CN115837862A - Abnormal data detection method and device, electronic equipment and storage medium - Google Patents
Abnormal data detection method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN115837862A CN115837862A CN202111111850.5A CN202111111850A CN115837862A CN 115837862 A CN115837862 A CN 115837862A CN 202111111850 A CN202111111850 A CN 202111111850A CN 115837862 A CN115837862 A CN 115837862A
- Authority
- CN
- China
- Prior art keywords
- battery
- data
- abnormal
- dimensional data
- learning model
- 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
Images
Landscapes
- Secondary Cells (AREA)
Abstract
The embodiment of the invention relates to an abnormal data detection method, an abnormal data detection device, electronic equipment and a storage medium, wherein the method comprises the steps of acquiring two-dimensional data of a battery sent by a vehicle in real time; wherein the battery two-dimensional data comprises two battery parameter data with intrinsic correlation; inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model; and determining whether the two-dimensional data of the battery is abnormal or not according to the prediction result. Because the two-dimensional data of the battery comprises two battery parameter data with intrinsic correlation, the two-dimensional data of the battery can be closer to the real situation of the battery, and the detection accuracy is improved. Furthermore, unsupervised learning models are trained in advance, and therefore the computational power requirements for the hardware architecture are low.
Description
Technical Field
The embodiment of the invention relates to the technical field of automobiles, in particular to an abnormal data detection method and device, electronic equipment and a storage medium.
Background
With the rapid development of the electric automobile industry, the holding capacity of the electric automobile is higher and higher, and the safety of the battery system as an important part of the electric automobile is particularly important. For a battery system, voltage data and temperature data of a battery are used as indexes directly reflecting the running state of the battery, and the battery system is especially important for research and classification early warning of the battery system.
Currently, the detection of battery abnormal conditions is mainly based on the battery voltage, the battery temperature or the change rate of the battery state of charge, and then whether the battery is abnormal is determined by setting a fixed threshold, for example, by comparing the battery voltage or the battery temperature with the fixed threshold.
However, in the running process of the vehicle, the voltage and the temperature of the battery can fluctuate due to the working condition of the vehicle, the external temperature and the thermal management state of the vehicle. The fluctuation of the battery voltage and the battery temperature data in the actual driving process often has certain correlation. It is possible that a false negative of the battery abnormality occurs by determining whether the battery is abnormal only through a comparison of the battery voltage or the battery temperature with a fixed threshold. That is, the conventional individual detection of each parameter may fail to detect a part of abnormal cells, and the detection result is inaccurate. In addition, the conventional detection method needs to calculate the battery voltage and the battery temperature change rate in seconds, and has high requirements on the computing capability of the system.
Disclosure of Invention
In order to solve at least one problem existing in the prior art, at least one embodiment of the present disclosure provides an abnormal data detection method, an abnormal data detection apparatus, an electronic device, and a storage medium.
In a first aspect, an embodiment of the present invention provides an abnormal data detection method, including:
acquiring two-dimensional data of a battery sent by a vehicle in real time; wherein the battery two-dimensional data comprises two battery parameter data with intrinsic correlation;
inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model;
and determining whether the two-dimensional data of the battery is abnormal or not according to the prediction result.
Optionally, the unsupervised learning model comprises a single-class support vector machine model.
Optionally, the inputting the two-dimensional battery data into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model includes:
inputting the two-dimensional data of the battery into the single-classification support vector machine model, and acquiring the distance between the two-dimensional data of the battery and the center point of a hyper-sphere of the single-classification support vector machine model;
outputting a prediction result according to a comparison result of the distance between the two-dimensional battery data and the center point of the hypersphere of the single-classification support vector machine model and the radius of the hypersphere;
the determining whether the two-dimensional data of the battery is abnormal according to the prediction result comprises the following steps:
and when the prediction result is that the distance between the two-dimensional battery data and the center point of the hypersphere of the single-classification support vector machine model is larger than the radius of the hypersphere, determining the two-dimensional battery data as abnormal data.
Optionally, the battery two-dimensional data includes at least one of battery voltage and battery temperature two-dimensional data, battery current and battery temperature two-dimensional data, battery state of charge and battery temperature two-dimensional data, or battery current and battery state of charge two-dimensional data.
Optionally, before obtaining the pre-trained unsupervised learning model, the method further includes:
and training and generating the unsupervised learning model according to the two-dimensional data of the historical battery.
Optionally, the generating the unsupervised learning model according to the historical battery two-dimensional data training includes:
acquiring historical battery data, wherein the historical battery data comprises first parameter data of a historical battery and second parameter data of the historical battery;
performing data cleaning on the historical battery data;
combining historical battery first parameter data and historical battery second parameter data of the historical battery data after data cleaning to generate historical battery two-dimensional data;
dividing the historical battery two-dimensional data into a training set and a verification set;
training through the training set, validating through the validation set to iteratively optimize the unsupervised learning model.
Optionally, after training through the training set and verifying through the verification set to iteratively optimize the unsupervised learning model, the method further includes:
and acquiring two-dimensional data of a plurality of preset abnormal batteries, inputting the two-dimensional data into the unsupervised learning model, and determining the unsupervised learning model as an optimal unsupervised learning model if the output accuracy of the unsupervised learning model is greater than a preset value.
Optionally, if the result of performing abnormal data detection on the two-dimensional battery data by using the unsupervised learning model is abnormal, sending abnormal early warning information to a vehicle end.
In a second aspect, an embodiment of the present invention provides an abnormal data detection method, including:
monitoring two-dimensional data of the battery in real time; wherein the battery two-dimensional data comprises two battery parameter data with intrinsic correlation;
inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model;
and determining whether the two-dimensional data of the battery is abnormal or not according to the prediction result.
Optionally, if it is determined that the two-dimensional battery data is abnormal, the two-dimensional battery data is sent to a server, and/or abnormality warning information is generated.
In a third aspect, an embodiment of the present invention provides an abnormal data detection apparatus, including:
the first acquisition module is used for acquiring two-dimensional data of the battery sent by the vehicle in real time;
the first prediction module is used for inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model;
the first abnormity judgment module is used for determining whether the two-dimensional data of the battery is abnormal according to the prediction result;
wherein the battery two-dimensional data comprises two battery parameter data having an intrinsic correlation.
In a fourth aspect, an embodiment of the present invention provides an abnormal data detecting apparatus, including:
a second acquisition module; the device is used for monitoring the two-dimensional data of the battery in real time;
the second prediction module is used for inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model;
the second abnormity judgment module is used for determining whether the two-dimensional data of the battery is abnormal according to the prediction result;
wherein the battery two-dimensional data comprises two battery parameter data having an intrinsic correlation.
In a fifth aspect, an embodiment of the present invention provides an electronic device, including: one or more processors, and a storage device;
the storage device is configured to store one or more programs that, when executed by the one or more processors, cause the one or more processors to implement the abnormal data detecting method according to any of the embodiments of the first aspect, or the abnormal data detecting method according to any of the embodiments of the second aspect.
In a sixth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, the computer program, when being executed by a processor, is configured to implement the abnormal data detecting method according to any embodiment of the first aspect, or the abnormal data detecting method according to any embodiment of the second aspect.
Therefore, in at least one embodiment of the disclosure, the two-dimensional data of the battery sent by the vehicle end is obtained in real time, the two-dimensional data of the battery is input into the pre-trained unsupervised learning model, the prediction result output by the unsupervised learning model is obtained, and whether the two-dimensional data of the battery is abnormal or not is determined according to the prediction result. Wherein the battery two-dimensional data comprises two battery parameter data with intrinsic correlation. According to the method and the device, the unsupervised learning model is adopted to detect the abnormal data of the two-dimensional data of the battery, and the automatic real-time supervision of the battery in the electric automobile can be realized. Because the two-dimensional data of the battery comprises two battery parameter data with intrinsic correlation, the two-dimensional data of the battery can be closer to the real situation of the battery, and the detection accuracy is improved. Furthermore, unsupervised learning models have been trained in advance, and therefore the computational power requirements for the hardware architecture are low.
Drawings
To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed for describing the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present disclosure, and other drawings can be obtained by those skilled in the art according to the drawings.
Fig. 1 is a schematic flowchart of an abnormal data detection method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a training process of an unsupervised learning model according to an embodiment of the present invention;
FIG. 3 is a diagram of a historical battery two-dimensional data distribution;
FIG. 4 is a diagram of a One-Class-SVM machine learning unsupervised model screening data distribution;
FIG. 5 is a schematic flow chart illustrating another abnormal data detection method according to an embodiment of the present invention;
fig. 6 is a block diagram of an abnormal data detection apparatus according to an embodiment of the present invention;
fig. 7 is a block diagram of an abnormal data detection apparatus according to an embodiment of the present invention;
fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order that the above objects, features and advantages of the present disclosure can be more clearly understood, the present disclosure will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the embodiments described are only a few embodiments of the present disclosure, and not all embodiments. The specific embodiments described herein are merely illustrative of the disclosure and are not intended to be limiting. All other embodiments derived by one of ordinary skill in the art from the described embodiments of the disclosure are intended to be within the scope of the disclosure.
It is noted that, in this document, relational terms such as "first" and "second," and the like, may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions.
With the rapid development of the electric automobile industry, the holding capacity of the electric automobile is higher and higher, and the safety of the battery system as an important part of the electric automobile is particularly important. For a battery system, voltage data and temperature data of a battery are used as indexes directly reflecting the running state of the battery, and the battery system is especially important for research and classification early warning of the battery system. In the running process of a vehicle, the working condition of the vehicle, the external temperature and the heat management state all cause the fluctuation of the voltage and the temperature of the battery. The fluctuation of the battery voltage and the battery temperature data in the actual driving process often has certain correlation. However, the current detection of the abnormal condition of the battery mainly uses the change rate of the battery voltage, the battery temperature or the battery state of charge as a standard, and then determines whether the battery is abnormal or not by setting a fixed threshold, for example, by comparing the battery voltage or the battery temperature with the fixed threshold. The detection of disassembling each parameter may not detect part of abnormal batteries, the detection result is not accurate, and the existing detection mode has high requirement on the computing capability of hardware.
In view of the above-mentioned shortcomings of the prior art, embodiments of the present invention provide an abnormal data detection method. Before describing in detail the embodiments of the present disclosure, an application scenario of the present disclosure will be described below, and the present disclosure may be applied to an electric vehicle, which may be an autonomous vehicle or a non-autonomous vehicle. The scheme of the abnormal data detection method provided by the embodiment of the invention can be applied to real-time detection in the driving process or the charging process of the electric vehicle.
Fig. 1 is a schematic flow chart of an abnormal data detection method according to an embodiment of the present invention. The execution main body of the abnormal data detection method provided by the embodiment of the invention is a server, such as a cloud server. As shown in fig. 1, the abnormal data detecting method includes S110 to S130:
and S110, acquiring two-dimensional data of the battery sent by the vehicle in real time.
During the running process or the charging process of the vehicle, the vehicle end can send the two-dimensional data of the battery to the server end in real time. Wherein the battery two-dimensional data comprises two battery parameter data with intrinsic correlation. For example, the battery two-dimensional data includes battery voltage and battery temperature two-dimensional data. There is an inherent correlation between battery voltage and battery temperature. The fluctuations of the battery voltage and the battery temperature during driving often have a certain correlation. The battery voltage data and the battery temperature can respectively reflect the operation state of the battery. However, due to the inherent correlation between the battery voltage and the battery temperature, the battery may have an operational failure when the battery voltage data and the battery temperature are both within the standard range of the normal operating parameters of the battery.
And S120, inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model.
And S130, determining whether the battery is abnormal or not according to the prediction result.
The unsupervised learning model refers to a learning model with training data being label-free, and the unsupervised learning model can continuously mine and search the relation among data in a pile of data without selecting independent variables to predict dependent variables. Therefore, the unsupervised learning model can discover data intrinsic connections and solve various problems in pattern recognition from training samples of unknown classes.
The method and the device for detecting the abnormal data have the advantages that the trained unsupervised learning model is deployed at the server side, the server side calls the pre-trained unsupervised learning model in real time when the abnormal data are detected, the two-dimensional data of the battery are input into the pre-trained unsupervised learning model, the unsupervised learning model can output a prediction result, and the server side determines whether the two-dimensional data of the battery are abnormal or not according to the prediction result. Compared with the mode that whether the battery is abnormal or not is determined only through comparison of one-dimensional battery data and a fixed threshold value in the prior art, the embodiment of the disclosure can detect abnormal data of two battery parameter data with internal correlation, so that the problem that detection of each parameter of the battery alone causes abnormal detection of the battery and leakage detection of the battery can be avoided. Because normal data of the vehicle battery are relatively easy to obtain, a large amount of labor cost is spent on manually marking abnormal data, and the inherent correlation among all parameters of the vehicle battery is difficult to be described by applying a characteristic function, the embodiment of the invention inputs the two-dimensional data of the battery into the pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model, and further determines whether the battery is abnormal according to the prediction result, so that the problems of high labor cost and difficulty in training a proper characteristic function by the supervised learning model due to the prediction can be solved. Compared with the conventional unsupervised learning model which directly adopts high-dimensional data or performs dimension reduction mapping processing, the embodiment of the invention directly selects two battery parameter data with internal correlation as input, can reduce the operation complexity and is closer to the real battery performance.
In One embodiment, optionally, the unsupervised learning model may comprise, for example, a single Class support vector machine model, i.e., a One-Class-SVM machine learning unsupervised model. The One-Class-SVM machine learning unsupervised model does not need to mark output labels of a training set and does not have category labels. The positive sample in the sample is circled by searching a hyper-sphere, and the hyper-sphere is adopted for decision making, so that the sample in the hyper-sphere circle is the positive sample. The One-Class-SVM machine learning unsupervised model only focuses on the similarity or matching degree with the sample, and does not draw a conclusion on an unknown part, so that the automatic abnormality detection of the two-dimensional battery data with two intrinsically-associated battery parameter data in the embodiment of the invention can be accurately judged.
In an embodiment, optionally, if the unsupervised learning model is a single-class support vector machine model, S120 inputs the battery two-dimensional data into the pre-trained unsupervised learning model to obtain the prediction result output by the unsupervised learning model, including:
inputting the two-dimensional data of the battery into the single-classification support vector machine model, and acquiring the distance between the two-dimensional data of the battery and the center point of a hyper-sphere of the single-classification support vector machine model;
and outputting a prediction result according to a comparison result of the distance between the two-dimensional data of the battery and the center point of the hyper-sphere of the single-classification support vector machine model and the radius of the hyper-sphere.
Correspondingly, the step S130 of determining whether the two-dimensional data of the battery is abnormal according to the prediction result includes: and when the prediction result is that the distance between the battery two-dimensional data and the center point of the hyper-sphere of the single-classification support vector machine model is larger than the radius of the hyper-sphere, determining the battery two-dimensional data as abnormal data.
After the single classification support vector machine model is trained, the center point of the hyper-sphere and the radius of the hyper-sphere are known. If the center point of the hypersphere of the single classification support vector machine model is O and the radius of the hypersphere is R, the prediction result can be output according to the comparison result of the distance between the two-dimensional data of the battery and the center point of the hypersphere of the single classification support vector machine model and the radius of the hypersphere R. For example, if the distance between the two-dimensional battery data and the center point of the hypersphere of the single classification support vector machine model is smaller than or equal to the radius R of the hypersphere, the output prediction result is +1, and if the distance between the two-dimensional battery data and the center point of the hypersphere of the single classification support vector machine model is larger than or equal to the radius R of the hypersphere, the output prediction result is-1. If the distance between the input two-dimensional data of the battery and the center point of the hyper-sphere is larger than the radius of the hyper-sphere, the two-dimensional data of the battery is positioned outside the hyper-sphere ring, and then the two-dimensional data of the battery is determined as abnormal data.
In one embodiment, the battery two-dimensional data may optionally include battery voltage and battery temperature two-dimensional data, battery current and battery temperature two-dimensional data, battery state of charge and battery temperature two-dimensional data, or battery current and battery state of charge two-dimensional data.
In the battery parameter data, the battery voltage and the battery temperature have an intrinsic correlation, the battery current and the battery temperature have an intrinsic correlation, the battery state of charge and the battery temperature have an intrinsic correlation, and the battery current and the battery state of charge have an intrinsic correlation, so that when the abnormal data of the battery is detected, it can be set that the two-dimensional data of the battery comprises at least one of two-dimensional data of the battery voltage and the battery temperature, two-dimensional data of the battery current and the battery temperature, two-dimensional data of the battery state of charge and the battery temperature, or two-dimensional data of the battery current and the battery state of charge.
In one embodiment, optionally, before obtaining the pre-trained unsupervised learning model, further comprising: and training and generating the unsupervised learning model according to historical battery two-dimensional data.
And training an unsupervised learning model by using historical battery two-dimensional data generated in the vehicle driving process or the charging process. The historical battery two-dimensional data also includes two battery parameter data having an intrinsic correlation, including at least one of battery voltage and battery temperature two-dimensional data, battery current and battery temperature two-dimensional data, battery state of charge and battery temperature two-dimensional data, or battery current and battery state of charge two-dimensional data, for example.
In an embodiment, optionally, fig. 2 is a schematic diagram of a training process of an unsupervised learning model according to an embodiment of the present invention. As shown in fig. 2, the unsupervised learning model is generated according to the two-dimensional data training of the historical battery, and includes:
s210, obtaining historical battery data, wherein the historical battery data comprises first parameter data of a historical battery and second parameter data of the historical battery.
For example, historical battery data stored in the server and generated during driving or charging of the vehicle may be used. The historical battery data includes historical battery first parameter data and historical battery second parameter data. The historical battery first parameter data and the historical battery second parameter data have an intrinsic correlation. The historical battery first parameter data may be, for example, historical battery voltage data, and the historical battery second parameter data may be, for example, historical battery temperature data.
And S220, performing data cleaning on the historical battery data.
The data cleaning of the historical battery data includes operations such as elimination of null data and repeated data. Optionally, for example, if the voltage data of the battery cells and the temperature data of the corresponding battery cells at a certain time are used in the driving process, if the electric vehicle is designed such that a plurality of battery cells share one temperature sensor, the acquired temperature data acquired by the acquired temperature sensor needs to be split. Namely, a plurality of battery cores sharing one temperature sensor are endowed with the temperature data acquired by the temperature sensor. If the collected temperature data and the collected voltage data are stored in rows, row-column conversion can be performed according to needs so as to facilitate data input of subsequent model training.
And S230, combining the historical battery first parameter data and the historical battery second parameter data of the historical battery data after data cleaning to generate historical battery two-dimensional data.
The historical battery first parameter data and the historical battery second parameter data of the historical battery data after data cleaning are combined to generate historical battery two-dimensional data, and the historical battery two-dimensional data are converted into an array format required by subsequent model training, for example, a numpy array format.
S240, dividing the historical battery two-dimensional data into a training set and a verification set.
The historical battery two-dimensional data can be divided into a training set and a verification set, the data splitting ratio is not specifically limited in this embodiment, for example, the data ratio of the training set to the verification set is 7:3. the training set is used for model training, and the verification set is used for verifying the iterative optimization model.
And S250, training through the training set, and verifying through the verification set to iteratively optimize the unsupervised learning model.
After a training set sample is input into a model, setting all parameters of the model, wherein the set model parameters needing to be adjusted comprise abnormal point proportion nu, kernel function kernel and kernel function gamma parameter gamma in training, verifying through a verification set, continuously changing abnormal point proportion nu, kernel function kernel and kernel function gamma parameter gamma for training, and obtaining an unsupervised learning model through iterative optimization.
Optionally, when performing model training, a script file using Python as a programming language may be established, and numpy, matplotlib, and sklern software packages are introduced to perform data processing during the model training. For example, after model training, the effect of the One-Class-SVM machine learning unsupervised model can be checked by using a matplotlib built-in mapping module. FIG. 3 is a two-dimensional data distribution diagram of a historical battery, and FIG. 4 is a screening data distribution diagram of an One-Class-SVM machine learning unsupervised model. As can be seen from fig. 4, the One-Class-SVM machine learning unsupervised model can perform the abnormal data screening of two battery parameter data having intrinsic correlation by performing two classifications on the abnormal data, i.e., distinguishing the abnormal data points from the normal data points. In fig. 3, the abscissa represents the battery voltage and the ordinate represents the battery temperature.
In one embodiment, optionally, after training through the training set and verifying through the validation set to iteratively optimize the unsupervised learning model, the method may further include:
and acquiring two-dimensional data of a plurality of preset abnormal batteries, inputting the two-dimensional data into the unsupervised learning model, and determining the unsupervised learning model as an optimal unsupervised learning model if the output accuracy of the unsupervised learning model is greater than a preset value.
For example, a random function in the numpy software package can be used to create and generate a plurality of preset abnormal vehicle two-dimensional data for detecting the detection capability of the model. And if the output accuracy of the unsupervised learning model is greater than a preset value, determining the unsupervised learning model as an optimal unsupervised learning model, otherwise, continuing to pass through the training set and set verification, training and debugging model parameters, and iteratively optimizing the unsupervised learning model.
The following describes the training process of the single-class support vector machine model in detail by using a specific example.
Part of the code for the training of the single-class support vector machine is as follows:
fit (X _ train) training sample data of the training set;
y _ pred _ train = clf
y _ pred _ test = clf
Inputting abnormal points into the model by y _ pred _ outputs = clf
n _ error _ train = y _ pred _ train [ y _ pred _ train = = -1] number of anomalies for size training set predictor
n _ error _ test = y _ pred _ test [ y _ pred _ test = = -1] number of anomalies for size validation set prediction value
n _ error _ outliers = y _ pred _ outliers [ y _ pred _ outliers = = -1] size manual entry of the number of anomalies of the predicted values in the anomaly points
Where train represents a training set and test represents a prediction set.
Fit (X _ train) represents the detection of the boundary of the hyper-sphere from the training samples, the outlier ratio nu and the kernel function gamma parameter gamma.
y _ pred _ train = clf.
y _ pred _ test = clf.
y _ pred _ entries = clf.predict (X _ entries) indicates that the singular classification support vector machine is inputted with an outlier to obtain a predicted value.
n _ error _ train = y _ pred _ train [ y _ pred _ train = = -1] size represents the number of anomalies for the corresponding predictor of the training set.
n _ error _ test = y _ pred _ test [ y _ pred _ test = = -1]. Size represents the number of anomalies for the corresponding predicted values of the validation set.
n _ error _ outliere = y _ pred _ outliers [ y _ pred _ outliers = = -1]. Size represents the abnormal number of the predicted values corresponding to the preset abnormal vehicle two-dimensional data which are manually input.
In an embodiment, optionally, if the result of detecting the abnormal data of the two-dimensional battery data by using the unsupervised learning model is abnormal, the abnormality warning information is sent to the vehicle side.
If the result of abnormal data detection of the two-dimensional data of the battery by the server side through the unsupervised learning model is abnormal, the abnormal data detection indicates that the battery is possible to be abnormal at the moment, and therefore abnormal early warning information is sent to the vehicle side to prompt a user of the vehicle side. The vehicle end user can select to carry out verification through other data source modes, and can also directly go to a maintenance point to monitor and maintain, so that the battery is prevented from continuously running under the abnormal condition, and the damage to the battery or other parts of the vehicle is avoided.
The abnormal data detection method provided by the embodiment of the invention can also be executed at the vehicle end, namely, a device with the detected abnormal data is deployed at the vehicle end. Fig. 5 is a schematic flowchart of another abnormal data detection method according to an embodiment of the present invention. The execution main body of the abnormal data detection method provided by the embodiment of the invention is a vehicle end. As shown in fig. 5, the abnormal data detecting method includes S310 to S330:
and S310, monitoring the two-dimensional data of the battery in real time.
During the driving or charging of the vehicle, the two-dimensional data of the battery can be monitored in real time by sensors, for example. Wherein the battery two-dimensional data comprises two battery parameter data with intrinsic correlation. There is an inherent correlation between, for example, battery voltage and battery temperature. The fluctuations of the battery voltage and the battery temperature during driving often have a certain correlation.
And S320, inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model.
The embodiment of the invention can pre-train the unsupervised learning model and deploy the trained unsupervised learning model at the vehicle end. When abnormal data is detected, the vehicle terminal calls a pre-trained unsupervised learning model in real time. The unsupervised learning model can continuously mine and search the relation between data in a pile of data without predicting dependent variables by selecting independent variables. Therefore, the unsupervised learning model can discover data intrinsic relations and solve various problems in pattern recognition according to training samples with unknown classes. For battery two-dimensional data comprising two battery parameter data with intrinsic correlation in the present disclosure, good classification of abnormal data can be achieved.
S330, determining whether the two-dimensional data of the battery are abnormal or not according to the prediction result.
According to the embodiment of the invention, the trained unsupervised learning model is deployed at the vehicle end, when abnormal data is detected, the vehicle end calls the pre-trained unsupervised learning model in real time, the battery two-dimensional data is input into the pre-trained unsupervised learning model, the unsupervised learning model can output a prediction result, and the vehicle end determines whether the battery two-dimensional data is abnormal or not according to the prediction result. Compared with the mode that whether the battery is abnormal or not is determined only through comparison of one-dimensional battery data and a fixed threshold value in the prior art, the embodiment of the disclosure can detect abnormal data of two battery parameter data with intrinsic correlation, so that the problem that detection of each parameter of the battery is independent to cause abnormal detection of the battery can be avoided. Because normal data of the vehicle battery are relatively easy to obtain, a large amount of labor cost is spent on manually marking abnormal data, and the inherent correlation among all parameters of the vehicle battery is difficult to be described by applying a characteristic function, the embodiment of the invention inputs the two-dimensional data of the battery into the pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model, and further determines whether the battery is abnormal according to the prediction result, so that the problems of high labor cost and difficulty in training a proper characteristic function by the supervised learning model due to the prediction can be solved. Compared with the conventional unsupervised learning model which directly adopts high-dimensional data or performs dimension reduction mapping processing, the embodiment of the invention directly selects two battery parameter data with internal correlation as input, can reduce the operation complexity and is closer to the real battery performance. Furthermore, unsupervised learning models are trained in advance, and therefore the computational power requirements for the hardware architecture are low.
Optionally, the unsupervised learning model in the embodiment of the present invention may also include a single-classification support vector machine model.
Optionally, the two-dimensional battery data in the embodiment of the present invention may include at least one of two-dimensional battery voltage and battery temperature data, two-dimensional battery current and battery temperature data, two-dimensional battery state of charge and battery temperature data, or two-dimensional battery current and battery state of charge data.
Optionally, before acquiring the pre-trained unsupervised learning model, the embodiment of the present invention further includes: and training and generating the unsupervised learning model according to the two-dimensional data of the historical battery.
Optionally, the generating the unsupervised learning model according to the two-dimensional data training of the historical battery in the embodiment of the present invention includes:
acquiring historical battery data, wherein the historical battery data comprises first parameter data of a historical battery and second parameter data of the historical battery; the historical battery first parameter data and the historical battery second parameter data have intrinsic correlation;
performing data cleaning on the historical battery data;
combining first parameter data of the historical battery data after data cleaning with second parameter data of the historical battery to generate two-dimensional data of the historical battery;
dividing the historical battery two-dimensional data into a training set and a verification set;
training through the training set, verifying through the verification set, and iteratively optimizing the unsupervised learning model.
Optionally, in the embodiment of the present invention, after training through the training set and verifying through the verification set to iteratively optimize the unsupervised learning model, the method further includes:
and acquiring two-dimensional data of a plurality of preset abnormal batteries, inputting the two-dimensional data into the unsupervised learning model, and determining the unsupervised learning model as the optimal unsupervised learning model if the output accuracy of the unsupervised learning model is greater than a preset value.
Optionally, if it is determined that the two-dimensional data of the battery is abnormal, the abnormal data is sent to a server, and/or abnormal early warning information is generated.
And if the result of abnormal data detection on the real-time monitored battery two-dimensional data by adopting the unsupervised learning model at the vehicle end is abnormal, the abnormal data can be sent to the server so that the server can perform statistical analysis to determine the abnormal type of the battery, or abnormal early warning information is locally generated to prompt a local driver.
The embodiment of the invention also provides an abnormal data detection device, which can be configured at the server side. Fig. 6 is a block diagram of an abnormal data detecting apparatus according to an embodiment of the present invention, which includes a first obtaining module 11, a first predicting module 12, and a first abnormal determining module 13.
The first obtaining module 11 is configured to obtain two-dimensional data of a battery sent by a vehicle in real time. The first prediction module 12 is configured to input the two-dimensional battery data into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model. The first abnormality determining module 13 is configured to determine whether the two-dimensional data of the battery is abnormal according to the prediction result. Wherein the battery two-dimensional data comprises two battery parameter data with intrinsic correlation.
It should be noted that the foregoing explanation applied to the embodiment of the abnormal data detection method deployed in the server is also applicable to the abnormal data detection apparatus of the present embodiment. The specific manner in which each module in the embodiment of the abnormal data detecting apparatus performs the operation has been described in detail in the embodiment of the method, and will not be described in detail here.
The embodiment of the invention also provides another abnormal data detection device, and the abnormal data detection device can be configured at a server side. Fig. 7 is a block diagram of an abnormal data detecting apparatus according to an embodiment of the present invention, which includes a second obtaining module 21, a second predicting module 22, and a second abnormal determining module 23.
The second obtaining module 21 is configured to monitor the two-dimensional data of the battery in real time. The second prediction module 22 is configured to input the two-dimensional battery data into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model. The second abnormality determining module 23 is configured to determine whether the two-dimensional data of the battery is abnormal according to the prediction result. Wherein the battery two-dimensional data comprises two battery parameter data with intrinsic correlation.
It should be noted that the foregoing explanation of the embodiment of the abnormal data detecting method applied to the vehicle end is also applicable to the abnormal data detecting apparatus of the present embodiment. The specific manner in which each module in the embodiment of the abnormal data detecting apparatus performs operations has been described in detail in the embodiment of the method, and will not be elaborated herein.
An embodiment of the present invention further provides an electronic device, which includes one or more processors and a storage device. The storage device is configured to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the abnormal data detection method according to any embodiment applied to the service side, or the abnormal data detection method according to any embodiment applied to the vehicle side.
Fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present invention. FIG. 8 illustrates a block diagram of an exemplary electronic device suitable for use in implementing embodiments of the present invention. The electronic device shown in fig. 8 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiment of the present invention.
As shown in fig. 8, the components of the electronic device may include, but are not limited to: one or more processors or processors 31, memory 32, and a bus 33 that couples the various system components. Optionally, the electronic device may further include one or more of an input/output (I/O) interface 34, and a communications component 35.
Electronic devices typically include a variety of computer system readable media. Such media may be any available media that is accessible by the electronic device and includes both volatile and nonvolatile media, removable and non-removable media.
The memory 32 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM) and/or cache memory. The electronic device may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, the storage system may be used to read from and write to non-removable, nonvolatile magnetic media (not shown in FIG. 8 and commonly referred to as a "hard drive"). Although not shown in FIG. 8, a magnetic disk drive for reading from and writing to a removable, nonvolatile magnetic disk (e.g., a "floppy disk") and an optical disk drive for reading from or writing to a removable, nonvolatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 33 by one or more data media interfaces. Memory 32 may include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the invention.
Embodiments also provide a computer-readable storage medium, on which a computer program is stored, where the computer program is used to implement the abnormal data detection method applied to any embodiment of the server side or the abnormal data detection method applied to any embodiment of the vehicle side when the computer program is executed by a processor.
In the above embodiments, all or part of the implementation may be realized by software, hardware, firmware or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the application to occur, in whole or in part.
Any reference to memory, database, or other medium used in the embodiments provided herein may include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM is available in many forms, such as Static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), and the like.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (13)
1. A method for detecting abnormal data of a battery, comprising:
acquiring two-dimensional data of a battery sent by a vehicle in real time; wherein the battery two-dimensional data comprises two battery parameter data with intrinsic correlation;
inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model;
and determining whether the two-dimensional data of the battery is abnormal or not according to the prediction result.
2. The abnormal data detection method according to claim 1, wherein the unsupervised learning model comprises a single-class support vector machine model.
3. The abnormal data detection method according to claim 2, wherein the step of inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model comprises the following steps:
inputting the two-dimensional data of the battery into the single-classification support vector machine model, and acquiring the distance between the two-dimensional data of the battery and the center point of a hyper-sphere of the single-classification support vector machine model;
outputting a prediction result according to a comparison result of the distance between the two-dimensional battery data and the center point of the hypersphere of the single-classification support vector machine model and the radius of the hypersphere;
the determining whether the two-dimensional data of the battery is abnormal according to the prediction result comprises the following steps:
and when the prediction result is that the distance between the two-dimensional battery data and the center point of the hypersphere of the single-classification support vector machine model is larger than the radius of the hypersphere, determining the two-dimensional battery data as abnormal data.
4. The abnormal data detection method according to claim 1, wherein the battery two-dimensional data includes at least one of battery voltage and battery temperature two-dimensional data, battery current and battery temperature two-dimensional data, battery state of charge and battery temperature two-dimensional data, or battery current and battery state of charge two-dimensional data.
5. The abnormal data detecting method according to claim 1, further comprising:
and training and generating the unsupervised learning model according to the two-dimensional data of the historical battery.
6. The abnormal data detection method of claim 5, wherein the training to generate the unsupervised learning model from the historical battery two-dimensional data comprises:
acquiring historical battery data, wherein the historical battery data comprises first parameter data of a historical battery and second parameter data of the historical battery; the historical battery first parameter data and the historical battery second parameter data have intrinsic correlation;
performing data cleaning on the historical battery data;
combining first parameter data of the historical battery data after data cleaning with second parameter data of the historical battery to generate two-dimensional data of the historical battery;
dividing the historical battery two-dimensional data into a training set and a verification set;
training through the training set, verifying through the verification set, and iteratively optimizing the unsupervised learning model.
7. The abnormal data detection method of claim 6, further comprising, after training through the training set, verifying through the validation set to iteratively optimize the unsupervised learning model:
and acquiring two-dimensional data of a plurality of preset abnormal batteries, inputting the two-dimensional data into the unsupervised learning model, and determining the unsupervised learning model as the optimal unsupervised learning model if the output accuracy of the unsupervised learning model is greater than a preset value.
8. An abnormal data detection method, comprising:
monitoring two-dimensional data of the battery in real time; wherein the battery two-dimensional data comprises two battery parameter data with intrinsic correlation;
inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model;
and determining whether the two-dimensional data of the battery is abnormal or not according to the prediction result.
9. The abnormal data detection method according to claim 1, wherein if it is determined that the two-dimensional battery data is abnormal, the two-dimensional battery data is sent to a server, and/or abnormal early warning information is generated.
10. An abnormal data detection apparatus, comprising:
the first acquisition module is used for acquiring two-dimensional data of the battery sent by the vehicle in real time;
the first prediction module is used for inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model;
the first abnormity judgment module is used for determining whether the two-dimensional data of the battery is abnormal according to the prediction result;
wherein the battery two-dimensional data comprises two battery parameter data having an intrinsic correlation.
11. An abnormal data detecting apparatus, comprising:
a second acquisition module; the device is used for monitoring the two-dimensional data of the battery in real time;
the second prediction module is used for inputting the two-dimensional data of the battery into a pre-trained unsupervised learning model to obtain a prediction result output by the unsupervised learning model;
the second abnormity judgment module is used for determining whether the two-dimensional data of the battery is abnormal according to the prediction result;
wherein the battery two-dimensional data comprises two battery parameter data having an intrinsic correlation.
12. An electronic device, comprising: one or more processors, and a storage device;
the storage device is configured to store one or more programs which, when executed by the one or more processors, cause the one or more processors to implement the abnormal data detecting method according to any one of claims 1 to 7 or the abnormal data detecting method according to any one of claims 8 to 9.
13. A computer-readable storage medium, characterized in that a computer program is stored thereon, which, when being executed by a processor, is adapted to carry out the abnormal data detecting method according to any one of claims 1 to 7, or the abnormal data detecting method according to any one of claims 8 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111111850.5A CN115837862A (en) | 2021-09-18 | 2021-09-18 | Abnormal data detection method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111111850.5A CN115837862A (en) | 2021-09-18 | 2021-09-18 | Abnormal data detection method and device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115837862A true CN115837862A (en) | 2023-03-24 |
Family
ID=85574505
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111111850.5A Pending CN115837862A (en) | 2021-09-18 | 2021-09-18 | Abnormal data detection method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115837862A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116736173A (en) * | 2023-08-10 | 2023-09-12 | 长江三峡集团实业发展(北京)有限公司 | Energy storage battery model construction and energy storage battery state judgment method and device |
-
2021
- 2021-09-18 CN CN202111111850.5A patent/CN115837862A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116736173A (en) * | 2023-08-10 | 2023-09-12 | 长江三峡集团实业发展(北京)有限公司 | Energy storage battery model construction and energy storage battery state judgment method and device |
| CN116736173B (en) * | 2023-08-10 | 2023-10-20 | 长江三峡集团实业发展(北京)有限公司 | Energy storage battery model construction and energy storage battery state judgment method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10650616B2 (en) | Fault diagnosis using distributed PCA architecture | |
| US7930122B2 (en) | Evaluating anomaly for one-class classifiers in machine condition monitoring | |
| US10354462B1 (en) | Fault diagnosis in power electronics using adaptive PCA | |
| US20230013544A1 (en) | Method, Apparatus and System for Detecting Abnormal Operating States of a Device | |
| CN109726058B (en) | Detection method and device and computer equipment | |
| CN110008082B (en) | Abnormal task intelligent monitoring method, device, equipment and storage medium | |
| CN116453438B (en) | Display screen parameter detection method, device, equipment and storage medium | |
| CN116304909A (en) | Abnormality detection model training method, fault scene positioning method and device | |
| CN113807418A (en) | Injection molding machine energy consumption abnormity detection method and system based on Gaussian mixture model | |
| CN115098962A (en) | Method for predicting residual life of mechanical equipment in degradation state based on hidden half Markov model | |
| CN115837862A (en) | Abnormal data detection method and device, electronic equipment and storage medium | |
| CN115905450A (en) | Unmanned aerial vehicle monitoring-based water quality abnormity tracing method and system | |
| CN112232370A (en) | Fault analysis and prediction method for engine | |
| CN116572747A (en) | Battery fault detection method, device, computer equipment and storage medium | |
| CN116714437B (en) | Hydrogen fuel cell automobile safety monitoring system and monitoring method based on big data | |
| CN116008820B (en) | Method, device and medium for detecting inconsistency of vehicle battery cells | |
| CN111027679A (en) | Abnormal data detection method and system | |
| US11334053B2 (en) | Failure prediction model generating apparatus and method thereof | |
| Gotz et al. | Application of anomaly detection algorithms in lithium-ion battery packs-a case study | |
| CN110751747A (en) | Data processing method and device | |
| CN110865939A (en) | Application program quality monitoring method and device, computer equipment and storage medium | |
| EP4379671A1 (en) | Assessment of input-output datasets using local complexity values and associated data structure | |
| CN117556331B (en) | AI-enhancement-based air compressor maintenance decision method and system | |
| CN117117923B (en) | Big data-based energy storage control grid-connected management method and system | |
| US20220269984A1 (en) | Continuous learning process using concept drift monitoring |
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 |