CN116644865A - Commercial vehicle fuel consumption prediction method, electronic equipment and storage medium - Google Patents
Commercial vehicle fuel consumption prediction method, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN116644865A CN116644865A CN202310926480.3A CN202310926480A CN116644865A CN 116644865 A CN116644865 A CN 116644865A CN 202310926480 A CN202310926480 A CN 202310926480A CN 116644865 A CN116644865 A CN 116644865A
- Authority
- CN
- China
- Prior art keywords
- data
- dynamic
- matrix
- distance
- representing
- 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 86
- 239000000446 fuel Substances 0.000 title claims abstract description 52
- 230000003068 static effect Effects 0.000 claims abstract description 76
- 230000004927 fusion Effects 0.000 claims abstract description 33
- 230000007246 mechanism Effects 0.000 claims abstract description 32
- 238000012545 processing Methods 0.000 claims abstract description 24
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 12
- 239000011159 matrix material Substances 0.000 claims description 127
- 230000008569 process Effects 0.000 claims description 48
- 239000013598 vector Substances 0.000 claims description 32
- 230000009466 transformation Effects 0.000 claims description 24
- 238000012549 training Methods 0.000 claims description 20
- 230000006870 function Effects 0.000 claims description 13
- 230000003044 adaptive effect Effects 0.000 claims description 7
- 238000000926 separation method Methods 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 5
- 230000001131 transforming effect Effects 0.000 claims description 2
- 239000002184 metal Substances 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 4
- 238000013473 artificial intelligence Methods 0.000 abstract description 3
- 239000003921 oil Substances 0.000 description 13
- 238000004590 computer program Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000000513 principal component analysis Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 230000000670 limiting effect Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000007499 fusion processing Methods 0.000 description 2
- 230000006855 networking Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000036961 partial effect Effects 0.000 description 2
- 230000002441 reversible effect Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000003197 catalytic effect Effects 0.000 description 1
- 238000010531 catalytic reduction reaction Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000000295 fuel oil Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000012847 principal component analysis method Methods 0.000 description 1
- 230000002829 reductive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Game Theory and Decision Science (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the field of artificial intelligence, and discloses a commercial vehicle fuel consumption prediction method, electronic equipment and a storage medium, wherein the method comprises the following steps: acquiring current vehicle data of a commercial vehicle, and determining initial dynamic data and initial static data from the current vehicle data according to dynamic factors and static factors; processing the initial dynamic data based on a pre-trained attention mechanism for fusing the relative positions and the distances to obtain dynamic characteristics to be fused; processing the initial static data based on a factorizer algorithm to obtain static features to be fused; according to the dynamic characteristics to be fused and the static characteristics to be fused, carrying out self-adaptive characteristic fusion to obtain target characteristics; and inputting the target characteristics into the multi-layer sensor which is trained in advance to obtain oil consumption prediction data. The method and the device can achieve the effect of improving the accuracy of the fuel consumption prediction of the commercial vehicle.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to a commercial vehicle fuel consumption prediction method, electronic equipment and a storage medium.
Background
With the rapid development of the internet of vehicles and artificial intelligence technology, big data analysis based on vehicle operation data is widely researched and applied. Commercial vehicles bear main road transportation tasks, and the problems of energy shortage and environmental pollution are caused by the continuous increase of the maintenance quantity and the increase of fuel oil emission of the commercial vehicles. Therefore, the reliable fuel consumption prediction model is established, and has important research significance for assisting driving and improving fuel utilization rate.
The existing fuel consumption prediction analysis method mainly comprises the steps of exploring the factors related to fuel consumption in vehicle operation dynamic data, and then predicting the fuel consumption by utilizing a multi-element linear model or a simple time sequence model. However, the relationship between commercial fuel consumption and numerous other influencing factors is complex and nonlinear, vehicle operation data is also dynamically changed, and static data such as tire specifications, vehicle types and the like are also crucial to fuel consumption prediction. Therefore, the existing fuel consumption prediction analysis method ignores a lot of static data and also ignores the time dependence of dynamic data, which can lead to the problems of strong subjectivity, large limitation and inaccurate predicted fuel consumption.
In view of this, the present invention has been made.
Disclosure of Invention
In order to solve the technical problems, the invention provides a commercial vehicle fuel consumption prediction method, electronic equipment and a storage medium, which realize the effect of improving the accuracy of commercial vehicle fuel consumption prediction.
The embodiment of the invention provides a method for predicting oil consumption of a commercial vehicle, which comprises the following steps:
acquiring current vehicle data of a commercial vehicle, and determining initial dynamic data and initial static data from the current vehicle data according to dynamic factors and static factors; the dynamic factors and the static factors are determined by main component analysis of sample vehicle data and sample oil consumption data;
processing the initial dynamic data based on a pre-trained attention mechanism for fusing the relative positions and the distances to obtain dynamic characteristics to be fused;
processing the initial static data based on a factorizer algorithm to obtain static features to be fused;
according to the dynamic characteristics to be fused and the static characteristics to be fused, carrying out self-adaptive characteristic fusion to obtain target characteristics;
inputting the target characteristics into a multi-layer sensor which is trained in advance to obtain oil consumption prediction data;
The multi-layer sensor is trained based on the sample vehicle data and the sample fuel consumption data, and the attention mechanism fusing the relative position and the distance is an attention mechanism fusing the relative position relationship and the distance relationship between the data on time sequence.
The embodiment of the invention provides electronic equipment, which comprises:
a processor and a memory;
the processor is configured to execute the steps of the commercial vehicle fuel consumption prediction method according to any one of the embodiments by calling a program or an instruction stored in the memory.
The embodiment of the invention provides a computer readable storage medium, which stores a program or instructions for causing a computer to execute the steps of the commercial vehicle fuel consumption prediction method according to any embodiment.
The embodiment of the invention has the following technical effects:
according to the dynamic factors and the static factors, initial dynamic data and initial static data are determined from the current vehicle data of the commercial vehicle so as to analyze the static factors and the dynamic factors at the same time, further, based on a concentration mechanism of fusion relative positions and distances which are completed through pre-training, the initial dynamic data are processed to obtain dynamic features to be fused, based on a factorizer algorithm, the initial static data are processed to obtain the static features to be fused, the features are extracted according to the characteristics of the dynamic factors and the static factors, and according to the dynamic features to be fused and the static features to be fused, adaptive feature fusion is conducted to obtain target features, the dynamic features to be fused and the static features to be fused are fused, so that the accuracy of subsequent prediction is convenient to improve, the target features are input into a multi-layer perceptron which is completed through pre-training, the oil consumption prediction data is obtained, the synchronous extraction of time relevance of the static features and the dynamic features is achieved, and the oil consumption prediction accuracy of the commercial vehicle is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for predicting fuel consumption of a commercial vehicle provided by an embodiment of the invention;
FIG. 2 is a schematic diagram of a relative position matrix according to an embodiment of the present invention;
FIG. 3 is a flowchart of another fuel consumption prediction method for a commercial vehicle according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below. It will be apparent that the described embodiments are only some, but not all, embodiments of the invention. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the invention, are within the scope of the invention.
The method for predicting the fuel consumption of the commercial vehicle is mainly suitable for predicting the fuel consumption of a period of time in the future through analysis of dynamic factors and static factors of the commercial vehicle. The commercial vehicle fuel consumption prediction method provided by the embodiment of the invention can be executed by the electronic equipment.
Fig. 1 is a flowchart of a method for predicting fuel consumption of a commercial vehicle according to an embodiment of the present invention. Referring to fig. 1, the method for predicting the fuel consumption of the commercial vehicle specifically includes:
s110, acquiring current vehicle data of the commercial vehicle, and determining initial dynamic data and initial static data from the current vehicle data according to dynamic factors and static factors.
The dynamic factors and the static factors are determined by principal component analysis of the sample vehicle data and the sample oil consumption data. The sample vehicle data and the sample fuel consumption data are data acquired in advance and used for training various required parameters in the commercial vehicle fuel consumption prediction method, and the sample fuel consumption data can comprise sample current fuel consumption data and sample future fuel consumption data. The current vehicle data is internet of vehicles data of the commercial vehicle, and may include dynamic vehicle data that changes with time and static vehicle data that does not change with time. The initial dynamic data is a portion of the current vehicle data corresponding to the dynamic factor. The initial static data is a portion of the current vehicle data corresponding to the static factor.
Specifically, current vehicle data of the commercial vehicle is acquired based on a vehicle networking platform or in real time. And taking the partial data corresponding to the dynamic factors in the current vehicle data as initial dynamic data, and taking the partial data corresponding to the static factors in the current vehicle data as initial static data.
Optionally, principal component analysis is performed on the sample vehicle data according to the sample fuel consumption data, and the static factor and the dynamic factor are determined. Illustratively, the dynamic factors determined by principal component analysis include: vehicle speed, engine speed, net output torque, friction torque, air intake, barometric pressure, water temperature, tank level, DPF (Diesel Particulate Filter, diesel particulate trap) differential pressure, SCR (Selective Catalytic Reducation, selective catalytic reduction) inlet temperature, SCR outlet temperature, accumulated mileage, and 18 acceleration, speed range, vehicle speed variance, engine speed variance, torque variance, water temperature variance are calculated therefrom; the static factors determined by principal component analysis include: the weight of the vehicle, the vehicle type, the engine displacement, the tire specification, the wind resistance coefficient and the windward area are 6.
S120, processing initial dynamic data based on a pre-trained attention mechanism for fusing the relative positions and the distances to obtain dynamic features to be fused.
The attention mechanism fusing the relative position and the distance is obtained based on the combined training of the sample vehicle data and the sample oil consumption data, and the attention mechanism fusing the relative position relationship and the distance relationship between the time series data. The dynamic characteristics to be fused are output data obtained by fusing the attention mechanism of the relative position and the distance after processing the initial dynamic data, and the output data is used for carrying out self-adaptive characteristic fusion processing subsequently.
Specifically, initial dynamic data is input into a pre-trained attention mechanism for fusing relative positions and distances, and the obtained output data is the dynamic characteristics to be fused.
Based on the above example, the attention mechanism fusing the relative position and the distance has multiple layers, and the initial dynamic data can be processed based on the attention mechanism fusing the relative position and the distance which is trained in advance through the following steps to obtain the dynamic characteristics to be fused:
step one, the initial dynamic data is used as the dynamic data to be processed, and the current layer number is determined.
Wherein the dynamic data to be processed is data for a subsequent loop processing. The current layer number may be data recorded starting from 0; under the structure of overlapping the attention mechanisms of the relative positions and the distances by multi-layer fusion, the current layer number is used for indicating the currently overlapped layer level.
Based on the first weight matrix, the second weight matrix and the third weight matrix of the current layer, respectively performing linear transformation on the dynamic data to be processed to obtain first dynamic data, second dynamic data and third dynamic data, and performing head separation processing on the first dynamic data, the second dynamic data and the third dynamic data according to a preset number to obtain a head separation dynamic sub-data group with a preset number.
The first weight matrix, the second weight matrix and the third weight matrix are matrices for performing linear transformation on dynamic data to be processed, and each layer of the first weight matrix, the second weight matrix and the third weight matrix which are fused with the attention mechanism of the relative position and the distance are obtained based on joint training of sample vehicle data and sample oil consumption data. The first dynamic data, the second dynamic data and the third dynamic data are products of the first weight matrix, the second weight matrix and the third weight matrix and the dynamic data to be processed respectively. The preset number is the number of heads at the time of the head separation process. The sub-head dynamic sub-data group is a data group after the sub-head processing of the first dynamic data, the second dynamic data and the third dynamic data, and each sub-head dynamic sub-data group comprises a group of the first dynamic sub-data, the second dynamic sub-data and the third dynamic sub-data.
Specifically, the product of the first weight matrix and the dynamic data to be processed is used as the first dynamic data, the product of the second weight matrix and the dynamic data to be processed is used as the second dynamic data, and the product of the third weight matrix and the dynamic data to be processed is used as the third dynamic data. And further, the first dynamic data, the second dynamic data and the third dynamic data are subjected to head separation processing, the first dynamic data is divided into a preset number of first dynamic sub-data, the second dynamic data is divided into a preset number of second dynamic sub-data, and the third dynamic data is divided into a preset number of third dynamic sub-data. Further, the first dynamic sub-data, the second dynamic sub-data and the third dynamic sub-data corresponding to each group of the heads are used as a group of dynamic sub-data groups of the heads, and accordingly, the head dynamic sub-data groups with the preset number of groups can be obtained.
Illustratively, the first dynamic data, the second dynamic data, and the third dynamic data are determined by the following formulas:
wherein ,Xfor the dynamic data to be processed,Qas a result of the first dynamic data being,Kas a result of the second dynamic data,Vas a result of the third dynamic data, the data,for the first weight matrix,/a >For the second weight matrix,/a>And is a third weight matrix.
And thirdly, determining a time sequence data distance matrix according to the distance between the data corresponding to any two moments in the dynamic data to be processed.
The time sequence data distance matrix is a matrix formed by the distances between the data corresponding to any two moments and is used for describing the distances between the data at all the moments.
Specifically, the distance calculation is carried out on the data corresponding to any two moments in the dynamic data to be processed, and each calculated distance forms a time sequence data distance matrix.
Based on the above example, the time sequence data distance matrix can be determined according to the distance between the data corresponding to any two moments in the dynamic data to be processed in the following manner:
determining Euclidean distance and Manhattan distance corresponding to the data corresponding to any two moments in the dynamic data to be processed;
and determining a time sequence data distance matrix according to the pre-trained first distance transformation matrix corresponding to the Euclidean distance, the second distance transformation matrix corresponding to the Manhattan distance, the Euclidean distance between the data corresponding to any two moments in the dynamic data to be processed and the Manhattan distance.
The first distance transformation matrix is a pre-trained matrix used for weighting Euclidean distances between data corresponding to any two moments. The second distance conversion matrix is a pre-trained matrix for weighting manhattan distances between data corresponding to any two times.
Specifically, in order to better capture the variation trend of the sequence node, for the data corresponding to every arbitrary two moments, the euclidean distance and the manhattan distance between the data corresponding to the two moments can be calculated. Further, each euclidean distance is weighted by a first distance transformation matrix, each manhattan distance is weighted by a second distance transformation matrix, and the weighted euclidean distances and the weighted manhattan distances are summed to obtain a time sequence data distance matrix.
Based on the above example, the euclidean distance and the manhattan distance corresponding to the data corresponding to any two moments can be determined by the following formulas:
wherein ,Ed(X i ，X j ) Representing time of day in dynamic data to be processediData and time of day of (a)jIs a euclidean distance between the data of (a),Md(X i ，X j ) Representing time of day in dynamic data to be processed iData and time of day of (a)jIs a manhattan distance between the data of (a),X i representing time of day in dynamic data to be processediIs a function of the data of (a),X j representing time of day in dynamic data to be processedjIs a function of the data of (a),nrepresenting the data dimension for each instant in the dynamic data to be processed,x im representation ofX i Middle (f)mThe data of the plurality of data,x jm representation ofX j Middle (f)mData.
Based on the above example, the time series data distance matrix may be determined by the following formula:
wherein ,Ed(X i ，X j ) Representing time of day in dynamic data to be processediData and time of day of (a)jIs a euclidean distance between the data of (a),Md(X i ，X j ) Representing time of day in dynamic data to be processediData and time of day of (a)jIs a manhattan distance between the data of (a),D ij representing elements of the time series data distance matrix corresponding to time instant i and time instant j, i.e. the firstiLine 1jThe elements of the column are arranged such that,representing a first distance transformation matrix,/a>Representing a second distance transformation matrix. Wherein (1)> and />The second distance transformation matrix is a pre-trained matrix.
And step four, aiming at each group of sub-head dynamic sub-data group, determining the dynamic characteristics to be processed corresponding to the sub-head dynamic sub-data group based on the sub-head dynamic sub-data group, the relative position matrix and the time sequence data distance matrix.
The relative position matrix is obtained based on the combined training of the sample vehicle data and the sample fuel consumption data, the time dimension of the relative position matrix is 2T-1, T is the time dimension of the initial dynamic data, and the schematic diagram of the relative position matrix is shown in fig. 2, where the time dimension of the relative position matrix can be understood as 2 (T-1) +1=2t-1. The dynamic characteristics to be processed are the processing results of the sub-head dynamic sub-data sets after the relative position matrix and the time sequence distance matrix are processed. It will be appreciated that the relative position matrix has a time dimension of 2T-1 and a characteristic dimension of N, and thus the relative position matrix has a size of (2T-1) ×n, as shown in fig. 2, and can represent relative positions in the front and rear directions, where the right half of the relative position 0 represents a forward relative position and the left half represents a reverse relative position.
Specifically, for each group of sub-head dynamic sub-data sets, the dynamic characteristics to be processed corresponding to the sub-head dynamic sub-data sets may be determined in the same manner, and one group is taken as an example for illustration. And processing the first dynamic sub data, the second dynamic sub data and the third dynamic sub data in the group of sub head dynamic sub data based on the relative position matrix and the time sequence data distance matrix, and taking the processed result as the dynamic characteristics to be processed corresponding to the sub head dynamic sub data group.
Based on the above example, the dynamic characteristics to be processed corresponding to the split dynamic sub-data set may be determined by the following formulas:
wherein ,indicating time of dayiAnd time of dayjThe importance coefficient of the two-dimensional space,Q [i,:] representing the time and the middle of the first dynamic sub-dataiCorresponding parts, i.e. the firstiThe number of rows of the device is,K T [:,j] representing transpose of second dynamic sub-data and time of dayjCorresponding parts, i.e. the firstjColumn (S)/(S)>Representing time of day in a relative position matrixjRelative to the moment of timeiIs used to determine the relative position vector of (a),d k representing a preset variance; divided by->The importance coefficient is scaled to prevent the individual value from becoming too large.
wherein ,indicating time of dayiAnd time of dayjImportance coefficient between->Indicating time of day iAnd time of daykThe importance coefficient of the two-dimensional space,a ij indicating time of dayiAnd time of dayjNormalized importance coefficients in between, exp (·) represents an exponential function based on a natural constant e, t represents the number of columns of importance coefficients, i.e., the total number of moments;
wherein ,a ij indicating time of dayiAnd time of dayjThe normalized importance coefficient between them,Z i representing the dynamic characteristics to be processed corresponding to the split dynamic sub-data set,V [j,:] representing the time and the middle of the third dynamic sub-datajCorresponding data, i.e. the firstjAnd (3) row.
It should be noted that, the conventional attention mechanism is used for learning the context of time sequence, that is, each time corresponds to one feature vector, and after the feature vectors of a plurality of continuous time are input into the attention mechanism, a new time sequence can be obtained, where the feature vector of each time in the new time sequence is related to the feature vectors of other time. However, in the conventional attention mechanism, the relative positional relationship between the respective moments in the time series is ignored, for example: who is closer to who is further from who is. That is, the feature vectors of a plurality of consecutive moments are input into the attention mechanism network in a disordered order, and the feature corresponding to each moment in the obtained new sequence is unchanged. However, in the course of running a vehicle, a sudden change in signal or the like often occurs, and the relative positional relationship between the respective moments is very important for the sudden change position. Therefore, the invention improves the attention mechanism, and uses the attention mechanism fusing the relative position and the distance to represent the position relationship between the feature vectors at each moment, namely the time far-near relationship through the relative position matrix. This requires that the features at each time are input in the same positional relationship every time during training, and cannot be out of order. The relative position matrix is initialized randomly, and parameters are fixed after training is completed and used for representing the time positions between input vectors.
And fifthly, connecting the dynamic characteristics to be processed to obtain single dynamic characteristics, and updating the current layer number.
The single dynamic feature is a feature of each dynamic feature to be processed after connection and is used for subsequent cyclic calculation.
Specifically, the to-be-processed dynamic features corresponding to each component head dynamic sub-data group can be connected according to a reverse processing mode during the component head processing, so as to obtain the connected single dynamic features. In order to improve the robustness, the single dynamic characteristic can be subjected to linear transformation to obtain a new single dynamic characteristic. After the single dynamic feature is obtained, the current layer number is added by one and used as a new current layer number.
On the basis of the above example, the dynamic characteristics to be processed can be connected through the following formula, so as to obtain single dynamic characteristics:
wherein ,a single dynamic characteristic is represented and is used to represent,Qthe first dynamic data is represented by a first set of data,Kthe second dynamic data is represented by a representation of the second dynamic data,Vrepresenting third dynamic data,/->Representing a matrix of relative positions of the two elements,Drepresenting a distance matrix of time series data,Concat(. Cndot.) represents the function of the connection,head 1 ，…，head s representing the dynamic characteristics of each to-be-processed,sgroup number representing the dynamic feature to be processed, +.>Representing a multi-headed join transform matrix. The multi-head connection transformation matrix is also obtained based on sample vehicle data and sample fuel consumption data through combined training.
Step six, determining the single dynamic characteristic as the dynamic characteristic to be fused under the condition that the current layer number is equal to the preset layer number, updating the single dynamic characteristic as the dynamic data to be processed under the condition that the current layer number is smaller than the preset layer number, and returning to execute the first weight matrix, the second weight matrix and the third weight matrix based on the current layer to respectively perform linear transformation on the dynamic data to be processed to obtain the first dynamic data, the second dynamic data and the third dynamic data.
Wherein the preset layer number is a preset super parameter. The multi-layer structure can ensure that the problem of gradient disappearance is avoided while deepening the depth (layer number) of the network, and ensure that the model prediction precision at least cannot be reduced along with the increase of the depth.
Specifically, when the current layer number is equal to the preset layer number, it indicates that the superposition operation of the multi-layer improved attention mechanism has been completed, so that the single dynamic feature obtained currently can be used as the dynamic feature to be fused for the subsequent adaptive feature fusion. And under the condition that the current layer number is smaller than the preset layer number, the superposition feature calculation still needs to be continued, so that the single dynamic feature can be used as new dynamic data to be processed, the first weight matrix, the second weight matrix and the third weight matrix based on the current layer are returned to be executed, and the first dynamic data, the second dynamic data and the third dynamic data are obtained by respectively carrying out linear transformation on the dynamic data to be processed, so that the new single dynamic feature and the new current layer number are obtained and used for next judgment and confirmation.
S130, processing the initial static data based on a factorization machine algorithm to obtain the static features to be fused.
The factorizer (Factorization Machine, FM) algorithm considers the interaction between the features, is a nonlinear model, and can automatically learn the weights of different features, so that the factorizer (Factorization Machine, FM) algorithm has higher calculation efficiency compared with the SVM (Support Vector Machine ) algorithm. The static features to be fused are output data processed through an FM algorithm and are used for carrying out self-adaptive feature fusion processing subsequently.
Specifically, the initial static data is processed through an FM algorithm, and the extracted features are the static features to be fused.
And S140, carrying out self-adaptive feature fusion according to the dynamic features to be fused and the static features to be fused to obtain target features.
The target features are features obtained by adaptively fusing the features of the dynamic dimension and the static dimension.
Specifically, according to a preset self-adaptive feature fusion mode, the dynamic features to be fused and the static features to be fused are fused, and the obtained fusion result is the target feature.
Based on the above example, the adaptive feature fusion can be performed according to the dynamic feature to be fused and the static feature to be fused through the following steps to obtain the target feature:
Step one, carrying out time dimension average on dynamic characteristics to be fused to obtain a first process characteristic, and transforming the static characteristics to be fused according to the dimension of the first process characteristic to obtain a second process characteristic.
The first process feature is a feature vector obtained by carrying out averaging calculation on the dynamic features to be fused in the time dimension. The second process vector is a feature vector obtained by dimension transformation of the static features to be fused. The vector dimensions of the first process feature and the second process feature are the same.
Specifically, the dynamic features to be fused are subjected to average processing in the time dimension, namely, the sum of features corresponding to all moments in the dynamic features to be fused is divided by the number of moments for which the dynamic features to be fused are located, and the processed vector value is the first process feature. And carrying out dimension transformation on the static features to be fused according to the dimension of the first process feature to obtain a vector value identical to the dimension of the first process feature, namely the second process feature.
And step two, determining a target weight vector according to the first process feature, the first fusion weight matrix corresponding to the first process feature, the second fusion weight matrix corresponding to the second process feature and the preset bias matrix.
The first fusion weight matrix is a matrix which is obtained through pre-training and used for weighting the first process features. The second fusion weight matrix is a pre-trained matrix for weighting the second process feature. The preset bias matrix is a matrix which is obtained through pre-training and used for carrying out bias adjustment when the first process characteristic and the second process characteristic are processed. The target weight vector is the fusion proportion of the dynamic characteristic and the static characteristic when the self-adaptive characteristic fusion is carried out subsequently.
Specifically, the first process feature is multiplied by a first fusion weight matrix to obtain a first product, and the second process feature is multiplied by a second fusion weight matrix to obtain a second product. Further, the sum of the first product, the second product and the preset bias matrix is used as a target weight vector.
Based on the above example, the target weight vector may be determined according to the first process feature, the first fused weight matrix corresponding to the first process feature, the second fused weight matrix corresponding to the second process feature, and the preset bias matrix by the following formula:
wherein ,Zthe weight vector of the object is represented as,representing a first process feature- >Representing a first fused weight matrix corresponding to a first process feature,/a first process feature>Representing a second process feature->Representing a second fused weight matrix corresponding to a second process feature,b z representing a preset bias matrix->、/>Andb z all are learning parameters, and are added with->Is a sigmod function.
Through the formula, each element in the target weight vector can be positioned between 0 and 1 and used for adaptively controlling the proportion of dynamic and static characteristics to serve as the weight of the subsequent adaptive characteristic fusion.
And thirdly, determining target characteristics according to the target weight vector, the dynamic characteristics to be fused and the static characteristics to be fused.
Specifically, weighting the dynamic features to be fused by using the target weight vector, weighting the static features to be fused by using 1 minus the target weight vector, and adding the results of the two weighting processes to obtain the target feature.
Based on the above example, the target feature may be determined according to the target weight vector, the dynamic feature to be fused, and the static feature to be fused by the following formula:
wherein ,Zthe weight vector of the object is represented as,representing a first process feature->A second process characteristic is indicated and is indicated,Hrepresenting the target feature- >Representing the hadamard product operation.
By the method, dynamic and static characteristics can be adaptively fused, and a hidden state, namely a target characteristic, is obtained.
And S150, inputting target characteristics into the multi-layer sensor which is trained in advance to obtain oil consumption prediction data.
The multi-layer perceptron is obtained based on sample vehicle data and sample oil consumption data through combined training. The fuel consumption prediction data is fuel consumption data of a subsequent period predicted based on current vehicle data of the commercial vehicle.
Specifically, the target features are input into the structure of the multi-layer sensor, so that the fuel consumption sequence of a period of time in the future can be predicted, namely fuel consumption prediction data. In training the multi-layer perceptron, training may be performed by a smoth L1 loss function.
Optionally, various learning parameters in the operation can be obtained through combined training of the sample vehicle data and the sample fuel consumption data. The learning parameters include: the first fusion weight matrix, the second fusion weight matrix, the bias matrix and the parameters in the multi-layer perceptron in the adaptive feature fusion are used for realizing the adaptation of each module to the fuel consumption data.
Fig. 3 is a flowchart of another method for predicting fuel consumption of a commercial vehicle according to an embodiment of the present invention. Referring to fig. 3, the method for predicting the fuel consumption of the commercial vehicle specifically includes:
and step 1, finding out relevant factors affecting oil consumption, namely dynamic factors and static factors, according to the vehicle networking data by a principal component analysis method.
And 2, extracting dynamic characteristics (dynamic characteristics to be fused) from data (initial dynamic data) corresponding to dynamic factors through an improved attention mechanism (attention mechanism fusing relative positions and distances), and extracting static characteristics (static characteristics to be fused) from data (initial static data) corresponding to static factors through an FM algorithm.
And 3, carrying out self-adaptive weighted fusion on the dynamic characteristics and the static characteristics through a self-adaptive fusion module (self-adaptive characteristic fusion) to obtain target characteristics.
And 4, predicting the oil consumption (oil consumption prediction data) for a period of time in the future through processing the target characteristics by the multi-layer sensor structure.
The embodiment has the following technical effects: according to the dynamic factors and the static factors, initial dynamic data and initial static data are determined from the current vehicle data of the commercial vehicle so as to analyze the static factors and the dynamic factors at the same time, further, based on a concentration mechanism of fusion relative positions and distances which are completed through pre-training, the initial dynamic data are processed to obtain dynamic characteristics to be fused, based on a factorizer algorithm, the initial static data are processed to obtain the static characteristics to be fused, characteristics are extracted according to the characteristics of the dynamic factors and the static factors, and according to the dynamic characteristics to be fused and the static characteristics to be fused, adaptive characteristic fusion is conducted to obtain target characteristics, so that the dynamic characteristics to be fused and the static characteristics to be fused are fused, the accuracy of subsequent prediction is convenient to improve, the target characteristics are input into a multi-layer sensor which is completed through pre-training, the predicted fuel consumption data is obtained, the time relevance of the static factors and the dynamic factors is considered is realized, and the prediction accuracy of the commercial fuel consumption is improved.
Fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention. As shown in fig. 4, electronic device 400 includes one or more processors 401 and memory 402.
The processor 401 may be a Central Processing Unit (CPU) or other form of processing unit having data processing capabilities and/or instruction execution capabilities and may control other components in the electronic device 400 to perform desired functions.
Memory 402 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random Access Memory (RAM) and/or cache memory (cache), and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like. One or more computer program instructions may be stored on the computer readable storage medium that may be executed by the processor 401 to implement the commercial vehicle fuel consumption prediction method and/or other desired functions of any of the embodiments of the present invention described above. Various content such as initial arguments, thresholds, etc. may also be stored in the computer readable storage medium.
In one example, the electronic device 400 may further include: an input device 403 and an output device 404, which are interconnected by a bus system and/or other forms of connection mechanisms (not shown). The input device 403 may include, for example, a keyboard, a mouse, and the like. The output device 404 may output various information to the outside, including early warning prompt information, braking force, etc. The output device 404 may include, for example, a display, speakers, a printer, and a communication network and remote output devices connected thereto, etc.
Of course, only some of the components of the electronic device 400 that are relevant to the present invention are shown in fig. 4 for simplicity, components such as buses, input/output interfaces, etc. are omitted. In addition, electronic device 400 may include any other suitable components depending on the particular application.
In addition to the methods and apparatus described above, embodiments of the present invention may also be a computer program product comprising computer program instructions which, when executed by a processor, cause the processor to perform the steps of the commercial vehicle fuel consumption prediction method provided by any of the embodiments of the present invention.
The computer program product may write program code for performing operations of embodiments of the present invention in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device, partly on a remote computing device, or entirely on the remote computing device or server.
Furthermore, embodiments of the present application may also be a computer-readable storage medium, on which computer program instructions are stored, which, when executed by a processor, cause the processor to perform the steps of the commercial vehicle fuel consumption prediction method provided by any embodiment of the present application.
The computer readable storage medium may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium would include the following: an electrical connection having one or more wires, a portable disk, a hard disk, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of the present application. As used in this specification, the terms "a," "an," "the," and/or "the" are not intended to be limiting, but rather are to be construed as covering the singular and the plural, unless the context clearly dictates otherwise. The terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method or apparatus comprising such elements.
It should also be noted that the positional or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. are based on the positional or positional relationship shown in the drawings, are merely for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the apparatus or element in question must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present invention. Unless specifically stated or limited otherwise, the terms "mounted," "connected," and the like are to be construed broadly and may be, for example, fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.
Claims (10)
1. The utility model provides a commercial vehicle oil consumption prediction method which is characterized by comprising the following steps:
acquiring current vehicle data of a commercial vehicle, and determining initial dynamic data and initial static data from the current vehicle data according to dynamic factors and static factors; the dynamic factors and the static factors are determined by main component analysis of sample vehicle data and sample oil consumption data;
processing the initial dynamic data based on a pre-trained attention mechanism for fusing the relative positions and the distances to obtain dynamic characteristics to be fused;
processing the initial static data based on a factorizer algorithm to obtain static features to be fused;
according to the dynamic characteristics to be fused and the static characteristics to be fused, carrying out self-adaptive characteristic fusion to obtain target characteristics;
inputting the target characteristics into a multi-layer sensor which is trained in advance to obtain oil consumption prediction data;
the multi-layer sensor is trained based on the sample vehicle data and the sample fuel consumption data, and the attention mechanism fusing the relative position and the distance is an attention mechanism fusing the relative position relationship and the distance relationship between the data on time sequence.
2. The method of claim 1, wherein the attention mechanism fusing relative position and distance has multiple layers; the method for processing the initial dynamic data based on the attention mechanism fusing the relative positions and the distances which are completed through pre-training, to obtain dynamic characteristics to be fused, comprises the following steps:
taking the initial dynamic data as dynamic data to be processed, and determining the current layer number;
based on a first weight matrix, a second weight matrix and a third weight matrix of a current layer, respectively performing linear transformation on the dynamic data to be processed to obtain first dynamic data, second dynamic data and third dynamic data, and performing head separation processing on the first dynamic data, the second dynamic data and the third dynamic data according to a preset number to obtain a head separation dynamic sub-data group with a preset number; each group of dynamic sub-data groups of the sub-head comprises a group of first dynamic sub-data, a group of second dynamic sub-data and a group of third dynamic sub-data;
determining a time sequence data distance matrix according to the distance between the data corresponding to any two moments in the dynamic data to be processed;
determining the dynamic characteristics to be processed corresponding to the sub-head dynamic sub-data groups according to the sub-head dynamic sub-data groups, the relative position matrix and the time sequence data distance matrix;
Connecting each dynamic characteristic to be processed to obtain a single dynamic characteristic, and updating the current layer number;
determining the single dynamic characteristic as a dynamic characteristic to be fused when the current layer number is equal to a preset layer number, updating the single dynamic characteristic as dynamic data to be processed when the current layer number is smaller than the preset layer number, and returning to execute the first weight matrix, the second weight matrix and the third weight matrix based on the current layer to respectively perform linear transformation on the dynamic data to be processed to obtain first dynamic data, second dynamic data and third dynamic data;
the first weight matrix, the second weight matrix, the third weight matrix, the relative position matrix and the multilayer perceptron of the attention mechanism fusing the relative position and the distance are obtained by combined training based on the sample vehicle data and the sample fuel consumption data, and the time dimension of the relative position matrix is 2T-1, and T is the time dimension of the initial dynamic data.
3. The method according to claim 2, wherein the determining the time series data distance matrix according to the distance between the data corresponding to any two moments in the dynamic data to be processed includes:
Determining Euclidean distance and Manhattan distance corresponding to the data corresponding to any two moments in the dynamic data to be processed;
and determining a time sequence data distance matrix according to a first distance transformation matrix which is trained in advance and corresponds to the Euclidean distance, a second distance transformation matrix which corresponds to the Manhattan distance, the Euclidean distance between the data corresponding to any two moments in the dynamic data to be processed and the Manhattan distance.
4. A method according to claim 3, wherein determining the euclidean distance and the manhattan distance corresponding to the data corresponding to the arbitrary two moments comprises:
the Euclidean distance and the Manhattan distance corresponding to the data corresponding to any two moments are determined through the following formulas:
wherein ,Ed(X i ，X j ) Representing time of day in dynamic data to be processediData and time of day of (a)jIs a euclidean distance between the data of (a),Md(X i ，X j ) Representing time of day in dynamic data to be processediData and time of day of (a)jIs a manhattan distance between the data of (a),X i representing time of day in dynamic data to be processediIs a function of the data of (a),X j representing time of day in dynamic data to be processedjIs a function of the data of (a),nrepresenting the data dimension for each instant in the dynamic data to be processed, x im Representation ofX i Middle (f)mThe data of the plurality of data,x jm representation ofX j Middle (f)mData;
correspondingly, the determining a time sequence data distance matrix according to a first distance transformation matrix which is trained in advance and corresponds to the Euclidean distance, a second distance transformation matrix which corresponds to the Manhattan distance, the Euclidean distance between the data corresponding to any two moments in the dynamic data to be processed, and the Manhattan distance comprises the following steps:
determining a time series data distance matrix by the following formula:
wherein ,D ij representing time sequence data distance matrix and timeiAnd time of dayjThe corresponding element(s),representing a first distance transformation matrix,/a>Representing a second distance transformation matrix.
5. The method of claim 2, wherein the determining the dynamic feature to be processed corresponding to the split dynamic sub-data set based on the split dynamic sub-data set, the relative position matrix, and the time series data distance matrix comprises:
determining the dynamic characteristics to be processed corresponding to the sub-head dynamic data group through the following formulas:
wherein ,indicating time of dayiAnd time of dayjThe importance coefficient of the two-dimensional space,Q [i,:] representing the time and the middle of the first dynamic sub-dataiThe corresponding portion of the first and second metal plates,K T [:,j] representing transpose of second dynamic sub-data and time of day jCorresponding part->Representing time of day in a relative position matrixjRelative to the moment of timeiIs used to determine the relative position vector of (a),d k representing a preset variance;
wherein ,indicating time of dayiAnd time of dayjNormalized importance coefficient between +.>Indicating time of dayiAnd time of daykThe importance coefficient between the two values, exp (·) represents an exponential function based on a natural constant e, and t represents the number of columns of the importance coefficient;
wherein ,Z i representing the dynamic characteristics to be processed corresponding to the headedness sub-data set,V [j,:] representing the time and the middle of the third dynamic sub-datajCorresponding data.
6. The method according to claim 2, wherein the connecting each dynamic feature to be processed to obtain a single dynamic feature includes:
the single dynamic feature is determined by the following formula:
wherein ,a single dynamic characteristic is represented and is used to represent,Qthe first dynamic data is represented by a first set of data,Kthe second dynamic data is represented by a representation of the second dynamic data,Vrepresenting third dynamic data,/->Representing a matrix of relative positions of the two elements,Drepresenting a distance matrix of time series data,Concat(. Cndot.) represents the function of the connection,head 1 ，…，head s representing the dynamic characteristics of each to-be-processed,sgroup number representing the dynamic feature to be processed, +.>Representing a multi-headed join transform matrix.
7. The method of claim 1, wherein the performing adaptive feature fusion according to the dynamic feature to be fused and the static feature to be fused to obtain the target feature includes:
Performing time dimension average on the dynamic characteristics to be fused to obtain a first process characteristic, and transforming the static characteristics to be fused according to the dimension of the first process characteristic to obtain a second process characteristic;
determining a target weight vector according to the first process feature, a first fusion weight matrix corresponding to the first process feature, the second process feature, a second fusion weight matrix corresponding to the second process feature and a preset bias matrix;
determining target features according to the target weight vector, the dynamic features to be fused and the static features to be fused;
the first fusion weight matrix, the second fusion weight matrix and the preset bias matrix are used for fusing the attention mechanism of the relative position and the distance and the multi-layer perceptron, and are obtained based on the sample vehicle data and the sample fuel consumption data through combined training.
8. The method of claim 7, wherein the determining the target weight vector from the first process feature, the first fused weight matrix corresponding to the first process feature, the second fused weight matrix corresponding to the second process feature, and a preset bias matrix comprises:
The target weight vector is determined by the following formula:
wherein ,Zthe weight vector of the object is represented as,representing a first process feature->Representing a first fused weight matrix corresponding to said first process feature,/for>Representing a second process feature->Representing a second fused weight matrix corresponding to the second process feature,b z representing a preset bias matrix->、/>Andb z all are learning parameters, and are added with->Is a sigmod function;
correspondingly, the determining the target feature according to the target weight vector, the dynamic feature to be fused and the static feature to be fused includes:
the target feature is determined by the following formula:
wherein ,Hthe characteristics of the object are represented and,representing the hadamard product operation.
9. An electronic device, the electronic device comprising:
a processor and a memory;
the processor is configured to execute the steps of the commercial fuel consumption prediction method according to any one of claims 1 to 8 by calling a program or instructions stored in the memory.
10. A computer-readable storage medium storing a program or instructions that cause a computer to execute the steps of the commercial vehicle fuel consumption prediction method according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310926480.3A CN116644865B (en) | 2023-07-27 | 2023-07-27 | Commercial vehicle fuel consumption prediction method, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310926480.3A CN116644865B (en) | 2023-07-27 | 2023-07-27 | Commercial vehicle fuel consumption prediction method, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116644865A true CN116644865A (en) | 2023-08-25 |
| CN116644865B CN116644865B (en) | 2023-10-20 |
Family
ID=87625148
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310926480.3A Active CN116644865B (en) | 2023-07-27 | 2023-07-27 | Commercial vehicle fuel consumption prediction method, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116644865B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117235508A (en) * | 2023-11-15 | 2023-12-15 | 天津市普迅电力信息技术有限公司 | Vehicle fuel consumption prediction method and system based on big data |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2016108175A (en) * | 2016-03-09 | 2017-09-14 | Государственное бюджетное образовательное учреждение высшего образования "Альметьевский государственный нефтяной институт" | Method for determining the current oil saturation coefficient of a developed oil reservoir in a well |
| CN110228470A (en) * | 2019-06-03 | 2019-09-13 | 吉林大学 | A kind of rate of economizing gasoline real-time computing technique based on the prediction of hiding auto model |
| CN110415071A (en) * | 2019-07-03 | 2019-11-05 | 西南交通大学 | A kind of competing product control methods of automobile based on opining mining analysis |
| CN110705797A (en) * | 2019-10-09 | 2020-01-17 | 浙江海洋大学 | Ship oil consumption data prediction method based on ship sensor network |
| CN110849411A (en) * | 2018-08-21 | 2020-02-28 | 上海博泰悦臻网络技术服务有限公司 | Internet of vehicles server, vehicle and vehicle condition evaluation method based on driving oil consumption data |
| CN111179465A (en) * | 2019-12-13 | 2020-05-19 | 同济大学 | Automobile oil consumption prediction method |
| CN112163299A (en) * | 2020-09-30 | 2021-01-01 | 三一重机有限公司 | Excavator oil consumption prediction method and system and electronic equipment |
| CN112883648A (en) * | 2021-02-23 | 2021-06-01 | 一汽解放汽车有限公司 | Training method and device for automobile fuel consumption prediction model and computer equipment |
| CN113435666A (en) * | 2021-07-20 | 2021-09-24 | 山东大学 | Commercial vehicle oil consumption prediction method and system based on vehicle running state |
| CN113624291A (en) * | 2021-07-27 | 2021-11-09 | 三一专用汽车有限责任公司 | Oil consumption monitoring method, oil consumption monitoring device and engineering vehicle |
| CN114659593A (en) * | 2022-03-28 | 2022-06-24 | 陕西天行健车联网信息技术有限公司 | System and method for estimating fuel quantity of automobile fuel tank |
| CN114781698A (en) * | 2022-03-31 | 2022-07-22 | 山东大学 | Vehicle oil consumption prediction method and system based on spatio-temporal attention mechanism chart convolution network |
-
2023
- 2023-07-27 CN CN202310926480.3A patent/CN116644865B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2016108175A (en) * | 2016-03-09 | 2017-09-14 | Государственное бюджетное образовательное учреждение высшего образования "Альметьевский государственный нефтяной институт" | Method for determining the current oil saturation coefficient of a developed oil reservoir in a well |
| CN110849411A (en) * | 2018-08-21 | 2020-02-28 | 上海博泰悦臻网络技术服务有限公司 | Internet of vehicles server, vehicle and vehicle condition evaluation method based on driving oil consumption data |
| CN110228470A (en) * | 2019-06-03 | 2019-09-13 | 吉林大学 | A kind of rate of economizing gasoline real-time computing technique based on the prediction of hiding auto model |
| CN110415071A (en) * | 2019-07-03 | 2019-11-05 | 西南交通大学 | A kind of competing product control methods of automobile based on opining mining analysis |
| CN110705797A (en) * | 2019-10-09 | 2020-01-17 | 浙江海洋大学 | Ship oil consumption data prediction method based on ship sensor network |
| CN111179465A (en) * | 2019-12-13 | 2020-05-19 | 同济大学 | Automobile oil consumption prediction method |
| CN112163299A (en) * | 2020-09-30 | 2021-01-01 | 三一重机有限公司 | Excavator oil consumption prediction method and system and electronic equipment |
| CN112883648A (en) * | 2021-02-23 | 2021-06-01 | 一汽解放汽车有限公司 | Training method and device for automobile fuel consumption prediction model and computer equipment |
| CN113435666A (en) * | 2021-07-20 | 2021-09-24 | 山东大学 | Commercial vehicle oil consumption prediction method and system based on vehicle running state |
| CN113624291A (en) * | 2021-07-27 | 2021-11-09 | 三一专用汽车有限责任公司 | Oil consumption monitoring method, oil consumption monitoring device and engineering vehicle |
| CN114659593A (en) * | 2022-03-28 | 2022-06-24 | 陕西天行健车联网信息技术有限公司 | System and method for estimating fuel quantity of automobile fuel tank |
| CN114781698A (en) * | 2022-03-31 | 2022-07-22 | 山东大学 | Vehicle oil consumption prediction method and system based on spatio-temporal attention mechanism chart convolution network |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117235508A (en) * | 2023-11-15 | 2023-12-15 | 天津市普迅电力信息技术有限公司 | Vehicle fuel consumption prediction method and system based on big data |
| CN117235508B (en) * | 2023-11-15 | 2024-01-30 | 天津市普迅电力信息技术有限公司 | Vehicle fuel consumption prediction method and system based on big data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116644865B (en) | 2023-10-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110377984B (en) | Industrial equipment residual effective life prediction method and system and electronic equipment | |
| CN116644865B (en) | Commercial vehicle fuel consumption prediction method, electronic equipment and storage medium | |
| CN110231156B (en) | Service robot motion system fault diagnosis method and device based on time sequence characteristics | |
| CN107944648B (en) | Large ship speed and oil consumption rate prediction method | |
| WO2021212496A1 (en) | Battery detection method and apparatus | |
| Chen et al. | Application of nonlinear output frequency response functions and deep learning to RV reducer fault diagnosis | |
| CN112580149B (en) | Vehicle following model generation method based on generation of countermeasure network and driving duration | |
| CN113326981A (en) | Atmospheric environment pollutant prediction model based on dynamic space-time attention mechanism | |
| CN111475546A (en) | Financial time sequence prediction method for generating confrontation network based on double-stage attention mechanism | |
| CN112256866A (en) | Text fine-grained emotion analysis method based on deep learning | |
| CN111144027A (en) | Approximation method based on BP neural network full characteristic curve function | |
| CN114399064A (en) | Equipment health index construction method based on multi-source sensor data fusion | |
| CN116503029B (en) | Module data cooperative processing method and system for automatic driving | |
| CN117710841A (en) | Small target detection method and device for aerial image of unmanned aerial vehicle | |
| CN117251695A (en) | Training method, device, equipment and storage medium of data prediction model | |
| Gao et al. | A personalized model for driver lane-changing behavior prediction using deep neural network | |
| CN116486285A (en) | Aerial image target detection method based on class mask distillation | |
| CN112163284B (en) | Causal analysis method for operating stability of underwater vehicle | |
| CN115273081A (en) | Point cloud semantic segmentation method and system based on self-adaptive feature sampling | |
| CN114723049A (en) | Class activation mapping method and device based on gradient optimization | |
| CN113139606A (en) | Heating and ventilation fan fault detection method and system based on artificial intelligence algorithm | |
| CN114139783A (en) | Wind power short-term power prediction method and device based on nonlinear weighted combination | |
| Yu et al. | A CNN-LSTM model for road surface recognition of electric balance vehicles | |
| CN112529637A (en) | Service demand dynamic prediction method and system based on context awareness | |
| Hao et al. | Research on Offline Handwritten Chinese Character Recognition Based on Deep Learning |
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 |