CN116913104B - Average vehicle speed prediction method, device and storage medium based on tensor robust decomposition - Google Patents
Average vehicle speed prediction method, device and storage medium based on tensor robust decomposition Download PDFInfo
- Publication number
- CN116913104B CN116913104B CN202310999172.3A CN202310999172A CN116913104B CN 116913104 B CN116913104 B CN 116913104B CN 202310999172 A CN202310999172 A CN 202310999172A CN 116913104 B CN116913104 B CN 116913104B
- Authority
- CN
- China
- Prior art keywords
- vehicle speed
- average vehicle
- tensor
- road section
- average
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 22
- 238000005457 optimization Methods 0.000 claims abstract description 14
- 238000004364 calculation method Methods 0.000 claims abstract description 7
- 239000011159 matrix material Substances 0.000 claims description 74
- 230000006870 function Effects 0.000 claims description 30
- 238000013500 data storage Methods 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 7
- 238000012544 monitoring process Methods 0.000 claims description 6
- 238000005065 mining Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 4
- 230000000737 periodic effect Effects 0.000 description 2
- 230000003190 augmentative effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/052—Detecting movement of traffic to be counted or controlled with provision for determining speed or overspeed
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Abstract
The invention discloses a method, a device and a storage medium for predicting average vehicle speed based on tensor robust decomposition, which comprise the following steps of S1: receiving an instruction for predicting the historical average speed of the road section sent by a server, and acquiring and storing historical speed data of the road section from a sensor; s2: constructing an average vehicle speed tensor according to stored road section historical vehicle speed data; s3: and constructing an average vehicle speed prediction target loss function according to the average vehicle speed tensor, performing iterative optimization, and outputting an average vehicle speed predicted value. According to the invention, the non-negative tensor decomposition model is constructed by utilizing the historical vehicle speed data, and the objective function (cauchy loss) is constructed to measure the difference between the observed average vehicle speed value and the predicted value, so that the interference of outliers on the feature mining is effectively reduced, and the prediction accuracy is improved. Meanwhile, an objective function is updated by adopting an alternate direction multiplier method, so that higher calculation efficiency is obtained.
Description
Technical Field
The invention relates to the technical field of road traffic, in particular to an average vehicle speed prediction method, an average vehicle speed prediction device and a storage medium based on tensor robust decomposition.
Background
With the great development of modern industry, thousands of small sensors record massive traffic flow data, which also promotes the development of a series of intelligent transportation industries, such as intelligent transportation systems, automatic driving technology, internet of vehicles and the like. However, the traffic data actually collected usually has different degrees of loss phenomenon of traffic data sets due to equipment damage, transmission distortion and the like, and meanwhile, a large number of outliers exist, which restricts the development of intelligent traffic systems.
At present, most models adopt L2 norms to construct an objective function to realize prediction of traffic loss data. However, the raw data is typically mixed with outlier data, which will not guarantee the robustness of the model, which may greatly reduce the accuracy of the average vehicle speed prediction.
Disclosure of Invention
Aiming at the problem of lower accuracy of average vehicle speed prediction in the prior art, the invention provides an average vehicle speed prediction method, an average vehicle speed prediction device and a storage medium based on tensor robust decomposition.
In order to achieve the above object, the present invention provides the following technical solutions:
the average vehicle speed prediction method based on tensor robust decomposition specifically comprises the following steps:
s1: receiving an instruction for predicting the historical average speed of the road section sent by a server, and acquiring and storing historical speed data of the road section from a sensor;
s2: constructing an average vehicle speed tensor according to stored road section historical vehicle speed data;
s3: and constructing an average vehicle speed prediction target loss function according to the average vehicle speed tensor, performing iterative optimization, and outputting an average vehicle speed predicted value.
Preferably, in the step S1, the collected historical vehicle speed data of the road section is stored in the form of a quadruple, and the quadruple is expressed in the form of q= (r, d, t, S), where Q represents the historical vehicle speed data of the road section, r represents the road section monitored by the sensor, d represents the date monitored by the sensor, t represents the time window of the day monitored by the sensor, and S represents the average vehicle speed detected by the sensor in the t time window of the d-th day of the r-th road section.
Preferably, in the step S2, the method for constructing the average vehicle speed tensor includes:
s2-1: dividing the quaternion q= (r, d, t, s) into K time slices according to the time window t, and applying the quaternion q= (r, d,1, s) with t=1 to obtain Q (1) To construct an I J slice matrix A (1) Wherein I is the number of road sections monitored by the sensor, and J is the number of days recorded by the sensor; analogically, construct slice matrix A (2) 、A (3) 、...、A (K) ;
S2-2: sequentially from front to back in three-dimensional space according to divided K time slices by using K slice matrixesPost-permutation construction of average vehicle speed tensor S epsilon R I×J×K Wherein R represents a real set; the set of all data components of the average vehicle speed tensor S is represented by the known data set Λ.
Preferably, the S3 includes:
s3-1: initializing process parameters involved in the average vehicle speed prediction process;
s3-2: constructing an average vehicle speed prediction target loss function according to the average vehicle speed tensor and the process parameters;
s3-3: performing iterative optimization on the average vehicle speed prediction target loss function;
s3-4: judging whether the iteration process of the target loss function L (phi) reaches an iteration termination condition, if so, terminating, and if not, continuing iteration;
s3-5: and calculating a road section average vehicle speed predicted value.
Preferably, the process parameters in S3-1 include: road section average speed tensor S, hidden characteristic matrix R, D, T and corresponding auxiliary variable matrixNon-negative constraint parameters of Lagrangian submatrices corresponding to three latent feature matricesρ jp 、ψ kp Learning penalty parameter tau corresponding to three hidden feature matrices i 、ν j 、ω k The hidden feature matrix dimension P finally converges the threshold tau;
the size of the hidden feature matrix R, D, T is determined by each dimension value of the road section average speed tensor S and the hidden feature matrix dimension P, namely, the hidden feature matrix with R being I rows and P columns, the hidden feature matrix with P being J rows and P columns and the hidden feature matrix with T being K rows and P columns;
auxiliary variable matrixRespectively an I row, a P column, a J row, a P column and a K row, a P column.
Preferably, in S3-2, the constructed average vehicle speed prediction target loss function is:
in formula (1), (i, j, k) ∈Λ represents the subscript of the known element contained in tensor S; s is S ijk Representing the known entity in the road section average speed tensor S, namely the average speed value of the kth time window on the jth day of the road section monitored by the ith sensor;representing the auxiliary variable matrix +.>The value of row i and column p->Representing the auxiliary variable matrix +.>The value of the j-th row and p-th column in (a)>Representing the auxiliary variable matrix +.>The value of row k and column p; gamma represents a cauchy loss control parameter; r is (r) ip A value representing the ith row and the ith column in the hidden characteristic matrix R; d, d jp A value representing the jth row and p-th column in the hidden feature matrix D; t is t kp A value representing the kth row and p-th column in the latent feature matrix T; />ρ jp 、ψ kp Respectively representing non-negative constraint parameters of Lagrangian submatrices corresponding to the latent feature matrix R, D, T; τ i 、ν j 、ω k Respectively represent the learning corresponding to the latent feature matrix R, D, TPenalty parameters; i represents the number of road sections monitored by the sensor; j represents the number of days of sensor monitoring record; k represents the number of slice matrices; p represents the dimension of the latent feature matrix.
Preferably, in S3-3, the formula of iterative optimization is:
in the formula (2),wherein (1)>Representing an average vehicle speed predicted value calculated according to the hidden characteristic matrix; /> Representing the auxiliary variable matrix +.>The value of row i and column p,/>representing the auxiliary variable matrix +.>The value of row j and column p->Representing the auxiliary variable matrix +.>The value of row k and column p, q represents a positive constant.
Preferably, in S3-5, the calculation formula of the average vehicle speed predicted value is:
in the formula (3),representing the average vehicle speed predicted value.
The invention also provides an average vehicle speed prediction device based on tensor robust decomposition, which is used for executing an average vehicle speed prediction method and comprises a data receiving module, a data storage module, a tensor construction module and a prediction module;
the data receiving module is used for collecting road section historical vehicle speed data from the sensor;
the data storage module is used for storing the collected historical road section speed data and the average speed predicted value output by the prediction module;
the tensor construction module is used for constructing an average vehicle speed tensor according to the road section historical vehicle speed data;
and the prediction module is used for constructing an objective function according to the average vehicle speed tensor, performing iterative optimization and outputting an average vehicle speed predicted value.
The present invention also provides a storage medium having stored thereon a computer program which, when run on a processor, implements the steps of a tensor-robust decomposition based average vehicle speed prediction method.
In summary, due to the adoption of the technical scheme, compared with the prior art, the invention has at least the following beneficial effects:
according to the invention, the non-negative tensor decomposition model is constructed by utilizing the historical vehicle speed data, and the objective function (cauchy loss) is constructed to measure the difference between the observed average vehicle speed value and the predicted value, so that the interference of the outlier on the feature mining is effectively reduced, and the prediction accuracy is improved. Meanwhile, an objective function is updated by adopting an alternate direction multiplier method, so that higher calculation efficiency is obtained.
The method is specially applied to the road section average speed data in the intelligent traffic data, can be used for carrying out road section average time prediction without being influenced by outliers and maintaining high accuracy at the same time, so as to solve the problem of dynamic road section average speed prediction aiming at periodic time sequence information, and can be widely applied to the fields of computer service, traffic and the like.
Description of the drawings:
fig. 1 is a schematic diagram of an average vehicle speed prediction method based on tensor robust decomposition according to an exemplary embodiment of the present invention.
Fig. 2 is a schematic diagram of an average vehicle speed prediction apparatus based on tensor robust decomposition according to an exemplary embodiment of the present invention.
Fig. 3 is a schematic diagram of a data storage module structure according to an exemplary embodiment of the present invention.
Fig. 4 is a schematic diagram of a prediction module structure according to an exemplary embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to examples and embodiments. It should not be construed that the scope of the above subject matter of the present invention is limited to the following embodiments, and all techniques realized based on the present invention are within the scope of the present invention.
In the description of the present invention, it should be understood that the terms "longitudinal," "transverse," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," and the like indicate orientations or positional relationships based on the orientation or positional relationships shown in the drawings, merely to facilitate describing the present invention and simplify the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and therefore should not be construed as limiting the present invention.
As shown in fig. 1, the invention provides an average vehicle speed prediction method based on tensor robust decomposition, which specifically comprises the following steps:
s1: and receiving an instruction for predicting the historical average speed of the road section, which is sent by the server, and acquiring and storing the historical speed data of the road section from the sensor.
In this embodiment, the form of the server sending the instruction is various, such as periodically sending a prediction instruction or sending an instruction according to the notification of the device.
In this embodiment, the received road section historical vehicle speed data is stored in the form of a quadruple, where the quadruple is represented by q= (r, d, t, s), where Q represents road section historical vehicle speed data, r represents road section monitored by the sensor, d represents date monitored by the sensor, t represents time window of day monitored by the sensor, and s represents average vehicle speed detected by the sensor in t time window of day d of the r-th road section.
S2: constructing an average vehicle speed tensor S epsilon R according to stored road section historical vehicle speed data I×J×K I represents the number of road segments monitored by the sensor, J represents the number of days the sensor monitors were recorded, and K represents the number of slice matrices.
S2-1: and dividing the quadruple Q= (r, d, t, s) into K time slices according to the time window t, and constructing a corresponding slice matrix.
For example, Q can be obtained by using the quadruple q= (r, d,1, s) with t=1 (1) To construct an I J slice matrix A (1) Where I is the number of road segments monitored by the sensor and J is the number of days recorded by the sensor monitoring. Analogize to construct the slice matrix A (2) 、A (3) 、...、A (K) 。
S2-2: k slice matrixes are arranged in a three-dimensional space to construct an average vehicle speed tensor S epsilon R I×J×K (R represents a real number set). Since the average vehicle speed of each road segment sensor in the time segment of the return part of a specific day is positive, the constructed average vehicle speed tensor S is a nonnegative sparse tensor, and the known data set Λ is used for representing the set formed by all data (including the number I of the road segments monitored by the sensor, the number J of the days of the monitoring records of the sensor, the number K of the slice matrixes, the road segments r monitored by the sensor, the date d monitored by the sensor, the time window t in the day of the monitoring of the sensor and the average vehicle speed S detected by the sensor in the t time window of the d-th day of the r-th road segment).
S3: and constructing an average vehicle speed prediction target loss function according to the average vehicle speed tensor, performing iterative optimization, and outputting an average vehicle speed predicted value.
S3-1: process parameters involved in the average vehicle speed prediction process are initialized.
The process parameters comprise a road section average speed tensor S, a hidden characteristic matrix R, D, T and a corresponding auxiliary variable matrixNon-negative constraint parameter ++of Lagrangian submatrix corresponding to three hidden feature matrices>ρ jp 、ψ kp Learning penalty parameter tau corresponding to three hidden feature matrices i 、ν j 、ω k The feature matrix dimension P is hidden, and the threshold tau is finally converged (10-5).
The size of the hidden feature matrix R, D, T is determined by each dimension value of the road section average speed tensor S and the hidden feature matrix dimension P, namely, the hidden feature matrix with R being I rows and P columns, the hidden feature matrix with P being J rows and P columns and the hidden feature matrix with T being K rows and P columns.
Auxiliary variable matrixRespectively is I rowA matrix of P columns, J rows P columns and K rows P columns.
S3-2: and constructing an average vehicle speed prediction target loss function according to the average vehicle speed tensor and the process parameters.
In the present embodiment, cauchy loss is usedConstruction of an augmented Lagrangian function L (phi) (i.e. target loss function) optimization as an optimization objective, introducing the auxiliary variable +.> As a decision parameter to separate the non-negative constraints from the learning process:
in formula (1), (i, j, k) ∈Λ represents the subscript of the known element contained in tensor S; s is S ijk Representing the known entity in the road section average speed tensor S, namely the average speed value of the kth time window on the jth day of the road section monitored by the ith sensor;representing the auxiliary variable matrix +.>The value of row i and column p->Representing the auxiliary variable matrix +.>The value of the j-th row and p-th column in (a)>Representing auxiliary variable momentMatrix->The value of row k and column p; gamma represents a cauchy loss control parameter; r is (r) ip A value representing the ith row and the ith column in the hidden characteristic matrix R; d, d jp A value representing the jth row and p-th column in the hidden feature matrix D; t is t kp A value representing the kth row and p-th column in the latent feature matrix T; />ρ jp 、ψ kp Respectively representing non-negative constraint parameters of Lagrangian submatrices corresponding to the latent feature matrix R, D, T; τ i 、ν j 、ω k Respectively representing learning penalty parameters corresponding to the hidden characteristic matrix R, D, T; i represents the number of road sections monitored by the sensor; j represents the number of days of sensor monitoring record; k represents the number of slice matrices; p represents the dimension of the latent feature matrix.
S3-3: and (3) adopting an alternate learning strategy, namely fixing other variables in the objective loss function L (phi), carrying out iterative optimization on the auxiliary variables so as to minimize the values, wherein the training iterative formula is as follows:
in the formula (2),wherein (1)>Representing an average vehicle speed predicted value calculated according to the hidden characteristic matrix; /> Representing the auxiliary variable matrix +.>The value of row i and column q->Representing the auxiliary variable matrix +.>The value of row j and column q ∈j>Representing the auxiliary variable matrix +.>The value of the kth row and the qth column, q being a positive constant.
S3-4: judging whether the iteration process of the target loss function L (phi) reaches an iteration termination condition, if so, terminating, and if not, continuing iteration.
In this embodiment, the iteration termination conditions include: the number of training iteration rounds reaches the maximum number of training iteration rounds, or the absolute value of the difference between the value of the target loss function calculated after the iteration of the round is finished and the value of the target loss function of the previous round is smaller than the final convergence threshold tau.
S3-5: calculating a road section average vehicle speed predicted value, which specifically comprises the following steps:
after the iteration of the objective loss function is finished, the hidden characteristic matrix R, D, T obtained by training when the objective loss function reaches the minimum value is utilized to calculate the average vehicle speed predicted value detected by the sensor with highest calculation accuracy in the t time window of the road section r and the d dayThe calculation formula is as follows:
in the formula (3),representing the average vehicle speed predicted value.
The method is specially applied to the average road section speed data in intelligent traffic data, and can prevent the method from being influenced by outliers when predicting the average vehicle speed by adopting a Cauchy loss structure target loss function, further uses tensor description to acquire historical vehicle speed data so as to save space-time information contained in the vehicle speed data, and adopts an alternate direction multiplier method to optimize the target function so as to maintain high-accuracy average road section time prediction, thereby solving the problem of dynamic average road section vehicle speed prediction aiming at periodic time sequence information.
Based on the method, as shown in fig. 2, the invention also provides an average vehicle speed prediction device based on tensor robust decomposition, which comprises a data receiving module, a data storage module, a tensor construction module and a prediction module.
The output end of the data receiving module is connected with the first input end of the data storage module, the output end of the data storage module is connected with the input end of the prediction module, and the output end of the prediction module is connected with the second input end of the data storage module.
The data receiving module is used for collecting road section historical vehicle speed data from the sensor;
the data storage module is used for storing the collected historical road section speed data and the average speed predicted value output by the prediction module;
the tensor construction module is used for constructing an average vehicle speed tensor according to the road section historical vehicle speed data;
and the prediction module is used for constructing an objective function according to the average vehicle speed tensor, performing iterative optimization and outputting an average vehicle speed predicted value.
In this embodiment, as shown in fig. 3, the data storage module includes a first storage unit and a second storage unit.
The first storage unit is used for storing the received road section historical vehicle speed data in the form of four-tuple, wherein the four-tuple is represented by Q= (r, d, t, s), Q represents road section historical vehicle speed data, r represents road sections monitored by the sensor, d represents the date monitored by the sensor, t represents a time window in one day monitored by the sensor, and s represents the average vehicle speed detected by the sensor in a t time window in the d-th day of the r-th road section.
And the second storage unit is used for storing the average vehicle speed predicted value output by the prediction module, and meanwhile, the average vehicle speed predicted value is also stored in a four-element mode.
In this embodiment, as shown in fig. 4, the prediction module includes an initialization unit, a training unit, and a calculation unit;
the initialization unit is used for initializing process parameters and corresponding auxiliary variables involved in the process of predicting the average road section vehicle speed;
the training unit is used for constructing an average vehicle speed prediction objective function by combining the constructed average vehicle speed tensor and the initialized process parameters and carrying out iterative training;
and the calculating unit is used for calculating the average vehicle speed predicted value according to the iteration parameter of the objective function.
The device can be deployed in an existing server or a separately arranged server special for predicting the average speed of the road section.
The present invention also provides an electronic device comprising a processor for running a computer program stored in a memory, to cause the electronic device to implement the steps of the average vehicle speed prediction method based on tensor robust decomposition in the above embodiment.
The present invention also provides a computer readable storage medium having stored therein a computer program which when run on a processor implements the steps of the average vehicle speed prediction method based on tensor robust decomposition in the above embodiments.
The computer program comprises computer program code which may be in source code form, object code form, executable file or in some intermediate form, etc. The computer readable medium may include at least: any entity or device capable of carrying computer program code to an electronic device, a recording medium, a computer Memory, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), an electrical carrier signal, a telecommunications signal, and a software distribution medium. Such as a U-disk, removable hard disk, magnetic or optical disk, etc. In some jurisdictions, computer readable media may not be electrical carrier signals and telecommunications signals in accordance with legislation and patent practice.
It will be understood by those of ordinary skill in the art that the foregoing embodiments are specific examples of carrying out the invention and that various changes in form and details may be made therein without departing from the spirit and scope of the invention.
Claims (6)
1. The average vehicle speed prediction method based on tensor robust decomposition is characterized by comprising the following steps of:
s1: receiving an instruction for predicting the historical average speed of the road section sent by a server, and acquiring and storing historical speed data of the road section from a sensor;
in the step S1, the collected historical vehicle speed data of the road section is stored in the form of a quadruple, wherein the quadruple is represented by q= (r, d, t, S), Q represents the historical vehicle speed data of the road section, r represents the road section monitored by the sensor, d represents the date monitored by the sensor, t represents the time window of the sensor during the day monitored by the sensor, and S represents the average vehicle speed detected by the sensor in the t time window of the d-th day of the r-th road section;
s2: constructing an average vehicle speed tensor according to stored road section historical vehicle speed data;
s3: constructing an average vehicle speed prediction target loss function according to the average vehicle speed tensor, performing iterative optimization, and outputting an average vehicle speed predicted value;
the step S3 comprises the following steps:
s3-1: initializing process parameters involved in the average vehicle speed prediction process;
s3-2: constructing an average vehicle speed prediction target loss function according to the average vehicle speed tensor and the process parameters;
in the step S3-2, the constructed average vehicle speed prediction target loss function is as follows:
in formula (1), (i, j, k) ∈Λ represents the subscript of the known element contained in tensor S; s is S ijk Representing the known entity in the road section average speed tensor S, namely the average speed value of the kth time window on the jth day of the road section monitored by the ith sensor;representing the auxiliary variable matrix +.>The value of row i and column p->Representing the auxiliary variable matrix +.>The value of the j-th row and p-th column in (a)>Representation assisted transformationQuantity matrix->The value of row k and column p; gamma represents a cauchy loss control parameter; r is (r) ip A value representing the ith row and the ith column in the hidden characteristic matrix R; d, d jp A value representing the jth row and p-th column in the hidden feature matrix D; t is t kp A value representing the kth row and p-th column in the latent feature matrix T; />ρ jp 、ψ kp Respectively representing non-negative constraint parameters of Lagrangian submatrices corresponding to the latent feature matrix R, D, T; τ i 、ν j 、ω k Respectively representing learning penalty parameters corresponding to the hidden characteristic matrix R, D, T; i represents the number of road sections monitored by the sensor; j represents the number of days of sensor monitoring record; k represents the number of slice matrices; p represents the dimension of the hidden feature matrix;
s3-3: performing iterative optimization on the average vehicle speed prediction target loss function;
in the step S3-3, the formula of iterative optimization is as follows:
in the formula (2),wherein (1)>Representing an average vehicle speed predicted value calculated according to the hidden characteristic matrix; />Representing the auxiliary variable matrix +.>The value of row i and column p->Representing the auxiliary variable matrix +.>The value of row j and column p->Representing the auxiliary variable matrix +.>The value of the kth row and the kth column, q represents a positive constant;
s3-4: judging whether the iteration process of the target loss function L (phi) reaches an iteration termination condition, if so, terminating, and if not, continuing iteration;
s3-5: and calculating a road section average vehicle speed predicted value.
2. The method for predicting the average vehicle speed based on tensor robust decomposition according to claim 1, wherein in S2, the method for constructing the average vehicle speed tensor is as follows:
s2-1: dividing the quaternion q= (r, d, t, s) into K time slices according to the time window t, and applying the quaternion q= (r, d,1, s) with t=1 to obtain Q (1) To construct an I J slice matrix A (1) Wherein I is the number of road sections monitored by the sensor, and J is the number of days recorded by the sensor; analogically, construct slice matrix A (2) 、A (3) 、...、A (K) ;
S2-2: the K slice matrixes are used for sequentially arranging and constructing an average vehicle speed tensor S epsilon R from front to back in a three-dimensional space according to the divided K time slices I×J×K Wherein R represents a real set; the set of all data components of the average vehicle speed tensor S is represented by the known data set Λ.
3. The method for predicting average vehicle speed based on tensor robust decomposition of claim 1, wherein said process parameters in S3-1 include: road section average speed tensor S, hidden characteristic matrix R, D, T and corresponding auxiliary variable matrixNon-negative constraint parameter ++of Lagrangian submatrix corresponding to three hidden feature matrices>ρ jp 、ψ kp Learning penalty parameter tau corresponding to three hidden feature matrices i 、ν j 、ω k The hidden feature matrix dimension P finally converges the threshold tau;
the size of the hidden feature matrix R, D, T is determined by each dimension value of the road section average speed tensor S and the hidden feature matrix dimension P, namely, the hidden feature matrix with R being I rows and P columns, the hidden feature matrix with P being J rows and P columns and the hidden feature matrix with T being K rows and P columns;
auxiliary variable matrixRespectively I row and P column,A matrix of J rows P columns and K rows P columns.
4. The method for predicting an average vehicle speed based on tensor robust decomposition according to claim 1, wherein in S3-5, the calculation formula of the average vehicle speed predicted value is:
in the formula (3),representing the average vehicle speed predicted value.
5. Average vehicle speed prediction device based on tensor robust decomposition, for performing the average vehicle speed prediction method according to any one of claims 1-4, characterized by comprising a data receiving module, a data storage module, a tensor construction module and a prediction module;
the data receiving module is used for collecting road section historical vehicle speed data from the sensor;
the data storage module is used for storing the collected historical road section speed data and the average speed predicted value output by the prediction module;
the tensor construction module is used for constructing an average vehicle speed tensor according to the road section historical vehicle speed data;
and the prediction module is used for constructing an objective function according to the average vehicle speed tensor, performing iterative optimization and outputting an average vehicle speed predicted value.
6. A storage medium having stored thereon a computer program which, when run on a processor, implements the steps of the tensor-robust decomposition based average vehicle speed prediction method of any of claims 1-4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310999172.3A CN116913104B (en) | 2023-08-09 | 2023-08-09 | Average vehicle speed prediction method, device and storage medium based on tensor robust decomposition |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310999172.3A CN116913104B (en) | 2023-08-09 | 2023-08-09 | Average vehicle speed prediction method, device and storage medium based on tensor robust decomposition |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116913104A CN116913104A (en) | 2023-10-20 |
CN116913104B true CN116913104B (en) | 2024-03-19 |
Family
ID=88364872
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310999172.3A Active CN116913104B (en) | 2023-08-09 | 2023-08-09 | Average vehicle speed prediction method, device and storage medium based on tensor robust decomposition |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116913104B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110717627A (en) * | 2019-09-29 | 2020-01-21 | 浙江大学 | Full traffic prediction method based on dual graph framework |
CN112734100A (en) * | 2020-12-31 | 2021-04-30 | 北京航空航天大学 | Road network travel time prediction method based on tensor neural network |
CN113971885A (en) * | 2020-07-06 | 2022-01-25 | 华为技术有限公司 | Vehicle speed prediction method, device and system |
-
2023
- 2023-08-09 CN CN202310999172.3A patent/CN116913104B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110717627A (en) * | 2019-09-29 | 2020-01-21 | 浙江大学 | Full traffic prediction method based on dual graph framework |
CN113971885A (en) * | 2020-07-06 | 2022-01-25 | 华为技术有限公司 | Vehicle speed prediction method, device and system |
CN112734100A (en) * | 2020-12-31 | 2021-04-30 | 北京航空航天大学 | Road network travel time prediction method based on tensor neural network |
Also Published As
Publication number | Publication date |
---|---|
CN116913104A (en) | 2023-10-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113051822A (en) | Industrial system anomaly detection method based on graph attention network and LSTM automatic coding model | |
CN114445461A (en) | Visible light infrared target tracking training method and device based on non-paired data | |
CN116957167B (en) | Method and device for predicting charging duration of electric vehicle charging pile | |
CN116913104B (en) | Average vehicle speed prediction method, device and storage medium based on tensor robust decomposition | |
CN113361194B (en) | Sensor drift calibration method based on deep learning, electronic equipment and storage medium | |
US20220027727A1 (en) | Online training of neural networks | |
CN116861262B (en) | Perception model training method and device, electronic equipment and storage medium | |
CN114819325A (en) | Score prediction method, system, device and storage medium based on graph neural network | |
CN115169721B (en) | Method and system for predicting single ton energy consumption in rectification process based on migration identification | |
CN114399901B (en) | Method and equipment for controlling traffic system | |
CN115731430A (en) | Method for generating deep learning image data set by finite element model and server | |
CN112668797B (en) | Long-short-period traffic prediction method | |
CN114841461A (en) | Air quality integration prediction method based on time sequence missing perception and multi-source factor fusion | |
CN111028245B (en) | Multi-mode composite high-definition high-speed video background modeling method | |
CN116934603A (en) | Logging curve missing segment completion method and device, storage medium and electronic equipment | |
CN114461619A (en) | Energy internet multi-source data fusion method and device, terminal and storage medium | |
CN117456738B (en) | Expressway traffic volume prediction method based on ETC portal data | |
CN113095328A (en) | Self-training-based semantic segmentation method guided by Gini index | |
CN117235508B (en) | Vehicle fuel consumption prediction method and system based on big data | |
EP4280172A1 (en) | Vehicle fault diagnosis method, device, and computer-readable storage medium | |
EP3975063A1 (en) | Device for and computer implemented method of machine learning | |
CN113538536B (en) | SAR image information-assisted remote sensing optical image dense cloud detection method and system | |
CN117349770B (en) | Structural health monitoring multi-strain sensor data anomaly detection and repair method | |
CN114330089B (en) | Rare earth element content change prediction method and system | |
Bolz | Multistep prediction of vehicle states using transformers |
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 | ||
GR01 | Patent grant |