CN111222542B - Based on L 1 Regularized effective characteristic selection method for working conditions of hybrid power bus - Google Patents
Based on L 1 Regularized effective characteristic selection method for working conditions of hybrid power bus Download PDFInfo
- Publication number
- CN111222542B CN111222542B CN201911298772.7A CN201911298772A CN111222542B CN 111222542 B CN111222542 B CN 111222542B CN 201911298772 A CN201911298772 A CN 201911298772A CN 111222542 B CN111222542 B CN 111222542B
- Authority
- CN
- China
- Prior art keywords
- working condition
- speed
- characteristic parameters
- segment
- beta
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/211—Selection of the most significant subset of features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
Abstract
The invention relates to a method based on L 1 The regularized effective characteristic selection method for the working condition of the hybrid power bus comprises the following steps: s1, acquiring working condition data of a hybrid bus, and acquiring speed working condition data of the hybrid bus corresponding to each time point; s2, dividing speed working condition data of the hybrid bus into N segments; s3, extracting M characteristic parameters in each segment, and recording as F 1 ,F 2 ,…,F M Wherein, the characteristic parameters comprise average speed, maximum speed, minimum speed, maximum acceleration, minimum acceleration and idle time of the segment; s4, classifying the N working condition fragments by using the extracted characteristic parameters to obtain the traffic fluency of different working condition types, and further obtaining a sample: (F) 1i ,F 2i ,…,F Mi ,y i ) I=1, …, N wherein y i The traffic fluency of the working condition type corresponding to the ith speed working condition segment; s5, obtaining effective characteristic parameters of the working condition by solving the characteristic extraction model.
Description
Technical Field
The invention relates to a method based on L 1 A method for selecting effective characteristics of working conditions of a hybrid power bus by a regularized linear classifier relates to the technical field of hybrid power buses.
Background
With the increasing severity of petroleum crisis and urban air pollution, hybrid buses are also widely used in the public transportation field. Since hybrid systems have multiple modes, how to control the energy flow in these modes to achieve optimal fuel and electric consumption is an important area of research.
Fuel consumption and electricity consumption are heavily dependent on the type of operating conditions of the vehicle, so how to identify and predict vehicle operating conditions is a challenging task. In order to effectively study vehicle operating conditions, it is necessary to extract effective features from the operating condition data to represent the type of operating condition. The traditional method for extracting the effective characteristics of the working conditions mainly comprises principal component analysis and a neural network, and the technical problem of the method is that the physical meaning of the extracted effective characteristics is difficult to explain.
Disclosure of Invention
For the aboveProblem, it is an object of the present invention to provide an L-based system that can efficiently find effective operating characteristics 1 A method for selecting effective characteristics of working conditions of a hybrid bus by using a normalized linear classifier.
In order to achieve the above purpose, the present invention adopts the following technical scheme: l-based 1 The regularized effective characteristic selection method for the working condition of the hybrid power bus comprises the following steps:
s1, acquiring working condition data of a hybrid bus, and acquiring speed working condition data of the hybrid bus corresponding to each time point;
s2, dividing speed working condition data of the hybrid bus into N segments;
s3, extracting M characteristic parameters in each segment, and recording as F 1 ,F 2 ,…,F M Wherein, the characteristic parameters comprise average speed, maximum speed, minimum speed, maximum acceleration, minimum acceleration and idle time of the segment;
s4, classifying the N working condition fragments by using the extracted characteristic parameters to obtain traffic smoothness of different working condition types, and further obtaining a sample:
(F 1i ,F 2i ,…,F Mi ,y i ),i=1,…,N
wherein y is i The traffic fluency of the working condition type corresponding to the ith speed working condition segment;
s5, obtaining effective characteristic parameters of the working condition by solving the characteristic extraction model.
Further, the specific process of S5 is as follows:
s51, establishing a feature extraction model:
wherein beta is 0 、β j Are all coefficients to be determined, lambda is a penalty parameter;
s52, determining input and output of a model:
model input: sample (F) 1i ,F 2i ,…,F Mi ,y i ) A fault tolerance constant epsilon;
model output: a non-zero component in beta;
s53, will (F) 1i ,F 2i ,…,F Mi ) Normalization, initial beta extraction 1 =β 2 =…=β M =0,y represents the fluency combination vector of the working condition, +.>An average value representing the fluency of the working condition type;
s54, by solvingDetermining the sum r 0 F most relevant j And defines an active set a= { j } and a vector F j Matrix F formed A ;
S55, selecting effective characteristic parameters for representing the working condition information.
Further, the specific process of S55 is as follows:
s551, defining least square directionThe current least squares direction is obtained, where k=1, 2, …, k=min (N-1, m), r k-1 Is the residual, for the M-dimensional vector delta, let delta be A =δ, the other elements are all 0;
s552, in delta direction, from beta k-1 Initially, the motion coefficients β are solved towards their least squares, yielding a new F A And a new residual r (lambda), wherein,
F A :β(λ)=β k-1 +(λ k-1 -λ)Δ,0≤λ≤λ k-1
r(λ)=y-Fβ(λ)=r k-1 -(λ k-1 -λ)FΔ
s553, focusing on λ= | < F l R (λ) > | whereLet j' th variable be such that lambda reaches a maximum, i.e. < F j R (λ) > |=λ, j should be added to a, and λ is given by k =λ, expanding the range of the active set;
s554, let a=a { j }, β k =β(λ)=β k-1 +(λ k-1 -λ k )Δ,r k =y-Fβ k Updating the cycle;
s555, return sequenceTruncating a constant beta equal to zero in the optimal solution j Let j=1, 2, …, L, the remainder β L+1 ,β L+2 ,…,β M Corresponding feature F (L+1) ,F (L+2) ,…,F M Is an effective characteristic parameter of the working condition.
The invention adopts the technical proposal and has the following characteristics: because the invention is a statistical method based on samples, and effective working condition characteristics can be efficiently found out from a plurality of characteristics for representing the working conditions of the vehicle, the invention has great advantages in terms of time complexity and space complexity of an algorithm.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described in conjunction with the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
L-based provided in the present embodiment 1 The method for selecting the effective characteristics of the working conditions of the hybrid power bus of the regularized linear classifier comprises the following steps:
s1, collecting working condition data of vehicles in a hybrid bus which is actually operated to obtain speed working condition data of the hybrid bus corresponding to each time point, wherein the speed working condition data in the embodiment refer to the speed data of the vehicles at each time point.
S2, dividing the whole speed working condition data of the hybrid bus into N segments, wherein the time interval of each segment can be 128S, for example, and the method is not limited to the above.
S3, extracting characteristic parameters in each segment, namely extracting the same characteristic parameters from each segment, wherein the extracted characteristic parameters comprise the average speed, the maximum speed, the minimum speed, the maximum acceleration, the minimum acceleration, the idle time (the idle time is the idle time of the engine) and the like of the segment, and the embodiment assumes that M characteristic parameters are extracted and is marked as F 1 ,F 2 ,…,F M 。
S4, classifying the N working condition fragments by using the extracted characteristic parameters to obtain traffic smoothness of different working condition types, wherein the traffic smoothness D epsilon [ a, b ], wherein D=a represents the condition that the traffic smoothness is the lowest, namely the most congested condition; d=b represents the highest traffic fluency, and further a sample is obtained:
(F 1i ,F 2i ,…,F Mi ,y i ),i=1,…,N
wherein y is i Is the traffic fluency of the working condition type corresponding to the ith speed working condition segment.
S5, obtaining effective characteristic parameters of different working conditions by solving a characteristic extraction model, wherein the specific process is as follows:
s51, establishing a feature extraction model:
wherein beta is 0 、β j Are undetermined coefficients, lambda is penalty parameter, F in order to give the relation between the characteristic and the working condition type ji Representing the components of the feature vector.
S52, determining the input and output of the feature extraction model
Model input: training sample (F) 1i ,F 2i ,…,F Mi ,y i ) And a fault tolerance constant epsilon=0.05.
Model output: a non-zero component in beta.
S53, will (F) 1i ,F 2i ,…,F Mi ) Normalizing to obtain average value of 0, norm of 1, and initial taking beta 1 =β 2 =…=β M =0,y represents the fluency combination vector of the condition (i.e., y in S52 i Formed vector), ->An average value representing the fluency of the operating mode type, wherein the average value is subtracted from each characteristic value and then divided by the standard deviation;
s54, by solvingDetermining the sum r 0 F most relevant j And defines an active set a= { j } and a vector F j Matrix F formed A Wherein F is j Representing the j-th feature vector.
S55, selecting the effective characteristic of the characteristic working condition information, wherein the specific process is as follows:
s551, defining least square directionThe current least squares direction is obtained, where k=1, 2, …, k=min (N-1, m), F A Is a matrix formed by vectors corresponding to indexes of the active set, r k-1 Is the residual, for the M-dimensional vector delta, let delta be A =δ, the other elements are all 0.
S552, in delta direction, from beta k-1 Initially, the motion coefficients β are solved towards their least squares, yielding a new F A Wherein, the method comprises the steps of, wherein,
F A :β(λ)=β k-1 +(λ k-1 -λ)Δ,0≤λ≤λ k-1 。
further, a new residual r (λ) =y-fβ (λ) =r is obtained k-1 -(λ k-1 -lambda) Fdelta, wherein F is a matrix, the index A part of which is equal to F A The other part is equal to 0.
S553, focusing on λ= | < F l R (λ) > | whereLet j variables maximize λ, i.e. | < F j R (λ) > |=λ, j should be added to a, and λ is given by k =λ, expanding the range of the active set.
S554, let a=a { j }, β k =β(λ)=β k-1 +(λ k-1 -λ k )Δ,r k =y-Fβ k The cycle is updated.
S555, return sequenceTruncating a constant beta equal to zero in the optimal solution j Here, j=1, 2, …, L is not limited, and then, the remaining β L+1 ,β L+2 ,…,β M Corresponding feature F (L+1) ,F (L+2) ,…,F M Is an effective feature of the working condition.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the scope of protection thereof, although the present application is described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: various alterations, modifications, and equivalents may be suggested to the detailed description of the invention as described herein, which would occur to persons skilled in the art upon reading the disclosure and are intended to be within the scope of the appended claims.
Claims (1)
1. L-based 1 The regularized effective characteristic selection method for the working condition of the hybrid power bus is characterized by comprising the following steps:
s1, acquiring working condition data of a hybrid bus, and acquiring speed working condition data of the hybrid bus corresponding to each time point;
s2, dividing speed working condition data of the hybrid bus into N segments;
s3, extracting M characteristic parameters in each segment, and recording as F 1 ,F 2 ,…,F M Wherein, the characteristic parameters comprise average speed, maximum speed, minimum speed, maximum acceleration, minimum acceleration and idle time of the segment;
s4, classifying the N working condition fragments by using the extracted characteristic parameters to obtain the traffic fluency of different working condition types, and further obtaining a sample:
(F 1i ,F 2i ,…,F Mi ,y i ),i=1,…,N
wherein y is i The traffic fluency of the working condition type corresponding to the ith speed working condition segment;
s5, obtaining effective characteristic parameters of working conditions by solving a characteristic extraction model, wherein the specific process is as follows:
s51, establishing a feature extraction model:
wherein beta is 0 、β j Are all coefficients to be determined, lambda is a penalty parameter;
s52, determining input and output of a model:
model input: sample (F) 1i ,F 2i ,…,F Mi ,y i ) A fault tolerance constant epsilon;
model output: a non-zero component in beta;
s53, will (F) 1i ,F 2i ,…,F Mi ) Normalization, initial beta extraction 1 =β 2 =…=β M =0,y represents the fluency combination vector of the working condition, +.>Representing a class of operating conditionsAn average value of the type fluency;
s54, by solvingDetermining the sum r 0 F most relevant j And defines an active set a= { j } and a vector F j Matrix F formed A ;
S55, selecting characteristic parameters of effective characterization working condition information, wherein the specific process is as follows:
s551, defining least square directionThe current least squares direction is obtained, where k=1, 2, …, k=min (N-1, m), r k-1 Is the residual, for the M-dimensional vector delta, let delta be A =δ, the other elements are all 0;
s552, in delta direction, from beta k-1 Initially, the motion coefficients β are solved towards their least squares, yielding a new F A And a new residual r (lambda), wherein,
F A :β(λ)=β k-1 +(λ k-1 -λ)Δ,0≤λ≤λ k-1
r(λ)=y-Fβ(λ)=r k-1 -(λ k-1 -λ)FΔ;
s553, focusing on λ= | < F l R (λ) > | whereLet j' th variable be such that lambda reaches a maximum, i.e. < F j R (λ) > |=λ, j should be added to a, and λ is given by k =λ, expanding the range of the active set;
s554, let a=a { j }, β k =β(λ)=β k-1 +(λ k-1 -λ k )Δ,r k =y-Fβ k Updating the cycle;
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911298772.7A CN111222542B (en) | 2019-12-17 | 2019-12-17 | Based on L 1 Regularized effective characteristic selection method for working conditions of hybrid power bus |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911298772.7A CN111222542B (en) | 2019-12-17 | 2019-12-17 | Based on L 1 Regularized effective characteristic selection method for working conditions of hybrid power bus |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111222542A CN111222542A (en) | 2020-06-02 |
CN111222542B true CN111222542B (en) | 2023-06-20 |
Family
ID=70806136
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911298772.7A Active CN111222542B (en) | 2019-12-17 | 2019-12-17 | Based on L 1 Regularized effective characteristic selection method for working conditions of hybrid power bus |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111222542B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111724505A (en) * | 2020-06-24 | 2020-09-29 | 北汽福田汽车股份有限公司 | Method and device for constructing driving condition |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103606271A (en) * | 2013-11-27 | 2014-02-26 | 大连理工大学 | Method for controlling hybrid power urban buses |
CN105046070A (en) * | 2015-07-07 | 2015-11-11 | 吉林大学 | Method for constructing city comprehensive working condition with turning performance |
US9804603B1 (en) * | 2015-03-18 | 2017-10-31 | Ag Leader Technology | Curved path approximation in vehicle guidance systems and methods |
CN108569297A (en) * | 2017-03-14 | 2018-09-25 | 厦门雅迅网络股份有限公司 | A kind of recognition methods of vehicle driving-cycle and system |
CN110254417A (en) * | 2019-06-27 | 2019-09-20 | 清华大学苏州汽车研究院(吴江) | Method for controlling hybrid power vehicle based on actual condition and the double identifications of driving style |
CN110298579A (en) * | 2019-06-26 | 2019-10-01 | 重庆大学 | A kind of construction method of new energy passenger car urban operating condition |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8164484B2 (en) * | 2007-10-03 | 2012-04-24 | University Of Southern California | Detection and classification of running vehicles based on acoustic signatures |
-
2019
- 2019-12-17 CN CN201911298772.7A patent/CN111222542B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103606271A (en) * | 2013-11-27 | 2014-02-26 | 大连理工大学 | Method for controlling hybrid power urban buses |
US9804603B1 (en) * | 2015-03-18 | 2017-10-31 | Ag Leader Technology | Curved path approximation in vehicle guidance systems and methods |
CN105046070A (en) * | 2015-07-07 | 2015-11-11 | 吉林大学 | Method for constructing city comprehensive working condition with turning performance |
CN108569297A (en) * | 2017-03-14 | 2018-09-25 | 厦门雅迅网络股份有限公司 | A kind of recognition methods of vehicle driving-cycle and system |
CN110298579A (en) * | 2019-06-26 | 2019-10-01 | 重庆大学 | A kind of construction method of new energy passenger car urban operating condition |
CN110254417A (en) * | 2019-06-27 | 2019-09-20 | 清华大学苏州汽车研究院(吴江) | Method for controlling hybrid power vehicle based on actual condition and the double identifications of driving style |
Non-Patent Citations (4)
Title |
---|
Liu Huawen, et al.Regularized partial least squares for multi-label learning.International Journal of Machine Learning and Cybernetics.2016,第335-346页. * |
Schmidt, Mark.Least squares optimization with L1-norm regularization.CS542B Project Report 504.2005,第195-221页. * |
朱华等.混合动力汽车行驶工况数据库的建立与分析.安徽工业大学学报(自然科学版).2019,第136-143页. * |
陈征等.基于进化-增强学习方法的插电式混合动力公交车能量管理策略.机械工程学报.2017,第86-93页. * |
Also Published As
Publication number | Publication date |
---|---|
CN111222542A (en) | 2020-06-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109766950B (en) | Industrial user short-term load prediction method based on morphological clustering and LightGBM | |
CN111038485A (en) | Hybrid electric vehicle control method and system based on driving style recognition | |
CN109934375B (en) | Power load prediction method | |
CN108569297B (en) | Vehicle driving condition identification method and system | |
CN109861211B (en) | Dynamic reconfiguration method of power distribution network based on data driving | |
CN108960586B (en) | Non-invasive load identification method adaptive to scene change | |
CN112052895A (en) | Pure electric vehicle driving style clustering method | |
CN105301508A (en) | Prediction method for electric automobile endurance mileage through redial basis function neural network | |
CN112766378B (en) | Cross-domain small sample image classification model method focusing on fine granularity recognition | |
CN115829105A (en) | Photovoltaic power prediction method based on historical data feature search | |
CN109376797B (en) | Network traffic classification method based on binary encoder and multi-hash table | |
CN114626487B (en) | Linear transformation relation checking method based on random forest classification algorithm | |
CN111681132A (en) | Typical power consumption mode extraction method suitable for massive class unbalanced load data | |
CN111222542B (en) | Based on L 1 Regularized effective characteristic selection method for working conditions of hybrid power bus | |
Hu et al. | Technology topic identification and trend prediction of new energy vehicle using LDA modeling | |
CN113627685B (en) | Wind driven generator power prediction method considering wind power internet load limit | |
CN109272058B (en) | Integrated power load curve clustering method | |
CN116596129A (en) | Electric vehicle charging station short-term load prediction model construction method | |
CN113997925B (en) | Energy management method for plug-in hybrid power system | |
CN117114160A (en) | Short-term photovoltaic power prediction method | |
CN114898136B (en) | Small sample image classification method based on characteristic self-adaption | |
CN116484299A (en) | Charging pile fault diagnosis method based on integration of gradient lifting tree and multi-layer perceptron | |
CN116307111A (en) | Reactive load prediction method based on K-means clustering and random forest algorithm | |
Wong et al. | A clustering-based method for fuzzy modeling | |
CN113419172A (en) | New energy automobile three-phase asynchronous motor fault identification based on GCN and VMD-ED |
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 |