CN117494585A - Commercial vehicle actual load prediction method based on mutual information and data blurring - Google Patents
Commercial vehicle actual load prediction method based on mutual information and data blurring Download PDFInfo
- Publication number
- CN117494585A CN117494585A CN202311849690.3A CN202311849690A CN117494585A CN 117494585 A CN117494585 A CN 117494585A CN 202311849690 A CN202311849690 A CN 202311849690A CN 117494585 A CN117494585 A CN 117494585A
- Authority
- CN
- China
- Prior art keywords
- data
- actual
- blurring
- simulated
- commercial vehicle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 65
- 238000004088 simulation Methods 0.000 claims abstract description 76
- 238000012549 training Methods 0.000 claims abstract description 14
- 238000005457 optimization Methods 0.000 claims abstract description 13
- 230000001133 acceleration Effects 0.000 claims description 14
- 230000005540 biological transmission Effects 0.000 claims description 12
- 239000012634 fragment Substances 0.000 claims description 10
- 238000007781 pre-processing Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 5
- 101100421536 Danio rerio sim1a gene Proteins 0.000 claims description 4
- 101100495431 Schizosaccharomyces pombe (strain 972 / ATCC 24843) cnp1 gene Proteins 0.000 claims description 4
- 101100365794 Schizosaccharomyces pombe (strain 972 / ATCC 24843) sim3 gene Proteins 0.000 claims description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 241000135164 Timea Species 0.000 claims description 2
- 238000013528 artificial neural network Methods 0.000 claims description 2
- 230000001174 ascending effect Effects 0.000 claims description 2
- 230000002068 genetic effect Effects 0.000 claims description 2
- 238000013507 mapping Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 7
- 238000012795 verification Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/02—Computing arrangements based on specific mathematical models using fuzzy logic
-
- 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
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Geometry (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Computer Hardware Design (AREA)
- Pure & Applied Mathematics (AREA)
- Automation & Control Theory (AREA)
- Mathematical Optimization (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Fuzzy Systems (AREA)
- Aviation & Aerospace Engineering (AREA)
- Molecular Biology (AREA)
- Algebra (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Traffic Control Systems (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a commercial vehicle actual load prediction method based on mutual information and data blurring, which adopts a mutual information theory to extract the optimal parameters of sub-segments of simulation driving data for load prediction, thereby realizing the simplification of characteristic parameters of redundant vehicles; according to the model prediction deviation and the actual running data, selecting an optimal blurring factor, blurring the simulated running data according to the optimal blurring factor, determining a blurring factor mode is defined, and meanwhile coverage of the simulated running data to the actual running data is improved; based on the fuzzy simulation running data, an ultra-parameter optimizing method combining grid search and Bayesian optimization is adopted to obtain an optimal model parameter combination, and a commercial vehicle actual load prediction model is obtained through training, so that the model training cost is reduced, meanwhile, the accurate prediction of the commercial vehicle actual load is realized, and the method has higher prediction precision and good engineering application performance.
Description
Technical Field
The invention relates to the technical field of transportation, in particular to a commercial vehicle actual load prediction method based on mutual information and data blurring.
Background
The accurate prediction of the actual load of the commercial vehicle is a key for making a control strategy of the commercial vehicle and improving the economy of the whole vehicle, and meanwhile, the obtained load information has important engineering practical significance for efficient scheduling in the logistics transportation industry, product performance optimization of a host factory and a part enterprise and effective supervision of government departments. However, due to the large loading range and complex driving conditions of commercial vehicles, how to accurately predict the actual loading of the vehicles is always a problem to be solved by commercial vehicle enterprises.
The load prediction method used in the traditional technology comprises two modes of using actual running data to load prediction model and using simulation running data to load prediction model, when the actual running data is adopted to train the model, the method has larger limitation and high landing cost due to the problems of difficult actual data acquisition, high cost, low efficiency and the like; when the model is trained by adopting the simulation running data, the more complex running condition of the vehicle on the actual road is difficult to fully cover by the simulation data, so that the model prediction effect is poor.
Disclosure of Invention
Aiming at the technical problems pointed out in the background art, the invention aims to provide a commercial vehicle actual load prediction method based on mutual information and data blurring.
In order to achieve the above purpose, the technical scheme provided by the invention is as follows:
a commercial vehicle actual load prediction method based on mutual information and data blurring comprises the following steps:
step 1: determining vehicle characteristic information of a target commercial vehicle, and acquiring corresponding load values and road running data of the target commercial vehicle under different load segments, wherein the road running data comprises simulation running data and actual running data;
the simulation running data comprise simulation time, simulation vehicle speed, simulation acceleration change rate, simulation rotating speed change rate, simulation torque change rate, simulation engine instantaneous power change rate, simulation specific power change rate, simulation gradient, simulation elevation difference and simulation transmission ratio;
the actual driving data comprise actual time, actual vehicle speed, actual acceleration change rate, actual rotating speed change rate, actual torque change rate, actual engine instantaneous power change rate, actual specific power change rate, actual gradient, actual elevation difference and actual transmission ratio;
step 2: preprocessing the simulated running data under different load sections according to the simulated acceleration and the simulated elevation difference in the simulated running data; cutting the preprocessed simulated running data according to the simulation time and the simulation speed in the simulated running data to obtain simulated running data subfragments;
step 3: extracting preferred parameters from the characteristic parameters of the simulated driving data subfragments by adopting a mutual information theory, and selecting the first five characteristic parameters with the maximum mutual information value as the preferred parameters of the simulated driving data subfragments;
step 4: establishing a blurring factor, generating a blurring set, carrying out blurring treatment on the preferred parameters, generating a preliminary sub-segment data set, and constructing a commercial vehicle load prediction model according to the preliminary sub-segment data set;
step 5: determining an optimal value of the blurring factor, and obtaining the optimal blurring factor;
step 6: carrying out fuzzification according to the optimal fuzzification factors, obtaining an optimal sub-fragment data set as training data, adopting a super-parameter optimizing method combining grid search and Bayesian optimization to optimize model parameters of the commercial vehicle load prediction model, obtaining an optimal model parameter combination, and obtaining a trained commercial vehicle load prediction model based on the optimal model parameter combination and the optimal sub-fragment data set;
step 7: and dividing travel chains based on actual travel data, and predicting a predicted load value of each travel chain based on a trained commercial vehicle load prediction model.
Compared with the prior art, the invention has the beneficial effects that:
according to the technical scheme provided by the invention, the mutual information theory is adopted, the optimal parameters of the simulated driving data sub-segments are extracted and used for load prediction, so that the simplification of the characteristic parameters of the redundant vehicles is realized; according to the model prediction deviation and the actual running data, selecting an optimal blurring factor, blurring the simulated running data according to the optimal blurring factor, determining a blurring factor mode is defined, and meanwhile coverage of the simulated running data to the actual running data is improved; based on the fuzzy simulation running data, an ultra-parameter optimizing method combining grid search and Bayesian optimization is adopted to obtain an optimal model parameter combination, and a commercial vehicle actual load prediction model is obtained through training, so that the model training cost is reduced, meanwhile, the accurate prediction of the commercial vehicle actual load is realized, and the method has higher prediction precision and good engineering application performance.
Drawings
FIG. 1 is a schematic flow chart of a method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of vehicle characteristic information of a target commercial vehicle in an embodiment of the present invention;
FIG. 3 is a diagram showing the distribution of mutual information values between characteristic parameters and load values in an embodiment of the present invention;
FIG. 4 is a schematic diagram of simulated driving data before simulation and coverage of the simulated driving data with actual driving data according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of simulated driving data and coverage thereof on actual driving data after being simulated in an embodiment of the present invention;
FIG. 6 is a schematic diagram of an actual driving data acquisition mode of a target commercial vehicle according to an embodiment of the present invention;
fig. 7 is a schematic diagram of an actual load prediction result of a target commercial vehicle according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only 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.
The embodiment provides a commercial vehicle actual load prediction method based on mutual information and data blurring, which adopts a mutual information theory to extract the preferred parameters of the sub-segments of the simulated driving data for load prediction, thereby realizing the simplification of the characteristic parameters of the redundant vehicles; according to the model prediction deviation and the actual running data, selecting an optimal blurring factor, blurring the simulated running data according to the optimal blurring factor, determining a blurring factor mode is defined, and meanwhile coverage of the simulated running data to the actual running data is improved; based on the fuzzy simulation running data, an ultra-parameter optimizing method combining grid search and Bayesian optimization is adopted to obtain an optimal model parameter combination, and a commercial vehicle actual load prediction model is obtained through training, so that the model training cost is reduced, meanwhile, the accurate prediction of the commercial vehicle actual load is realized, and the method has higher prediction precision and good engineering application performance.
As shown in fig. 1, the embodiment provides a commercial vehicle actual load prediction method based on mutual information and data blurring, which includes the following steps:
step 1: determining vehicle characteristic information of a target commercial vehicle, and acquiring corresponding load values and road running data of the target commercial vehicle under different load segments, wherein the road running data comprises simulation running data and actual running data; wherein the vehicle characteristic information of the target commercial vehicle is shown in fig. 2.
The simulation running data comprise 15 vehicle parameters such as simulation time, simulation vehicle speed, simulation acceleration change rate, simulation rotating speed change rate, simulation torque change rate, simulation engine instantaneous power change rate, simulation specific power change rate, simulation gradient, simulation elevation difference, simulation transmission ratio and the like; the simulation running data are obtained through simulation calculation by using a commercial vehicle running simulation model.
The actual driving data comprise actual time, actual vehicle speed, actual acceleration change rate, actual rotating speed change rate, actual torque change rate, actual engine instantaneous power change rate, actual specific power change rate, actual gradient, actual elevation difference and actual transmission ratio; the actual driving data are mainly read and calculated to obtain 15 vehicle parameters through the vehicle-mounted data acquisition terminal connected with the vehicle OBD interface. The method has the advantages that the actual driving data demand is small, and the method is mainly used for verification of the follow-up simulation driving data processing result and verification of the load prediction result. As shown in fig. 6, a schematic diagram of the actual driving data acquisition mode of the target commercial vehicle is given.
Step 2: preprocessing the simulated running data under different load sections according to the simulated acceleration and the simulated elevation difference in the simulated running data; cutting the preprocessed simulated running data according to the simulation time and the simulation speed in the simulated running data to obtain simulated running data subfragments;
the method specifically comprises the following steps:
step 2.1: for the simulated running data, if the simulated acceleration occurs at any timea< -0.1 m/s 2 Or simulate elevation difference<-0.5 m, deleting the simulated running data corresponding to the moment to obtain the preprocessed simulated running data;
step 2.2: cutting the preprocessed simulated running data into a plurality of sub-segments with the duration of 10s according to the simulation time and the simulation speed to obtain simulated running data sub-segments, namely non-deceleration and non-downhill simulated running data sub-segments;
step 2.3: and calculating the average value of the simulated vehicle speed, the simulated acceleration change rate, the simulated rotating speed change rate, the simulated torque change rate, the simulated engine instantaneous power change rate, the simulated specific power change rate, the simulated gradient, the simulated elevation difference and the simulated transmission ratio in each simulated driving data sub-segment, and taking the average value as the characteristic parameter of the simulated driving data sub-segment.
Step 3: extracting preferred parameters from characteristic parameters of the simulated driving data subfragments by adopting a mutual information theory, and predicting commercial vehicle load;
the method specifically comprises the following steps:
step 3.1: normalizing the characteristic parameters and the load values of all the simulated driving data subfragments, and mapping the characteristic parameters and the load values to the [0,1] interval;
step 3.2: calculating a mutual information value between the characteristic parameter and the load value; the mutual information value represents the correlation between the characteristic parameter and the load value, and the larger the mutual information value is, the higher the correlation between the characteristic parameter and the load value is; as shown in fig. 3, a mutual information value distribution between the characteristic parameter and the load value is shown;
step 3.3: and selecting the first five characteristic parameters with the maximum mutual information value as the optimal parameters of the simulated driving data subfragments according to the size of the mutual information value, and predicting the commercial vehicle load. As shown in FIG. 3, the simulated torque, simulated instantaneous power, simulated vehicle speed, simulated gear ratio, and simulated elevation difference may be selected as preferred parameters for the simulated travel data subfragments.
Step 4: establishing a blurring factor, generating a blurring set, carrying out blurring treatment on preferred parameters, generating a primary sub-segment data set, expanding the coverage of simulated driving data to actual driving data, and constructing a commercial vehicle load prediction model according to the primary sub-segment data set;
the method comprises the steps of constructing a blurring factor, generating a blurring set, blurring preferred parameters, and generating a primary sub-segment data set, and specifically comprises the following steps:
step 4.1: establishing a blurring factor, generating a mean value of 0, a standard deviation of the blurring factor, and a sample size equal to the number of the sub-fragments of the simulated driving datalIs used as fuzzy setU = {u i }，i = 1,2,…,l，The blurring factor represents the expansion degree of the coverage of the simulated driving data to the actual driving data, and the larger the blurring factor is, the larger the expansion degree is, and the selection range is 0-15%;
step 4.2: respectively carrying out blurring processing on the optimized parameters to generate a blurring data set with the same data quantity as the simulated driving data subfragments, and integrating the blurring data set with the simulated driving data subfragments to form a preliminary subfragment data set, wherein the blurring processing is shown in a formula (1):
(1)
in the method, in the process of the invention,P i represent the firstiPreferred parameters for each simulated driving data subfragment;represent the firstiPreferred parameters after blurring the sub-segments of the simulated driving data;u i represent the firstiAnd the fuzzy coefficients corresponding to the simulated driving data subfragments.
Furthermore, the blurring process of the preferred parameters is selective. Only when the preferred parameters comprise parameters such as simulation torque, simulation transmission ratio and the like, the blurring process is carried out, the blurring process is only carried out on the simulation torque and the simulation transmission ratio, and the blurring process is not needed for the rest parameters and the rest conditions. Therefore, for the preferred parameters of the embodiment, the blurring process is required for the simulation torque and the simulation transmission ratio, and the blurring process is not required for the simulation instantaneous power, the simulation vehicle speed and the simulation elevation difference.
Preferably, a BP neural network method (GA-BP) based on genetic algorithm optimization is adopted to build a commercial vehicle load prediction model.
The commercial vehicle load prediction model takes the optimal parameters as input, takes the load value as output and the hidden layer number as 12, namely, the commercial vehicle load prediction model with the input layer as 5, the hidden layer as 12 and the output layer as 1 is constructed; model parameter population size (selection range is [20,120 ]), evolution times (selection range is [50,550 ]), crossover probability (selection range is [0.4,0.9 ]), variation probability (selection range is [0.01,0.2 ]).
In addition, when the load prediction model is trained subsequently, the provided sub-segment data sets are adopted, 70% of the sub-segment data sets are randomly extracted for training the model, the rest 30% of the sub-segment data sets are used for prediction, and the prediction deviation of the model is obtained.
Step 5: determining an optimal value of the blurring factor, and obtaining the optimal blurring factor;
the method specifically comprises the following steps:
step 5.1: the method comprises the steps of respectively carrying out fuzzification on data in a preliminary sub-segment data set according to preset fuzzification factor values of 2.5%, 5%, 7.5%, 10%, 12.5% and 15%, and obtaining corresponding fuzzification sub-segment data sets under different preset fuzzification factor values;
step 5.2: aiming at different preset fuzzification factor values, the obtained fuzzification sub-segment data sets are taken as training data, default model parameters are selected based on the commercial vehicle load prediction model constructed in the step 4, training and prediction of the commercial vehicle load prediction model are carried out, 3 fuzzification factors with the minimum prediction deviation are selected as candidate fuzzification factors according to the prediction deviation of the commercial vehicle load prediction model, and are respectively recorded as sim1, sim2 and sim3; as shown in table 1, the predicted deviation at different preset values of the blurring factor is given. According to the prediction deviation, 3 blurring factors with the smallest prediction deviation are selected as candidate blurring factors, and the specific candidate blurring factors are 5%, 7.5% and 10%.
TABLE 1 prediction bias at different preset fuzzification factor values
| Presetting a fuzzification factor value | 2.5% | 5% | 7.5% | 10% | 12.5% | 15% |
| Prediction bias | 1.11% | 0.81% | 0.71% | 0.97% | 2.14% | 2.31% |
Step 5.3: based on the simulation torque and the simulation rotating speed, according to a convex hull algorithm, calculating a convex polygon of the fuzzy sub-segment data set as a boundary of a coverage area of a data area, and further respectively obtaining convex hulls C of simulation running data corresponding to different alternative fuzzy factors j Convex hull C of actual driving data real Wherein, the method comprises the steps of, wherein,j = sim1, sim2, sim3；
step 5.4: according to (2), calculating a convex hull C of the simulated driving data j And convex hull C of actual driving data real Convex hull C of intersection with actual driving data real Area coverage area ratio of (2)rThe method comprises the steps of carrying out a first treatment on the surface of the According to different alternative blurring factorsrThe value is selectedrThe candidate blurring factor corresponding to the maximum value of (2) is used as the optimal blurring factor:
(2)
in the method, in the process of the invention,A() Representing an area function;representing the coverage area of an intersection area of the simulated driving data and the actual driving data; />The area coverage representing the actual travel data. As shown in fig. 4 and 5, the coverage expansion degree of the simulated running data before and after the blurring and the coverage of the simulated running data after the blurring to the actual running data are shown under the optimal blurring factor. As can be seen from fig. 4 and fig. 5, compared with the simulation running data before the blurring, the simulation running data after the blurring can completely cover the actual running data, and the coverage of the simulation running data to the actual running data is greatly improved.
Step 6: carrying out fuzzification according to the optimal fuzzification factors, obtaining an optimal sub-fragment data set as training data, adopting a super-parameter optimizing method combining grid search and Bayesian optimization to optimize model parameters of the commercial vehicle load prediction model, obtaining an optimal model parameter combination, and obtaining a trained commercial vehicle load prediction model based on the optimal model parameter combination and the optimal sub-fragment data set;
the method specifically comprises the following steps:
step 6.1: adopting a grid searching method, taking 1/10 of the maximum value range of each model parameter of the commercial vehicle load prediction model as a step length in a high-dimensional space, wherein the step length is equal to the maximum value range of each model parameter of the commercial vehicle load prediction modelc = (b k -a k )/10，k =1, 2,3,4, performing a traversal search, calculating a first model prediction bias under each model parameter combination, and performing ascending arrangement according to the magnitude of the first model prediction bias; wherein,b k represents the firstkThe maximum value range of the individual model parameters,a k represents the firstkMinimum value range of the individual model parameters;
step 6.2: adopting a Bayesian optimization method, aiming at the model parameter combination corresponding to the first 10% of the first model prediction deviation, taking 1% of the maximum value range of each model parameter as a step length, carrying out super-parameter optimization, calculating a second model prediction deviation under the new model parameter combination, and taking the model parameter combination corresponding to the minimum second model prediction deviation as an optimal model parameter combination; at this time, the population size was 20, the evolution number was 100, the crossover probability was 0.4, and the mutation probability was 0.1.
Step 6.3: and obtaining an optimal sub-segment data set based on the optimal blurring factor blurring, and obtaining a trained commercial vehicle load prediction model based on the optimal model parameter combination and the optimal sub-segment data set.
Step 7: and dividing travel chains based on actual travel data, and predicting a predicted load value of each travel chain based on a trained commercial vehicle load prediction model.
The method specifically comprises the following steps:
step 7.1: dividing actual driving data into a plurality of travel chains by using flameout duration of a target commercial vehicle, wherein the travel chains are divided by taking flameout time of the target commercial vehicle as a starting time and taking flameout 30 minutes as an interval; meanwhile, before the target commercial vehicle starts each time, the load is weighed by the wagon balance, and the actual load of the target commercial vehicle is recorded and used for checking the load prediction capacity of the constructed model.
Step 7.2: preprocessing and cutting each travel chain to obtain travel chain fragments; and (3) carrying out load prediction on the travel chain segments by adopting the trained commercial vehicle load prediction model in the step (6) to obtain the predicted load corresponding to each travel chain segment, and simultaneously selecting the predicted load median between 5% and 95% as the final predicted value of the actual load of the travel chain. And verifying the model prediction accuracy according to the recorded actual load, wherein the prediction result is shown in fig. 7.
The mode of preprocessing and cutting each travel chain is identical to the mode of preprocessing and cutting the simulated travel data.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the above-described embodiments, and that the above-described embodiments and descriptions are only preferred embodiments of the present invention, and are not intended to limit the invention, and that various changes and modifications may be made therein without departing from the spirit and scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (10)
1. The commercial vehicle actual load prediction method based on mutual information and data blurring is characterized by comprising the following steps of:
step 1: determining vehicle characteristic information of a target commercial vehicle, and acquiring corresponding load values and road running data of the target commercial vehicle under different load segments, wherein the road running data comprises simulation running data and actual running data;
the simulation running data comprise simulation time, simulation vehicle speed, simulation acceleration change rate, simulation rotating speed change rate, simulation torque change rate, simulation engine instantaneous power change rate, simulation specific power change rate, simulation gradient, simulation elevation difference and simulation transmission ratio;
the actual driving data comprise actual time, actual vehicle speed, actual acceleration change rate, actual rotating speed change rate, actual torque change rate, actual engine instantaneous power change rate, actual specific power change rate, actual gradient, actual elevation difference and actual transmission ratio;
step 2: preprocessing the simulated running data under different load sections according to the simulated acceleration and the simulated elevation difference in the simulated running data; cutting the preprocessed simulated running data according to the simulation time and the simulation speed in the simulated running data to obtain simulated running data subfragments;
step 3: extracting preferred parameters from the characteristic parameters of the simulated driving data subfragments by adopting a mutual information theory, and selecting the first five characteristic parameters with the maximum mutual information value as the preferred parameters of the simulated driving data subfragments;
step 4: establishing a blurring factor, generating a blurring set, carrying out blurring treatment on the preferred parameters, generating a preliminary sub-segment data set, and constructing a commercial vehicle load prediction model according to the preliminary sub-segment data set;
step 5: determining an optimal value of the blurring factor, and obtaining the optimal blurring factor;
step 6: carrying out fuzzification according to the optimal fuzzification factors, obtaining an optimal sub-fragment data set as training data, adopting a super-parameter optimizing method combining grid search and Bayesian optimization to optimize model parameters of the commercial vehicle load prediction model, obtaining an optimal model parameter combination, and obtaining a trained commercial vehicle load prediction model based on the optimal model parameter combination and the optimal sub-fragment data set;
step 7: and dividing travel chains based on actual travel data, and predicting a predicted load value of each travel chain based on a trained commercial vehicle load prediction model.
2. The method for predicting the actual load of the commercial vehicle based on mutual information and data blurring according to claim 1, wherein the step 2 specifically comprises:
step 2.1: for the simulated running data, if the simulated acceleration occurs at any timea < -0.1 m/s 2 Or simulate elevation difference<-0.5 m, deleting the simulated running data corresponding to the moment to obtain the preprocessed simulated running data;
step 2.2: cutting the preprocessed simulated running data into a plurality of sub-segments with the duration of 10s according to the simulation time and the simulation speed to obtain simulated running data sub-segments;
step 2.3: and calculating the average value of the simulated vehicle speed, the simulated acceleration change rate, the simulated rotating speed change rate, the simulated torque change rate, the simulated engine instantaneous power change rate, the simulated specific power change rate, the simulated gradient, the simulated elevation difference and the simulated transmission ratio in each simulated driving data sub-segment, and taking the average value as the characteristic parameter of the simulated driving data sub-segment.
3. The method for predicting the actual load of the commercial vehicle based on mutual information and data blurring according to claim 2, wherein the step 3 specifically comprises:
step 3.1: normalizing the characteristic parameters and the load values of all the simulated driving data subfragments, and mapping the characteristic parameters and the load values to the [0,1] interval;
step 3.2: calculating a mutual information value between the characteristic parameter and the load value; the mutual information value represents the correlation between the characteristic parameter and the load value, and the larger the mutual information value is, the higher the correlation between the characteristic parameter and the load value is;
step 3.3: and selecting the first five characteristic parameters with the maximum mutual information value as the preferred parameters of the simulated driving data subfragments according to the size of the mutual information value.
4. The method for predicting actual load of commercial vehicle based on mutual information and data blurring as set forth in claim 3, wherein in step 4, the step of constructing blurring factors, generating a blurring set, blurring preferred parameters, and generating a preliminary sub-segment dataset specifically includes:
step 4.1: establishing a blurring factor, generating a mean value of 0, a standard deviation of the blurring factor, and a sample size equal to the number of the sub-fragments of the simulated driving datalIs used as fuzzy setU = {u i }，i = 1,2,…,l，The blurring factor represents the expansion degree of the coverage of the simulated driving data to the actual driving data, and the larger the blurring factor is, the larger the expansion degree is, and the selection range is 0-15%;
step 4.2: respectively carrying out blurring processing on the optimized parameters to generate a blurring data set with the same data quantity as the simulated driving data subfragments, and integrating the blurring data set with the simulated driving data subfragments to form a preliminary subfragment data set, wherein the blurring processing is shown in a formula (1):
(1)
in the method, in the process of the invention,P i represent the firstiPreferred parameters for each simulated driving data subfragment;represent the firstiPreferred parameters after blurring the sub-segments of the simulated driving data;u i represent the firstiAnd the fuzzy coefficients corresponding to the simulated driving data subfragments.
5. The method for predicting the actual load of the commercial vehicle based on mutual information and data blurring according to claim 4, wherein in the step 4, a BP neural network method based on genetic algorithm optimization is adopted to build a commercial vehicle load prediction model.
6. The method for predicting actual load of commercial vehicle based on mutual information and data blurring according to claim 5, wherein in the step 4, the load prediction model of the commercial vehicle takes the optimized parameter as input, takes the load value as output, and the hidden layer number is 12; model parameters of the commercial vehicle load prediction model comprise population size, evolution times, crossover probability and variation probability.
7. The method for predicting the actual load of the commercial vehicle based on mutual information and data blurring as set forth in claim 6, wherein the step 5 specifically includes:
step 5.1: the method comprises the steps of respectively carrying out fuzzification on data in a preliminary sub-segment data set according to preset fuzzification factor values of 2.5%, 5%, 7.5%, 10%, 12.5% and 15%, and obtaining corresponding fuzzification sub-segment data sets under different preset fuzzification factor values;
step 5.2: aiming at different preset fuzzification factor values, the obtained fuzzification sub-segment data sets are taken as training data, default model parameters are selected based on the commercial vehicle load prediction model constructed in the step 4, training and prediction of the commercial vehicle load prediction model are carried out, 3 fuzzification factors with the minimum prediction deviation are selected as candidate fuzzification factors according to the prediction deviation of the commercial vehicle load prediction model, and are respectively recorded as sim1, sim2 and sim3;
step 5.3: based on the simulation torque and the simulation rotating speed, according to a convex hull algorithm, calculating a convex polygon of the fuzzy sub-segment data set as a boundary of a coverage area of a data area, and further respectively obtaining convex hulls C of simulation running data corresponding to different alternative fuzzy factors j Convex hull C of actual driving data real Wherein, the method comprises the steps of, wherein,j = sim1, sim2, sim3；
step 5.4: according to (2), calculating simulated running dataConvex hull C j And convex hull C of actual driving data real Convex hull C of intersection with actual driving data real Area coverage area ratio of (2)rThe method comprises the steps of carrying out a first treatment on the surface of the According to different alternative blurring factorsrThe value is selectedrThe candidate blurring factor corresponding to the maximum value of (2) is used as the optimal blurring factor:
(2)
in the method, in the process of the invention,A() Representing an area function;representing the coverage area of an intersection area of the simulated driving data and the actual driving data; />The area coverage representing the actual travel data.
8. The method for predicting the actual load of the commercial vehicle based on mutual information and data blurring according to claim 7, wherein the step 6 specifically comprises:
step 6.1: adopting a grid searching method, taking 1/10 of the maximum value range of each model parameter of the commercial vehicle load prediction model as a step length in a high-dimensional space, wherein the step length is equal to the maximum value range of each model parameter of the commercial vehicle load prediction modelc = (b k -a k )/10，k =1, 2,3,4, performing a traversal search, calculating a first model prediction bias under each model parameter combination, and performing ascending arrangement according to the magnitude of the first model prediction bias; wherein,b k represents the firstkThe maximum value range of the individual model parameters,a k represents the firstkMinimum value range of the individual model parameters;
step 6.2: adopting a Bayesian optimization method, aiming at the model parameter combination corresponding to the first 10% of the first model prediction deviation, taking 1% of the maximum value range of each model parameter as a step length, carrying out super-parameter optimization, calculating a second model prediction deviation under the new model parameter combination, and taking the model parameter combination corresponding to the minimum second model prediction deviation as an optimal model parameter combination;
step 6.3: and obtaining an optimal sub-segment data set based on the optimal blurring factor blurring, and obtaining a trained commercial vehicle load prediction model based on the optimal model parameter combination and the optimal sub-segment data set.
9. The method for predicting the actual load of the commercial vehicle based on mutual information and data blurring according to claim 8, wherein the step 7 specifically comprises:
step 7.1: dividing actual driving data into a plurality of travel chains by using flameout duration of a target commercial vehicle, wherein the travel chains are divided by taking flameout time of the target commercial vehicle as a starting time and taking flameout 30 minutes as an interval;
step 7.2: preprocessing and cutting each travel chain to obtain travel chain fragments; and (3) carrying out load prediction on the travel chain segments by adopting the trained commercial vehicle load prediction model in the step (6) to obtain the predicted load corresponding to each travel chain segment, and simultaneously selecting the predicted load median between 5% and 95% as the final predicted value of the actual load of the travel chain.
10. The method for predicting the actual load of the commercial vehicle based on mutual information and data blurring according to claim 1 is characterized in that in the step 1, the actual running data is obtained in the following manner, and the actual time, the actual vehicle speed, the actual acceleration change rate, the actual rotating speed change rate, the actual torque change rate, the actual engine instantaneous power change rate, the actual specific power change rate, the actual gradient, the actual elevation difference and the actual transmission ratio in the actual running process of the vehicle are read and calculated through the vehicle-mounted data acquisition terminal connected with the vehicle OBD interface.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311849690.3A CN117494585B (en) | 2023-12-29 | 2023-12-29 | Commercial vehicle actual load prediction method based on mutual information and data blurring |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311849690.3A CN117494585B (en) | 2023-12-29 | 2023-12-29 | Commercial vehicle actual load prediction method based on mutual information and data blurring |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117494585A true CN117494585A (en) | 2024-02-02 |
| CN117494585B CN117494585B (en) | 2024-04-02 |
Family
ID=89683276
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311849690.3A Active CN117494585B (en) | 2023-12-29 | 2023-12-29 | Commercial vehicle actual load prediction method based on mutual information and data blurring |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117494585B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2505448A1 (en) * | 2011-03-28 | 2012-10-03 | Isbak Istanbul Ulasim Haberlesme Ve Guvenlik Teknolojileri Sanayi Ve Ticaret Anonim Sirketi | On-board real-time weight prediction system by using CAN data bus |
| CN110148230A (en) * | 2019-05-20 | 2019-08-20 | 兴民智通(武汉)汽车技术有限公司 | A kind of vehicle load-carrying prediction technique based on LSTM neural network |
| CN112819031A (en) * | 2021-01-04 | 2021-05-18 | 中国汽车技术研究中心有限公司 | Vehicle-mounted weight prediction method and system, electronic device and medium |
| CN114386892A (en) * | 2021-11-30 | 2022-04-22 | 江铃汽车股份有限公司 | Transportation vehicle load management method, system, storage medium and equipment |
| CN114644001A (en) * | 2021-05-20 | 2022-06-21 | 长城汽车股份有限公司 | Vehicle load prediction method and device, storage medium and vehicle |
| CN114954494A (en) * | 2022-06-14 | 2022-08-30 | 广西玉柴机器股份有限公司 | Heavy commercial vehicle load rapid estimation method |
| WO2023126023A1 (en) * | 2021-12-30 | 2023-07-06 | 中南大学 | Heavy-haul train and longitudinal dynamics traction operation optimization control system and method thereof |
| CN116662897A (en) * | 2023-08-01 | 2023-08-29 | 中汽研汽车检验中心（天津）有限公司 | Vehicle load prediction method based on hub data |
| CN116821775A (en) * | 2023-08-29 | 2023-09-29 | 陕西重型汽车有限公司 | Load estimation method based on machine learning |
-
2023
- 2023-12-29 CN CN202311849690.3A patent/CN117494585B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2505448A1 (en) * | 2011-03-28 | 2012-10-03 | Isbak Istanbul Ulasim Haberlesme Ve Guvenlik Teknolojileri Sanayi Ve Ticaret Anonim Sirketi | On-board real-time weight prediction system by using CAN data bus |
| CN110148230A (en) * | 2019-05-20 | 2019-08-20 | 兴民智通(武汉)汽车技术有限公司 | A kind of vehicle load-carrying prediction technique based on LSTM neural network |
| CN112819031A (en) * | 2021-01-04 | 2021-05-18 | 中国汽车技术研究中心有限公司 | Vehicle-mounted weight prediction method and system, electronic device and medium |
| CN114644001A (en) * | 2021-05-20 | 2022-06-21 | 长城汽车股份有限公司 | Vehicle load prediction method and device, storage medium and vehicle |
| CN114386892A (en) * | 2021-11-30 | 2022-04-22 | 江铃汽车股份有限公司 | Transportation vehicle load management method, system, storage medium and equipment |
| WO2023126023A1 (en) * | 2021-12-30 | 2023-07-06 | 中南大学 | Heavy-haul train and longitudinal dynamics traction operation optimization control system and method thereof |
| CN114954494A (en) * | 2022-06-14 | 2022-08-30 | 广西玉柴机器股份有限公司 | Heavy commercial vehicle load rapid estimation method |
| CN116662897A (en) * | 2023-08-01 | 2023-08-29 | 中汽研汽车检验中心（天津）有限公司 | Vehicle load prediction method based on hub data |
| CN116821775A (en) * | 2023-08-29 | 2023-09-29 | 陕西重型汽车有限公司 | Load estimation method based on machine learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117494585B (en) | 2024-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2022077693A1 (en) | Load prediction model training method and apparatus, storage medium, and device | |
| CN110610260B (en) | Driving energy consumption prediction system, method, storage medium and equipment | |
| CN110775065B (en) | Hybrid electric vehicle battery life prediction method based on working condition recognition | |
| CN111619545B (en) | Hybrid electric vehicle energy management method based on traffic information | |
| CN109815523B (en) | Multi-target differential evolution algorithm for train operation based on decomposition | |
| CN111047085B (en) | Hybrid vehicle working condition prediction method based on meta-learning | |
| CN113525396B (en) | Hybrid electric vehicle layered prediction energy management method integrating deep reinforcement learning | |
| CN110245390B (en) | Automobile engine oil consumption prediction method based on RS-BP neural network | |
| CN114103924A (en) | Energy management control method and device for hybrid vehicle | |
| Montazeri-Gh et al. | Driving condition recognition for genetic-fuzzy HEV control | |
| CN111198501B (en) | Method for determining fuel equivalent factor by RBF neural network | |
| Montazeri-Gh et al. | Intelligent approach for parallel HEV control strategy based on driving cycles | |
| CN114777192B (en) | Secondary network heat supply autonomous optimization regulation and control method based on data association and deep learning | |
| CN113479187B (en) | Layered different-step-length energy management method for plug-in hybrid electric vehicle | |
| CN113135113A (en) | Global SOC (System on chip) planning method and device | |
| CN112298155B (en) | Method for predicting energy consumption of hybrid power truck based on variable time domain model | |
| CN110807508A (en) | Bus peak load prediction method considering complex meteorological influence | |
| CN117494585B (en) | Commercial vehicle actual load prediction method based on mutual information and data blurring | |
| CN117465301A (en) | Fuel cell automobile real-time energy management method based on data driving | |
| CN116804706A (en) | Temperature prediction method and device for lithium battery of electric automobile | |
| CN114707239A (en) | Electric energy resource allocation planning method and device and electronic equipment | |
| CN111199304A (en) | Multi-target combustion optimization method based on data-driven fusion strategy | |
| CN114372640A (en) | Wind power prediction method based on fluctuation sequence classification correction | |
| EP3876045A1 (en) | A method of evaluating operation of at least one vehicle | |
| CN114298387A (en) | Hadoop-based energy consumption prediction method for parallelization BP neural network |
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 |