CN117422590A - Construction monitoring method, device, equipment and medium based on load decomposition - Google Patents
Construction monitoring method, device, equipment and medium based on load decomposition Download PDFInfo
- Publication number
- CN117422590A CN117422590A CN202311415380.0A CN202311415380A CN117422590A CN 117422590 A CN117422590 A CN 117422590A CN 202311415380 A CN202311415380 A CN 202311415380A CN 117422590 A CN117422590 A CN 117422590A
- Authority
- CN
- China
- Prior art keywords
- power
- load
- data
- model
- equipment
- 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
- 238000012544 monitoring process Methods 0.000 title claims abstract description 72
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 68
- 238000010276 construction Methods 0.000 title claims abstract description 57
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000012549 training Methods 0.000 claims abstract description 31
- 239000013598 vector Substances 0.000 claims description 64
- 238000004590 computer program Methods 0.000 claims description 17
- 230000007246 mechanism Effects 0.000 claims description 15
- 238000010606 normalization Methods 0.000 claims description 6
- 238000003860 storage Methods 0.000 claims description 6
- 238000012806 monitoring device Methods 0.000 claims description 4
- 238000011176 pooling Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 abstract description 8
- 230000005856 abnormality Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 8
- 239000011159 matrix material Substances 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000005070 sampling Methods 0.000 description 6
- 230000005611 electricity Effects 0.000 description 5
- 238000009435 building construction Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/08—Construction
-
- 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
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Economics (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Primary Health Care (AREA)
- Marketing (AREA)
- Human Resources & Organizations (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention belongs to the technical field of construction monitoring, and particularly discloses a construction monitoring method, device, equipment and medium based on load decomposition. Acquiring load power data and influence factor data in a preset area; extracting features according to the load power data to obtain equipment power features and time features of each equipment, and establishing a power feature model of each equipment according to the equipment power features and time features of each equipment; training a CAF non-invasive load decomposition model according to the load power data and the influence factor data, and obtaining a monitoring result according to the CAF non-invasive load decomposition model; and comparing the monitoring result with a power characteristic model of equipment corresponding to the monitoring result to obtain a construction state result. According to the invention, the CAF non-invasive load decomposition model is trained by using the load power data and the influence factor data, and the decomposed monitoring result is compared with the power characteristic model, so that whether the equipment works normally or not in the construction process is judged, and the construction abnormality is found in time.
Description
Technical Field
The invention belongs to the technical field of construction monitoring, and particularly relates to a construction monitoring method, device, equipment and medium based on load decomposition.
Background
With the development of internet technology and the increasing degree of social informatization, big data has penetrated into various industries of society, becoming increasingly bulky and complex. Therefore, how to stand on the large electric power data, enable the large electric power data to enable industry, develop the electricity consumption behavior analysis of building construction based on the common electric power data of building day, reflect the conditions of starting work, construction and shutdown with the electric power data, assist the work of 'building maintenance and stable folk life', prevent the delay of building maintenance and building rotting, and maintain the safety and stability of society, which is the current urgent problem to be solved.
Invasive load monitoring (Intrusive Load Monitoring, ILM) techniques require the installation of sensors for each consumer of the user, requiring significant costs and creating a potential safety hazard to the user's privacy. The Non-invasive load monitoring (Non-Intrusive Load Monitoring, NILM) technology refers to that only one sensor is installed at the user inlet, and the specific condition of each power consumption load is obtained by collecting and analyzing the total power consumption load of the user, so that the power consumption condition and the power consumption rule of each electric appliance of the user are known. This approach can greatly reduce the enormous economic costs associated with installing the sensing device and is therefore relatively widely applicable. Non-intrusive load splitting enables splitting of total power data into individual consumer power consumption.
In recent years, many researchers have explored non-invasive load-splitting techniques. Researchers have proposed a hierarchical Dirichlet hidden Markov model, which completes the decomposition of the unsupervised power, and has the advantages of simple and quick algorithm, but the defects also exist, namely the recognition error is obvious; the sparse Viterbi algorithm is provided, so that the time required by the HMM in load decomposition is greatly reduced, the algorithm is still imperfect, and the time required by the algorithm is still overlong when the number of power loads is continuously increased and reaches a certain value; the method can extract the total V-I curve of the power load and put the total V-I curve into the cells under the binary background, so that the characteristics of the curve graph can be effectively extracted, and the load decomposition is carried out. The above is a load-resolving operation by using a high-frequency sampling method. However, what is needed for high frequency sampling is an extremely accurate sampling device, and the amount of load data sampled is very large, limiting the widespread popularization of algorithms, both in terms of sampling cost and operational speed. For the load decomposition under the low-frequency sampling condition, researchers compare the decomposition results of a plurality of algorithms according to a multi-label classification algorithm, however, the method can only aim at high-power load, the decomposition result is not ideal, and the accuracy is low; it is proposed that based on the sum-to-k constraint, a non-negative matrix factorization is used to decompose the individual electrical load information from the total electrical load information on the basis of the constraint conditions applied, and even under low frequency sampling conditions, the accuracy of the load decomposition is still high, but the time required for training the sample data is high, because the training process needs to be restarted when the load type changes.
Based on non-invasive construction electricity load decomposition, daily electricity data and electric equipment monitoring analysis are carried out on a building site of a building, so that building construction conditions are truly reflected, and long-term shutdown and tail rot risks are analyzed. By research: on one hand, a normalized building supervision means can be provided for a building department, so that social responsibility of a company is reflected; on the other hand, the system can assist urban management and environmental protection departments to perform law enforcement inspection on building trays.
Disclosure of Invention
The invention aims to provide a construction monitoring method, device, equipment and medium based on load decomposition, which are used for solving the technical problems that the existing building construction process is difficult to monitor and construction abnormality cannot be found in time.
In order to achieve the above purpose, the invention adopts the following technical scheme:
in a first aspect, the present invention provides a construction monitoring method based on load decomposition, including the steps of:
acquiring load power data and influence factor data in a preset area;
extracting features according to the load power data to obtain equipment power features and time features of each equipment, and establishing a power feature model of each equipment according to the equipment power features and time features of each equipment;
training a CAF non-invasive load decomposition model according to the load power data and the influence factor data, and obtaining a monitoring result according to the CAF non-invasive load decomposition model;
and comparing the monitoring result with a power characteristic model of equipment corresponding to the monitoring result to obtain a construction state result.
The invention further improves that: the step of extracting the characteristics according to the load power data to obtain the power characteristics and the time characteristics of the equipment of each equipment and establishing the power characteristic model of each equipment according to the power characteristics and the time characteristics of the equipment of each equipment specifically comprises the following steps:
extracting power value feature vectors of all the devices according to the load power data;
extracting time feature vectors of all the devices according to the load power data;
and inputting the time feature vector and the power value feature vector into a full connection layer to obtain a power feature model.
The invention further improves that: the step of extracting the power value characteristic vector of each device according to the load power data specifically comprises the following steps:
generating an active power sequence of each device according to the load power data of each device;
inputting the active power sequence of each device into a convolution layer to obtain the power characteristic information of each device;
and inputting the active power sequences of the devices into an average pooling layer, and reserving power characteristic information to obtain a power value characteristic vector.
The invention further improves that: the step of training the CAF non-invasive load decomposition model according to the load power data and the influence factor data and obtaining the monitoring result according to the CAF non-invasive load decomposition model specifically comprises the following steps:
inputting the influence factor data into a CNN model, and extracting influence factor characteristics;
model training is carried out according to the influence factor characteristics and the load power data as the input of the CAF non-invasive load decomposition model, and a decoder result is obtained;
and inputting the decoder result into the full connection layer to obtain a monitoring result.
The invention further improves that: the step of training the model according to the influence factor characteristics and the load power data as the input of the CAF non-invasive load decomposition model to obtain the decoder result specifically comprises the following steps:
inputting the load power data into an Autoformer encoder, and calculating a query vector, a key vector and a value vector of the Autoformer encoder;
substituting values of the query vector and the key vector into a second layer autocorrelation mechanism of the Autoformer decoder, inputting the influence factor features into a first layer autocorrelation mechanism of the Autoformer decoder to obtain a second query vector, a second key vector and a second value vector, and introducing the second value vector into the second layer autocorrelation mechanism of the Autoformer decoder to obtain a decoder result.
The invention further improves that: the CAF non-invasive load decomposition model is screened according to the average absolute error, the accuracy, the precision, the recall rate and the F1 value.
The invention further improves that: the step of acquiring load power data and influence factor data in a preset area specifically comprises the following steps:
acquiring load power data and influence factor data in a preset area;
filling the missing values of the obtained load power data and influence factor data to form a complete load power data set and an influence factor data set;
zero-mean normalization is performed on the load power dataset.
In a second aspect, the present invention provides a construction monitoring device based on load decomposition, including:
and a data acquisition module: the method comprises the steps of acquiring load power data and influence factor data in a preset area;
the power characteristic model building module: the power characteristic model is used for extracting the characteristics according to the load power data to obtain the equipment power characteristics and the time characteristics of each equipment and establishing the power characteristic model of each equipment according to the equipment power characteristics and the time characteristics of each equipment;
model training module: the system is used for training a CAF non-invasive load decomposition model according to the load power data and the influence factor data, and obtaining a monitoring result according to the CAF non-invasive load decomposition model;
and a construction state monitoring module: and the power characteristic model is used for comparing the monitoring result with the power characteristic model of the equipment corresponding to the monitoring result to obtain a construction state result.
In a third aspect, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor implementing a load decomposition-based construction monitoring method as described above when executing the computer program.
In a fourth aspect, the present invention provides a computer readable storage medium storing a computer program which when executed by a processor implements a load decomposition based construction monitoring method as described above.
Compared with the prior art, the invention at least comprises the following beneficial effects:
(1) According to the invention, the CAF non-invasive load decomposition model is trained by using the load power data and the influence factor data, so that the load decomposition is realized, and the decomposed monitoring result is compared with the power characteristic model, so that whether the equipment works normally or not in the construction process is judged, and the construction abnormality is found in time.
(2) According to the invention, an Autoformer is selected, a sequence decomposition module breaks through a traditional method of taking time sequence decomposition as pretreatment, a sequence decomposition unit is designed to be embedded into a depth model, progressive prediction is realized, and components with stronger predictability are obtained gradually; the Auto-Correlation (Auto-Correlation) mechanism is a self-attention mechanism for dropping point-to-point (point-wise) connections based on random process theory, realizes the Auto-Correlation mechanism for sequence-level (series-wise) connections, has complexity, and breaks the information utilization bottleneck.
(3) Based on non-invasive construction electricity load decomposition, daily electricity data and electric equipment monitoring analysis are carried out on a building site of a building, so that building construction conditions are truly reflected, and long-term shutdown and tail rot risks are analyzed. By research: on one hand, a normalized building supervision means can be provided for a building department, so that social responsibility of a company is reflected; on the other hand, the system can assist urban management and environmental protection departments to perform law enforcement inspection on building trays.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention.
In the drawings:
FIG. 1 is a flow chart of a construction monitoring method based on load decomposition according to the present invention;
FIG. 2 is a schematic diagram of a CAF non-invasive load decomposition model in a load decomposition-based construction monitoring method according to the present invention;
fig. 3 is a block diagram of a construction monitoring device based on load decomposition according to the present invention.
Detailed Description
The invention will be described in detail below with reference to the drawings in connection with embodiments. It should be noted that, without conflict, the embodiments of the present invention and features of the embodiments may be combined with each other.
The following detailed description is exemplary and is intended to provide further details of the invention. Unless defined otherwise, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments in accordance with the invention.
Example 1
A construction monitoring method based on load decomposition, as shown in figure 1, comprises the following steps:
s1, acquiring load power data and influence factor data in a preset area;
specifically, in S1, the preset area is a construction site for performing construction monitoring;
specifically, in S1, the load power data includes total active power in a preset area and active power data of each electric device in the preset area;
the influence factor data comprise time factor data, weather factor data and the like;
the time factor data comprise working days, non-working days, limiting and stopping days, heavy pollution days and the like;
specifically, the step S1 includes the following steps:
s11, acquiring load power data and influence factor data in a preset area;
s12, filling the missing values of the obtained load power data and the influence factor data to form a complete load power data set and an influence factor data set;
s13, carrying out zero-mean normalization on the load power data set;
the normalization formula:
wherein x is the data in the load power data set, x 0 For load power data, μ is the power mean of the device and σ is the power standard deviation of the device.
S2, extracting features according to the load power data to obtain the power features and time features of the devices, and establishing a power feature model of each device according to the power features and time features of the devices;
specifically, the step S2 includes the following steps:
s21, extracting power value feature vectors of all the devices according to the load power data;
the power value feature vector is marked as F 1 (x)；
Specifically, S21 includes the following steps:
s211, generating an active power sequence of each device according to the load power data of each device;
s212, inputting the active power sequences of the devices into a convolution layer to obtain power characteristic information of the devices;
s213, inputting the active power sequences of the devices into an average pooling layer, and reserving power characteristic information to obtain a power value characteristic vector;
the power characteristic information extracted by the convolution layer is reserved through step S213, and meanwhile training parameters of the neural network are reduced, so that overfitting is avoided.
S22, extracting time feature vectors of all the devices according to the load power data;
the time feature vector is denoted as F 2 (x)；
Specifically, S22 includes the following steps:
s221, generating an active power sequence of each device according to the load power data of each device;
s222, randomly initializing a time matrix with the same size according to the active power sequences of the devices;
s223, iterating the time matrix along with model training, so as to obtain a time feature vector;
specifically, in S223, during the training process, the model continuously updates the weight, and the time matrix changes accordingly to the time information that can more accurately represent the real load sequence;
s23, inputting the time feature vector and the power value feature vector into a full connection layer to obtain a power feature model;
the power characteristic model is marked as F (x);
s3, training a CAF non-invasive load decomposition model according to the load power data and the influence factor data, and obtaining a monitoring result according to the CAF non-invasive load decomposition model;
a schematic of a CAF non-invasive load decomposition model is shown in fig. 2.
Specifically, the step S3 includes the following steps:
s31, inputting influence factor data into a CNN model, and extracting influence factor characteristics;
s32, performing model training according to the influence factor characteristics and the load power data as the input of a CAF non-invasive load decomposition model to obtain a decoder result;
specifically, the step S32 includes the steps of:
s321, inputting load power data into an Autoformer encoder, and calculating query vectors, key vectors and value vectors of the Autoformer encoder;
s322, substituting values of the query vector and the key vector into a second-layer autocorrelation mechanism of the Autoformer decoder, inputting the influence factor characteristics into a first-layer autocorrelation mechanism of the Autoformer decoder to obtain a second query vector, a second key vector and a second value vector, and introducing the second value vector into the second-layer autocorrelation mechanism of the Autoformer decoder to obtain a decoder result;
specifically, in step S32, optimization is performed by an AdamW optimizer, the training batch size is 64, the learning rate is 0.0001, the weight decay is 0.01, and the maximum training round number is 100. And reserving the optimal round of model as an optimal model.
Specifically, the optimal model is evaluated by using the average absolute error MAE, the accuracy Acc, the Precision, the Recall, and the F1 value, which are respectively:
wherein x is t Andrespectively decomposing power value and actual power value of a certain device at time T, wherein T is a period of time;
in the formula, TP is True (True Positive), the True value is the Positive class and is judged as the Positive class, FP is False Positive (False Positive), the True value is the Negative class and TN is True Negative (True Negative), the True value is the Negative class and is judged as the Negative class; FN False Negative (False Negative), the true value is the positive class and is determined to be the Negative class.
S33, inputting the decoder result into the full connection layer to obtain a monitoring result;
s4, comparing the monitoring result with a power characteristic model of equipment corresponding to the monitoring result to obtain a construction state result;
specifically, in S4, when the monitoring result is matched with the power feature model of the device corresponding to the monitoring result and the working time length accords with the construction operation time length arrangement, the construction can be considered to be performed normally.
For example, a power characteristic model of the tower crane is obtained in S2, a component of the tower crane appears in the monitoring result in S5, and the duration is 5 hours, so that normal construction can be considered.
Example 2
A construction monitoring device based on load decomposition, as shown in fig. 3, comprises the following steps:
and a data acquisition module: the method comprises the steps of acquiring load power data and influence factor data in a preset area;
specifically, in the data acquisition module, a preset area is a construction site for construction monitoring;
specifically, the load power data in the data acquisition module comprises total active power in a preset area and active power data of each electric equipment in the preset area;
the influence factor data comprise time factor data, weather factor data and the like;
the time factor data comprise working days, non-working days, limiting and stopping days, heavy pollution days and the like;
specifically, the data acquisition module includes the following steps:
acquiring load power data and influence factor data in a preset area;
filling the missing values of the obtained load power data and influence factor data to form a complete load power data set and an influence factor data set;
zero-mean normalization is performed on the load power data set;
the normalization formula:
wherein x is the data in the load power data set, x 0 For load power data, μ is the power mean of the device and σ is the power standard deviation of the device.
The power characteristic model building module: the power characteristic model is used for extracting the characteristics according to the load power data to obtain the equipment power characteristics and the time characteristics of each equipment and establishing the power characteristic model of each equipment according to the equipment power characteristics and the time characteristics of each equipment;
specifically, the power characteristic model building module comprises the following steps:
extracting power value feature vectors of all the devices according to the load power data;
the power value feature vector is marked as F 1 (x)；
Specifically, the method comprises the following steps:
generating an active power sequence of each device according to the load power data of each device;
inputting the active power sequence of each device into a convolution layer to obtain the power characteristic information of each device;
inputting the active power sequences of all the devices into an average pooling layer, and reserving power characteristic information to obtain a power value characteristic vector;
the power characteristic information extracted by the convolution layer is reserved, and meanwhile training parameters of the neural network are reduced, so that overfitting is avoided.
Extracting time feature vectors of all the devices according to the load power data;
the time feature vector is denoted as F 2 (x)；
Specifically, the method comprises the following steps:
generating an active power sequence of each device according to the load power data of each device;
randomly initializing a time matrix with the same size according to the active power sequences of the devices;
iterating the time matrix along with model training, so as to obtain a time feature vector;
specifically, during the training process, the model continuously updates the weight, and the time matrix changes to the time information which can more accurately represent the real load sequence;
inputting the time feature vector and the power value feature vector into a full connection layer to obtain a power feature model;
the power characteristic model is marked as F (x);
model training module: the system is used for training a CAF non-invasive load decomposition model according to the load power data and the influence factor data, and obtaining a monitoring result according to the CAF non-invasive load decomposition model;
a schematic of a CAF non-invasive load decomposition model is shown in fig. 2.
Specifically, the model training module comprises the following steps:
inputting the influence factor data into a CNN model, and extracting influence factor characteristics;
model training is carried out according to the influence factor characteristics and the load power data as the input of the CAF non-invasive load decomposition model, and a decoder result is obtained;
specifically, the method comprises the following steps:
inputting the load power data into an Autoformer encoder, and calculating a query vector, a key vector and a value vector of the Autoformer encoder;
substituting values of the query vector and the key vector into a second layer autocorrelation mechanism of the Autoformer decoder, inputting the influence factor features into a first layer autocorrelation mechanism of the Autoformer decoder to obtain a second query vector, a second key vector and a second value vector, and introducing the second value vector into the second layer autocorrelation mechanism of the Autoformer decoder to obtain a decoder result;
specifically, when the training batch size is 64, the learning rate is 0.0001, the weight attenuation is 0.01, and the maximum training round number is 100. And reserving the optimal round of model as an optimal model.
Specifically, the optimal model is evaluated by using the average absolute error MAE, the accuracy Acc, the Precision, the Recall, and the F1 value, which are respectively:
wherein x is t Andrespectively decomposing power value and actual power value of a certain device at time T, wherein T is a period of time;
in the formula, TP is True (True Positive), the True value is the Positive class and is judged as the Positive class, FP is False Positive (False Positive), the True value is the Negative class and TN is True Negative (True Negative), the True value is the Negative class and is judged as the Negative class; FN False Negative (False Negative), the true value is the positive class and is determined to be the Negative class.
Inputting the decoder result into the full connection layer to obtain a monitoring result;
and a construction state monitoring module: the power characteristic model is used for comparing the monitoring result with the power characteristic model of the equipment corresponding to the monitoring result to obtain a construction state result;
specifically, when the monitoring result is matched with the power characteristic model of the equipment corresponding to the monitoring result and the working time length accords with the construction operation time length arrangement, the construction can be considered to be performed normally.
Specifically, when the monitoring result is not matched with the power characteristic model of the equipment corresponding to the monitoring result, judging that the construction is abnormal, and giving an alarm;
and when the power characteristic model of the equipment corresponding to the monitoring result is matched, judging that the construction is normal.
Example 3
A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor implementing a load break-down based construction monitoring method of embodiment 1 when the computer program is executed by the processor.
Example 4
A computer-readable storage medium storing a computer program which, when executed by a processor, implements a load-break-down-based construction monitoring method of embodiment 1.
It will be appreciated by those skilled in the art that the present invention can be carried out in other embodiments without departing from the spirit or essential characteristics thereof. Accordingly, the above disclosed embodiments are illustrative in all respects, and not exclusive. All changes that come within the scope of the invention or equivalents thereto are intended to be embraced therein.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it should be noted that: the above embodiments are only for illustrating the technical aspects of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: modifications and equivalents may be made to the specific embodiments of the invention without departing from the spirit and scope of the invention, which is intended to be covered by the claims.
Claims (10)
1. The construction monitoring method based on the load decomposition is characterized by comprising the following steps of:
acquiring load power data and influence factor data in a preset area;
extracting features according to the load power data to obtain equipment power features and time features of each equipment, and establishing a power feature model of each equipment according to the equipment power features and time features of each equipment;
training a CAF non-invasive load decomposition model according to the load power data and the influence factor data, and obtaining a monitoring result according to the CAF non-invasive load decomposition model;
and comparing the monitoring result with a power characteristic model of equipment corresponding to the monitoring result to obtain a construction state result.
2. The construction monitoring method based on load decomposition according to claim 1, wherein the step of extracting the characteristics according to the load power data to obtain the device power characteristics and the time characteristics of each device and establishing the power characteristic model of each device according to the device power characteristics and the time characteristics of each device specifically comprises the steps of:
extracting power value feature vectors of all the devices according to the load power data;
extracting time feature vectors of all the devices according to the load power data;
and inputting the time feature vector and the power value feature vector into a full connection layer to obtain a power feature model.
3. The construction monitoring method based on load decomposition according to claim 2, wherein the step of extracting the power value feature vector of each device according to the load power data specifically comprises:
generating an active power sequence of each device according to the load power data of each device;
inputting the active power sequence of each device into a convolution layer to obtain the power characteristic information of each device;
and inputting the active power sequences of the devices into an average pooling layer, and reserving power characteristic information to obtain a power value characteristic vector.
4. The construction monitoring method based on load decomposition according to claim 1, wherein the step of training a CAF non-invasive load decomposition model according to the load power data and the influence factor data and obtaining the monitoring result according to the CAF non-invasive load decomposition model specifically comprises the following steps:
inputting the influence factor data into a CNN model, and extracting influence factor characteristics;
model training is carried out according to the influence factor characteristics and the load power data as the input of the CAF non-invasive load decomposition model, and a decoder result is obtained;
and inputting the decoder result into the full connection layer to obtain a monitoring result.
5. The method for monitoring construction based on load decomposition according to claim 4, wherein the step of obtaining the decoder result by performing model training based on the influence factor characteristics and the load power data as input of the CAF non-invasive load decomposition model specifically comprises:
inputting the load power data into an Autoformer encoder, and calculating a query vector, a key vector and a value vector of the Autoformer encoder;
substituting values of the query vector and the key vector into a second layer autocorrelation mechanism of the Autoformer decoder, inputting the influence factor features into a first layer autocorrelation mechanism of the Autoformer decoder to obtain a second query vector, a second key vector and a second value vector, and introducing the second value vector into the second layer autocorrelation mechanism of the Autoformer decoder to obtain a decoder result.
6. The construction monitoring method based on load decomposition according to claim 1, wherein the CAF non-invasive load decomposition model is selected according to average absolute error, accuracy, precision, recall and F1 value.
7. The construction monitoring method based on load decomposition according to claim 1, wherein the step of obtaining load power data and influence factor data in a preset area specifically comprises:
acquiring load power data and influence factor data in a preset area;
filling the missing values of the obtained load power data and influence factor data to form a complete load power data set and an influence factor data set;
zero-mean normalization is performed on the load power dataset.
8. A construction monitoring device based on load decomposition, characterized by comprising:
and a data acquisition module: the method comprises the steps of acquiring load power data and influence factor data in a preset area;
the power characteristic model building module: the power characteristic model is used for extracting the characteristics according to the load power data to obtain the equipment power characteristics and the time characteristics of each equipment and establishing the power characteristic model of each equipment according to the equipment power characteristics and the time characteristics of each equipment;
model training module: the system is used for training a CAF non-invasive load decomposition model according to the load power data and the influence factor data, and obtaining a monitoring result according to the CAF non-invasive load decomposition model;
and a construction state monitoring module: and the power characteristic model is used for comparing the monitoring result with the power characteristic model of the equipment corresponding to the monitoring result to obtain a construction state result.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements a load decomposition based construction monitoring method according to any of claims 1-7 when the computer program is executed by the processor.
10. A computer-readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements a load decomposition based construction monitoring method according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311415380.0A CN117422590A (en) | 2023-10-27 | 2023-10-27 | Construction monitoring method, device, equipment and medium based on load decomposition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311415380.0A CN117422590A (en) | 2023-10-27 | 2023-10-27 | Construction monitoring method, device, equipment and medium based on load decomposition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117422590A true CN117422590A (en) | 2024-01-19 |
Family
ID=89527975
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311415380.0A Pending CN117422590A (en) | 2023-10-27 | 2023-10-27 | Construction monitoring method, device, equipment and medium based on load decomposition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117422590A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113837894A (en) * | 2021-08-06 | 2021-12-24 | 国网江苏省电力有限公司南京供电分公司 | Non-invasive resident user load decomposition method based on residual convolution module |
| CN113987913A (en) * | 2021-09-22 | 2022-01-28 | 吉林化工学院 | Improved non-invasive load decomposition model |
| CN115130732A (en) * | 2022-06-06 | 2022-09-30 | 中国计量大学 | Non-invasive load decomposition method based on sequence-to-point and residual error connection |
| CN115983347A (en) * | 2023-01-12 | 2023-04-18 | 上海电力大学 | Non-invasive load decomposition method, device and storage medium |
| CN116151118A (en) * | 2023-02-27 | 2023-05-23 | 常州大学 | Non-invasive load decomposition method based on BERT and CNN |
| CN116599036A (en) * | 2023-04-21 | 2023-08-15 | 安徽理工大学 | Two-stage non-invasive load decomposition method based on TCN and Informar |
| CN116865258A (en) * | 2023-07-14 | 2023-10-10 | 国网冀北电力有限公司电力科学研究院 | Hierarchical distributed power supply intelligent power grid construction method |
-
2023
- 2023-10-27 CN CN202311415380.0A patent/CN117422590A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113837894A (en) * | 2021-08-06 | 2021-12-24 | 国网江苏省电力有限公司南京供电分公司 | Non-invasive resident user load decomposition method based on residual convolution module |
| CN113987913A (en) * | 2021-09-22 | 2022-01-28 | 吉林化工学院 | Improved non-invasive load decomposition model |
| CN115130732A (en) * | 2022-06-06 | 2022-09-30 | 中国计量大学 | Non-invasive load decomposition method based on sequence-to-point and residual error connection |
| CN115983347A (en) * | 2023-01-12 | 2023-04-18 | 上海电力大学 | Non-invasive load decomposition method, device and storage medium |
| CN116151118A (en) * | 2023-02-27 | 2023-05-23 | 常州大学 | Non-invasive load decomposition method based on BERT and CNN |
| CN116599036A (en) * | 2023-04-21 | 2023-08-15 | 安徽理工大学 | Two-stage non-invasive load decomposition method based on TCN and Informar |
| CN116865258A (en) * | 2023-07-14 | 2023-10-10 | 国网冀北电力有限公司电力科学研究院 | Hierarchical distributed power supply intelligent power grid construction method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107391598B (en) | Automatic threat information generation method and system | |
| CN115578015A (en) | Sewage treatment overall process supervision method and system based on Internet of things and storage medium | |
| CN111340110B (en) | Fault early warning method based on industrial process running state trend analysis | |
| CN115296933B (en) | Industrial production data risk level assessment method and system | |
| CN115563880A (en) | Enterprise power consumption abnormity detection method based on Isolated forest-variable point enhancement | |
| CN116362785A (en) | Power user electricity consumption abnormality diagnosis method and device in big data environment | |
| CN117933531A (en) | Distributed photovoltaic power generation power prediction system and method | |
| CN110458383B (en) | Method and device for realizing demand processing servitization, computer equipment and storage medium | |
| CN117422590A (en) | Construction monitoring method, device, equipment and medium based on load decomposition | |
| CN117033923A (en) | Method and system for predicting crime quantity based on interpretable machine learning | |
| CN116302835A (en) | Fortune dimension abnormity detection device, method and storage medium | |
| Zhang et al. | Prediction and Detection of Sewage Treatment Process Using N-BEATS Autoencoder Network | |
| CN116108276A (en) | Information recommendation method and device based on artificial intelligence and related equipment | |
| CN113408210B (en) | Deep learning-based non-invasive load decomposition method, system, medium and equipment | |
| CN113326371B (en) | Event extraction method integrating pre-training language model and anti-noise interference remote supervision information | |
| CN111221704B (en) | Method and system for determining running state of office management application system | |
| CN114490645A (en) | Automatic machine learning method for power grid structured data | |
| Anishaa et al. | Identifying similar question pairs using machine learning techniques | |
| CN118444576B (en) | Sewage treatment monitoring method and system based on Internet of things | |
| Zhao et al. | Burst detection in district metering areas using flow subsequences clustering–reconstruction analysis | |
| CN117807046B (en) | Data driving model and system based on event map analysis | |
| Yu et al. | A regression prediction model based on incremental iteration for big industrial data | |
| CN117768220B (en) | Network security level protection evaluation method, system and device based on artificial intelligence | |
| CN117194576B (en) | Power grid customer information data integration processing method and system | |
| CN113988360B (en) | Wind power prediction method and device based on wind speed fluctuation characteristic typing |
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 |