CN117150421A - Novel low-voltage switch cabinet data monitoring method and system - Google Patents
Novel low-voltage switch cabinet data monitoring method and system Download PDFInfo
- Publication number
- CN117150421A CN117150421A CN202311434792.9A CN202311434792A CN117150421A CN 117150421 A CN117150421 A CN 117150421A CN 202311434792 A CN202311434792 A CN 202311434792A CN 117150421 A CN117150421 A CN 117150421A
- Authority
- CN
- China
- Prior art keywords
- data
- representing
- bald
- iteration
- monitoring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 79
- 238000000034 method Methods 0.000 title claims abstract description 56
- 230000002159 abnormal effect Effects 0.000 claims abstract description 37
- 230000006399 behavior Effects 0.000 claims abstract description 13
- 238000010801 machine learning Methods 0.000 claims abstract description 13
- 230000005856 abnormality Effects 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 9
- 230000002431 foraging effect Effects 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 32
- 238000004422 calculation algorithm Methods 0.000 claims description 24
- 230000036621 balding Effects 0.000 claims description 21
- 235000003642 hunger Nutrition 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 13
- 238000000354 decomposition reaction Methods 0.000 claims description 12
- 238000011161 development Methods 0.000 claims description 12
- 241000511338 Haliaeetus leucocephalus Species 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 10
- 230000007423 decrease Effects 0.000 claims description 7
- 238000013459 approach Methods 0.000 claims description 6
- 238000001228 spectrum Methods 0.000 claims description 6
- 230000036186 satiety Effects 0.000 claims description 4
- 235000019627 satiety Nutrition 0.000 claims description 4
- 206010000117 Abnormal behaviour Diseases 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 3
- 238000003745 diagnosis Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 2
- 230000003247 decreasing effect Effects 0.000 claims 1
- 238000001514 detection method Methods 0.000 abstract description 14
- 238000007405 data analysis Methods 0.000 abstract description 6
- 230000008030 elimination Effects 0.000 abstract description 3
- 238000003379 elimination reaction Methods 0.000 abstract description 3
- 238000006243 chemical reaction Methods 0.000 description 15
- 238000013461 design Methods 0.000 description 11
- 238000002955 isolation Methods 0.000 description 10
- 230000003287 optical effect Effects 0.000 description 10
- 238000004891 communication Methods 0.000 description 9
- 201000004384 Alopecia Diseases 0.000 description 6
- 230000003676 hair loss Effects 0.000 description 6
- 238000012423 maintenance Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000012549 training Methods 0.000 description 5
- 230000002093 peripheral effect Effects 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000010354 integration Effects 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 230000008439 repair process Effects 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 150000001875 compounds Chemical class 0.000 description 2
- 230000003750 conditioning effect Effects 0.000 description 2
- 230000001276 controlling effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000001105 regulatory effect Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 239000013078 crystal Substances 0.000 description 1
- 238000013523 data management Methods 0.000 description 1
- 230000006837 decompression Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012938 design process Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000017525 heat dissipation Effects 0.000 description 1
- 238000007654 immersion Methods 0.000 description 1
- 230000036039 immunity Effects 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 238000013450 outlier detection Methods 0.000 description 1
- 238000013021 overheating Methods 0.000 description 1
- 230000003449 preventive effect Effects 0.000 description 1
- 230000001902 propagating effect Effects 0.000 description 1
- 238000010079 rubber tapping Methods 0.000 description 1
- 239000003381 stabilizer Substances 0.000 description 1
- 230000000087 stabilizing effect Effects 0.000 description 1
- 230000037351 starvation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network
- H02J13/00002—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network characterised by monitoring
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2123/00—Data types
- G06F2123/02—Data types in the time domain, e.g. time-series data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
- G06F2218/10—Feature extraction by analysing the shape of a waveform, e.g. extracting parameters relating to peaks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Abstract
The invention discloses a novel low-voltage switch cabinet data monitoring method and system, comprising the following steps: acquiring monitoring data of a switch cabinet; preprocessing the monitoring data, and eliminating trend items of the monitoring data by adopting a least square method; extracting statistical characteristics and time domain characteristics of the data from the preprocessed monitoring data; the extracted feature data is input into a machine learning model, the input features are analyzed, and abnormal conditions are identified by comparing the normal modes. The trend item elimination operation is carried out on the data collected by the sensor, so that the accuracy and the authenticity of the signal are greatly enhanced, and the accuracy of the data analysis and the abnormal detection result of the switch cabinet is further improved. The problems are optimized by simulating foraging and sailing behaviors of the African bald, the optimizing capability is strong, the convergence speed is high, and the abnormality detection capability and reliability of the system are improved.
Description
Technical Field
The invention belongs to the technical field of low-voltage switch cabinet monitoring, and relates to a novel low-voltage switch cabinet data monitoring method and system.
Background
The low-voltage switch cabinet is a device formed by a plurality of switches and measuring equipment, and can integrate a circuit control and detection device into one cabinet, so that the use safety and the management efficiency are greatly improved, and the device is widely applied to distribution offices and becomes an important carrier for power transmission and allocation. Since the low-voltage switch cabinet is easy to generate various fault data in long-time operation, the fault position and the fault reason of the equipment are rapidly and accurately detected, and the method is a main step for rapidly recovering the power.
Modern power systems have higher requirements on low-voltage switching devices, and conventional low-voltage switching devices only can realize basic switching control functions, as shown in fig. 1, the conventional low-voltage switching devices generally need manual operation to complete various functions, and when a circuit is abnormal, the circuit and related devices need to be checked to find out the cause of the fault. Therefore, it is urgent to design an intelligent low-voltage switch cabinet data monitoring device based on the internet of things. The real-time monitoring and control of the switching equipment are realized through the internet of things technology, and the reliability, the safety and the economical efficiency of the power system can be improved.
In practical application, the traditional state monitoring method is difficult to accurately reflect the state of the equipment because of the interference of various factors in the operation process of the low-voltage switch equipment. In order to solve the problem, various monitoring technologies are required to be combined, so that the state of the low-voltage switch equipment is comprehensively monitored and early-warned. The application provides low-voltage switch equipment based on the internet of things and a monitoring system thereof, which can effectively solve the defects.
Disclosure of Invention
The application aims to provide a novel low-voltage switch cabinet data monitoring method and system, which are used for eliminating trend items of data acquired by a sensor, so that the accuracy and the authenticity of signals are greatly enhanced, and the accuracy of switch cabinet data analysis and abnormal detection results is further improved.
The technical solution for realizing the purpose of the invention is as follows:
a new low-voltage switch cabinet data monitoring method comprises the following steps:
s01: acquiring monitoring data of a switch cabinet;
s02: preprocessing the monitoring data, and eliminating trend items of the monitoring data by adopting a least square method;
s03: extracting statistical characteristics and time domain characteristics of the data from the preprocessed monitoring data;
s04: the extracted feature data is input into a machine learning model, the input features are analyzed, and abnormal conditions are identified by comparing the normal modes.
In a preferred embodiment, the trend term eliminating method in step S02 includes:
s11: let the collected signals beAnd the trend term therein is expressed as polynomial function +.>：
Wherein,；
s12: to find coefficients in trend term functionsConstructing an error energy function by using a least square method>By minimizing the trend term function +.>And signal->The quadratic sum of the errors between the two parameters is used for obtaining the optimal parameters:
regarding this functionPerforming partial derivative operation to obtain ∈>Partial derivatives, which are formed into a composition comprising +.>A system of linear equations, each of which is related to +. >The following variables:
thereby determining the coefficientA trend term expression of the signal is obtained, +.>Representing polynomial orders, deleting the obtained trend term.
In a preferred technical solution, the step S02 further includes denoising by a variation modal decomposition, and the method of the variation modal decomposition includes the following steps:
s21: performing Hilbert transformation on the signals with the trend items eliminated to obtain analytic signals of each IMF component, and extracting single-side frequency spectrums of the signals from the analytic signals:
wherein, the IMF is an intrinsic mode function,representing the unit pulse function +.>Representing imaginary units, ++>The time is represented by the time period of the day,representing convolution operation,/->Is->The IMF component is->A value of time of day;
s22: introducing an exponential weight, adjusting the estimated center frequency of each IMF component, and modulating the spectrum of each IMF component to a corresponding baseband:
s23: constructing a constrained variation model based on the square of the demodulated signal gradientThe norms of (2) are as follows:
wherein,representing the modal components>Representing modality component->Center frequency of>Representation pair->Deviation calculation of->Representing the sum of the modal components;
s24: by introducing Lagrangian number multiplication factorsAnd a secondary penalty factor->Converting the variation model into an unconstrained variation model, wherein the obtained new model is shown in the following formula:
Wherein,is Lagrangian multiplier +.>Is a primitive function;
s25: updating modal components by employing an alternating direction multiplier algorithmCenter frequency->Lagrangian multiplier +.>To find the saddle point in the above equation, which is the optimal solution, the modal component +.>Center frequency->Lagrangian number multiplied by a factor->The update of (c) follows the following equation:
by using this alternate update approach, each variable can be iteratively optimized to approach gradually toward an optimal solution, where,represents the center frequency +.>Is->The IMF is->Center frequency at multiple iterations, +.>For the noise margin the reference signal is transmitted,，/>、/>respectively correspond to->，/>、/>Fourier transform of->Represents->The +.>An estimate of the order iteration; />Updating constraint conditions representing the original functions; />A pair solution representing Lagrangian multiplier,>is->First->Performing iteration values; />Is->The IMF is iterated 1 timeCenter frequency at time;
s26: in the continuous iteration process, if the iteration result satisfies the following formula, the iteration is stopped:
wherein,the judgment precision is; />Is->Updating values of the secondary iteration modal components; />Is->Updated values of the secondary iteration modality components.
In a preferred technical scheme, the machine learning model is a bald-bald algorithm, and the algorithm optimizes by simulating foraging and sailing behavior problems of african bald-bald, and the method specifically comprises the following steps:
First stage-determine the optimal bald eagle in group:
grouping the characteristic data according to a certain rule, and determining the optimal balding in the group according to a balding algorithm for the data in each group, wherein the balding is a data point with outstanding characteristics or abnormal behaviors in the group of data; in determining the optimal balding, selecting the optimal balding within each group as representative of the group according to the set evaluation criteria or indices;
second phase-starvation rate of bald Condition:
the abnormal data point density around the bald eagle is measured through the calculation of the hunger rate and the judgment of the threshold value;
third stage-exploration:
for each bald-cone, exploring neighboring nodes in its vicinity, including critical data points adjacent to the bald-cone or data points of related sensors, tracking the source of the fault or abnormal event by exploring and analyzing the source and propagation paths of the abnormal data points;
fourth stage-development:
and optimizing the parameters and identifying the abnormality.
In a preferred embodiment, the determining the optimal bald eagle in the group comprises:
after the initial population is formed, performing fitness calculation on all solutions, selecting the solution with the highest fitness as the best solution of the first group, selecting the solution with the next highest fitness as the best solution of the second group, re-calculating the fitness of the whole population in each fitness iteration, and then moving other solutions to the positions of the best solutions of the first group and the second group by using the following formula:
Wherein,representing the probability of other balding to move to the optimal balding position, this probability being determined on the basis of two parameters L1 and L2, the values of which range between 0 and 1 and the sum of which is equal to 1,/o>Represents the optimal solution of->Representing a suboptimal solution;
the probability of selecting the best solution is obtained using the roulette wheel method and each best solution is selected for each group:
in the method, in the process of the invention,probability representing optimal solution, ++>Representing individual hunger status;
in a preferred embodiment, the calculating of the hunger rate of the bald-awhile comprises:
establishing mathematical modeling:
representing the hunger and satiety state of the bald, the Iteration representing the current Iteration number, the maximums representing the maximum Iteration number, +.>Is a random number between-1 and 1, and which varies in each iteration, # is a random number between-1 and 1>Is a random number between-2 and 2,/and a random number between-2 and 2>Is a random number between 0 and 1, ">Representing the time of calculating the rate of bald-Consumer in the algorithm;
when (when)When the value falls below 0, it indicates that the bald is in hunger, when +.>When the value increases above 0, indicating that the bald is already full, the proportion of the total number of bald will decrease and the magnitude of the decrease will be at each iterationGradually increase when- >When the value of (2) is greater than 1, the user can find food in different areas, and then the user enters an exploration stage; if->The value of (2) is less than 1, the bald is found to be food in the vicinity of the best solution, and the development phase is entered.
In a preferred embodiment, the exploring includes performing mathematical modeling:
wherein,is one of [0,1 ]]Random numbers between, used to explore strategies, +.>Is a preset exploration parameter for adjusting the degree of exploration strategy, +.>Is the bald eagle position vector in the next generation, which is the hunger saturation rate obtained by calculation and is related to the current iteration condition,>represents the position of one of the best solutions, +.>And->Is in [0,1 ]]Random number between->And->Representing the search upper and lower boundaries of the parameter, respectively.
In a preferred embodiment, the development comprises the following steps:
when entering the development phase, two different rotary flight and attack strategies are executed, the selection of which is based onThe selection is made by the following formula:
to determine each policy, parameters are used before performing the search operationAssignment, which must be between 0 and 1, first generates a name +.>The value of which ranges from 0 to 1, and then by comparing this random number with the parameter +. >The size of (a) determines which strategy to use if +.>Greater than or equal to the parameterThen the attack strategy is executed and slowly implemented; however, if->Less than parameter->Then execute the rotational flight strategyAnd is omitted.
The invention also discloses a novel low-voltage switch cabinet data monitoring system, which comprises:
the monitoring module is used for acquiring monitoring data of the switch cabinet;
and a pretreatment module: preprocessing the monitoring data, and eliminating trend items of the monitoring data by adopting a least square method;
the feature extraction module is used for extracting statistical features and time domain features of the data from the preprocessed monitoring data;
the fault diagnosis module inputs the extracted characteristic data into the machine learning model, analyzes the input characteristics and identifies abnormal conditions by comparing the normal modes;
the invention also discloses a computer storage medium, on which a computer program is stored, which when executed realizes the novel low-voltage switch cabinet data monitoring method.
Compared with the prior art, the invention has the remarkable advantages that:
the trend item elimination operation is carried out on the data acquired by the sensor, the problems of unstable power supply voltage of a lower computer acquisition system, zero drift caused by temperature change of the sensor and the like are solved, VMD decomposition is adopted, each component is gradually extracted, the time-frequency characteristics of the component are reserved, unwanted noise and aliasing are eliminated, the accuracy and the authenticity of signals are greatly enhanced, and the accuracy of the data analysis and the abnormal detection result of the switch cabinet is further improved.
The extracted six characteristic data are input into a bald and awl algorithm, so that the mode of the working state of the normal switch cabinet is automatically learned, abnormal conditions are detected, the input characteristics are analyzed, and potential abnormal conditions are identified by comparing the behavior modes under normal conditions. The baldness algorithm can be effectively applied to a switch cabinet data monitoring system, problems are found and optimized by simulating foraging and sailing behaviors of African baldness, the optimizing capability is strong, the convergence speed is high, the abnormality detection capability and reliability of the system are improved, and the accurate monitoring and abnormality identification of switch cabinet data are realized.
Drawings
FIG. 1 is a diagram showing the main working steps of a conventional low voltage switchgear;
fig. 2 is a structure of a low-voltage switch cabinet based on the internet of things in the embodiment;
FIG. 3 is a block diagram of an embedded system in a control module according to the present embodiment;
fig. 4 is a hardware structure of the embedded controller of the present embodiment;
fig. 5 is a flowchart of an intelligent control procedure of the low-voltage switchgear of the present embodiment;
several classifications are common in the normal case of the switchgear data monitoring system shown in fig. 6;
FIG. 7 is a graph of ROC for evaluating the level of training of a switchgear data monitoring system model;
FIG. 8 is a confusion matrix for evaluating the level of training of a switchgear data monitoring system model;
FIG. 9 is a scatter plot of the level of model training for evaluating switchgear data monitoring systems.
Detailed Description
The principle of the invention is as follows: the trend item elimination operation is carried out on the data acquired by the sensor, the problems of unstable power supply voltage of a lower computer acquisition system, zero drift caused by temperature change of the sensor and the like are solved, VMD decomposition is adopted to gradually extract each component, the time-frequency characteristics of the components are reserved, unwanted noise and aliasing are eliminated, and the accuracy and the authenticity of signals are greatly enhanced. The bald-and-Convergence algorithm is effectively applied to the switch cabinet data monitoring system, problems are found and optimized by simulating foraging and sailing behaviors of the African bald-and-Convergence, the optimizing capability is strong, the convergence speed is high, the abnormality detection capability and the reliability of the system are improved, and the accurate monitoring and abnormality identification of the switch cabinet data are realized.
Example 1:
fig. 2 is a structure of a low-voltage switch cabinet based on the internet of things in the embodiment. The intelligent switch system is divided into two parts: intelligent household appliances and embedded tablet computers. The smart home appliances include various measuring instruments for monitoring key electrical parameters such as voltage, current, power, etc. in order to monitor the state of the device in real time.
In order to quickly achieve the design objective, the intelligent switching system is constructed by assembling each module. Meanwhile, the setting and monitoring management of the equipment are realized by configuring software, and the software has the functions of monitoring display, alarming, user authority management, data management, equipment maintenance, network communication and the like. The configuration software can greatly shorten the system integration time, improve the working efficiency and make the operation very simple.
The intelligent switch system has the characteristics of high efficiency, stability and reliability, and the integrated monitoring and control of the low-voltage switch cabinet are realized through the combination of the embedded tablet personal computer and the intelligent household appliance. The intelligent control system has the advantages that the design structure is clear, the layout is reasonable, the combination of hardware and software realizes intelligent and automatic control, the working efficiency is improved, the manpower resources are saved, and the intelligent control system is safer and more reliable.
FIG. 3 is a block diagram of an embedded system according to the present application. The embedded system mainly comprises a hardware system and a software system, wherein the hardware system comprises a core processor, a storage device and a peripheral device. Wherein the core processor is a computing center of the embedded system. Storage devices provide space for running programs and storing data. Peripheral devices include a/D converters, D/a converters, input-output devices, and the like. Communication between the controller and the outside world is achieved by the peripheral device. A software system is a specific system consisting of real-time operating system initialization code, drivers, and applications. The real-time operating system is a core program for the embedded software system to realize multi-task scheduling.
Fig. 4 shows a hardware structure of the embedded controller of the present application. When the intelligent controller operates, the A/D converter is required to collect sensor data, the D/A converter is required to output control signals, and the serial screen is required to realize man-machine interaction. In addition, the normal operation of the switch cabinet is guaranteed, and safety problems are avoided. Therefore, the hardware of the embedded intelligent controller is divided into nine modules: the system comprises a minimum system module taking an STM32 chip as a core, an A/D acquisition and D/A output module, a serial port screen and upper computer communication module, a power isolation module, a clock module, an optical isolation module, a temperature sensor module, a control output module and a safety module. The nine modules will be described separately:
(1) Minimum system module: in the intelligent controller design, the dominant product STM32 is selected as the processor. The processor adopts a Cortex kernel, has the data width of 32 bits and the clock frequency of 168MHz, provides a plurality of communication interfaces (such as RS232, RS485 and CAN bus interfaces) and the like, has strong data processing capacity and large-capacity memory (such as 1024K FLASH and 193K SRAM), and CAN meet the requirements of data processing and storage. In addition, the system also provides rich peripheral interfaces, such as ADC, PWM, DAC, and the like, so that the system can easily realize multifunctional monitoring and control. The high clock frequency of the processor can realize more accurate voltage and pulse signal output, can collect various monitoring parameters and realize multifunctional monitoring and alarming.
(2) a/D acquisition and D/a output module: in intelligent controller designs, it is critical to monitor operating parameters such as load current and operating current. To achieve this, it is necessary to design an a/D conversion circuit to convert these analog signals into digital signals for processing, and to use a D/a conversion circuit to convert the processed digital signals into analog signals for output to a controller. The a/D conversion circuit is typically composed of a signal conditioning circuit and an a/D conversion module. The signal conditioning circuit mainly completes the functions of filtering, amplifying, adjusting and the like of an input signal so as to meet the input requirement of the A/D conversion module. The A/D conversion module converts the analog signals into discrete digital signals for processing and analysis by the controller. The conversion process can be completed through an A/D converter built in the singlechip. In the design process, proper A/D converters are required to be selected for configuration so as to meet the parameter requirements of required precision, speed and the like. Meanwhile, a sample hold circuit is required to be used for stabilizing an input signal in the conversion process, so that the accuracy of a conversion result is ensured. The inverse process of the D/A conversion circuit is opposite to that of the A/D conversion circuit, and the D/A conversion circuit converts the processed digital signals into analog signals and outputs the analog signals to the controller so as to realize the operation of the controller. The D/A conversion circuit mainly comprises a digital signal processor, a reference voltage source, an output amplifier and the like, wherein the digital signal processor converts the processed digital signal into an analog signal and outputs the analog signal to the audio output amplifier, and the audio output amplifier outputs the analog signal to a loudspeaker or other external equipment. The input and output signals are converted, so that the monitoring and control of the control operation parameters are realized, and the safe and stable operation is further ensured.
(3) Serial screen and host computer communication module: in the design of the intelligent controller, a serial screen design scheme is adopted for realizing man-machine interaction. The serial port screen is communicated with the STM32 chip mainly through an RS232 protocol, and an RS232 communication circuit is designed by adopting a MAX3232 chip. Through the scheme, a user can directly interact with the intelligent controller through the serial port screen, such as inputting voltage and current set values, adjusting the rotating speed of the fan and the like, and meanwhile, the running state, such as load current, running current and the like, can be read in real time. In addition, the serial port screen can also support the customization of the communication protocol, and personalized customization can be carried out according to the needs so as to meet the requirements of different users. The serial port screen is used as an important component of intelligent controller design, can provide a high-efficiency and convenient man-machine interaction mode for a user, and improves the intelligent degree and reliability of the equipment.
(4) And a power isolation module: in the intelligent controller design, voltage power supplies of 3.3V, 5V and 12V are required to be provided according to the power supply requirements of different modules. First, the 12V DC power supply provided needs to be isolated and step down to ensure their stability and safety. The isolated regulated power supply module VRB1205AP-6WR3 is selected to accomplish this task. The power supply module is capable of converting 12V DC power to 5V DC power and providing an output power of up to 6W. Next, the 5V DC power needs to be converted to 3.3V to provide the required power for STM32 and other modules. Forward low drop-out voltage regulators ams1117-3.3 were selected to accomplish this task. The voltage stabilizer can easily and stably reduce the input voltage to 3.3V and support the range of the input voltage from 4V to 12V. Finally, it is also necessary to provide a 5V power supply to the D/a circuit to enable proper operation. Since the 12V DC power supply is converted into the 5V DC power supply through the isolated regulated power supply module, the step-down conversion is not needed to realize the output of the 5V power supply.
(5) And (3) a clock module: the clock module not only provides a basic clock signal, but also supports a real-time clock function. It can record the current date and time and support the operations of time increase, decrease, adjustment and presentation. The accuracy of the clock module is critical to the proper operation and precise control of the embedded controller. By using crystal elements and related circuitry, the clock module is able to provide a highly stable time-based signal to ensure long-term stability and time accuracy of the controller under different circumstances. Therefore, the embedded controller of the switch cabinet can accurately execute various operations according to a preset time plan, follow a time control strategy and meet the requirements of practical application.
(6) An optical isolation module: the optical isolation module is used for realizing photoelectric isolation between the internal circuit of the switch cabinet and external equipment. The main function of the circuit is to isolate the input signal and the output signal by the transmission of the optical signal in the circuit transmission process, thereby improving the safety and noise immunity of the system. The optical isolation module is composed of a light emitting device and a photodiode. The light emitting device is used to convert an electrical signal into an optical signal. The photodiode is an electronic device with a photoelectric conversion function, which can convert a current excited by an optical signal into a voltage signal, thereby converting the optical signal into an electrical signal. In a switchgear circuit, an optical isolation module is generally used to isolate input/output signals, such as switching signals, analog signals, or communication interfaces, to prevent external interference from propagating to an internal circuit or from adversely affecting external devices. By using optical isolation, interference conduction between the electrical ground and the common ground can be blocked, providing a higher electrical isolation effect.
(7) Temperature sensor module: this is a device for monitoring the temperature conditions inside a switchgear cabinet. The temperature sensor can measure the temperature of the surrounding environment and convert temperature data into electric signals to be output so as to realize the functions of temperature protection and fault early warning. The temperature protection function may be used for automatic power off or triggering safety measures such as alarms. When the temperature sensor module detects that the temperature exceeds the set safety range, the temperature sensor module performs power-off protection on related equipment or circuits through control signals so as to prevent accidents. In addition, the temperature sensor module can also cooperate with a monitoring system to realize a fault early warning function. Through continuous monitoring of temperature data, the system can judge the trend of temperature rise and take preventive measures, such as timely notifying maintenance personnel to perform heat dissipation or repair, so as to avoid damage caused by equipment overheating.
(8) And a control output module: and the equipment module is used for controlling the internal equipment of the switch cabinet. It implements control operations on the device, including opening, closing, adjusting, etc., by driving external actuators. Through the use of the control output module, a user can conveniently control the operation of various devices in the switch cabinet, and automation and remote control are realized. The starting and stopping of the equipment are realized by controlling the output module to turn on or off a motor; the contact state of the relay can be controlled by the control output module, so that the on-off control of the circuit is realized.
(9) And (3) a safety module: an equipment module for detecting and handling safety events in a switchgear. Its main function is to monitor the current, voltage and leakage quantity parameters in the circuit, and to react quickly and take corresponding measures to protect the safe operation of the switch cabinet and its internal equipment. Can be matched with other devices such as a controller, a protection device and the like for interactive communication and linkage operation. In addition, the safety module can also provide related alarm signals, state indication and fault diagnosis functions, so that a user can be helped to find and solve a safety event in time.
Fig. 5 shows an introduction to the data detection system of the low-voltage switchgear of the present embodiment. The working principle of the system will be explained in detail as follows:
s01: in order to monitor various parameters of the switch cabinet in real time, proper sensors and data acquisition equipment are required to be deployed. These sensors may include temperature sensors, humidity sensors, current sensors, voltage sensors, and the like. These sensors should be able to accurately measure the environmental and electrical parameters inside the switchgear.
S02: once the raw data is collected from the sensors, the next step is to pre-process the data to ensure its accuracy and reliability. The preprocessing process includes data cleansing, outlier detection and correction, and data decompression and formatting, converting the data into a form suitable for algorithmic processing.
S03: feature extraction is a critical step in the pre-processed data that helps transform the raw data into meaningful feature representations. The purpose of feature extraction is to capture key information in the data, reduce data dimensionality, and provide useful feature data that facilitates algorithmic analysis and decision making. One common feature extraction method is to extract statistical features of the data, such as mean, maximum, minimum, standard deviation, percentile, etc. These statistics provide information on the central tendency, degree of dispersion and distribution of the data. And time domain features such as peaks, waveform features, etc.
S04: the extracted characteristic data is input into the machine learning model to enable the model to automatically learn the mode of the normal switch cabinet working state, once the machine learning model is trained, the input characteristic data is analyzed, and potential abnormal conditions can be identified by comparing the behavior modes under normal conditions. This may be achieved by comparing the similarity of the input data to the normal mode or calculating an anomaly score. A larger anomaly score or data that differs from the normal mode by a larger amount may be indicative of a potential anomaly.
S05: according to the output result of the machine learning model, the system can judge whether an abnormal condition exists currently. When the model detects that the input characteristic data has a significant difference or abnormal mode from the normal behavior mode, the model is identified as abnormal condition occurrence. Through an alarm mechanism and abnormal information recording, the system can timely respond and process abnormal conditions, and the sensing and coping capacity of abnormal events is improved. This helps to ensure the safety and reliability of the switchgear operation and provides valuable data for improving the design, maintenance and operation of the system.
S06: the visual display of the monitoring data and the analysis results of the abnormal conditions is an effective way, so that operators or related personnel can more intuitively know the state and the abnormal conditions of the switch cabinet. Such visual displays are often presented in the form of charts, graphs or dashboards for rapid acquisition of critical information, through visual displays and monitoring reports, operators and related personnel can better understand the operation conditions of the switchgear, discover abnormal conditions in time, and take appropriate measures for maintenance and repair. This can improve operating efficiency, reduce potential risks, and optimize the performance and reliability of the switchgear system.
In the working step S02 of the low-voltage switch cabinet data monitoring system, in the mentioned data preprocessing operation, the trend term eliminating operation is performed first, and in the signal acquisition process, there are some problems, such as unstable power supply voltage of the lower computer acquisition system, zero drift caused by temperature change of the sensor, and the like. These problems can lead to spurious features in the acquired signal and mask the trend of its true features. If these trends are not addressed, the features extracted in the subsequent feature extraction process will become unreliable or even ineffective, resulting in a loss of signal accuracy and authenticity. The method adopts a least square method to perform the operation of eliminating the trend term. In the signal acquisition process, the acquired signal is set asAnd the trend term therein is expressed as polynomial function +.>。
，/>
To find coefficients in trend term functionsThe error energy function is constructed using a least squares method. Hopefully by minimizing the trend term function +.>And signal->And the error of the two parameters is calculated as the quadratic sum of the errors to obtain the optimal parameters.
Consider a device withA function of variables, one of which is +.>. Regarding this functionPerforming partial derivative operation to obtain ∈ >And the partial derivatives. These partial derivatives constitute a composition comprising +>A system of linear equations, each of which is related to +.>Variable number.
，/>
Thereby determining the coefficientA trend term expression of the signal is obtained. />Representing polynomial orders, in practical engineering applications,/->The value range of (2) is usually [1,3 ]]。
Next, the denoising operation is required, and VMD decomposition is adopted. VMD decomposition can effectively improve problems such as end-point effects and modal aliasing, and therefore, better performance in terms of accuracy and error. The VMD algorithm breaks down the signal into a plurality of amplitude modulated frequency modulated (AM-FM) signals. This means that the VMD can accurately identify and separate out the different frequency components in the signal. By applying wiener adaptive filtering, the VMD can extract each component step by step, preserving its time-frequency characteristics and eliminating unwanted noise and aliasing. More accurate and reliable signal decomposition results can be provided.
The VMD decomposition steps are as follows:
the first step: the VMD decomposition algorithm first performs Hilbert transform on the signal to obtain an resolved signal for each IMF component. Subsequently, the single-sided spectrum of the signal extracted from the resolved signal can be seen as follows:
wherein,representing a unit pulse function; / >Representing imaginary units; />Time of presentation->Representing convolution operation,/->Is->The IMF component is->A value of the time of day.
And a second step of: an exponential weight is introduced with the aim of adjusting the estimated center frequency of each IMF component and modulating the spectrum of each IMF component to the corresponding baseband.
And a third step of: constructing a constrained variation model based on the square of the demodulated signal gradientThe norm of (c) can be seen as follows:
wherein,representing the modal components>Representing modality component->Center frequency of>Representation pair->Deviation calculation of->Representing the sum of the modal components.
Fourth step: by introducing Lagrangian number multiplication factorsAnd a secondary penalty factor->In the equation of the third step, the above-mentioned variational model can be converted into an unconstrained variational model, so that it can be solved. After these factors are introduced, the new model is shown as the formula: />
Fifth step: updating modal components by employing an alternating direction multiplier algorithmCenter frequency->Lagrangian multiplier +.>The saddle point in the above equation is obtained, and the saddle point is the optimal solution. Modal component->Center frequency->Lagrangian number multiplied by a factor->The updating of (2) follows the following equation. By using this alternate update approach, each variable can be iteratively optimized to approach gradually towards the optimal solution, where +. >Represents the center frequency +.>Is->The IMF is->Center frequency at multiple iterations, +.>For noise margin +.>，/>、/>Respectively correspond to->，/>、/>Is used for the fourier transform of (a),represents->The +.>An estimate of the order iteration;
updating constraint conditions representing the original functions; />A pair solution representing Lagrangian multiplier,>is thatFirst->Performing iteration values; />Is->Center frequency of each IMF at 1 iteration.
Sixth step: in the continuous iteration process, if the iteration result satisfies the following formula, the iteration is stopped. Wherein,the judgment precision is; />Is->Updating values of the secondary iteration modal components; />Is->Updated values of the secondary iteration modality components.
In VMD decomposition, a suitable number of modal components is selectedAnd regularization parameter->Is very critical. For the selection of the number of the modal components, the average value of the instantaneous frequencies of the IMF components can be used as an effective judgment basis. When the number of decomposed modal components is excessive, the decomposed IMF components may appear to be intermittent in the high frequency portion, resulting in a decrease in average instantaneous frequency. This further results in a pronounced sag in the IMF component instantaneous frequency mean curve. This critical number is a reasonable number of modal components. When the number of decomposed modal components +. >At 4, the curve clearly shows a sag phenomenon. Thus, it selectsAs the number of modal components. Regarding regularization parameters->In general engineering applications, the value is usually between 1000 and 3000, which can be a reasonable choice interval.
After the data preprocessing operation of S02, feature extraction is further performed on the processed data, as mentioned in step S03. Statistical feature extraction is used herein to refer to computing some statistics or features from a set of data to describe certain attributes or potential patterns of the data. It is a pre-processing step commonly used in data analysis and machine learning to simplify and compress data for further analysis and modeling. The mean, standard deviation, variance, correlation, kurtosis and skewness values of the processed data are used herein.
After the six features are extracted, the extracted six feature data are input into a machine learning model according to the steps S04-S05, so that the machine learning model automatically learns the mode of the working state of the normal switch cabinet, abnormal conditions are detected, the input features are analyzed, and potential abnormal conditions are identified by comparing the behavior modes under the normal conditions. The adopted algorithm is a bald-and-Convergence algorithm, and the algorithm is mainly used for optimizing through simulating the problems of foraging and sailing behaviors of African bald and has the characteristics of strong optimizing capability, high convergence speed and the like, and the algorithm comprises four stages, and specifically comprises the following steps:
1. First stage-determine the optimal bald eagle in group:
and grouping the characteristic data according to a certain rule or condition. These groupings may be based on temporal, spatial, or other relevant attributes. For example, the data may be grouped by different time periods, different switch cabinets, or different sensors. For the data within each group, the optimal balding within the group was determined by the balding algorithm. Balding may be understood as a data point with prominent features or abnormal behavior in the set of data. In determining the optimal bald eagle, operations of feature extraction and abnormality detection are required. Feature extraction is used to extract useful features from the data and provide an index for subsequent judgment. The optimal bald eagle is selected within each group as representative of the group according to a particular evaluation criterion or index.
Specifically, after the initial population is formed, fitness calculation is performed on all solutions, the solution with the highest fitness is selected as the best solution of the first group, and the solution with the next highest fitness is selected as the best solution of the second group. The other solutions are then moved to where the best solutions of the first and second sets are located using the following formula. In each fitness iteration, the fitness of the whole population is recalculated.
In the formula (I), the compound (II) is a compound (III),indicating the probability of other balding to move to the optimal balding position. This probability is determined on the basis of two parameters L1 and L2, whose values range between 0 and 1 and whose sum is equal to 1. Method for obtaining selected best solutions by using roulette wheelProbability, and select each best solution for each group.
In the method, in the process of the invention,probability representing optimal solution, ++>Representing individual hunger status.
2. Second phase-starvation rate of bald Condition:
to measure the abnormal data point density around the balding. Through calculation of the starvation rate and judgment of the threshold value, abnormal conditions in the switch cabinet data can be better identified, and corresponding measures such as alarming, maintenance or repair can be further adopted to ensure normal operation and safety of the switch cabinet.
In particular, bald-Convergences store higher energy in the body as they feel full during foraging, which allows them to fly longer distances to find food. However, when the bald is hungry, they are not energy efficient and cannot fly for a long period of time. In this case, they would look for food near strong baldness and appear more 21636and 21636when starved. This phenomenon can be mathematically modeled by using the following equation.
In this model, some parameters and random numbers are used to represent the behaviour of the baldness. First of all,indicating hunger and satiety of the bald and the Condition of the Condition. By controlling->Can simulate the hunger state of a bald-Condition. The Iteration represents the current Iteration number, while the maximums represent the maximum Iteration number. />Is a random number between-1 and which varies in each iteration. />Is a random number between-2 and 2. />Is a random number between 0 and 1, ">Representing the algorithm time to calculate the rate of immersion.
By observingCan determine the hunger and satiety state of the bald-Condition. When->When the value falls below 0, this indicates that the bald is hungry. When->When the value increases above 0, this indicates that the bald is full. In each iteration, the proportion of the total number of bald-evels decreases and the magnitude of the decrease increases gradually. When->When the value of (2) is greater than 1, the bald will find food in a different area, at which point the system enters the exploration phase. If->If the value of (2) is less than 1, the bald-coat will find food in the vicinity of the best solution, at which point the system enters the development stage. By the above description, it is possible to keep the number of words approximately the sameMeaning of parameters and random numbers in the model and their effect on the behavior of the bald-Condition are expressed.
3. Third stage-exploration:
the anomaly detection capability is further extended based on the selected baldness of the first stage and the calculated hunger rate of the second stage. By exploring the surrounding data points, potentially more anomalies or fault conditions can be detected. For each bald-Cone, the neighboring nodes in its vicinity are explored. This may include critical data points adjacent to the bald Condition or data points of related sensors. The root cause of a fault source or abnormal event is tracked by exploring and analyzing the source and propagation paths of the abnormal data points. Based on the analysis results of the exploration phase, further decision support for the switchgear data monitoring system is provided. Specifically, the following model was used:
in this model, some parameters and random numbers are used to control the behaviour of the balding. First of all,is one of [0,1 ]]Random numbers in between, used to explore strategies. />Is a preset exploration parameter used for adjusting the degree of exploration strategy.Is the bald eagle position vector in the next generation, which is the hunger saturation rate obtained by calculation and is related to the current iteration situation. />Representing the position of one of the best solutions. />And->Is in [0,1 ]]Random numbers in between. />And->Representing the search upper and lower boundaries of the parameter, respectively.
4. Fourth stage-development:
equivalent value ofBetween 0.5 and 1, the development phase is entered. Two different rotary flight and tapping strategies are performed. Policy selection is based on->Selection is made. The specific body process can be represented by the following formula: />
Wherein, in order to determine the selection of each policy, parameters need to be parameters before performing the search operationA value must be assigned that is between 0 and 1. When entering this stage, a name +.>And the value of the random number is in the range of 0 to 1. Next, by comparing this random number with the parameter +.>Is used to decide which strategy to use. If->Greater than or equal to parameter->Then the attack strategy is executed and slowly implemented; however, if->Less than parameter->Then a rotary flight strategy is performed.
Tasks in the development stage include implementation and integration of algorithms, parameter tuning, performance assessment, system integration and optimization, and deployment and continuous optimization. The tasks aim to effectively apply a bald-Convergence algorithm to a switch cabinet data monitoring system, improve the abnormality detection capability and reliability of the system, and realize accurate monitoring and abnormality identification of switch cabinet data. Several categories of switch cabinet data monitoring systems are common in the normal case shown in fig. 6, respectively: normal, short circuit, overload, leakage. In a switchgear data monitoring system, a short circuit may appear as an abnormality such as a sudden increase in current, an abnormal fluctuation in current, or an exceeding of a rated value by current; overload conditions can be identified in the data monitoring system by monitoring the continuously high value of the current and conditions outside the rated range; typically due to poor line contact, equipment failure or insulation damage. The switch cabinet data monitoring system can detect the increase or abnormal difference of leakage current and is used for judging whether the leakage current problem exists in the system.
Fig. 7 shows that the switch cabinet data monitoring system based on the bald-and-race algorithm, which evaluates the level of model training of the switch cabinet data monitoring system using ROC curves (subject operating characteristics), exhibited excellent model performance, and the area under ROC curve (AUC value) was 0.97. The AUC value of this height indicates that the system is able to effectively identify and monitor anomalies in the switchgear data. The switch cabinet data monitoring system in the bald-Convergence algorithm exhibited excellent model performance and demonstrated its excellent performance in anomaly detection by the ROC curve of high AUC values. This will provide a reliable tool for users to monitor and manage the switchgear data in real time, ensuring safe operation and efficient maintenance of the equipment.
Fig. 8 shows the use of a confusion matrix to evaluate the level of training of a switchgear data monitoring system model that performs data analysis and anomaly detection by a bald-and-race algorithm for the switchgear data monitoring system to identify anomalies in the switchgear data. From the generated confusion matrix, the model shows good performance, which means that it can classify normal and abnormal operations with high accuracy.
Fig. 9 shows a data analysis and anomaly detection for a switchgear data monitoring system using a baldness algorithm to identify anomalies in the switchgear data. By generating a scatter plot, good performance of the model can be observed, indicating that it can accurately distinguish normal data from abnormal data points.
The foregoing examples are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the foregoing examples, and any other changes, modifications, substitutions, combinations, and simplifications that do not depart from the spirit and principles of the present invention should be made therein and are intended to be equivalent substitutes within the scope of the present invention.
Claims (10)
1. The novel low-voltage switch cabinet data monitoring method is characterized by comprising the following steps of:
s01: acquiring monitoring data of a switch cabinet;
s02: preprocessing the monitoring data, and eliminating trend items of the monitoring data by adopting a least square method;
s03: extracting statistical characteristics and time domain characteristics of the data from the preprocessed monitoring data;
s04: the extracted feature data is input into a machine learning model, the input features are analyzed, and abnormal conditions are identified by comparing the normal modes.
2. The new low-voltage switchgear data monitoring method according to claim 1, wherein the trend term eliminating method in step S02 comprises:
s11: let the collected signals beAnd the trend term therein is expressed as polynomial function +.>：
，
Wherein,；
s12: to find coefficients in trend term functionsConstructing an error energy function by using a least square method>By minimizing the trend term function +.>And signal->The quadratic sum of the errors between the two parameters is used for obtaining the optimal parameters:
，
regarding this functionPerforming partial derivative operation to obtain ∈>Partial derivatives of theseThe partial derivatives form a composition comprisingA system of linear equations, each of which is related to +.>The following variables:，
thereby determining the coefficientA trend term expression of the signal is obtained, +.>Representing polynomial orders, deleting the obtained trend term.
3. The new low-voltage switchgear data monitoring method according to claim 2, wherein the step S02 further comprises denoising by means of a variant modal decomposition, the variant modal decomposition method comprising the steps of:
s21: performing Hilbert transformation on the signals with the trend items eliminated to obtain analytic signals of each IMF component, and extracting single-side frequency spectrums of the signals from the analytic signals:
，
Wherein, the IMF is an intrinsic mode function,representing the unit pulse function +.>Representing imaginary units, ++>Time of presentation->Representing convolution operation,/->Is->The IMF component is->A value of time of day;
s22: introducing an exponential weight, adjusting the estimated center frequency of each IMF component, and modulating the spectrum of each IMF component to a corresponding baseband:
，
s23: constructing a constrained variation model based on the square of the demodulated signal gradientThe norms of (2) are as follows:
，
wherein,representing the modal components>Representing modality component->Center frequency of>Representation pair->Deviation calculation of->Representing the sum of the modal components;
s24: by introducing Lagrangian number multiplication factorsAnd a secondary penalty factor->Converting the variation model into an unconstrained variation model, wherein the obtained new model is shown in the following formula:
，
wherein,is Lagrangian multiplier +.>Is a primitive function;
s25: updating modal components by employing an alternating direction multiplier algorithmCenter frequency->Lagrangian multiplierTo find the saddle point, which is the optimal solution, the modal component +.>Center frequency->Lagrangian number multiplied by a factor->The update of (c) follows the following equation:
，
，
，
by using this alternate update approach, each variable can be iteratively optimized to approach gradually toward an optimal solution, where, Represents the center frequency +.>Is->The IMF is->Center frequency at multiple iterations, +.>For noise margin +.>，、/>Respectively correspond to->，/>、/>Fourier transform of->Represents->The +.>An estimate of the order iteration; />Updating constraint conditions representing the original functions; />A pair solution representing Lagrangian multiplier,>is->First->Performing iteration values; />Is->Center frequency of each IMF at 1 iteration;
s26: in the continuous iteration process, if the iteration result satisfies the following formula, the iteration is stopped:
，
wherein,for distinguishing accuracy +.>Is->Updated values of the sub-iterative modal component +.>Is->Updated values of the secondary iteration modality components.
4. The method for monitoring data of a new low voltage switchgear as claimed in claim 1, wherein the machine learning model is a bald-Convergence algorithm which optimizes by simulating the problems of foraging and sailing behaviour of African bald-Convergence, comprising the specific steps of:
first stage-determine the optimal bald eagle in group:
grouping the characteristic data according to a certain rule, and determining the optimal balding in the group according to a balding algorithm for the data in each group, wherein the balding is a data point with outstanding characteristics or abnormal behaviors in the group of data; in determining the optimal balding, selecting the optimal balding within each group as representative of the group according to the set evaluation criteria or indices;
Second phase-starvation rate of bald Condition:
the abnormal data point density around the bald eagle is measured through the calculation of the hunger rate and the judgment of the threshold value;
third stage-exploration:
for each bald-cone, exploring neighboring nodes in its vicinity, including critical data points adjacent to the bald-cone or data points of related sensors, tracking the source of the fault or abnormal event by exploring and analyzing the source and propagation paths of the abnormal data points;
fourth stage-development:
and optimizing the parameters and identifying the abnormality.
5. The method of claim 4, wherein determining an optimal balk in a group comprises:
after the initial population is formed, performing fitness calculation on all solutions, selecting the solution with the highest fitness as the best solution of the first group, selecting the solution with the next highest fitness as the best solution of the second group, re-calculating the fitness of the whole population in each fitness iteration, and then moving other solutions to the positions of the best solutions of the first group and the second group by using the following formula:
，
wherein,representing the probability of other balding to move to the optimal balding position, this probability being determined on the basis of two parameters L1 and L2, the values of which range between 0 and 1 and the sum of which is equal to 1,/o >Represents the optimal solution of->Representing a suboptimal solution;
the probability of selecting the best solution is obtained using the roulette wheel method and each best solution is selected for each group:
，
in the method, in the process of the invention,probability representing optimal solution, ++>Representing individual hunger status.
6. The new low voltage switchgear data monitoring method according to claim 4, wherein the calculation of the hunger rate of bald-irises comprises:
establishing mathematical modeling:
，
，
representing the hunger and satiety state of the bald, the Iteration representing the current Iteration number, the maximums representing the maximum Iteration number, +.>Is a random number between-1 and 1, and which varies in each iteration, # is a random number between-1 and 1>Is a random number between-2 and 2,/and a random number between-2 and 2>Is a random number between 0 and 1, ">Representing the time of calculating the rate of bald-Consumer in the algorithm;
when (when)When the value falls below 0, it indicates that the bald is in hunger, when +.>When the value increases above 0, which means that the bald is already full, the proportion of the total number of bald is decreased and the magnitude of the decrease is gradually increased in each iteration, when +.>When the value of (2) is greater than 1, the user can find food in different areas, and then the user enters an exploration stage; if->The value of (2) is less than 1, the bald is found to be food in the vicinity of the best solution, and the development phase is entered.
7. The method of claim 4, wherein the exploring comprises mathematically modeling:
，
，
wherein,is one of [0,1 ]]Random numbers between, used to explore strategies, +.>Is a preset exploration parameter for adjusting the degree of exploration strategy, +.>Is the bald eagle position vector in the next generation, which is the hunger saturation rate obtained by calculation and is related to the current iteration condition,>represents the position of one of the best solutions, +.>And->Is in [0,1 ]]Random number between->And->Representing the search upper and lower boundaries of the parameter, respectively.
8. The new low voltage switchgear data monitoring method according to claim 4, characterized in that the development comprises the steps of:
when entering the development phase, two different rotary flight and attack strategies are executed, the selection of which is based onThe selection is made by the following formula:
，
to determine each policy, parameters are used before performing the search operationAssignment, which must be between 0 and 1, first generates a name +.>The value of which ranges from 0 to 1, and then by comparing this random number with the parameter +. >The size of (a) determines which strategy to use if +.>Greater than or equal to parameter->Then the attack strategy is executed and slowly implemented; however, if->Less than parameter->Then a rotary flight strategy is performed.
9. A novel low voltage switchgear data monitoring system, comprising:
the monitoring module is used for acquiring monitoring data of the switch cabinet;
and a pretreatment module: preprocessing the monitoring data, and eliminating trend items of the monitoring data by adopting a least square method;
the feature extraction module is used for extracting statistical features and time domain features of the data from the preprocessed monitoring data;
the fault diagnosis module inputs the extracted characteristic data into the machine learning model, analyzes the input characteristics and identifies abnormal conditions by comparing the normal modes.
10. A computer storage medium having stored thereon a computer program, characterized in that the computer program, when executed, implements the new low-voltage switchgear data monitoring method of any of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311434792.9A CN117150421A (en) | 2023-11-01 | 2023-11-01 | Novel low-voltage switch cabinet data monitoring method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311434792.9A CN117150421A (en) | 2023-11-01 | 2023-11-01 | Novel low-voltage switch cabinet data monitoring method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117150421A true CN117150421A (en) | 2023-12-01 |
Family
ID=88901225
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311434792.9A Pending CN117150421A (en) | 2023-11-01 | 2023-11-01 | Novel low-voltage switch cabinet data monitoring method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117150421A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113935384A (en) * | 2021-11-25 | 2022-01-14 | 中铁二院工程集团有限责任公司 | Bridge health monitoring method and system for signal adaptive decomposition and recognition |
| CN114325361A (en) * | 2021-11-15 | 2022-04-12 | 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 | Transformer tap switch state online monitoring method based on vibration principle |
| CN115575125A (en) * | 2022-09-21 | 2023-01-06 | 三峡大学 | Bearing fault diagnosis method based on GADF-GAN-AVOA-CNN |
| CN116089813A (en) * | 2022-12-07 | 2023-05-09 | 中广核(乌兰察布)风力发电有限公司 | Wind turbine generator set transmission chain fault diagnosis method based on machine learning |
| CN116306306A (en) * | 2023-03-31 | 2023-06-23 | 南昌大学 | Micro-grid inertia constant estimation method based on African bald-Condition algorithm |
| CN116644970A (en) * | 2023-05-10 | 2023-08-25 | 福建工程学院 | Photovoltaic power prediction method based on VMD decomposition and lamination deep learning |
-
2023
- 2023-11-01 CN CN202311434792.9A patent/CN117150421A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114325361A (en) * | 2021-11-15 | 2022-04-12 | 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 | Transformer tap switch state online monitoring method based on vibration principle |
| CN113935384A (en) * | 2021-11-25 | 2022-01-14 | 中铁二院工程集团有限责任公司 | Bridge health monitoring method and system for signal adaptive decomposition and recognition |
| CN115575125A (en) * | 2022-09-21 | 2023-01-06 | 三峡大学 | Bearing fault diagnosis method based on GADF-GAN-AVOA-CNN |
| CN116089813A (en) * | 2022-12-07 | 2023-05-09 | 中广核(乌兰察布)风力发电有限公司 | Wind turbine generator set transmission chain fault diagnosis method based on machine learning |
| CN116306306A (en) * | 2023-03-31 | 2023-06-23 | 南昌大学 | Micro-grid inertia constant estimation method based on African bald-Condition algorithm |
| CN116644970A (en) * | 2023-05-10 | 2023-08-25 | 福建工程学院 | Photovoltaic power prediction method based on VMD decomposition and lamination deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 尹进田: "《高速列车牵引电机微弱故障诊断与溯源研究》", 北京航空航天大学出版社, pages: 102 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11605036B2 (en) | System and methods for power system forecasting using deep neural networks | |
| WO2011027607A1 (en) | Anomaly detection and diagnostic method, anomaly detection and diagnostic system, and anomaly detection and diagnostic program | |
| CA2972973A1 (en) | Machine learning-based fault detection system | |
| US10310495B2 (en) | Systems and methods for monitoring automation systems | |
| TWI692196B (en) | Fault detection system and method for solar photovoltaic | |
| CN112434799B (en) | Non-invasive load identification method based on full convolution neural network | |
| CN110971677B (en) | Electric power internet of things terminal equipment side channel safety monitoring method based on countermeasure reinforcement learning | |
| CN117113218A (en) | Visual data analysis system and method thereof | |
| CN116961242B (en) | Real-time data management method and system for secondary fusion on-column circuit breaker | |
| CN117150421A (en) | Novel low-voltage switch cabinet data monitoring method and system | |
| CN117114454B (en) | DC sleeve state evaluation method and system based on Apriori algorithm | |
| CN106356994B (en) | A kind of grid stability method of discrimination based on power grid PMU big datas | |
| Wu et al. | Alarms‐related wind turbine fault detection based on kernel support vector machines | |
| Geurts et al. | Early prediction of electric power system blackouts by temporal machine learning | |
| CN112099438B (en) | Machine tool energy-saving control method and device based on current signals | |
| US20230054387A1 (en) | System for monitoring and analyzing electric parameters | |
| CN114662576A (en) | Non-invasive storage battery car charging detection method and system based on supervised classification | |
| Balouji et al. | Eigen-analysis based power quality event data clustering and classification | |
| Cui et al. | Multi-mode Big Data Mining and Analysis Based on Internet of Things on Power | |
| KR102604219B1 (en) | Method and System for Detecting Faults on High Resolution Data | |
| Viereck et al. | Driving reliability with machine learning and improving operation by digitalization of power transformers | |
| CN117111446B (en) | Fuzzy PID control optimization method for magnetic suspension flywheel motor | |
| CN117713221B (en) | Micro-inversion photovoltaic grid-connected optimization system | |
| Wang et al. | Impedance Ground Faults Detection and Classification Method for DC Microgrid | |
| Keramati et al. | Non-intrusive Load Monitoring Using Water Consumption Patterns |
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 |