CN113327430A - Method and device for predicting speed of underground trackless rubber-tyred vehicle based on LSTM - Google Patents
Method and device for predicting speed of underground trackless rubber-tyred vehicle based on LSTM Download PDFInfo
- Publication number
- CN113327430A CN113327430A CN202110488042.4A CN202110488042A CN113327430A CN 113327430 A CN113327430 A CN 113327430A CN 202110488042 A CN202110488042 A CN 202110488042A CN 113327430 A CN113327430 A CN 113327430A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- vehicle speed
- data
- speed prediction
- tyred
- 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
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/052—Detecting movement of traffic to be counted or controlled with provision for determining speed or overspeed
-
- G—PHYSICS
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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
Abstract
The invention discloses a method for predicting the speed of an underground trackless rubber-tyred vehicle based on LSTM, which comprises the steps of collecting the running data of trackless rubber-tyred vehicles of the same model; screening and processing the effective driving data to obtain training data; dividing training data into M sample sets according to the vehicle load index; constructing a vehicle speed prediction model based on LSTM; respectively training the vehicle speed prediction models through M sample sets to obtain M vehicle speed prediction models of trackless rubber-tyred vehicles of the same model; and selecting a corresponding vehicle speed prediction model according to the vehicle load index of the target vehicle in the latest stop state, and inputting m pieces of characteristic index data acquired in real time into the vehicle speed prediction model to obtain vehicle speed prediction data of the target vehicle. By utilizing the invention, the speed of the trackless rubber-tyred vehicle can be more reasonably predicted, and once an overspeed or a dangerous event occurs, a driver is timely reminded, so that the occurrence of accidents is reduced.
Description
Technical Field
The invention relates to the technical field of intelligent transportation, in particular to a method for predicting the speed of an underground trackless rubber-tyred vehicle based on LSTM.
Background
The trackless rubber-tyred vehicle gradually becomes the main mode of the underground coal mine auxiliary transportation system due to the characteristics of high transportation efficiency, few operation links, strong environmental adaptability and the like, and the safe driving of the trackless rubber-tyred vehicle has increasingly important influence on the overall efficient and stable operation of a coal mine.
At present, the safe driving of a trackless rubber-tyred vehicle under a coal mine faces a plurality of challenges, and firstly, the trackless rubber-tyred vehicle is provided with complex road conditions in a driving roadway, such as long-distance ramps and curved paths, and more concave-convex road surfaces and potholes; secondly, insufficient illumination and excessive dust in the roadway can cause serious interference to the sight of a driver or sensing equipment (such as a laser radar, a millimeter wave radar, a camera and the like) in an automatic driving system; in addition, uncontrollable factors such as fatigue driving of a driver, distraction, unfamiliarity with road conditions and the like may occur in the driving process of the vehicle.
The overspeed behavior is taken as the maximum potential safety hazard of the underground trackless rubber-tyred vehicle in safe running and must be paid sufficient attention. In addition to relying on driver literacy, the main technical approaches in current practice include: when the locomotive protection device detects that the speed of a vehicle exceeds a preset fixed threshold, an alarm prompt is carried out, overspeed snapshot facilities can be installed in partial coal mine roadways, recording and linkage alarm are carried out when the vehicle is overspeed, and stall protection devices can be installed in some vehicles or roadways to carry out emergency stop on the vehicles in an out-of-control state. In general, the overspeed control of trackless rubber-tyred vehicles is still based on a cut-off fixed threshold constraint (according to general regulations, trackless rubber-tyred vehicles must not exceed 40km/h for the transport of objects and 25km/h for the transport of people), and mainly takes rigid, protective, result-oriented control measures. Because the influence of vehicle working condition factors, road environment factors and traffic behavior factors on the actual safe vehicle speed threshold is not considered, the safe, reliable and efficient vehicle speed early warning effect is difficult to realize. For example, when a large downhill slope or a short distance from the front person is about to occur, the actual reasonable vehicle speed should be well below the fixed upper limit required by regulations, and if the driver is inexperienced or in a tired driving situation, the risk of accident is likely to increase due to too fast vehicle speed.
The time sequence data processing and predicting model based on the Recurrent Neural Network (RNN) is particularly suitable for processing time sequence data because the time sequence relevance characteristics of a group of input data or intermediate state data can be effectively extracted. The Long Short Term Memory (LSTM) is a special RNN type, which solves the problem that a general RNN is difficult to train (gradient disappears), introduces Memory forgetting and strengthening characteristics closer to human characteristics, has a data processing capability of a longer time domain span, and is widely applied to the technical fields such as natural language processing, video analysis, health status prediction, and the like.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: in order to solve the technical problems that a trackless rubber-tyred vehicle speed early warning mode is single and poor in effect in the prior art, the invention provides an LSTM-based underground trackless rubber-tyred vehicle speed prediction method.
The technical scheme adopted by the invention for solving the technical problems is as follows: a method for predicting the speed of an underground trackless rubber-tyred vehicle based on LSTM comprises the following steps:
s101: collecting driving data of different shifts from driving into an underground roadway to driving out of the underground roadway of the trackless rubber-tyred vehicle with the same model;
s102: screening the driving data, eliminating invalid driving data, and performing normalization processing on the remaining valid driving data to obtain training data;
s103: dividing the training data into M sample sets according to the vehicle load index;
s104: constructing a vehicle speed prediction model based on LSTM, and determining a loss function, a gradient optimizer and a learning rate of the vehicle speed prediction model;
s105: respectively training the vehicle speed prediction models through the M sample sets to obtain M vehicle speed prediction models of the trackless rubber-tyred vehicle with the same model;
s106: and selecting a corresponding vehicle speed prediction model according to the vehicle load index of the target vehicle in the latest stop state, and inputting m pieces of characteristic index data acquired in real time into the vehicle speed prediction model to obtain vehicle speed prediction data of the target vehicle.
According to the method for predicting the vehicle speed of the underground trackless rubber-tyred vehicle based on the LSTM, the driving data of the same model and different shifts are used as learning samples, a deep learning model for predicting the current vehicle speed is constructed according to the current actual vehicle working condition, road environment and traffic behavior, the vehicle speed of the trackless rubber-tyred vehicle can be monitored more reasonably, and once an overspeed or dangerous event occurs, a driver is reminded in time, and accidents are reduced; meanwhile, the method can be used for guiding and evaluating the driving behavior of the novice driver, and the operation safety and the learning efficiency of the novice driver are improved.
Further, specifically, the constructing of the LSTM-based vehicle speed prediction model specifically includes: input matrix at time t
Wherein the content of the first and second substances,
the above-mentioned
Denotes a column vector consisting of m pieces of feature index data at time t, and W denotes
The length in the time dimension and the output data at the time t are Yt＝[yt]Wherein, ytAnd (5) representing the vehicle speed data at the time t, and setting the number of hidden layers of the LSTM neural network model.
Further, specifically, the driving data includes vehicle working condition data, road environment data and traffic behavior data. Due to the fact that the road conditions of the roadways under the mine are complex, all data which affect the speed of the vehicle need to be contained when the speed prediction model is trained, and therefore the speed prediction model can output more accurate results.
Further, specifically, the vehicle operating condition data includes vehicle load, transmission gear, vehicle speed, vehicle steering angle, vehicle acceleration, accelerator pedal travel, brake pedal pressure, and tire pressure. The working condition of the vehicle has direct influence on the vehicle speed, and the training data needs to contain a plurality of working condition data influencing the vehicle speed.
Further, the road environment data specifically includes a real-time position of the vehicle, a curvature of the road at the current position, a road gradient at the current position, and a distance of a nearest intersection within a preset distance (e.g., 100 meters) before the vehicle is currently driven. The road environment has indirect influence on the vehicle speed, for example, when going uphill, the vehicle speed may be slowed down, when going downhill, the vehicle speed may be accelerated due to inertia, and the vehicle speed may also need to be slowed down at the intersection, and when predicting the current vehicle speed of the vehicle, the road condition in the roadway needs to be considered, so that the predicted vehicle speed is more reasonable and accurate.
Further, specifically, the traffic behavior data includes the number of dynamic objects (people and vehicles) in a roadway in a preset distance ahead of the current driving of the vehicle, the distance between the dynamic objects and the vehicle, the moving speed and direction of the dynamic objects, and the signal state of a signal lamp of the nearest intersection in the preset distance ahead of the current driving of the vehicle. For example, when other vehicles in front of the vehicle are running or people walk, the vehicle speed is reduced, and when the vehicle encounters a red light, the vehicle speed is also reduced, so that when the vehicle speed is predicted, traffic behaviors in a roadway need to be considered, and the predicted vehicle speed is more reasonable and accurate.
Further, the vehicle load index specifically includes no load, light load, medium load, heavy load, and full load. The vehicle speed is different when the vehicle is in different load states, in order to predict the vehicle speed more accurately, the vehicle speed prediction models are classified according to different vehicle load indexes, and each vehicle load condition is suitable for the corresponding vehicle speed prediction model.
Further, specifically, the characteristic index data includes a vehicle rotation speed, a vehicle steering angle, a vehicle acceleration, an accelerator pedal travel, a brake pedal pressure, a tire pressure, a vehicle real-time position, a road curvature of a current position, a road gradient of the current position, a distance of a nearest intersection ahead of the current driving of the vehicle, a number of dynamic objects ahead of the current driving of the vehicle, a distance between the dynamic objects and the vehicle, a moving speed and a moving direction of the dynamic objects, and a signal state of a nearest intersection signal lamp within a preset distance ahead of the current driving of the vehicle. The characteristic index data is important data used for training a vehicle speed prediction model, and the more comprehensive the characteristic index data is, the more accurate the vehicle speed prediction is.
The underground trackless rubber-tyred vehicle speed prediction method based on the LSTM has the advantages that the driving data of the same model and different shifts are used as learning samples, a deep learning model for predicting the current vehicle speed is constructed according to the current actual vehicle working condition, road environment and traffic behavior, the vehicle speed of the trackless rubber-tyred vehicle can be monitored more reasonably, and once an overspeed or dangerous event occurs, a driver is reminded in time, so that accidents are reduced; meanwhile, the method can be used for guiding and evaluating the driving behavior of the novice driver, and the operation safety and the learning efficiency of the novice driver are improved; the method can also be used for dynamically setting the real-time overspeed threshold, and compared with the traditional method for setting the fixed overspeed threshold, the method effectively improves the safety of the vehicle. Compared with the prior art, the LSTM predicted vehicle speed only predicts the short-time speed of the vehicle according to the working condition parameters of the vehicle, the practical application effect is poor, the vehicle speed prediction model of the method is specific to the trackless rubber-tyred vehicle under a mine, the running environment of the trackless rubber-tyred vehicle is fixed, the vehicle speed prediction can be carried out according to the working condition of the vehicle, the roadway environment and the traffic behavior, and then the reasonable speed value of each vehicle configuration (vehicle type and load) in each place is given, and the safe running of the vehicle is restrained.
Drawings
The invention is further illustrated with reference to the following figures and examples.
FIG. 1 is a flow chart of a method for predicting vehicle speed of an LSTM-based underground trackless rubber-tyred vehicle.
FIG. 2 is a schematic structural diagram of the vehicle speed prediction device of the underground trackless rubber-tyred vehicle based on the LSTM.
In the figure: 1. the system comprises an acquisition module, a processing module, a data partitioning module, a model building module, a training module, a prediction module and a prediction module, wherein the acquisition module 2 comprises a processing module, 3 comprises a data partitioning module, 4 comprises a model building module, 5 comprises a training module, and 6 comprises a prediction module.
Detailed Description
The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic views illustrating only the basic structure of the present invention in a schematic manner, and thus show only the constitution related to the present invention.
In the description of the present invention, it is to be understood that the terms "central," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," "circumferential," and the like are used in the orientations and positional relationships indicated in the drawings for convenience in describing the invention and to simplify the description, and are not intended to indicate or imply that the referenced devices or elements must have a particular orientation, be constructed and operated in a particular orientation, and are therefore not to be considered limiting of the invention. Furthermore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present invention, "a plurality" means two or more unless otherwise specified.
In the description of the present invention, it should be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
As shown in FIG. 1, the method for predicting the vehicle speed of the underground trackless rubber-tyred vehicle based on the LSTM comprises the following steps:
s101: the method comprises the steps of collecting driving data of trackless rubber-tyred vehicles of the same model from driving into an underground roadway to driving out of the underground roadway in different shifts.
It should be noted that, a skilled driver drives the trackless rubber-tyred vehicle and then acquires driving data, and the skilled driver knows the roadway road conditions under the mine, so that the acquired driving data is reliable. The driving data of each shift comprises vehicle working condition data, road environment data and traffic behavior data. Vehicle operating condition data includes vehicle load, transmission gear, vehicle speed, vehicle steering angle, vehicle acceleration, accelerator pedal travel, brake pedal pressure, and tire pressure. The road environment data comprises the real-time position of the vehicle, the road curvature of the current position, the road gradient of the current position and the distance of the nearest intersection within the preset distance in front of the current running of the vehicle. The traffic behavior data comprises the number of dynamic targets in a roadway in a preset distance in front of the current driving of the vehicle, the distance between the dynamic targets and the vehicle, the moving speed and direction of the dynamic targets, and the signal state of a signal lamp of the nearest intersection in the preset distance in front of the current driving of the vehicle. The method can be used for sequentially acquiring the running data of a plurality of trackless rubber-tyred vehicles of different models in the same mine, and each trackless rubber-tyred vehicle is provided with necessary sensors and equipment, such as a load sensor, a gear switch, a speed sensor, a rotating speed sensor, a steering angle sensor, an acceleration sensor, a pressure sensor, a travel sensor, a tire pressure sensor, a UWB (ultra wide band) tag card, a road cooperative gateway and the like. For example, the real-time position of the vehicle can be acquired through an underground UWB accurate positioning system, the vehicle-mounted UWB tag card and a UWB base station are in underground communication, the base station can calculate the position information of the vehicle through a positioning algorithm, and finally the UWB substation distributes the position information to the vehicle through an underground 4G/5G/WIFI or V2I vehicle-road direct-connection communication interface. In addition to the host vehicle location information, the UWB system may also transmit other information including the location, category, moving speed and direction, signal light status, etc. of other vehicles or persons. For example, the curvature of the road at the current position and the gradient of the road at the current position may be acquired by a roadway map pre-stored in the vehicle. For example, the distance of the nearest intersection within a preset distance ahead of the current driving of the vehicle can be obtained through a roadway map pre-stored in the vehicle and the real-time position calculation of the vehicle. For example, the number of dynamic targets in a roadway within a preset distance ahead of the current driving of the vehicle, the distance between the dynamic target and the vehicle, and the moving speed and direction of the dynamic target can be acquired by an underground UWB accurate positioning system. For example, the signal state of the signal lamp of the nearest intersection within a preset distance in front of the current driving of the vehicle can be acquired by an underground UWB accurate positioning system. In the embodiment, vehicle working condition data, road environment data and traffic behavior data are all very necessary for predicting the vehicle speed, the road conditions of the mine roadway are very complex, and the conditions of uphill, downhill, road surface pothole and the like exist, and the road conditions have certain influence on the vehicle running.
S102: and screening the driving data, eliminating invalid driving data, and performing normalization processing on the remaining valid driving data to obtain training data.
It should be noted that the gear of the transmission can be used to screen effective data, such as a low-speed state of the vehicle (the trackless rubber-tyred vehicle is in a state where the vehicle speed is lower than a preset threshold (e.g. 3 km/h)), a reverse state of the vehicle (the trackless rubber-tyred vehicle is in a reverse gear state), a continuous deceleration state of the vehicle (the trackless rubber-tyred vehicle is in a state where the duration of the speed reduction exceeds the preset threshold (e.g. 20 seconds)), and these data are not involved in training or are not directly used to predict a characteristic index of the vehicle speed. The normalization processing is carried out on the remaining effective driving data, so that all data can be unified, and the model can be conveniently trained and used.
S103: and dividing the training data into M sample sets according to the vehicle load index.
It should be noted that the vehicle load index may be divided into no-load, light load, medium load, heavy load and full load, and the training data is classified according to the vehicle load index according to the vehicle load data obtained from the driving data.
S104: and constructing a vehicle speed prediction model based on the LSTM, and determining a loss function, a gradient optimizer and a learning rate of the vehicle speed prediction model.
It should be noted that the constructing of the vehicle speed prediction model based on the LSTM specifically includes: input matrix at time t
Wherein
Denotes a column vector consisting of m pieces of feature index data at time t, and W denotes
The length in the time dimension and the output data at the time t are Yt＝[yt]Wherein, ytRepresenting the vehicle speed data at the time t, setting the layer number of hidden layers of the LSTM neural network model to be 2, and setting the layer number of the hidden layersThe more the number of layers, the more complicated the network structure, the more the network training difficulty can be increased, and more computing power also needs to be consumed when actual operation is performed after the network training is completed. When the speed of a vehicle at a certain moment is predicted, characteristic indexes in a period of time T before the moment need to be collected, and the time T is set every interval0All feature indicators in the current real time form a column vector, W is the number of these column vectors, i.e. W ═ T/T0. In the embodiment, the loss function is preferably Mean Square Error (MSE), the gradient optimizer is preferably adaptive moment estimation (Adam), the learning rate is set to be 0.01, and the adaptive moment estimation (Adam) algorithm dynamically adjusts the learning rate of each parameter by using the first moment estimation and the second moment estimation of the gradient. The learning rate can influence the speed of the deep learning training reaching the optimal value, when the learning rate is too large, the parameter updating amplitude is too large, the optimal value is possibly crossed in a certain step, and when the learning rate is too small, the model can not be converged for a long time, so that the operation speed is influenced.
S105: and training the vehicle speed prediction models respectively through the M sample sets to obtain M vehicle speed prediction models of the trackless rubber-tyred vehicles of the same model.
It should be noted that, in this embodiment, M sample sets represent training data of five states, i.e., no-load, light-load, medium-load, heavy-load and full-load, and five vehicle speed prediction models of the trackless rubber-tyred vehicle of the same model can be obtained by training the vehicle speed prediction models through the five sample sets, and since the vehicle has different weights when no-load, light-load, medium-load, heavy-load and full-load, the influence on the vehicle speed is different, different vehicle speed prediction models are established to predict the vehicle speed. The data of each sample set can be divided into a training set, a verification set and a test set according to a certain proportion, then training parameters such as hardware resources, batch size, iteration turns and the like are appointed, whether training loss errors gradually decrease or not is observed in the training process, overfitting problems are prevented by using the verification set, and finally, the generated optimal weight file is used for reasoning in the test set to confirm the effectiveness of the model.
S106: and selecting a corresponding vehicle speed prediction model according to the vehicle load index of the target vehicle in the latest stop state, and inputting m pieces of characteristic index data acquired in real time into the vehicle speed prediction model to obtain vehicle speed prediction data of the target vehicle.
It should be noted that, when the vehicle is in a stopped state, an event that the load of the vehicle is changed such as getting on or off a cargo, getting on or off a person, etc. may occur, so that a corresponding vehicle speed prediction model needs to be selected according to a vehicle load index of the target vehicle in the latest stopped state, so that the vehicle speed prediction is more accurate. The characteristic index data comprises the rotation speed of the vehicle, the steering angle of the vehicle, the acceleration of the vehicle, the travel of an accelerator pedal, the pressure of a brake pedal, the pressure of tires, the real-time position of the vehicle, the curvature of a road at the current position, the gradient of the road at the current position, the distance of a nearest intersection in front of the current running of the vehicle, the number of dynamic targets in front of the current running of the vehicle, the distance between the dynamic targets and the vehicle, the moving speed and the direction of the dynamic targets and the signal state of a signal lamp at the nearest intersection in a preset distance in front of the current running of the vehicle, and the characteristic index data acquired in real time form a plurality of column vector data
Then, a plurality of column vectors are combined into a matrix
The vehicle speed prediction data is input into the selected vehicle speed prediction model, and the vehicle speed prediction model can output corresponding vehicle speed prediction data. Furthermore, the current overspeed threshold value of the vehicle can be estimated after the vehicle speed prediction model is obtained, and if the vehicle possibly needs to overspeed, the driver is reminded in time to prevent accidents.
As shown in FIG. 2, the invention also provides a vehicle speed prediction system of the underground trackless rubber-tyred vehicle based on the LSTM, and the vehicle speed prediction method of the underground trackless rubber-tyred vehicle based on the LSTM comprises the following steps: the device comprises an acquisition module 1, a processing module 2, a data dividing module 3, a model building module 4, a training module 5 and a prediction module 6. The acquisition module 1 is configured to acquire driving data of trackless rubber-tyred vehicles of the same model from different shifts from driving into an underground roadway to driving out of the underground roadway; the processing module 2 is configured to screen the driving data, eliminate invalid driving data, and perform normalization processing on the remaining valid driving data to obtain training data; the data dividing module 3 is configured to divide the training data into M sample sets according to the vehicle load index; the model building module 4 is configured to build an LSTM-based vehicle speed prediction model, and determine a loss function, a gradient optimizer and a learning rate of the vehicle speed prediction model; the training module 5 is configured to train the vehicle speed prediction models respectively through M sample sets to obtain M vehicle speed prediction models of trackless rubber-tyred vehicles of the same model; the prediction module 6 is configured to select a corresponding vehicle speed prediction model according to the vehicle load index of the target vehicle in the latest stop state, and input m pieces of feature index data acquired in real time into the vehicle speed prediction model to obtain vehicle speed prediction data of the target vehicle.
The specific implementation of the LSTM-based speed prediction device for the underground trackless rubber-tyred vehicle in the embodiments of the present invention may refer to the above-mentioned embodiments of the LSTM-based speed prediction method for the underground trackless rubber-tyred vehicle, and will not be described herein again.
In conclusion, the method and the device for predicting the vehicle speed of the underground trackless rubber-tyred vehicle based on the LSTM, provided by the invention, have the advantages that the driving data of the same model and different shifts are used as learning samples, and a deep learning model for predicting the current vehicle speed is constructed according to the current actual vehicle working condition, road environment and traffic behavior, so that the vehicle speed of the trackless rubber-tyred vehicle can be monitored more reasonably, and once an overspeed or dangerous event occurs, a driver is reminded in time, and accidents are reduced; meanwhile, the method can be used for guiding and evaluating the driving behavior of the novice driver, and the operation safety and the learning efficiency of the novice driver are improved; the method can also be used for dynamically setting the real-time overspeed threshold, and compared with the traditional method for setting the fixed overspeed threshold, the method effectively improves the safety of the vehicle.
In light of the foregoing description of the preferred embodiment of the present invention, many modifications and variations will be apparent to those skilled in the art without departing from the spirit and scope of the invention. The technical scope of the present invention is not limited to the contents of the specification, and must be determined by the scope of the claims.
Claims (9)
1. A vehicle speed prediction method of an underground trackless rubber-tyred vehicle based on LSTM is characterized by comprising the following steps:
s101: collecting driving data of different shifts from driving into an underground roadway to driving out of the underground roadway of the trackless rubber-tyred vehicle with the same model;
s102: screening the driving data, eliminating invalid driving data, and performing normalization processing on the remaining valid driving data to obtain training data;
s103: dividing the training data into M sample sets according to the vehicle load index;
s104: constructing a vehicle speed prediction model based on LSTM, and determining a loss function, a gradient optimizer and a learning rate of the vehicle speed prediction model;
s105: respectively training the vehicle speed prediction models through the M sample sets to obtain M vehicle speed prediction models of the trackless rubber-tyred vehicle with the same model;
s106: and selecting a corresponding vehicle speed prediction model according to the vehicle load index of the target vehicle in the latest stop state, and inputting m pieces of characteristic index data acquired in real time into the vehicle speed prediction model to obtain vehicle speed prediction data of the target vehicle.
2. The LSTM-based underground trackless rubber-tyred vehicle speed prediction method of claim 1, wherein the constructing the LSTM-based vehicle speed prediction model specifically comprises:
input matrix at time t
Wherein the content of the first and second substances,
the above-mentioned
Denotes a column vector consisting of m pieces of feature index data at time t, and W denotes
The length in the time dimension and the output data at the time t are Yt＝[yt]Wherein, ytAnd (5) representing the vehicle speed data at the time t, and setting the number of hidden layers of the LSTM neural network model.
3. A LSTM based downhole trackless rubber-tyred vehicle speed prediction method according to claim 1, wherein the driving data comprises vehicle condition data, road environment data, and traffic behavior data.
4. A LSTM-based downhole trackless rubber-tyred vehicle speed prediction method according to claim 3, wherein the vehicle operating condition data comprises vehicle load, transmission gear, vehicle speed, vehicle steering angle, vehicle acceleration, accelerator pedal travel, brake pedal pressure, and tire pressure.
5. A LSTM based downhole trackless rubber-tyred vehicle speed prediction method according to claim 3, wherein the road environment data comprises the vehicle real-time position, the road curvature at the current position, the road grade at the current position, and the distance to the nearest intersection within a preset distance ahead of the vehicle's current travel.
6. A LSTM based downhole trackless rubber-tyred vehicle speed prediction method according to claim 3, wherein the traffic behavior data comprises the number of dynamic objects in the roadway within a preset distance in front of the vehicle currently driving, the distance of the dynamic objects from the vehicle, the moving speed and direction of the dynamic objects, and the signal status of the signal light of the nearest intersection within a preset distance in front of the vehicle currently driving.
7. The LSTM-based downhole trackless rubber-tyred vehicle speed prediction method of claim 1, wherein the vehicle load indicators comprise no-load, light-load, medium-load, heavy-load, and full-load.
8. A LSTM-based downhole trackless rubber-tyred vehicle speed prediction method according to claim 1, wherein the characteristic index data comprises vehicle speed, vehicle steering angle, vehicle acceleration, accelerator pedal travel, brake pedal pressure, tire pressure, vehicle real-time position, road curvature of current position, road slope of current position, distance of the nearest intersection ahead of the vehicle currently driving, number of dynamic objects ahead of the vehicle currently driving, distance of the dynamic objects from the vehicle, moving speed and direction of the dynamic objects, and signal status of the signal light of the nearest intersection within a preset distance ahead of the vehicle currently driving.
9. A vehicle speed prediction system of an underground trackless rubber-tyred vehicle based on LSTM is characterized in that the vehicle speed prediction method of the underground trackless rubber-tyred vehicle based on LSTM according to any one of claims 1 to 8 is adopted, and the method comprises the following steps:
the system comprises an acquisition module, a display module and a display module, wherein the acquisition module is configured to acquire driving data of different shifts from driving of the trackless rubber-tyred vehicle into an underground roadway to driving of the trackless rubber-tyred vehicle out of the underground roadway with the same model;
the processing module is configured to screen the driving data, eliminate invalid driving data and normalize the remaining valid driving data to obtain training data;
a data partitioning module configured to partition the training data into M sample sets according to a vehicle load indicator;
a model construction module configured to construct an LSTM-based vehicle speed prediction model, determine a loss function, a gradient optimizer, and a learning rate of the vehicle speed prediction model;
the training module is configured to train the vehicle speed prediction models respectively through the M sample sets to obtain M vehicle speed prediction models of the trackless rubber-tyred vehicle with the same model;
and the prediction module is configured to select a corresponding vehicle speed prediction model according to the vehicle load index of the target vehicle in the latest stop state, and input m pieces of feature index data acquired in real time into the vehicle speed prediction model to obtain vehicle speed prediction data of the target vehicle.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110488042.4A CN113327430B (en) | 2021-05-06 | 2021-05-06 | Method and device for predicting speed of underground trackless rubber-tyred vehicle based on LSTM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110488042.4A CN113327430B (en) | 2021-05-06 | 2021-05-06 | Method and device for predicting speed of underground trackless rubber-tyred vehicle based on LSTM |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113327430A true CN113327430A (en) | 2021-08-31 |
| CN113327430B CN113327430B (en) | 2022-11-01 |
Family
ID=77414206
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110488042.4A Active CN113327430B (en) | 2021-05-06 | 2021-05-06 | Method and device for predicting speed of underground trackless rubber-tyred vehicle based on LSTM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113327430B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115116231A (en) * | 2022-08-26 | 2022-09-27 | 深圳市城市交通规划设计研究中心股份有限公司 | Vehicle-road cooperative microscopic simulation system and method, electronic device and storage medium |
| CN115588184A (en) * | 2022-10-28 | 2023-01-10 | 摩尔线程智能科技(北京)有限责任公司 | Method and device for detecting target running device |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103914985A (en) * | 2014-04-25 | 2014-07-09 | 大连理工大学 | Method for predicting future speed trajectory of hybrid power bus |
| CN109783843A (en) * | 2018-12-05 | 2019-05-21 | 交控科技股份有限公司 | Analogue system vehicle dynamic model speed predicting method based on LSTM |
| US20200114926A1 (en) * | 2018-10-16 | 2020-04-16 | Toyota Motor Engineering & Manufacturing North America, Inc. | Vehicle velocity predictor using neural networks based on v2x data augmentation to enable predictive optimal control of connected and automated vehicles |
| CN111080018A (en) * | 2019-12-20 | 2020-04-28 | 南京航空航天大学 | Intelligent internet automobile speed prediction method based on road traffic environment |
| CN111459995A (en) * | 2020-03-11 | 2020-07-28 | 南京航空航天大学 | Multi-mode vehicle speed prediction method based on driving data |
| CN112265546A (en) * | 2020-10-26 | 2021-01-26 | 吉林大学 | Networked automobile speed prediction method based on time-space sequence information |
| CN112373483A (en) * | 2020-11-23 | 2021-02-19 | 浙江天行健智能科技有限公司 | Vehicle speed and steering prediction method based on forward neural network |
-
2021
- 2021-05-06 CN CN202110488042.4A patent/CN113327430B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103914985A (en) * | 2014-04-25 | 2014-07-09 | 大连理工大学 | Method for predicting future speed trajectory of hybrid power bus |
| US20200114926A1 (en) * | 2018-10-16 | 2020-04-16 | Toyota Motor Engineering & Manufacturing North America, Inc. | Vehicle velocity predictor using neural networks based on v2x data augmentation to enable predictive optimal control of connected and automated vehicles |
| CN109783843A (en) * | 2018-12-05 | 2019-05-21 | 交控科技股份有限公司 | Analogue system vehicle dynamic model speed predicting method based on LSTM |
| CN111080018A (en) * | 2019-12-20 | 2020-04-28 | 南京航空航天大学 | Intelligent internet automobile speed prediction method based on road traffic environment |
| CN111459995A (en) * | 2020-03-11 | 2020-07-28 | 南京航空航天大学 | Multi-mode vehicle speed prediction method based on driving data |
| CN112265546A (en) * | 2020-10-26 | 2021-01-26 | 吉林大学 | Networked automobile speed prediction method based on time-space sequence information |
| CN112373483A (en) * | 2020-11-23 | 2021-02-19 | 浙江天行健智能科技有限公司 | Vehicle speed and steering prediction method based on forward neural network |
Non-Patent Citations (3)
| Title |
|---|
| 徐进 等: "山区道路弯坡组合路段重载车辆行驶速度模型", 《长安大学学报（自然科学版）》 * |
| 符锌砂 等: "高速公路纵坡路段货车运行车速预测", 《公路交通科技》 * |
| 雷斌 等: "长大上坡路段载重汽车运行速度预测模型", 《长安大学学报（自然科学版）》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115116231A (en) * | 2022-08-26 | 2022-09-27 | 深圳市城市交通规划设计研究中心股份有限公司 | Vehicle-road cooperative microscopic simulation system and method, electronic device and storage medium |
| CN115588184A (en) * | 2022-10-28 | 2023-01-10 | 摩尔线程智能科技(北京)有限责任公司 | Method and device for detecting target running device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113327430B (en) | 2022-11-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3099552B1 (en) | A vehicle driver feedback system and corresponding method | |
| US8478500B1 (en) | System and method for utilizing traffic signal information for improving fuel economy and reducing trip time | |
| Nouveliere | Experimental vehicle longitudinal control using a second order sliding mode technique | |
| CN113327430B (en) | Method and device for predicting speed of underground trackless rubber-tyred vehicle based on LSTM | |
| CN109415043A (en) | A kind of automatic Pilot traveling planing method based on comfort level | |
| CN106297340A (en) | A kind of driving vehicle pre-warning system for monitoring and method | |
| CN107272683A (en) | Parallel intelligent vehicle control based on ACP methods | |
| CN109100155A (en) | A kind of unmanned vehicle is in ring high-speed simulation test macro and method | |
| CN107010075A (en) | Vehicle accelerates to determine | |
| CN100589143C (en) | Method and appaatus for judging the traveling state of a floating vehicle | |
| CN103661397A (en) | Assessment of driving behavior of a driver of a motor vehicle | |
| CN110316198A (en) | A kind of safe-guard system and operation method for highway speed-raising | |
| CN111383465B (en) | Highway danger early warning and speed control system based on car networking | |
| CN103347757A (en) | System and method for optimizing fuel economy using predictive environment and driver behavior information | |
| CN103935364A (en) | Automobile active anti-collision early warning system based on millimeter-wave radars | |
| CN102136190A (en) | Dispatching management system and method for event emergency response of urban bus passenger transport | |
| CN112884288B (en) | Highway rain and fog scene driving safety evaluation system | |
| CN111028504A (en) | Urban expressway intelligent traffic control method and system | |
| CN107038860A (en) | A kind of user's driving behavior methods of marking based on ADAS technologies and regression model | |
| CN115782905B (en) | Automatic driving safety degree quantification system for driving vehicle | |
| Aryal | Optimization of geometric road design for autonomous vehicle | |
| Wang et al. | ARIMA model and few-shot learning for vehicle speed time series analysis and prediction | |
| CN113335293A (en) | Highway road surface detection system of drive-by-wire chassis | |
| CN116229762A (en) | Intersection right-turning collision early warning method based on intelligent luminous line | |
| CN110610611B (en) | Driving safety evaluation method for intelligent network-connected vehicle in mixed-driving traffic flow |
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 |