CN115973173A - Pure electric vehicle load estimation method based on intelligent optimization algorithm - Google Patents
Pure electric vehicle load estimation method based on intelligent optimization algorithm Download PDFInfo
- Publication number
- CN115973173A CN115973173A CN202310024350.0A CN202310024350A CN115973173A CN 115973173 A CN115973173 A CN 115973173A CN 202310024350 A CN202310024350 A CN 202310024350A CN 115973173 A CN115973173 A CN 115973173A
- Authority
- CN
- China
- Prior art keywords
- data
- vehicle
- load estimation
- model
- optimization algorithm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- 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
Landscapes
- Electric Propulsion And Braking For Vehicles (AREA)
Abstract
The invention discloses a pure electric vehicle load estimation method based on an intelligent optimization algorithm, which is characterized in that vehicle dynamic balance equations of adjacent states of vehicles are inferred, then two specific driving states with relatively accurate estimation quality and relatively stable vehicle states are obtained by a data analysis method, boundary parameters of the two estimation states are optimized by the intelligent optimization algorithm, and vehicle loads are estimated by combining vehicle speed, acceleration and total driving torque information.
Description
Technical Field
The invention belongs to the technical field of intelligent networked automobiles, and particularly relates to a pure electric vehicle load estimation method based on an intelligent optimization algorithm.
Background
With the high-speed development of the intelligent internet automobile industry, related national departments develop a series of policy and regulations, and for vehicles and new energy automobiles with the current emission standard of national VI, monitoring facilities capable of monitoring the running state of the automobiles in real time and early warning the running fault information of the automobiles are required to be installed; for automobile enterprises, vehicles and fleet management companies, how to extract valuable information from a large amount of data provides a basis for after-sale service of products, improvement of products, fleet management of the vehicle management companies, vehicle user supervision and the like, and is a research hotspot in the technical field of current vehicle networking.
The application of the vehicle load estimation method is mainly divided into two aspects, on one hand, the vehicle load estimation method is used as a key vehicle parameter in a vehicle control algorithm or a control strategy, on the other hand, in order to improve vehicle performances such as vehicle safety, comfort, dynamic property, economy and the like, all current large vehicle enterprises introduce some advanced control algorithms, and the considered vehicle state is changed into a variable load state from a single fixed load state, so that in order to accurately control the vehicle, a more accurate vehicle load needs to be obtained; on the other hand, load estimation is applied to vehicle supervision of enterprises, vehicle management companies or individuals, vehicle load information is obtained, vehicle enterprises are used for vehicle after-sale service and product tracking, load data information is provided for after-sale vehicles, and the vehicle management companies can conveniently manage the vehicles or the fleet of vehicles, detect whether the vehicles are overloaded and the like. In conclusion, accurate load estimation has a great significance for vehicle performance improvement and vehicle management.
The current vehicle load estimation method mainly comprises two modes, one mode is that the load capacity sensor is installed and the deformation of a vehicle suspension is combined to directly measure, the method is high in cost, complex in installation mode, greatly interfered by the external environment state, large in scaling workload when being applied to large-scale vehicles of different models, and not beneficial to large-scale popularization and application of enterprises; the other method is that a vehicle longitudinal dynamics model is utilized, and load is pre-estimated in a mode of combining vehicle sensor information, a specific implementation mode can perform cloud offline analysis through vehicle networking data, or a vehicle-mounted controller performs real-time online analysis and the like, and application and deployment are simple and flexible; at present, the anti-interference capability of the environment is poor by combining a vehicle longitudinal dynamics model with load estimation models such as a Kalman filtering algorithm, a cyclic recursive least square algorithm and the like, because the maximum influence factor for load estimation in the dynamics model is a reverse road gradient factor, the gradient information of a running road is generally difficult to obtain when a vehicle runs, the load estimation error is large, the problems that the complexity of the model algorithm is high, the convergence speed of the algorithm is low and the like by a cyclic recursive method are solved, the estimated load quality and the actual quality have large access, and the actual requirement cannot be met.
The information disclosed in this background section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art.
Disclosure of Invention
The object of the present invention is to provide a solution to the above-mentioned drawbacks of the prior art.
In order to achieve the purpose, the invention provides a pure electric vehicle load estimation method based on an intelligent optimization algorithm.
The technical scheme for solving the problems is as follows: a pure electric vehicle load estimation method based on an intelligent optimization algorithm specifically comprises the following steps:
(1) Acquiring required vehicle data, wherein the data required for estimating the load of the pure electric vehicle is data information such as vehicle data sampling time (timestamp), driving speed, vehicle acceleration information, driving torque and the like;
(2) Preprocessing the required vehicle data, wherein the missing values, the abnormal values, the repeated values and the like are preprocessed;
(3) And establishing a dynamic balance equation of the vehicles in two adjacent states, analyzing and comparing the importance of the running resistance of each of the two adjacent states, and deducing a vehicle load estimation model.
(4) Carrying out load estimation on the obtained data by using the established load estimation model to obtain estimated vehicle mass, finding vehicle data which correspond to the estimated mass more accurately, analyzing the vehicle state, finding a driving state suitable for the load estimation model, and counting the data ranges of driving speed, acceleration and driving torque between adjacent moments;
(5) And establishing a constraint condition of the estimated state of the application load estimation model, and optimizing boundary parameters of the constraint conditions of two specific estimated states by using an intelligent optimization algorithm.
(6) Modifying the model parameters, verifying and analyzing and estimating and counting the vehicle load.
Further, the step (1) of vehicle data acquisition comprises:
(1.1) uploading the vehicle CAN data to an offline data or vehicle-mounted real-time online data in an enterprise online cloud platform database through a T-box from a data source obtained in the step (1);
(1.2) the Data content and Data = [ T, v, a, T ] obtained, respectively expressed as time, speed, acceleration, total drive torque.
Further, the vehicle data preprocessing step of step (2) includes:
(2.1) the data preprocessing comprises missing values, abnormal values, repeated values and the like, and the data preprocessing is respectively operated according to different data sources;
(2.2) for offline data in the enterprise online cloud platform database, because vehicle data are uploaded to the cloud database in a mobile communication mode, data are possibly mixed with noise in a transmission process due to influence factors such as hardware equipment error reporting or signal network difference, the problem of repeated values mainly exists, and only data with effective data are reserved by carrying out duplicate removal on data acquisition time;
(2.3) processing the missing values and the abnormal values, and performing smooth filtering by using a Gaussian average filtering method to reduce the influence caused by the missing values and the abnormal values;
further, the step (3) of building a vehicle load estimation model comprises:
(3.1) vehicle longitudinal dynamics models such as:
wherein T is tq Total drive torque to the front of the retarder/differential; i all right angle 0 Is the retarder/differential reduction ratio; eta T For transmission system mechanical efficiency; r is the wheel radius; m is the mass of the whole vehicle; f is a rolling resistance coefficient; g is the gravity of the whole vehicle; alpha is the longitudinal gradient of the road surface; a is the windward area; c D Is the air resistance coefficient; u. u a Is the vehicle speed; du/dt is the longitudinal acceleration;
(3.2) at two moments before and after a short time, the vehicle is in a state, and the dynamic equation in (3.1) can be deduced:
(3.3) in two moments before and after a short time, the air resistance can be ignored, the rolling resistance coefficient can be regarded as a constant, and a load estimation model can be obtained according to (3.2):
further, the load estimation of the load estimation model in the step (4) on the preprocessed data in the step (2) comprises:
(4.1) carrying out load estimation on all data in the preprocessed data in the step (2) by using the load estimation model in the step (3.3) to obtain estimated quality;
(4.2) analyzing all the estimated results, and selecting data with the quality similar to the real quality of the vehicle target (within the error range of 5% of the real value);
(4.3) analyzing the selected vehicle data, wherein the selected data can be described by two vehicle states, namely a vehicle acceleration tail stage, two adjacent moments, namely a vehicle speed increasing state, an acceleration decreasing state and a driving torque decreasing state, and a vehicle maintaining constant speed state, namely two relatively stable states, namely a vehicle speed slightly decreasing state, an acceleration increasing state and a total driving torque increasing state;
and (4.5) counting the variation ranges of the speed, the acceleration and the total driving torque of the two vehicle states in the data.
In the step (5), the load estimation model estimation state constraint establishment and the intelligent optimization algorithm optimization step comprise:
(5.1) analyzing the vehicle load estimation result in the step (4.3) to obtain two stable load estimation states, and establishing estimation state constraint conditions;
(5.2) constraint 1 Uniform State Retention can be expressed as:
(5.3) constraint 2 the end of acceleration phase can be expressed as:
(5.4) [ x ] in the formula 1 ，x 2 ，x 3 ，x 4 ，x 5 ，x 6 ]、[y 1 ，y 2 ，y 3 ，y 4 ，y 5 ，y 6 ]Boundary parameters of two state characteristics are respectively, the numerical values of the boundary parameters represent the maximum limits of the speed, the acceleration and the total driving torque variation at adjacent moments, and the determination of the boundary parameters is directly related to the load prediction accuracy;
(5.5) in order to improve the prediction precision of the load prediction model and reduce the noise in the prediction result, multi-objective optimization can be carried out on the model constraint condition parameters through an intelligent optimization algorithm;
(5.6) estimating the mean root mean square error F between the effective mass and the real target mass of the driving data 1 (x) And the ratio F of the number of mean effective values 2 (x) As an optimization objective, the optimization process objective function:
(5.7) in the formula, N is the optimized number of the driving data, N i The estimated effective mass quantity of the ith data; m is a unit of j Calculating a jth quality point in the ith data effective quality by using a model added with constraint conditions; m is a unit of target The real target quality corresponding to the ith data; l is i The total length of the ith piece of data;
(5.8) optimizing the used vehicle data to be the driving data obtained when the vehicle is in different masses, for example, the vehicle model with no load of 2890kg and full load of 4490kg, considering that the mass change space of the vehicle is large, the vehicle can take 2890kg, 3690kg, 4490kg, 6000kg and the like, the driving working conditions of the vehicle include daily common driving working conditions as much as possible, and then obtaining the corresponding driving data;
(5.9) optimization of the parameters [ x ] 1 ，x 2 ，x 3 ，x 4 ，x 5 ，x 6 ]And [ y 1 ，y 2 ，y 3 ，y 4 ，y 5 ，y 6 ]The range is the statistical parameter range, x, of step (4.5) 1 ，x 2 ∈[-0.8，0.5]，x 3 ，x 4 ∈[-0.3，0.6]，x 5 ，x 6 ∈[-15，40]，y 1 ，y 2 ∈[0，0.6]，y 3 ，y 4 ∈[-1.0,0],y 5 ，y 6 ∈[-100,0]。
In the step (6), the model verification and estimation statistics include:
(6.1) the model verification is that after optimization, model parameters are adjusted, production activities are carried out on vehicles with different loads, data are obtained for result verification, if the accuracy of the estimated result is high, the error is less than 5%, and effective quality points are concentrated, the requirement is met, and on the contrary, the steps (4), (5) and (6) are repeated until the requirement is met;
(6.2) the estimation result statistics is estimated by adding constraint conditions, and the estimated points meet two constraint conditions, so that the final load estimation of the vehicle in a certain stroke can be obtained in a statistical mode, the average value of a certain number (more than 10 effective mass points) of effective value mass points or the average value of all points in the previous n seconds (if n is more than 500 s) can be determined, the random noise problem among data can be reduced by the average value mode, and the load estimation accuracy is greatly improved.
Compared with the prior art, the invention has the following beneficial effects:
the invention belongs to a pure electric vehicle load estimation method based on an intelligent optimization algorithm, which is a method for analyzing vehicle driving data by combining a vehicle dynamic balance equation to find two vehicle load estimation states and optimizing boundary parameters based on constraint conditions. The method is simple and efficient, has high estimation accuracy and strong generalization capability, can be deployed and applied under various platforms, and is beneficial to actual production and application.
Drawings
FIG. 1 is a drawing of the abstract of the present invention;
FIG. 2 is a flow chart of a pure electric vehicle load estimation method based on an intelligent optimization algorithm, which is described in the invention;
FIG. 3 is a flow chart illustrating the processing of travel data according to the present invention;
FIG. 4 is a graph showing the result of the load prediction value of the preprocessed data according to the load prediction model of the present invention;
FIG. 5 is a visualization diagram of the prediction result of the unoptimized load prediction model according to the present invention
FIG. 6 is a flowchart illustrating an optimization process based on an intelligent optimization algorithm-genetic algorithm according to the present invention;
FIG. 7 is a diagram illustrating the results of the optimized load estimation model of the present invention;
FIG. 8 is a flow chart illustrating the statistics of final estimates according to the present invention.
Detailed Description
The following detailed description of specific embodiments of the invention is provided, but it should be understood that the scope of the invention is not limited to the specific embodiments.
Throughout the specification and claims, unless explicitly stated otherwise, the word "comprise", or variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated element or component but not the exclusion of any other element or component.
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in the following with reference to the accompanying drawings and the detailed description. It should be understood that the specific embodiments described herein are merely illustrative of and not restrictive on the broad application.
As shown in fig. 2, a pure electric vehicle load estimation method based on an intelligent optimization algorithm specifically includes the following steps:
1, acquiring required vehicle data, namely acquiring data such as vehicle data sampling time (time stamp), driving speed, vehicle acceleration information, driving torque and the like for load estimation of the pure electric vehicle;
2, preprocessing the data of the required vehicle, and preprocessing the missing value, the abnormal value, the repeated value and the like;
and 3, establishing a dynamic balance equation of the vehicles in the two adjacent states, analyzing and comparing the importance of the running resistance in the two adjacent states, and deducing a vehicle load estimation model.
4, carrying out load estimation on the obtained data by using the established load estimation model to obtain estimated vehicle mass, finding vehicle data which correspond to the estimated mass more accurately, analyzing the vehicle state, finding a driving state suitable for the load estimation model, and counting the data ranges of driving speed, acceleration and driving torque between adjacent moments;
and 5, establishing a constraint condition of the estimated state of the application load estimation model, and optimizing the boundary parameter of the constraint condition by using an intelligent optimization algorithm.
And 6, modifying the model parameters, verifying and analyzing and predicting and counting the vehicle load.
As shown in fig. 3, the step 1 of vehicle data acquisition includes:
1.1, uploading the vehicle CAN data to offline data or vehicle-mounted real-time online data in an enterprise online cloud platform database through a T-box from the data source obtained in the step 1;
1.2 Data = [ T, v, a, T ], time, velocity, acceleration, total drive torque, respectively.
As shown in fig. 3, the step 2 vehicle data preprocessing step includes:
2.1, preprocessing the data, wherein the data comprise missing values, abnormal values, repeated values and the like, and the data are respectively operated according to different data sources;
2.2 for the offline data in the enterprise online cloud platform database, as the vehicle data is uploaded to the cloud end database in a mobile communication mode, data noise may be generated in the transmission process due to influence factors such as hardware equipment error reporting or signal network difference, the problem of repeated values mainly exists, and the only data with effective data is reserved by carrying out duplicate removal on data acquisition time;
2.3 processing the missing value and the abnormal value, and performing smooth filtering by using a Gaussian mean filtering method to reduce the influence caused by the missing value and the abnormal value;
further, the step 3 of building the vehicle load estimation model comprises the following steps:
3.vehicle longitudinal dynamics models such as:
wherein T is tq Total drive torque to the front of the retarder/differential; i all right angle 0 Is the retarder/differential reduction ratio; eta T For transmission system mechanical efficiency; r is the wheel radius; m is the mass of the whole vehicle; f is a rolling resistance coefficient; g is the gravity of the whole vehicle; alpha is the longitudinal gradient of the road surface; a is the windward area; c D Is the air resistance coefficient; u. u a Is the vehicle speed; du/dt is the longitudinal acceleration;
3.2 the vehicle loading state at two moments before and after a short time can be deduced according to the dynamic equation in (3.1):
3.3 in two moments before and after the short time, air resistance can be ignored, the rolling resistance coefficient can be regarded as a constant, and a load estimation model can be obtained according to 3.2:
the load estimation of the load estimation model on the preprocessed data in the step (2) comprises the following steps:
4.1, performing load estimation on all data in the preprocessed data in the step 2 by using a 3.3 load estimation model to obtain estimated quality, wherein the obtained data result is shown in a figure 4 and a figure 5;
4.2 analyzing all the estimated results, and selecting data which is similar to the real quality of the vehicle target (within the error range of 5% of the real value), such as data which is similar to the target quality in the data shown in FIG. 3, wherein the target quality is 4490kg;
4.3 by analyzing the selected vehicle data, the selected data can be described by two vehicle states, one is the vehicle acceleration end stage, two adjacent moments are time periods, the vehicle speed is increased, the acceleration is reduced, the driving torque is reduced, and the other is two relatively stable states, namely the vehicle speed is slightly reduced, the acceleration is increased, and the total driving torque is increased to maintain a constant speed state, wherein the two states are represented as a mass effective point approaching 4490kg in fig. 4;
and 4.5, carrying out state statistics on the multiple groups of running data, wherein the two vehicle states in the statistical data are the speed, the acceleration and the total driving torque variation ranges at the acceleration tail stage and the constant speed stage.
As shown in fig. 6, the load estimation model estimation state constraint establishment and intelligent optimization algorithm optimization steps include:
5.1, analyzing the vehicle load estimation result in the step 4.3 to obtain two stable load estimation states, and establishing an estimation state constraint condition;
5.2 constraint 1 Uniform State Retention can be expressed as:
5.3 constraint 2 the end of acceleration phase can be expressed as:
5.4 formula [ x 1 ，x 2 ，x 3 ，x 4 ，x 5 ，x 6 ]、[y 1 ，y 2 ，y 3 ，y 4 ，y 5 ，y 6 ]Boundary parameters of two state characteristics are respectively, the numerical values of the boundary parameters represent the maximum limits of the speed, the acceleration and the total driving torque variation at adjacent moments, and the determination of the boundary parameters is directly related to the load prediction accuracy;
5.5 in order to improve the prediction precision of the load prediction model and reduce the noise in the prediction result, multi-objective optimization can be carried out on the model constraint condition parameters through an intelligent optimization algorithm;
5.6 estimating the average between the effective mass and the real target mass of the driving dataRoot mean square error of value F 1 (x) And the ratio F of the average number of effective values 2 (x) As an optimization objective, optimizing a process objective function:
5.7 in the formula, N is the optimized driving data number, N i The estimated effective mass quantity for the ith data; m is j Calculating a jth quality point in the effective quality of the ith data by using a model added with constraint conditions; m is target The real target quality corresponding to the ith data; l is i The total length of the ith piece of data;
5.8 optimizing the used vehicle data to be driving data obtained when the vehicle is in different masses, for example, the vehicle model with no load of 2890kg and full load of 4490kg, considering that the mass change space of the vehicle is large, the vehicle can take 2890kg, 3690kg, 4490kg, 6000kg and the like, the driving working conditions of the vehicle include daily common driving working conditions as much as possible, and then obtaining the corresponding driving data;
5.9 optimization parameters [ x ] 1 ，x 2 ，x 3 ，x 4 ，x 5 ，x 6 ]And [ y 1 ，y 2 ，y 3 ，y 4 ，y 5 ，y 6 ]The range is the statistical parameter range, x, of step (4.5) 1 ，x 2 ∈[0.8，0.5]，x 3 ，x 4 ∈[-0.3，0.6]，x 5 ，x 6 ∈[-15，40]，y 1 ，y 2 ∈[0，0.6]，y 3 ，y 4 ∈[1.0，0]，y 5 ，y 6 ∈[100，0]。
5.10 the intelligent optimization algorithm has global optimization performance and strong universality and is suitable for parallel processing, and the commonly used intelligent optimization algorithm comprises the following steps: genetic algorithm, simulated annealing algorithm, tabu search algorithm, particle swarm algorithm, ant colony algorithm and the like.
5.11 performing multi-objective optimization on the model constraint condition parameters by using an intelligent optimization algorithm, for example, using a genetic algorithm NSGA-II, as shown in FIG. 6, the optimization principle is that the genetic algorithm performs selection, crossing, mutation of the population and merging of offspring parents according to fitness function values, and pruning the population through non-dominated sorting until the convergence condition is satisfied, the initial set population size is 100, the maximum genetic algebra is 1000, and the crossing probability is 0.80.
5.12 after optimization, model parameters are modified, the estimation result is visualized as shown in FIG. 7, the estimated points of the optimized model are accurate, the estimated effective mass points are all located in the error range of 5% of the target mass, and the error between the estimated mass and the true value is less than 3%.
As shown in fig. 8, the estimation statistics include:
6.1, counting the estimation results, estimating by adding constraint conditions, and estimating points meeting two constraint conditions;
6.2 for the final load estimation of the vehicle in a certain travel, the final load estimation can be obtained in a statistical mode, the average value can be obtained through a certain number (more than 10 effective mass points) of effective value mass points, or the average value of all the points in the previous n seconds (for example, n is more than 500 s) is determined, the random noise problem among data can be reduced through the average value obtaining mode, and the load estimation accuracy is greatly improved.
When the system is used for specific production and deployment, the system can be operated in an online cloud platform server or a vehicle-mounted controller at a user terminal, the application and deployment are flexible, the system can be applied to various service frameworks, and the system can be used as a simple and efficient analysis and identification method tool, analyze and identify by calling an internal Internet of vehicles service cloud platform database of an enterprise, transmit an identification result to an Internet of vehicles remote monitoring service cloud platform, realize information tracking of products, provide accurate data basis for after-sales service and vehicle insurance service, facilitate overload monitoring of a fleet by a vehicle management mechanism, and guarantee driving safety of drivers. The method is applied to the vehicle-mounted controller, provides key parameters for other advanced vehicle control algorithms in the controller, and gives full play to the advantages of the algorithms, so that the vehicle is accurately controlled, and the performance of the vehicle is improved.
Generally, the invention provides a simple, efficient and flexible load estimation model with strong generalization capability, vehicle driving data is utilized for analysis, the load estimation state of effective mass is obtained from a large amount of driving data, and a set of vehicle load estimation method with low cost, high accuracy, low model complexity and simple model calibration is explored.
The foregoing description of specific exemplary embodiments of the invention has been presented for the purposes of illustration and description. It is not intended to limit the invention to the precise form disclosed, and obviously many modifications and variations are possible in light of the above teaching. The exemplary embodiments were chosen and described in order to explain certain principles of the invention and its practical application to enable one skilled in the art to make and use various exemplary embodiments of the invention and various alternatives and modifications as are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the claims and their equivalents.
Claims (7)
1. A pure electric vehicle load estimation method based on an intelligent optimization algorithm specifically comprises the following steps:
s1, acquiring required vehicle data, namely, estimating the load of the pure electric vehicle to obtain data such as vehicle data sampling time (timestamp), driving speed, vehicle acceleration information, driving torque and the like;
s2, preprocessing the data of the required vehicle data, and preprocessing missing values, abnormal values and repeated values contained in the data;
and S3, establishing a dynamic balance equation of the vehicles in the two adjacent states, analyzing and comparing the importance of the driving resistance in the two adjacent states, and deducing a vehicle load estimation model.
S4, carrying out load estimation on the obtained data by using the established load estimation model to obtain estimated vehicle mass, finding vehicle data which correspond to the estimated mass more accurately, analyzing the vehicle state, finding a driving state suitable for the load estimation model, and counting the data ranges of driving speed, acceleration and driving torque between adjacent moments;
s5, establishing constraint conditions of two specific estimation states of the application load estimation model, and optimizing constraint condition boundary parameters by using an intelligent optimization algorithm;
s6, modifying the model parameters, verifying and analyzing and predicting and counting the vehicle load.
2. The pure electric vehicle load estimation method based on the intelligent optimization algorithm according to claim 1, characterized in that: the step S1 of collecting vehicle data comprises the following steps:
s1.1, uploading the acquired data source to offline data or vehicle-mounted real-time online data in an enterprise online cloud platform database through a T-box;
the Data content obtained in S1.2 and Data = [ T, v, a, T ], are represented as T time, v speed, a acceleration sensor information, T total driving torque, respectively.
3. The pure electric vehicle load estimation method based on the intelligent optimization algorithm according to claim 1, characterized in that: the step S2 of preprocessing vehicle data includes: s2.1, the data preprocessing is respectively operated according to different data sources;
s2.2, for offline data in the enterprise online cloud platform database, vehicle data are uploaded to the cloud end database in a mobile communication mode, the problem of repeated values of data doped noise caused by error reporting of hardware equipment or influence factors of signal network difference in the transmission process can be solved, and unique data with effective data are reserved by removing duplication of data acquisition time;
s2.3, processing the missing value and the abnormal value, and performing smooth filtering by using a Gaussian mean filtering method to reduce the influence caused by the missing value and the abnormal value.
4. The pure electric vehicle load estimation method based on the intelligent optimization algorithm according to claim 1, characterized in that: the step S3 of building a vehicle load estimation model comprises the following steps:
s3.1 vehicle longitudinal dynamics models such as:
wherein T is tq Total drive torque to the front of the retarder/differential; i.e. i 0 Is a reduction/differential reduction ratio; eta T For transmission system mechanical efficiency; r is the wheel radius; m is the mass of the whole vehicle; f is a rolling resistance coefficient; g is the gravity of the whole vehicle; alpha is the longitudinal gradient of the road surface; a is the windward area; c D Is the air resistance coefficient; u. u a Is the vehicle speed; du/dt is the longitudinal acceleration;
s3.2, vehicle loading states at two moments before and after a short time can be deduced according to the kinetic equation in (3.1):
s3.3, in two moments before and after a short time, the air resistance can be ignored, the rolling resistance coefficient can be regarded as a constant, and a load estimation model can be obtained according to S3.2:
5. the pure electric vehicle load estimation method based on the intelligent optimization algorithm according to claim 1, characterized in that: the step S4 of carrying out load estimation on the preprocessed data in the step S2 by the load estimation model comprises the following steps:
s4.1, carrying out load estimation on all data in the data preprocessed in the step S2 by using an S3.3 load estimation model to obtain estimated quality;
s4.2, analyzing all the estimated results, and selecting data which is similar to the real quality of the vehicle target or within the error range of 5% of the real value;
s4.3, analyzing the selected vehicle data, wherein the selected data can be described through two specific vehicle states, one is a vehicle acceleration tail stage, the vehicle speed is increased, the acceleration is reduced and the driving torque is reduced at two adjacent moments, and the other is a relatively stable state that the vehicle speed is slightly reduced, the acceleration is increased and the total driving torque is increased for maintaining a constant speed state;
and S4.4, counting the variation ranges of the speed, the acceleration and the total driving torque of the two vehicle states in the data.
6. The pure electric vehicle load estimation method based on the intelligent optimization algorithm according to claim 1, characterized in that: in step S5, the load estimation model estimation state constraint establishment and the intelligent optimization algorithm optimization step include:
s5.1, analyzing the vehicle load estimation result in the step S4.3 to obtain two specific load estimation states, and establishing an estimation state constraint condition;
s5.2 constraint 1 uniform state retention can be expressed as:
s5.3 constraint 2 the end of acceleration phase can be expressed as:
s5.4 in formula [ x 1 ，x 2 ，x 3 ，x 4 ，x 5 ，x 6 ]、[y 1 ，y 2 ，y 3 ，y 4 ，y 5 ，y 6 ]Boundary parameters of two state characteristics are respectively, the numerical values of the boundary parameters represent the maximum limits of speed, acceleration and total driving torque variation at adjacent moments, and the determination of the boundary parameters is directly related to the load estimation accuracy;
s5.5, in order to improve the prediction precision of the load prediction model and reduce the noise in the prediction result, multi-objective optimization can be carried out on the model constraint condition parameters through an intelligent optimization algorithm;
S5.6the mean root mean square error F between the estimated effective mass and the real target mass of the driving data 1 (x) And the ratio F of the number of mean effective values 2 (x) As an optimization objective, optimizing a process objective function:
s5.7, N is the optimized number of driving data, N i The estimated effective mass quantity of the ith data; m is j Calculating a jth quality point in the effective quality of the ith data by using a model added with constraint conditions; m is target The real target quality corresponding to the ith data; l is a radical of an alcohol i The total length of the ith data;
s5.8, optimizing the used vehicle data to be the driving data obtained when the vehicle is in different qualities, wherein the driving working conditions of the vehicle comprise daily common driving working conditions as much as possible, and then obtaining the corresponding driving data;
s5.9 optimization of the parameters [ x ] 1 ，x 2 ，x 3 ，x 4 ，x 5 ，x 6 ]And [ y 1 ，y 2 ，y 3 ，y 4 ，y 5 ，y 6 ]The range is the statistical parameter range, x, of step S4.5 1 ，x 2 ∈[-0.8，0.5]，x 3 ，x 4 ∈[-0.3，0.6]，x 5 ，x 6 ∈[-15，40]，y 1 ，y 2 ∈[0，0.6]，y 3 ，y 4 ∈[-1.0，0]，y 5 ，y 6 ∈[-100，0]。
7. The pure electric vehicle load estimation method based on the intelligent optimization algorithm according to claim 1, characterized in that: in step S6, the model verification and estimation statistics include:
s6.1, model verification is that after optimization, model parameters are adjusted, production activities are carried out on vehicles with different loads, result verification is carried out on obtained data, if the accuracy of the estimated result is high, the error is less than 5%, and effective quality points are concentrated, the requirement is met, and on the contrary, the steps S4, S5 and S6 are repeated until the requirement is met;
s6.2, the estimation result statistics is estimated by adding a constraint condition mode, and estimated points meeting two constraint conditions are estimated, so that the final load estimation of the vehicle in a certain stroke can be obtained in a statistical mode, the average value of a certain number of effective value quality points can be obtained, or the average value of all points in the previous n seconds can be determined, the random noise problem among data can be reduced by the average value, and the load estimation accuracy is greatly improved.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310024350.0A CN115973173A (en) | 2023-01-09 | 2023-01-09 | Pure electric vehicle load estimation method based on intelligent optimization algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310024350.0A CN115973173A (en) | 2023-01-09 | 2023-01-09 | Pure electric vehicle load estimation method based on intelligent optimization algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115973173A true CN115973173A (en) | 2023-04-18 |
Family
ID=85970074
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310024350.0A Pending CN115973173A (en) | 2023-01-09 | 2023-01-09 | Pure electric vehicle load estimation method based on intelligent optimization algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115973173A (en) |
-
2023
- 2023-01-09 CN CN202310024350.0A patent/CN115973173A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110610260B (en) | Driving energy consumption prediction system, method, storage medium and equipment | |
| CN110148230B (en) | Whole vehicle load prediction method based on LSTM neural network | |
| CN107688343B (en) | Energy control method of hybrid power vehicle | |
| CN111301426B (en) | Method for predicting energy consumption in future driving process based on GRU network model | |
| CN110459053B (en) | Highway mixed traffic flow speed estimation method based on intelligent network vehicle connection information | |
| CN114783183A (en) | Monitoring method and system based on traffic situation algorithm | |
| CN109002622B (en) | Method for estimating overall load response of large-span bridge under random traffic flow effect | |
| CN110346518B (en) | Traffic emission pollution visualization early warning method and system thereof | |
| CN112950128A (en) | Intelligent logistics transportation system and method based on big data fusion | |
| CN112200932A (en) | Method and equipment for evaluating energy consumption of heavy-duty diesel vehicle | |
| CN113297685B (en) | Vehicle operation condition mode identification method | |
| CN113222385B (en) | Method for constructing and evaluating driving condition of electric automobile | |
| CN112542049A (en) | Comprehensive management and control platform for intelligent traffic | |
| CN116882981B (en) | Intelligent battery management system based on data analysis | |
| CN114510778A (en) | Track irregularity prediction method based on hybrid intelligent optimization LSTM | |
| CN113642768A (en) | Vehicle running energy consumption prediction method based on working condition reconstruction | |
| CN113535816A (en) | Driving performance evaluation method and system for intelligent network cloud control vehicle | |
| CN111198032A (en) | Real-time estimation method for automobile mass | |
| CN114595721A (en) | Vehicle part strain load prediction method based on road condition recognition | |
| CN113837473A (en) | Charging equipment fault rate analysis system and method based on BP neural network | |
| Kan et al. | A deep learning engine power model for estimating the fuel consumption of heavy-duty trucks | |
| CN115973173A (en) | Pure electric vehicle load estimation method based on intelligent optimization algorithm | |
| CN112035536A (en) | Electric automobile energy consumption prediction method considering dynamic road network traffic flow | |
| Xing et al. | Driving cycle recognition for hybrid electric vehicle | |
| Rui et al. | High-accuracy transient fuel consumption model based on distance correlation analysis |
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 |