CN112487714B - Method for generating cable shaft fire state identification decision tree model - Google Patents
Method for generating cable shaft fire state identification decision tree model Download PDFInfo
- Publication number
- CN112487714B CN112487714B CN202011347416.2A CN202011347416A CN112487714B CN 112487714 B CN112487714 B CN 112487714B CN 202011347416 A CN202011347416 A CN 202011347416A CN 112487714 B CN112487714 B CN 112487714B
- Authority
- CN
- China
- Prior art keywords
- value
- decision tree
- preset
- residual current
- characteristic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000003066 decision tree Methods 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000012549 training Methods 0.000 claims abstract description 46
- 238000001514 detection method Methods 0.000 claims abstract description 19
- 239000000779 smoke Substances 0.000 claims abstract description 14
- 238000011156 evaluation Methods 0.000 claims abstract description 13
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 230000000737 periodic effect Effects 0.000 claims description 3
- 238000000926 separation method Methods 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 abstract description 11
- 230000000694 effects Effects 0.000 abstract description 6
- 238000004458 analytical method Methods 0.000 description 6
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000015556 catabolic process Effects 0.000 description 2
- 238000009413 insulation Methods 0.000 description 2
- 206010000369 Accident Diseases 0.000 description 1
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical compound [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 229910052782 aluminium Inorganic materials 0.000 description 1
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 229910052802 copper Inorganic materials 0.000 description 1
- 239000010949 copper Substances 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 229910052751 metal Inorganic materials 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R19/00—Arrangements for measuring currents or voltages or for indicating presence or sign thereof
-
- 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
- G01R31/12—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing
- G01R31/1209—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing using acoustic measurements
-
- 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
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/16—Cables, cable trees or wire harnesses
-
- 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
- G06F2218/14—Classification; Matching by matching peak patterns
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Water Supply & Treatment (AREA)
- Strategic Management (AREA)
- Computer Hardware Design (AREA)
- Public Health (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Geometry (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Acoustics & Sound (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Locating Faults (AREA)
- Testing Relating To Insulation (AREA)
Abstract
The invention provides a generation method of a cable shaft fire state identification decision tree model, which comprises the following steps of S1, collecting a residual current value and a residual voltage value of a cable in a cable shaft, obtaining a resistive residual current component in the residual current, and obtaining fault arc characteristics of the residual voltage; step S2, collecting ultrasonic signals generated by partial discharge in the cable shaft, and positioning according to the ultrasonic signals to determine the size and the position of the partial discharge as partial discharge information; s3, collecting the environment temperature, the detection point temperature and the smoke concentration in the cable shaft; s4, evaluating the characteristic items through preset rules to generate an evaluation result; and inputting the training data serving as training data into a preset model to train according to the evaluation result and the input corresponding characteristics, and generating a decision tree model. The invention effectively reduces the precision requirement of the monitoring equipment, and has strong real-time performance, high monitoring effect and high response speed.
Description
Technical Field
The invention relates to the technical field of power system automation, in particular to a method for generating a decision tree model for identifying fire states of a cable shaft.
Background
The cable shaft is a vertical shaft (comprising a floor power distribution booth) through which various wires and cables in a building pass, and the power wires and cables, weak current circuits, equipment and the like are erected on the shaft wall, so that the cable shaft is a strong current and weak current transmission and distribution hub of the building. Most of electric wires, cables, equipment and the like in the cable well are inflammable and combustible, live running is carried out, the risk of fire is caused, once a fire accident occurs to the cable, the fire is rapidly developed, and the fire is rapidly spread, so that serious safety accidents and economic losses can be caused. The prior art fire prediction system for cable shafts mainly detects the temperature and smoke in the shaft to judge whether a fire occurs. The method based on the smoke and temperature has certain limitation, and the occurrence of the smoke and the fire light indicates that the fire disaster happens, so that the optimal extinguishing time is easily missed. And the method for analyzing the gas needs a certain time, so that the fire prediction is not real-time enough. Because of the special characteristics of the cable shaft, the fire disaster is basically in a closed and semi-closed state and is not easy to extinguish, so that the rapid and accurate monitoring of the fire disaster in the shaft is particularly critical. The existing detection methods have high requirements on equipment precision, poor monitoring effect and low response speed.
Disclosure of Invention
The invention aims to provide a generation method of a fire state identification decision tree model of a cable shaft, which solves the technical problems of high requirements on equipment precision, poor monitoring effect and low response speed in the prior art.
In one aspect of the invention, a method for generating a cable shaft fire state identification decision tree model is provided, which comprises the following steps:
step S1, collecting a residual current value and a residual voltage value of a cable in a cable shaft, and obtaining a voltage waveform of the residual voltage value; the resistive residual current component in the residual current is obtained according to the residual current value, and the fault arc characteristic of the residual voltage is obtained according to the voltage waveform of the residual voltage value;
step S2, collecting ultrasonic signals generated by partial discharge in the cable shaft, and positioning according to the ultrasonic signals to determine the size and the position of the partial discharge as partial discharge information;
s3, collecting the environment temperature, the detection point temperature and the smoke concentration in the cable shaft;
s4, taking the collected residual current value, residual voltage value, resistive residual current component, fault arc characteristics of residual voltage, partial discharge information, ambient temperature, detection point temperature and smoke concentration as characteristic items, and evaluating the characteristic items through preset rules to generate an evaluation result; and inputting the training data serving as training data into a preset model to train according to the evaluation result and the input corresponding characteristics, and generating a decision tree model.
Preferably, the step S1 includes: converting the collected residual current value according to a preset conversion rule to generate the phase and amplitude of a residual current fundamental wave and corresponding higher harmonic waves;
acquiring original voltage of a cable in a cable shaft, converting the acquired original voltage according to a preset conversion rule, and generating fundamental waves of the original voltage and corresponding amplitude values and phases of various higher harmonic components;
and carrying out resistive component separation on the residual current according to the phase and amplitude of the corresponding higher harmonic wave of the residual current and the amplitude and phase of the corresponding higher harmonic wave component of the original voltage to obtain resistive residual current in the residual current.
Preferably, the step S1 further includes: acquiring the wavelet high-frequency component energy ratio of each layer through a wavelet basis function according to the fault arc characteristics of the residual voltage, and identifying similar waveforms through the wavelet high-frequency component periodic variance value;
and counting the times that the fault arc characteristic value of the residual voltage exceeds a set fault threshold value, and judging that the fault arc is detected if the times that the fault arc characteristic value exceeds the set fault threshold value within 0.6s is larger than a preset times threshold value.
Preferably, the step S2 includes: acquiring attenuation characteristics of the acquired ultrasonic signals, and positioning a discharge position according to a peak value or an effective value in the attenuation characteristics, wherein the peak value or the effective value in the attenuation characteristics becomes larger, and judging that the signal source is closer;
and calculating space coordinates through a simultaneous spherical equation or a hyperboloid equation set according to the time difference of the ultrasonic signals reaching the sensor, and carrying out accurate positioning.
Preferably, the step S4 includes: and evaluating the fire state of the cable shaft by taking the collected residual current value, residual voltage value, resistive residual current component, fault arc characteristics of residual voltage, partial discharge information, ambient temperature, detection point temperature and smoke concentration as characteristics to generate an evaluation result.
Preferably, the step S4 further includes: and taking the evaluation result and all the input corresponding characteristic values as a training data set, judging the categories of all the characteristic values in the training data set according to a preset category identification rule, judging the decision tree as a single node tree if the categories of all the characteristic values in the training data set belong to the same category, taking the category of the characteristic values as the single node category of the decision tree, and returning the decision tree.
Preferably, the step S4 further includes: judging whether the feature set in the training data set is equal to a preset judgment value, if the feature set is equal to the preset judgment value, judging the decision tree as a single-node tree, taking the class with the largest number of feature values in the training data set as the class of the single node of the decision tree, and returning to the decision tree;
if the feature set is not equal to the preset judgment value, calculating the information gain ratio of each feature in the feature set to the training data set, and determining the feature Ag with the maximum information gain ratio value.
Preferably, the step S4 further includes: judging whether the value of the information gain ratio of the characteristic Ag is smaller than a preset judging threshold value, if the value of the information gain ratio of the characteristic Ag is smaller than the preset judging threshold value, judging the decision tree as a single-node tree, taking the class with the largest quantity of the characteristic values in the training data set as the class of the single node of the decision tree, and returning to the decision tree;
if the value of the information gain ratio of the feature Ag is not smaller than the preset judging threshold value, according to each possible value ai of the feature Ag, the training data set is divided into a plurality of non-empty subsets Di, the category with the largest number of the feature values in the non-empty subsets Di is used as a mark, the mark is used as a root node and sub-nodes thereof to form a decision tree, and the decision tree is returned.
Preferably, the step S4 further includes: and for the marked child nodes, the non-empty subset Di is taken as a training set, the characteristic Ag is taken as a characteristic set, and a preset model is recursively called for training to obtain a child decision tree.
Preferably, the method further comprises: updating the generated decision tree into a preset model, and identifying the state of the cable shaft through the updated preset model to generate an identification result; the identification result comprises normal, early warning and alarming.
In summary, the embodiment of the invention has the following beneficial effects:
according to the method for generating the decision tree model for identifying the fire state of the cable shaft, the decision tree model is obtained through residual current detection, fault arc detection and partial discharge detection based on the conventional cable shaft example training, the model is perfected, the condition of the cable shaft is identified through the decision tree model, the accuracy requirement of monitoring equipment is effectively reduced, the real-time performance is high, the monitoring effect is high, and the response speed is high.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are required in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that it is within the scope of the invention to one skilled in the art to obtain other drawings from these drawings without inventive faculty.
Fig. 1 is a schematic flow chart of a method for generating a decision tree model for identifying fire status of a cable shaft according to an embodiment of the present invention.
FIG. 2 is a schematic diagram of a decision tree in an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings, for the purpose of making the objects, technical solutions and advantages of the present invention more apparent.
Fig. 1 is a schematic diagram of an embodiment of a method for generating a decision tree model for identifying fire status of a cable shaft according to the present invention. In this embodiment, the method comprises the steps of:
step S1, collecting a residual current value and a residual voltage value of a cable in a cable shaft, and obtaining a voltage waveform of the residual voltage value; the resistive residual current component in the residual current is obtained according to the residual current value, and the fault arc characteristic of the residual voltage is obtained according to the voltage waveform of the residual voltage value;
it can be understood that the residual current value in the cable is collected through the residual current sensor, meanwhile, the voltage value in the cable is collected, the voltage waveform is obtained, the sampling data of the original voltage is formed, the residual current and the voltage are preprocessed, the resistive component of the residual current is separated and calculated, and the resistive residual current component in the residual current is obtained and transmitted to the cloud in real time.
In a specific embodiment, the leakage current in the power distribution system is formed by various impedances, and is generally caused by leakage resistance and capacitance, the wires and the cable core wires are aluminum wires or copper wires, the distributed capacitance exists between the metal object and the ground, and the current is necessarily generated by the distributed capacitance, and is called capacitive current or leakage current, which is the reason that the leakage current can be detected by the cable with good insulation. Since i_0 is not a leakage current due to actual insulation degradation, it is a combination of a capacitive component leakage current i_oc (a leakage current due to the presence of distribution lines and equipment to ground distributed capacitance even if the grid has no ground fault) and a resistive component current i_or (a leakage current due to a ground fault due to dielectric breakdown of the grid, which is an actual source of various accidents and induced electrical fire). In this embodiment, for detecting the residual current, specifically, converting the collected residual current value according to a preset conversion rule, to generate a phase and an amplitude of a residual current fundamental wave and a corresponding higher harmonic wave; acquiring original voltage of a cable in a cable shaft, converting the acquired original voltage according to a preset conversion rule, and generating fundamental waves of the original voltage and corresponding amplitude values and phases of various higher harmonic components; and carrying out resistive component separation on the residual current according to the phase and amplitude of the corresponding higher harmonic wave of the residual current and the amplitude and phase of the corresponding higher harmonic wave component of the original voltage to obtain resistive residual current in the residual current. Detecting residual voltage, namely acquiring the wavelet high-frequency component energy ratio of each layer according to the fault arc characteristics of the residual voltage through a wavelet basis function, and identifying similar waveforms through the wavelet high-frequency component periodic variance value; and counting the times that the fault arc characteristic value of the residual voltage exceeds a set fault threshold value, and judging that the fault arc is detected when the times that the fault arc characteristic value exceeds the set fault threshold value exceeds the threshold value for more than 8 times within 0.6 s.
Step S2, collecting ultrasonic signals generated by partial discharge in the cable shaft, and positioning according to the ultrasonic signals to determine the size and the position of the partial discharge as partial discharge information;
it will be appreciated that the electrical device will generate sound waves during discharge. The spectrum of sound waves generated by the discharge is very broad, ranging from tens of Hz to MHz, where signals below 20kHz can be heard by the human ear, and ultrasonic signals above this frequency must be received with an ultrasonic sensor. The intensity of the discharge can be estimated by measuring the sound pressure of the ultrasonic signal, based on the relationship between the energy released by the discharge and the acoustic energy, using the change in the sound pressure of the ultrasonic signal to represent the change in the energy released by the partial discharge.
In a specific embodiment, acquiring attenuation characteristics of an acquired ultrasonic signal, and positioning a discharge position according to a peak value or an effective value in the attenuation characteristics, wherein the peak value or the effective value in the attenuation characteristics becomes larger, and then determining that the signal source is closer; and calculating space coordinates through a simultaneous spherical equation or a hyperboloid equation set according to the time difference of the ultrasonic signals reaching the sensor, and carrying out accurate positioning. It can be understood that the ultrasonic partial discharge positioning has two types of amplitude positioning and time difference positioning. The amplitude positioning is based on the attenuation characteristic of the ultrasonic signal, and the amplitude positioning is based on the peak value or the effective value, wherein the closer to the signal source, the larger the signal is; the time difference positioning is to calculate the space coordinates through the simultaneous spherical equation or the hyperboloid equation set according to the time difference of the ultrasonic signal reaching the sensor, and the precision can reach 10cm. The method can adopt an amplitude method to perform preliminary positioning, and then whether further accurate positioning is needed or not is determined according to the field requirement; in addition, due to different structures inside the equipment, the ultrasonic signal propagation has certain complexity, and a positioning method such as acoustic-electric combination can be adopted. When the power equipment detects that the ultrasonic partial discharge signal is abnormal, short-term online monitoring or detection by other methods should be performed. Specifically, the ultrasonic abnormal signal analysis is preferably performed by a comparison method of typical waveforms, a lateral analysis method, and a trend analysis method. The typical waveform comparison method is to comprehensively consider field interference factors, and then obtain an ultrasonic signal truly representing the internal abnormality of the target and compare the ultrasonic signal with a typical waveform chart library; the transverse analysis method is to compare the signal of the target part with the signal of the adjacent area or the signal of the other same part; the trend analysis method is to compare the signal of the target part with the historical data to determine whether the signal has a clear growing trend; the influence of the working condition factors should be comprehensively considered during the analysis of the abnormal signals.
S3, collecting the environment temperature, the detection point temperature and the smoke concentration in the cable shaft; it will be appreciated that the temperature sensor is used to collect the ambient temperature and the detection point temperature and the smoke sensor is used to detect the smoke concentration in the shaft.
S4, taking the collected residual current value, residual voltage value, resistive residual current component, fault arc characteristics of residual voltage, partial discharge information, ambient temperature, detection point temperature and smoke concentration as characteristic items, and evaluating the characteristic items through preset rules to generate an evaluation result; and inputting the training data serving as training data into a preset model to train according to the evaluation result and the input corresponding characteristics, and generating a decision tree model.
It will be appreciated that, based on the information gain ratio of the different features (the information gain ratio: the information gain ratio of the feature a to the training data set D is defined as the ratio of the information gain of the feature a to the entropy of the training data set about the value of the feature a), the magnitudes of the information gain ratios are compared, the feature with the largest information gain ratio is selected as the root node until each instance is covered by a path or a rule, and is covered by only a path or a rule, the collected multiple sets of data are used as input features to evaluate the fire state of the cable shaft in combination with the actual situation, the input feature value and the output judgment result are used as training data, and a decision tree model is generated by applying a C4.5 algorithm, and the decision tree generation schematic diagram is shown in fig. 2. And verifying the accuracy of decision tree discrimination in the test set, and continuously adjusting the parameters of the decision tree model, such as pruning, generating new attributes by combining attribute association degree and the like until the effect is satisfied.
In a specific embodiment, the collected residual current value, residual voltage value, resistive residual current component, fault arc characteristics of residual voltage, partial discharge information, ambient temperature, detection point temperature and smoke concentration are used as characteristics to evaluate the fire disaster state of the cable shaft, and an evaluation result is generated. It can be understood that the collected residual current, voltage, current, partial discharge amplitude (for example, the sound wave size reflects the physical quantity for the ultrasonic method), environmental temperature, detection point temperature and other index data in the monitoring line are taken as input features, the fire state of the cable shaft is evaluated in combination with actual conditions, and the input feature values and the output judging result are taken as training data.
Specifically, a decision tree model is generated from the training set, wherein the inputs: training a data set D, and a feature set A threshold epsilon; and (3) outputting: decision tree T. And taking the evaluation result and all the input corresponding characteristic values as a training data set, judging the categories of all the characteristic values in the training data set according to a preset category identification rule, judging the decision tree as a single node tree if the categories of all the characteristic values in the training data set belong to the same category, taking the category of the characteristic values as the single node category of the decision tree, and returning the decision tree. It will be appreciated that if all instances in D belong to the same class Ck, then the decision tree is a single junction tree and Ck is returned to T as the class of the junction.
Judging whether the feature set in the training data set is equal to a preset judgment value, if the feature set is equal to the preset judgment value, judging the decision tree as a single-node tree, and taking the category with the largest number of feature values in the training data set as the categoryReturning the class of single nodes of the decision tree to the decision tree; it will be appreciated that ifSetting T as a single node tree, taking the class Ck with the largest instance number in D as the class of the node, and returning to T.
If the feature set is not equal to the preset judgment value, calculating the information gain ratio of each feature in the feature set to the training data set, and determining the feature Ag with the maximum information gain ratio value. It is understood that the information gain ratio of each feature in a to D is calculated, and the feature Ag having the largest information gain ratio is selected.
More specifically, judging whether the value of the information gain ratio of the characteristic Ag is smaller than a preset judging threshold value, if the value of the information gain ratio of the characteristic Ag is smaller than the preset judging threshold value, judging the decision tree as a single-node tree, taking the class with the largest number of the characteristic values in the training data set as the class of the single node of the decision tree, and returning to the decision tree; if the value of the information gain ratio of the feature Ag is not smaller than the preset judging threshold value, according to each possible value ai of the feature Ag, the training data set is divided into a plurality of non-empty subsets Di, the category with the largest number of the feature values in the non-empty subsets Di is used as a mark, the mark is used as a root node and sub-nodes thereof to form a decision tree, and the decision tree is returned. It can be understood that if the information gain ratio of Ag is smaller than the threshold epsilon, setting T as a single-node tree, and returning T by taking the class Ck with the largest number of instances in D as the class of the node; otherwise, for each possible value ai of Ag, dividing D into a subset of a plurality of non-empty Di according to ag=ai, using the class with the largest number of instances in Di as a marker, constructing sub-nodes, forming a tree T by the nodes and the sub-nodes thereof, and returning to T.
And for the marked child nodes, the non-empty subset Di is taken as a training set, the characteristic Ag is taken as a characteristic set, and a preset model is recursively called for training to obtain a child decision tree. It can be understood that, for the node i, di is used as a training set, A- { Ag } is used as a feature set, and step four is recursively invoked to obtain a subtree Ti, and Ti is returned.
Step S5, updating the generated decision tree into a preset model, and identifying the state of the cable shaft through the updated preset model to generate an identification result; the identification result comprises normal, early warning and alarming.
In summary, the embodiment of the invention has the following beneficial effects:
according to the method for generating the cable shaft fire state identification decision tree model, the decision tree is obtained through residual current detection, fault arc detection and partial discharge detection based on the conventional cable shaft example training, the model is perfected, the state of the cable shaft is identified through the decision tree, the accuracy requirement of monitoring equipment is effectively reduced, the real-time performance is high, the monitoring effect is high, and the response speed is high.
The above disclosure is only a preferred embodiment of the present invention, and it is needless to say that the scope of the invention is not limited thereto, and therefore, the equivalent changes according to the claims of the present invention still fall within the scope of the present invention.
Claims (8)
1. The method for generating the cable shaft fire state identification decision tree model is characterized by comprising the following steps of:
step S1, collecting a residual current value and a residual voltage value of a cable in a cable shaft, and obtaining a voltage waveform of the residual voltage value; the resistive residual current component in the residual current is obtained according to the residual current value, and the fault arc characteristic of the residual voltage is obtained according to the voltage waveform of the residual voltage value;
step S2, collecting ultrasonic signals generated by partial discharge in the cable shaft, and positioning according to the ultrasonic signals to determine the size and the position of the partial discharge as partial discharge information;
s3, collecting the environment temperature, the detection point temperature and the smoke concentration in the cable shaft;
s4, taking the collected residual current value, residual voltage value, resistive residual current component, fault arc characteristics of residual voltage, partial discharge information, ambient temperature, detection point temperature and smoke concentration as characteristic items, and evaluating the characteristic items through preset rules to generate an evaluation result; inputting the training data into a preset model according to the evaluation result and the input corresponding characteristics to train, and generating a decision tree model;
wherein, the step S1 includes:
converting the collected residual current value according to a preset conversion rule to generate the phase and amplitude of a residual current fundamental wave and corresponding higher harmonic waves;
acquiring original voltage of a cable in a cable shaft, converting the acquired original voltage according to a preset conversion rule, and generating fundamental waves of the original voltage and corresponding amplitude values and phases of various higher harmonic components;
and carrying out resistive component separation on the residual current according to the phase and amplitude of the corresponding higher harmonic wave of the residual current and the amplitude and phase of the corresponding higher harmonic wave component of the original voltage to obtain resistive residual current in the residual current.
2. The method of claim 1, wherein the step S1 further comprises:
acquiring the wavelet high-frequency component energy ratio of each layer through a wavelet basis function according to the fault arc characteristics of the residual voltage, and identifying similar waveforms through the wavelet high-frequency component periodic variance value;
and counting the times that the fault arc characteristic value of the residual voltage exceeds a set fault threshold value, and judging that the fault arc is detected if the times that the fault arc characteristic value exceeds the set fault threshold value within 0.6s is larger than a preset times threshold value.
3. The method according to claim 2, wherein the step S2 includes:
acquiring attenuation characteristics of the acquired ultrasonic signals, and positioning a discharge position according to a peak value or an effective value in the attenuation characteristics, wherein the peak value or the effective value in the attenuation characteristics becomes larger, and judging that the signal source is closer;
and calculating space coordinates through a simultaneous spherical equation or a hyperboloid equation set according to the time difference of the ultrasonic signals reaching the sensor, and carrying out accurate positioning.
4. The method of claim 3, wherein said step S4 further comprises:
and taking the evaluation result and all the input corresponding characteristic values as a training data set, judging the categories of all the characteristic values in the training data set according to a preset category identification rule, judging the decision tree as a single node tree if the categories of all the characteristic values in the training data set belong to the same category, taking the category of the characteristic values as the single node category of the decision tree, and returning the decision tree.
5. The method of claim 4, wherein said step S4 further comprises:
judging whether the feature set in the training data set is equal to a preset judgment value, if the feature set is equal to the preset judgment value, judging the decision tree as a single-node tree, taking the class with the largest number of feature values in the training data set as the class of the single node of the decision tree, and returning to the decision tree;
if the feature set is not equal to the preset judgment value, calculating the information gain ratio of each feature in the feature set to the training data set, and determining the feature Ag with the maximum information gain ratio value.
6. The method of claim 5, wherein said step S4 further comprises:
judging whether the value of the information gain ratio of the characteristic Ag is smaller than a preset judging threshold value, if the value of the information gain ratio of the characteristic Ag is smaller than the preset judging threshold value, judging the decision tree as a single-node tree, taking the class with the largest quantity of the characteristic values in the training data set as the class of the single node of the decision tree, and returning to the decision tree;
if the value of the information gain ratio of the feature Ag is not smaller than the preset judging threshold value, according to each possible value ai of the feature Ag, the training data set is divided into a plurality of non-empty subsets Di, the category with the largest number of the feature values in the non-empty subsets Di is used as a mark, the mark is used as a root node and sub-nodes thereof to form a decision tree, and the decision tree is returned.
7. The method of claim 6, wherein the step S4 further comprises:
and for the marked child nodes, the non-empty subset Di is taken as a training set, the characteristic Ag is taken as a characteristic set, and a preset model is recursively called for training to obtain a child decision tree.
8. The method of claim 1, wherein the method further comprises: updating the generated decision tree into a preset model, and identifying the state of the cable shaft through the updated preset model to generate an identification result; the identification result comprises normal, early warning and alarming.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011347416.2A CN112487714B (en) | 2020-11-26 | 2020-11-26 | Method for generating cable shaft fire state identification decision tree model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011347416.2A CN112487714B (en) | 2020-11-26 | 2020-11-26 | Method for generating cable shaft fire state identification decision tree model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112487714A CN112487714A (en) | 2021-03-12 |
| CN112487714B true CN112487714B (en) | 2023-12-22 |
Family
ID=74935497
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011347416.2A Active CN112487714B (en) | 2020-11-26 | 2020-11-26 | Method for generating cable shaft fire state identification decision tree model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112487714B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015086745A1 (en) * | 2013-12-11 | 2015-06-18 | Ecosynrg Ltd | Electric charging system and method |
| CN106251567A (en) * | 2016-10-11 | 2016-12-21 | 广东工业大学 | A kind of Intellective Fire Alarm System |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2762869C (en) * | 2011-12-20 | 2021-09-14 | Premier Lighting Ltd. | Wireless lighting and electrical device control system |
| WO2017058112A1 (en) * | 2015-09-28 | 2017-04-06 | Temasek Polytechnic | Apparatus, system and method for energy management |
-
2020
- 2020-11-26 CN CN202011347416.2A patent/CN112487714B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015086745A1 (en) * | 2013-12-11 | 2015-06-18 | Ecosynrg Ltd | Electric charging system and method |
| CN106251567A (en) * | 2016-10-11 | 2016-12-21 | 广东工业大学 | A kind of Intellective Fire Alarm System |
Non-Patent Citations (2)
| Title |
|---|
| 遏制电气火灾多发的阻性漏电检测技术分析;吴志强等;消防科学与技术;第39卷(第7期);991-993 * |
| 采棉机火灾预警检测装置的设计与试验研究;鱼博;中国优秀硕士学位论文全文数据库 农业科技辑(第02期);D044-169 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112487714A (en) | 2021-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110275094B (en) | High-voltage cable partial discharge acousto-optic combined detection method and system | |
| CN110007197B (en) | Series arc fault detection system and detection method | |
| Rudd et al. | A generic knowledge-based approach to the analysis of partial discharge data | |
| CN109116126B (en) | Method and system for monitoring and evaluating shielding effectiveness of shielding layer of communication cable on line | |
| CN109799434A (en) | PD Pattern Recognition system and method | |
| KR101550588B1 (en) | Junction box for solar power generating system having a function of diagnosing high temperature and arc/corona discharge by detecting infrared and ultrasonic waves | |
| CN103576059A (en) | Integrated fault diagnosis method and system for turn-to-turn discharging of transformer | |
| CN103830866A (en) | Fire detection, prevention and control system of electronic control cabinet and control method thereof | |
| KR101533948B1 (en) | Monitoring system for electric power equipment | |
| CN110907755A (en) | Power transmission line online monitoring fault recognition method | |
| CN110647924B (en) | GIS equipment state evaluation method based on support vector description and K-nearest neighbor algorithm | |
| Wang et al. | Reflectometry-based cable insulation aging diagnosis and prognosis | |
| CN115128345B (en) | Power grid safety early warning method and system based on harmonic monitoring | |
| CN115542099B (en) | Online GIS partial discharge detection method and device | |
| CN112782540A (en) | High-voltage cable on-line monitoring and fault point positioning device | |
| JP2002090412A (en) | Partial-discharge measuring apparatus | |
| CN110673075B (en) | Method for evaluating electromagnetic interference resistance of ultrahigh frequency partial discharge detector | |
| CN112487714B (en) | Method for generating cable shaft fire state identification decision tree model | |
| CN109635430A (en) | Grid power transmission route transient signal monitoring method and system | |
| CN110398363A (en) | It is a kind of based on the RV retarder performance state monitoring method graphically converted | |
| CN104297595A (en) | Method for improving electronic equipment radiated susceptibility test repeatability in reverberation room constant stirring mode | |
| CN110501154B (en) | GIS equipment fault detection and positioning method based on MOSVR and boxplot analysis | |
| CN112162181A (en) | Monitoring method, monitoring device and computer readable storage medium | |
| Liu et al. | Partial Discharge Detection Method for Distribution Network Based on Feature Engineering | |
| Oh et al. | A study on improvement of resource efficiency for IoT-based pipe leak detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |