CN117113218A - Visual data analysis system and method thereof - Google Patents
Visual data analysis system and method thereof Download PDFInfo
- Publication number
- CN117113218A CN117113218A CN202310978238.0A CN202310978238A CN117113218A CN 117113218 A CN117113218 A CN 117113218A CN 202310978238 A CN202310978238 A CN 202310978238A CN 117113218 A CN117113218 A CN 117113218A
- Authority
- CN
- China
- Prior art keywords
- time sequence
- training
- feature vector
- noise
- vector
- 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
- 238000007405 data analysis Methods 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims abstract description 42
- 230000000007 visual effect Effects 0.000 title claims abstract description 20
- 239000013598 vector Substances 0.000 claims abstract description 228
- 238000004458 analytical method Methods 0.000 claims abstract description 16
- 238000012549 training Methods 0.000 claims description 102
- 239000011159 matrix material Substances 0.000 claims description 27
- 238000013527 convolutional neural network Methods 0.000 claims description 22
- 230000007246 mechanism Effects 0.000 claims description 15
- 238000000605 extraction Methods 0.000 claims description 9
- 238000007781 pre-processing Methods 0.000 claims description 9
- 230000001419 dependent effect Effects 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 abstract description 18
- 238000013473 artificial intelligence Methods 0.000 abstract description 5
- 238000013135 deep learning Methods 0.000 abstract description 5
- 238000005516 engineering process Methods 0.000 abstract description 5
- 230000006870 function Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 10
- 238000012544 monitoring process Methods 0.000 description 7
- 238000009826 distribution Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 239000000284 extract Substances 0.000 description 4
- 238000010606 normalization Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000011176 pooling Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000001276 controlling effect Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 230000007257 malfunction Effects 0.000 description 2
- 230000002633 protecting effect Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000005355 Hall effect Effects 0.000 description 1
- 238000012300 Sequence Analysis Methods 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000009022 nonlinear effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
- G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Probability & Statistics with Applications (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A visual data analysis system and method thereof are disclosed. The method comprises the steps of firstly obtaining temperature values, current values and noise values of a plurality of preset time points of monitored secondary equipment in a preset time period, then extracting context-related feature vectors from the temperature values, the current values and the noise values of the preset time points, and finally determining an analysis result based on the context-related feature vectors. By the mode, temperature, current and noise data of the secondary equipment at different time points are comprehensively utilized, and abnormal conditions of the secondary equipment are monitored and identified by combining deep learning and artificial intelligence technology, so that safe and stable operation of the power system is ensured.
Description
Technical Field
The present disclosure relates to the field of data analysis, and more particularly, to a visualized data analysis system and method thereof.
Background
The secondary electric power equipment is auxiliary equipment for monitoring, measuring, controlling, protecting and adjusting primary equipment in an electric power system, namely equipment which is not directly connected with electric energy generation, the types of secondary equipment are various, and a secondary equipment system in a power grid is more complicated.
The operation state of the secondary power equipment directly affects the safe, economical and stable operation of the power system. If the secondary equipment is out of order or abnormal, the power grid accident can be caused, and huge economic loss and social influence are caused. Therefore, a data analysis scheme for the secondary equipment is expected, and abnormal conditions of the secondary equipment are timely found and processed, so that the reliability of the power grid is guaranteed.
Disclosure of Invention
In view of this, the present disclosure proposes a visual data analysis system and method thereof, which comprehensively utilizes temperature, current and noise data of secondary devices at different time points, and combines deep learning and artificial intelligence techniques to monitor and identify abnormal conditions of the secondary devices of the power, so as to ensure safe and stable operation of the power system.
According to an aspect of the present disclosure, there is provided a visualized data analysis method including:
acquiring temperature values, current values and noise values of the monitored secondary equipment at a plurality of preset time points in a preset time period;
extracting context-associated feature vectors from the temperature values, the current values, and the noise values of the plurality of predetermined time points; and
and determining an analysis result based on the context-associated feature vector.
According to another aspect of the present disclosure, there is provided a visualized data analysis system comprising:
the data acquisition module is used for acquiring temperature values, current values and noise values of the monitored secondary equipment at a plurality of preset time points in a preset time period;
a feature vector extraction module for extracting context-associated feature vectors from the temperature values, the current values, and the noise values of the plurality of predetermined time points; and
and the analysis result determining module is used for determining an analysis result based on the context-associated feature vector.
According to an embodiment of the present disclosure, it first acquires temperature values, current values, and noise values of a monitored secondary device at a plurality of predetermined time points within a predetermined period of time, then extracts context-associated feature vectors from the temperature values, current values, and noise values of the plurality of predetermined time points, and finally determines an analysis result based on the context-associated feature vectors. By the mode, temperature, current and noise data of the secondary equipment at different time points are comprehensively utilized, and abnormal conditions of the secondary equipment are monitored and identified by combining deep learning and artificial intelligence technology, so that safe and stable operation of the power system is ensured.
Other features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments, which proceeds with reference to the accompanying drawings.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate exemplary embodiments, features and aspects of the present disclosure and together with the description, serve to explain the principles of the disclosure.
Fig. 1 illustrates a flow chart of a method of visual data analysis according to an embodiment of the present disclosure.
Fig. 2 shows an architectural diagram of a visualized data analysis method according to an embodiment of the disclosure.
Fig. 3 shows a flowchart of sub-step S120 of a visualized data analysis method according to an embodiment of the present disclosure.
Fig. 4 shows a flowchart of sub-step S130 of a visualized data analysis method according to an embodiment of the disclosure.
Fig. 5 shows a flowchart of training steps further included in a visualized data analysis method according to an embodiment of the present disclosure.
Fig. 6 illustrates a block diagram of a visualized data analysis system according to an embodiment of the disclosure.
Fig. 7 illustrates an application scenario diagram of a visualized data analysis method according to an embodiment of the disclosure.
Detailed Description
The following description of the embodiments of the present disclosure will be made clearly and fully with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some, but not all embodiments of the disclosure. All other embodiments, which can be made by one of ordinary skill in the art without undue burden based on the embodiments of the present disclosure, are also within the scope of the present disclosure.
As used in this disclosure and in the claims, the terms "a," "an," "the," and/or "the" are not specific to a singular, but may include a plurality, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that the steps and elements are explicitly identified, and they do not constitute an exclusive list, as other steps or elements may be included in a method or apparatus.
Various exemplary embodiments, features and aspects of the disclosure will be described in detail below with reference to the drawings. In the drawings, like reference numbers indicate identical or functionally similar elements. Although various aspects of the embodiments are illustrated in the accompanying drawings, the drawings are not necessarily drawn to scale unless specifically indicated.
In addition, numerous specific details are set forth in the following detailed description in order to provide a better understanding of the present disclosure. It will be understood by those skilled in the art that the present disclosure may be practiced without some of these specific details. In some instances, methods, means, elements, and circuits well known to those skilled in the art have not been described in detail in order not to obscure the present disclosure.
The secondary power equipment refers to auxiliary equipment for monitoring, measuring, controlling, protecting and regulating primary equipment of the power system. The electric secondary equipment is not directly connected with the electric energy generation, but the signals and the data of the primary equipment are collected, processed and analyzed to ensure the safe, economical and stable operation of the electric power system. The types of the electric power secondary equipment are many, including but not limited to relays, protection devices, automation devices, measuring instruments, communication equipment and the like, the electric power secondary equipment plays an important role in an electric power system, the operation state of the electric power system is monitored and controlled, and the reliability and the stability of a power grid are guaranteed.
Aiming at the technical problems, the technical concept of the application is to comprehensively utilize the temperature, current and noise data of the secondary equipment at different time points, and monitor and identify the abnormal condition of the secondary equipment by combining deep learning and artificial intelligence technology so as to ensure the safe and stable operation of the power system.
Fig. 1 illustrates a flow chart of a method of visual data analysis according to an embodiment of the present disclosure. Fig. 2 shows an architectural diagram of a visualized data analysis method according to an embodiment of the disclosure. As shown in fig. 1 and 2, a visual data analysis method according to an embodiment of the present disclosure includes the steps of: s110, acquiring temperature values, current values and noise values of the monitored secondary equipment at a plurality of preset time points in a preset time period; s120, extracting context-associated feature vectors from the temperature values, the current values and the noise values of the plurality of preset time points; and S130, determining an analysis result based on the context-associated feature vector.
More specifically, in step S110, temperature values, current values, and noise values of the monitored secondary device at a plurality of predetermined time points within a predetermined period of time are acquired. These data may provide information on the performance and operating status of the monitored secondary device. That is, it can be known from the temperature data whether the device is overheated or supercooled, the current can be used to determine whether the device is operating normally and whether there is a current leakage problem, etc., and the noise value can provide status information about the device during operation. Accordingly, the temperature value, the current value and the noise value of the monitored secondary equipment at a plurality of preset time points in a preset time period are acquired, data acquisition can be performed through sensors, corresponding sensor equipment such as a temperature sensor, a current sensor, a noise sensor and the like are installed and connected to the monitored secondary equipment, and the sensors can measure the data such as the temperature, the current and the noise in real time and convert the data into electric signals for acquisition. The temperature sensor is used for measuring the temperature of the secondary equipment, such as a thermistor (such as a thermistor NTC and PTC), a thermocouple, an infrared sensor and the like, and can convert the temperature into an electric signal to be recorded or transmitted to the monitoring system; the current sensor is used for measuring the current of the secondary equipment, and can be used for recording and monitoring by using a current transformer (such as a current transformer CT), a Hall effect sensor, a resistance sensor and the like; noise sensors are used to measure the level of noise produced by secondary equipment, which can be converted into electrical signals by microphones or piezoelectric sensors, etc., and are commonly used to detect vibration, mechanical operating conditions, environmental noise, etc. of the equipment. These sensors are typically connected directly to or mounted near the secondary equipment to accurately measure temperature, current, noise, and the like. The choice of sensor depends on the specific application requirements and monitoring requirements. The accuracy, reliability and stability of the sensor are ensured when the sensor is installed and used.
It should be appreciated that in other examples of the application, a data logger may also be used to record data such as temperature, current and noise of the secondary device being monitored, the data logger may perform data collection at predetermined points in time over a predetermined period of time and store the data in an internal memory or external storage medium; or, the data of the monitored secondary equipment can be acquired by using a remote monitoring system, and the data such as temperature, current, noise and the like can be acquired in real time by connecting the monitored equipment with the monitoring system, and the system usually uses a sensor and a communication device to acquire and transmit the data.
More specifically, in step S120, a context-associated feature vector is extracted from the temperature values, the current values, and the noise values at the plurality of predetermined time points. Accordingly, in one possible implementation, as shown in fig. 3, extracting the context-associated feature vector from the temperature values, the current values, and the noise values at the plurality of predetermined time points includes: s121, respectively performing data preprocessing on the temperature values, the current values and the noise values of the plurality of preset time points to obtain a temperature time sequence input vector, a current time sequence input vector and a noise time sequence input vector; s122, performing time sequence feature extraction on the temperature time sequence input vector, the current time sequence input vector and the noise time sequence input vector based on a deep convolutional neural network model to obtain a temperature time sequence feature vector, a current time sequence feature vector and a noise time sequence feature vector; and S123, fusing the temperature time sequence feature vector, the current time sequence feature vector and the noise time sequence feature vector to obtain the context correlation feature vector.
Accordingly, in one possible implementation manner, the data preprocessing is performed on the temperature values, the current values and the noise values at the plurality of predetermined time points to obtain a temperature time sequence input vector, a current time sequence input vector and a noise time sequence input vector, which includes: and arranging the temperature values, the current values and the noise values of the plurality of preset time points into the temperature time sequence input vector, the current time sequence input vector and the noise time sequence input vector according to time dimensions. That is, the time-series discrete data of the temperature value, the current value, and the noise value are converted into the structured temperature-series input vector, the current-series input vector, and the noise-series input vector. Meanwhile, the vector obtained according to the operation mode of the time dimension arrangement can reflect the change and trend of the parameters in the time dimension, and an important data source is provided for subsequent analysis.
It should be appreciated that in one specific example, converting temperature values, current values, and noise values at a plurality of predetermined time points into time-series input vectors, data preprocessing may be performed as follows: 1. the time dimension arrangement, namely arranging the temperature value, the current value and the noise value of each time point according to the time sequence to form a time sequence, wherein the sequence of the time sequence is required to be ensured to be consistent with the time sequence of the acquired data; 2. normalization, the normalization of the values of each parameter (temperature, current, noise) scales it to a uniform range, for example [0,1], which eliminates the dimensional differences between the different parameters, so that they can be compared and analyzed; 3. dividing the time sequence into windows with fixed sizes, wherein each window comprises continuous time points, the size of the window can be set according to requirements, for example, each window comprises 5 time points, and thus time sequence data can be converted into a matrix form; 4. feature extraction, for the data in each window, some statistical features, such as mean, variance, maximum, minimum, etc., can be extracted, which can reflect the overall features of the data in each window; 5. the vectors are constructed and the features extracted in each window are combined into a vector which is used as the representation of the window, so that a temperature time sequence input vector, a current time sequence input vector and a noise time sequence input vector are obtained. Through the above steps, the temperature values, current values, and noise values at a plurality of predetermined points in time can be converted into structured time series input vectors, which can be used for subsequent analysis and modeling tasks.
Accordingly, in one possible implementation, performing timing feature extraction on the temperature timing input vector, the current timing input vector, and the noise timing input vector based on a deep convolutional neural network model to obtain a temperature timing feature vector, a current timing feature vector, and a noise timing feature vector, including: and respectively passing the temperature time sequence input vector, the current time sequence input vector and the noise time sequence input vector through a time sequence feature extractor based on a one-dimensional convolutional neural network model to obtain the temperature time sequence feature vector, the current time sequence feature vector and the noise time sequence feature vector. Changes in temperature, current and noise data at different points in time are considered to reflect the operating state of the device. Specifically, the temperature is typically maintained within a certain range during normal operation of the device. However, when there is a malfunction or abnormality in the apparatus, an abnormal change in temperature may occur. High temperatures may indicate equipment overload or failure, while low temperatures may suggest problems with energy transfer or other abnormal conditions. The current value is related to the load of the device, the power consumption situation and the stability of the power supply. Abnormal current fluctuations may suggest that the device is overloaded, power problems, shorted, etc. Noise refers to an undesired sound or signal produced by the device during operation, and abnormal noise may be caused by loosening, wear, or other malfunctions of the device's components. Therefore, in the technical scheme of the application, the one-dimensional convolutional neural network model is expected to be utilized for carrying out time sequence analysis and feature extraction on temperature, current and noise data, so as to capture time sequence implicit change feature distribution information in the data.
It should be understood that a one-dimensional convolutional neural network (1D CNN) is a deep learning model for processing data having a time-series structure, and is mainly applied to processing one-dimensional sequence data, such as time-series, audio signals, text data, etc., unlike a conventional convolutional neural network (2D CNN) for processing image data. The one-dimensional convolutional neural network model extracts features in the input data through a convolutional operation and a pooling operation. The convolution operation uses a filter (also called a convolution kernel) to slide over the input data to calculate a characteristic representation of each location. This captures local patterns and features in the input data. The pooling operation is used for reducing the dimension of the features and extracting the most remarkable features. In one-dimensional convolutional neural network models, multiple convolutional layers and pooling layers are typically used to progressively extract higher level features. After the convolutional layer, an activation function (e.g., reLU) may be added to introduce non-linear properties. Finally, the extracted features are mapped to the desired output categories through the fully connected layer. The one-dimensional convolutional neural network model is suitable for many tasks, such as text classification, voice recognition, time sequence prediction and the like, and can automatically learn time sequence characteristics in input data, so that the performance and generalization capability of the model are improved.
Accordingly, in one possible implementation, fusing the temperature timing feature vector, the current timing feature vector, and the noise timing feature vector to obtain the context-dependent feature vector includes: and taking the current time sequence feature vector as a query vector, taking the temperature time sequence feature vector as a key vector and taking the noise time sequence feature vector as a value vector, and fusing the temperature time sequence feature vector, the current time sequence feature vector and the noise time sequence feature vector by using a self-attention mechanism to obtain the context correlation feature vector. That is, a self-attention mechanism is utilized to establish a contextual relationship among the current timing feature vector, and the noise timing feature vector. Specifically, the current timing feature vector is taken as a query vector Q, the temperature timing feature vector is taken as a key vector K, and the noise timing feature vector is taken as a value vector V, and then the context correlation feature vector is obtained by calculation according to the following formula:
by introducing the context information in this way, the temperature, current and noise characteristic interaction and mutual influence relations at different time points are integrated into the context correlation characteristic vector, so that the model can more accurately understand the running state and abnormal condition of the equipment.
It is worth mentioning that the self-attention mechanism (self-attention mechanism) is a mechanism for capturing context associations in sequence data, commonly used in natural language processing and sequence modeling tasks. The method can dynamically calculate the weight of each position according to the information of different positions in the sequence, thereby realizing the self-adaptive adjustment of the attention degree of different positions. In the self-attention mechanism, the attention weight (attention weights) is obtained by calculating the similarity between the query vector (query vector) and the key vector (key vector), and then the attention weight and the value vector (value vector) are weighted and summed to obtain the context-associated feature vector (contextualized feature vector). Specifically, with the current timing feature vector as the query vector, the temperature timing feature vector as the key vector, and the noise timing feature vector as the value vector, they can be fused using a self-attention mechanism. The following is the basic steps of the self-attention mechanism: 1. calculating the similarity, namely obtaining the attention weight by calculating the similarity between the query vector and the key vector, wherein a common method for calculating the similarity is to use dot products or bilinear functions; 2. normalizing the attention weights, for which normalization processing may be performed such that their sum is equal to 1, which can ensure that the distribution of the attention weights conforms to the nature of the probability distribution; 3. and (3) carrying out weighted summation on the normalized attention weight and the value vector to obtain a context-associated feature vector, wherein the weighted summation can be realized in a matrix multiplication mode. By fusing the self-attention mechanism, the association information between different time sequence feature vectors can be captured, and the context association feature vectors are obtained, which is helpful to better represent the overall information and mode of the sequence data. The advantage of the self-care mechanism is that it is able to adaptively learn the correlation between different positions and can be computed in parallel, suitable for processing long sequence data.
More specifically, in step S130, an analysis result is determined based on the context-associated feature vector. Accordingly, in one possible implementation, as shown in fig. 4, determining the analysis result based on the context-associated feature vector includes: s131, the context association feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the running state of the monitored secondary equipment is normal or not; and S132, displaying the classification result on a display screen. The classifier can learn the mapping relation between the feature vector and the classification label in the training process. That is, by training on known device states, the classifier is able to learn the feature patterns and differences between normal and abnormal states. In this way, in the inference phase, the input context-associated feature vector may be automatically mapped into a corresponding classification label, i.e. "the operation state of the monitored secondary device is normal" or "the operation state of the monitored secondary device is abnormal". In practical application, the output of the classifier, namely the classification result, is displayed in a display screen. Thus, people can intuitively know the output of the machine learning model, namely, whether the running state of the equipment is normal or not can be quickly known. The visual feedback can help people to monitor and identify problems of secondary equipment better, so that measures can be taken in time to maintain or adjust, reliability and safety of a power grid are improved, and risks of accidental loss are reduced.
It should be appreciated that a classifier is a machine learning model that is used to classify input data into different categories or labels. The role of the classifier is to learn the classification rules and classifier using a given class, known training data, and then classify (or predict) the unknown data. Logistic regression (logistics), SVM, etc. are commonly used to solve the classification problem, and for multi-classification problems (multi-class classification), logistic regression or SVM can be used as well, but multiple bi-classifications are required to compose multiple classifications, but this is error-prone and inefficient, and the commonly used multi-classification method is the Softmax classification function. Common classifiers include Support Vector Machines (SVMs), decision trees, random forests, neural networks, etc.
Accordingly, in one possible implementation manner, the context-associated feature vector is passed through a classifier to obtain a classification result, where the classification result is used to indicate whether the operation state of the monitored secondary device is normal, and the method includes: performing full-connection coding on the context-associated feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
Accordingly, in one possible implementation, in step S132, the classification result will be displayed on the display screen. This may be achieved by outputting corresponding text or icons on the display screen. For example, text information such as "normal" or "abnormal" may be displayed, or green and red icons may be used to represent normal and abnormal states.
Further, the visual data analysis method further comprises a training step: and training the time sequence feature extractor and the classifier based on the one-dimensional convolutional neural network model. It should be appreciated that for the training step of the timing feature extractor and classifier based on a one-dimensional convolutional neural network model, it generally comprises the following stages: 1. data preparation, collecting and sorting a data set for training, wherein the data set comprises marked time sequence data samples, and each sample has a corresponding running state label (normal or abnormal); 2. data preprocessing, namely preprocessing data so as to facilitate training and use of a model, wherein the preprocessing can comprise data cleaning, normalization, standardization, smoothing and the like, and the aim of preprocessing is to eliminate noise and improve the interpretability of the data and the stability of the model; 3. the feature extractor is trained by using a one-dimensional convolutional neural network model as the feature extractor and inputting time sequence data into the model for training. In the training process, the model extracts useful time sequence features through learning patterns and rules in data, and the training aim is to minimize a loss function of a feature extractor so that the feature extractor can accurately capture the features related to the running state; 4. training a classifier, namely inputting the extracted time sequence features into the classifier for training by using a trained feature extractor, wherein the classifier can be a traditional machine learning algorithm or other neural network models, and in the training process, the classifier classifies the time sequence features by learning the relationship between the features and the running state, and the training aim is to minimize the loss function of the classifier so that the time sequence features can be accurately classified as normal or abnormal; 5. and evaluating the model, namely evaluating the trained model by using a verification set or a cross verification method and the like, wherein the evaluation indexes can comprise accuracy, recall rate, precision, F1 score and the like. The purpose of the evaluation is to examine the performance and generalization ability of the model to determine if adjustments or improvements are needed.
More specifically, as shown in fig. 5, in one specific example, the training step includes: s210, training data are acquired, wherein the training data comprise training temperature values, training current values and training noise values of a plurality of preset time points of the monitored secondary equipment in a preset time period, and real values of whether the running state of the monitored secondary equipment is normal or not; s220, training temperature values, training current values and training noise values of the plurality of preset time points are respectively arranged into training temperature time sequence input vectors, training current time sequence input vectors and training noise time sequence input vectors according to time dimensions; s230, respectively passing the training temperature time sequence input vector, the training current time sequence input vector and the training noise time sequence input vector through the time sequence feature extractor based on the one-dimensional convolutional neural network model to obtain a training temperature time sequence feature vector, a training current time sequence feature vector and a training noise time sequence feature vector; s240, using the training current time sequence feature vector as a query vector, the training temperature time sequence feature vector as a key vector and the training noise time sequence feature vector as a value vector, and using a self-attention mechanism to fuse the training temperature time sequence feature vector, the training current time sequence feature vector and the training noise time sequence feature vector to obtain a training context correlation feature vector; s250, the training context associated feature vectors pass through a classifier to obtain a classification loss function value; and S260, training the time sequence feature extractor based on the one-dimensional convolutional neural network model and the classifier based on the classification loss function value and through gradient descent direction propagation, wherein in each round of iteration of the training, the weight matrix of the classifier is subjected to half-space structuring constraint iteration of weight intrinsic support.
In the technical scheme of the application, when the training temperature time sequence feature vector, the training current time sequence feature vector and the training noise time sequence feature vector are fused through a self-attention mechanism to obtain the training context correlation feature vector, the self-attention mechanism has different feature fitting directions, such as over-fitting, under-fitting and normal fitting, aiming at the training temperature time sequence feature vector, the training current time sequence feature vector and the training noise time sequence feature vector due to the fact that the training temperature time sequence feature vector, the training current time sequence feature vector and the training noise time sequence feature vector respectively express local correlation features of temperature values, current values and noise values in time sequence, and the difference of distribution of source data in time sequence can further cause the difference of feature distribution among the training temperature time sequence feature vector, the training current time sequence feature vector and the training noise time sequence feature vector.
In this way, when the training context-associated feature vector is classified by the classifier, the part of the training context-associated feature vector corresponding to the training temperature time sequence feature vector, the training current time sequence feature vector and the training noise time sequence feature vector respectively have different weight fitting directions relative to the part corresponding to the weight matrix of the classifier, so that the overall feature distribution of the training context-associated feature vector has the problem of poor convergence relative to the weight matrix of the classifier, thereby affecting the training speed of the classifier.
Based on this, the applicant of the present application associates feature vectors in the context, e.g. denoted asThe weight matrix of each classifier is e.g. denoted +.>In the iterative process of (1), weight matrix +.>And performing half-space structuring constraint of weight intrinsic support.
Accordingly, in one possible implementation, in each iteration of the training, performing a half-space structured constraint iteration of weight eigen support on the weight matrix of the classifier includes: performing half-space structuring constraint iteration of weight intrinsic support on the weight matrix of the classifier by using the following constraint formula; wherein, the constraint formula is:wherein (1)>Is a weight matrix of the classifier, +.>Is a matrix->Eigenvector of eigenvalues of (a)>Representing the transposed matrix of the matrix +.>Is the training context associated feature vector,is a weight matrix after the half-space structuring constraint of the weight intrinsic support is carried out, and the weight matrix is +.>Representing multiplication->Representing matrix addition.
Here, the weighted eigen-supported half-space structuring is constrained by the weight matrix of the classifierIs associated with the training context to be classified>Is used as support for the correlation integration of the weight matrix +.>Expressed feature vector +_for associating with the training context to be classified>Is coupled to a high-dimensional manifold of (a)Half-space (half-space) is subject to structural support constraints of hyperplane as decision boundary, such that the training context associated feature vector +.>Can be in the weight matrix +.>The represented open domain of the half space effectively converges with respect to the hyperplane, thereby improving the training speed of the classifier.
In summary, according to the visual data analysis method disclosed by the embodiment of the disclosure, temperature, current and noise data of the secondary equipment at different time points can be comprehensively utilized, and abnormal conditions of the secondary equipment can be monitored and identified by combining deep learning and artificial intelligence technology, so that safe and stable operation of the power system is ensured.
Fig. 6 shows a block diagram of a visualized data analysis system 100 according to an embodiment of the disclosure. As shown in fig. 6, a visualized data analysis system 100 according to an embodiment of the present disclosure includes: a data acquisition module 110, configured to acquire temperature values, current values, and noise values of the monitored secondary device at a plurality of predetermined time points within a predetermined period of time; a feature vector extraction module 120 for extracting context-associated feature vectors from the temperature values, the current values, and the noise values of the plurality of predetermined time points; and an analysis result determining module 130, configured to determine an analysis result based on the context-associated feature vector.
Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the above-described visual data analysis system 100 have been described in detail in the above description of the visual data analysis method with reference to fig. 1 to 5, and thus, repetitive descriptions thereof will be omitted.
As described above, the visualized data analysis system 100 according to the embodiment of the present disclosure may be implemented in various wireless terminals, for example, a server or the like having a visualized data analysis algorithm. In one possible implementation, the visualized data analysis system 100 according to embodiments of the disclosure may be integrated into the wireless terminal as one software module and/or hardware module. For example, the visual data analysis system 100 may be a software module in the operating system of the wireless terminal or may be an application developed for the wireless terminal; of course, the visual data analysis system 100 could equally be one of many hardware modules of the wireless terminal.
Alternatively, in another example, the visual data analysis system 100 and the wireless terminal may be separate devices, and the visual data analysis system 100 may be connected to the wireless terminal through a wired and/or wireless network and communicate the interaction information in accordance with a agreed data format.
Fig. 7 illustrates an application scenario diagram of a visualized data analysis method according to an embodiment of the disclosure. As shown in fig. 7, in this application scenario, first, temperature values (e.g., D1 illustrated in fig. 7), current values (e.g., D2 illustrated in fig. 7), and noise values (e.g., D3 illustrated in fig. 7) of a plurality of predetermined time points of the monitored secondary device within a predetermined period of time are acquired, and then the temperature values, current values, and noise values of the plurality of predetermined time points are input to a server (e.g., S illustrated in fig. 7) in which a visualized data analysis algorithm is deployed, wherein the server is able to process the temperature values, current values, and noise values of the plurality of predetermined time points using the visualized data analysis algorithm to obtain a classification result for indicating whether the operation state of the monitored secondary device is normal.
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The foregoing description of the embodiments of the present disclosure has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the various embodiments described. The terminology used herein was chosen in order to best explain the principles of the embodiments, the practical application, or the improvement of technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.
Claims (10)
1. A method of visual data analysis, comprising:
acquiring temperature values, current values and noise values of the monitored secondary equipment at a plurality of preset time points in a preset time period;
extracting context-associated feature vectors from the temperature values, the current values, and the noise values of the plurality of predetermined time points; and
and determining an analysis result based on the context-associated feature vector.
2. The visualized data analysis method of claim 1, wherein extracting context-dependent feature vectors from the temperature, current, and noise values for the plurality of predetermined points in time comprises:
respectively carrying out data preprocessing on the temperature values, the current values and the noise values of the plurality of preset time points to obtain a temperature time sequence input vector, a current time sequence input vector and a noise time sequence input vector;
performing time sequence feature extraction on the temperature time sequence input vector, the current time sequence input vector and the noise time sequence input vector based on a deep convolutional neural network model to obtain a temperature time sequence feature vector, a current time sequence feature vector and a noise time sequence feature vector; and
and fusing the temperature time sequence feature vector, the current time sequence feature vector and the noise time sequence feature vector to obtain the context correlation feature vector.
3. The visualized data analysis method of claim 2, wherein the data preprocessing of the temperature values, the current values, and the noise values at the plurality of predetermined time points to obtain a temperature timing input vector, a current timing input vector, and a noise timing input vector, respectively, comprises:
and arranging the temperature values, the current values and the noise values of the plurality of preset time points into the temperature time sequence input vector, the current time sequence input vector and the noise time sequence input vector according to time dimensions.
4. The visualized data analysis method of claim 3 wherein performing a time series feature extraction on the temperature time series input vector, the current time series input vector, and the noise time series input vector based on a deep convolutional neural network model to obtain a temperature time series feature vector, a current time series feature vector, and a noise time series feature vector comprises:
and respectively passing the temperature time sequence input vector, the current time sequence input vector and the noise time sequence input vector through a time sequence feature extractor based on a one-dimensional convolutional neural network model to obtain the temperature time sequence feature vector, the current time sequence feature vector and the noise time sequence feature vector.
5. The method of visual data analysis of claim 4, wherein fusing the temperature timing feature vector, the current timing feature vector, and the noise timing feature vector to obtain the context-dependent feature vector comprises:
and taking the current time sequence feature vector as a query vector, taking the temperature time sequence feature vector as a key vector and taking the noise time sequence feature vector as a value vector, and fusing the temperature time sequence feature vector, the current time sequence feature vector and the noise time sequence feature vector by using a self-attention mechanism to obtain the context correlation feature vector.
6. The visualized data analysis method of claim 5 wherein determining an analysis result based on the contextually relevant feature vector comprises:
the context association feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the running state of the monitored secondary equipment is normal or not; and
and displaying the classification result on a display screen.
7. The method of visual data analysis of claim 6, further comprising the step of training: and training the time sequence feature extractor and the classifier based on the one-dimensional convolutional neural network model.
8. The visualized data analysis method of claim 7 wherein the training step comprises:
acquiring training data, wherein the training data comprises training temperature values, training current values and training noise values of a plurality of preset time points of the monitored secondary equipment in a preset time period, and a true value of whether the running state of the monitored secondary equipment is normal or not;
respectively arranging the training temperature values, the training current values and the training noise values of the plurality of preset time points into training temperature time sequence input vectors, training current time sequence input vectors and training noise time sequence input vectors according to time dimensions;
respectively passing the training temperature time sequence input vector, the training current time sequence input vector and the training noise time sequence input vector through the time sequence feature extractor based on the one-dimensional convolutional neural network model to obtain a training temperature time sequence feature vector, a training current time sequence feature vector and a training noise time sequence feature vector;
taking the training current time sequence feature vector as a query vector, taking the training temperature time sequence feature vector as a key vector and taking the training noise time sequence feature vector as a value vector, and fusing the training temperature time sequence feature vector, the training current time sequence feature vector and the training noise time sequence feature vector by using a self-attention mechanism to obtain a training context correlation feature vector;
the training context associated feature vectors pass through a classifier to obtain a classification loss function value; and
training the time sequence feature extractor based on the one-dimensional convolutional neural network model and the classifier based on the classification loss function value and through gradient descent direction propagation, wherein in each round of iteration of the training, a weight matrix of the classifier is subjected to half-space structuring constraint iteration of weight intrinsic support.
9. The visualized data analysis method of claim 8, wherein, in each iteration of the training, performing a half-space structured constraint iteration of weight eigen support on the weight matrix of the classifier comprises:
performing half-space structuring constraint iteration of weight intrinsic support on the weight matrix of the classifier by using the following constraint formula;
wherein, the constraint formula is:wherein (1)>Is a weight matrix of the classifier, +.>Is a matrix->Eigenvector of eigenvalues of (a)>Representing the transposed matrix of the matrix +.>Is the training context associated feature vector, < >>Is a weight matrix after the half-space structuring constraint of the weight intrinsic support is carried out, and the weight matrix is +.>Representing multiplication->Representing matrix addition.
10. A visual data analysis system, comprising:
the data acquisition module is used for acquiring temperature values, current values and noise values of the monitored secondary equipment at a plurality of preset time points in a preset time period;
a feature vector extraction module for extracting context-associated feature vectors from the temperature values, the current values, and the noise values of the plurality of predetermined time points; and
and the analysis result determining module is used for determining an analysis result based on the context-associated feature vector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310978238.0A CN117113218A (en) | 2023-08-04 | 2023-08-04 | Visual data analysis system and method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310978238.0A CN117113218A (en) | 2023-08-04 | 2023-08-04 | Visual data analysis system and method thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117113218A true CN117113218A (en) | 2023-11-24 |
Family
ID=88811929
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310978238.0A Pending CN117113218A (en) | 2023-08-04 | 2023-08-04 | Visual data analysis system and method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117113218A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117388893A (en) * | 2023-12-11 | 2024-01-12 | 深圳市移联通信技术有限责任公司 | Multi-device positioning system based on GPS |
| CN117579513A (en) * | 2024-01-16 | 2024-02-20 | 北京中科网芯科技有限公司 | Visual operation and maintenance system and method for convergence and diversion equipment |
| CN117590223A (en) * | 2024-01-18 | 2024-02-23 | 南京飞腾电子科技有限公司 | Online monitoring system and method for circuit breaker |
| CN118254624A (en) * | 2024-04-26 | 2024-06-28 | 深圳市国汇计量质量检测有限公司 | Visual early warning system and method based on big data |
-
2023
- 2023-08-04 CN CN202310978238.0A patent/CN117113218A/en active Pending
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117388893A (en) * | 2023-12-11 | 2024-01-12 | 深圳市移联通信技术有限责任公司 | Multi-device positioning system based on GPS |
| CN117388893B (en) * | 2023-12-11 | 2024-03-12 | 深圳市移联通信技术有限责任公司 | Multi-device positioning system based on GPS |
| CN117579513A (en) * | 2024-01-16 | 2024-02-20 | 北京中科网芯科技有限公司 | Visual operation and maintenance system and method for convergence and diversion equipment |
| CN117579513B (en) * | 2024-01-16 | 2024-04-02 | 北京中科网芯科技有限公司 | Visual operation and maintenance system and method for convergence and diversion equipment |
| CN117590223A (en) * | 2024-01-18 | 2024-02-23 | 南京飞腾电子科技有限公司 | Online monitoring system and method for circuit breaker |
| CN117590223B (en) * | 2024-01-18 | 2024-04-30 | 南京飞腾电子科技有限公司 | Online monitoring system and method for circuit breaker |
| CN118254624A (en) * | 2024-04-26 | 2024-06-28 | 深圳市国汇计量质量检测有限公司 | Visual early warning system and method based on big data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117113218A (en) | Visual data analysis system and method thereof | |
| WO2021098634A1 (en) | Non-intrusive data analytics system for adaptive intelligent condition monitoring of lifts | |
| CN116639010B (en) | Intelligent control system and method for charging pile | |
| Khelif et al. | Direct remaining useful life estimation based on support vector regression | |
| CN116625438B (en) | Gas pipe network safety on-line monitoring system and method thereof | |
| CN109462521B (en) | Network flow abnormity detection method suitable for source network load interaction industrial control system | |
| Yuan et al. | Learning-based real-time event identification using rich real PMU data | |
| CN116683648B (en) | Intelligent power distribution cabinet and control system thereof | |
| CN116893336B (en) | High-low voltage switching device and method thereof | |
| CN116934304B (en) | Intelligent power distribution room equipment operation maintenance management system and method thereof | |
| CN117784710B (en) | Remote state monitoring system and method for numerical control machine tool | |
| Shao et al. | A Multichannel LSTM‐CNN Method for Fault Diagnosis of Chemical Process | |
| CN116930042B (en) | Building waterproof material performance detection equipment and method | |
| Huang et al. | Metric learning-based fault diagnosis and anomaly detection for industrial data with intraclass variance | |
| CN116929815A (en) | Equipment working state monitoring system and method based on Internet of things | |
| CN116824517B (en) | Substation operation and maintenance safety control system based on visualization | |
| CN117289736A (en) | Temperature control method and system for high-temperature anti-corrosion storage tank | |
| CN117689278B (en) | Construction quality intelligent management system and method | |
| CN117614060A (en) | Wireless charging method | |
| CN117192416A (en) | Battery monitoring system and method based on BMS system | |
| CN117621875B (en) | Electric automobile charging management system and method based on low-temperature compensation | |
| CN117951646A (en) | Data fusion method and system based on edge cloud | |
| CN117856442A (en) | Power equipment remote monitoring system and method based on big data | |
| CN117831743A (en) | Household care health monitoring management system and method based on Internet of things technology | |
| CN117351659A (en) | Hydrogeological disaster monitoring device and monitoring method |
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 |