the transmission data to a state space according to the difference value; constructing a Markov chain mapped to state space data, and calculating a state transition probability matrix to obtain time characteristics of the data; calculating a cross state transition probability matrix to capture the spatial dependency of a target node and adjacent nodes, and extracting the spatial characteristics of a target node data stream; the calculated space-time characteristics are used as the input of a multi-classification convolutional neural network model, and the abnormal value identification of the transmission data is realized through the calculation of a multi-convolutional neural network; and pushing the identified abnormal value to a terminal, reminding operation and maintenance personnel or a data manager, recording the abnormal value in a database, and developing a load abnormal value identification method based on load data transmitted by the power system under the condition of not depending on data characteristic value distribution and data correlation.
The method for identifying the abnormal value of the power load based on the time-space characteristics of the data stream can accurately identify the abnormal value and the normal data in the transmission data of the power system, has low false detection and missing detection rates, can timely transmit the abnormal data to a data terminal, records the abnormal value and performs data verification. The method provided by the invention is analyzed through big data and a deep learning algorithm, does not need to invest a large amount of equipment, and is strong in practicability and easy to popularize.
Drawings
FIG. 1 is a block diagram of a method for identifying abnormal values of power loads based on spatiotemporal features of data streams.
Fig. 2 is a flowchart of a convolutional neural network algorithm.
Detailed Description
FIGS. 1-2 illustrate preferred embodiments of the present invention, and the present invention will be further described with reference to FIGS. 1-2.
The present invention is further described with reference to the following detailed description, however, it should be understood by those skilled in the art that the detailed description given herein with respect to the accompanying drawings is for better explanation and that the present invention is not necessarily limited to the specific embodiments, but rather, for equivalent alternatives or common approaches, may be omitted from the detailed description, while still remaining within the scope of the present application.
As shown in fig. 1: a method for identifying abnormal values of power loads based on data flow space-time characteristics comprises the following steps:
step 1) obtaining a real-time data stream in a space-time sliding window, and mapping the real-time data stream to a corresponding state space.
In the data value and data selection range in the time-space sliding window, the transmission sequence G of data is determined, in this embodiment, the data values on 9 time scales are selected to determine the transmission sequence G of data:
G={G1,L,G9},
G1refer to target node, { G1,L,G9And F time points are detected by the target node, and the time-space sliding window is formed by the detection data of F time points near the target node, and the window size is 9 multiplied by F.
Deleting the last end data of the transmission sequence G, adding the latest data, completing the acquisition of real-time data, and determining a new data transmission sequence G':
G′={G1,L,G9}+G10-G1={G2,L,G10}。
wherein G is10Representing data newly generated and added to a data transmission sequence;
according to the 3 sigma criterion, the data acquisition sequence is subjected to basic state division:
{Δst-3λ,Δst+3λ},
wherein Δ stλ is the standard deviation of the sequence, which is the sample mean of the transmitted data at time t.
Dividing the eigenvalue region into nine sub-regions, wherein G ═ { a, b, c, d, e, f, G, h, i }, mapping the transmission data to the state space according to the size of the difference, and the corresponding mapping function is as follows:
and 2) constructing the state sequence of the target node into a Markov chain form, and then extracting the time characteristic of the data stream of the target node by calculating a state transition probability matrix.
The first-order Markov chain model in the step 2) is T ═ T { (T)1,T2,L,TtThe corresponding formState space G ═ a, b, c, d, e, f, G, h, i }, for any G1、g2、L、gtE.g. G, has
L{Tt=gt|T0=g0,T1=g1,L,Tt-1=gt-1}=L{Tt=gt|Tt-1=gt-1}。
Calculating the State transition Rate Ki,j:
Wherein, giAnd gjRepresenting the states of the first-order Markov chain model T at times T-1 and T, respectively, and Z being the total number of occurrences of the states.
Calculating the State transition probability Ki,jThe state transition probability matrix K formed in the state space G:
and 3) respectively constructing the state sequences of the target node and the neighbor nodes into a Markov chain form, and then calculating a cross state probability transition matrix to extract the spatial characteristics.
Let A be a target node, B be a neighbor node of A, and Markov chain forms of state sequences of nodes A and B are respectivelyThe state space is GAAnd GB。
As a model TAIn the state at the time t-1,as a model TBRecording the cross state transition probability for the state at time t
Where Z is the total number of occurrences of the state.
Calculating the formed cross state transition probability matrix KAB:
And 4) incorporating the extracted space-time characteristics into a trained convolutional neural network model, and carrying out anomaly detection and classification identification on the target node data stream.
As shown in fig. 2: the convolutional neural network realizes the abnormal value detection and type identification of data transmission. A multi-classification convolutional neural network model is selected as a method for classifying and identifying the spatio-temporal features. The designed convolutional neural network model has 8 layers in total, and consists of an input layer, a maximum pooling layer, a convolutional layer, a full-link layer and an output layer. The training process includes back and forward propagation: in forward propagation, different features of the input layer data are extracted through the pooling layer, the convolutional layer. And then, integrating features through a full connection layer, obtaining a classification result through Softmax, and calculating cross entropy loss. In back propagation, gradient values are calculated according to the chain rule and each layer weight is updated by a random gradient descent method.
The convolutional neural network training method comprises the following steps:
step 4001) giving an input vector and an objective function output;
step 4002) obtaining a hidden layer, and outputting the output of each unit of the hidden layer;
step 4003) calculating a target value and an actual output bias value e;
step 4004) judging whether e is within an allowable range, if so, finishing training; if not, execute step 4005;
step 4005) calculating errors of neurons in the network layer;
step 4006) calculating a gradient of the error;
step 4007) calculating errors of neurons in the network layer;
step 4008) performs step 4002.
The method for detecting the abnormity and identifying the classification of the target node data flow comprises the following steps:
step a, the model input layer is formed by m 9 multiplied by 9 matrixes to form a space-time characteristic matrix setWhich includes a state probability transition matrixAnd cross state probability transition matrix
Step b, the model adopts alternate pooling layers and convolution layers to extract a space-time feature matrix setFeature maps in different regions, where the first alternate layer passes through a set of 64 convolution kernel matrices of 1 × 1 step size and 3 × 3 step sizePerforming convolution operation, and performing ReLU functionThe feature map is output and then compressed in conjunction with the pooling layer with a stride of 1 x 1 and a window size of 2 x 2. The second alternating layer keeps the other parameters unchanged, the size is 3 × 3, the number of convolution kernels is 128, the final characteristics are obtained, and the convolution layer form is expressed as:
where b is the offset, l is the current layer, k is the convolution kernel, MjIs a convolution window.
The pooling layer is expressed as:
where down is the down-sampling function, b is the offset, and β is the weighting coefficient.
Step c, the model integrates and reduces dimensions of the characteristic diagrams of the pooling layer and the convolution layer in a multilayer artificial neural network mode, 2 full-connection layers are adopted, the number of the neurons of the full-connection layers is 64 and 128, after the full-connection layers pass through, the characteristic diagrams are changed into 64-dimensional vectors, and the form of the full-connection layers is expressed as follows:
wherein, ω isijIs a weight value.
And d, connecting the 64-dimensional vectors to an output layer through a Softmax classifier, calculating output probability, wherein 4 neurons of the output layer represent types of target node data streams, namely normal, data mutation abnormity, data reporting abnormity and data dislocation fault, judging whether the data is abnormal according to results, and transmitting abnormal values to a data terminal for recording and data display, so that data verification is facilitated.
The foregoing is directed to preferred embodiments of the present invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof, and the scope thereof is determined by the claims that follow. However, any simple modification, equivalent change and modification of the above embodiments according to the technical essence of the present invention are within the protection scope of the technical solution of the present invention.
Claims (10)
1. A method for identifying abnormal values of power loads based on time-space characteristics of data streams is characterized by comprising the following steps: the method comprises the following steps:
step 1) obtaining a real-time data stream in a time-space sliding window, and mapping the real-time data stream to a corresponding state space;
step 2) constructing the state sequence of the target node into a Markov chain form, and then extracting the time characteristic of the target node data stream by calculating a state transition probability matrix;
step 3) respectively constructing the state sequences of the target node and the neighbor nodes into a Markov chain form, and then calculating a cross state probability transition matrix to extract spatial features;
and 4) incorporating the extracted space-time characteristics into a trained convolutional neural network model, and carrying out anomaly detection and classification identification on the target node data stream.
2. The method for identifying abnormal values of power loads based on spatiotemporal features of data streams as claimed in claim 1, wherein: determining a data transmission sequence G in the data value and data selection range in the space-time sliding window in the step 1):
G={G1,L,Gn},
wherein G is1Refer to target node, { G2,L,GnThe method comprises the steps that (1) a neighboring node sequence of a target node is referred, a time-space sliding window is formed by detection data of F moments near the node, the window size is n multiplied by F, and n refers to the transmission moment of the data;
deleting the last end data of the transmission sequence G, adding the latest data, completing the acquisition of real-time data, and determining a new data transmission sequence G':
G′={G1,L,Gn}+Gn+1-G1={G2,L,Gn+1},
wherein G isn+1Representing data newly generated and added to a data transmission sequence;
according to the 3 sigma criterion, the data acquisition sequence is subjected to basic state division:
{Δst-3λ,Δst+3λ},
wherein Δ stλ is the standard deviation of the sequence, which is the sample mean of the transmitted data at time t.
3. The method for identifying abnormal values of power loads based on spatiotemporal features of data streams as claimed in claim 2, wherein: dividing the eigenvalue region into nine sub-regions, G ═ { a, b, c, d, e, f, G, h, i }, mapping the transmission data to a state space according to the size of the difference, and the corresponding mapping function is:
4. the method for identifying abnormal values of power loads based on spatiotemporal features of data streams as claimed in claim 1, wherein: the first-order Markov chain model in the step 2) is T ═ T { (T)1,T2,L,TtCorresponding to the state space G ═ a, b, c, d, e, f, G, h, i, for any G1、g2、L、gtE.g. G, has
L{Tt=gt|T0=g0,T1=g1,L,Tt-1=gt-1}=L{Tt=gt|Tt-1=gt-1}。
5. The method for identifying abnormal values of power loads based on spatiotemporal features of data streams as claimed in claim 4, wherein: computing state transitionsRate of shift κi,j:
Wherein, giAnd gjRepresenting the states of the first-order Markov chain model T at times T-1 and T, respectively, and Z being the total number of occurrences of the states.
6. The method for identifying abnormal values of power loads based on spatiotemporal features of data streams as claimed in claim 1, wherein: calculating the State transition probability Ki,jThe state transition probability matrix K constructed in the state space G:
7. The method for identifying abnormal values of power loads based on spatiotemporal features of data streams as claimed in claim 1, wherein: let A be a target node, B be a neighbor node of A, and the Markov chain forms of the state sequences of nodes A and B are respectivelyThe state space is GAAnd GB。
8. The method for identifying abnormal values of power loads based on spatiotemporal features of data streams as claimed in claim 7, wherein:as a model TAIn the state at the time t-1,as a model TBRecording the cross state transition probability for the state at time t
Where Z is the total number of occurrences of the state.
10. The method for identifying abnormal values of power loads based on spatiotemporal features of data streams as claimed in claim 1, wherein: the convolutional neural network in the step 4) comprises an input layer, a maximum pooling layer, a convolutional layer, a full-link layer and an output layer;
the training method of the convolutional neural network comprises the following steps:
step 4001) giving an input vector and an objective function output;
step 4002) obtaining a hidden layer, and outputting the output of each unit of the hidden layer;
step 4003) calculating a target value and an actual output bias value e;
step 4004) judging whether e is within an allowable range, if so, finishing training; if not, execute step 4005;
step 4005) calculating errors of neurons in the network layer;
step 4006) calculating a gradient of the error;
step 4007) calculating errors of neurons in the network layer;
step 4008) performs step 4002.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110680396.9A CN113469228A (en) | 2021-06-18 | 2021-06-18 | Power load abnormal value identification method based on data flow space-time characteristics |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110680396.9A CN113469228A (en) | 2021-06-18 | 2021-06-18 | Power load abnormal value identification method based on data flow space-time characteristics |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113469228A true CN113469228A (en) | 2021-10-01 |
Family
ID=77868905
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110680396.9A Pending CN113469228A (en) | 2021-06-18 | 2021-06-18 | Power load abnormal value identification method based on data flow space-time characteristics |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113469228A (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110213788A (en) * | 2019-06-15 | 2019-09-06 | 福州大学 | WSN abnormality detection and kind identification method based on data flow space-time characteristic |
CN111242276A (en) * | 2019-12-27 | 2020-06-05 | 国网山西省电力公司大同供电公司 | One-dimensional convolution neural network construction method for load current signal identification |
-
2021
- 2021-06-18 CN CN202110680396.9A patent/CN113469228A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110213788A (en) * | 2019-06-15 | 2019-09-06 | 福州大学 | WSN abnormality detection and kind identification method based on data flow space-time characteristic |
CN111242276A (en) * | 2019-12-27 | 2020-06-05 | 国网山西省电力公司大同供电公司 | One-dimensional convolution neural network construction method for load current signal identification |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109492822B (en) | Air pollutant concentration time-space domain correlation prediction method | |
CN116757534B (en) | Intelligent refrigerator reliability analysis method based on neural training network | |
CN109034191B (en) | ELM-based one-dimensional telemetry data abnormal interpretation method | |
CN112132430B (en) | Reliability evaluation method and system for distributed state sensor of power distribution main equipment | |
CN112966714B (en) | Edge time sequence data anomaly detection and network programmable control method | |
CN108133225A (en) | A kind of icing flashover fault early warning method based on support vector machines | |
CN113762329A (en) | Method and system for constructing state prediction model of large rolling mill | |
CN110738355A (en) | urban waterlogging prediction method based on neural network | |
CN116684878B (en) | 5G information transmission data safety monitoring system | |
CN110674858A (en) | Traffic jam detection method based on space-time association and big data mining | |
CN114266301A (en) | Intelligent power equipment fault prediction method based on graph convolution neural network | |
CN112464996A (en) | Intelligent power grid intrusion detection method based on LSTM-XGboost | |
CN114265913A (en) | Space-time prediction algorithm based on federal learning on industrial Internet of things edge equipment | |
CN116021981A (en) | Method, device, equipment and storage medium for predicting ice coating faults of power distribution network line | |
Ye et al. | A deep learning-based method for automatic abnormal data detection: Case study for bridge structural health monitoring | |
CN117114913A (en) | Intelligent agricultural data acquisition system based on big data | |
CN117272196A (en) | Industrial time sequence data anomaly detection method based on time-space diagram attention network | |
CN110689140A (en) | Method for intelligently managing rail transit alarm data through big data | |
CN114444582A (en) | Mechanical equipment fault diagnosis method based on convolutional neural network and Bayesian network | |
CN116842459A (en) | Electric energy metering fault diagnosis method and diagnosis terminal based on small sample learning | |
CN116485029A (en) | Method, device, equipment and medium for detecting accuracy of dynamic data in oilfield development | |
CN113469228A (en) | Power load abnormal value identification method based on data flow space-time characteristics | |
CN113884807B (en) | Power distribution network fault prediction method based on random forest and multi-layer architecture clustering | |
CN115062686A (en) | Multi-KPI (Key performance indicator) time sequence abnormity detection method and system based on multi-angle features | |
Sumalatha et al. | Real Time Big Data Analytics for Agricultural Land Hotspot Prediction |
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 |