CN115909712A - Training method of travel speed determination model, travel speed determination method and device - Google Patents
Training method of travel speed determination model, travel speed determination method and device Download PDFInfo
- Publication number
- CN115909712A CN115909712A CN202111151474.2A CN202111151474A CN115909712A CN 115909712 A CN115909712 A CN 115909712A CN 202111151474 A CN202111151474 A CN 202111151474A CN 115909712 A CN115909712 A CN 115909712A
- Authority
- CN
- China
- Prior art keywords
- automated guided
- guided vehicle
- speed
- configuration information
- configuration parameter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012549 training Methods 0.000 title claims abstract description 127
- 238000000034 method Methods 0.000 title claims abstract description 86
- 238000013528 artificial neural network Methods 0.000 claims description 78
- 210000002569 neuron Anatomy 0.000 claims description 38
- 230000032258 transport Effects 0.000 description 38
- 230000008569 process Effects 0.000 description 36
- 238000010586 diagram Methods 0.000 description 10
- 238000012360 testing method Methods 0.000 description 10
- 238000010801 machine learning Methods 0.000 description 9
- 239000011159 matrix material Substances 0.000 description 9
- 238000013500 data storage Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 6
- 230000003044 adaptive effect Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000001174 ascending effect Effects 0.000 description 1
- 238000005452 bending Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000000969 carrier Substances 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000000446 fuel Substances 0.000 description 1
- 239000005431 greenhouse gas Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012821 model calculation Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000011022 operating instruction Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 229920006395 saturated elastomer Polymers 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/60—Other road transportation technologies with climate change mitigation effect
- Y02T10/70—Energy storage systems for electromobility, e.g. batteries
Landscapes
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
The embodiment of the application relates to the technical field of intelligent transportation, in particular to a training method of a running speed determination model, a running speed determination method and equipment.
Description
Technical Field
The embodiment of the application relates to the technical field of intelligent transportation, in particular to a training method of a driving speed determination model, a driving speed determination method and equipment.
Background
The electric automobile can solve the problems of energy consumption, greenhouse gas emission and the like caused by the traditional oil vehicle by adopting electric energy, and realizes energy conservation, emission reduction, environmental protection and sustainable development. For promoting the popularization of electric vehicles, a matched device for solving the problem of endurance is an important item. As main corollary equipment, the charging pile is long in charging time, and continuous operation of the electric automobile is influenced. Adopt and trade the power station, can reduce the time that electric automobile's insufficient voltage battery was changed to full charge battery to be close with the time that the fuel truck refueled, realize electric automobile's battery quick replacement. Therefore, the continuous operation of the electric automobile is not influenced while the endurance mileage of the electric automobile is increased by the battery replacement station.
In the battery replacement process, the running speed of the unmanned transport vehicle for transporting the battery is an important factor influencing the battery replacement time. At present, the speed of the automated guided vehicle is set according to experience, an excessively slow running speed affects the battery replacement time (carrying time), and an excessively fast running speed affects the stability in the running process, for example, an excessively fast speed causes the battery to fall off from the automated guided vehicle, that is, balance cannot be achieved in both the carrying time and the running stability.
Disclosure of Invention
The embodiment of the application provides a training method of a running speed determination model, a running speed determination method and equipment.
In a first aspect, an embodiment of the present application provides a method for training a travel speed determination model, including obtaining a training set, where the training set includes a plurality of configuration information, each configuration information corresponds to a measured speed, the configuration information includes a first configuration parameter that affects a transport time of an automated guided vehicle and a second configuration parameter that affects a travel stability of the automated guided vehicle, and after obtaining the training set, training a preset neural network by using the training set until the neural network converges, so as to obtain a travel speed determination model.
In the above embodiment of the present application, a training process of a driving speed determination model is specifically described, in which a training set used in the training process considers a first configuration parameter affecting a transportation time of an automated guided vehicle and a second configuration parameter affecting a driving stability of the automated guided vehicle, where the transportation time is a time required for the automated guided vehicle to complete a transportation task, and the driving stability refers to a stability of the automated guided vehicle to maintain a transported object during driving, for example, the transported object does not fall off from the automated guided vehicle, and therefore, the neural network can learn an internal relationship and a rule between the first configuration parameter and the second configuration parameter and a driving speed, respectively, so that the trained driving speed determination model can consider an influence of the transportation time and the driving stability on the driving speed when outputting the driving speed, that is, a driving speed that takes into account the transportation time and the driving stability can be output, and it is advantageous for the automated guided vehicle to complete the transportation task as quickly and stably as possible.
In one possible implementation of the first aspect, the first configuration parameter comprises a rotational speed of a servo motor in the automated guided vehicle and/or a torque of a servo motor in the automated guided vehicle.
In the above-described embodiments of the present application, the first configuration parameter includes at least one of a rotational speed of a servo motor in the automated guided vehicle and a torque of the servo motor in the automated guided vehicle. The speed and the torque of the servo motor can directly influence the running speed, namely the conveying time. At least one of the rotating speed and the torque of the servo motor is used as a first configuration parameter, a training set is introduced, so that the neural network can learn the internal relation and the rule between the rotating speed and/or the torque of the servo motor and the running speed respectively, the trained running speed determination model can take the influence of the rotating speed and/or the torque of the servo motor on the running speed into consideration when outputting the running speed, and the running speed can be adapted to the rotating speed and/or the torque of the servo motor.
In one possible implementation of the first aspect, the second configuration parameter comprises a load bearing surface size and/or a load weight of the automated guided vehicle.
In the above embodiments of the present application, the second configuration parameter includes at least one of a load bearing surface size and a load weight of the automated guided vehicle. The size of the bearing surface is the area of the plane of the unmanned transport vehicle for bearing the transported object. Different bearing surfaces correspond to the transported goods with different volume specifications, and the volume specifications of the transported goods influence the stability in the driving process, for example, the transported goods with large volume are easy to overturn when the speed is too high. The load weight is the weight of the transported object carried by the automated guided vehicle, and affects the stability during the driving process, for example, when the load is heavy, the inertia of the automated guided vehicle is large, and when the speed is too high, the risk that the transported object flies out due to the inertia is large. Therefore, the size of the bearing surface and the load weight of the unmanned transport vehicle are used as second configuration parameters, and a training set is introduced, so that the neural network can learn the internal relation and rule between the size of the bearing surface and the running speed and the internal relation and rule between the load weight and the running speed, the influence of the size of the bearing surface and the load weight on the running speed can be considered when the trained running speed determination model outputs the running speed, and the running speed can be adapted to the size of the bearing surface and the load weight.
In a possible implementation manner of the first aspect, the obtaining a training set specifically includes: acquiring the value range of each configuration parameter, wherein one configuration parameter takes at least three levels; combining at least three horizontal numbers of each configuration parameter according to a preset combination rule to obtain a plurality of configuration information; and acquiring the measured speeds corresponding to the plurality of pieces of configuration information respectively, and taking the plurality of pieces of configuration information and the measured speeds corresponding to the plurality of pieces of configuration information respectively as a training set.
In the above embodiment of the present application, at least three level numbers are taken from the value range of the configuration parameters, and then the configuration parameters at different level numbers are combined to obtain a plurality of configuration information, so that the training set covers the corresponding relationship between the configuration parameter combinations (configuration information) at different level numbers and the measured speeds, and thus, the neural network can learn the internal relations and rules between the configuration parameter combinations (configuration information) at different level numbers and the measured speeds, so that the trained travel speed determination model has universality, that is, the trained travel speed determination model can output adaptive travel speeds for unmanned vehicles at different configuration parameters.
In a possible implementation manner of the first aspect, the preset combination rule is an orthogonal trial combination.
In the above embodiments of the present application, at least three horizontal numbers of each configuration parameter are combined according to an orthogonal test, on one hand, the orthogonal test combination can make the configuration parameter combination under each horizontal number representative, so that the training set has a sufficient sample size, which is beneficial to improving the accuracy of the trained driving speed determination model, and on the other hand, the orthogonal test combination can reduce the test cost compared with the full combination.
In one possible implementation manner of the first aspect, at least three horizontal numbers of a configuration parameter are distributed in an arithmetic progression.
In the above embodiment of the present application, by setting at least three horizontal numbers of each configuration parameter to be distributed in an arithmetic progression, the horizontal numbers of each configuration parameter can uniformly cover the range of the configuration parameter, and there is no bias, so that the neural network can learn the influence of the configuration parameter on the running speed in the whole value range, and there is no possibility that a trained running speed determination model cannot output an accurate running speed when facing an unmanned transportation vehicle in which the configuration parameter 1# is near the lower limit of the value range because a certain configuration parameter 1# (for example, the size of a carrying surface) is concentrated near the upper limit of the corresponding value range in a training set. That is, at least three horizontal numbers of each configuration parameter are distributed in an arithmetic progression, so that the accuracy of the model can be improved.
In one possible implementation manner of the first aspect, the neural network is an error back propagation neural network, the error back propagation neural network includes an input layer, an implied layer and an output layer, the number of the implied layer is 1, and an initial value of the number of neuron nodes in the implied layer is (m × n) ^0.5+1, where m is the number of neuron nodes in the input layer and n is the number of neuron nodes in the output layer.
In the above embodiments of the present application, the error back propagation neural network has a strong nonlinear mapping capability, and can fully learn the intrinsic relation and rule between the configuration information and the measured speed, and secondly, the error back propagation neural network includes a hidden layer, the calculation amount is small, over-fitting does not occur, and finally, the initial value of the number of neuron nodes in the hidden layer is (mxn) ^0.5+1, so that the number of neuron nodes in the hidden layer is between the number of neuron nodes in the input layer and the number of neuron nodes in the output layer, model convergence can be accelerated, and meanwhile, accuracy is ensured.
In a second aspect, an embodiment of the present application provides a method for determining a traveling speed of an automated guided vehicle, including: configuration information of the automated guided vehicle is acquired, the configuration information including a first configuration parameter that affects a transportation time of the automated guided vehicle and a second configuration parameter that affects a driving stability of the automated guided vehicle is input to a driving speed determination model trained by the method according to the first aspect to acquire a driving speed of the automated guided vehicle.
In the above-described embodiment of the present application, it is found that the training method according to the first aspect can output the traveling speed that can satisfy both the transport time and the traveling stability by taking into account the influence of the transport time and the traveling stability on the traveling speed when the traveling speed is output, and therefore, when the traveling speed of an unmanned vehicle needs to be determined, the traveling speed that can satisfy both the transport time and the traveling stability can be output only by inputting the configuration information of the unmanned vehicle to the traveling speed determination model.
In a third aspect, an embodiment of the present application provides a training apparatus, including: a processor, and a memory communicatively coupled to the processor, wherein the memory stores instructions executable by the processor to enable the processor to perform the method of training a travel speed determination model of the first aspect.
In a fourth aspect, embodiments of the present application provide a readable storage medium storing a program or instructions, which when executed by a processor, implement the method for training a travel speed determination model of the first aspect, and implement the method for determining a travel speed of the second aspect.
In a fifth aspect, an embodiment of the present application provides an automated guided vehicle, including a vehicle body, a programmable logic controller, a processor, and a memory, where the programmable logic controller is configured to control operation of the vehicle body; the memory stores a program or instructions, and the processor is used for executing the program or instructions to realize the running speed determination method of the second aspect so as to obtain the running speed of the automated guided vehicle;
the processor is also used for sending the running speed to the programmable logic controller so that the programmable logic controller controls the vehicle body to run according to the running speed.
In the above embodiments of the present application, the automated guided vehicle can automatically determine the driving speed that takes into account the transportation time and the driving stability, and operate at the driving speed, so that the transportation process is fast, safe and stable.
In a sixth aspect, the embodiment of the application provides an electric vehicle battery replacement system, which comprises a plurality of unmanned vehicles in the fifth aspect, wherein the driving speed which takes transport time and driving stability into consideration can be automatically determined based on the unmanned vehicles, and the electric vehicle battery replacement system runs at the driving speed, so that the transport process is rapid, safe and stable, and the electric vehicle battery replacement system has high battery replacement efficiency and high stability.
The foregoing description is only an overview of the technical solutions of the present application, and the present application can be implemented according to the content of the description in order to make the technical means of the present application more clearly understood, and the following detailed description of the present application is given in order to make the above and other objects, features, and advantages of the present application more clearly understandable.
Drawings
One or more embodiments are illustrated by way of example in the accompanying drawings, which correspond to the figures in which like reference numerals refer to similar elements and which are not to scale unless otherwise specified.
Fig. 1 (a) is a schematic diagram of a power swapping process in a power swapping station according to some embodiments of the present application;
fig. 1 (b) is a schematic diagram of a power swapping process in a power swapping station according to some embodiments of the present application;
fig. 1 (c) is a schematic diagram of a power swapping process in a power swapping station according to some embodiments of the present application;
fig. 1 (d) is a schematic diagram of a power swapping process in a power swapping station according to some embodiments of the present application;
FIG. 2 is a schematic diagram of a neural network according to some embodiments of the present application;
FIG. 3 is a schematic block diagram of a travel speed determination system according to some embodiments of the present application;
FIG. 4 is a flow chart illustrating a method for training a travel speed determination model according to some embodiments of the present disclosure;
FIG. 5 is a schematic sub-flowchart of step S21 of the method shown in FIG. 4;
FIG. 6 is a flow chart illustrating a method of determining a travel speed according to some embodiments of the present application;
FIG. 7 is a schematic diagram of a training apparatus according to some embodiments of the present application;
fig. 8 is a schematic structural view of an automated guided vehicle according to some embodiments of the present application.
Detailed Description
The present application will be described in detail with reference to specific examples. The following examples will aid those skilled in the art in further understanding the present application, but are not intended to limit the present application in any way. It should be noted that various changes and modifications can be made by one skilled in the art without departing from the spirit of the application. All falling within the scope of protection of the present application.
In order to make the objects, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of and not restrictive on the broad application.
It should be noted that, if not conflicted, the various features of the embodiments of the present application may be combined with each other within the scope of protection of the present application. Additionally, while functional block divisions are performed in apparatus schematics, with logical sequences shown in flowcharts, in some cases, steps shown or described may be performed in sequences other than block divisions in apparatus or flowcharts. Further, the terms "first," "second," "third," and the like, as used herein do not limit the order of data and execution, but merely distinguish between identical or similar items that have substantially the same function or effect.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description of the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
In addition, the technical features mentioned in the embodiments of the present application described below may be combined with each other as long as they do not conflict with each other.
At present, the market scale of the electric vehicle is rapidly expanded and the yield and the sales volume are rapidly increased in view of the development of policies and market conditions of the electric vehicle. With the popularization and development of electric vehicles, the infrastructure for solving the problem of electric vehicle endurance is used as a new infrastructure, and the construction is accelerated. The foundation supporting facilities mainly comprise charging piles and power exchanging stations. The charging pile charges the whole electric vehicle for dozens of minutes to several hours, and continuous operation of the electric vehicle is affected. Trade the power station and be through the battery of changing the electric motor car, be about to insufficient voltage battery and take out the back and install full-power battery, whole trade electric power process needs several minutes just can solve the continuation of the journey problem, compares in charging pile, can shorten parking time, improves the operation efficiency of electric motor car.
In the battery replacement process, the batteries are generally transported by the unmanned transport vehicle, so that the labor cost can be saved, and the automatic battery replacement is realized. Specifically, as shown in fig. 1 (a) to 1 (d), the electric vehicle 100 to be replaced with a battery is stopped at a predetermined position of the battery replacement station, the automated guided vehicle 200 travels from the standby area 300 to the bottom of the electric vehicle 100 to be aligned with the battery installation area of the electric vehicle 100 after receiving a battery replacement request, the carrying surface of the automated guided vehicle 200 is raised to contact the insufficient battery 101, the insufficient battery 101 is released on the carrying surface, the carrying surface is lowered, the automated guided vehicle 200 starts traveling to the buffer area 400, the forks 402 place the insufficient battery 101 on the buffer rack 401 disposed in the buffer area 400, and thereafter, the forks 402 place the full battery 102 on the carrying surface of the automated guided vehicle 200, the automated guided vehicle 200 transports the full battery 102 to the bottom of the electric vehicle 100, aligns with the battery installation area of the electric vehicle 100, and is raised, so that the full battery 102 is installed on the electric vehicle 100. When the fully charged battery 102 is mounted, the automated guided vehicle 200 is separated from the bottom of the electric vehicle 100.
The present inventors have noticed that the traveling speed of the automated guided vehicle is an important factor affecting the battery replacement efficiency. However, at present, an operator sets a preset speed according to experience and stores the preset speed in a programmable logic controller of the automated guided vehicle, and the programmable logic controller controls the automated guided vehicle to move at the preset speed to transport the battery. On one hand, the preset speed is set according to experience, so that the problem of too slow or too fast exists, when the preset speed is too slow, the carrying time (battery replacement time) is influenced, and when the preset speed is too fast, the stability in the driving process is influenced, for example, the battery falls off from the unmanned carrier due to too fast preset speed, namely, the balance cannot be achieved in the carrying time and the driving stability; on the other hand, the experience of the operator is limited, the driving speed cannot be determined when the operator faces various types of unmanned transportation vehicles, multiple experimental debugging is often required to determine a general driving speed, for example, the operator a is familiar with a small unmanned transportation vehicle, the driving speed can be set according to the experience, when the operator faces an unfamiliar large unmanned transportation vehicle, the driving speed cannot be estimated, and multiple experiments may be required to determine a general driving speed, so that the battery replacement efficiency is affected.
In order to solve the problem that the driving speed of the automated guided vehicle cannot take into account the transportation time and the driving stability, and the problem that the driving speed of different types of automated guided vehicles cannot be determined, the applicant researches and discovers that a universal machine learning model can be trained through a training set, the training set comprises a plurality of configuration information, and each configuration information corresponds to one measured speed. When configuration information of other automated guided vehicles is input into the machine learning model, the machine learning model outputs the traveling speed of the automated guided vehicle, so that the traveling speed can be determined in the presence of various types of automated guided vehicles without additional experimental debugging. For example, the machine learning model is stored in a memory of any one of the automated guided vehicles, and when the automated guided vehicle acquires its own configuration information, a processor of the automated guided vehicle calls the machine learning model in the memory, and the machine learning model receives the configuration information of the automated guided vehicle and outputs the traveling speed of the automated guided vehicle.
Specifically, the preset neural network is trained by using the training set, so that the neural network continuously learns the internal relation and rule between the configuration information and the measured speed, the parameters of the neural network are adjusted during each iterative learning, and the parameters are stopped being adjusted after the neural network converges, so that the driving speed determination model (namely, the machine learning model) is obtained.
On the basis, the configuration information in the training set comprises a first configuration parameter influencing the transportation time of the unmanned transport vehicle and a second configuration parameter influencing the running stability of the unmanned transport vehicle, wherein the transportation time is the time required by the unmanned transport vehicle to complete a transportation task (such as transportation of a power-deficient battery and a full-power battery), and the running stability refers to the stability of the unmanned transport vehicle for maintaining a transported object during running, for example, the transported object cannot fall off the unmanned transport vehicle.
Since the method provided by the embodiment of the present application relates to machine learning, for understanding, the neural network and the training process related to the embodiment of the present application will be described first.
(1) Neural network
As shown in fig. 2, the neural network may be composed of neural units, and may be specifically understood as a neural network having an input layer, an implicit layer, and an output layer, where generally, the first layer is an input layer, the last layer is an output layer, and the middle layers are implicit layers. Among them, a neural network with many hidden layers is called a Deep Neural Network (DNN). The work of each layer in the neural network can be described by the mathematical expression y = a (W · x + b), and from the physical level, the work of each layer in the neural network can be understood as performing the transformation of the input space to the output space (i.e., the row space to the column space of the matrix) through five operations on the input space (the set of input vectors), including 1, ascending/descending; 2. zooming in/out; 3. rotating; 4. translating; 5. "bending". The operations 2 and 3 are completed by W.x, the operation 4 is completed by + b, and the operation 5 is realized by a (). The expression "space" is used here because the object to be classified is not a single thing, but a class of things, and space refers to the set of all individuals of such things, where W is the weight matrix of each layer of the neural network, and each value in the matrix represents the weight value of one neuron of the layer. The matrix W determines the spatial transformation of the input space to the output space described above, i.e. W at each layer of the neural network controls how the space is transformed. The purpose of training the neural network is to finally obtain the weight matrix of all layers of the trained neural network. Therefore, the training process of the neural network is essentially a way of learning the control space transformation, and more specifically, the weight matrix.
It should be noted that, in the embodiment of the present application, based on the model adopted by machine learning, the essence is a neural network.
(2) Training process
In the process of training the neural network, because the output of the neural network is expected to be as close as possible to the value really expected to be predicted, the weight matrix of each layer of the neural network can be updated according to the difference between the predicted value of the current network and the really expected target value (of course, an initialization process is usually carried out before the first updating, namely, parameters are configured in advance for each layer in the neural network), for example, if the predicted value of the network is high, the weight matrix is adjusted to be lower in prediction, and the adjustment is carried out continuously until the neural network can predict the really expected target value. Therefore, it is necessary to define in advance "how to compare the difference between the predicted value and the target value", which are loss functions (loss functions) or objective functions (objective functions), which are important equations for measuring the difference between the predicted value and the target value. Taking the loss function as an example, if the higher the output value (loss) of the loss function indicates the larger the difference, the training of the neural network becomes a process of reducing the loss as much as possible.
In the training process of the neural network, an error Back Propagation (BP) algorithm can be adopted to correct the size of parameters in the initial neural network, so that the reconstruction error loss of the neural network is smaller and smaller. The back propagation algorithm is an error loss dominated back propagation motion aimed at obtaining optimal neural network parameters, such as weight matrix. Specifically, an input signal is transmitted in a forward direction until an error loss is generated in output, and parameters in an initial neural network are updated by back propagation of error loss information, so that when the error meets a preset condition or the number of times of updating the parameters reaches a preset threshold value, the neural network converges, and a trained model is obtained.
Next, the generation and application of the model are described. Specifically, referring to fig. 3, a system architecture of the travel speed determination system according to the embodiment of the present application is described, please refer to fig. 3, and fig. 3 is a system architecture diagram of the travel speed determination system according to the embodiment of the present application. In fig. 3, the driving speed determination system 500 includes an execution device 510, a training device 520, a database 530, a client device 540, a data storage system 550, and a data acquisition device 560, where the execution device 510 includes a calculation module 511, where the data acquisition device 560 is configured to obtain a large-scale data set (i.e., a training set including configuration information and a measured speed corresponding to the configuration information) required by a developer, and store the training set in the database 530, and the training device 520 trains the driving speed determination model 501 constructed in the present application based on the training set in the database 530, where the driving speed determination model 501 may be a structure of a neural network described in the embodiment corresponding to fig. 2, and it is specifically referred to the embodiment corresponding to fig. 2, and details of the embodiment are not described here. The trained travel speed determination model 501 is then used by the execution device 510. The execution device 510 may call data, code, etc. from the data storage system 550 and may store data, instructions, etc. in the data storage system 550. The data storage system 550 may reside within the execution device 510 or the data storage system 550 may be external to the execution device 510.
The driving speed determination model 501 trained by the training device 520 may be applied to different systems or devices (i.e., the execution device 510), specifically, an unmanned transportation vehicle, or a terminal-side device, such as a mobile phone, a tablet, or a computer. In FIG. 3, the execution device 510 is configured with an I/O interface to which a "user" may input data via the client device 540 for data interaction with an external device. For example, the client device 540 may be a remote controller of the system 500, the user inputs configuration information into the remote controller, the configuration information is input into the computing module 511 of the execution device 510 through the remote controller, the computing module 511 performs computing processing on the input configuration information to obtain a driving speed, and the driving speed is stored on a storage medium of the execution device 510 for subsequent calling. In addition, in some embodiments of the present application, the client device 540 may also be integrated in the execution device 510, for example, when the execution device 510 is a computer, the configuration information may be directly obtained through an input end (a keyboard or a mouse, etc.) of the computer, and then the configuration information is calculated by a calculation module in the computer to obtain the driving speed; if the execution device 510 is an automated guided vehicle, the configuration information may be directly obtained from a memory in which the configuration information is stored inside the automated guided vehicle, and the configuration information may be calculated by a calculation module inside the automated guided vehicle to obtain the traveling speed. The product forms of the execution device 510 and the client device 540 are not limited herein.
It should be noted that fig. 3 is only a schematic diagram of a system architecture provided by the embodiment of the present application, and the position relationship between the devices, modules, etc. shown in the diagram does not constitute any limitation, for example, in fig. 3, the data storage system 550 is an external memory with respect to the execution device 510, and in other cases, the data storage system 550 may also be disposed in the execution device 510; in fig. 3, the client device 540 is an external device with respect to the execution device 510, but in other cases, the client device 540 may be integrated in the execution device 510.
It should be further noted that the training of the running speed determination model 501 according to the embodiment of the present application may be implemented on the cloud side, for example, a training device on the cloud side (the training device may be disposed on one or more servers or virtual machines) may obtain a training set, train the neural network according to multiple sets of configuration information and measured speeds in the training set to obtain a trained running speed determination model 501, and then send the trained running speed determination model 501 to the execution device 510 for application, for example, send to an unmanned transport vehicle (execution device) to determine a running speed. It is understood that the training of the travel speed determination model in the above embodiment may also be implemented on the terminal side, that is, the training device 520 may also be an automated guided vehicle or a terminal side device, that is, the trained travel speed determination model 501 may be directly used in the automated guided vehicle or the terminal side device. In the embodiment of the present application, there is no limitation on which device (cloud side or terminal side) the driving speed determination model 501 is trained or applied.
The driving speed determination model has two stages, namely a training stage and an inference stage, wherein the training stage corresponds to the training of the driving speed determination model, and the inference stage corresponds to the application of the driving speed determination model. The following describes specific flows of the training method of the travel speed determination model and the travel speed determination method provided by the embodiment of the present application from these two stages, respectively.
1. Training phase
In the embodiment of the present application, the training phase refers to a process of obtaining the driving speed determination model 501 by performing a training operation on the neural network by the training device 520 in fig. 3. Specifically, referring to fig. 4, fig. 4 is a schematic flow chart of a training method of a travel speed determination model according to an embodiment of the present application, where the method S20 specifically includes the following steps:
s21: a training set is obtained, wherein the training set comprises a plurality of configuration information, each configuration information corresponds to one measured speed, and the configuration information comprises a first configuration parameter influencing the carrying time of the unmanned transport vehicle and a second configuration parameter influencing the driving stability of the unmanned transport vehicle.
S22: and training a preset neural network by adopting the training set until the neural network is converged to obtain a driving speed determination model.
The training set is training data for the neural network to learn. The training set includes a plurality of configuration information, each configuration information corresponds to a measured speed, for example, the training set includes 300 (configuration information, measured speed), the data amount of a specific training set can be set by those skilled in the art according to actual situations, and it is understood that the data amount can also be 100, 200, or 500, etc.
The configuration information reflects a state of the automated guided vehicle. Each configuration information includes a first configuration parameter that affects a transport time of the automated guided vehicle and a second configuration parameter that affects a driving stability of the automated guided vehicle. The transport time is the time required for an automated guided vehicle to perform each transport task, such as the time required for an automated guided vehicle to service an electric vehicle to transport a low-powered battery and a full-powered battery. The driving stability refers to stability of the automated guided vehicle in maintaining a conveyed object (for example, a battery) during driving, and for example, the conveyed object does not fall off the automated guided vehicle.
It can be understood that, in the power exchanging station, the traveling path of the automated guided vehicle is a structured path, that is, a predetermined path, the length of the traveling path is constant, the greater the traveling speed, the shorter the required transportation time, the higher the power exchanging efficiency, but the lower the traveling stability (rollover easily occurs due to the excessively high traveling speed); the smaller the running speed, the higher the running stability, but the longer the required transportation time and the lower the battery replacement efficiency.
The first configuration parameter is a configuration parameter that affects the transportation time of the automated guided vehicle, and corresponds to a configuration parameter that affects the traveling speed of the automated guided vehicle, for example, the first configuration parameter may be the power of a motor in the automated guided vehicle or the wheel size of the automated guided vehicle. The second configuration parameter is a configuration parameter affecting the driving stability of the automated guided vehicle, for example, the second configuration parameter is a parameter reflecting the loading condition of the automated guided vehicle or a parameter reflecting the driving path of the automated guided vehicle.
It will be appreciated that the configuration information in the training set is taken from different automated guided vehicles. In some embodiments, the measured speed corresponding to the configuration information may be a better speed obtained by a person skilled in the art through actual testing. For example, the automated guided vehicle having the configuration information is allowed to transport the battery at a plurality of speeds, and a speed at which the automated guided vehicle can travel as quickly as possible while keeping the traveling process stable and safe is determined as the measured speed. It is understood that the more the configuration information in the training set is, the more the trained travel speed determination model is applicable to the type of automated guided vehicle.
And after the training set is obtained, training a preset neural network by adopting the training set until the neural network converges to obtain the driving speed determination model. Here, regarding the structure and training of the neural network, as described in the aforementioned "(1) neural network and (2) training process", details thereof are not repeated here.
Based on the training set including a first configuration parameter influencing the carrying time of the unmanned carrying vehicle and a second configuration parameter influencing the running stability of the unmanned carrying vehicle, the neural network can learn the internal relation and rule between the first configuration parameter and the second configuration parameter and the running speed respectively, so that the trained running speed determining model can take the influence of the carrying time and the running stability on the running speed respectively into consideration when outputting the running speed, namely, the running speed which takes the carrying time and the running stability into consideration can be output, and the unmanned carrying vehicle can complete the carrying task as quickly, stably and safely as possible.
According to some embodiments of the application, optionally, the first configuration parameter comprises a rotational speed of a servo motor in the automated guided vehicle and/or a torque of a servo motor in the automated guided vehicle.
The servo motor is main hardware of the automated guided vehicle, converts a voltage signal into torque and rotating speed to drive the automated guided vehicle to run, and can accurately control the speed and the position of the automated guided vehicle. The power of the servo motor is in direct proportion to the product of the torque and the rotating speed, namely the torque and the rotating speed of the servo motor directly determine the power of the servo motor, and the power of the servo motor directly influences the running speed, namely the rotating speed and the torque of the servo motor can directly influence the running speed, namely the carrying time.
It will be appreciated that the different types of automated guided vehicles are configured with servo motors having different rotational speeds and torques. In the market, the rotating speed and the torque of the servo motor have a working range, for example, the rotating speed and the torque of a large-sized unmanned carrying vehicle for carrying heavy objects can be relatively large, and the rotating speed and the torque of a small-sized unmanned carrying vehicle for carrying light objects can be relatively small. Further, since the specification (size and weight) of the battery of the electric vehicle is within a certain range, the rotation speed and torque of the automated guided vehicle for transporting the battery of the electric vehicle are also within a certain range in order to match the transportation of the battery of the electric vehicle.
The first configuration parameter comprises at least one of a rotational speed of a servo motor in the automated guided vehicle and a torque of the servo motor in the automated guided vehicle, i.e. at least one of the rotational speed and the torque of the servo motor is introduced into the training set. It will be appreciated that the training set covers configuration information for different types of automated guided vehicles, and therefore the torque and speed of the servo motors in the training set also covers the torque and speed of the servo motors of different types of automated guided vehicles.
At least one of the rotating speed and the torque of the servo motor is used as a first configuration parameter, a training set is introduced, so that the neural network can learn the internal relation and the law between the rotating speed and/or the torque of the servo motor and the running speed respectively, the influence of the rotating speed and/or the torque of the servo motor on the running speed can be considered when the trained running speed determination model outputs the running speed, and the running speed can be adapted to the rotating speed and/or the torque of the servo motor. In addition, the torque and the rotating speed of the servo motor in the training set also cover the torque and the rotating speed of the servo motor of the unmanned conveying vehicle of different types, so that when the trained running speed determination model faces the torque and the rotating speed of different servo motors, the running speed corresponding to the rotating speed and/or the torque of the servo motor can be output, namely the application range of the running speed determination model is wide.
According to some embodiments of the application, optionally, the second configuration parameter comprises a load bearing surface size and/or a load weight of the automated guided vehicle.
The carrying surface of the automated guided vehicle is a horizontal surface for carrying an object to be carried (for example, a battery of an electric vehicle) during transportation. It will be appreciated that for an automated guided vehicle for carrying batteries, the carrying surface needs to match the floor area of the battery being carried, for an automated guided vehicle with a larger carrying surface the volume of the battery being carried is larger, and for an automated guided vehicle with a smaller carrying surface the volume of the battery being carried is smaller. In the transportation and driving process of the automated guided vehicle, the size of the battery affects the driving stability of the automated guided vehicle, for example, if the size of the battery is large, the vehicle is easy to overturn if the driving speed is high, and if the size of the battery is small, the driving speed is slow, the stability is excessive, and the conveying efficiency is affected. Therefore, one factor influencing the driving stability can be accurately characterized through the size of the bearing surface. Based on that, the battery volume in the market is in a certain range, and the size of the bearing surface is also in a certain range.
The load weight of the automated guided vehicle is a rated weight, i.e., the maximum weight that can be loaded with a transported object (e.g., a battery of an electric vehicle). It is understood that the batteries are different in size and weight, and thus, an automated guided vehicle corresponding to different load weights is required. Since the weight (load weight) of the battery during transportation and traveling of the automated guided vehicle affects the traveling stability of the automated guided vehicle, for example, if the weight of the battery is large, the battery flies out due to large inertia when the traveling speed is high, and if the weight of the battery is small, the traveling speed is too slow, and the stability is excessive, which affects the traveling efficiency. Thus, a factor influencing the driving stability can be accurately characterized by the load weight.
The size of the bearing surface and the load weight of the unmanned transport vehicle are used as second configuration parameters, and a training set is introduced, so that the neural network can learn the internal relation and rule between the size of the bearing surface and the running speed and the internal relation and rule between the load weight and the running speed, the influence of the size of the bearing surface and the load weight on the running speed can be considered when the trained running speed determination model outputs the running speed, and the running speed can be adapted to the size of the bearing surface and the load weight. In addition, the bearing surface size and the load weight of the unmanned transport vehicles of different types are covered by the training concentrated bearing surface size and the load weight, so that when the trained running speed determination model faces different bearing surface sizes and different load weights, the running speed corresponding to the bearing surface sizes and the load weights can be output, and the application range of the running speed determination model is wide.
According to some embodiments of the present application, optionally, referring to fig. 5, the step S21 specifically includes:
s211: the value range of each configuration parameter is obtained, and one configuration parameter takes at least three levels.
S212: and combining at least three horizontal numbers of each configuration parameter according to a preset combination rule to obtain a plurality of configuration information.
S213: and acquiring the measured speeds corresponding to the plurality of pieces of configuration information respectively, and taking the plurality of pieces of configuration information and the measured speeds corresponding to the plurality of pieces of configuration information respectively as a training set.
The configuration parameters herein include first and second configuration parameters in a training set, including, for example, servo motor torque, servo motor speed, bearing surface size, and load weight. The configuration parameter range can be obtained by a person skilled in the art through statistics of configuration parameters of the existing automated guided vehicle, for example, the value range of the rotation speed of the servo motor is [ P1, P3] (rpm), the value range of the torque of the servo motor is [ α 1, α 3] (n.m), the value range of the bearing surface size is [ T1, T3] (s ^ 2), and the value range of the load weight is [ Ω 1, Ω 3] (kg).
A configuration parameter may take at least three levels, for example, 3 levels, 4 levels, or 5 levels. Taking three horizontal numbers as an example for illustration, for example, the rotation speed of the servo motor takes three horizontal numbers P1, P2 and P3, and it can be understood that P2 is within the value range [ P1, P3 ]. Correspondingly, the servo motor torque takes three horizontal numbers of alpha 1, alpha 2 and alpha 3, the carrying surface size takes three horizontal numbers of T1, T2 and T3, and the load carrying capacity takes three horizontal numbers of omega 1, omega 2 and omega 3.
And combining at least three horizontal numbers of each configuration parameter according to a preset combination rule to obtain a plurality of configuration information. For example, a configuration message may be (P1, α 2, T1, Ω 2). The preset combination rule can be a comprehensive experimental combination, 3 × 3=81 combinations exist under the comprehensive experimental combination, it can be understood that each combination is configuration information, the level number of each configuration parameter determines the type of the combination, the types of the combinations are richer, the number of samples in a training set is larger, the driving speed determination model obtained through training has better performance, when the number of samples in the training set exceeds a certain value, an overfitting phenomenon can occur, and the performance of the driving speed determination model is saturated and is not increased any more. Therefore, the performance of the model can be determined according to the running speed, the number of samples in the training set is determined, and the level number of each configuration parameter is further determined.
After the plurality of pieces of configuration information are acquired, the measured speeds respectively corresponding to the plurality of pieces of configuration information can be acquired through means such as experimental tests, and therefore the plurality of pieces of configuration information and the measured speeds respectively corresponding to the plurality of pieces of configuration information can form a training set.
At least three level numbers are taken from the value range of the configuration parameters, and then the configuration parameters under different level numbers are combined to obtain a plurality of configuration information, so that the training set covers the corresponding relation between the configuration parameter combinations (configuration information) under different level numbers and the measured speed, and the neural network can learn the internal relation and the rule between the configuration parameter combinations (configuration information) under different level numbers and the measured speed, so that the trained driving speed determination model has universality, namely, the trained driving speed determination model can output the adaptive driving speed in the face of unmanned carriers under different configuration parameters.
According to some embodiments of the application, optionally, the predetermined combination rule is an orthogonal trial combination. For example, in the example where each of the above-described arrangement parameters has three horizontal numbers, the three horizontal numbers of each of the arrangement parameters are combined in an orthogonal experiment combination manner to obtain 9 combinations as shown in table 1 below.
TABLE 1 results using orthogonal experimental combinations
Reference numerals | Bearing surface size | Speed of servo motor | Torque of servo motor | Weight of load | Measured |
|
1 | T1 | P1 | | Ω1 | V1 | |
2 | T1 | P2 | α2 | Ω2 | V2 | |
3 | T1 | P3 | α3 | Ω3 | V3 | |
4 | T2 | P1 | α2 | Ω3 | V4 | |
5 | T2 | P2 | α3 | Ω1 | V5 | |
6 | T2 | P3 | α1 | Ω2 | V6 | |
7 | T3 | P1 | α3 | Ω2 | V7 | |
8 | T3 | P2 | α1 | Ω3 | V8 | |
9 | T3 | P3 | α2 | Ω1 | V9 |
By combining at least three horizontal numbers of each configuration parameter according to an orthogonal test, on one hand, the orthogonal test combination can enable the configuration parameter combination under each horizontal number to be representative, so that a training set has enough sample size, and the accuracy of a driving speed determination model obtained by training is improved, and on the other hand, compared with a full combination, the orthogonal test combination can reduce the test cost.
According to some embodiments of the present application, optionally, at least three horizontal numbers of a configuration parameter are distributed in an arithmetic progression. For example, if the servomotor speed takes three horizontal numbers P1, P2, and P3, P2=0.5 × (P1 + P3), i.e., P1, P2, and P3 are distributed in an arithmetic progression. For example, the size of the bearing surface is 4T 1, T2, T4 and T3, and then T1, T2, T4 and T3 are distributed in an arithmetic progression.
At least three horizontal numbers of each configuration parameter are arranged in an arithmetic progression distribution, so that the horizontal numbers of each configuration parameter can uniformly cover the range of the configuration parameter without bias, for example, the size of the bearing surface is in a value range [ T1, T3], the horizontal numbers are uniformly distributed in the range, and the situation of being concentrated near T1 or near T3 can not occur. Therefore, the influence of the size of the bearing surface on the running speed in the whole value range can be learned by the neural network, and the situation that the trained running speed determination model cannot output the accurate running speed when facing an unmanned transport vehicle with the size of the bearing surface near the lower limit T3 of the value range due to the fact that the size of the bearing surface is concentrated near the upper limit T1 of the corresponding value range can be avoided. That is, at least three horizontal numbers of each configuration parameter are distributed in an arithmetic progression, so that the accuracy of the model can be improved.
According to some embodiments of the present application, optionally, the neural network is an error back propagation neural network, the error back propagation neural network includes an input layer, an implied layer and an output layer, the number of the implied layer is 1, and an initial value of the number of neuron nodes in the implied layer is (m × n) ^0.5+1, where m is the number of neuron nodes in the input layer and n is the number of neuron nodes in the output layer.
The structure and training process of the neural network are as described in the aforementioned "(1) neural network and (2) training process", and thus, the description thereof is omitted.
In some embodiments, the number m of neuron nodes in the input layer is equal to the number of configuration parameters in the configuration information, for example, when the configuration information includes 4 configuration parameters of servo motor torque, servo motor rotation speed, bearing surface size and load weight, the number m of neuron nodes in the input layer may be 4. In some embodiments, the number n of the neuron nodes in the output layer is the same as the number of the output quantity, and the output quantity is the driving speed, so that the number n of the neuron nodes in the output layer is 1. The initial value of the number of neuron nodes in the hidden layer is 3.
In the iterative training of the error back propagation neural network, along with the adjustment of model parameters, the number of neuron nodes in the hidden layer is gradually increased from an initial value, and the model parameters and the number of the neuron nodes in the hidden layer during the convergence of the neural network are selected as the model parameters and the number of the neuron nodes in the hidden layer of the driving speed determination model. It can be understood that, compared with the hidden layer neuron node number with an initial value of 1, the hidden layer neuron node number with an initial value of (m × n) ^0.5+1, so that the hidden layer neuron node number is between the input layer neuron node number and the output layer neuron node number, the model convergence can be accelerated, and the accuracy is ensured.
The error back propagation neural network has strong nonlinear mapping capability and can fully learn the internal relation and rule between configuration information and measured speed, and secondly, the error back propagation neural network comprises a hidden layer, so that the calculated amount is small, overfitting cannot occur, and finally, the initial value of the number of the neuron nodes in the hidden layer is (m multiplied by n) ^0.5+1, so that the number of the neuron nodes in the hidden layer is between the number of the neuron nodes in the input layer and the number of the neuron nodes in the output layer, the model convergence can be accelerated, and the accuracy is ensured.
2. Inference phase
In the embodiment of the present application, the inference phase refers to a process of performing, by the execution device 510 in fig. 3, calculation processing on the input configuration information by using the trained travel speed determination model 501.
Specifically, referring to fig. 6, fig. 6 is a schematic flow chart of a method for determining a driving speed according to an embodiment of the present application, where the method S30 may specifically include the following steps:
s31: configuration information of the automated guided vehicle is acquired, and the configuration information comprises a first configuration parameter influencing the transportation time of the automated guided vehicle and a second configuration parameter influencing the driving stability of the automated guided vehicle.
S32: and inputting the configuration information into the driving speed determination model obtained by training in the training stage to obtain the driving speed of the unmanned transport vehicle.
Here, the automated guided vehicle is an automated guided vehicle whose traveling speed is to be determined, and first, the arrangement information thereof is acquired. The configuration information reflects the state of the automated guided vehicle. The configuration information includes a first configuration parameter that affects a transport time of the automated guided vehicle and a second configuration parameter that affects a driving stability of the automated guided vehicle. The transfer time is the time required for the automated guided vehicle to complete a transfer task, such as the time required for the automated guided vehicle to transport a low-powered battery and a full-powered battery. The driving stability refers to stability of the automated guided vehicle for maintaining a transported object (for example, a battery) during driving, and for example, the transported object does not fall off the automated guided vehicle.
And inputting the configuration information into a driving speed determination model obtained by training in a training stage, and outputting the driving speed by the driving speed determination model.
The driving speed determining model can take the influence of the transportation time and the driving stability on the driving speed into consideration when outputting the driving speed, namely, the driving speed which can take both the transportation time and the driving stability into consideration can be output, so that when the driving speed of the unmanned transport vehicle needs to be determined, the driving speed which can take both the transportation time and the driving stability into consideration can be output only by inputting the configuration information of the unmanned transport vehicle to the driving speed determining model.
Referring to fig. 7, according to some embodiments of the present application, the present application further provides a training device 520, including a processor 521 and a memory 522 communicatively connected to the processor 521, where the memory 522 stores instructions executable by the processor 521, and the instructions are executed by the processor 521, so that the processor 521 can perform the processes of the embodiment of the training method for determining a driving speed model in the training phase, and can achieve the same technical effects, and the details are not repeated herein to avoid repetition.
Memory 522 may include both read-only memory and random access memory, among other things, and provides instructions and data to the processor. A portion of memory 522 may also include non-volatile random access memory (NVRAM). Memory 522 stores operating instructions, executable modules, or data structures, or a subset thereof, or an expanded set thereof.
The processor 521 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the training method may be implemented by hardware integrated logic circuits or instructions in the form of software in the processor 521. The processor 521 may be a general-purpose processor, a Digital Signal Processor (DSP), a microprocessor or a microcontroller, and may further include an Application Specific Integrated Circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. The processor may implement or perform the training method in embodiments corresponding to the training phase.
According to some embodiments of the present application, a readable storage medium is further provided, where a program or an instruction is stored, and when the program or the instruction is executed by a processor, the program or the instruction implements each process of the embodiment of the training method for determining a driving speed of a guided vehicle in the training phase and each process of the embodiment of the method for determining a driving speed of an automated guided vehicle in the inference phase, and can achieve the same technical effect, and in order to avoid repetition, the description is omitted here.
The readable storage medium includes a computer readable storage medium, such as a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and so on.
According to some embodiments of the present application, referring to fig. 8, the present application further provides an automated guided vehicle 200, including a vehicle body 210, a programmable logic controller 220, a processor 230, and a memory 240, wherein the programmable logic controller 220 is configured to control the operation of the vehicle body 210; the memory 240 stores a program or instructions, and the processor 230 is configured to execute the program or instructions to implement the method for determining the driving speed of the automated guided vehicle 200 in the inference phase to obtain the driving speed of the automated guided vehicle 200; the processor 230 is further configured to send the driving speed to the programmable logic controller 220, so that the programmable logic controller 220 controls the vehicle body 210 to operate according to the driving speed.
The vehicle body 210 is a hardware structure of the automated guided vehicle 200, and includes a mechanical module, a power module, and a driving module. The mechanical module comprises a vehicle body, wheels, a steering device and the like, the power module comprises a power supply battery, and the driving module comprises a servo motor. The power supply battery supplies power to the servo motor, and the servo motor rotates to drive the wheels to rotate, so that the vehicle body moves.
The plc 220 is a digital operation controller with a microprocessor for automatic control, and can load control instructions into a memory at any time for storage and execution. Programmable controller 220 is comprised of functional units such as a microprocessor, instruction and data memory, input/output interfaces, power supply, digital to analog conversion, etc.
When the automated guided vehicle 200 receives the battery to be transported, the configuration information thereof is acquired. It will be appreciated that the configuration information may be stored in memory 240 for processor 230 to determine a corresponding travel speed from the travel speed determination model calculations stored in memory 240 based on the configuration information. In some embodiments, portions of the configuration information regarding the hardware configuration of the automated guided vehicle 200 itself may be stored in the memory 240, such as servo motor torque, rotational speed, and bearing surface size, and portions of the configuration information regarding the load (e.g., load weight) may be collected by weight sensors on the automated guided vehicle 200 and transmitted to the processor.
After the processor 230 determines the traveling speed, a speed command reflecting the traveling speed is transmitted to the programmable logic controller 220, and thus the programmable logic controller 220 controls the vehicle body 210 to travel at the traveling speed.
The automated guided vehicle 200 can automatically determine a traveling speed that takes into account both a traveling time and traveling stability, and operate at the traveling speed, so that a traveling process is fast, safe, and stable.
According to some embodiments of the present application, the present application further provides an electric vehicle battery replacement system, which includes a plurality of the above-mentioned unmanned vehicles.
An electric vehicle battery replacing system for replacing an electric vehicle battery is a main facility in a battery replacing station. The electric vehicle battery changing system can further comprise a center console for controlling and dispatching the unmanned transport vehicle. It is understood that the console may be a computer or a server.
The driving speed which takes transport time and driving stability into consideration can be automatically determined based on each unmanned transport vehicle, and the electric vehicle runs at the driving speed, so that the transport process is fast, safe and stable, and the electric vehicle battery replacement system has high battery replacement efficiency.
According to some embodiments of the present application, there is provided a method for training a travel speed determination model, wherein a training set is used, the training set comprises a plurality of configuration information, each configuration information corresponds to a measured speed, and a configuration information comprises 4 configuration parameters: the automatic guided vehicle comprises a servo motor, a first configuration parameter, a second configuration parameter and a load weight, wherein the rotation speed, the torque, the bearing surface size and the load weight of the servo motor in the automatic guided vehicle influence the transportation time of the automatic guided vehicle, and the bearing surface size and the load weight influence the driving stability of the automatic guided vehicle. Specifically, each configuration parameter takes 3 horizontal numbers in its value range, for example, the value range of the servo motor rotation speed is [ P1, P3] (rpm), the servo motor rotation speed takes three horizontal numbers P1, P2, and P3 distributed in an arithmetic progression, where P2=0.5 (P1, P3), and the 3 horizontal numbers of each configuration parameter are combined according to orthogonal experimental combinations to obtain 9 combinations, where each combination may correspond to multiple configuration information. And after the training set is obtained, training the error back propagation neural network by adopting the training set until convergence, and obtaining a driving speed determination model. The error back propagation neural network comprises an input layer, a hidden layer and an output layer, wherein the number of the hidden layer is 1, the initial value of the number of the neuron nodes in the hidden layer is (m multiplied by n) ^0.5+1, wherein m is the number of the neuron nodes in the input layer, and n is the number of the neuron nodes in the output layer.
Based on the setting of the training set, the rotating speed and the torque of the servo motor are used as first configuration parameters, the training set is introduced, so that the neural network can learn the internal relation and the rule between the rotating speed and the torque of the servo motor and the driving speed respectively, the influence of the rotating speed and the torque of the servo motor on the driving speed can be considered when the trained driving speed determining model outputs the driving speed, and the driving speed can be adapted to the rotating speed and the torque of the servo motor. The size and the load weight of the bearing surface of the unmanned transport vehicle are used as second configuration parameters, a training set is introduced, so that the neural network can learn the internal connection and the rule between the size and the running speed of the bearing surface and the internal connection and the rule between the load weight and the running speed, the influence of the size and the load weight of the bearing surface on the running speed can be considered when the trained running speed determining model outputs the running speed, and the running speed can be adapted to the size and the load weight of the bearing surface. Combining 3 horizontal numbers of each configuration parameter distributed in an arithmetic progression according to orthogonal experiment combinations to obtain a plurality of configuration information, and forming a training set, so that the training set can uniformly cover the corresponding relation between the configuration parameter combinations (configuration information) under different horizontal numbers and the measured speed, and a neural network can learn the internal relation and the rule between the configuration parameter combinations (configuration information) under different horizontal numbers and the measured speed, so that the trained driving speed determination model has universality, namely, the trained driving speed determination model can output the adaptive driving speed in the face of unmanned vehicles under various different configuration parameters.
In addition, the error back propagation neural network has stronger nonlinear mapping capability, can fully learn the internal relation and rule between the configuration information and the measured speed, and comprises a hidden layer, so that the calculation amount is less, overfitting cannot occur, and finally, the initial value of the number of the neuron nodes in the hidden layer is (m multiplied by n) ^0.5+1, so that the number of the neuron nodes in the hidden layer is between the number of the neuron nodes in the input layer and the number of the neuron nodes in the output layer, the model convergence can be accelerated, and the accuracy is ensured.
It should be noted that the above-described device embodiments are merely illustrative, where the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a general hardware platform, and certainly can also be implemented by hardware. It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a computer readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), or the like.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; within the context of the present application, where technical features in the above embodiments or in different embodiments may also be combined, the steps may be implemented in any order and there are many other variations of the different aspects of the present application described above which are not provided in detail for the sake of brevity; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and these modifications or substitutions do not depart from the scope of the technical solutions of the embodiments of the present application.
Claims (12)
1. A method of training a travel speed determination model, the method comprising:
acquiring a training set, wherein the training set comprises a plurality of configuration information, each configuration information corresponds to one measured speed, and the configuration information comprises a first configuration parameter influencing the carrying time of the unmanned transport vehicle and a second configuration parameter influencing the driving stability of the unmanned transport vehicle;
and training a preset neural network by adopting the training set until the neural network is converged to obtain the driving speed determination model.
2. The method of claim 1, wherein the first configuration parameter includes a speed of rotation of a servo motor in the automated guided vehicle and/or a torque of a servo motor in the automated guided vehicle.
3. The method of claim 1 or 2, wherein the second configuration parameter comprises a load bearing surface size and/or a load weight of the automated guided vehicle.
4. The method of any one of claims 1-3, wherein the obtaining the training set comprises:
obtaining the value range of each configuration parameter, wherein one configuration parameter takes at least three levels;
combining at least three horizontal numbers of each configuration parameter according to a preset combination rule to obtain a plurality of configuration information;
and acquiring the measured speeds corresponding to the configuration information respectively, and taking the configuration information and the measured speeds corresponding to the configuration information respectively as the training set.
5. The method of claim 4, wherein the predetermined combination rule is an orthogonal trial combination.
6. The method of claim 4, wherein at least three horizontal values of a configuration parameter are distributed in an arithmetic progression.
7. The method of claim 1, wherein the neural network is an error back propagation neural network, the error back propagation neural network comprises an input layer, an implied layer and an output layer, the number of the implied layer is 1, and an initial value of the number of neuron nodes in the implied layer is (m x n) ^0.5+1, where m is the number of neuron nodes in the input layer and n is the number of neuron nodes in the output layer.
8. A travel speed determination method, characterized by comprising:
acquiring configuration information of an automated guided vehicle, wherein the configuration information comprises a first configuration parameter influencing the transporting time of the automated guided vehicle and a second configuration parameter influencing the driving stability of the automated guided vehicle;
inputting the configuration information into a travel speed determination model trained using the method of any one of claims 1-7 to obtain a travel speed of the automated guided vehicle.
9. An exercise apparatus, comprising:
a processor, and
a memory communicatively coupled to the processor, wherein,
the memory stores instructions executable by the processor to enable the processor to perform the method of any one of claims 1-7.
10. A readable storage medium, characterized in that it stores a program or instructions which, when executed by a processor, implement the method according to any one of claims 1-8.
11. An automated guided vehicle, comprising:
a vehicle body;
the programmable logic controller is used for controlling the vehicle body to operate;
a processor and a memory, the memory storing a program or instructions for executing the program or instructions to implement the method of claim 8 to obtain a travel speed of the automated guided vehicle;
the processor is also used for sending the running speed to the programmable logic controller so that the programmable logic controller controls the vehicle body to run according to the running speed.
12. An electric vehicle charging system comprising a plurality of the automated guided vehicle of claim 11.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111151474.2A CN115909712A (en) | 2021-09-29 | 2021-09-29 | Training method of travel speed determination model, travel speed determination method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111151474.2A CN115909712A (en) | 2021-09-29 | 2021-09-29 | Training method of travel speed determination model, travel speed determination method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115909712A true CN115909712A (en) | 2023-04-04 |
Family
ID=86495224
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111151474.2A Pending CN115909712A (en) | 2021-09-29 | 2021-09-29 | Training method of travel speed determination model, travel speed determination method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115909712A (en) |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20110090185A (en) * | 2010-02-03 | 2011-08-10 | 한국과학기술원 | System and method for charging the battery of electric vehicle |
CN110347155A (en) * | 2019-06-26 | 2019-10-18 | 北京理工大学 | A kind of intelligent vehicle automatic Pilot control method and system |
KR20190140316A (en) * | 2018-06-11 | 2019-12-19 | 현대모비스 주식회사 | Charging robot for electric vehicle and control method thereof |
CN110936824A (en) * | 2019-12-09 | 2020-03-31 | 江西理工大学 | Electric automobile double-motor control method based on self-adaptive dynamic planning |
CN111624992A (en) * | 2020-04-28 | 2020-09-04 | 北京科技大学 | Path tracking control method of transfer robot based on neural network |
CN111619388A (en) * | 2020-07-02 | 2020-09-04 | 孙旭阳 | Intelligent charging device for electric automobile |
US11037320B1 (en) * | 2016-03-01 | 2021-06-15 | AI Incorporated | Method for estimating distance using point measurement and color depth |
-
2021
- 2021-09-29 CN CN202111151474.2A patent/CN115909712A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20110090185A (en) * | 2010-02-03 | 2011-08-10 | 한국과학기술원 | System and method for charging the battery of electric vehicle |
US11037320B1 (en) * | 2016-03-01 | 2021-06-15 | AI Incorporated | Method for estimating distance using point measurement and color depth |
KR20190140316A (en) * | 2018-06-11 | 2019-12-19 | 현대모비스 주식회사 | Charging robot for electric vehicle and control method thereof |
CN110347155A (en) * | 2019-06-26 | 2019-10-18 | 北京理工大学 | A kind of intelligent vehicle automatic Pilot control method and system |
CN110936824A (en) * | 2019-12-09 | 2020-03-31 | 江西理工大学 | Electric automobile double-motor control method based on self-adaptive dynamic planning |
CN111624992A (en) * | 2020-04-28 | 2020-09-04 | 北京科技大学 | Path tracking control method of transfer robot based on neural network |
CN111619388A (en) * | 2020-07-02 | 2020-09-04 | 孙旭阳 | Intelligent charging device for electric automobile |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wu et al. | Continuous reinforcement learning of energy management with deep Q network for a power split hybrid electric bus | |
Vepsäläinen et al. | Development and validation of energy demand uncertainty model for electric city buses | |
CN110991757B (en) | Comprehensive prediction energy management method for hybrid electric vehicle | |
Abousleiman et al. | Electric vehicle modelling and energy‐efficient routing using particle swarm optimisation | |
Taghavipour et al. | An optimal power management strategy for power split plug-in hybrid electric vehicles | |
Drungilas et al. | Deep reinforcement learning based optimization of automated guided vehicle time and energy consumption in a container terminal | |
Chen et al. | A new hybrid model predictive controller design for adaptive cruise of autonomous electric vehicles | |
Sun et al. | Power management in hybrid electric vehicles using deep recurrent reinforcement learning | |
CN117371739A (en) | Vehicle charging management method and device, electronic equipment and storage medium | |
Bazhynova et al. | Neural Network Model of Assessing the Technical Condition of the Power Unit of a Hybrid Vehicle | |
CN115339326A (en) | Torque distribution method for four-wheel drive vehicle, terminal equipment and storage medium | |
Sun et al. | Intelligent estimation for electric vehicle mass with unknown uncertainties based on particle filter | |
CN115909712A (en) | Training method of travel speed determination model, travel speed determination method and device | |
Ceraolo et al. | Hybridisation of forklift trucks | |
CN114881287A (en) | Energy optimization method for industrial wireless chargeable sensor network | |
CN114971152A (en) | Cooperative control method, system, equipment and medium for truck group | |
Bian et al. | Control parameter optimization for automobile cruise control system via improved differential evolution algorithm | |
Jung et al. | Design space exploration for powertrain electrification using Gaussian processes | |
Hailong et al. | Study on HEV’s driving condition recognition method based on PSO algorithm | |
Kwon et al. | Driving Profile Optimization Using a Deep Q‐Network to Enhance Electric Vehicle Battery Life | |
CN110516845A (en) | A kind of course continuation mileage prediction technique and device | |
CN114003030B (en) | Path tracking control method of two-wheel differential mobile robot considering centroid change | |
CN114756025B (en) | Autonomous trolley cruise control method and device | |
CN114952839B (en) | Cloud edge cooperation-based two-stage robot motion decision system | |
CN114118926B (en) | Vehicle task distribution method based on co-evolution algorithm of denoising self-encoder |
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 |