CN113095197A - Vehicle driving state identification method and device, electronic equipment and readable storage medium - Google Patents
Vehicle driving state identification method and device, electronic equipment and readable storage medium Download PDFInfo
- Publication number
- CN113095197A CN113095197A CN202110366312.4A CN202110366312A CN113095197A CN 113095197 A CN113095197 A CN 113095197A CN 202110366312 A CN202110366312 A CN 202110366312A CN 113095197 A CN113095197 A CN 113095197A
- Authority
- CN
- China
- Prior art keywords
- inertial sensor
- vehicle
- data
- sensor data
- driving state
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000003860 storage Methods 0.000 title claims abstract description 18
- 230000009183 running Effects 0.000 claims abstract description 56
- 238000012937 correction Methods 0.000 claims abstract description 28
- 238000012545 processing Methods 0.000 claims abstract description 23
- 238000012549 training Methods 0.000 claims description 30
- 230000002457 bidirectional effect Effects 0.000 claims description 11
- 238000009826 distribution Methods 0.000 claims description 9
- 238000012795 verification Methods 0.000 claims description 8
- 238000005070 sampling Methods 0.000 claims description 4
- 241000288105 Grus Species 0.000 claims description 3
- 230000001174 ascending effect Effects 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 abstract description 8
- 238000013473 artificial intelligence Methods 0.000 abstract description 6
- 238000004422 calculation algorithm Methods 0.000 abstract description 4
- 239000010410 layer Substances 0.000 description 46
- 230000001133 acceleration Effects 0.000 description 18
- 238000010586 diagram Methods 0.000 description 9
- 238000004891 communication Methods 0.000 description 7
- 230000003068 static effect Effects 0.000 description 7
- 238000005303 weighing Methods 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 5
- 238000013145 classification model Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 108010076504 Protein Sorting Signals Proteins 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000002356 single layer Substances 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Navigation (AREA)
Abstract
The invention provides a vehicle driving state identification method, a vehicle driving state identification device, electronic equipment and a readable storage medium, wherein the method comprises the following steps: acquiring inertial sensor data within a preset time window length; carrying out coordinate system correction, zero offset correction and standardization processing on the inertial sensor data to obtain standardized inertial data; and inputting the standardized inertia data into a vehicle running state recognition model to obtain the vehicle running state output by the vehicle running state recognition model. According to the vehicle running state identification method provided by the embodiment of the invention, after the data of the inertial sensor is subjected to standardized processing, the vehicle running state is determined by the vehicle running state identification model, the traditional GPS and external equipment are not required, the vehicle running state can be identified in real time and at high precision only by carrying out simple vehicle body inertial sensor deployment, and meanwhile, the vehicle running state identification method has the characteristics of high reliability and strong robustness by means of data preprocessing and an artificial intelligence algorithm model.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to a vehicle driving state identification method and device, electronic equipment and a readable storage medium.
Background
The vehicle running state is used as an important characteristic for describing vehicle behaviors and plays a key role in a vehicle-mounted weighing system, an automobile auxiliary driving system and a driving safety monitoring system.
In the prior art, the vehicle running state is mainly obtained through GPS/Beidou satellite positioning, but the method can only identify the static state and the motion state of the vehicle and cannot meet a more refined vehicle running state identification task; in addition, for example, in a garbage collection and transportation truck with a vehicle-mounted weighing system, the vehicle-mounted weighing system needs to determine whether the vehicle is in a static state, and the vehicle may enter an underground parking lot of a residential area during the collection and transportation process, so that a GPS signal is lost, and the identification of the driving state of the vehicle is affected.
Disclosure of Invention
In view of this, an object of an embodiment of the present invention is to provide a method, an apparatus, an electronic device and a readable storage medium for identifying a driving state of a vehicle, which specifically include:
in a first aspect, an embodiment of the present invention provides a vehicle driving state identification method, where the method includes:
acquiring inertial sensor data within a preset time window length;
carrying out coordinate system correction, zero offset correction and standardization processing on the inertial sensor data to obtain standardized inertial data;
inputting the standardized inertia data into a vehicle running state recognition model to obtain a vehicle running state output by the vehicle running state recognition model;
the vehicle driving state identification model is obtained by taking a sample inertial sensor data set formed by sample inertial sensor data as a training sample and taking a corresponding sample vehicle driving state as a label for training.
Optionally, the performing coordinate system correction, zero offset correction, centering, and normalization on the inertial sensor data to obtain normalized inertial data specifically includes:
rotating the inertial sensor data from a corresponding reference coordinate system to a vehicle coordinate system;
correcting the inertial sensor data for zero offset error;
and adjusting the numerical distribution interval of the inertial sensor data in each coordinate axis to obtain standardized inertial data.
Optionally, the vehicle driving state recognition model comprises an input layer, a hidden layer and an output layer;
the input layer is to receive inertial sensor data over the time window length;
the hidden layer sequentially comprises at least one-dimensional convolution layer, a flattening layer and at least one two-way gating circulation layer; each one-dimensional convolution layer is provided with at least one-dimensional convolution kernel and is used for acquiring the inertial sensor data received by the input layer and extracting the characteristics of the inertial sensor data; the flattening layer is used for connecting the characteristics extracted by the one-dimensional convolution layer in series; each bidirectional gated loop layer has at least one bidirectional GRU for memorizing long-short term historical characteristics of the inertial sensor data;
the output layer is used for determining the vehicle running state corresponding to the inertial sensor data according to the output of the hidden layer.
Optionally, the method further comprises: determining network hyper-parameters of a vehicle driving state identification model;
the network hyper-parameter comprises the length of the time window, the value position of the corresponding state in the window, the number of the one-dimensional convolution layers, the number of the one-dimensional convolution kernels in each one-dimensional convolution layer, the number of the two-way gating circulation layers and the number of the two-way GRUs in each two-way gating circulation layer.
Optionally, the determining a network hyper-parameter of the vehicle driving state identification model specifically includes:
carrying out disorder processing on the sample inertial sensor data in the sample inertial sensor data set to obtain a disorder sample;
dividing the out-of-order samples into an out-of-order sample training set and an out-of-order sample verification set;
based on the out-of-order sample training set, carrying out grid search on different network hyper-parameter combinations, and recording the classification accuracy of all parameter combinations on the out-of-order sample verification set;
and selecting the network hyper-parameter combination with the highest accuracy as the network hyper-parameter of the vehicle driving state identification model.
Optionally, the acquiring inertial sensor data within a preset time window length specifically includes:
installing and fixing an inertial sensor comprising a three-axis gyroscope and a three-axis accelerometer on a vehicle;
periodically sampling data of an inertial sensor using an on-board terminal device connected to the inertial sensor.
Optionally, the vehicle driving condition comprises stationary, driving, stationary key-off, stationary load handling, smooth driving, bump driving, hard braking, up/down hill, left/right turn, or rollover.
In a second aspect, an embodiment of the present invention provides a vehicle driving state recognition apparatus, including:
the data acquisition module is used for acquiring inertial sensor data within a preset time window length;
the data standardization module is used for carrying out coordinate system correction, zero offset correction, centralization and standardization processing on the inertial sensor data to obtain standardized inertial data;
the state determination module is used for inputting the standardized inertia data into a vehicle running state recognition model to obtain a vehicle running state output by the vehicle running state recognition model;
the vehicle driving state identification model is obtained by taking a sample inertial sensor data set formed by sample inertial sensor data as a training sample and taking a corresponding sample vehicle driving state as a label for training.
In a third aspect, an embodiment of the present invention provides an electronic device, including:
one or more processors;
a memory for storing one or more programs;
wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the method of the first aspect.
In a fourth aspect, embodiments of the present invention provide a computer-readable storage medium having stored thereon executable instructions, which when executed by a processor, cause the processor to perform the method according to the first aspect.
According to the vehicle driving state identification method, the vehicle driving state identification device, the electronic equipment and the readable storage medium, after the data of the inertial sensor are subjected to standardized processing, the vehicle driving state is determined through the vehicle driving state identification model, the traditional GPS and external equipment are not needed, the vehicle driving state can be identified in real time and at high precision only through simple vehicle body inertial sensor deployment, and meanwhile, the vehicle driving state identification method has the advantages of being high in reliability and strong in robustness by means of data preprocessing and an artificial intelligence algorithm model.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings without creative work. The foregoing and other objects, features and advantages of the application will be apparent from the accompanying drawings. Like reference numerals refer to like parts throughout the drawings. The drawings are not intended to be to scale as practical, emphasis instead being placed upon illustrating the subject matter of the present application.
Fig. 1 is a flowchart illustrating a method for identifying a driving state of a vehicle according to an embodiment of the present invention.
Fig. 2 is a schematic flow chart illustrating an inertial sensor data preprocessing method according to an embodiment of the present invention.
FIG. 3 shows a schematic diagram of an inertial sensor coordinate system provided in accordance with an embodiment of the invention.
FIG. 4 illustrates another schematic diagram of an inertial sensor coordinate system provided in accordance with an embodiment of the invention.
Fig. 5 is a schematic structural diagram illustrating a vehicle driving state recognition model according to an embodiment of the present invention.
Fig. 6 is a flowchart illustrating a network hyper-parameter determination method of a vehicle driving state recognition model according to an embodiment of the present invention.
Fig. 7 is a flow chart of an inertial sensor data acquisition method according to an embodiment of the invention.
FIG. 8 is a schematic structural diagram illustrating a vehicle driving state recognition apparatus according to an embodiment of the present invention
FIG. 9 is a schematic diagram illustrating an electronic device according to an embodiment of the present invention
Detailed Description
Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood that the description is illustrative only and is not intended to limit the scope of the present disclosure. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present disclosure.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The words "a", "an" and "the" and the like as used herein are also intended to include the meanings of "a plurality" and "the" unless the context clearly dictates otherwise. Furthermore, the terms "comprises," "comprising," and the like, as used herein, specify the presence of stated features, steps, operations, and/or components, but do not preclude the presence or addition of one or more other features, steps, operations, or components.
All terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art unless otherwise defined. It is noted that the terms used herein should be interpreted as having a meaning that is consistent with the context of this specification and should not be interpreted in an idealized or overly formal sense.
The vehicle running state is used as an important characteristic for describing vehicle behaviors and plays a key role in a vehicle-mounted weighing system, an automobile auxiliary driving system and a driving safety monitoring system.
In the prior art, the vehicle running state is mainly obtained through GPS/Beidou satellite positioning, but the method can only identify the static state and the motion state of the vehicle and cannot meet a more refined vehicle running state identification task; in addition, for example, in a garbage collection truck equipped with an on-board weighing system, the on-board weighing system needs to determine whether the truck is in a stationary state, and the truck may enter a parking lot in a residential area during collection and transportation, thereby causing GPS signal loss and recognition of the running state of the truck.
In view of the above, an object of the embodiments of the present disclosure is to provide a method and an apparatus for identifying a driving state of a vehicle, an electronic device and a readable storage medium, and the following describes details of the embodiments of the present disclosure in conjunction with the accompanying drawings.
Fig. 1 shows a schematic flow chart of a vehicle driving state identification method provided by an embodiment of the present invention, which includes the following specific contents:
and step S110, acquiring inertial sensor data within a preset time window length.
The Inertial sensor in the embodiment of the present invention may also be referred to as an Inertial Measurement Unit (IMU), which is a sensor device capable of accurately measuring acceleration or angular velocity information of a measured object in various directions. Inertial sensors typically include a three-axis gyroscope and a three-axis accelerometer, and some additionally include a three-axis magnetometer primarily responsible for inertial data acquisition.
Generally, inertial sensor data of a measured object is measured by using an inertial sensor, and the inertial sensor data within a time window length needs to be collected so as to determine a vehicle driving state at a time near the time window. The time window length in this step is pre-calculated and set.
And step S120, carrying out coordinate system correction, zero offset correction and standardization processing on the inertial sensor data to obtain standardized inertial data.
The originally collected inertial sensor data is influenced by the structural difference and the installation position error of different vehicles, and the reference coordinate systems and the vehicle coordinate systems of different vehicle inertial sensors have larger difference, so that the complexity of the inertial sensor data is increased. Therefore, coordinate system corrections need to be made to the inertial sensor data, thereby reducing the complexity of the data distribution.
The zero offset error is one of the main errors of an inertial sensor, expressed as the degree of dispersion of the sampled values it outputs when the input angular rate of the inertial sensor is zero. When the vehicle driving state is recognized, the zero offset error affects the determination of the recognition result, and for example, the vehicle in a stationary state is recognized as a moving state, and therefore, it is necessary to perform zero offset correction on the inertial sensor data.
In a multi-index evaluation system, the evaluation indexes are different in size and magnitude due to different properties. When the levels of the indexes are greatly different, if the original index values are directly used for analysis, the function of the indexes with higher numerical values in the comprehensive analysis is highlighted, and the function of the indexes with lower numerical levels is relatively weakened. Specifically, the inertia characteristics of different vehicles in the embodiment of the invention may be different, and the numerical distribution of the three-axis inertia sensor between the axes is also greatly different. Therefore, in order to ensure the reliability of the result, it is necessary to further standardize the original index data to obtain standardized inertia data for the subsequent vehicle driving state identification step.
Step S130, inputting the normalized inertia data into a vehicle driving state identification model, and obtaining a vehicle driving state output by the vehicle driving state identification model.
According to the embodiment of the invention, the application of the inertial sensor is combined with the artificial intelligence technology, and the standardized inertial data obtained in the previous steps are specifically input into the vehicle driving state identification model, and the model can output the final identification result of the vehicle driving state.
The vehicle driving state identification model is obtained by taking a sample inertial sensor data set formed by sample inertial sensor data as a training sample and taking a corresponding sample vehicle driving state as a label for training. In order to construct a labeled training data set for supervised training of the model, the inertial sensor data needs to be labeled manually, that is, a corresponding vehicle driving state label is assigned to each moment in the inertial sensor data.
A sample inertial sensor data set formed from sample inertial sensor data in an embodiment of the present invention, and a label corresponding thereto, will be described below. Assuming a total length of the inertial sensor data set D is T, the format of a single sample in the data set D is:
wherein StObtained by means of manual marking and represents the normalized inertial data of the t-th moment
The corresponding vehicle driving state one-hot is coded, and
and
acceleration data and gyroscope data in the normalized inertial data, respectively. Thus, the structure of the data set D is as follows:
in order to construct a vehicle driving state recognition model based on time series, the data set D needs to be subjected to sliding window processing according to the window size w, and the vehicle driving state S at the kth moment in each window is processedkObtaining a data set D after sliding the window as the vehicle running state corresponding to the windoww：
Finally at DwThe windows containing different vehicle sensor data generated by continuous sliding windows are deleted so as to avoid polluting the data set, and thus the sample inertial sensor data set formed by the final sample inertial sensor data and the corresponding label are obtained. DwThe method can be further used for training the vehicle driving state recognition model.
The vehicle running state in the embodiment of the invention can be various states such as static, running, static flameout, static loading and unloading, stable running, bumpy running, sudden braking, ascending/descending, left/right turning or rollover and the like, and the number and the type of the actually selected states can be set according to the actual requirements of users. In practical application, when labels of a training data set are manually labeled, the labels needed by a user are given to a sample. Compared with the vehicle running state identification technology based on the GPS, the vehicle running state identification method provided by the embodiment of the invention supports a more refined and user-customizable vehicle state identification result.
According to the vehicle running state identification method provided by the embodiment of the invention, after the data of the inertial sensor is subjected to standardized processing, the vehicle running state is determined by the vehicle running state identification model, the traditional GPS and external equipment are not required, the vehicle running state can be identified in real time and at high precision only by carrying out simple vehicle body inertial sensor deployment, and meanwhile, the vehicle running state identification method has the characteristics of high reliability and strong robustness by means of data preprocessing and an artificial intelligence algorithm model.
Based on any of the above embodiments, fig. 2 shows a schematic flow chart of the inertial sensor data preprocessing method provided by the embodiment of the present invention, which is specifically described as follows.
Step S121, rotating the inertial sensor data from the corresponding reference coordinate system to a vehicle coordinate system.
The three-axis acceleration and the three-axis gyroscope raw data in the inertial sensor generally use the sensor itself as a fixed reference coordinate system, and fig. 3 shows a schematic diagram of an inertial sensor coordinate system with a model of MPU-6000/MPU-6050. As shown in fig. 4, there is a large difference between the reference coordinate system and the vehicle coordinate system of different vehicle inertial sensors due to the difference in the structure of different vehicles and the error in the installation position. Such a difference is a main factor affecting the accuracy of classification of the vehicle running state. Therefore, it is necessary to rotate the reference coordinate system of the inertial sensor data on different vehicles to the carrier coordinate system, i.e. the vehicle coordinate system, so that the measurement sensitivity directions of all the inertial sensor data are consistent.
Assuming an initial moment inertial sensor coordinate system (ox)syszs) Vehicle body coordinate system (ox)cyczc). Setting a sensor acceleration threshold K capable of ensuring that the vehicle is in a completely static state, judging the acceleration in time sequence, and acquiring a time-period acceleration mean value under the condition that the acceleration is smaller than the threshold K, namely the vehicle is static
Since the vehicle is only influenced by the acceleration of the earth's center at this time, the vehicle is driven by the acceleration of the earth's center
I.e. the components of the geocentric acceleration in the three axes of the sensor coordinate system. Roll angle gamma, pitch angle theta and yaw angle of inertial sensor at the moment
And
the relationship of (1) is:
thereby can pass through
Calculating a corresponding roll angle gamma and a corresponding pitch angle theta:
then defaults the yaw angle
The coordinate system rotation matrix R is thus obtained as follows:
thus in the inertial sensor coordinate system (ox)syszs) Three-axis acceleration signal acquired in
And three-axis gyroscope signals
Can be rotated to a vehicle body coordinate system (ox) through a matrix Rcyczc) The method comprises the following steps:
wherein
And
respectively a vehicle body coordinate system (ox)cyczc) Acceleration and gyroscope signals.
After the coordinate system is corrected, no matter where the inertial sensor is installed on the vehicle, no matter how large deflection angle exists between the inertial sensor and the vehicle, the corrected data can be guaranteed to have the same reference coordinate system, namely a vehicle body coordinate system, so that the complexity of data distribution is reduced, and the method has important help for improving the accuracy of a subsequent vehicle running state classification model.
And step S122, correcting the zero offset error of the inertial sensor data.
Acquiring the mean value of the acceleration data rotating within a period of time under the condition that the acceleration value of the sensor meets the threshold value K
And mean value of gyroscope data
And
i.e. the zero offset error of the inertial sensor. Thus, with respect to the vehicle body coordinate system (ox)cyczc) The following sensor signals are zero offset corrected as follows:
wherein A iscAnd GcRespectively a vehicle body coordinate system (ox)cyczc) Addition after zero offset correctionVelocity and gyroscope data.
And S123, adjusting the numerical distribution interval of the inertial sensor data in each coordinate axis to obtain standardized inertial data.
After the inertial sensor data are subjected to coordinate system correction and zero offset correction, the data are required to be used for constructing a vehicle driving state classification model in the step. Firstly, selecting K trolleys, wherein a triaxial acceleration signal sequence A with data length of F and a triaxial gyroscope signal sequence G are used for constructing a data set:
A＝[A1，1,A1,2,…,A1,F],…,[AK,1,AK,2,…,AK,F]]
G＝[G1,1,G1,2,…,G1,F],…,[GK,1,GK,2,…,GK,F]]
wherein A isi,j，Gi，j(i ═ 1,2, …, K, j ═ 1,2, …, F) are in the vehicle body coordinate system (ox), respectivelycyczc) Next, acceleration and gyro data at the jth time of the ith carriage are obtained. The data is then standardized, and the effect of the standardized data is described in the foregoing embodiments, which is not described herein again.
The specific processing mode is to calculate the mean value of the data sets A and G
And standard deviation sigmaA＝std(A)，σGThe data was finally centralized and normalized as follows:
through the above processing, data with a mean value of 0 and a standard deviation of 1, which obey the standard normal distribution, are finally obtained. The processing effectively reduces the complexity of the data and enables the data distribution to be consistent.
According to the inertial sensor data preprocessing method provided by the embodiment of the invention, the coordinate system correction, the zero offset correction and the standardization processing are carried out on the inertial sensor data to obtain the standardized inertial data, so that the complexity of the distribution of the inertial data is reduced, the precision of the inertial data is improved, and the method is of great help for improving the accuracy of the subsequent vehicle driving state classification model.
Based on any of the above embodiments, fig. 5 shows a schematic structural diagram of a vehicle driving state identification model provided by an embodiment of the present invention, and the specific content is as follows.
The vehicle driving state classification model in the embodiment of the invention is an artificial Neural Network model based on a one-dimensional CNN (Convolutional Neural Network) and a GRU (Gate Recurrent Unit), and the structure of the model is divided into an input layer, a hidden layer and an output layer. The input layer receives acceleration and gyroscope timing data X with a time window length w as follows:
wherein
Respectively in the vehicle body coordinate system (ox)cyczc) The normalized triaxial acceleration and triaxial gyroscope values at the next tth moment, X being 6 × w.
The hidden layer sequentially comprises LCA one-dimensional convolution layer, each layer having n in sequencei(i＝1,2,…,LC) One-dimensional convolution kernels with the length of 3 are used for feature extraction; 1 flattening layer for connecting the features extracted from the convolution layer in series; l isGA bidirectional gated cyclic layer, each layer having mj(j＝1,2,…,LG) And the bidirectional GRU is used for memorizing long-term and short-term historical characteristic information.
And the output layer receives the output of the hidden layer, and finally obtains one-hot codes of different driving state categories of the vehicle through a softmax function, so that the driving state of the vehicle corresponding to the kth moment in the data of the input layer is reflected.
The vehicle driving state identification model provided by the embodiment of the invention uses the neural network based on the one-dimensional CNN and the bidirectional GRU, so that the automatic extraction of the sensor sequence data characteristics is realized, the long-term and short-term characteristic information of the time sequence data is effectively utilized, and the identification accuracy of the model is obviously improved.
Based on any of the above embodiments, fig. 6 is a schematic flow chart illustrating a method for determining a network hyper-parameter of a vehicle driving state identification model according to an embodiment of the present invention, and the specific content is as follows.
The network hyper-parameter in the embodiment of the invention comprises the length of the time window, the value position of the corresponding state in the window, the number of the one-dimensional convolution layers, the number of the one-dimensional convolution kernels in each one-dimensional convolution layer, the number of the two-way gating circulation layers and the number of the two-way GRUs in each two-way gating circulation layer.
In order to improve the classification accuracy of the model, the optimal time window length w, the value position k of the corresponding state of the window, and the number L of convolution layers need to be foundCAnd the number n of convolution kernels per layeri(i＝1,2,…,LC) (ii) a Bidirectional GRU layer number LGAnd the number m of bidirectional GRU units per layerj(j＝1,2,…,LG). The 6 parameters were subjected to a parameter optimization test. The method comprises the following specific steps:
step S610, carrying out disorder processing on the sample inertial sensor data in the sample inertial sensor data set to obtain a disorder sample;
step S620, dividing the out-of-order samples into an out-of-order sample training set and an out-of-order sample verification set;
step S630, based on the out-of-order sample training set, carrying out grid search on different network hyper-parameter combinations, and recording the classification accuracy of all parameter combinations on the out-of-order sample verification set;
and step S640, selecting the network hyper-parameter combination with the highest accuracy as the network hyper-parameter of the vehicle driving state identification model.
The above steps are explained below by a specific example. And after disorder processing is carried out on the inertial sensor data set S, the inertial sensor data set S is divided into a training set and a verification set according to the proportion of 8:2, and the hyper-parameters of the neural network are optimized on the basis. The super-parameter optimizing range: the time window size w ∈ {10,30,60,120,180 }; the value position k of the window state belongs to {1,2,3, …, w }, and the number L of the one-dimensional CNN layersCE {1,2,3,4,5}, the number n of single-layer convolution kernelsi(i＝1,2,…,LC) E {5,10,20,50,100 }; bidirectional GRU layer number LGE to {1,2} layer, the data value range of the single-layer unit is mj(j＝1,2,…,LG) E.g., 5,10,20,50, 100. The network was encoded using the tenserflow machine learning framework under the python programming language, using the Adam optimizer, with a fixed learning rate of 0.001. Carrying out hyper-parameter grid search to obtain an optimal verification set with the accuracy rate of 98.93% and the corresponding time window size of 180 seconds; the window state value position is 100. The optimal network structure is 2 layers of one-dimensional CNN, and the number of convolution kernels is 60 and 30 respectively; 1 layer of bidirectional GRU, and the number of nodes is 50.
The network hyper-parameter determining method of the vehicle driving state recognition model provided by the embodiment of the invention can be used for determining the network hyper-parameter related to the model in advance before training the vehicle driving state model, thereby improving the classification accuracy of the model.
Based on any of the above embodiments, fig. 7 shows a schematic flow chart of the inertial sensor data acquisition method provided by the embodiment of the present invention, which is specifically described as follows.
And step S710, installing and fixing the inertial sensor comprising the three-axis gyroscope and the three-axis accelerometer on the vehicle.
The inertial sensor is mounted on the vehicle body, comprises a three-axis accelerometer and a three-axis gyroscope and is used for collecting vehicle acceleration data and angular velocity data, the inertial sensor needs to be fixed on the vehicle body or a vehicle frame during mounting, and the phenomenon that the sensor and the vehicle body generate relative displacement when the vehicle runs to influence the accuracy of the data is avoided.
And S720, periodically sampling the data of the inertial sensor by using the vehicle-mounted terminal equipment connected with the inertial sensor.
The inertial sensor is connected to the vehicle-mounted terminal device, the device uploads data to the cloud server for storage in real time through the mobile communication network, and the stored data are used for training the vehicle driving state recognition model.
The vehicle-mounted terminal device is generally an embedded terminal including an MCU chip, an a/D conversion chip, and a wireless communication chip. The equipment is mainly used for transmitting inertial data to the cloud storage unit and performing subsequent real-time data preprocessing and model reasoning calculation. The cloud storage unit comprises a cloud server with a large-capacity storage device, is used for storing a large amount of inertial data and provides a large amount of original data for a subsequent model training unit. In a specific implementation process, the embedded vehicle-mounted terminal equipment based on the ARM architecture can be used for sampling the original sensor data with the period of 1 second, and the original sensor data are uploaded through a wireless communication network and stored in a cloud server.
Based on any of the above embodiments, fig. 8 is a schematic structural diagram of a vehicle driving state recognition device according to an embodiment of the present invention, and the specific content is as follows.
The data acquisition module 801 is used for acquiring inertial sensor data within a preset time window length;
a data standardization module 802, configured to perform coordinate system correction, zero offset correction, centralization, and standardization processing on the inertial sensor data to obtain standardized inertial data;
a state determining module 803, configured to input the normalized inertial data into a vehicle driving state identification model, so as to obtain a vehicle driving state output by the vehicle driving state identification model;
the vehicle driving state identification model is obtained by taking a sample inertial sensor data set formed by sample inertial sensor data as a training sample and taking a corresponding sample vehicle driving state as a label for training.
According to the vehicle running state recognition device provided by the embodiment of the invention, after the data of the inertial sensor is subjected to standardized processing, the vehicle running state is determined by the vehicle running state recognition model, the traditional GPS and external equipment are not required, the vehicle running state can be recognized in real time and at high precision only by carrying out simple vehicle body inertial sensor deployment, and meanwhile, the vehicle running state recognition device has the characteristics of high reliability and strong robustness by means of data preprocessing and an artificial intelligence algorithm model.
Fig. 9 shows a schematic physical structure diagram illustrating an electronic device, and as shown in fig. 9, the electronic device may include: a processor (processor)910, a communication Interface (Communications Interface)920, a memory (memory)930, and a communication bus 940, wherein the processor 910, the communication Interface 920, and the memory 930 communicate with each other via the communication bus 940. Processor 910 may invoke logic instructions in memory 930 to perform the following method: acquiring inertial sensor data within a preset time window length; carrying out coordinate system correction, zero offset correction and standardization processing on the inertial sensor data to obtain standardized inertial data; inputting the standardized inertia data into a vehicle running state recognition model to obtain a vehicle running state output by the vehicle running state recognition model; the vehicle driving state identification model is obtained by taking a sample inertial sensor data set formed by sample inertial sensor data as a training sample and taking a corresponding sample vehicle driving state as a label for training.
Furthermore, the logic instructions in the memory 930 may be implemented in software functional units and stored in a computer readable storage medium when the logic instructions are sold or used as independent products. Based on such understanding, the technical solutions of the embodiments of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method described in the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In another aspect, an embodiment of the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program is implemented by a processor to perform the method provided by the foregoing embodiments, for example, including: acquiring inertial sensor data within a preset time window length; carrying out coordinate system correction, zero offset correction and standardization processing on the inertial sensor data to obtain standardized inertial data; inputting the standardized inertia data into a vehicle running state recognition model to obtain a vehicle running state output by the vehicle running state recognition model; the vehicle driving state identification model is obtained by taking a sample inertial sensor data set formed by sample inertial sensor data as a training sample and taking a corresponding sample vehicle driving state as a label for training.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A vehicle driving state recognition method, characterized by comprising:
acquiring inertial sensor data within a preset time window length;
carrying out coordinate system correction, zero offset correction and standardization processing on the inertial sensor data to obtain standardized inertial data;
inputting the standardized inertia data into a vehicle running state recognition model to obtain a vehicle running state output by the vehicle running state recognition model;
the vehicle driving state identification model is obtained by taking a sample inertial sensor data set formed by sample inertial sensor data as a training sample and taking a corresponding sample vehicle driving state as a label for training.
2. The method for recognizing a driving state of a vehicle according to claim 1, wherein the performing coordinate system correction, zero offset correction, centering, and normalization processing on the inertial sensor data to obtain normalized inertial data specifically includes:
rotating the inertial sensor data from a corresponding reference coordinate system to a vehicle coordinate system;
correcting the inertial sensor data for zero offset error;
and adjusting the numerical distribution interval of the inertial sensor data in each coordinate axis to obtain standardized inertial data.
3. The vehicle running state recognition method according to claim 1, wherein the vehicle running state recognition model includes an input layer, a hidden layer, and an output layer;
the input layer is to receive inertial sensor data over the time window length;
the hidden layer sequentially comprises at least one-dimensional convolution layer, a flattening layer and at least one two-way gating circulation layer; each one-dimensional convolution layer is provided with at least one-dimensional convolution kernel and is used for acquiring the inertial sensor data received by the input layer and extracting the characteristics of the inertial sensor data; the flattening layer is used for connecting the characteristics extracted by the one-dimensional convolution layer in series; each bidirectional gated loop layer has at least one bidirectional GRU for memorizing long-short term historical characteristics of the inertial sensor data;
the output layer is used for determining the vehicle running state corresponding to the inertial sensor data according to the output of the hidden layer.
4. The vehicle travel state recognition method according to claim 3, characterized by further comprising: determining network hyper-parameters of a vehicle driving state identification model;
the network hyper-parameter comprises the length of the time window, the value position of the corresponding state in the window, the number of the one-dimensional convolution layers, the number of the one-dimensional convolution kernels in each one-dimensional convolution layer, the number of the two-way gating circulation layers and the number of the two-way GRUs in each two-way gating circulation layer.
5. The vehicle driving state identification method according to claim 4, wherein the determining of the network hyper-parameter of the vehicle driving state identification model specifically comprises:
carrying out disorder processing on the sample inertial sensor data in the sample inertial sensor data set to obtain a disorder sample;
dividing the out-of-order samples into an out-of-order sample training set and an out-of-order sample verification set;
based on the out-of-order sample training set, carrying out grid search on different network hyper-parameter combinations, and recording the classification accuracy of all parameter combinations on the out-of-order sample verification set;
and selecting the network hyper-parameter combination with the highest accuracy as the network hyper-parameter of the vehicle driving state identification model.
6. The method for recognizing the driving state of the vehicle according to claim 1, wherein the collecting the inertial sensor data within a preset time window length specifically comprises:
installing and fixing an inertial sensor comprising a three-axis gyroscope and a three-axis accelerometer on a vehicle;
periodically sampling data of an inertial sensor using an on-board terminal device connected to the inertial sensor.
7. The vehicle driving state identification method according to any one of claims 1 to 6, wherein the vehicle driving state includes stationary, driving, stationary key-off, stationary loading and unloading, smooth driving, bumpy driving, hard braking, ascending/descending, left/right turning, or rollover.
8. A vehicle driving state recognition apparatus characterized by comprising:
the data acquisition module is used for acquiring inertial sensor data within a preset time window length;
the data standardization module is used for carrying out coordinate system correction, zero offset correction, centralization and standardization processing on the inertial sensor data to obtain standardized inertial data;
the state determination module is used for inputting the standardized inertia data into a vehicle running state recognition model to obtain a vehicle running state output by the vehicle running state recognition model;
the vehicle driving state identification model is obtained by taking a sample inertial sensor data set formed by sample inertial sensor data as a training sample and taking a corresponding sample vehicle driving state as a label for training.
9. An electronic device, comprising:
one or more processors;
a memory for storing one or more programs;
wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the method of any of claims 1-7.
10. A computer readable storage medium having stored thereon executable instructions, which when executed by a processor, cause the processor to perform the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110366312.4A CN113095197B (en) | 2021-04-06 | Vehicle driving state identification method and device, electronic equipment and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110366312.4A CN113095197B (en) | 2021-04-06 | Vehicle driving state identification method and device, electronic equipment and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113095197A true CN113095197A (en) | 2021-07-09 |
| CN113095197B CN113095197B (en) | 2024-07-16 |
Family
ID=
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116189332A (en) * | 2022-10-20 | 2023-05-30 | 开源网安物联网技术(武汉)有限公司 | Vehicle health scoring method and device, electronic equipment and storage medium |
| CN116502982A (en) * | 2023-06-26 | 2023-07-28 | 深圳市汉德网络科技有限公司 | Canned ship cargo flow direction control method and system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105320966A (en) * | 2015-10-30 | 2016-02-10 | 东软集团股份有限公司 | Vehicle driving state recognition method and apparatus |
| CN108563891A (en) * | 2018-04-23 | 2018-09-21 | 吉林大学 | A method of traffic accident is intelligently prevented based on Inertial Measurement Unit |
| US20190050711A1 (en) * | 2017-08-08 | 2019-02-14 | Neusoft Corporation | Method, storage medium and electronic device for detecting vehicle crashes |
| CN110263836A (en) * | 2019-06-13 | 2019-09-20 | 南京师范大学 | A kind of bad steering state identification method based on multiple features convolutional neural networks |
| CN110329271A (en) * | 2019-06-18 | 2019-10-15 | 北京航空航天大学杭州创新研究院 | A kind of multisensor vehicle driving detection system and method based on machine learning |
| CN111767933A (en) * | 2019-05-17 | 2020-10-13 | 北京京东尚科信息技术有限公司 | Method and device for identifying vehicle driving state |
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105320966A (en) * | 2015-10-30 | 2016-02-10 | 东软集团股份有限公司 | Vehicle driving state recognition method and apparatus |
| US20190050711A1 (en) * | 2017-08-08 | 2019-02-14 | Neusoft Corporation | Method, storage medium and electronic device for detecting vehicle crashes |
| CN108563891A (en) * | 2018-04-23 | 2018-09-21 | 吉林大学 | A method of traffic accident is intelligently prevented based on Inertial Measurement Unit |
| CN111767933A (en) * | 2019-05-17 | 2020-10-13 | 北京京东尚科信息技术有限公司 | Method and device for identifying vehicle driving state |
| CN110263836A (en) * | 2019-06-13 | 2019-09-20 | 南京师范大学 | A kind of bad steering state identification method based on multiple features convolutional neural networks |
| CN110329271A (en) * | 2019-06-18 | 2019-10-15 | 北京航空航天大学杭州创新研究院 | A kind of multisensor vehicle driving detection system and method based on machine learning |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116189332A (en) * | 2022-10-20 | 2023-05-30 | 开源网安物联网技术(武汉)有限公司 | Vehicle health scoring method and device, electronic equipment and storage medium |
| CN116189332B (en) * | 2022-10-20 | 2024-03-19 | 开源网安物联网技术(武汉)有限公司 | Vehicle health scoring method and device, electronic equipment and storage medium |
| CN116502982A (en) * | 2023-06-26 | 2023-07-28 | 深圳市汉德网络科技有限公司 | Canned ship cargo flow direction control method and system |
| CN116502982B (en) * | 2023-06-26 | 2023-09-08 | 深圳市汉德网络科技有限公司 | Canned ship cargo flow direction control method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111947671B (en) | Method, apparatus, computing device and computer-readable storage medium for positioning | |
| CN111142091B (en) | Automatic driving system laser radar online calibration method fusing vehicle-mounted information | |
| CN111415533B (en) | Bend safety early warning monitoring method, device and system | |
| CN110986938B (en) | Bumpy road identification method and device and electronic equipment | |
| CN111398989A (en) | Performance analysis method and test equipment of driving assistance system | |
| CN111027381A (en) | Method, device, equipment and storage medium for recognizing obstacle by monocular camera | |
| CN110914837A (en) | Method and system for determining orientation of vehicle head | |
| CN113514109A (en) | Unmanned aerial vehicle flight fault detection method and system | |
| CN111649740A (en) | Method and system for high-precision positioning of vehicle based on IMU | |
| CN117325903A (en) | High confidence decision system for intelligent flaw detection of rail | |
| CN113962301B (en) | Multi-source input signal fused pavement quality detection method and system | |
| CN114897409A (en) | Method and system for evaluating road risk based on vehicle driving | |
| CN113095197B (en) | Vehicle driving state identification method and device, electronic equipment and readable storage medium | |
| CN114417729A (en) | Mining area environment safety early warning method based on BP neural network | |
| CN113095197A (en) | Vehicle driving state identification method and device, electronic equipment and readable storage medium | |
| CN116664028B (en) | Cargo flow direction control method and device of transport vehicle and transport vehicle | |
| CN114021080A (en) | Trajectory prediction model training and trajectory prediction method, apparatus, device and medium | |
| CN111103578B (en) | Laser radar online calibration method based on deep convolutional neural network | |
| CN110263836B (en) | Bad driving state identification method based on multi-feature convolutional neural network | |
| US20210206515A1 (en) | Aviation component inspection using image recognition | |
| CN114074693A (en) | Train positioning method, device and system with multiple sensors integrated and train | |
| CN113112061B (en) | Method and device for predicting fuel consumption of vehicle | |
| Abdelgawad et al. | Estimating vehicle speed on highway roads from smartphone sensors using deep learning models | |
| CN114279452A (en) | Unmanned truck trailer attitude detection method and detection system | |
| CN115512541B (en) | Pavement accurate information service system for safe driving of vehicle |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |