CN109308306A - A kind of user power utilization anomaly detection method based on isolated forest - Google Patents
A kind of user power utilization anomaly detection method based on isolated forest Download PDFInfo
- Publication number
- CN109308306A CN109308306A CN201811151326.9A CN201811151326A CN109308306A CN 109308306 A CN109308306 A CN 109308306A CN 201811151326 A CN201811151326 A CN 201811151326A CN 109308306 A CN109308306 A CN 109308306A
- Authority
- CN
- China
- Prior art keywords
- electricity consumption
- data
- user
- value
- trend
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The present invention provides a kind of user power utilization anomaly detection methods based on isolated forest, include the following steps: S1, obtain electricity consumption time series data by data acquisition modes;S2, data are cleaned, incomplete data, wrong data, repeated data is removed;S3, the feature extraction based on statistics;S4, data prediction;S5, to matrix YM×KIt is normalized to obtain new matrix YM×K';S6, judge that multiplexing electric abnormality or electricity consumption are normal using isolated forest model: S61, from new matrix YM×KMiddle extraction, each user extracts ψ statistical nature, if the quantity t, y of iTree treeijIt is new matrix YM×KIn the i-th row jth column element;S62, y is calculatedijAbnormal score s (yij,ψ);S63, judge s (yij, ψ) whether it is less than 1- Δ e, Δ e is the constant in 0.22~0.07 range;If so, being multiplexing electric abnormality;If it is not, then electricity consumption is normal.User power utilization anomaly detection method based on isolated forest solves the problems, such as to cause to lead to analytical calculation long operational time due to subsequent arithmetic is larger because not carrying out processing to data in the prior art.
Description
Technical field
The present invention relates to electricity consumption monitoring fields, and in particular to a kind of user power utilization unusual checking based on isolated forest
Method.
Background technique
The multiplexing electric abnormality monitoring method of relatively early stage is determining each multiplexing electric abnormality index, determines the threshold of each abnormal index
Value, and different weight score values is assigned to each abnormal index, the stealing suspicion coefficient of each user is calculated after cumulative.Generally
Multiplexing electric abnormality index be briefly divided into that line loss is abnormal and abnormal two classes of instantaneous flow.According to these design stealing identifications extremely
Model identifies stealing user by calculating suspicion coefficient.
However for the detection of this kind of equipment fault and user power utilization abnormal index, what is mostly used in early days is on-site test side
Method, i.e. technical staff are checked to electricity consumption scene.This processing mode extremely spends human and material resources resource, low efficiency, effect
Difference can only monitor daily power consumption some areas realize centralized automatic meter-reading, and can not get the electricity of metering device
The instantaneous flows data such as pressure, electric current, power.Meanwhile this mode is unfavorable for power industry there is also great human factor
Management.
Chinese patent discloses that a kind of application No. is the electricity consumption based on fuzzy neural network of CN201810104000.4 is different
Normal Activity recognition method, the initial data of extraction section user is as sample data from electricity consumption data library;Data are carried out to locate in advance
Reason;On the basis of analysis of history multiplexing electric abnormality behavior case, multiplexing electric abnormality behavior evaluation index system is designed;Utilize pre- place
Data after reason construct expert's sample;It is output with abnormal electricity consumption suspicion coefficient using abnormal electricity consumption behavior mark as input item
, construct fuzzy neural network model;By the constructed fuzzy neural network model of test data input, abnormal electricity consumption is carried out
Behavior diagnosis;Evaluation is made to abnormal electricity consumption diagnostic result, sets objective appraisal, Optimized model.It is different that the present invention realizes electricity consumption
The automatic identification of Chang Hangwei diagnoses, and using the method for fuzzy neural network, realizes the automatic training study of system and builds
Mould reaches and quickly accurately positions suspicion user again, to obtain the convenience that the unlawful practice of various abnormal electricity consumptions provides.But
Cause subsequent arithmetic larger due to not carrying out processing to data, long operational time easily causes to occur when machine phenomenon.
Summary of the invention
The present invention will provide a kind of user power utilization anomaly detection method based on isolated forest, solve in the prior art
Lead to the problem of leading to analytical calculation long operational time due to subsequent arithmetic is larger because not carrying out processing to data.
To achieve the above object, present invention employs the following technical solutions:
A kind of user power utilization anomaly detection method based on isolated forest, includes the following steps:
S1, electricity consumption time series data is obtained by data acquisition modes;
S2, data are cleaned, incomplete data, wrong data, repeated data is removed;
S3, the feature extraction based on statistics:
S31, data definition: S311, data set is enabled to be X={ xn, n takes 1 to N, includes N number of daily electricity consumption in data set
User, each user are divided into D days, M months, the electricity consumption data in Q season;The daily power consumption sequence of S312, each user:
xn={ xnd, d takes 1 to D;The moon electricity consumption sequence of S313, each user: yn={ ynm, m takes 1 to M,
The season electricity consumption sequence of S114, each user: zn={ znq, q takes 1 to Q,
S32, user power utilization behavioural characteristic is divided as unit of year, season, month in time, and calculates each use
Unit time mean value, standard deviation and the coefficient of dispersion sequence at family are to calculate: the Urban Annual Electrical Power Consumption amount standard deviation D1 of each user,
The Urban Annual Electrical Power Consumption amount coefficient of dispersion D2 of each user, quarterly electricity consumption standard deviation D3~D6, quarterly electricity consumption coefficient of dispersion
On D7~D10, monthly electricity consumption standard deviation D11~D21, monthly electricity consumption coefficient of dispersion D22~D32, every monthly average electricity consumption
Liter downward trend D33~D41, the difference of adjacent two months electricity consumption mean values and maximum value D42~D43 of ratio, adjacent electricity consumption in two months are equal
Maximum value D46~D47, adjacent of the difference and minimum value D44~D45 of ratio of value, the difference of adjacent season electricity consumption mean value and ratio
The difference of season electricity consumption mean value and minimum value D48~D49 of ratio, wherein D1~D49 is statistical nature;
S4, data prediction: assuming that initial data is used to be formed with n-dimensional vector after the mode based on statistical nature is handled
M sample value, M indicates user's number, and N indicates the number for the statistical nature that each user extracts, and enable its for a M ×
X in the matrix X of N, matrix XmnIndicate the occurrence of m-th user n-th statistical nature;Matrix X is dropped by pretreated model
For the matrix Y of M × KM×K, K < N;
S5, judge that multiplexing electric abnormality or electricity consumption are normal using isolated forest model:
S51, from new matrix YM×KMiddle extraction, each user extracts ψ statistical nature, if the quantity t, y of iTree treeijIt is
New matrix YM×KIn the i-th row jth column element;
S52, detection process be exactly to allow the statistical characteristics y of each userijEach iTree tree is traversed, is then calculated
Y in ergodic processijBy the path length h (y of every iTree treeij), y is finally calculated according to all path lengthsij's
Abnormal score s (yij, ψ), calculation formula are as follows:
Wherein, c (ψ) is used to calculate the average path length of binary search tree, and effect is that result is normalized;
The calculation of H (ψ) is:γ is Euler's constant;E (h (yij)) is that yij owns in isolated forest
The average path length of iTree tree;
S53, judge s (yij, ψ) whether it is less than 1- Δ e, Δ e is the constant in 0.22~0.07 range;If so, to use
Electrical anomaly;If it is not, then electricity consumption is normal.
Compared with the prior art, the invention has the following beneficial effects:
By realizing that statistical nature extracts, effective data are obtained;By realizing dimension-reduction treatment, reduce operand
According to improving arithmetic speed, avoid the generation when machine phenomenon, while being chosen by condition, ensure that operational data has generation
Table, surface are calculated because choosing some statistical natures and lead to occur failing to judge phenomenon appearance, ensure that judging result
Precision.
Further advantage, target and feature of the invention will be partially reflected by the following instructions, and part will also be by this
The research and practice of invention and be understood by the person skilled in the art.
Detailed description of the invention
Fig. 1 is the algorithm realization procedure chart of isolated forest model;
Fig. 2 is autocoder network structure;
Fig. 3 is that the ReLU of autocoder activates letter image graph;
Fig. 4 is that the training majorized function algorithm of autocoder realizes figure;
Fig. 5 is deep layer autocoder network structure;
The autocoder network structure that Fig. 6 utilizes keras tool to establish;
Fig. 7 is the deep layer autocoder network structure established using keras tool.
Specific embodiment
In order to make the present invention realize technological means, creation characteristic, reach purpose and effect more clearly and be apparent to,
The present invention is further elaborated with reference to the accompanying drawings and detailed description:
Embodiment 1:
A kind of user power utilization anomaly detection method based on isolated forest, includes the following steps:
S1, electricity consumption time series data is obtained by data acquisition modes;
S2, data are cleaned, incomplete data, wrong data, repeated data is removed;
S3, the feature extraction based on statistics:
S31, data definition: S311, data set is enabled to be X={ xn, n takes 1 to N, includes N number of daily electricity consumption in data set
User, each user are divided into D days, M months, the electricity consumption data in Q season;The daily power consumption sequence of S312, each user:
xn={ xnd, d takes 1 to D;The moon electricity consumption sequence of S313, each user: yn={ ynm, m takes 1 to M,
The season electricity consumption sequence of S114, each user: zn={ znq, q takes 1 to Q,
S32, user power utilization behavioural characteristic is divided as unit of year, season, month in time, and calculates each use
Unit time mean value, standard deviation and the coefficient of dispersion sequence at family are to calculate: the Urban Annual Electrical Power Consumption amount standard deviation D1 of each user,
The Urban Annual Electrical Power Consumption amount coefficient of dispersion D2 of each user, quarterly electricity consumption standard deviation D3~D6, quarterly electricity consumption coefficient of dispersion
On D7~D10, monthly electricity consumption standard deviation D11~D21, monthly electricity consumption coefficient of dispersion D22~D32, every monthly average electricity consumption
Liter downward trend D33~D41, the difference of adjacent two months electricity consumption mean values and maximum value D42~D43 of ratio, adjacent electricity consumption in two months are equal
Maximum value D46~D47, adjacent of the difference and minimum value D44~D45 of ratio of value, the difference of adjacent season electricity consumption mean value and ratio
The difference of season electricity consumption mean value and minimum value D48~D49 of ratio, wherein D1~D49 is statistical nature;
S4, data prediction: assuming that initial data is used to be formed with n-dimensional vector after the mode based on statistical nature is handled
M sample value, M indicates user's number, and N indicates the number for the statistical nature that each user extracts, and enable its for a M ×
X in the matrix X of N, matrix XmnIndicate the occurrence of m-th user n-th statistical nature;Matrix X is dropped by pretreated model
For the matrix Y of M × KM×K, K < N;
S5, judge that multiplexing electric abnormality or electricity consumption are normal using isolated forest model:
S51, from new matrix YM×KMiddle extraction, each user extracts ψ statistical nature, if the quantity t, y of iTree treeijIt is
New matrix YM×KIn the i-th row jth column element;
S52, detection process be exactly to allow the statistical characteristics y of each userijEach iTree tree is traversed, is then calculated
Y in ergodic processijBy the path length h (y of every iTree treeij) (walking manner is not walked as isolated forest model
The metering of one step is 1), finally to calculate y according to all path lengthsijAbnormal score s (yij, ψ), calculation formula are as follows:
Wherein, c (ψ) is used to calculate the average path length of binary search tree, and effect is that result is normalized;
The calculation of H (ψ) is:γ is Euler's constant;E (h (yij)) is that yij owns in isolated forest
The average path length of iTree tree;
S53, judge s (yij, ψ) whether it is less than 1- Δ e, Δ e is the constant in 0.22~0.07 range;If so, to use
Electrical anomaly;If it is not, then electricity consumption is normal.
In order to obtain isolated forest model, as shown in Figure 1, the acquisition step of isolated forest model includes:
S711, assume that raw data set is indicated with F, a sample point of F ' is randomly choosed from data set and is put into as subsample
The root node of tree,
One S712, random selection dimension q, are randomly generated a cut-point p, this cut-point in present node data
P is resulted from present node data between the maximum value and minimum value of specified dimension q;
S713, a hyperplane is generated with this cut-point p, present node data space is then divided into 2 sub-spaces:
The data of q<p in specified dimension are put into the left subtree Fl of present node, the data of q>=p are put into the right subtree of present node
Fr;
S714, the recursion step S712 and S713 in child node, constantly construct new children tree nodes, until in children tree nodes
Only one data or children tree nodes arrived restriction height, do not continue to divide, obtain t iTree tree.
In the present embodiment, pretreated model is PCA dimension-reduction treatment.
In order to obtain more effective statistical natures, also followed the steps below after step S12:
S13, electricity consumption tendency is divided into three kinds of alteration trend, fluctuation tendency and lifting trend trend types;
S14, alteration trend, fluctuation tendency and lifting trend are calculated:
S141, fluctuation tendency: the possible variation of assessment sequence or degree of fluctuation, standard are used in statistics Plays difference
Difference is bigger, and the range of numerical fluctuations is bigger;So calculating electricity consumption standard deviation std here to indicate the fluctuation of electricity consumption data
Trend feature;Meanwhile electricity consumption coefficient of dispersion cv is calculated to measure the dispersion degree of user power utilization, enable certain time period level
Mean value is μ, then:
Electricity consumption standard deviation:
Electricity consumption coefficient of dispersion:
Cv=std/ μ (2.2)
S142, alteration trend: alteration trend feature refers to the front and back Diversity measure of user power consumption, i.e., for the moment by certain
Between section compared with the average electricity consumption of previous time adjacent segments, difference and ratio reflect speed degree that electricity consumption changes,
It is as follows to define calculation:
The difference of the adjacent k month or k season electricity consumption mean value:
The ratio of the adjacent k month or k season electricity consumption mean value:
S143, lifting trend: lifting trend feature refers to by being made next time according to continuous several days electricity consumptions of user
The prediction of electricity consumption, and compared with practical electricity consumption next time, obtain a possibility that rising or falling;It is moved used here as simple
The dynamic method of average determines the feature vector of lifting trend;The simple method of moving average elapses item by item according to time series, successively calculates
One cell mean of fixed item number, and as predicted value next time;Enable k for mobile item number, t moment actual value is xnt, then
The calculation method of lifting trend feature:
T moment predicted value:
Ft=(xn(t-1)+xn(t-2)+…xn(t-k))/k (2.5)
T moment lifting trend:
Tr=xnt-Ft (2.6)
If tr < 0, show that electricity consumption trend declines;If tr > 0, electricity consumption trend rises;
Wherein, the difference avg of electric standard difference std, electric coefficient of dispersion cv, the adjacent k month or k season electricity consumption mean valuea, phase
The ratio avg of the adjacent k month or k season electricity consumption mean valueb, t moment lifting trend tr be statistical characteristics.
Preferably, steps are as follows for PCA dimensionality reduction in step s 2:
S21, the mean value that each column of X are subtracted to the corresponding column, i.e., carry out zero averaging for every a line feature of data X,
Obtain X ':
S22, X ' covariance matrix C, vector x are calculatediAnd xjCovariance, in (3.1) formula,
S23, the N number of eigenvalue λ for finding out covariance matrix C and the corresponding feature vector V of each eigenvalue λ:
CV=λ V (3.2)
S24, by all eigenvalue λs according to being arranged in a queue { λ from big to small1..., λi..., λN, according to feature
Feature vector V is rearranged the matrix W of a N*N by value from big to small, and the element of the i-th column is i-th in queue in matrix W
Eigenvalue λiThe element of corresponding feature vector V, and the corresponding feature vector of preceding K characteristic value is taken from matrix W, obtain one
The matrix A of a N × KN×K;
S25, K is calculated according to formula 3.3, takes first K value for meeting 3.3 formulas:
S26, calculation formula 3.4, wherein YM×KNew feature data as after dimensionality reduction to k dimension;
YM×K=XM×NAN×K (3.4)
1. example introduction: experimental data derives from the every of nearly 10000 users of whole year in 2015 that national grid was collected
Daily power consumption tables of data, user's daily power consumption table has recorded the every daily power consumption kilowatt hour of all users, the same day and the previous day
Total electricity consumption expression value, each user possess the time series data that one group of dimension is 334.User's inventory has determined that user identifier is believed
Breath, provide reference numeral user whether be multiplexing electric abnormality user mark.
2. data cleansing: user power utilization initial data concentration obtains 334 valid data dimensions after cleaning treatment,
In comprising 1394 abnormal electricity consumption behavior users and 8562 unknown electricity consumption behavior users, abnormal user ratio is 14.00%.
3. data prediction:
1) based on the data prediction of autocoder: to the data set after cleaning do based on autocoder and depth from
The data prediction of encoder two ways.Data are normalized first, each characteristic dimension data is indicated
Between [0,1], then according to the self-encoding encoder network model of design, the neural network tool based on TensorFlow is utilized
Keras establishes the network layer structure of two kinds of self-encoding encoders, as shown in Figure 4.The activation primitive ReLU of middle layer is set, and training is excellent
Change function is adadelta, and loss function binary_crossentropy, frequency of training is 100 times.
Data are pre-processed by the autocoder and depth self-encoding encoder model of foundation, after 100 training,
Model tends towards stability, and loss value respectively reaches 0.0313 and 0.0311.
After initial data pretreatment, the dimension of data is compressed to 32 dimensions.In order to intuitively to based on autocoder
The validity and performance superiority and inferiority of the preprocess method of model are tested, and pretreated new data set is mapped to such as Fig. 6
It is observed under two-dimensional visualization plane.
Wherein white point is represented without multiplexing electric abnormality suspicion user, and red point represents the user for determining and having multiplexing electric abnormality behavior.
On the one hand, it can be seen that the data point of most of white is gathered in (0,0) areas adjacent in figure, and has fraction diffusion outward,
And the data point of most red is obvious to external diffusion, there are the trend of bias data concentrated area, show the spy of outlier
Property.On the other hand, compared to autocoder model, the exceptional data point based on depth self-encoding encoder model preprocessing is presented
The distribution more dispersed out is analyzed two class data points using the similarity measurements flow function that similarity function (formula 7) defines, takes α
=0.1, calculating measures that the results are shown in Table 1.
Wherein dist is distance function, and when two data samples are similar, dist levels off to 0, Lp=1;Otherwise Lp is approached
In 0.
The similarity measurement of 1 autocoder result of table compares (α=0.1)
Dist calculates the average distance calculated between same class data using Euclidean distance method in this experiment.It can from table
To find out, the Lp value of normal data points is much larger than the Lp value of exceptional data point, show that normal users similarity degree is high, shows
Distribution is more assembled, and apart from each other between abnormal electricity consumption behavior user, shows that data dispersion is larger.Meanwhile comparing autocoding
Device and depth self-encoding encoder pretreated model, it can be seen that the normal user data Lp value that depth self-encoding encoder model training goes out
More more assemble greatly, and abnormal user data Lp value is smaller more dispersed.Therefore, based on depth self-encoding encoder in the part Experiment
Preprocess method compares traditional autocoder, and being applied to effect performance in multiplexing electric abnormality Data Detection can be more preferable.
Data prediction based on principal component analytical method: it is pre- that the data based on linear PCA are done to the data set after cleaning
Processing.Obtained principal component is arranged by descending order, and is chosen the corresponding feature space of preceding 32 principal components and calculated newly
Characteristic dimension, in order to comparative analysis.
The step is established respectively pre-processes initial data based on linear PCA Method of Data with Adding Windows, selects after pretreatment
Former data, are mapped to the new feature space of 32 dimensions by the corresponding feature vector of preceding 32 main compositions.Select preceding 32 main compositions
Purpose is that the result of all preprocess methods is unified to the same dimension.
Wherein white point is represented without multiplexing electric abnormality suspicion user, and red point represents the user for determining and having multiplexing electric abnormality behavior.
Firstly, had the tendency that as we can see from the figure based on the pretreated data of PCA from a certain accumulation point to external diffusion, and
Comparatively, all opposite aggregation of white data point, red data point all relatively more disperse.Then, it is pre-processed from based on PCA
From the point of view of figure afterwards, white data point still has major part to be overlapped with red data point, it is seen that the preprocess method is to two class numbers
According to division effect it is unobvious.
Two class data points are analyzed using the similarity measurements flow function that formula (7) defines, take α=0.03, and calculating measures knot
Fruit such as following table.
The similarity measurement of 2 principal component analysis result of table compares (α=0.03)
Dist calculates the average distance still calculated using Euclidean distance method between same class data in this experiment.From table
In as can be seen that the method based on PCA reached good effect.
Isolated forest model is established: the new data set obtained to above-mentioned four kinds of data prediction modes carries out two dimension can
Show depending on changing, compares the effect of different pretreatments method.
Next, for four kinds of data preprocessing methods used by isolated forest model, the correspondence finally obtained is obscured
Matrix, Precision-Recall index and its P-R curve graph are respectively as shown in table 3, table 4.
The confusion matrix result of forest model is isolated under 3 different pretreatments method of table
Table 4 is directed to the Precision-Recall index and overall precision of abnormal data
Firstly, as can be seen that different pre- from the confusion matrix and Precision-Recall index result tested above
The abnormality detection model based on isolated forest has all reached higher overall precision under processing model.Meanwhile different data are pre-
Processing method selects also have difference to the influence of the detection effect of model.Observation has the user data detection feelings of abnormal electricity consumption behavior
Condition, it can be found that the model abnormality detection Precision value based on depth self-encoding encoder and Recall value ratio are based on autocoding
The index of device method high 0.07 and 0.14 is better than autocoder method in effect.And the pretreatment side based on linear PCA
Method is also more preferable than autocoder method in the performance boost to model abnormality detection, Precision value and Recall value
High 0.05 and 0.04, but it is big to the performance boost of model abnormality detection not as good as depth self-encoding encoder.
Embodiment 2:
The present embodiment and the difference of embodiment 1 are only that: the present embodiment is on the basis of embodiment 1 only to pretreated model
Changed, the present embodiment uses autocoder.
Firstly, establishing a traditional single hidden layer autocoder model, it is a full Connection Neural Network, such as Fig. 2
It is shown.
In Fig. 2, the first half of model is as autocoding part, and latter half is as automatic decoding part.The model
It will be used as simultaneously and output and input from cleaned 334 characteristic dimensions of initial data, is i.e. the neuron number of input layer
As being with the neuron number of output layer.Here the node number of middle layer is set as 32, is less than input layer and output
The node number of layer, plays the role of data compression.
Next, configuring relevant parameter for autocoder model.Wherein the middle layer activation primitive of network uses ReLU
Activation primitive, ReLU activation primitive figure is as shown in figure 3, its basic mathematical form is as follows:
F (x)=max (0, wTx+b) (5.1)
For nonlinear function, compared to traditional sigmoid activation primitive, ReLU first is due to non-negative section
Gradient is constant, therefore applies and gradient disappearance and gradient explosion issues are not present in depth network, so that the convergence speed of model
Degree maintains a stable state.Then, ReLU only needs a threshold value to can be obtained by activation value, calculates one greatly without spending
The operation of heap complexity, simplifies calculating process.
The training majorized function of model uses adadelta gradient decreasing function, it is a kind of optimization that learning rate is adaptive
Method can have faster convergence rate when training depth complex network.The specific calculating process Fig. 4 institute of its algorithm
Show.
It is binary_crossentropy to the loss function that model selects, i.e. logarithm loss function is mainly used to do pole
Maximum-likelihood estimation, its calculation formula is as shown in 4.2.Last set trained the number of iterations is 100 times.
L (Y, P (Y | X))=- logP (Y | X) (5.2)
The software algorithm is realized as shown in Figure 6.
Embodiment 3:
The present embodiment the difference from example 2 is that: the present embodiment only on the basis of embodiment 2 to autocoder increase
A hidden layer is added.
Previous autocoder data processing model only establishes single hidden layer, this time establishes to pending data
For one deeper from encoding model, network structure is as shown in Figure 5:
Basic configuration parameter is identical as previous model configuration, and the training majorized function of allocation models is adadelta, damage
Mistake function is binary_crossentropy, and frequency of training is 100 times, and intermediate coding layer and decoding layer activation primitive use
ReLU activation primitive.The software algorithm is as shown in Figure 7.
Finally, it is stated that the above examples are only used to illustrate the technical scheme of the present invention and are not limiting, although referring to compared with
Good embodiment describes the invention in detail, those skilled in the art should understand that, it can be to skill of the invention
Art scheme is modified or replaced equivalently, and without departing from the objective and range of technical solution of the present invention, should all be covered at this
In the scope of the claims of invention.
Claims (5)
1. a kind of user power utilization anomaly detection method based on isolated forest, which comprises the steps of:
S1, electricity consumption time series data is obtained by data acquisition modes;
S2, data are cleaned, incomplete data, wrong data, repeated data is removed;
S3, the feature extraction based on statistics:
S31, data definition: S311, data set is enabled to be X={ xn, n takes 1 to N, includes N number of daily electricity consumption user in data set, often
A user is divided into D days, M months, the electricity consumption data in Q season;The daily power consumption sequence of S312, each user: xn={ xnd,
D takes 1 to D;The moon electricity consumption sequence of S313, each user: yn={ ynm, m takes 1 to M,S114, Mei Geyong
The season electricity consumption sequence at family: zn={ znq, q takes 1 to Q,
S32, user power utilization behavioural characteristic is divided as unit of year, season, month in time, and calculates the list of each user
Position time average, standard deviation and coefficient of dispersion sequence are to calculate: Urban Annual Electrical Power Consumption amount standard deviation D1, Mei Geyong of each user
The Urban Annual Electrical Power Consumption amount coefficient of dispersion D2 at family, quarterly electricity consumption standard deviation D3~D6, quarterly electricity consumption coefficient of dispersion D7~
Under D10, monthly electricity consumption standard deviation D11~D21, monthly electricity consumption coefficient of dispersion D22~D32, every monthly average electricity consumption rise
Drop trend D33~D41, maximum value D42~D43 of the difference of adjacent two months electricity consumption mean values and ratio, adjacent two months electricity consumption mean values it
Minimum value D44~D45 of difference and ratio, the difference of adjacent season electricity consumption mean value and maximum value D46~D47, the adjacent season of ratio
The difference of electricity consumption mean value and minimum value D48~D49 of ratio, wherein D1~D49 is statistical nature;
S4, data prediction: assuming that initial data to be used to after the mode based on statistical nature is handled be formed with the M of n-dimensional vector
A sample value, M indicate user's number, and N indicates the number for the statistical nature that each user extracts, and enable the square that it is a M × N
X in battle array X, matrix XmnIndicate the occurrence of m-th user n-th statistical nature;Matrix X is reduced to M × K by pretreated model
Matrix YM×K, K < N;
S5, judge that multiplexing electric abnormality or electricity consumption are normal using isolated forest model:
S51, from new matrix YM×KMiddle extraction, each user extracts ψ statistical nature, if the quantity t, y of iTree treeijIt is new matrix
YM×KIn the i-th row jth column element;
S52, detection process be exactly to allow the statistical characteristics y of each userijEach iTree tree is traversed, traversal is then calculated
Y in the processijBy the path length h (y of every iTree treeij), y is finally calculated according to all path lengthsijException
Score value s (yij, ψ), calculation formula are as follows:
Wherein, c (ψ) is used to calculate the average path length of binary search tree, and effect is that result is normalized;H(ψ)
Calculation be:γ is Euler's constant;E (h (yij)) is yij all iTree in isolated forest
The average path length of tree;
S53, judge s (yij, ψ) whether it is less than 1- Δ e, Δ e is the constant in 0.22~0.07 range;If so, different for electricity consumption
Often;If it is not, then electricity consumption is normal.
2. a kind of user power utilization anomaly detection method based on isolated forest according to claim 1, feature exist
In the acquisition step of isolated forest model includes:
S711, assume that raw data set is indicated with F, randomly choose a sample point of F ' as subsample from data set and be put into tree
Root node,
One S712, random selection dimension q, are randomly generated a cut-point p in present node data, this cut-point p is produced
It is born in present node data between the maximum value and minimum value of specified dimension q;
S713, a hyperplane is generated with this cut-point p, present node data space is then divided into 2 sub-spaces: referring to
The data for determining q<p in dimension are put into the left subtree Fl of present node, the data of q>=p are put into the right subtree Fr of present node;
S714, the recursion step S712 and S713 in child node, constantly construct new children tree nodes, until only having in children tree nodes
One data or children tree nodes arrived restriction height, do not continue to divide, obtain t iTree tree.
3. a kind of user power utilization anomaly detection method based on isolated forest according to claim 1, feature exist
In in step s 4, pretreated model is autocoder, deep layer autocoder or PCA dimension-reduction treatment.
4. a kind of user power utilization anomaly detection method based on isolated forest according to claim 3, feature exist
In also being followed the steps below after step S12:
S13, electricity consumption tendency is divided into three kinds of alteration trend, fluctuation tendency and lifting trend trend types;
S14, alteration trend, fluctuation tendency and lifting trend are calculated:
S141, fluctuation tendency: the possible variation of assessment sequence is used in statistics Plays difference or degree of fluctuation, standard deviation are got over
Greatly, the range of numerical fluctuations is bigger;So it is special come the fluctuation tendency for indicating electricity consumption data to calculate electricity consumption standard deviation std here
Sign;Meanwhile electricity consumption coefficient of dispersion cv is calculated to measure the dispersion degree of user power utilization, enabling certain time period electricity consumption average value is μ,
Then:
Electricity consumption standard deviation:
Electricity consumption coefficient of dispersion:
Cv=std/ μ (2.2)
S142, alteration trend: alteration trend feature refers to the front and back Diversity measure of user power consumption, i.e., by certain time period with
The average electricity consumption of previous time adjacent segments compares, and difference and ratio reflect speed degree that electricity consumption changes, definition meter
Calculation mode is as follows:
The difference of the adjacent k month or k season electricity consumption mean value:
The ratio of the adjacent k month or k season electricity consumption mean value:
S143, lifting trend: lifting trend feature refers to by making electricity consumption next time according to continuous several days electricity consumptions of user
The prediction of amount, and compared with practical electricity consumption next time, obtain a possibility that rising or falling;Used here as simple rolling average
Method determines the feature vector of lifting trend;The simple method of moving average elapses item by item according to time series, successively calculates fixterm
A several cell means, and as predicted value next time;Enable k for mobile item number, t moment actual value is xnt, then lifting trend
The calculation method of feature:
T moment predicted value:
Ft=(xn(t-1)+xn(t-2)+…+xn(t-k))/k (2.5)
T moment lifting trend:
Tr=xnt-Ft (2.6)
If tr < 0, show that electricity consumption trend declines;If tr > 0, electricity consumption trend rises;
Wherein, the difference avg of electric standard difference std, electric coefficient of dispersion cv, the adjacent k month or k season electricity consumption mean valuea, adjacent k month or
The ratio avg of k season electricity consumption mean valueb, t moment lifting trend tr be statistical characteristics.
5. a kind of user power utilization anomaly detection method based on isolated forest according to claim 4, feature exist
In steps are as follows for PCA dimensionality reduction in step s 2:
S21, the mean value that each column of X are subtracted to the corresponding column, i.e., carry out zero averaging for every a line feature of data X, obtain
X ':
S22, X ' covariance matrix C, vector x are calculatediAnd xjCovariance, in (3.1) formula,
S23, the N number of eigenvalue λ for finding out covariance matrix C and the corresponding feature vector V of each eigenvalue λ:
CV=λ V (3.2)
S24, by all eigenvalue λs according to being arranged in a queue { λ from big to small1..., λi..., λN, according to characteristic value from big
To the small matrix W that feature vector V is rearranged to a N*N, the element of the i-th column is ith feature value λ in queue in matrix Wi
The element of corresponding feature vector V, and the corresponding feature vector of preceding K characteristic value is taken from matrix W, obtain the square of a N × K
Battle array AN×K;
S25, K is calculated according to formula 3.3, takes first K value for meeting 3.3 formulas:
S26, calculation formula 3.4, wherein YM×KNew feature data as after dimensionality reduction to k dimension;
YM×K=XM×NAN×K (3.4)。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811151326.9A CN109308306B (en) | 2018-09-29 | 2018-09-29 | User power consumption abnormal behavior detection method based on isolated forest |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811151326.9A CN109308306B (en) | 2018-09-29 | 2018-09-29 | User power consumption abnormal behavior detection method based on isolated forest |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109308306A true CN109308306A (en) | 2019-02-05 |
CN109308306B CN109308306B (en) | 2021-07-06 |
Family
ID=65224976
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811151326.9A Active CN109308306B (en) | 2018-09-29 | 2018-09-29 | User power consumption abnormal behavior detection method based on isolated forest |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109308306B (en) |
Cited By (36)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109948738A (en) * | 2019-04-11 | 2019-06-28 | 合肥工业大学 | Energy consumption method for detecting abnormality, the apparatus and system of coating drying room |
CN109977107A (en) * | 2019-04-02 | 2019-07-05 | 电子科技大学 | A kind of electricity consumption acquisition data cleaning method |
CN110135614A (en) * | 2019-03-26 | 2019-08-16 | 广东工业大学 | It is a kind of to be tripped prediction technique based on rejecting outliers and the 10kV distribution low-voltage of sampling techniques |
CN110149258A (en) * | 2019-04-12 | 2019-08-20 | 北京航空航天大学 | A kind of automobile CAN-bus network data method for detecting abnormality based on isolated forest |
CN110188799A (en) * | 2019-04-29 | 2019-08-30 | 国网上海市电力公司 | A kind of continuous data multidimensional analysis and intelligent O&M method |
CN110243599A (en) * | 2019-07-02 | 2019-09-17 | 西南交通大学 | Multidimensional peels off train EMU axle box bearing temperature anomaly state monitoring method |
CN110288383A (en) * | 2019-05-31 | 2019-09-27 | 国网上海市电力公司 | Group behavior power distribution network multiplexing electric abnormality detection method based on user property label |
CN110502883A (en) * | 2019-08-23 | 2019-11-26 | 四川长虹电器股份有限公司 | A kind of keystroke abnormal behavior detection method based on PCA |
CN110929643A (en) * | 2019-11-21 | 2020-03-27 | 西北工业大学 | Hyperspectral anomaly detection method based on multiple features and isolated trees |
CN111008662A (en) * | 2019-12-04 | 2020-04-14 | 贵州电网有限责任公司 | Online monitoring data anomaly analysis method for power transmission line |
CN111062590A (en) * | 2019-12-02 | 2020-04-24 | 深圳供电局有限公司 | Electricity abnormal behavior detection method and device, computer equipment and storage medium |
CN111160647A (en) * | 2019-12-30 | 2020-05-15 | 第四范式(北京)技术有限公司 | Money laundering behavior prediction method and device |
CN111275576A (en) * | 2020-01-19 | 2020-06-12 | 烟台海颐软件股份有限公司 | Identification method and identification system for abnormal electricity price execution user |
CN111612037A (en) * | 2020-04-24 | 2020-09-01 | 平安直通咨询有限公司上海分公司 | Abnormal user detection method, device, medium and electronic equipment |
CN111767951A (en) * | 2020-06-29 | 2020-10-13 | 上海积成能源科技有限公司 | Method for discovering abnormal data by applying isolated forest algorithm in residential electricity safety analysis |
CN111931834A (en) * | 2020-07-31 | 2020-11-13 | 广东工业大学 | Method, equipment and storage medium for detecting abnormal flow data in aluminum profile extrusion process based on isolated forest algorithm |
WO2020237540A1 (en) * | 2019-05-29 | 2020-12-03 | 西门子股份公司 | Power grid user classification method and device, and computer-readable storage medium |
CN112561251A (en) * | 2020-11-30 | 2021-03-26 | 广东电网有限责任公司广州供电局 | Power distribution network abnormal point detection method and device, computer equipment and storage medium |
CN112633412A (en) * | 2021-01-05 | 2021-04-09 | 南方电网深圳数字电网研究院有限公司 | Abnormal electricity consumption detection method, equipment and storage medium |
CN112668614A (en) * | 2020-12-11 | 2021-04-16 | 浙江成功软件开发有限公司 | Anti-money laundering studying and judging method |
CN112821556A (en) * | 2021-01-19 | 2021-05-18 | 许磊 | Power detection control system and method |
CN112906744A (en) * | 2021-01-20 | 2021-06-04 | 湖北工业大学 | Fault single battery identification method based on isolated forest algorithm |
CN113128567A (en) * | 2021-03-25 | 2021-07-16 | 云南电网有限责任公司 | Abnormal electricity consumption behavior identification method based on electricity consumption data |
CN113420816A (en) * | 2021-06-24 | 2021-09-21 | 北京市生态环境监测中心 | Data abnormal value determination method for full-spectrum water quality monitoring equipment |
CN113469235A (en) * | 2021-06-24 | 2021-10-01 | 珠海卓邦科技有限公司 | Water fluctuation abnormality identification method and device, computer device and storage medium |
CN113496440A (en) * | 2021-06-28 | 2021-10-12 | 国网上海市电力公司 | User abnormal electricity utilization detection method and system |
CN114495137A (en) * | 2022-04-15 | 2022-05-13 | 深圳高灯计算机科技有限公司 | Bill abnormity detection model generation method and bill abnormity detection method |
CN114755002A (en) * | 2022-04-06 | 2022-07-15 | 燕山大学 | Buffer balance valve fault diagnosis method based on full-connection neural network |
CN115099291A (en) * | 2022-08-29 | 2022-09-23 | 同方德诚(山东)科技股份公司 | Building energy-saving monitoring method |
CN116304962A (en) * | 2023-05-25 | 2023-06-23 | 湖南东润智能仪表有限公司 | Intelligent anomaly monitoring method for water meter metering data |
CN116611000A (en) * | 2023-07-17 | 2023-08-18 | 东营市恒盛农业科技有限公司 | Intelligent hairy crab culture environment monitoring system based on machine learning |
CN116645097A (en) * | 2023-03-30 | 2023-08-25 | 广东盛迪嘉电子商务股份有限公司 | Payment clearing platform monitoring and early warning system |
CN117312997A (en) * | 2023-11-21 | 2023-12-29 | 乾程电力有限公司 | Intelligent diagnosis method and system for power management system |
CN117349764A (en) * | 2023-12-05 | 2024-01-05 | 河北三臧生物科技有限公司 | Intelligent analysis method for stem cell induction data |
CN117786587A (en) * | 2024-02-28 | 2024-03-29 | 深圳市福山自动化科技有限公司 | Power grid data quality abnormality diagnosis method based on data analysis |
CN113469235B (en) * | 2021-06-24 | 2024-04-26 | 珠海卓邦科技有限公司 | Water fluctuation abnormality recognition method and device, computer device and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107657288A (en) * | 2017-10-26 | 2018-02-02 | 国网冀北电力有限公司 | A kind of power scheduling flow data method for detecting abnormality based on isolated forest algorithm |
US10045218B1 (en) * | 2016-07-27 | 2018-08-07 | Argyle Data, Inc. | Anomaly detection in streaming telephone network data |
CN108494747A (en) * | 2018-03-08 | 2018-09-04 | 上海观安信息技术股份有限公司 | Traffic anomaly detection method, electronic equipment and computer program product |
-
2018
- 2018-09-29 CN CN201811151326.9A patent/CN109308306B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10045218B1 (en) * | 2016-07-27 | 2018-08-07 | Argyle Data, Inc. | Anomaly detection in streaming telephone network data |
CN107657288A (en) * | 2017-10-26 | 2018-02-02 | 国网冀北电力有限公司 | A kind of power scheduling flow data method for detecting abnormality based on isolated forest algorithm |
CN108494747A (en) * | 2018-03-08 | 2018-09-04 | 上海观安信息技术股份有限公司 | Traffic anomaly detection method, electronic equipment and computer program product |
Non-Patent Citations (3)
Title |
---|
DONG XU等: "An Improved Data Anomaly Detection Method Based on Isolation Forest", 《IEEE》 * |
FEI TONY LIU等: "Isolation Forest", 《IEEE》 * |
张荣昌: "基于数据挖掘的用电数据异常的分析与研宄", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (52)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110135614A (en) * | 2019-03-26 | 2019-08-16 | 广东工业大学 | It is a kind of to be tripped prediction technique based on rejecting outliers and the 10kV distribution low-voltage of sampling techniques |
CN109977107B (en) * | 2019-04-02 | 2022-04-05 | 电子科技大学 | Method for cleaning power utilization collected data |
CN109977107A (en) * | 2019-04-02 | 2019-07-05 | 电子科技大学 | A kind of electricity consumption acquisition data cleaning method |
CN109948738A (en) * | 2019-04-11 | 2019-06-28 | 合肥工业大学 | Energy consumption method for detecting abnormality, the apparatus and system of coating drying room |
CN110149258A (en) * | 2019-04-12 | 2019-08-20 | 北京航空航天大学 | A kind of automobile CAN-bus network data method for detecting abnormality based on isolated forest |
CN110188799A (en) * | 2019-04-29 | 2019-08-30 | 国网上海市电力公司 | A kind of continuous data multidimensional analysis and intelligent O&M method |
WO2020237540A1 (en) * | 2019-05-29 | 2020-12-03 | 西门子股份公司 | Power grid user classification method and device, and computer-readable storage medium |
CN110288383A (en) * | 2019-05-31 | 2019-09-27 | 国网上海市电力公司 | Group behavior power distribution network multiplexing electric abnormality detection method based on user property label |
CN110288383B (en) * | 2019-05-31 | 2024-02-02 | 国网上海市电力公司 | Group behavior power distribution network electricity utilization abnormality detection method based on user attribute tags |
CN110243599A (en) * | 2019-07-02 | 2019-09-17 | 西南交通大学 | Multidimensional peels off train EMU axle box bearing temperature anomaly state monitoring method |
CN110243599B (en) * | 2019-07-02 | 2020-05-05 | 西南交通大学 | Method for monitoring temperature abnormal state of multi-dimensional outlier train motor train unit axle box bearing |
CN110502883A (en) * | 2019-08-23 | 2019-11-26 | 四川长虹电器股份有限公司 | A kind of keystroke abnormal behavior detection method based on PCA |
CN110502883B (en) * | 2019-08-23 | 2022-08-19 | 四川长虹电器股份有限公司 | PCA-based keystroke behavior anomaly detection method |
CN110929643A (en) * | 2019-11-21 | 2020-03-27 | 西北工业大学 | Hyperspectral anomaly detection method based on multiple features and isolated trees |
CN110929643B (en) * | 2019-11-21 | 2022-04-26 | 西北工业大学 | Hyperspectral anomaly detection method based on multiple features and isolated trees |
CN111062590A (en) * | 2019-12-02 | 2020-04-24 | 深圳供电局有限公司 | Electricity abnormal behavior detection method and device, computer equipment and storage medium |
CN111008662A (en) * | 2019-12-04 | 2020-04-14 | 贵州电网有限责任公司 | Online monitoring data anomaly analysis method for power transmission line |
CN111160647A (en) * | 2019-12-30 | 2020-05-15 | 第四范式(北京)技术有限公司 | Money laundering behavior prediction method and device |
CN111160647B (en) * | 2019-12-30 | 2023-08-22 | 第四范式(北京)技术有限公司 | Money laundering behavior prediction method and device |
CN111275576A (en) * | 2020-01-19 | 2020-06-12 | 烟台海颐软件股份有限公司 | Identification method and identification system for abnormal electricity price execution user |
CN111612037A (en) * | 2020-04-24 | 2020-09-01 | 平安直通咨询有限公司上海分公司 | Abnormal user detection method, device, medium and electronic equipment |
CN111767951A (en) * | 2020-06-29 | 2020-10-13 | 上海积成能源科技有限公司 | Method for discovering abnormal data by applying isolated forest algorithm in residential electricity safety analysis |
CN111931834A (en) * | 2020-07-31 | 2020-11-13 | 广东工业大学 | Method, equipment and storage medium for detecting abnormal flow data in aluminum profile extrusion process based on isolated forest algorithm |
CN112561251A (en) * | 2020-11-30 | 2021-03-26 | 广东电网有限责任公司广州供电局 | Power distribution network abnormal point detection method and device, computer equipment and storage medium |
CN112668614A (en) * | 2020-12-11 | 2021-04-16 | 浙江成功软件开发有限公司 | Anti-money laundering studying and judging method |
CN112668614B (en) * | 2020-12-11 | 2022-11-01 | 浙江成功软件开发有限公司 | Anti-money laundering studying and judging method |
CN112633412A (en) * | 2021-01-05 | 2021-04-09 | 南方电网深圳数字电网研究院有限公司 | Abnormal electricity consumption detection method, equipment and storage medium |
CN112821556A (en) * | 2021-01-19 | 2021-05-18 | 许磊 | Power detection control system and method |
CN112821556B (en) * | 2021-01-19 | 2023-04-07 | 深圳市迅捷光通科技有限公司 | Power detection control system and method |
CN112906744A (en) * | 2021-01-20 | 2021-06-04 | 湖北工业大学 | Fault single battery identification method based on isolated forest algorithm |
CN112906744B (en) * | 2021-01-20 | 2023-08-04 | 湖北工业大学 | Fault single battery identification method based on isolated forest algorithm |
CN113128567A (en) * | 2021-03-25 | 2021-07-16 | 云南电网有限责任公司 | Abnormal electricity consumption behavior identification method based on electricity consumption data |
CN113469235A (en) * | 2021-06-24 | 2021-10-01 | 珠海卓邦科技有限公司 | Water fluctuation abnormality identification method and device, computer device and storage medium |
CN113469235B (en) * | 2021-06-24 | 2024-04-26 | 珠海卓邦科技有限公司 | Water fluctuation abnormality recognition method and device, computer device and storage medium |
CN113420816A (en) * | 2021-06-24 | 2021-09-21 | 北京市生态环境监测中心 | Data abnormal value determination method for full-spectrum water quality monitoring equipment |
CN113496440B (en) * | 2021-06-28 | 2023-12-12 | 国网上海市电力公司 | User abnormal electricity consumption detection method and system |
CN113496440A (en) * | 2021-06-28 | 2021-10-12 | 国网上海市电力公司 | User abnormal electricity utilization detection method and system |
CN114755002A (en) * | 2022-04-06 | 2022-07-15 | 燕山大学 | Buffer balance valve fault diagnosis method based on full-connection neural network |
CN114495137B (en) * | 2022-04-15 | 2022-08-02 | 深圳高灯计算机科技有限公司 | Bill abnormity detection model generation method and bill abnormity detection method |
CN114495137A (en) * | 2022-04-15 | 2022-05-13 | 深圳高灯计算机科技有限公司 | Bill abnormity detection model generation method and bill abnormity detection method |
CN115099291B (en) * | 2022-08-29 | 2022-11-11 | 同方德诚(山东)科技股份公司 | Building energy-saving monitoring method |
CN115099291A (en) * | 2022-08-29 | 2022-09-23 | 同方德诚(山东)科技股份公司 | Building energy-saving monitoring method |
CN116645097A (en) * | 2023-03-30 | 2023-08-25 | 广东盛迪嘉电子商务股份有限公司 | Payment clearing platform monitoring and early warning system |
CN116304962A (en) * | 2023-05-25 | 2023-06-23 | 湖南东润智能仪表有限公司 | Intelligent anomaly monitoring method for water meter metering data |
CN116304962B (en) * | 2023-05-25 | 2023-08-04 | 湖南东润智能仪表有限公司 | Intelligent anomaly monitoring method for water meter metering data |
CN116611000B (en) * | 2023-07-17 | 2023-10-24 | 东营市恒盛农业科技有限公司 | Intelligent hairy crab culture environment monitoring system based on machine learning |
CN116611000A (en) * | 2023-07-17 | 2023-08-18 | 东营市恒盛农业科技有限公司 | Intelligent hairy crab culture environment monitoring system based on machine learning |
CN117312997A (en) * | 2023-11-21 | 2023-12-29 | 乾程电力有限公司 | Intelligent diagnosis method and system for power management system |
CN117312997B (en) * | 2023-11-21 | 2024-03-08 | 乾程电力有限公司 | Intelligent diagnosis method and system for power management system |
CN117349764A (en) * | 2023-12-05 | 2024-01-05 | 河北三臧生物科技有限公司 | Intelligent analysis method for stem cell induction data |
CN117349764B (en) * | 2023-12-05 | 2024-02-27 | 河北三臧生物科技有限公司 | Intelligent analysis method for stem cell induction data |
CN117786587A (en) * | 2024-02-28 | 2024-03-29 | 深圳市福山自动化科技有限公司 | Power grid data quality abnormality diagnosis method based on data analysis |
Also Published As
Publication number | Publication date |
---|---|
CN109308306B (en) | 2021-07-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109308306A (en) | A kind of user power utilization anomaly detection method based on isolated forest | |
CN109377409A (en) | A kind of user power utilization anomaly detection method based on BP neural network | |
CN103135009B (en) | Electric appliance detection method and system based on user feedback information | |
CN109142946A (en) | Transformer fault detection method based on ant group algorithm optimization random forest | |
CN108616120A (en) | A kind of non-intrusive electrical load decomposition method based on RBF neural | |
US20130138669A1 (en) | System and method employing a hierarchical load feature database to identify electric load types of different electric loads | |
CN109214503B (en) | Power transmission and transformation project cost prediction method based on KPCA-LA-RBM | |
Bidoki et al. | Evaluating different clustering techniques for electricity customer classification | |
Shchetinin | Cluster-based energy consumption forecasting in smart grids | |
Justo et al. | Behavioral similarity of residential customers using a neural network based on adaptive resonance theory | |
CN110321493A (en) | A kind of abnormality detection of social networks and optimization method, system and computer equipment | |
CN111563827A (en) | Load decomposition method based on electrical appliance physical characteristics and residential electricity consumption behaviors | |
CN112488192A (en) | Regional power grid short-term load prediction method | |
CN114118269A (en) | Energy big data aggregation analysis method based on typical service scene | |
Kotanchek et al. | Symbolic regression via genetic programming as a discovery engine: Insights on outliers and prototypes | |
CN111027841A (en) | Low-voltage transformer area line loss calculation method based on gradient lifting decision tree | |
CN116842459B (en) | Electric energy metering fault diagnosis method and diagnosis terminal based on small sample learning | |
CN116819423A (en) | Method and system for detecting abnormal running state of gateway electric energy metering device | |
Wang et al. | Mutual information-weighted principle components identified from the depth features of stacked autoencoders and original variables for oil dry point soft sensor | |
CN110059752A (en) | A kind of statistical learning querying method based on comentropy Sampling Estimation | |
Li et al. | Kennard-Stone balance algorithm for time-series big data stream mining | |
CN110852628B (en) | Rural medium-long term load prediction method considering development mode influence | |
CN114240102A (en) | Line loss abnormal data identification method and device, electronic equipment and storage medium | |
de Almeida et al. | A case study on environmental sustainability: A study of the trophic changes in fish species as a result of the damming of rivers through clustering analysis | |
Liao et al. | Reducing Annotation Efforts in Electricity Theft Detection through Optimal Sample Selection |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |