CN115795323B - Malignant load identification method, equipment and storage medium - Google Patents
Malignant load identification method, equipment and storage medium Download PDFInfo
- Publication number
- CN115795323B CN115795323B CN202310052973.9A CN202310052973A CN115795323B CN 115795323 B CN115795323 B CN 115795323B CN 202310052973 A CN202310052973 A CN 202310052973A CN 115795323 B CN115795323 B CN 115795323B
- Authority
- CN
- China
- Prior art keywords
- model
- sample
- sub
- malignant load
- output result
- 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.)
- Active
Links
Images
Abstract
The invention discloses a malignant load identification method, malignant load identification equipment and a storage medium, belonging to the technical field of malignant loads, wherein the method comprises the following steps: acquiring circuit characteristic data of a target circuit; inputting the circuit characteristic data into a first model to obtain an output result of the first model, wherein the output result of the first model comprises whether a malignant load exists in the target circuit or not; inputting the circuit characteristic data into a second model to obtain an output result of the second model, wherein the output result of the second model comprises whether a load matched with a preset malignant load type exists in the target circuit or not; and outputting the identification result of the malignant load according to the output result of the first model and the output result of the second model. The invention aims to improve robustness of malignant load identification and maintenance efficiency.
Description
Technical Field
The present invention relates to the technical field of malignant load, and in particular, to a malignant load identification method, a malignant load identification device, and a storage medium.
Background
With the development of economy, people's settlement and corresponding electricity scale are expanding, and the situation that the number of students and the electricity demand of the students are increased also appears in college student apartments, however, this also leads to the serious safety threat of electricity consumption fire accidents. In fire accidents occurring in college student apartments, most of the fire accidents are caused by illegal use of high-power malignant loads such as rapid heating, induction cookers and the like. Therefore, the identification and monitoring of the malignant load are of great significance in guaranteeing the electricity safety.
At present, a non-invasive load monitoring method is generally adopted for identifying malignant load, and a model for classifying electric equipment is established for identifying the malignant load, however, the method can influence the identification accuracy due to the fact that the newly added malignant load type cannot be identified when the malignant load type is updated, so that the robustness of the method is not high.
Disclosure of Invention
The main object of the present invention is to provide a malignant load recognition method, a malignant load recognition apparatus and a storage medium, aiming at improving the robustness of the malignant load recognition and improving the maintenance efficiency.
In order to achieve the above object, the present invention provides a malignant load identification method comprising the steps of:
acquiring circuit characteristic data of a target circuit;
inputting the circuit characteristic data into a first model to obtain an output result of the first model, wherein the output result of the first model comprises whether a malignant load exists in the target circuit or not;
inputting the circuit characteristic data into a second model to obtain an output result of the second model, wherein the output result of the second model comprises whether a load matched with a preset malignant load type exists in the target circuit or not;
and outputting the identification result of the malignant load according to the output result of the first model and the output result of the second model.
Optionally, the step of outputting the identification result of the malignant load according to the output result of the first model and the output result of the second model includes:
when the output result of the first model is matched with the output result of the second model, determining that the identification result comprises the output result of the first model and the output result of the second model, and outputting the identification result;
when the output result of the first model is not matched with the output result of the second model, determining that the identification result comprises the output result of the first model and warning information, and outputting the identification result.
Optionally, the number of the circuit feature data is a plurality, and the step of outputting the identification result of the malignant load according to the output result of the first model and the output result of the second model includes:
and when the output result of the first model corresponding to each circuit characteristic data is not matched with the output result of the corresponding second model, determining that the identification result comprises the output result of the first model and the warning information, and outputting the identification result.
Optionally, before the step of acquiring the circuit characteristic data of the target circuit, the method further includes:
receiving harmonic data of the target circuit acquired at a first moment uploaded by equipment and storing the harmonic data into a database;
sequentially receiving a plurality of sub-electrical signal data uploaded by the equipment, storing the sub-electrical signal data in a cache to form first electrical signal data of the equipment until receiving electrical signal data of a second moment uploaded by the equipment, wherein the plurality of sub-electrical signal data are formed by dividing the electrical signal data of the target circuit acquired by the equipment at the first moment;
storing the first electrical signal data in the buffer in association with the harmonic data to the database;
the step of acquiring the circuit characteristic data of the target circuit comprises the following steps:
the harmonic data and the first electrical signal data stored in association are acquired from the database as the circuit characteristic data.
Optionally, before the step of inputting the circuit feature data into the first model, the method further includes:
acquiring a sample data set;
determining that a sample with malignant load in the sample data set is a first positive sample, and a sample without malignant load is a first negative sample;
training a first learning model according to the first positive sample and the first negative sample, and obtaining the first model.
Optionally, the second model includes a plurality of second sub-models, each of the second sub-models is configured to identify a corresponding one of the preset malignant load types, and before the step of inputting the circuit characteristic data into the second model, the method further includes:
acquiring a sample data set;
determining a target malignant load type according to the preset malignant load type;
determining a second positive sample and a second negative sample corresponding to the target malignant load type according to the sample data set and the target malignant load type;
training a second learning sub-model according to the second positive sample and the second negative sample to obtain a second sub-model for identifying the target malignant load type.
Optionally, the malignant load identification method further includes:
when a model updating instruction is received, a newly added malignant load type is obtained;
determining that a sample of the sample data set, in which the newly added malignant load type exists, is a third positive sample, and a sample of the sample data set, in which the newly added malignant load type does not exist, is a third negative sample;
training the second learning sub-model according to the third positive sample and the third negative sample to obtain a new sub-model;
the new additive model is added to the second model.
Optionally, the step of determining a second positive sample and a second negative sample corresponding to the target malignant load type according to the sample data set and the target malignant load type includes:
determining that a sample of the sample dataset in which the target malignant load type exists is a fourth positive sample and a sample of the sample dataset in which the target malignant load type does not exist is the second negative sample;
when the number of samples of the second negative sample is greater than that of the fourth positive sample, selecting one sub-sample in the fourth positive sample as a first sub-sample, determining a plurality of second sub-samples meeting preset conditions in the fourth positive sample, and obtaining a corresponding fifth positive sample according to the first sub-sample and the plurality of second sub-samples;
when the total number of samples of the fourth positive sample and all the fifth positive samples is smaller than the number of samples of the second negative sample, returning to the step of executing the selection of one sub-sample in the fourth positive sample as a first sub-sample, determining a plurality of second sub-samples meeting preset conditions in the fourth positive sample, and obtaining a corresponding fifth positive sample according to the first sub-sample and the plurality of second sub-samples;
determining that a sample set of the fourth positive sample and all the fifth positive samples is the second positive sample when a total number of samples of the fourth positive sample and all the fifth positive samples is equal to a number of samples of the second negative sample;
the preset conditions include that the Euclidean distance between each second subsamples and the first subsamples is smaller than the Euclidean distance between any subsamples in the first subsamples and the second negative samples.
In addition, in order to achieve the above object, the present application also proposes a malignant load recognition apparatus including: a memory, a processor, and a malignancy identification program stored on the memory and executable on the processor, the malignancy identification program configured to implement the steps of the malignancy identification method as set forth in any one of the preceding claims.
In order to achieve the above object, the present application further proposes a storage medium having stored thereon a malignancy load identification program which, when executed by a processor, implements the steps of the malignancy load identification method according to any one of the above.
According to the malignant load identification method provided by the invention, the malignant load condition of the target circuit is analyzed by acquiring the circuit characteristic data of the target circuit and utilizing the circuit characteristic data; inputting the circuit characteristic data into a first model, obtaining an output result of the first model, and identifying whether a malignant load exists in the target circuit through the first model; inputting the circuit characteristic data into a second model to obtain an output result of the second model, and identifying whether a load matched with a preset malignant load type exists in the target circuit or not through the second model, so that double identification is carried out from two angles through the two models; and outputting the identification result of the malignant load according to the output result of the first model and the output result of the second model. Compared with the method for establishing only one model for classifying the electric equipment, the method and the device for identifying the malignant load type by using the double-recognition can ensure the accuracy of the identification when the malignant load type to be identified is increased by using the double-recognition method and the device for identifying the electric equipment, so that the robustness of the malignant load identification is improved, the first model and/or the second model can be selectively maintained according to the requirements, the maintenance of the model for classifying the electric equipment is not required to be repeated, and the maintenance efficiency is improved.
Drawings
FIG. 1 is a schematic diagram of a hardware architecture involved in the operation of an embodiment of a malignant load recognition apparatus according to the present invention;
FIG. 2 is a flow chart of a method for identifying a malignant load according to an embodiment of the present invention;
FIG. 3 is a flowchart of another embodiment of a malignant load recognition method according to the present invention;
fig. 4 is a flowchart of a malignant load identification method according to another embodiment of the invention.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
The embodiment of the invention provides malignant load identification equipment. As shown in fig. 1, the malignant load recognition apparatus may include: a processor 1001, such as a central processing unit (Central Processing Unit, CPU), a communication bus 1002, a user interface 1003, a network interface 1004, a memory 1005. Wherein the communication bus 1002 is used to enable connected communication between these components. The user interface 1003 may include a Display, an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may further include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a WIreless interface (e.g., a WIreless-FIdelity (WI-FI) interface). The Memory 1005 may be a high-speed random access Memory (Random Access Memory, RAM) Memory or a stable nonvolatile Memory (NVM), such as a disk Memory. The memory 1005 may also optionally be a storage device separate from the processor 1001 described above.
It will be appreciated by those skilled in the art that the structure shown in fig. 1 does not constitute a limitation of the malignant load recognition apparatus, and the malignant load recognition apparatus may further comprise functional modules for realizing other functions, including more or less components than illustrated, or some components may be combined, or different arrangements of components.
As shown in fig. 1, a malignant load recognition program may be included in a memory 1005 as one type of storage medium. In the malignancy identification device shown in fig. 1, the network interface 1004 is mainly used for data communication with other devices; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 may be configured to call a malignancy recognition program stored in the memory 1005 and execute the steps of the malignancy recognition method provided by the embodiment of the present invention.
The embodiment of the invention also provides a malignant load identification method which is applied to the malignant load identification equipment.
Referring to fig. 2, an embodiment of a malignant load identification method of the present application is provided. In this embodiment, the malignant load identification method includes the steps of:
step S10, obtaining circuit characteristic data of a target circuit;
the embodiment provides a method for identifying and analyzing malignant load conditions in a target circuit, wherein the malignant load is also called a resistive load and generally refers to purely resistive direct heating type electric appliances such as electric soldering iron, electric hair drier, electric heating furnace and the like; the target circuit may be a circuit in which a large number of electric appliances exist in a student apartment or the like, or may be a test circuit designed for a specific electric appliance.
Optionally, the malignancy load identification device obtains circuit characteristic data of the target circuit and analyzes according to the circuit characteristic data to monitor malignancy load conditions in the target circuit.
Step S20, inputting the circuit characteristic data into a first model, and obtaining an output result of the first model, wherein the output result of the first model comprises whether a malignant load exists in the target circuit or not;
the first model is a pre-trained machine learning model, and can identify and output whether malignant load exists in a corresponding target circuit according to the input circuit characteristic data, so that coarse-grained identification is performed on the malignant load condition in the target circuit.
Illustratively, the data samples learned while the first model is trained include electric blowers and electric ovens, and then when electric blowers and/or electric ovens are present in the target circuit, the first model outputs a result that a malignant load is present in the target circuit.
Step S30, inputting the circuit characteristic data into a second model, and obtaining an output result of the second model, wherein the output result of the second model comprises whether a load matched with a preset malignant load type exists in the target circuit;
the second model is also a pre-trained machine learning model, and can output whether a load matched with a preset malignant load type exists in a corresponding target circuit according to the input circuit characteristic data, so that fine-grained identification is carried out on the malignant load condition in the target circuit; the preset malignant load type is the malignant load type which can be identified by the second model.
The second model may determine whether the electric hair drier exists in the target circuit and whether the electric heating furnace exists in the target circuit according to the circuit characteristic data.
And step S40, outputting the identification result of the malignant load according to the output result of the first model and the output result of the second model.
Alternatively, the above steps S20 and S30 may be performed simultaneously, and the circuit characteristic data is input into the first model and the second model at the same time, so as to save time required for data analysis.
From the output of the first model, it may be determined whether a malignant load exists in the target circuit, and from the output of the second model, it may be determined whether a load matching a preset malignant load type exists in the target circuit (i.e., whether an appliance in the preset malignant load type exists). And determining and outputting a recognition result of the malignant load of the target circuit based on the output result of the first model and the output result of the second model.
According to the malignant load identification method provided by the embodiment of the invention, the malignant load condition of the target circuit is analyzed by acquiring the circuit characteristic data of the target circuit and utilizing the circuit characteristic data; inputting the circuit characteristic data into a first model, obtaining an output result of the first model, and identifying whether a malignant load exists in the target circuit through the first model; inputting the circuit characteristic data into a second model to obtain an output result of the second model, and identifying whether a load matched with a preset malignant load type exists in the target circuit or not through the second model, so that double identification is carried out from two angles through the two models; and outputting the identification result of the malignant load according to the output result of the first model and the output result of the second model. Compared with the method for establishing only one model for classifying the electric equipment, the method and the device for identifying the malignant load type by using the double-recognition can ensure the accuracy of the identification when the malignant load type to be identified is increased by using the double-recognition method and the device for identifying the electric equipment, so that the robustness of the malignant load identification is improved, the first model and/or the second model can be selectively maintained according to the requirements, the maintenance of the model for classifying the electric equipment is not required to be repeated, and the maintenance efficiency is improved.
Further, in this embodiment, before the step of obtaining the circuit characteristic data of the target circuit, the method further includes:
receiving harmonic data of the target circuit acquired at a first moment uploaded by equipment and storing the harmonic data into a database;
sequentially receiving a plurality of sub-electrical signal data uploaded by the equipment, storing the sub-electrical signal data in a cache to form first electrical signal data of the equipment until receiving electrical signal data of a second moment uploaded by the equipment, wherein the plurality of sub-electrical signal data are formed by dividing the electrical signal data of the target circuit acquired by the equipment at the first moment;
storing the first electrical signal data in the buffer in association with the harmonic data to the database;
the step of acquiring the circuit characteristic data of the target circuit comprises the following steps:
the harmonic data and the first electrical signal data stored in association are acquired from the database as the circuit characteristic data.
In this embodiment, the circuit characteristic data includes harmonic data and electric signal data, wherein the electric signal data includes current data and voltage data, and the harmonic data includes current harmonic data and voltage harmonic data. The device is used for being connected with the target circuit, collecting circuit characteristic data of the target circuit at a plurality of moments at preset intervals, uploading the circuit characteristic data to the cloud platform, and the cloud platform is configured with the malignant load identification device.
Optionally, the device is connected with the target circuit to collect harmonic data and electric signal data, wherein the harmonic data can collect 2-31 times of harmonic waves, and the accuracy of the identification result is improved by improving the accuracy of the collected data; after circuit characteristic data at a first moment is acquired, harmonic data are uploaded and stored in a database, the electric signal data are divided into a plurality of sub-electric signal data to be sequentially uploaded due to the fact that data blocks are large, and after the divided sub-electric signal data are sequentially received by malignant load identification equipment, the sub-electric signal data are stored in a buffer to form the first electric signal data, and after electric signal data at a second moment are received, the first electric signal data in the buffer and the harmonic data at the corresponding moment are stored in the database in a correlated mode. When data analysis is performed, the malignant load identification device acquires harmonic data at the first moment and corresponding first electric signal data from the database, so that the malignant load condition of the target circuit at the first moment is analyzed.
By receiving the circuit characteristic data uploaded by the equipment connected with the target circuit, the cloud platform can be used for carrying out data analysis by utilizing the malignant load identification equipment, so that equipment required to be deployed in the current place of the target circuit is reduced, and the deployment cost is reduced; and based on the cloud platform, the malignant load identification equipment can simultaneously identify the malignant load conditions of a plurality of target circuits, so that the identification efficiency of the malignant load is improved.
Further, based on the above embodiment, another embodiment of the malignant load identification method of the present application is proposed. In this embodiment, referring to fig. 3, the step of outputting the identification result of the malignant load according to the output result of the first model and the output result of the second model includes:
step S41, when the output result of the first model is matched with the output result of the second model, determining that the identification result comprises the output result of the first model and the output result of the second model, and outputting the identification result;
the output result of the first model is matched with the output result of the second model, namely, the first model judges that a malignant load exists in the target circuit, and the second model judges that a load matched with a preset malignant load type exists in the target circuit, for example, the first model judges that the target circuit has a malignant load, and the second model judges that the target circuit has an electric hair drier; meanwhile, it is also possible that the first model judges that no malignant load exists in the target circuit, and the second model judges that no load matched with the preset malignant load type exists in the target circuit.
And step S42, when the output result of the first model is not matched with the output result of the second model, determining that the identification result comprises the output result of the first model and warning information, and outputting the identification result.
The output result of the first model is not matched with the output result of the second model, namely the first model judges that malignant load exists in the target circuit, but the second model judges that load matched with the preset malignant load type does not exist in the target circuit; it is also possible that the first model judges that no malignant load exists in the target circuit, whereas the second model judges that a load matching a preset malignant load type exists in the target circuit.
Optionally, when the output result of the first model is matched with the output result of the second model, the output results of both models can be output as the identification result of the malignant load; when the output result of the first model is not matched with the output result of the second model, the output results of the two models are contradictory, and the first model with coarse granularity identification has more reliable identification accuracy relative to the second model, at the moment, the output result of the first model and corresponding warning information are taken as identification results of malignant load, and a worker can know that the two models are contradictory according to the warning information.
By comparing whether the output results of the two models are matched, different recognition directions of the two models can be combined, the accuracy of the malignant load recognition effect is further improved, and the robustness is improved. In addition, when two models are judged to be contradictory, a recognition result with more reference value can be provided for the staff.
Further, in this embodiment, the number of the circuit feature data is plural, and the step of outputting the identification result of the malignant load according to the output result of the first model and the output result of the second model includes:
and when the output result of the first model corresponding to each circuit characteristic data is not matched with the output result of the corresponding second model, determining that the identification result comprises the output result of the first model and the warning information, and outputting the identification result.
Optionally, when the output result of the first model is not matched with the output result of the second model, multiple determinations are made using the multiple circuit feature data, and whether the output result of the first model corresponding to each determination is matched with the output result of the second model is compared. And when the output results of the first model are compared for a plurality of times and are not matched with the output results of the second model, the output results of the first model and the corresponding warning information are used as identification results of the malignant load. Further, a preset threshold value may be set as a threshold value for a plurality of determinations, for example, 3 times. And when the output result of the first model is changed to be matched with the output result of the second model in the multiple judgments, the matched output result of the first model and the matched output result of the second model are used as identification results of malignant loads.
By utilizing multiple data analysis of multiple circuit characteristic data and comparing whether the first model output result and the second model output result corresponding to each circuit characteristic data are matched or not, the judgment contradiction of the first model and the second model caused by the abnormal data at individual moments can be prevented, and the robustness of malignant load is improved.
Further, based on the above embodiment, still another embodiment of the malignant load identification method of the present application is provided. In this embodiment, referring to fig. 4, before the step of inputting the circuit characteristic data into the first model, the method further includes:
step S21, acquiring a sample data set;
the sample dataset is a collection of data samples for training the first model and the second model, including various types of loads and their combinations corresponding circuit characteristic data.
Step S22, determining that a sample with malignant load in the sample data set is a first positive sample, and determining that a sample without malignant load is a first negative sample;
because the first model is used for identifying whether the malignant load exists in the target circuit, the first model is optionally a classifier, wherein a random forest algorithm is adopted, and according to the input circuit characteristic data, two results of whether the malignant load exists or not can be output.
Optionally, determining that the sample with the malignant load in the sample data set is a first positive sample, determining that the sample without the malignant load is a first negative sample, and using the first positive sample and the first negative sample as training materials of the first model.
Step S23, training a first learning model according to the first positive sample and the first negative sample, and obtaining the first model.
Optionally, training a first learning model according to the first positive sample and the first negative sample, wherein the first learning model is a binary base learner based on a random forest algorithm, and obtaining the first model for identifying whether malignant load exists according to circuit characteristic data after training.
By determining the first positive sample and the first negative sample in the data sample set to train the first learning model, a first model for identifying whether malignant load exists or not can be obtained, a coarse-grained malignant load identification direction is provided for malignant load identification equipment, and accuracy of malignant load identification is improved.
Further, in this embodiment, the second model includes a plurality of second sub-models, each of the second sub-models is configured to identify a corresponding one of the preset malignant load types, and before the step of inputting the circuit feature data into the second model, the method further includes:
acquiring a sample data set;
determining a target malignant load type according to the preset malignant load type;
determining a second positive sample and a second negative sample corresponding to the target malignant load type according to the sample data set and the target malignant load type;
training a second learning sub-model according to the second positive sample and the second negative sample to obtain a second sub-model for identifying the target malignant load type.
Since the second model is used for identifying whether a load matching a preset malignant load type exists in the target circuit, the second model comprises a corresponding number of second sub-models for a plurality of malignant load types in the preset malignant load types, and each second sub-model is used for identifying one malignant load type. Optionally, each second sub-model is a classifier of a random forest algorithm, and according to the input circuit characteristic data, two results of the corresponding malignant load type or the corresponding malignant load type are output. The preset malignant load type comprises an electric hair drier and an electric heating furnace, and the second model comprises two second sub-models which are respectively used for identifying whether the electric hair drier exists in the target circuit or not and whether the electric heating furnace exists or not.
Optionally, determining one of the preset malignant load types as a target malignant load type, determining a sample with the target malignant load type in the sample data set as a second positive sample, and training a second learning sub-model according to the second positive sample and the second negative sample, wherein the second learning sub-model is also a binary base learner based on a random forest algorithm, obtaining a second sub-model for identifying the target malignant load type after training, and training other second sub-models by using the same method for other malignant load types in the preset malignant load type so as to obtain a second model.
By training a plurality of second sub-models for identifying a plurality of malignant load types in the preset malignant load types, two-classification identification can be performed for each malignant load type, fine-granularity malignant load identification directions are provided for malignant load identification equipment, and accuracy of malignant load identification is improved. And when the newly added malignant load type exists, the identification effect of other second sub-models is not influenced, and the robustness of malignant load identification is improved.
Further, in this embodiment, the method for identifying a malignant load further includes:
when a model updating instruction is received, a newly added malignant load type is obtained;
determining that a sample of the sample data set, in which the newly added malignant load type exists, is a third positive sample, and a sample of the sample data set, in which the newly added malignant load type does not exist, is a third negative sample;
training the second learning sub-model according to the third positive sample and the third negative sample to obtain a new sub-model;
the new additive model is added to the second model.
The model update instruction is an instruction for a malignant load type increase in the preset malignant load types, indicating that the second model needs to be updated so that the second model can identify the newly increased malignant load type.
Optionally, when the newly identified malignant load type is required, determining the newly identified malignant load type, determining the sample with the newly identified malignant load type as a third positive sample in the sample data set, and training a second learning sub-model according to the third positive sample and the third negative sample, thereby obtaining a new sub-model for identifying the newly identified malignant load type, and adding the new sub-model to the second model, wherein the second model can be used for identifying the newly identified malignant load type. The algorithm and learning method used by the new sub-model is identical to the other second sub-models, except that the training material is a third positive and negative sample associated with the new added malignant load type.
In addition, if it is determined that the newly added malignant load type affects the recognition accuracy of the first model, the first model can be trained again by combining the training material of the newly added malignant load type, so that the first model is maintained.
The second model can be updated by training the new sub-model for identifying the newly added malignant load type, and because the second model comprises a plurality of second sub-models, each second sub-model is used for identifying a corresponding malignant load type, when the malignant load type is newly added, only the corresponding new sub-model is required to be trained, and retraining of all the second sub-models is not required, so that the maintenance efficiency is improved.
Further, in this embodiment, the step of determining, according to the sample data set and the target malignant load type, a second positive sample and a second negative sample corresponding to the target malignant load type includes:
determining that a sample of the sample dataset in which the target malignant load type exists is a fourth positive sample and a sample of the sample dataset in which the target malignant load type does not exist is the second negative sample;
when the number of samples of the second negative sample is greater than that of the fourth positive sample, selecting one sub-sample in the fourth positive sample as a first sub-sample, determining a plurality of second sub-samples meeting preset conditions in the fourth positive sample, and obtaining a corresponding fifth positive sample according to the first sub-sample and the plurality of second sub-samples;
when the total number of samples of the fourth positive sample and all the fifth positive samples is smaller than the number of samples of the second negative sample, returning to the step of executing the selection of one sub-sample in the fourth positive sample as a first sub-sample, determining a plurality of second sub-samples meeting preset conditions in the fourth positive sample, and obtaining a corresponding fifth positive sample according to the first sub-sample and the plurality of second sub-samples;
determining that a sample set of the fourth positive sample and all the fifth positive samples is the second positive sample when a total number of samples of the fourth positive sample and all the fifth positive samples is equal to a number of samples of the second negative sample;
the preset conditions include that the Euclidean distance between each second subsamples and the first subsamples is smaller than the Euclidean distance between any subsamples in the first subsamples and the second negative samples.
Since the number of positive samples in the data sample set for a particular one of the malignant load types is generally smaller than its negative number of samples, and there is a large number gap, in order to ensure the training effect of the second model, the number of positive samples needs to be increased by oversampling.
Optionally, when it is determined that the number of samples of the fourth positive samples with the target malignant load type in the data sample set is smaller than the number of samples of the second negative samples without the target malignant load type, selecting one of the fourth positive samples as the first sub-sample, and then determining a plurality of second sub-samples adjacent to the first sub-sample that meet a preset condition, where the preset condition includes euclidean distances between the first sub-sample and each of the second sub-samples, which are smaller than euclidean distances between any of the first sub-samples and the second negative samples, and the euclidean distances are calculated according to circuit feature data corresponding to the sub-samples. After the first sub-sample and the second sub-sample are determined, corresponding simulation sub-samples are randomly generated between the first sub-sample and each second sub-sample, and the generated simulation sub-samples form a fifth positive sample. Further, judging whether the total number of the samples of the fourth positive sample and the fifth positive sample is equal to the second negative sample, if the total number of the samples of the fourth positive sample and the fifth positive sample is still smaller than the second negative sample, selecting another first sub-sample again, and determining that the second sub-sample generates more fifth positive samples until the total number of the samples of the fourth positive sample and the fifth positive sample is equal to the number of the samples of the second negative sample, thereby obtaining a second positive sample composed of the fourth positive sample and the fifth positive sample.
And obtaining a second positive sample by oversampling the fourth positive sample, balancing the number between the positive sample and the negative sample of the second model, and improving the training effect of the second model, thereby improving the identification accuracy of the malignant load.
In addition, the embodiment of the invention also provides a storage medium, wherein the storage medium is stored with a malignant load identification program, and the malignant load identification program realizes the relevant steps of any embodiment of the malignant load identification method when being executed by a processor.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as described above, comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method according to the embodiments of the present invention.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (8)
1. A method of identifying a malignant load, the method comprising the steps of:
acquiring circuit characteristic data of a target circuit, wherein the number of the circuit characteristic data is a plurality of, and the circuit characteristic data comprises harmonic data of the target circuit;
inputting the circuit characteristic data into a first model to obtain an output result of the first model, wherein the output result of the first model comprises whether a malignant load exists in the target circuit or not;
inputting the circuit characteristic data into a second model to obtain an output result of the second model, wherein the output result of the second model comprises whether a load matched with a preset malignant load type exists in the target circuit or not;
outputting a recognition result of the malignant load according to the output result of the first model and the output result of the second model;
the step of outputting the identification result of the malignant load according to the output result of the first model and the output result of the second model comprises the following steps:
when the output result of the first model is matched with the output result of the second model, determining that the identification result comprises the output result of the first model and the output result of the second model, and outputting the identification result;
when the output result of the first model corresponding to each circuit characteristic data is not matched with the output result of the corresponding second model, determining that the identification result comprises the output result of the first model and warning information, and outputting the identification result, wherein the warning information is used for prompting that the first model and the second model judging result are contradictory;
wherein the first model and the second model each comprise a classifier based on a random forest algorithm.
2. The method of identifying a malignant load of claim 1, wherein prior to the step of obtaining circuit characteristic data of the target circuit, further comprising:
receiving harmonic data of the target circuit acquired at a first moment uploaded by equipment and storing the harmonic data into a database;
sequentially receiving a plurality of sub-electrical signal data uploaded by the equipment, storing the sub-electrical signal data in a cache to form first electrical signal data of the equipment until receiving electrical signal data of a second moment uploaded by the equipment, wherein the plurality of sub-electrical signal data are formed by dividing the electrical signal data of the target circuit acquired by the equipment at the first moment;
storing the first electrical signal data in the buffer in association with the harmonic data to the database;
the step of acquiring the circuit characteristic data of the target circuit comprises the following steps:
the harmonic data and the first electrical signal data stored in association are acquired from the database, and the circuit characteristic data further comprises the first electrical signal data of the target circuit.
3. The method of identifying a malignant load according to any one of claims 1 to 2, further comprising, prior to the step of inputting the circuit characteristic data into the first model:
acquiring a sample data set;
determining that a sample with malignant load in the sample data set is a first positive sample, and a sample without malignant load is a first negative sample;
training a first learning model according to the first positive sample and the first negative sample, and obtaining the first model.
4. The method of identifying a malignancy of any one of claims 1 to 2, wherein the second model comprises a plurality of second sub-models, each of the second sub-models being for identifying a corresponding one of the preset malignancy types, the step of inputting the circuit characteristic data into the second model further comprising:
acquiring a sample data set;
determining a target malignant load type according to the preset malignant load type;
determining a second positive sample and a second negative sample corresponding to the target malignant load type according to the sample data set and the target malignant load type;
training a second learning sub-model according to the second positive sample and the second negative sample to obtain a second sub-model for identifying the target malignant load type.
5. The method of identifying a malignant load of claim 4, further comprising:
when a model updating instruction is received, a newly added malignant load type is obtained;
determining that a sample of the sample data set, in which the newly added malignant load type exists, is a third positive sample, and a sample of the sample data set, in which the newly added malignant load type does not exist, is a third negative sample;
training the second learning sub-model according to the third positive sample and the third negative sample to obtain a new sub-model;
the new additive model is added to the second model.
6. The method of malignancy identification as defined in claim 4 wherein determining a second positive sample and a second negative sample corresponding to the target malignancy type from the sample data set and the target malignancy type comprises:
determining that a sample of the sample dataset in which the target malignant load type exists is a fourth positive sample and a sample of the sample dataset in which the target malignant load type does not exist is the second negative sample;
when the number of samples of the second negative sample is greater than that of the fourth positive sample, selecting one sub-sample in the fourth positive sample as a first sub-sample, determining a plurality of second sub-samples meeting preset conditions in the fourth positive sample, and obtaining a corresponding fifth positive sample according to the first sub-sample and the plurality of second sub-samples;
when the total number of samples of the fourth positive sample and all the fifth positive samples is smaller than the number of samples of the second negative sample, returning to the step of executing the selection of one sub-sample in the fourth positive sample as a first sub-sample, determining a plurality of second sub-samples meeting preset conditions in the fourth positive sample, and obtaining a corresponding fifth positive sample according to the first sub-sample and the plurality of second sub-samples;
determining that a sample set of the fourth positive sample and all the fifth positive samples is the second positive sample when a total number of samples of the fourth positive sample and all the fifth positive samples is equal to a number of samples of the second negative sample;
the preset conditions include that the Euclidean distance between each second subsamples and the first subsamples is smaller than the Euclidean distance between any subsamples in the first subsamples and the second negative samples.
7. A malignant load recognition apparatus, characterized by comprising: a memory, a processor and a malignancy identification program stored on the memory and executable on the processor, the malignancy identification program configured to implement the steps of the malignancy identification method according to any one of claims 1 to 6.
8. A storage medium having stored thereon a malignancy identification program which, when executed by a processor, implements the steps of the malignancy identification method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310052973.9A CN115795323B (en) | 2023-02-03 | 2023-02-03 | Malignant load identification method, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310052973.9A CN115795323B (en) | 2023-02-03 | 2023-02-03 | Malignant load identification method, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115795323A CN115795323A (en) | 2023-03-14 |
| CN115795323B true CN115795323B (en) | 2023-05-05 |
Family
ID=85429569
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310052973.9A Active CN115795323B (en) | 2023-02-03 | 2023-02-03 | Malignant load identification method, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115795323B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106980055A (en) * | 2017-03-13 | 2017-07-25 | 东华大学 | A kind of student dormitory based on data mining electrical equipment violating the regulations uses monitoring system |
| CN114689959A (en) * | 2020-12-31 | 2022-07-01 | 无锡小净共享网络科技有限公司 | Malignant load recognition device, method, system and storage medium |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2304449B1 (en) * | 2008-07-17 | 2012-05-16 | ISIS Innovation Limited | Utility metering |
| CN205335811U (en) * | 2016-02-05 | 2016-06-22 | 上海用悦信息技术服务中心 | Pernicious load recognition device |
| CN108573288B (en) * | 2018-05-29 | 2021-11-23 | 福建新开普信息科技有限公司 | Resistive load identification and learning method based on electric energy meter |
| CN110135452B (en) * | 2019-03-27 | 2023-08-08 | 广东浩迪创新科技有限公司 | Illegal electric appliance identification method and system based on intelligent electric meter |
| CN110929563A (en) * | 2019-10-14 | 2020-03-27 | 广东浩迪创新科技有限公司 | Electrical appliance identification method and device |
| CN110967585B (en) * | 2019-12-20 | 2022-03-15 | 武汉盛帆电子股份有限公司 | Malignant load identification method and device |
| CN112816816A (en) * | 2020-12-31 | 2021-05-18 | 无锡小净共享网络科技有限公司 | Malignant load identification method, device, equipment and storage medium |
| CN113673480A (en) * | 2021-09-03 | 2021-11-19 | 武汉盛帆电子股份有限公司 | Non-intrusive load identification method and device |
-
2023
- 2023-02-03 CN CN202310052973.9A patent/CN115795323B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106980055A (en) * | 2017-03-13 | 2017-07-25 | 东华大学 | A kind of student dormitory based on data mining electrical equipment violating the regulations uses monitoring system |
| CN114689959A (en) * | 2020-12-31 | 2022-07-01 | 无锡小净共享网络科技有限公司 | Malignant load recognition device, method, system and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115795323A (en) | 2023-03-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110007220B (en) | Method and device for diagnosing operating state of circuit breaker mechanism | |
| CN110633222B (en) | Method and device for determining regression test case | |
| CN108198408B (en) | Self-adaptive anti-electricity-stealing monitoring method and system based on electricity information acquisition system | |
| CN112801423B (en) | Method and device for identifying abnormity of air quality monitoring data and storage medium | |
| CN111818198A (en) | Domain name detection method, domain name detection device, equipment and medium | |
| US10853829B2 (en) | Association method, and non-transitory computer-readable storage medium | |
| CN110097120B (en) | Network flow data classification method, equipment and computer storage medium | |
| CN115174250A (en) | Network asset safety assessment method and device, electronic equipment and storage medium | |
| CN114910756A (en) | Insulation performance evaluation method and system for low-voltage bus duct | |
| CN115795323B (en) | Malignant load identification method, equipment and storage medium | |
| CN113420295A (en) | Malicious software detection method and device | |
| Li et al. | Symbolization‐based differential evolution strategy for identification of structural parameters | |
| CN110704614B (en) | Information processing method and device for predicting user group type in application | |
| CN113033639A (en) | Training method of abnormal data detection model, electronic device and storage medium | |
| CN110705603B (en) | Method and system for dynamically judging similarity of user request data | |
| CN111784042A (en) | Power utilization node safety risk prediction method and device and storage medium | |
| US20220129774A1 (en) | Information processing system and information processing method | |
| CN114462040A (en) | Malicious software detection model training method, malicious software detection method and malicious software detection device | |
| CN110532304B (en) | Data processing method and device, computer readable storage medium and electronic device | |
| CN110414845B (en) | Risk assessment method and device for target transaction | |
| JP7274465B2 (en) | Information processing equipment | |
| CN107506600B (en) | Cancer type prediction method and device based on methylation data | |
| JP6793598B2 (en) | Inspection support equipment, inspection support methods, and inspection support programs | |
| Bayma et al. | Fault diagnosis methodology based on nonlinear system modelling and frequency analysis | |
| CN116740900B (en) | SVM-based power construction early warning method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |