CN113655391A - Energy storage power station battery fault diagnosis method based on LightGBM model - Google Patents
Energy storage power station battery fault diagnosis method based on LightGBM model Download PDFInfo
- Publication number
- CN113655391A CN113655391A CN202110985262.8A CN202110985262A CN113655391A CN 113655391 A CN113655391 A CN 113655391A CN 202110985262 A CN202110985262 A CN 202110985262A CN 113655391 A CN113655391 A CN 113655391A
- Authority
- CN
- China
- Prior art keywords
- battery
- battery pack
- lightgbm
- model
- data
- 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
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000004146 energy storage Methods 0.000 title claims abstract description 30
- 238000003745 diagnosis Methods 0.000 title abstract description 7
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000001514 detection method Methods 0.000 claims abstract description 22
- 238000004458 analytical method Methods 0.000 claims abstract description 12
- 238000002372 labelling Methods 0.000 claims abstract description 5
- 238000003066 decision tree Methods 0.000 claims description 18
- 238000012795 verification Methods 0.000 claims description 14
- 238000012360 testing method Methods 0.000 claims description 12
- 238000005259 measurement Methods 0.000 claims description 11
- 230000007547 defect Effects 0.000 claims description 9
- 238000013135 deep learning Methods 0.000 claims description 8
- 238000003384 imaging method Methods 0.000 claims description 8
- 230000002015 leaf growth Effects 0.000 claims description 6
- WHXSMMKQMYFTQS-UHFFFAOYSA-N Lithium Chemical compound [Li] WHXSMMKQMYFTQS-UHFFFAOYSA-N 0.000 claims description 5
- 229910052744 lithium Inorganic materials 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 238000012423 maintenance Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 abstract description 5
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 21
- 238000012545 processing Methods 0.000 description 9
- 238000004364 calculation method Methods 0.000 description 6
- 238000003860 storage Methods 0.000 description 5
- 239000002253 acid Substances 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 238000005086 pumping Methods 0.000 description 3
- 238000007599 discharging Methods 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 239000000523 sample Substances 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 238000000547 structure data Methods 0.000 description 2
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- OJIJEKBXJYRIBZ-UHFFFAOYSA-N cadmium nickel Chemical compound [Ni].[Cd] OJIJEKBXJYRIBZ-UHFFFAOYSA-N 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000012983 electrochemical energy storage Methods 0.000 description 1
- 229910052739 hydrogen Inorganic materials 0.000 description 1
- 239000001257 hydrogen Substances 0.000 description 1
- 229910001416 lithium ion Inorganic materials 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000015654 memory Effects 0.000 description 1
- 230000003446 memory effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
Images
Classifications
-
- 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
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
- G01R31/3842—Arrangements for monitoring battery or accumulator variables, e.g. SoC combining voltage and current measurements
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/04—Analysing solids
-
- 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
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/389—Measuring internal impedance, internal conductance or related variables
-
- 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
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/396—Acquisition or processing of data for testing or for monitoring individual cells or groups of cells within a battery
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2291/00—Indexing codes associated with group G01N29/00
- G01N2291/02—Indexing codes associated with the analysed material
- G01N2291/023—Solids
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2291/00—Indexing codes associated with group G01N29/00
- G01N2291/02—Indexing codes associated with the analysed material
- G01N2291/028—Material parameters
- G01N2291/0289—Internal structure, e.g. defects, grain size, texture
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
The invention discloses a battery fault diagnosis method of an energy storage power station based on a LightGBM model, in particular to the technical field of battery detection, firstly grouping and labeling battery packs, then measuring characteristic data of the battery packs through a measuring module, establishing a database by combining with the pack numbers, establishing a computer analysis module, establishing a battery pack parameter prediction model of the LightGBM based on characteristics, obtaining an optimal hyper-parameter by adjusting relevant parameters in a LightGBM algorithm and then training and verifying the model by utilizing massive data in the database, predicting whether the battery packs will be in fault or not according to the optimal hyper-parameter and the battery pack data characteristics monitored by the measuring module in real time, sending voice early warning through a mechanical module according to the prediction result, sending voice early warning when the battery suddenly fails in an abnormal state, the invention can accurately predict whether the battery has faults or not, find and repair the battery in time and stop the loss in time.
Description
Technical Field
The invention belongs to the technical field of battery detection, and particularly provides an energy storage power station battery fault diagnosis method based on a LightGBM model.
Background
In hydroelectric power generation, in order to adjust peak-valley power consumption in daily adjustment and annual adjustment and achieve balance of power consumption, an energy storage power station needs to be built, energy is usually stored in the energy storage power station in two modes, the first mode is a water pumping energy storage power station, a water pumping and power generating dual-purpose unit is installed in the water pumping energy storage power station, water can be pumped and generated, water is discharged from a reservoir in daytime and in the first night, water with high water level passes through the dual-purpose unit, the dual-purpose unit is used as a generator at the moment, mechanical energy of the water with high water level is converted into electric energy to be transmitted to a power grid, the problem that the power grid is in low valley when the power consumption is in peak time and in the second night, the power grid cannot store the electric energy, the dual-purpose unit is used as a water pump (the dual-purpose unit can rotate reversely), the redundant electric energy in the power grid is utilized, the water with low water level is pumped to high water level and is injected into a reservoir with high water level, thus, when the electricity consumption is low, the redundant electric energy in the power grid is converted into mechanical energy of water to be stored in the reservoir, when the electricity consumption is high, the reservoir discharges water, the mechanical energy of the water is converted into electric energy through the generator to be transmitted to the power grid, the water in the reservoir is used for multiple times, the two units are used together to finish multiple conversion of energy, the high-water level reservoir stores a large amount of water with low water level, which is equivalent to the redundant electric energy in the power grid, and the generated electric energy is stored through the secondary battery pack completely, so that an energy storage power station battery is needed, and the energy storage of a rechargeable battery, namely a secondary battery, also called a storage battery, is an electrochemical energy storage mode, and due to the limitation of factors such as price, energy storage density and the like, the rechargeable battery is not placed in the energy storage range of the energy field in the past, but along with the technical progress, the situation that the rechargeable battery is used for large-scale energy storage is gradually increased, the storage battery has become basic energy storage equipment, the battery has many types, the lead-acid battery is the most familiar storage battery, the sealed maintenance-free lead-acid battery has become the mainstream of the battery, the cadmium-nickel battery in the alkaline battery has been gradually replaced by the nickel-hydrogen battery, compared with the alkaline power supply, the lead-acid battery has the advantages of large capacity, firm structure, many times of charge and discharge cycles and the like, but the price is much higher, thereby limiting the application of the lead-acid battery in the field of energy. The other battery with excellent performance is a lithium ion secondary battery widely used in recent years, thoroughly solves the memory effect of charging and discharging, greatly facilitates use, basically avoids pollution to the environment in the manufacturing process, is called as a green battery, and is used in an energy storage power station.
The existing battery detection system adopts a hardware closed-loop feedback method, detects the battery through special detection hardware, and realizes the detection of the charging and discharging voltage value of the battery, but in a large-scale battery pack in an energy storage power station, the detection method is inflexible, can detect the battery only when the battery has a problem, cannot early warn, and cannot simultaneously meet the problems of quick adjustment and accurate adjustment, and aiming at the defects of the detection method, a new solution is necessary to be provided.
In order to solve the problems that an existing detection method is inflexible and cannot achieve early warning, a method for diagnosing battery faults of an energy storage power station based on a LightGBM model is provided.
Disclosure of Invention
The invention aims to provide a LightGBM model-based energy storage power station battery fault diagnosis method, which solves the problems that the existing energy storage power station battery detection is not perfect enough, the battery detection is not flexible enough, and early warning cannot be achieved.
The invention is realized by the following technical scheme: a method for diagnosing battery faults of an energy storage power station based on a LightGBM model comprises the following steps:
s1: dividing a plurality of batteries of the energy storage power station into battery packs according to an access circuit, labeling each battery pack, connecting a measurement module to each battery pack, and measuring the conventional battery characteristics of the charging current, the charging voltage and the internal resistance of each battery pack and the internal structure data of the battery;
s2: uploading the obtained data to a computer analysis module, and establishing a corresponding database by the computer analysis module according to the uploaded battery characteristic data and the label of the battery pack;
s3: the computer analysis module establishes a battery pack parameter prediction model of the LightGBM based on characteristics, distributes the weight of each data characteristic and factor by adjusting related parameters in the LightGBM algorithm, and then trains and verifies the model by utilizing massive data in a database to obtain the optimal hyper-parameter;
s4: after massive data training and verified deep learning, the LightGBM model predicts whether the battery pack will break down or not according to the optimal hyper-parameter and the battery pack data characteristics monitored by the measurement module in real time;
s5: according to the prediction result, the battery pack does not break down, the real-time monitoring is continued, the battery pack breaks down, the computer display module displays that the battery pack breaks down, and the mechanical module sends out voice early warning;
s6: the maintainer carries out the most accurate manual maintenance to this group battery according to the prediction result that shows, finds out the problem and in time maintains, accomplishes in time to stop the loss.
Preferably, the number of batteries, the battery capacity, the total voltage of the series-connected batteries, the same number of batteries, the same battery capacity, and the same longitudinal voltage of the series-connected batteries in each group of batteries are all the same, so that the initial data of each group are the same during detection, and the most accurate result of the LightGBM model after deep learning of mass data training and verification is ensured.
Preferably, the mode that the measuring module detects the battery and then acquires data is that the current measuring module and the voltage measuring module are arranged at the input end and the output end, the ultrasonic C imaging scanning system is used for internal scanning, and the scanned image is processed by a digital image of a computer to obtain the internal structure of the battery pack.
Preferably, the ultrasonic imaging scanning system is based on the fundamental principle of ultrasonic flaw detection, and performs point-by-point line-by-line scanning processing on a detection workpiece in a certain area according to a set path in a mode of sound wave transmission or reflection, and extracts echo signals of a transverse section to form a two-dimensional image, the ultrasonic scanning detects and identifies defects, and determines the positions of the defects in the lithium battery pack to be detected, and the ultrasonic scanning result can be more accurate and efficient by combining digital automatic control equipment in detection, so the technology is widely applied to the field of nondestructive detection by virtue of the advantages of high sensitivity, high automation, high intuition and the like, in order to express the defects of the lithium battery most intuitively, an ultrasonic probe not only needs to scan along the X direction, but also needs to scan along the Y direction, performs certain surface scanning, and receives the echo signals from a transducer to extract signal amplitude, and obtaining an imaging result picture, uploading the image to a computer, and accurately and quickly detecting whether the steam pocket, the solid block and other factors appear in the battery pack or not through gray image segmentation and filtering processing.
Preferably, the battery pack parameter prediction model of the LightGBM is characterized in that a plurality of weak regression trees are firstly established, a strong regression tree is obtained through linear combination, the LightGBM model mainly comprises an improved histogram algorithm and a leaf growth strategy with depth limitation, the histogram algorithm divides continuous current and voltage data into N integers and constructs a histogram with the width of N, discretized values are accumulated in the histogram as index values during traversal, an optimal decision tree segmentation point is searched out, the leaf growth strategy with depth limitation is used, a leaf with the maximum gain is found to be split during each splitting, corresponding parameters are obtained during training of the model, and the battery pack parameter prediction model of the LightGBM is a probability-based decision tree model and has a high-precision prediction result.
Preferably, the prediction process of the battery pack parameter prediction model of LightGBM is as follows: firstly, preprocessing original sampling data, and dividing a training set, a verification set and a test set, wherein the training set is used for training common parameters of connection values of a battery pack parameter prediction model of the LightGBM, tuning of super parameters such as iteration times is desired according to evaluation of the verification set, the test set is used for an actual battery fault test, independent prediction is carried out on days to be predicted in the test set after model training and parameter tuning, and finally a model prediction result is obtained through an optimal weighting method, the LightGBM is an open-source, rapid and efficient lifting frame based on a decision tree algorithm, supports efficient parallel training and gradient lifting, and the idea is as follows: the method comprises the steps of iterating variables at one time, increasing submodels one by one in the iteration process, ensuring that a loss function is continuously reduced, and continuously reducing the loss function towards the gradient of a variable with the second highest information content after adding a new submodel each time, wherein a gradient lifting decision tree is a classical model, has the functional characteristics of gradient lifting and decision trees, and has the advantages of good training effect, difficulty in overfitting and the like, while LightGBM is one of the gradient lifting decision trees and is used for solving the problems of the gradient lifting decision trees in mass data processing, the LightGBM adopts a method of splitting according to leaves, has low calculation cost, avoids overfitting through controlling the depth of the tree and the minimum data volume of each leaf node, selects a decision tree algorithm based on Histopgram, can reduce storage cost and calculation cost, and in addition, the processing of class characteristics also ensures that the LightGBM has better performance improvement under specific data, realizing algorithm control and optimizing a main parameter Num _ leaves, the number of leaves of each tree; learning _ rate, Learning rate; max _ depth, maximum learning depth, for controlling the over-fitting phenomenon; min _ data, the minimum number of data in a leaf, can be used to control the overfitting phenomenon; feature _ fraction, selecting the proportion of the total feature number of the features, wherein the value is between 0 and 1, when the feature _ fraction is less than 0, the LightGBM randomly selects partial features during each iteration, and the feature _ fraction is used for controlling the proportion of the total feature number; and the Bagging _ fraction selects the proportion of the data accounting for the total data volume, takes a value between 0 and 1, is similar to the feature _ fraction, randomly and not repeatedly selects observation of corresponding proportion, sets the proportion to be larger than 0, and accurately predicts whether the battery will be in failure or not in the deep learning of parameter adjustment through the predictive verification, parameter adjustment, re-prediction and re-verification of the LightGBM model.
Preferably, when the battery is measured by the measuring module, if the measurement result is a battery fault, the mechanical module directly sends out a fault early warning to prompt a manager that a certain group of batteries have the battery fault and are repaired in time, and when the batteries in an abnormal state are suddenly damaged, the mechanical module directly sends out the early warning.
The invention has the following beneficial effects:
the invention measures the characteristic data of the battery pack through a measuring module, establishes a database by combining with the pack number, establishes a battery pack parameter prediction model of the LightGBM based on the characteristics through a computer analysis module, obtains the optimal hyper-parameter by continuously adjusting the relevant parameters in the LightGBM algorithm in each verification prediction and then training and verifying the model by utilizing massive data in the database, predicts whether the battery pack will have faults or not according to the optimal hyper-parameter and the battery pack data characteristics monitored in real time by the measuring module, sends out voice early warning through a mechanical module according to the prediction result, sends out voice early warning when the battery has sudden faults under abnormal conditions, the LightGBM is still an improved realization under a DT GBAlgorithm frame, and is a quick, distributed and high-performance GBDT frame based on a decision tree algorithm, the method mainly solves the problem that the efficiency and the expandability of the GBDT framework algorithm are improved when high-dimensionality large data are faced, and the method is realized through three technologies of single-side gradient sampling, mutually exclusive feature combination and a histogram algorithm through fewer samples, fewer features and fewer memories. In addition, in the aspect of engineering, the LightGBM is optimized in the aspect of parallel computing, parallel characteristic and data are supported, the optimization is performed according to respective parallel modes, communication traffic is reduced, the calculation amount is reduced, the calculation speed is higher, the prediction accuracy is higher, whether the battery fails or not can be accurately predicted, timely maintenance is found, loss is timely stopped, the loss of the energy storage battery is further reduced, and the service life of the battery is prolonged.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a diagram: the invention provides an overall operation flow chart of an energy storage power station battery fault diagnosis method based on a LightGBM model;
FIG. 2 is a diagram of: the invention relates to a system block diagram of a method for diagnosing battery faults of an energy storage power station based on a LightGBM model;
FIG. 3 is a diagram of: the invention relates to a LightGBM model prediction flow chart in an energy storage power station battery fault diagnosis method based on a LightGBM model.
Detailed Description
In order to make the technical means, the creation characteristics, the achievement purposes and the effects of the invention easy to understand, the invention is further described with the specific embodiments.
In the description of the present invention, it should be noted that the terms "upper", "lower", "inner", "outer", "front", "rear", "both ends", "one end", "the other end", and the like indicate orientations or positional relationships based on those shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the referred device or element must have a specific orientation, be constructed in a specific orientation, and be operated, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In the description of the present invention, it is to be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "disposed," "connected," and the like are to be construed broadly, such as "connected," which may be fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
Referring to fig. 1-3, a method for diagnosing battery faults of an energy storage power station based on a LightGBM model includes the following steps:
s1: dividing a plurality of batteries of the energy storage power station into battery packs according to an access circuit, labeling each battery pack, connecting a measurement module to each battery pack, and measuring the conventional battery characteristics of the charging current, the charging voltage and the internal resistance of each battery pack and the internal structure data of the battery;
s2: uploading the obtained data to a computer analysis module, and establishing a corresponding database by the computer analysis module according to the uploaded battery characteristic data and the label of the battery pack;
s3: the computer analysis module establishes a battery pack parameter prediction model of the LightGBM based on characteristics, distributes the weight of each data characteristic and factor by adjusting related parameters in the LightGBM algorithm, and then trains and verifies the model by utilizing massive data in a database to obtain the optimal hyper-parameter;
s4: after massive data training and verified deep learning, the LightGBM model predicts whether the battery pack will break down or not according to the optimal hyper-parameter and the battery pack data characteristics monitored by the measurement module in real time;
s5: according to the prediction result, the battery pack does not break down, the real-time monitoring is continued, the battery pack breaks down, the computer display module displays that the battery pack breaks down, and the mechanical module sends out voice early warning;
s6: the maintainer carries out the most accurate manual maintenance to this group battery according to the prediction result that shows, finds out the problem and in time maintains, accomplishes in time to stop the loss.
The purpose that the number of batteries, the capacity of the batteries and the total voltage of the series batteries in each group of batteries are the same, the number of the batteries and the capacity of the batteries are the same, and the longitudinal voltages of the series batteries are the same is to ensure that initial data of each group are the same during detection, and the LightGBM model is guaranteed to be most accurate after deep learning of mass data training and verification.
The mode that measuring module detected the battery and then gained data is, through all setting up current measurement module at input and output, voltage measurement module to and carry out the internal scanning through ultrasonic imaging scanning system, the scanning image passes through computer digital image processing, obtains group battery inner structure.
Wherein, the ultrasonic imaging scanning system is based on the fundamental principle of ultrasonic flaw detection, and carries out point-by-point line-by-line scanning processing of a certain area on a detection workpiece according to a set path in a mode of sound wave transmission or reflection, and extracts echo signals of a transverse section to form a two-dimensional image, the defects are detected and identified by ultrasonic scanning, and the positions of the defects in the lithium battery pack to be detected are determined, and the ultrasonic scanning result can be more accurate and efficient by combining digital automatic control equipment in detection, so the technology is widely applied to the field of nondestructive detection by virtue of the advantages of high sensitivity, high automation, high intuition and the like, in order to express the defects of the lithium battery most intuitively, an ultrasonic probe not only needs to scan along the X direction, but also needs to scan along the Y direction, carries out certain surface scanning, and extracts signal amplitude from the echo signals received by a transducer, and obtaining an imaging result picture, uploading the image to a computer, and accurately and quickly detecting whether the steam pocket, the solid block and other factors appear in the battery pack or not through gray image segmentation and filtering processing.
The battery pack parameter prediction model of the LightGBM is characterized in that a plurality of weak regression trees are firstly established, a strong regression tree is obtained through linear combination, the LightGBM mainly comprises an improved histogram algorithm and a leaf growth strategy with depth limitation, the histogram algorithm divides continuous current and voltage data into N integers and constructs a histogram with the width of N, discretized values are accumulated in the histogram as index values during traversal, an optimal decision tree segmentation point is searched out, the leaf growth strategy with depth limitation is used, a leaf with the maximum gain is found to be split during each splitting, corresponding parameters are obtained during model training, and the battery pack parameter prediction model of the LightGBM is a probability-based decision tree model and has a high-precision prediction result.
The prediction process of the battery pack parameter prediction model of the LightGBM is as follows: firstly, preprocessing original sampling data, and dividing a training set, a verification set and a test set, wherein the training set is used for training common parameters of connection values of a battery pack parameter prediction model of the LightGBM, tuning of super parameters such as iteration times is desired according to evaluation of the verification set, the test set is used for an actual battery fault test, independent prediction is carried out on days to be predicted in the test set after model training and parameter tuning, and finally a model prediction result is obtained through an optimal weighting method, the LightGBM is an open-source, rapid and efficient lifting frame based on a decision tree algorithm, supports efficient parallel training and gradient lifting, and the idea is as follows: the method comprises the steps of iterating variables at one time, increasing submodels one by one in the iteration process, ensuring that a loss function is continuously reduced, and continuously reducing the loss function towards the gradient of a variable with the second highest information content after adding a new submodel each time, wherein a gradient lifting decision tree is a classical model, has the functional characteristics of gradient lifting and decision trees, and has the advantages of good training effect, difficulty in overfitting and the like, while LightGBM is one of the gradient lifting decision trees and is used for solving the problems of the gradient lifting decision trees in mass data processing, the LightGBM adopts a method of splitting according to leaves, has low calculation cost, avoids overfitting through controlling the depth of the tree and the minimum data volume of each leaf node, selects a decision tree algorithm based on Histopgram, can reduce storage cost and calculation cost, and in addition, the processing of class characteristics also ensures that the LightGBM has better performance improvement under specific data, realizing algorithm control and optimizing a main parameter Num _ leaves, the number of leaves of each tree; learning _ rate, Learning rate; max _ depth, maximum learning depth, for controlling the over-fitting phenomenon; min _ data, the minimum number of data in a leaf, can be used to control the overfitting phenomenon; feature _ fraction, selecting the proportion of the total feature number of the features, wherein the value is between 0 and 1, when the feature _ fraction is less than 0, the LightGBM randomly selects partial features during each iteration, and the feature _ fraction is used for controlling the proportion of the total feature number; and the Bagging _ fraction selects the proportion of the data to the total data volume, takes a value between 0 and 1, is similar to the feature _ fraction, randomly and not repeatedly selects observation of corresponding proportion, sets the observation to be the proportion larger than 0, and accurately predicts whether the battery will be in failure or not in the deep learning of parameter adjustment through the predictive verification, parameter adjustment and re-verification of the LightGBM model.
When the battery is measured through the measuring module, the measuring result is a battery fault, then the mechanical module directly sends out fault early warning to prompt a manager that a certain group of batteries have the battery fault and are timely repaired, and when the batteries are suddenly damaged in an abnormal state, the early warning is directly sent out through the mechanical module.
In actual operation, firstly grouping and labeling the battery pack, then measuring the characteristic data of the battery pack through a measuring module, establishing a database by combining with a pack number, establishing a computer analysis module, establishing a light GBM battery pack parameter prediction model based on characteristics, distributing the weight of each data characteristic and factor by adjusting related parameters in a light GBM algorithm, then training and verifying the model by using massive data in the database to obtain an optimal hyper-parameter, after deep learning of mass data training and verification of the light GBM model, predicting whether the battery pack will fail or not according to the optimal hyper-parameter and the battery pack data characteristics monitored by the measuring module in real time,
according to the prediction result, the battery pack can not break down, the side continues to monitor in real time and can break down, the side displays that the battery pack is about to break down through the computer display module, and sends out voice early warning through the mechanical module, when the battery breaks down suddenly under an abnormal state, the mechanical module immediately sends out voice early warning to prompt a worker to overhaul in time.
The foregoing shows and describes the general principles and broad features of the present invention and advantages thereof. It will be understood by those skilled in the art that the present invention is not limited by the foregoing embodiments, which are merely illustrative of the principles of the invention, and that various changes and modifications may be made without departing from the spirit and scope of the invention, which is intended to be protected by the following claims. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (7)
1. A method for diagnosing battery faults of an energy storage power station based on a LightGBM model is characterized by comprising the following steps:
s1: dividing a plurality of batteries of the energy storage power station into battery packs according to an access circuit, labeling each battery pack, connecting a measurement module to each battery pack, measuring the conventional battery characteristics of charging current, charging voltage and internal resistance of each battery pack and the internal structure of the battery, and generating data;
s2: uploading the obtained data to a computer analysis module, and establishing a corresponding database by the computer analysis module according to the uploaded battery characteristic data and the label of the battery pack;
s3: the computer analysis module establishes a battery pack parameter prediction model of the LightGBM based on characteristics, distributes the weight of each data characteristic and factor by adjusting related parameters in the LightGBM algorithm, and then trains and verifies the model by utilizing massive data in a database to obtain the optimal hyper-parameter;
s4: after massive data training and verified deep learning, the LightGBM model predicts whether the battery pack will break down or not according to the optimal hyper-parameter and the battery pack data characteristics monitored by the measurement module in real time;
s5: according to the prediction result, the battery pack does not break down, the real-time monitoring is continued, the battery pack breaks down, the computer display module displays that the battery pack breaks down, and the mechanical module sends out voice early warning;
s6: the maintainer carries out the most accurate manual maintenance to this group battery according to the prediction result that shows, finds out the problem and in time maintains, accomplishes in time to stop the loss.
2. The method as claimed in claim 1, wherein the number of batteries in each group of batteries, the capacity of the batteries and the total voltage of the series batteries are the same.
3. The method as claimed in claim 1, wherein the manner of detecting the battery and acquiring the data by the measuring module is that the current measuring module and the voltage measuring module are arranged at the input end and the output end, and the internal scanning is performed by an ultrasonic imaging scanning system, and the scanned image is processed by a digital image of a computer to obtain the internal structure of the battery pack.
4. The method as claimed in claim 3, wherein the ultrasonic imaging scanning system is used for scanning the detection workpiece point by point and line by line according to a set path in a mode of sound wave transmission or reflection according to the basic principle of ultrasonic flaw detection, extracting an echo signal of a transverse section to form a two-dimensional image, scanning, detecting and identifying the defect, and determining the position of the defect in the detected lithium battery pack.
5. The method for diagnosing the battery faults of the energy storage power station based on the LightGBM model is characterized in that a battery pack parameter prediction model of the LightGBM is established by firstly establishing a plurality of weak regression trees, obtaining a strong regression tree through linear combination, and the LightGBM model mainly improves a histogram algorithm and a leaf growth strategy with depth limitation, wherein the histogram algorithm divides continuous current and voltage data into N integers and constructs a histogram with the width of N, discretized values are accumulated in the histogram as index values during traversal, an optimal decision tree division point is searched out, and the leaf growth strategy with depth limitation finds a leaf with the maximum gain to divide each time, so that corresponding parameters are obtained in the training of the model.
6. The method as claimed in claim 5, wherein the LightGBM battery parameter prediction model comprises a prediction process: the method comprises the steps of firstly preprocessing original sampling data, dividing a training set, a verification set and a test set, wherein the training set is used for training common parameters of connection values of a battery pack parameter prediction model of the LightGBM, adjusting and optimizing super parameters such as iteration times according to evaluation of the verification set, the test set is used for actual battery fault tests, independent prediction is carried out on days to be predicted in the test set after model training and parameter adjustment, and finally a model prediction result is obtained through an optimal weighting method.
7. The method as claimed in claim 1, wherein when the battery fault is measured by the measurement module, and the measurement result is a battery fault, the mechanical module directly sends out a fault early warning to prompt an administrator that a certain group of batteries have a battery fault and are repaired in time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110985262.8A CN113655391A (en) | 2021-08-26 | 2021-08-26 | Energy storage power station battery fault diagnosis method based on LightGBM model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110985262.8A CN113655391A (en) | 2021-08-26 | 2021-08-26 | Energy storage power station battery fault diagnosis method based on LightGBM model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113655391A true CN113655391A (en) | 2021-11-16 |
Family
ID=78482067
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110985262.8A Pending CN113655391A (en) | 2021-08-26 | 2021-08-26 | Energy storage power station battery fault diagnosis method based on LightGBM model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113655391A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115825756A (en) * | 2023-02-16 | 2023-03-21 | 中国华能集团清洁能源技术研究院有限公司 | Distributed energy storage power station fault multi-stage diagnosis method and system |
| CN117371561A (en) * | 2023-10-08 | 2024-01-09 | 杭州亚太化工设备有限公司 | Industrial production artificial intelligence system based on machine learning |
| CN117872168A (en) * | 2024-03-12 | 2024-04-12 | 苏州市洛肯电子科技有限公司 | Intelligent detection method and system for embedded RFID new energy battery |
Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106033113A (en) * | 2015-03-19 | 2016-10-19 | 国家电网公司 | Health state evaluation method for energy-storage battery pack |
| CN107340476A (en) * | 2016-04-29 | 2017-11-10 | 株式会社日立制作所 | The electrical state monitoring system and electrical state monitoring method of battery |
| CN109061495A (en) * | 2018-08-07 | 2018-12-21 | 中国电建集团福建省电力勘测设计院有限公司 | A kind of hybrid energy-storing battery failure diagnostic method |
| CN110118938A (en) * | 2019-04-11 | 2019-08-13 | 华中科技大学 | Method and device based on ultrasonic spectrum analysis lithium battery interior state |
| CN110413494A (en) * | 2019-06-19 | 2019-11-05 | 浙江工业大学 | A kind of LightGBM method for diagnosing faults improving Bayes's optimization |
| CN110535159A (en) * | 2019-07-25 | 2019-12-03 | 中国电力科学研究院有限公司 | A kind of method and system of scale energy-accumulating power station running unit fault pre-alarming |
| CN111898325A (en) * | 2020-08-14 | 2020-11-06 | 天津大学 | Method for predicting remaining service life of power battery of electric automobile |
| CN112485689A (en) * | 2020-10-26 | 2021-03-12 | 江苏慧智能源工程技术创新研究院有限公司 | Method for predicting residual cycle life of lithium battery in energy storage system based on Xgboost model |
| KR20210033851A (en) * | 2019-09-19 | 2021-03-29 | 주식회사 그릿에이트 | Battery safety status diagnostic monitoring system using ultrasonic sensor |
| CN112597691A (en) * | 2020-09-01 | 2021-04-02 | 新天绿色能源股份有限公司 | LightGBM algorithm-based fault early warning method for wind turbine generator variable pitch motor temperature sensor |
| CN112731159A (en) * | 2020-12-23 | 2021-04-30 | 江苏省电力试验研究院有限公司 | Method for pre-judging and positioning battery fault of battery compartment of energy storage power station |
| CN112763929A (en) * | 2020-12-31 | 2021-05-07 | 华东理工大学 | Method and device for predicting health of battery monomer of energy storage power station system |
| CN112785016A (en) * | 2021-02-20 | 2021-05-11 | 南京领行科技股份有限公司 | New energy automobile maintenance and fault monitoring and diagnosis method based on machine learning |
| CN112798961A (en) * | 2021-02-27 | 2021-05-14 | 天津大学 | Method for predicting remaining service life of power battery of electric automobile |
| US20210175553A1 (en) * | 2019-12-04 | 2021-06-10 | Feasible, Inc. | Acoustic signal based analysis of batteries |
| CN113049976A (en) * | 2021-04-27 | 2021-06-29 | 武汉理工大学 | Vehicle battery health state assessment method and system |
| CN113189208A (en) * | 2021-03-17 | 2021-07-30 | 东莞理工学院 | Ultrasonic characteristic detection method and detection system for lithium battery |
| CN113221468A (en) * | 2021-05-31 | 2021-08-06 | 福州大学 | Photovoltaic array fault diagnosis method based on ensemble learning |
-
2021
- 2021-08-26 CN CN202110985262.8A patent/CN113655391A/en active Pending
Patent Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106033113A (en) * | 2015-03-19 | 2016-10-19 | 国家电网公司 | Health state evaluation method for energy-storage battery pack |
| CN107340476A (en) * | 2016-04-29 | 2017-11-10 | 株式会社日立制作所 | The electrical state monitoring system and electrical state monitoring method of battery |
| CN109061495A (en) * | 2018-08-07 | 2018-12-21 | 中国电建集团福建省电力勘测设计院有限公司 | A kind of hybrid energy-storing battery failure diagnostic method |
| CN110118938A (en) * | 2019-04-11 | 2019-08-13 | 华中科技大学 | Method and device based on ultrasonic spectrum analysis lithium battery interior state |
| CN110413494A (en) * | 2019-06-19 | 2019-11-05 | 浙江工业大学 | A kind of LightGBM method for diagnosing faults improving Bayes's optimization |
| CN110535159A (en) * | 2019-07-25 | 2019-12-03 | 中国电力科学研究院有限公司 | A kind of method and system of scale energy-accumulating power station running unit fault pre-alarming |
| KR20210033851A (en) * | 2019-09-19 | 2021-03-29 | 주식회사 그릿에이트 | Battery safety status diagnostic monitoring system using ultrasonic sensor |
| US20210175553A1 (en) * | 2019-12-04 | 2021-06-10 | Feasible, Inc. | Acoustic signal based analysis of batteries |
| CN111898325A (en) * | 2020-08-14 | 2020-11-06 | 天津大学 | Method for predicting remaining service life of power battery of electric automobile |
| CN112597691A (en) * | 2020-09-01 | 2021-04-02 | 新天绿色能源股份有限公司 | LightGBM algorithm-based fault early warning method for wind turbine generator variable pitch motor temperature sensor |
| CN112485689A (en) * | 2020-10-26 | 2021-03-12 | 江苏慧智能源工程技术创新研究院有限公司 | Method for predicting residual cycle life of lithium battery in energy storage system based on Xgboost model |
| CN112731159A (en) * | 2020-12-23 | 2021-04-30 | 江苏省电力试验研究院有限公司 | Method for pre-judging and positioning battery fault of battery compartment of energy storage power station |
| CN112763929A (en) * | 2020-12-31 | 2021-05-07 | 华东理工大学 | Method and device for predicting health of battery monomer of energy storage power station system |
| CN112785016A (en) * | 2021-02-20 | 2021-05-11 | 南京领行科技股份有限公司 | New energy automobile maintenance and fault monitoring and diagnosis method based on machine learning |
| CN112798961A (en) * | 2021-02-27 | 2021-05-14 | 天津大学 | Method for predicting remaining service life of power battery of electric automobile |
| CN113189208A (en) * | 2021-03-17 | 2021-07-30 | 东莞理工学院 | Ultrasonic characteristic detection method and detection system for lithium battery |
| CN113049976A (en) * | 2021-04-27 | 2021-06-29 | 武汉理工大学 | Vehicle battery health state assessment method and system |
| CN113221468A (en) * | 2021-05-31 | 2021-08-06 | 福州大学 | Photovoltaic array fault diagnosis method based on ensemble learning |
Non-Patent Citations (1)
| Title |
|---|
| 陈纬楠等: "基于长短期记忆网络和LightGBM组合模型的短期负荷预测", 电力系统自动化, vol. 45, no. 4, 25 February 2021 (2021-02-25), pages 91 - 97 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115825756A (en) * | 2023-02-16 | 2023-03-21 | 中国华能集团清洁能源技术研究院有限公司 | Distributed energy storage power station fault multi-stage diagnosis method and system |
| CN117371561A (en) * | 2023-10-08 | 2024-01-09 | 杭州亚太化工设备有限公司 | Industrial production artificial intelligence system based on machine learning |
| CN117872168A (en) * | 2024-03-12 | 2024-04-12 | 苏州市洛肯电子科技有限公司 | Intelligent detection method and system for embedded RFID new energy battery |
| CN117872168B (en) * | 2024-03-12 | 2024-06-11 | 苏州市洛肯电子科技有限公司 | Intelligent detection method and system for embedded RFID new energy battery |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113655391A (en) | Energy storage power station battery fault diagnosis method based on LightGBM model | |
| WO2021169486A1 (en) | Method, system and apparatus for monitoring battery impedance abnormality on basis of charging process | |
| CN101067644B (en) | Storage battery performance analytical expert diagnosing method | |
| EP3770620B1 (en) | Deterioration estimation device and deterioration estimation method | |
| CN110161414A (en) | A kind of power battery thermal runaway on-line prediction method and system | |
| CN108254696A (en) | The health state evaluation method and system of battery | |
| CN110712528B (en) | Real-time monitoring method and device for power battery pack | |
| CN116401585B (en) | Energy storage battery failure risk assessment method based on big data | |
| CN109856545B (en) | The battery group residual capacity detection method and system of solar telephone | |
| Chang et al. | Micro-fault diagnosis of electric vehicle batteries based on the evolution of battery consistency relative position | |
| CN115792637A (en) | Lithium battery health state estimation method | |
| CN103413033A (en) | Method for predicting storage battery faults | |
| CN115236523A (en) | Power battery fault diagnosis and prediction method based on digital twinning | |
| CN113447817B (en) | Lithium battery online life prediction method based on two-point life characteristics | |
| CN117054892B (en) | Evaluation method, device and management method for battery state of energy storage power station | |
| CN111525197B (en) | Storage battery SOH real-time estimation system and method | |
| CN113484784A (en) | Lithium battery online aging diagnosis method based on two-point impedance aging characteristics | |
| CN115825756B (en) | Multi-stage fault diagnosis method and system for distributed energy storage power station | |
| CN102095953A (en) | On-line detection method for performance of accumulator charger | |
| CN115656837A (en) | Fault prediction method for series-connected battery | |
| CN113466700B (en) | Lithium battery online life prediction method based on two-point impedance life characteristics | |
| CN207586393U (en) | A kind of Monitored System of Industrial Storage Cell | |
| CN115825755B (en) | Method for evaluating consistency of voltages of battery cells of energy storage battery | |
| CN116505105B (en) | Storage battery on-line management method based on parallel module | |
| Huang et al. | Evaluation index of battery pack of energy storage station based on RB recession mechanism |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20220922 Address after: 226299 No. 500, Linyang Road, Qidong Economic Development Zone, Qidong City, Nantong City, Jiangsu Province Applicant after: Qidong wotai new energy Co.,Ltd. Address before: 211100 West unit, 1st floor, building D, Jinzhi science and Technology Park, 100 Jiangjun Avenue, Jiangning District, Nanjing City, Jiangsu Province Applicant before: JIANGSU HUIZHI ENERGY ENGINEERING TECHNOLOGY INNOVATION RESEARCH INSTITUTE Co.,Ltd. |