CN113761705A - Multi-sensor fusion method and system based on multi-dimensional attribute correlation analysis - Google Patents
Multi-sensor fusion method and system based on multi-dimensional attribute correlation analysis Download PDFInfo
- Publication number
- CN113761705A CN113761705A CN202110812171.4A CN202110812171A CN113761705A CN 113761705 A CN113761705 A CN 113761705A CN 202110812171 A CN202110812171 A CN 202110812171A CN 113761705 A CN113761705 A CN 113761705A
- Authority
- CN
- China
- Prior art keywords
- sensor data
- sensor
- data time
- acquiring
- correlation
- 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
- 238000010219 correlation analysis Methods 0.000 title claims abstract description 18
- 238000007500 overflow downdraw method Methods 0.000 title claims abstract description 16
- 230000004927 fusion Effects 0.000 claims abstract description 64
- 238000000034 method Methods 0.000 claims abstract description 56
- 230000008859 change Effects 0.000 claims description 44
- 230000008569 process Effects 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 5
- 230000001105 regulatory effect Effects 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 230000000875 corresponding effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000012935 Averaging Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000002596 correlated effect Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000035772 mutation Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention provides a multi-sensor fusion method and system based on multi-dimensional attribute correlation analysis, and relates to the technical field of multi-sensor data fusion. Firstly, acquiring consistency characteristics and stability characteristics of a sensor data time sequence; acquiring correlation characteristics of the sensor data time sequence based on an improved shape distance algorithm; then, calculating the reliability of the sensor data based on the consistency characteristics, the stability characteristics and the correlation characteristics; and finally, performing weighted fusion on the multi-sensor data based on the reliability of each sensor. The method solves the problems of large relative error and low precision of multi-sensor data fusion when most of sensor data are deviated, improves the accuracy of the data fusion of the multi-sensor in different scenes, and provides accurate guidance suggestion for the driving decision of the automatic driving vehicle.
Description
Technical Field
The invention relates to a multi-sensor data fusion technology, in particular to a multi-sensor fusion method and system based on multi-dimensional attribute correlation analysis.
Background
The multi-sensor data fusion is an important means for automatically sensing the environment of the vehicle, and the higher the accuracy of the data fusion is, the more accurate description of the target position can be obtained, which is very important for constructing an environment model and making an accurate driving decision. In the process of fusing the observation results of a plurality of radar sensors on a target to obtain a target position and construct an environment model, the data fusion of the plurality of radar sensors can be regarded as a homogeneous time series data fusion problem, and a weighting fusion method is generally adopted to solve the problem.
At present, for the problem of weighted fusion of multi-sensor data, the reliability of the sensor is generally defined by using the consistency characteristic and the stability characteristic of the data, and then the weighted fusion of the multi-sensor data is performed, which has a good effect when the sensor does not have large-area deviation. However, in practice, due to vehicle jolts or other environmental factors, it may happen that most of the sensors are biased at the same time. When most of sensor data have deviation, especially the deviation in the same direction, the sensor weight defined by data consistency and data stability cannot truly reflect the reliability of the sensor, so that the problems of high relative error, low precision and the like of data fusion can be caused.
Therefore, it is desirable to provide a new multi-sensor data fusion technology to overcome the problems of relatively large error and low precision of data fusion when most of the sensor data have deviations in the prior art.
Disclosure of Invention
Technical problem to be solved
Aiming at the defects of the prior art, the invention provides a multi-sensor fusion method and a multi-sensor fusion system based on multi-dimensional attribute correlation analysis, and solves the problems of large relative error and low precision existing in the prior multi-sensor data fusion technology when most of sensor data have deviation.
(II) technical scheme
In order to achieve the purpose, the invention is realized by the following technical scheme:
in a first aspect, the present invention provides a multi-sensor fusion method based on multi-dimensional attribute correlation analysis, the method including:
acquiring consistency characteristics and stability characteristics of a sensor data time sequence;
acquiring the distance between the sensor data time sequences based on the slope of the change of the sensor data time sequences and the standard Euclidean distance, and acquiring the correlation characteristics of the sensor data time sequences based on the distance between the sensor data time sequences;
calculating reliability of sensor data based on the consistency features, the stability features, and correlation features of the sensor;
performing weighted fusion of the multi-sensor data based on the reliability of each sensor.
Preferably, the consistency characteristic and the stability characteristic of the acquired sensor data include:
acquiring consistency characteristics of sensor data based on a support method; and acquiring stability characteristics of the sensor data based on the support degree variance.
Preferably, the acquiring the distance between the sensor data time series based on the slope of the change of the sensor data time series and the standard euclidean distance and the acquiring the correlation characteristic of the sensor data time series based on the distance between the sensor data time series includes:
s21, standardizing the sensor data time sequence;
s22, acquiring a standard Euclidean distance between sensor data time sequences based on the sensor data time sequences after the standardization processing;
s23, obtaining a difference value between adjacent time point data in the sensor data time sequence after the standardization processing, obtaining the change slope of the sensor data time sequence based on the difference value, and determining the change state mode of each data time sequence based on the change quantity of the slope;
and S24, acquiring correlation characteristics between any different attribute parameters in the sensor data based on the change state mode and the standard Euclidean distance.
Preferably, the method further comprises:
setting and adjusting adjustment factors and contribution factors when performing weighted fusion on multi-sensor data based on the reliability of each sensor, wherein the adjustment factors comprise a support degree adjustment factor, a stability adjustment factor and a correlation adjustment factor; the contribution factors include a support contribution factor, a stability contribution factor, and a relevance contribution factor.
Preferably, the method further comprises:
a multi-sensor data set is acquired and preprocessed to acquire a complete sensor data time series.
In a second aspect, the present invention further provides a multi-sensor fusion system based on multi-dimensional attribute correlation analysis, the system including:
the basic characteristic acquisition module is used for acquiring consistency characteristics and stability characteristics of the sensor data time sequence;
the correlation characteristic acquisition module is used for acquiring the distance between the sensor data time sequences based on the slope of the change of the sensor data time sequences and the standard Euclidean distance and acquiring the correlation characteristics of the sensor data time sequences based on the distance between the sensor data time sequences;
a reliability acquisition module that calculates reliability of sensor data based on the consistency characteristics, the stability characteristics, and the correlation characteristics of the sensor;
a data weighted fusion module that performs weighted fusion on the multi-sensor data based on the reliability of each sensor.
Preferably, the obtaining of the consistency characteristic and the stability characteristic of the sensor data by the basic characteristic obtaining module includes:
acquiring consistency characteristics of sensor data based on a support method; and acquiring stability characteristics of the sensor data based on the support degree variance.
Preferably, the correlation characteristic obtaining module obtains the distance between the sensor data time series based on the slope of the change of the sensor data time series and the standard euclidean distance, and obtains the correlation characteristic of the sensor data time series based on the distance between the sensor data time series, and the correlation characteristic includes:
s21, standardizing the sensor data time sequence;
s22, acquiring a standard Euclidean distance between sensor data time sequences based on the sensor data time sequences after the standardization processing;
s23, obtaining a difference value between adjacent time point data in the sensor data time sequence after the standardization processing, obtaining the change slope of the sensor data time sequence based on the difference value, and determining the change state mode of each data time sequence based on the change quantity of the slope;
and S24, acquiring correlation characteristics between any different attribute parameters in the sensor data based on the change state mode and the standard Euclidean distance.
Preferably, the system further comprises:
a parameter setting and adjusting module, configured to set and adjust adjustment factors and contribution factors when performing weighted fusion on multi-sensor data based on the reliability of each sensor, where the adjustment factors include a support degree adjustment factor, a stability adjustment factor, and a correlation adjustment factor; the contribution factors include a support contribution factor, a stability contribution factor, and a relevance contribution factor.
Preferably, the system further comprises:
the data acquisition and processing module acquires a multi-sensor data set and preprocesses the multi-sensor data set to acquire a complete sensor data time sequence.
(III) advantageous effects
The invention provides a multi-sensor fusion method and system based on multi-dimensional attribute correlation analysis. Compared with the prior art, the method has the following beneficial effects:
1. firstly, acquiring consistency characteristics and stability characteristics of a sensor data time sequence; acquiring correlation characteristics of the sensor data time sequence based on an improved shape distance algorithm; then, the reliability of the sensor data is calculated based on the consistency characteristics, the stability characteristics and the correlation characteristics; and finally, performing weighted fusion on the multi-sensor data based on the reliability of each sensor. The method solves the problems of large relative error and low precision of multi-sensor data fusion when most of sensor data are deviated, improves the accuracy of the data fusion of the multi-sensor in different scenes, and provides accurate guidance suggestion for the driving decision of the automatic driving vehicle.
2. The invention adopts an improved shape distance method to calculate the correlation between the same sensor observation parameters, can well measure the variation trend of the sensor data time sequence, thereby calculating the distance between the sensor data time sequences.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of a multi-sensor fusion method based on multi-dimensional attribute correlation analysis according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a plurality of sensors in an embodiment of the present invention when the sensors are offset in the same direction;
FIG. 3a is a schematic diagram illustrating a comparison of the variation trend of the x-coordinate value and the y-coordinate value when there is no abnormal data in the embodiment of the present invention;
FIG. 3b is a diagram illustrating a comparison of the variation trend of the x-coordinate value and the y-coordinate value when abnormal data exists according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described, and it is obvious that the described embodiments are a part of the embodiments of the present invention, but not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the application provides a multi-sensor fusion method and system based on multi-dimensional attribute correlation analysis, and solves the problems of large relative error and low precision existing in most sensor data in the existing multi-sensor data fusion technology when the sensor data are deviated.
In order to solve the technical problems, the general idea of the embodiment of the application is as follows:
in order to solve the problems that relative errors are large and precision is low in multi-sensor data fusion when most sensor data are deviated, the method adds an index of correlation characteristics among sensor observation parameters, calculates the reliability of the sensor data by combining consistency characteristics and stability characteristics of a sensor data time sequence, and then performs weighted fusion on the multi-sensor data based on the reliability of each sensor. The multi-sensor data are weighted and fused by the technology of the invention, so that the data fusion accuracy of the multi-sensor under different scenes is obviously improved.
In order to better understand the technical solution, the technical solution will be described in detail with reference to the drawings and the specific embodiments.
Example 1:
in a first aspect, the present invention first provides a multi-sensor fusion method based on multi-dimensional attribute correlation analysis, including:
s1, acquiring consistency characteristics and stability characteristics of the sensor data time sequence;
s2, acquiring the distance between the sensor data time sequences based on the slope of the change of the sensor data time sequences and the standard Euclidean distance, and acquiring the correlation characteristics of the sensor data time sequences based on the distance between the sensor data time sequences;
s3, calculating reliability of sensor data based on the consistency characteristics, the stability characteristics and the correlation characteristics of the sensor;
s4, performing weighted fusion on the multi-sensor data based on the reliability of each sensor.
Therefore, the consistency characteristic and the stability characteristic of the sensor data time sequence are firstly obtained; acquiring correlation characteristics of the sensor data time sequence based on an improved shape distance algorithm; then, the reliability of the sensor data is calculated based on the consistency characteristics, the stability characteristics and the correlation characteristics; and finally, performing weighted fusion on the multi-sensor data based on the reliability of each sensor. The method solves the problems of large relative error and low precision of multi-sensor data fusion when most of sensor data are deviated, improves the accuracy of the data fusion of the multi-sensor in different scenes, and provides accurate guidance suggestion for the driving decision of the automatic driving vehicle.
The following describes the implementation of an embodiment of the present invention in detail by taking multiple radar sensors as an example and combining the explanation of specific steps.
And S1, acquiring consistency characteristics and stability characteristics of the sensor data time series.
Before the consistency characteristic and the stability characteristic of the sequence are obtained, a multi-sensor data set needs to be obtained, preprocessing (preprocessing comprises missing data processing) is carried out on the multi-sensor data set, and then a complete sensor data time sequence is obtained. In particular, the method comprises the following steps of,
this example acquires multi-radar sensor data from a large scale autopilot dataset Nuscenes established by autopilots. Nuscenes is a typical unmanned multi-modal dataset, and comprises mainstream vehicle-mounted sensor data such as laser radar data, millimeter wave radar data and image data. In addition, in the process of environment sensing by the radar sensor, due to the influence of complicated road conditions such as shielding and the accuracy of a target identification algorithm, data loss is easy to occur during information fusion. Directly ignoring missing data can result in a reduction in accuracy of the fused result, especially when the sensor with the higher weight has missing data. And therefore the missing data needs to be handled. Because the sensor time sequence is a fixed-distance time sequence, the missing value can be interpolated by adopting the mean value of the data before and after the missing value, and a complete sensor data time sequence is obtained.
After obtaining the complete sensor data time series, consistency and stability characteristics are obtained based on the sensor data time series.
1) And acquiring consistency characteristics of the sensor data time sequence based on a support method.
Data consistency is an index used to express how close a certain sensor data is to all other sensor data. If the consistency of certain sensor data is high, the reliability of the sensor data is considered to be high; if the consistency of certain sensor data is low, the reliability of the sensor data is low. In describing the consistency of the sensor data, methods such as belief entropy, information entropy, probability distribution, preference relation, support degree and the like can be adopted. In order to make the calculation result of the data consistency more accurate, the invention preferentially uses a support degree method defined in an exponential decay function mode to describe the data consistency. The support degree can describe the difference between data, can amplify the influence of the tiny difference on the fusion result, and is more suitable for the actual application process. The support degree is expanded from the data difference of two sensors to the data difference of one sensor and other sensors, so that the data difference between one sensor and other sensors at the same time can be effectively expressed, and the weight of the sensor in data fusion can be obtained according to the support degree.
When the consistency of the sensor data is characterized by the support degree, the support degree defines the distance between the data in an exponential form, and is specifically defined as follows:
wherein r isA,BRefers to the support between A and B, mu is the support adjustment factor.
The support definition indicates that when the data distance is greater than a preset specific value, the data support is considered to be low.
Suppose a sensor SiAnd a sensor SjThe observation results at the time t are respectively
And
in the absence of an observation error,
and
the values of (a) should be equal. If there is an observation error, if
And
if the difference between the values of the two sensors is small, the mutual support degree of the two sensors is high when the x coordinate is observed; if it is not
And
if the difference between the values of (a) and (b) is large, the mutual support degree of the two sensors is low when the x coordinate is observed. According to the definition of mutual support degree, then the sensor S at the moment tiAnd a sensor SjMutual support with respect to x-coordinate observation at time t
Is represented as follows:
wherein mu is a support degree regulating factor.
In the same way, the sensor S at the time tiAnd a sensor SjMutual support of y coordinate observation at time t
Is represented as follows:
suppose there are n sensors { S1,S2,...,SnObserve the same object respectively, and the observed results at the time t are respectively
And
if the sensor SiThe observed values of (1) and other sensors are all small in mutual support degree, which indicates that the sensor S isiThe values of (a) are relatively inaccurate; if the sensor SiThe observed value of (2) and the observed values of other sensors have larger mutual support degree, which indicates that the sensor S isiThe value of (a) is relatively accurate. Taking the observation of x coordinate as an example, the mutual support degree is extended between one sensor and all other sensors, and the sensor S at the time t is usediSupport for observing x-coordinate
The data of the sensor is represented by the following data:
in a clear view of the above, it is known that,
the larger the x coordinate is, the more the sensor is observed at time t, the more the x coordinate is observedThe higher the proximity of the values of the sensors, the more accurate the value of the sensor is considered;
the smaller the sensor is, the lower the proximity of the sensor to all other sensors when the x-coordinate is observed at time t, the less accurate the sensor is considered to be.
Similarly, the sensor S at the time tiSupport for observing y-coordinate
Is defined as follows:
2) And acquiring the stability characteristic of the sensor data time sequence based on the support degree variance.
Data stability characterizes the behavior of the sensor over a certain time window. For time series data, the data at the current time is affected by the data at the previous time. Therefore, the sensor with high reliability at the historical moment can perform better at the current moment, namely, the sensor with higher data stability is more reliable. However, since the observed object is moving all the time in the driving environment, the stability of the observation data does not directly reflect the stability of the sensor. In this regard, the present invention uses the variance of sensor reliability as the basis for measuring sensor stability. Similar to the support, if the variance is greater than a certain value, the reliability of the stability-based sensor may be considered to be low; if the correlation change is greater than a certain value, the reliability of the sensor based on the correlation change may be considered low.
Taking the value of the x coordinate as an example, suppose that within a time window l, the sensor S is at time tiThe obtained support degree sequence is as follows:
then the average support of the sequence
The calculation is as follows:
considering that the new data has higher reference value, an exponential time decay factor is set, wherein the time decay factor eta of the moment qqComprises the following steps:
ηq＝e-λ(t-q)
wherein λ is an attenuation parameter.
Thus, for the observation of the x coordinate at time t within the time window l, the sensor SiVariance of (2)
Expressed as:
similarly, for the observation of the y coordinate at time t within time window l, sensor SiVariance of (2)
Expressed as:
s2, acquiring the distance between the sensor data time series based on the slope of the change of the sensor data time series and the standard Euclidean distance, and acquiring the correlation characteristic of the sensor data time series based on the distance between the sensor data time series.
When most of the sensor data in the sensor system are deviated, especially the same-direction deviation occurs, the reliability of the multi-sensor data cannot be described only by using the data consistency based on the support degree and the variance based on the change of the support degree. Since the deviated data may have a larger data support degree when the majority of sensor data is deviated, the variance based on the support degree may not reflect the change of the stability degree of the data. Referring to fig. 2, it can be seen that sensor 1 is the most reliable sensor, and sensor 2 and sensor 3 are offset in the same direction, which minimizes the support of sensor 1, maximizes variance, minimizes stability, and minimizes reliability, which is not in accordance with the actual situation. Therefore, new indexes are required to be introduced to characterize the reliability of multi-sensing data.
The results observed by the sensors have multi-dimensional attributes, and there is a correlation between the attributes. For example, the radar simultaneously observes multiple parameters of the same target, such as an x-axis value, a y-axis value, and a z-axis value representing the position of the target, and the offset angle and the velocity of the target, which are not independent of each other but are correlated. Considering that the observed object does not change suddenly in a certain time span, when a sensor is reliable, all parameters observed by the sensor should change with the same or similar trend. Thus, the correlation between the parameter sequences remains unchanged. As shown in fig. 3a, when there is no data anomaly, the trend of the relationship between the parameter x time series and the parameter y time series is stable. Based on this, the invention takes the correlation relationship (i.e. correlation) between the parameters as one of the criteria for measuring the reliability of the sensor, and the smaller the change of the correlation between the parameters is, the higher the reliability of the sensor is. Particularly, the method based on the shape distance is adopted to carry out more accurate division on the basis of the mode distance, and the correlation of the short sequence can be effectively analyzed. The following takes the correlation between the coordinate x parameter and the coordinate y parameter when the radar sensor observes the target as an example, to illustrate a specific process of obtaining the correlation between the parameters based on the improved shape distance method.
The shape distance method can measure the variation trend of the sequences well, thereby calculating the distance between the sequences. However, the distance of the traditional shape is focused on describing the relationship between line segments, and the description of the relationship between points has defects. As shown in fig. 3b, when t is 7, the x-axis data has abnormal mutation, and the mutation at the time t is 7 can be well described by using a shape distance method; however, the data at time t-8 is also considered abnormal data due to the slope drop of the abrupt change data. For this reason, the technical solution will improve the shape distance method, and the standard euclidean distance between the auxiliary data collectively describes the distance variation between different parameter time series at each time point. The method comprises the following specific steps:
first, to eliminate the influence of the data scale on the result, the data is normalized. The data are generally standardized by max-min to obtain a standard parameter time sequence X 'of an X coordinate axis'iAnd a standard parametric time series of Y 'coordinate axes'iWherein:
wherein,
and
respectively, representing the values of the x-coordinate and the y-coordinate at time t after normalization.
And secondly, calculating the distance between the time sequences with different parameters in the time window l at the time t. Calculating the difference between the adjacent time point data, and segmenting the sequence according to time to obtain:
wherein,
and
sets representing differences between x-axis and y-axis neighboring point-in-time data, respectively;
and
the difference between the x-axis and y-axis neighboring point-in-time data, respectively, is represented.
Calculating the slope k of the data change according to the difference between the datatAnd determining the mode of each piece of data. Wherein,
with the conventional shape distance method, the states of the sequence are considered to be divided into acceleration-down, horizontal-down, deceleration-down, invariant, deceleration-up, horizontal-up, and acceleration-up, which are described by the pattern M { -3, -2, -1,0,1,2,3}, and a threshold th is set for pattern division. The threshold th is one of the criteria for determining whether the sequence status changes. To prevent the influence of the minute jitter of the sequence on the distance analysis, th is typically set to a positive number close to 0, the value of which can be obtained experimentally and is typically set to 0.2.
When k istWhen < -th, the state of the sequence is descending. The falling pattern is now divided by the amount of change in slope.
When Δ ktWhen 0, the sequence state is horizontal falling, defining mt＝-2；
When Δ ktWhen < 0, the sequence state is accelerated and falls, defining mt＝-3；
When Δ ktAt > 0, the sequence state is a descending deceleration, defining mt＝-1。
When k istAt > th, the sequence status is up.
When Δ ktWhen 0, the sequence state is horizontal falling, defining mt＝2；
When Δ ktAt < 0, the sequence state is horizontal decline, defining mt＝1；
When Δ ktAt > 0, the sequence status is horizontally decreasing, defining mt＝3。
When-th is less than or equal to ktWhen the sequence is less than or equal to th, the sequence is considered to be approximately unchanged, and m is definedt0. The sequence state judgment criteria are specifically shown in table 1 below.
TABLE 1 sequence status decision Table
Furthermore, in order to be able to better characterize the distance between the sequences, the standard Euclidean distance d between corresponding time points within the time window is calculatedt:
Wherein,
Finally, time t, sequence XiAnd YiA distance D betweentComprises the following steps:
wherein m isxnA mode value representing a sequence of x coordinates at time n; m isynA mode value representing a y-coordinate sequence at time n; dnRepresenting the euclidean distance of the sequence of times x and y at n.
Calculating the variable quantity D of the incidence relation at the time ttVariation D of association relation with t-1 momentt-1Difference of (2)
S3, calculating reliability of the sensor data based on the consistency characteristics, the stability characteristics, and the correlation characteristics of the sensor.
After three data characteristics of consistency, stability, relevance and the like of the multi-sensor data are obtained, the reliability of the sensor is calculated based on the three data characteristics.
As can be seen from the above analysis, the greater the support degree of the sensor describing the data consistency, the closer the sensor data is to other sensor data, the higher the data reliability of the sensor is; the smaller the support variance describing the stability of the sensor, the higher the observation stability of the sensor in the time window is, and the higher the data reliability of the sensor is; the smaller the difference value of the variation of the incidence relation describing the variation of the incidence relation among the multiple parameters observed by the sensor is, the more stable the evolution trend of the incidence relation among the parameters is, and the higher the data reliability of the sensor is.
In order to make the metrics of the three aspects the same, the reliability of the sensor can be formulated as follows:
wherein,
and
respectively representing the sensor S at time tiReliability when x and y coordinates are observed, (α, β) are adjustment factors for sensor stability and correlation change (i.e., correlation), respectively, (a, b, c) are contribution factors for sensor consistency, stability and correlation change to sensor reliability;
and
respectively representing the sensor S at time tiThe support degree when x coordinates and y coordinates are observed;
and
respectively representing the x-coordinate and the y-coordinate of the target object observed by the sensoriThe variance of (a);
represents the variation D of the association relation at time ttVariation D of association relation with t-1 momentt-1The difference of (a).
In order to achieve more accurate data fusion, adjustment factor and contribution factor parameters are set and adjusted when multi-sensor data is subjected to weighted fusion based on the reliability of each sensor. Wherein the regulating factors comprise a support degree regulating factor mu, a stability regulating factor alpha and a correlation regulating factor beta; the contribution factors include a support contribution factor a, a stability contribution factor b, and a relevance contribution factor c.
S4, performing weighted fusion on the multi-sensor data based on the reliability of each sensor.
The weight of the sensor is positively correlated with the reliability of the sensor, that is, the reliability of the sensor is high, and thus the weight of the sensor data is large when data fusion is performed.
At time t, the sensorSiWeights in observing the x-coordinate
Is defined as follows:
wherein n represents the number of sensors to be fused,
representing the weight assigned by the kth sensor at time t when observing the x coordinate.
Similarly, at time t, sensor SiWeights when observing the y-coordinate
Is defined as follows:
wherein n represents the number of sensors to be fused,
representing the weight assigned by the kth sensor at time t when observing the y coordinate.
Therefore, the result of data fusion of n sensors at the time t can be obtained
And
respectively expressed as:
thus, the whole process of the multi-sensor fusion method based on the multi-dimensional attribute correlation analysis is completed.
In order to verify the effectiveness of the method, the experimental result of the method (SSR) is compared with an average value method (AVG) and a method (SS) considering data consistency and stability, the fusion results of the x-axis value and the y-axis value under the conditions of all conditions (configuration 1), most sensor data difference conditions (configuration 2), few sensor data difference conditions (configuration 3) and no sensor obvious difference conditions (configuration 4) are respectively compared, the fusion accuracy of different methods under different scenes is summarized, and the results are shown in the following table 2.
TABLE 2 comparison of fusion results for different scenarios
From the above results, it can be seen that the influence of the obvious offset value on the fusion result is not considered when the information fusion is performed by using the averaging method (AVG), which may cause the obvious deviation of the fusion result; compared with the direct averaging method, the method only considering the consistency and stability of the data has the advantage that the accuracy of the fusion result is obviously improved, but when most of the data has obvious deviation, especially the equidirectional deviation, the accuracy of the fusion result is lower. Compared with other two methods, the data fusion method provided by the method (SSR) of the invention has the advantages that the accuracy rate is obviously improved under different scenes, and particularly, the fusion result is obviously improved under the condition that most sensors have differences.
Example 2:
in a second aspect, the present invention further provides a multi-sensor fusion system based on multi-dimensional attribute correlation analysis, the system comprising:
the basic characteristic acquisition module is used for acquiring consistency characteristics and stability characteristics of the sensor data time sequence;
the correlation characteristic acquisition module is used for acquiring the distance between the sensor data time sequences based on the slope of the change of the sensor data time sequences and the standard Euclidean distance and acquiring the correlation characteristics of the sensor data time sequences based on the distance between the sensor data time sequences;
a reliability acquisition module that calculates reliability of sensor data based on the consistency characteristics, the stability characteristics, and the correlation characteristics of the sensor;
a data weighted fusion module that performs weighted fusion on the multi-sensor data based on the reliability of each sensor.
Optionally, the consistency characteristic and the stability characteristic of the sensor data acquired by the basic characteristic acquiring module include:
acquiring consistency characteristics of sensor data based on a support method; and acquiring stability characteristics of the sensor data based on the support degree variance.
Optionally, the correlation characteristic obtaining module obtains a distance between the sensor data time series based on a slope of a change of the sensor data time series and a standard euclidean distance, and obtains the correlation characteristic of the sensor data time series based on the distance between the sensor data time series, including:
s21, standardizing the sensor data time sequence;
s22, acquiring a standard Euclidean distance between sensor data time sequences based on the sensor data time sequences after the standardization processing;
s23, obtaining a difference value between adjacent time point data in the sensor data time sequence after the standardization processing, obtaining the change slope of the sensor data time sequence based on the difference value, and determining the change state mode of each data time sequence based on the change quantity of the slope;
and S24, acquiring correlation characteristics between any different attribute parameters in the sensor data based on the change state mode and the standard Euclidean distance.
Optionally, the system further includes:
a parameter setting and adjusting module, configured to set and adjust adjustment factors and contribution factors when performing weighted fusion on multi-sensor data based on the reliability of each sensor, where the adjustment factors include a support degree adjustment factor, a stability adjustment factor, and a correlation adjustment factor; the contribution factors include a support contribution factor, a stability contribution factor, and a relevance contribution factor.
Optionally, the system further includes:
the data acquisition and processing module acquires a multi-sensor data set and preprocesses the multi-sensor data set to acquire a complete sensor data time sequence.
It can be understood that the multi-sensor fusion system based on the multi-dimensional attribute correlation analysis provided in the embodiment of the present invention corresponds to the multi-sensor fusion method based on the multi-dimensional attribute correlation analysis, and the explanation, examples, and beneficial effects of the relevant contents thereof may refer to the corresponding contents in the multi-sensor fusion method based on the multi-dimensional attribute correlation analysis, and are not described herein again.
In summary, compared with the prior art, the method has the following beneficial effects:
1. firstly, acquiring consistency characteristics and stability characteristics of a sensor data time sequence; acquiring correlation characteristics of the sensor data time sequence based on an improved shape distance algorithm; then, the reliability of the sensor data is calculated based on the consistency characteristics, the stability characteristics and the correlation characteristics; and finally, performing weighted fusion on the multi-sensor data based on the reliability of each sensor. The method solves the problems of large relative error and low precision of multi-sensor data fusion when most of sensor data are deviated, improves the accuracy of the data fusion of the multi-sensor in different scenes, and provides accurate guidance suggestion for the driving decision of the automatic driving vehicle.
2. The invention adopts an improved shape distance method to calculate the correlation between the same sensor observation parameters, can well measure the variation trend of the sensor data time sequence, thereby calculating the distance between the sensor data time sequences.
It is noted that, herein, 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. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
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 multi-sensor fusion method based on multi-dimensional attribute correlation analysis, the method comprising:
acquiring consistency characteristics and stability characteristics of a sensor data time sequence;
acquiring the distance between the sensor data time sequences based on the slope of the change of the sensor data time sequences and the standard Euclidean distance, and acquiring the correlation characteristics of the sensor data time sequences based on the distance between the sensor data time sequences;
calculating reliability of sensor data based on the consistency features, the stability features, and correlation features of the sensor;
performing weighted fusion of the multi-sensor data based on the reliability of each sensor.
2. The method of claim 1, wherein the obtaining consistency and stability characteristics of the sensor data comprises:
acquiring consistency characteristics of sensor data based on a support method; and acquiring stability characteristics of the sensor data based on the support degree variance.
3. The method of claim 1, wherein the obtaining distances between the sensor data time series based on the slopes of the changes of the sensor data time series and the standard euclidean distances and the correlation features of the sensor data time series based on the distances between the sensor data time series comprises:
s21, standardizing the sensor data time sequence;
s22, acquiring a standard Euclidean distance between sensor data time sequences based on the sensor data time sequences after the standardization processing;
s23, obtaining a difference value between adjacent time point data in the sensor data time sequence after the standardization processing, obtaining the change slope of the sensor data time sequence based on the difference value, and determining the change state mode of each data time sequence based on the change quantity of the slope;
and S24, acquiring correlation characteristics between any different attribute parameters in the sensor data based on the change state mode and the standard Euclidean distance.
And acquiring correlation characteristics between any different attribute parameters in the sensor data based on the change state mode and the standard Euclidean distance.
4. The method of claim 1, wherein the method further comprises:
setting and adjusting adjustment factors and contribution factors when performing weighted fusion on multi-sensor data based on the reliability of each sensor, wherein the adjustment factors comprise a support degree adjustment factor, a stability adjustment factor and a correlation adjustment factor; the contribution factors include a support contribution factor, a stability contribution factor, and a relevance contribution factor.
5. The method of claim 1, wherein the method further comprises:
a multi-sensor data set is acquired and preprocessed to acquire a complete sensor data time series.
6. A multi-sensor fusion system based on multi-dimensional attribute correlation analysis, the system comprising:
the basic characteristic acquisition module is used for acquiring consistency characteristics and stability characteristics of the sensor data time sequence;
the correlation characteristic acquisition module is used for acquiring the distance between the sensor data time sequences based on the slope of the change of the sensor data time sequences and the standard Euclidean distance and acquiring the correlation characteristics of the sensor data time sequences based on the distance between the sensor data time sequences;
a reliability acquisition module that calculates reliability of sensor data based on the consistency characteristics, the stability characteristics, and the correlation characteristics of the sensor;
a data weighted fusion module that performs weighted fusion on the multi-sensor data based on the reliability of each sensor.
7. The system of claim 6, wherein the base feature acquisition module acquiring consistency and stability features of sensor data comprises:
acquiring consistency characteristics of sensor data based on a support method; and acquiring stability characteristics of the sensor data based on the support degree variance.
8. The system of claim 6, wherein the correlation feature acquisition module acquires a distance between the sensor data time series based on a slope of a change of the sensor data time series and a standard euclidean distance and acquires the correlation feature of the sensor data time series based on the distance between the sensor data time series, comprising:
s21, standardizing the sensor data time sequence;
s22, acquiring a standard Euclidean distance between sensor data time sequences based on the sensor data time sequences after the standardization processing;
s23, obtaining a difference value between adjacent time point data in the sensor data time sequence after the standardization processing, obtaining the change slope of the sensor data time sequence based on the difference value, and determining the change state mode of each data time sequence based on the change quantity of the slope;
and S24, acquiring correlation characteristics between any different attribute parameters in the sensor data based on the change state mode and the standard Euclidean distance.
9. The system of claim 6, wherein the system further comprises:
a parameter setting and adjusting module, configured to set and adjust adjustment factors and contribution factors when performing weighted fusion on multi-sensor data based on the reliability of each sensor, where the adjustment factors include a support degree adjustment factor, a stability adjustment factor, and a correlation adjustment factor; the contribution factors include a support contribution factor, a stability contribution factor, and a relevance contribution factor.
10. The system of claim 6, wherein the system further comprises:
the data acquisition and processing module acquires a multi-sensor data set and preprocesses the multi-sensor data set to acquire a complete sensor data time sequence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110812171.4A CN113761705A (en) | 2021-07-19 | 2021-07-19 | Multi-sensor fusion method and system based on multi-dimensional attribute correlation analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110812171.4A CN113761705A (en) | 2021-07-19 | 2021-07-19 | Multi-sensor fusion method and system based on multi-dimensional attribute correlation analysis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113761705A true CN113761705A (en) | 2021-12-07 |
Family
ID=78787724
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110812171.4A Pending CN113761705A (en) | 2021-07-19 | 2021-07-19 | Multi-sensor fusion method and system based on multi-dimensional attribute correlation analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113761705A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115469136A (en) * | 2022-10-25 | 2022-12-13 | 南方电网数字电网研究院有限公司 | Non-invasive three-phase voltage measuring method based on electric field sensor array |
| CN115659284A (en) * | 2022-12-27 | 2023-01-31 | 河北新龙科技集团股份有限公司 | Big data fusion platform |
| CN116861370A (en) * | 2023-09-05 | 2023-10-10 | 山东千颐科技有限公司 | Motion data processing method for underground explosion-proof rubber-tyred vehicle |
| CN117454316A (en) * | 2023-12-25 | 2024-01-26 | 安徽蔚来智驾科技有限公司 | Multi-sensor data fusion method, storage medium and intelligent device |
| CN117574321A (en) * | 2024-01-16 | 2024-02-20 | 中国电建集团西北勘测设计研究院有限公司 | Dam safety assessment method based on mass monitoring and inspection information fusion model |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060224357A1 (en) * | 2005-03-31 | 2006-10-05 | Taware Avinash V | System and method for sensor data validation |
| CN102306206A (en) * | 2011-07-18 | 2012-01-04 | 福州大学 | Self-adapting consistent data fusion method |
| CN109343051A (en) * | 2018-11-15 | 2019-02-15 | 众泰新能源汽车有限公司 | A kind of multi-Sensor Information Fusion Approach driven for advanced auxiliary |
| US20190361460A1 (en) * | 2017-02-08 | 2019-11-28 | Marquette University | Robotic tracking navigation with data fusion |
-
2021
- 2021-07-19 CN CN202110812171.4A patent/CN113761705A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060224357A1 (en) * | 2005-03-31 | 2006-10-05 | Taware Avinash V | System and method for sensor data validation |
| CN102306206A (en) * | 2011-07-18 | 2012-01-04 | 福州大学 | Self-adapting consistent data fusion method |
| US20190361460A1 (en) * | 2017-02-08 | 2019-11-28 | Marquette University | Robotic tracking navigation with data fusion |
| CN109343051A (en) * | 2018-11-15 | 2019-02-15 | 众泰新能源汽车有限公司 | A kind of multi-Sensor Information Fusion Approach driven for advanced auxiliary |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115469136A (en) * | 2022-10-25 | 2022-12-13 | 南方电网数字电网研究院有限公司 | Non-invasive three-phase voltage measuring method based on electric field sensor array |
| CN115659284A (en) * | 2022-12-27 | 2023-01-31 | 河北新龙科技集团股份有限公司 | Big data fusion platform |
| CN115659284B (en) * | 2022-12-27 | 2023-07-18 | 河北新龙科技集团股份有限公司 | Big data fusion platform |
| CN116861370A (en) * | 2023-09-05 | 2023-10-10 | 山东千颐科技有限公司 | Motion data processing method for underground explosion-proof rubber-tyred vehicle |
| CN116861370B (en) * | 2023-09-05 | 2023-12-01 | 山东千颐科技有限公司 | Motion data processing method for underground explosion-proof rubber-tyred vehicle |
| CN117454316A (en) * | 2023-12-25 | 2024-01-26 | 安徽蔚来智驾科技有限公司 | Multi-sensor data fusion method, storage medium and intelligent device |
| CN117454316B (en) * | 2023-12-25 | 2024-04-26 | 安徽蔚来智驾科技有限公司 | Multi-sensor data fusion method, storage medium and intelligent device |
| CN117574321A (en) * | 2024-01-16 | 2024-02-20 | 中国电建集团西北勘测设计研究院有限公司 | Dam safety assessment method based on mass monitoring and inspection information fusion model |
| CN117574321B (en) * | 2024-01-16 | 2024-04-12 | 中国电建集团西北勘测设计研究院有限公司 | Dam safety assessment method based on mass monitoring and inspection information fusion model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113761705A (en) | Multi-sensor fusion method and system based on multi-dimensional attribute correlation analysis | |
| CN110501671B (en) | Target tracking method and device based on measurement distribution | |
| CN113777600B (en) | Multi-millimeter wave radar co-location tracking method | |
| CN110109095B (en) | Target feature assisted multi-source data association method | |
| CN111829505A (en) | Multi-sensor track quality extrapolation track fusion method | |
| CN110929796B (en) | Multi-source sensor-based decision layer data fusion method and system and storage medium | |
| CN112763993A (en) | Method and device for calibrating radar parameters, electronic equipment and storage medium | |
| CN113269811A (en) | Data fusion method and device and electronic equipment | |
| CN111161324A (en) | Target tracking method based on adaptive multi-mode updating strategy | |
| CN115767714A (en) | Vehicle Bluetooth key positioning method and device, vehicle and storage medium | |
| Ren et al. | Improved shape-based distance method for correlation analysis of multi-radar data fusion in self-driving vehicle | |
| CN116184386A (en) | Target tracking method, device and storage medium | |
| CN111259332A (en) | Fuzzy data association method and multi-target tracking method in clutter environment | |
| CN116228776B (en) | Electromechanical equipment welding defect identification method and system | |
| CN111596277B (en) | Flight target threat degree identification method and system based on fuzzy comprehensive evaluation method | |
| CN116049158A (en) | Soil information acquisition method based on multi-sensor data fusion algorithm | |
| CN115220002B (en) | Multi-target data association tracking method and related device for fixed single station | |
| CN114037017B (en) | Data fusion method based on error distribution fitting | |
| CN113747270B (en) | Cloud data center thing allies oneself with management and control system | |
| CN113010834B (en) | Ground inspection robot inspection coverage verification method and device based on GIS | |
| CN112344966B (en) | Positioning failure detection method and device, storage medium and electronic equipment | |
| CN111191734B (en) | Sensor data fusion method, device, equipment and storage medium | |
| CN111160190A (en) | Vehicle-mounted pedestrian detection-oriented classification auxiliary kernel correlation filtering tracking method | |
| CN116503390B (en) | Hardware part defect detection method based on computer vision | |
| CN114694138B (en) | Road surface detection method, device and equipment applied to intelligent driving |
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 |