CN112946497A - Storage battery fault diagnosis method and device based on fault injection deep learning - Google Patents
Storage battery fault diagnosis method and device based on fault injection deep learning Download PDFInfo
- Publication number
- CN112946497A CN112946497A CN202011410618.7A CN202011410618A CN112946497A CN 112946497 A CN112946497 A CN 112946497A CN 202011410618 A CN202011410618 A CN 202011410618A CN 112946497 A CN112946497 A CN 112946497A
- Authority
- CN
- China
- Prior art keywords
- storage battery
- fault
- deep learning
- battery
- storage
- 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
- 238000003745 diagnosis Methods 0.000 title claims abstract description 86
- 238000002347 injection Methods 0.000 title claims abstract description 86
- 239000007924 injection Substances 0.000 title claims abstract description 86
- 238000000034 method Methods 0.000 title claims abstract description 78
- 238000013135 deep learning Methods 0.000 title claims abstract description 40
- 238000013136 deep learning model Methods 0.000 claims abstract description 60
- 238000007599 discharging Methods 0.000 claims abstract description 45
- 230000008569 process Effects 0.000 claims abstract description 44
- 238000012549 training Methods 0.000 claims abstract description 33
- 238000001514 detection method Methods 0.000 claims description 34
- 238000002474 experimental method Methods 0.000 claims description 26
- 230000009977 dual effect Effects 0.000 claims description 4
- 230000000087 stabilizing effect Effects 0.000 claims description 4
- 208000032953 Device battery issue Diseases 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 15
- 238000012423 maintenance Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000003750 conditioning effect Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000001052 transient 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]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D21/00—Measuring or testing not otherwise provided for
- G01D21/02—Measuring two or more variables by means not covered by a single other subclass
-
- 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
- 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]
- G01R31/385—Arrangements for measuring battery or accumulator variables
- G01R31/387—Determining ampere-hour charge capacity or SoC
- G01R31/388—Determining ampere-hour charge capacity or SoC involving voltage measurements
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
The invention discloses a storage battery fault diagnosis method and device based on fault injection deep learning, and the storage battery fault diagnosis method based on fault injection deep learning comprises the following steps: firstly, each storage battery in a storage battery pack to be detected is respectively connected with a storage battery fault diagnosis device, then, the storage battery fault diagnosis device is used for detecting battery performance parameters of each storage battery in the storage battery pack to be detected in the charging and discharging process, finally, the battery performance parameters of each storage battery in the charging and discharging process are input into a fault injection-based deep learning model, whether each storage battery has a fault or not and the fault type of the storage battery with the fault are determined, and the fault injection-based deep learning model is generated by deep learning of a training storage battery pack. The storage battery fault diagnosis method and device based on fault injection deep learning disclosed by the embodiment of the invention can improve the diagnosis efficiency of the storage battery of the power distribution network.
Description
Technical Field
The embodiment of the invention relates to an electric power technology, in particular to a storage battery fault diagnosis method and device based on fault injection deep learning.
Background
Because the storage battery is small in size, light in weight, high in discharge performance, safe, reliable and small in maintenance amount, the storage battery is usually configured in a distribution network distribution area to serve as a backup power supply, and the storage battery is generally applied to a communication power supply of the distribution network distribution area at present. Theoretically, the storage battery has higher reliability and longer service life, however, due to the lack of effective online diagnosis means, many storage battery packs in actual use far fail to reach the rated service life, and the problem of insufficient power supply capacity often occurs. In fact, after the storage battery is used for 2 to 3 years, most of the storage batteries are difficult to pass capacity detection, and even a part of single batteries are failed after being used for one or two years.
The battery cells of the distribution point network storage battery pack are connected in series to form the distribution point network storage battery pack, and the performance of the whole group of storage batteries is sharply reduced due to the abnormality of any storage battery. Especially, when the single body is opened, the whole storage battery pack is invalid, and the storage battery cannot provide a backup power supply for the relay protection equipment, so that the fault range is enlarged.
Aiming at the problem of open circuit caused by insufficient capacity of a single storage battery in a storage battery pack, the current operation and maintenance method is to test whether the battery meets the operation and maintenance requirements by adopting a charge-discharge and capacity-checking mode. However, the operation and maintenance personnel are required to continuously work on the site for a long time every time of checking the capacity, and the test can be carried out only after the extra battery pack is carried to replace the battery pack to be detected, so that the battery detection efficiency is extremely low.
Disclosure of Invention
The invention provides a storage battery fault diagnosis method and device based on fault injection deep learning, which can improve the detection efficiency of a distribution point network storage battery pack.
In a first aspect, an embodiment of the present invention provides a storage battery fault diagnosis method based on fault injection deep learning, including:
connecting each storage battery in the storage battery pack to be detected with a storage battery fault diagnosis device respectively;
detecting battery performance parameters of each storage battery in the storage battery pack to be detected in the charging and discharging process by using a storage battery fault diagnosis device;
the method comprises the steps of inputting battery performance parameters of each storage battery in the charging and discharging process into a fault injection-based deep learning model, determining whether each storage battery has a fault and the fault type of the storage battery with the fault, and generating the fault injection-based deep learning model after deep learning a training storage battery pack which comprises a plurality of storage batteries with different faults and a normal storage battery.
In a possible implementation manner of the first aspect, before each storage battery in the storage battery pack to be detected is connected to the storage battery fault diagnosis device, the method further includes:
respectively connecting a plurality of storage batteries with different faults and a normal storage battery in a training storage battery pack with a storage battery fault diagnosis device, wherein the plurality of storage batteries with different faults are connected with the normal storage battery in series;
carrying out a charge and discharge experiment on a plurality of storage batteries with different faults and a normal storage battery through a storage battery fault diagnosis device, and acquiring battery performance parameters of the plurality of storage batteries with different faults and the normal storage battery in the charge and discharge experiment process;
and constructing a deep learning model based on fault injection according to battery performance parameters of a plurality of storage batteries with different faults and a normal storage battery in the charging and discharging experiment process.
In a possible implementation manner of the first aspect, constructing a deep learning model based on fault injection according to battery performance parameters of a plurality of storage batteries with different faults and a normal storage battery in a charging and discharging experiment process includes:
based on a SENET improved ESPCN algorithm, a fault injection-based deep learning model is constructed according to battery performance parameters of a plurality of storage batteries with different faults and a normal storage battery in the process of a charge and discharge experiment.
In a possible implementation manner of the first aspect, inputting the battery performance parameters of each storage battery in the charging and discharging process into a deep learning model based on fault injection, and determining whether each storage battery has a fault and a fault type of the storage battery with the fault, where the method includes:
and inputting the battery performance parameters of each storage battery in the charging and discharging process into a fault injection-based deep learning model, and determining whether each storage battery has a fault and the fault type of the storage battery with the fault, wherein an SE module in the fault injection-based deep learning model is embedded into different networks for result comparison.
In a second aspect, an embodiment of the present invention provides a battery fault diagnosis apparatus based on fault injection deep learning, including: the system comprises a plurality of storage battery connecting ports, a parameter detection module and a fault diagnosis module;
the storage battery connection ports are respectively connected with each storage battery in the storage battery pack to be detected;
the parameter detection module is used for detecting battery performance parameters of each storage battery connected with the plurality of storage battery connection ports in the charging and discharging processes;
the fault diagnosis module is used for inputting battery performance parameters of each storage battery in the charging and discharging process into a fault injection-based deep learning model, determining whether each storage battery has a fault and the fault type of the storage battery with the fault, and the fault injection-based deep learning model is generated by deep learning a training storage battery pack which comprises a plurality of storage batteries with different faults and a normal storage battery.
In one possible implementation manner of the second aspect, the battery fault diagnosis device based on fault injection deep learning further includes: a charge and discharge control module;
the charging and discharging control module is used for performing charging and discharging experiments on a plurality of storage batteries with different faults and a normal storage battery when a plurality of storage battery connection ports are respectively connected with the plurality of storage batteries with different faults and the normal storage battery in the training storage battery pack, wherein the plurality of storage batteries with different faults are connected with the normal storage battery in series;
the parameter detection module is also used for acquiring battery performance parameters of a plurality of storage batteries with different faults and a normal storage battery in the charge and discharge experiment process;
the fault diagnosis module is also used for constructing a deep learning model based on fault injection according to battery performance parameters of a plurality of storage batteries with different faults and a normal storage battery in the process of a charging and discharging experiment.
In a possible implementation manner of the second aspect, the fault diagnosis module is specifically configured to construct a deep learning model based on fault injection according to battery performance parameters of a plurality of storage batteries with different faults and a normal storage battery in a charging and discharging experiment process based on a SENET improved ESPCN algorithm.
In a possible implementation manner of the second aspect, the fault diagnosis module is specifically configured to input the battery performance parameters of each storage battery in the charging and discharging processes into a fault injection-based deep learning model, and determine whether each storage battery has a fault and a fault type of the storage battery with the fault, where SE modules in the fault injection-based deep learning model are embedded into different networks for result comparison.
In a possible implementation manner of the second aspect, the plurality of storage batteries with different faults include a storage battery with a small battery capacity and a preset capacity threshold, a storage battery with a large battery internal resistance and a preset internal resistance threshold, and a storage battery with a battery SCO smaller than a preset SOC;
the battery performance parameters include current, voltage, temperature, and charge and discharge capacity.
In a possible implementation manner of the second aspect, the parameter detection module includes a voltage stabilizing circuit, a switching power supply, a dual power supply operational amplifier, a current acquisition circuit, and a signal processor.
The embodiment of the invention provides a storage battery fault diagnosis method and a storage battery fault diagnosis device based on fault injection deep learning, which are characterized in that each storage battery in a storage battery pack to be detected is connected with the storage battery fault diagnosis device, then the storage battery fault diagnosis device is used for detecting battery performance parameters of each storage battery in the storage battery pack to be detected in the charging and discharging process, finally the battery performance parameters of each storage battery in the charging and discharging process are input into a deep learning model based on fault injection to determine whether each storage battery has faults and the fault type of the storage battery with the faults, the deep learning model based on fault injection is generated by deep learning a training storage battery pack, the training storage battery pack comprises a plurality of storage batteries with different faults and a normal storage battery, and the faults of the storage batteries in the storage battery pack are detected by adopting the deep learning method, the workload of diagnosing the storage battery is reduced, the diagnosis efficiency of the storage battery of the power distribution network is improved, and the safety of a power grid is ensured.
Drawings
Fig. 1 is a flowchart of a method for diagnosing a fault of a storage battery based on deep learning of fault injection according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of the structure of the SE module;
fig. 3 is a schematic structural diagram of a storage battery fault diagnosis device based on fault injection deep learning according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a voltage regulator circuit in the parameter detection module;
FIG. 5 is a schematic diagram of a switching power supply in the parameter detection module;
FIG. 6 is a schematic diagram of a dual power operational amplifier in the parameter detection module;
FIG. 7 is a schematic diagram of a current collection circuit in the parameter detection module;
fig. 8 is a schematic structural diagram of a signal processor in the parameter detection module.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Fig. 1 is a flowchart of a battery fault diagnosis method based on deep fault injection learning according to an embodiment of the present invention, and as shown in fig. 1, the battery fault diagnosis method based on deep fault injection learning according to the embodiment includes:
and S101, respectively connecting each storage battery in the storage battery pack to be detected with a storage battery fault diagnosis device.
The storage battery fault diagnosis method based on fault injection deep learning is used for fault diagnosis of a power distribution network storage battery pack. The traditional power distribution network storage battery group diagnosis method needs to take out a storage battery to be detected from a power distribution network and then carry out complete charge and discharge tests on the storage battery to be detected, so that whether a storage battery with a fault exists in the storage battery group is determined. However, the traditional power distribution network storage battery pack diagnosis method is low in efficiency, needs to judge faults manually, is not timely in response to power grid faults, and can expand the power grid fault range and prolong fault events. Meanwhile, due to the short plate effect and the difference characteristic of the storage battery and the like, the hidden troubles of insufficient electric quantity of the storage battery pack exist.
According to the storage battery fault diagnosis method based on fault injection deep learning, the storage battery fault is diagnosed by adopting a deep learning method, so that the efficiency of storage battery fault diagnosis is improved. According to the storage battery fault diagnosis method based on fault injection deep learning, the storage battery fault diagnosis device is used for carrying out fault diagnosis on the storage battery pack of the power distribution network. The storage battery fault diagnosis device comprises a plurality of storage battery connection ports, and each storage battery in the storage battery pack to be detected is connected with the plurality of storage battery connection ports of the storage battery fault diagnosis device respectively. Because the storage battery pack to be detected comprises a plurality of storage batteries connected in series, the plurality of storage batteries in the storage battery pack to be detected are respectively connected with the storage battery connecting ports of the storage battery fault diagnosis device, and fault detection can be performed on the plurality of storage batteries in the storage battery pack to be detected at one time.
And S102, detecting the battery performance parameters of each storage battery in the storage battery pack to be detected in the charging and discharging process by using a storage battery fault diagnosis device.
After each storage battery in the storage battery pack to be detected is connected with the storage battery fault diagnosis device, the storage battery fault diagnosis device can be used for detecting the battery performance parameters of each storage battery. The storage battery fault diagnosis device is provided with one or more detection circuits required for detecting battery performance parameters of the storage battery. Because one or more battery performance parameters can change when the storage battery has a fault, whether the storage battery has the fault can be determined through detection and analysis of the one or more battery performance parameters.
The storage battery fault diagnosis device can be connected with each storage battery in the storage battery pack to be detected all the time, and detects the battery performance parameters of the storage batteries in the normal working state of the storage batteries; or after the storage battery fault diagnosis device is connected with each storage battery in the storage battery pack to be detected, each storage battery in the storage battery pack is subjected to charge-discharge control, so that each storage battery is subjected to complete charge and discharge processes to obtain battery performance parameters. The battery performance parameters include one or more of current, voltage, temperature and charge/discharge capacity, or the battery performance parameters may further include other parameters related to the battery performance.
Step S103, inputting the battery performance parameters of each storage battery in the charging and discharging process into a fault injection-based deep learning model, determining whether each storage battery has a fault and the fault type of the storage battery with the fault, wherein the fault injection-based deep learning model is generated by deep learning a training storage battery pack, and the training storage battery pack comprises a plurality of storage batteries with different faults and a normal storage battery.
After battery performance parameters of each storage battery in the storage battery pack to be detected in the charging and discharging process are obtained, the battery performance parameters can be input into a deep learning model based on fault injection, and the classification result of each storage battery is determined through the classification of the deep learning model, so that whether each storage battery in the storage battery pack to be detected has a fault or not and the fault type of the storage battery with the fault are determined. The deep learning model based on fault injection is generated by deep learning a training storage battery pack, wherein the training storage battery pack comprises a plurality of storage batteries with different faults and a normal storage battery.
That is, the storage battery fault diagnosis device stores a deep learning model based on fault injection in advance, or the storage battery fault diagnosis device generates the deep learning model based on fault injection through detection learning of a training storage battery pack before a storage battery pack to be detected is detected. The deep learning method is a method for classifying data by training a classification model through training sample data. The training battery pack includes a plurality of batteries with different known faults and a normal battery. The models of all storage batteries in the training storage battery pack are the same, and the models of all storage batteries in the training storage battery pack are the same as those of all storage batteries in the storage battery pack to be detected. The fault types of storage batteries with different faults in the training storage battery pack, and battery performance parameters of the storage batteries with different faults and normal storage batteries in the charging and discharging processes are training data for training the deep learning model based on fault injection.
After the deep learning model based on fault injection is trained, the classification result of each storage battery, namely whether each storage battery has a fault or not and the fault type of the storage battery with the fault, can be obtained after the battery performance parameters of each storage battery detected by the storage battery to be detected are input.
In this embodiment, the battery failure includes a condition that the battery capacity is small and a preset capacity threshold, the battery internal resistance is large and a preset internal resistance threshold, and the State of Charge (SCO) of the battery is smaller than a preset SOC. Or the failure of the battery may also include other failures that can be manifested by battery performance parameters.
It should be noted that, the storage battery fault diagnosis device may perform the training of the deep learning model based on fault injection before performing fault diagnosis on the storage battery pack to be detected, and specifically includes: respectively connecting a plurality of storage batteries with different faults and a normal storage battery in a training storage battery pack with a storage battery fault diagnosis device, wherein the plurality of storage batteries with different faults are connected with the normal storage battery in series; carrying out a charge and discharge experiment on a plurality of storage batteries with different faults and a normal storage battery through a storage battery fault diagnosis device, and acquiring battery performance parameters of the plurality of storage batteries with different faults and the normal storage battery in the charge and discharge experiment process; and constructing a deep learning model based on fault injection according to battery performance parameters of a plurality of storage batteries with different faults and a normal storage battery in the charging and discharging experiment process.
After battery performance parameters of a plurality of storage batteries with different faults and a normal storage battery in the charging and discharging experiment process are obtained, the obtained battery performance parameters can be classified and sorted, data are normalized, and the data are converted into a deep learning universal data format. Thus, the preparation work before the deep learning model is trained is completed, and then the deep learning model based on fault injection can be trained.
In the embodiment, a deep learning model based on fault injection can be trained by adopting a SENET-based improved ESPCN algorithm as an example because the difference of data between the storage batteries with faults is small. The ESPCN algorithm based on SENET improvement takes the ESPCN algorithm as the basis of a network structure, and respectively optimizes and improves the generation network, the judgment network and the loss function of the model.
According to the advantages of small calculation amount of the SEET algorithm, target detection and semantic segmentation efficiency and the like, after the ESPCN performs nonlinear processing on convolution calculation each time, the SEET structure is inserted in a module form, namely an SE module. Fig. 2 is a schematic structural diagram of an SE module, and as shown in fig. 2, the SE module first performs a Squeeze operation on a feature map obtained by convolution to obtain channel-level global features, then performs an Excitation operation on the global features, learns the relationship among channels to obtain weights of different channels, and finally multiplies the weights by an original feature information map to obtain final feature information. The channel characteristics with the largest information quantity are obtained, and the unimportant channel characteristics are suppressed.
On the basis of a deep learning model based on fault injection by a SENET improved ESPCN algorithm, the method for determining whether each storage battery has a fault and the fault type of the storage battery with the fault comprises the following steps: and inputting the battery performance parameters of each storage battery in the charging and discharging process into a fault injection-based deep learning model, and determining whether each storage battery has a fault and the fault type of the storage battery with the fault, wherein an SE module in the fault injection-based deep learning model is embedded into different networks for result comparison. The deep learning model based on fault injection trained by the SEET improved ESPCN algorithm has an improved SE module, so that the SE module can be embedded into different networks to compare results to obtain a classification result. On the other hand, the performance gain and the network depth of the deep learning model based on fault injection trained by the SENET-based improved ESPCN algorithm are superior to those of other deep learning networks, and the fault diagnosis accuracy and efficiency of the deep learning model based on fault injection are improved.
The storage battery fault diagnosis method based on fault injection deep learning provided by the embodiment comprises the steps of firstly connecting each storage battery in a storage battery pack to be detected with a storage battery fault diagnosis device, then detecting battery performance parameters of each storage battery in the storage battery pack to be detected in the charging and discharging processes by using the storage battery fault diagnosis device, finally inputting the battery performance parameters of each storage battery in the charging and discharging processes into a deep learning model based on fault injection to determine whether each storage battery has a fault and the fault type of the storage battery with the fault, generating the deep learning model based on fault injection by deep learning a training storage battery pack, wherein the training storage battery pack comprises a plurality of storage batteries with different faults and a normal storage battery, and the fault of the storage battery in the storage battery pack is detected by adopting the deep learning method, the workload of diagnosing the storage battery is reduced, the diagnosis efficiency of the storage battery of the power distribution network is improved, and the safety of a power grid is ensured.
Fig. 3 is a schematic structural diagram of a battery fault diagnosis device based on deep fault injection learning according to an embodiment of the present invention, and as shown in fig. 3, the battery fault diagnosis device based on deep fault injection learning according to the embodiment of the present invention includes:
a plurality of battery connection ports 31, a parameter detection module 32, and a failure diagnosis module 33.
The plurality of storage battery connection ports 31 are respectively connected with each storage battery 35 in the storage battery pack 34 to be detected; the parameter detection module 32 is configured to detect battery performance parameters of each storage battery 35 connected to the plurality of storage battery connection ports 31 in a charging and discharging process; the fault diagnosis module 33 is configured to input the battery performance parameters of each storage battery 35 in the charging and discharging processes into a fault injection-based deep learning model, determine whether each storage battery 35 has a fault and a fault type of the storage battery with the fault, and generate the fault injection-based deep learning model by performing deep learning on a training storage battery pack, where the training storage battery pack includes a plurality of storage batteries with different faults and a normal storage battery.
The storage battery fault diagnosis device based on fault injection deep learning provided by this embodiment is used for implementing the technical scheme of the storage battery fault diagnosis method based on fault injection deep learning shown in fig. 1, and the implementation principle and the technical effect are similar, and are not described herein again.
Further, in the embodiment shown in fig. 3, the battery failure diagnosis apparatus based on the deep learning of fault injection further includes: and a charge and discharge control module. The charging and discharging control module is used for performing charging and discharging experiments on a plurality of storage batteries with different faults and a normal storage battery when a plurality of storage battery connection ports 31 are respectively connected with the plurality of storage batteries with different faults and the normal storage battery in the training storage battery pack, wherein the plurality of storage batteries with different faults are connected with the normal storage battery in series; the parameter detection module 32 is further configured to obtain battery performance parameters of a plurality of storage batteries with different faults and a normal storage battery in a charging and discharging experiment process; the fault diagnosis module 33 is further configured to construct a deep learning model based on fault injection according to battery performance parameters of a plurality of storage batteries with different faults and a normal storage battery in a charging and discharging experiment process.
Further, on the basis of the embodiment shown in fig. 3, the fault diagnosis module 33 is specifically configured to construct a deep learning model based on fault injection according to battery performance parameters of a plurality of storage batteries with different faults and a normal storage battery during a charging and discharging experiment based on the SENET improved ESPCN algorithm.
Further, on the basis of the embodiment shown in fig. 3, the fault diagnosis module 33 is specifically configured to input the battery performance parameters of each storage battery during charging and discharging into a fault injection-based deep learning model, and determine whether a fault exists in each storage battery and a fault type of the storage battery with the fault, where SE modules in the fault injection-based deep learning model are embedded into different networks for result comparison.
Further, on the basis of the embodiment shown in fig. 3, the plurality of storage batteries with different faults include a storage battery with a low battery capacity and a preset capacity threshold, a storage battery with a high battery internal resistance and a preset internal resistance threshold, and a storage battery with a battery state of charge SCO smaller than a preset SOC; the battery performance parameters include current, voltage, temperature, and charge and discharge capacity.
Further, on the basis of the embodiment shown in fig. 3, the parameter detection module 33 includes a voltage stabilizing circuit, a switching power supply, a dual power supply operational amplifier, a current acquisition circuit, and a signal processor.
The following describes a parameter detection module in the battery fault diagnosis apparatus based on fault injection deep learning according to an embodiment of the present invention with a specific circuit structure. Fig. 4 to fig. 8 are schematic circuit diagrams of a parameter detection module in a battery fault diagnosis apparatus based on deep learning of fault injection according to an embodiment of the present invention, and fig. 4 to fig. 8 are only specific circuit structures of a parameter detection module. Fig. 4 is a schematic structural diagram of a voltage stabilizing circuit in the parameter detection module, fig. 5 is a schematic structural diagram of a switching power supply in the parameter detection module, fig. 6 is a schematic structural diagram of a dual-power operational amplifier in the parameter detection module, fig. 7 is a schematic structural diagram of a current acquisition circuit in the parameter detection module, and fig. 8 is a schematic structural diagram of a signal processor in the parameter detection module.
As shown in fig. 4, the voltage regulator circuit is mainly composed of basic parts such as a reference voltage source, a sampling circuit, a loop compensator, a power amplifier tube, etc., and the series power amplifier tube is an amplifier tube composed of NPN-type triodes VT2 and VT 3. The VT1 is a driving tube, and it uses PNP type transistor. U1 is the input voltage and U0 is the output voltage. R1 and R2 are sampling impedances, and the sampling voltage UQ is applied to the non-inverting input terminal of the error amplifier, which amplifies the difference between them to generate the error voltage Ur, compared to the reference voltage Uref applied to the inverting input terminal. It is used to regulate the voltage drop of the series regulator to stabilize the output voltage.
As shown in fig. 5, the switching power supply includes an LM5576 and a peripheral circuit, and implements the switching power supply function.
As shown in fig. 6, in the dual-power operational amplifier, when the positive input terminal and the negative input terminal are short-circuited, the output voltage is lower than 25 μ V, the parameter detection module does not need to perform extra zero setting, and the dual-power operational amplifier has extremely low input bias current and extremely high open loop gain. The dual-power operational amplifier is provided with two stages of operational amplifiers to form signal conditioning, wherein the first stage is used for converting a current signal output by a mutual inductor into a voltage signal, the second stage is used for converting a bipolar signal into a unipolar signal which can be converted normally by an A/D (analog/digital) converter, in order to prevent overcurrent or static electricity from damaging monitoring equipment, a transient suppression diode is arranged at the output end of a first-stage operation developed device, and the operating voltage of the operational amplifier is positive or negative 5V.
As shown in fig. 7, a reference resistor Rref in the current collection circuit. When the resistor R55 is R54 and the resistor R45 is R48, Va is 1.5V + Vb, and thus the output current of the constant current source is 2 mA. The constant current source circuit load can be grounded and can not be short-circuited due to no load. And the change of the constant current source can be realized by changing the input reference Vref or adjusting the magnitude of the reference resistance Rref. The voltage across the reference resistor R46 will be affected by its driving load voltage Vb due to the resistance accuracy. Therefore, the four resistors R55, R54, R45 and R48 are selected according to the principle that the mismatch is as small as possible, and the mismatch directions of each pair of resistors are consistent, so that the stability of the constant current source is improved.
As shown in fig. 8, the signal processor has a high-speed processing capability of 150MHz, and has a 32-bit floating point processing unit, 6 Direct Memory Access (DMA) channels supporting Analog-to-Digital Converter (ADC), multi-channel buffer serial Interface (McBSP), and External Memory Interface (EMIF), and at most 18 channels of PWM output, 6 of which are unique high-precision PWM output (HRPWM) and 12-bit 16-channel ADC.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments illustrated herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (10)
1. A storage battery fault diagnosis method based on fault injection deep learning is characterized by comprising the following steps:
connecting each storage battery in the storage battery pack to be detected with a storage battery fault diagnosis device respectively;
detecting battery performance parameters of each storage battery in the storage battery pack to be detected in the charging and discharging process by using a storage battery fault diagnosis device;
and inputting the battery performance parameters of each storage battery in the charging and discharging process into a fault injection-based deep learning model, and determining whether each storage battery has a fault and the fault type of the storage battery with the fault, wherein the fault injection-based deep learning model is generated by deep learning a training storage battery pack, and the training storage battery pack comprises a plurality of storage batteries with different faults and a normal storage battery.
2. The method according to claim 1, wherein before connecting each battery in the battery pack to be tested to a battery failure diagnosis device, the method further comprises:
connecting a plurality of storage batteries with different faults and a normal storage battery in the training storage battery pack with a storage battery fault diagnosis device respectively, wherein the plurality of storage batteries with different faults are connected with the normal storage battery in series;
performing a charge and discharge experiment on the plurality of storage batteries with different faults and a normal storage battery through the storage battery fault diagnosis device to obtain battery performance parameters of the plurality of storage batteries with different faults and the normal storage battery in the charge and discharge experiment process;
and constructing the deep learning model based on fault injection according to the battery performance parameters of the plurality of storage batteries with different faults and a normal storage battery in the charging and discharging experiment process.
3. The method according to claim 2, wherein the constructing the fault injection-based deep learning model according to the battery performance parameters of the plurality of storage batteries with different faults and a normal storage battery during a charge and discharge experiment comprises:
and constructing the fault injection-based deep learning model according to the battery performance parameters of the plurality of storage batteries with different faults and a normal storage battery in the charge and discharge experiment process based on the SENET improved ESPCN algorithm.
4. The method of claim 3, wherein the step of inputting the battery performance parameters of each storage battery in the charging and discharging process into a fault injection-based deep learning model to determine whether each storage battery has a fault and the fault type of the fault storage battery comprises the steps of:
and inputting the battery performance parameters of each storage battery in the charging and discharging processes into a fault injection-based deep learning model, and determining whether each storage battery has a fault and the fault type of the storage battery with the fault, wherein an SE module in the fault injection-based deep learning model is embedded into different networks for result comparison.
5. A storage battery fault diagnosis device based on fault injection deep learning is characterized by comprising the following components: the system comprises a plurality of storage battery connecting ports, a parameter detection module and a fault diagnosis module;
the storage battery connection ports are respectively connected with each storage battery in the storage battery pack to be detected;
the parameter detection module is used for detecting battery performance parameters of each storage battery connected with the plurality of storage battery connection ports in the charging and discharging processes;
the fault diagnosis module is used for inputting the battery performance parameters of each storage battery in the charging and discharging process into a fault injection-based deep learning model, and determining whether each storage battery has a fault and the fault type of the storage battery with the fault, wherein the fault injection-based deep learning model is generated by deep learning a training storage battery pack, and the training storage battery pack comprises a plurality of storage batteries with different faults and a normal storage battery.
6. The apparatus of claim 5, further comprising: a charge and discharge control module;
the charging and discharging control module is used for performing charging and discharging experiments on the plurality of storage batteries with different faults and a normal storage battery when the plurality of storage battery connection ports are respectively connected with the plurality of storage batteries with different faults and the normal storage battery in the training storage battery pack, wherein the plurality of storage batteries with different faults are connected with the normal storage battery in series;
the parameter detection module is also used for acquiring battery performance parameters of the plurality of storage batteries with different faults and a normal storage battery in the process of a charge and discharge experiment;
the fault diagnosis module is also used for constructing the deep learning model based on fault injection according to the battery performance parameters of the plurality of storage batteries with different faults and a normal storage battery in the charging and discharging experiment process.
7. The apparatus according to claim 6, wherein the fault diagnosis module is specifically configured to construct the fault injection-based deep learning model according to the battery performance parameters of the plurality of batteries with different faults and a normal battery during the charge and discharge experiment based on a SENET improved ESPCN algorithm.
8. The apparatus according to claim 7, wherein the fault diagnosis module is specifically configured to input the battery performance parameters of each storage battery during charging and discharging into a fault injection-based deep learning model, and determine whether a fault exists in each storage battery and a fault type of the fault storage battery, wherein SE modules in the fault injection-based deep learning model are embedded into different networks for comparison.
9. The device according to any one of claims 6 to 8, wherein the plurality of storage batteries with different faults comprises a storage battery with a low battery capacity and a preset capacity threshold, a storage battery with a high battery internal resistance and a preset internal resistance threshold, and a storage battery with a battery state of charge SCO smaller than a preset SOC;
the battery performance parameters include current, voltage, temperature, and charge and discharge capacity.
10. The device according to any one of claims 5 to 8, wherein the parameter detection module comprises a voltage stabilizing circuit, a switching power supply, a dual power supply operational amplifier, a current acquisition circuit and a signal processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011410618.7A CN112946497A (en) | 2020-12-04 | 2020-12-04 | Storage battery fault diagnosis method and device based on fault injection deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011410618.7A CN112946497A (en) | 2020-12-04 | 2020-12-04 | Storage battery fault diagnosis method and device based on fault injection deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112946497A true CN112946497A (en) | 2021-06-11 |
Family
ID=76234732
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011410618.7A Pending CN112946497A (en) | 2020-12-04 | 2020-12-04 | Storage battery fault diagnosis method and device based on fault injection deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112946497A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113608140A (en) * | 2021-06-25 | 2021-11-05 | 国网山东省电力公司泗水县供电公司 | Battery fault diagnosis method and system |
| CN113625692A (en) * | 2021-08-23 | 2021-11-09 | 公安部交通管理科学研究所 | Electric automobile battery security inspection system based on fault injection |
| CN115902646A (en) * | 2023-01-06 | 2023-04-04 | 中国电力科学研究院有限公司 | Energy storage battery fault identification method and system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109949224A (en) * | 2019-02-26 | 2019-06-28 | 北京悦图遥感科技发展有限公司 | A kind of method and device of the connection grade super-resolution rebuilding based on deep learning |
| CN110018425A (en) * | 2019-04-10 | 2019-07-16 | 北京理工大学 | A kind of power battery fault diagnosis method and system |
| CN110190306A (en) * | 2019-06-04 | 2019-08-30 | 昆山知氢信息科技有限公司 | A kind of on-line fault diagnosis method for fuel cell system |
| CN110426637A (en) * | 2019-07-04 | 2019-11-08 | 佛山科学技术学院 | A kind of battery failures diagnostic method neural network based and device |
| CN111007401A (en) * | 2019-12-16 | 2020-04-14 | 国网江苏省电力有限公司电力科学研究院 | Electric vehicle battery fault diagnosis method and device based on artificial intelligence |
| CN111308355A (en) * | 2020-03-14 | 2020-06-19 | 智洋创新科技股份有限公司 | Transformer substation storage battery state detection and analysis method based on deep learning |
| CN111340141A (en) * | 2020-04-20 | 2020-06-26 | 天津职业技术师范大学(中国职业培训指导教师进修中心) | Crop seedling and weed detection method and system based on deep learning |
-
2020
- 2020-12-04 CN CN202011410618.7A patent/CN112946497A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109949224A (en) * | 2019-02-26 | 2019-06-28 | 北京悦图遥感科技发展有限公司 | A kind of method and device of the connection grade super-resolution rebuilding based on deep learning |
| CN110018425A (en) * | 2019-04-10 | 2019-07-16 | 北京理工大学 | A kind of power battery fault diagnosis method and system |
| CN110190306A (en) * | 2019-06-04 | 2019-08-30 | 昆山知氢信息科技有限公司 | A kind of on-line fault diagnosis method for fuel cell system |
| CN110426637A (en) * | 2019-07-04 | 2019-11-08 | 佛山科学技术学院 | A kind of battery failures diagnostic method neural network based and device |
| CN111007401A (en) * | 2019-12-16 | 2020-04-14 | 国网江苏省电力有限公司电力科学研究院 | Electric vehicle battery fault diagnosis method and device based on artificial intelligence |
| CN111308355A (en) * | 2020-03-14 | 2020-06-19 | 智洋创新科技股份有限公司 | Transformer substation storage battery state detection and analysis method based on deep learning |
| CN111340141A (en) * | 2020-04-20 | 2020-06-26 | 天津职业技术师范大学(中国职业培训指导教师进修中心) | Crop seedling and weed detection method and system based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| JIE HU,AT AL.: "Squeeze-and-Excitation Networks", 《2018 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》 * |
| 都磊: "基于深度学习的动力电池故障诊断研究", 《中国优秀硕士学位论文全文数据库 (基础科学辑)》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113608140A (en) * | 2021-06-25 | 2021-11-05 | 国网山东省电力公司泗水县供电公司 | Battery fault diagnosis method and system |
| CN113625692A (en) * | 2021-08-23 | 2021-11-09 | 公安部交通管理科学研究所 | Electric automobile battery security inspection system based on fault injection |
| CN115902646A (en) * | 2023-01-06 | 2023-04-04 | 中国电力科学研究院有限公司 | Energy storage battery fault identification method and system |
| CN115902646B (en) * | 2023-01-06 | 2023-06-13 | 中国电力科学研究院有限公司 | Energy storage battery fault identification method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112946497A (en) | Storage battery fault diagnosis method and device based on fault injection deep learning | |
| Yang et al. | On-board diagnosis of soft short circuit fault in lithium-ion battery packs for electric vehicles using an extended Kalman filter | |
| CN109726452B (en) | Online proton exchange membrane fuel cell fault diagnosis method based on impedance spectrum | |
| US10859636B2 (en) | Uninterruptible power supply (UPS) modules for testing systems | |
| CN110133536B (en) | System, method and apparatus for determining an indicator of leakage current within a battery entity | |
| CN101680933B (en) | Method of diagnosing defective elements in a standalone system, powered by an intermittent power source | |
| CN105071453A (en) | Battery management system | |
| CN102608537B (en) | Feed network type heavy-power battery testing device | |
| CN105356528A (en) | Battery management system | |
| CN103165203B (en) | Detection method for nuclear power station circuit board component | |
| CN109061495A (en) | A kind of hybrid energy-storing battery failure diagnostic method | |
| JP2003309934A (en) | Dc backup power source and diagnostic method thereof | |
| CN105759147A (en) | Integrated test device for battery management system of electric vehicle | |
| US20070252555A1 (en) | Method and apparatus for determining to which battery cell in a string of battery cells a test instrument is connected | |
| JPWO2016157721A1 (en) | Failure detection device | |
| CN207481815U (en) | Electric system above and below electric automobile high-voltage | |
| CN105785169A (en) | Battery system pre-charging management testing device and testing method thereof | |
| CN109596927A (en) | Test method and test macro for battery management system BMS | |
| CN219204173U (en) | Direct-current grounding auxiliary power supply device capable of achieving road pulling and searching | |
| Wu et al. | Research on short-circuit fault-diagnosis strategy of lithium-ion battery in an energy-storage system based on voltage cosine similarity | |
| CN115575778A (en) | Calibration system and calibration method of direct-current insulation monitoring device | |
| CN105629178A (en) | Storage battery pack open circuit monitoring device and storage battery pack open circuit monitoring method | |
| KR20230162253A (en) | Battery diagnosis method, battery diagnosis apparatus and battery system providing the same | |
| US20210126295A1 (en) | Systems and methods involving measurement of battery cells | |
| CN104682466A (en) | Calibration method and calibration circuit for charging control circuit |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210611 |
|
| RJ01 | Rejection of invention patent application after publication |