CN106896723B - A kind of intelligent automatic gear shifting control system of bicycle - Google Patents

A kind of intelligent automatic gear shifting control system of bicycle Download PDF

Info

Publication number
CN106896723B
CN106896723B CN201710208591.5A CN201710208591A CN106896723B CN 106896723 B CN106896723 B CN 106896723B CN 201710208591 A CN201710208591 A CN 201710208591A CN 106896723 B CN106896723 B CN 106896723B
Authority
CN
China
Prior art keywords
bicycle
hidden layer
condition signal
signal
neural network
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.)
Active
Application number
CN201710208591.5A
Other languages
Chinese (zh)
Other versions
CN106896723A (en
Inventor
代晓萌
姚鑫骅
朱雨贺
王润秋
张鑫磊
傅建中
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN201710208591.5A priority Critical patent/CN106896723B/en
Publication of CN106896723A publication Critical patent/CN106896723A/en
Application granted granted Critical
Publication of CN106896723B publication Critical patent/CN106896723B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/0265—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion
    • G05B13/027—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion using neural networks only
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B62—LAND VEHICLES FOR TRAVELLING OTHERWISE THAN ON RAILS
    • B62M—RIDER PROPULSION OF WHEELED VEHICLES OR SLEDGES; POWERED PROPULSION OF SLEDGES OR SINGLE-TRACK CYCLES; TRANSMISSIONS SPECIALLY ADAPTED FOR SUCH VEHICLES
    • B62M25/00—Actuators for gearing speed-change mechanisms specially adapted for cycles
    • B62M25/08—Actuators for gearing speed-change mechanisms specially adapted for cycles with electrical or fluid transmitting systems
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance

Landscapes

  • Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Transportation (AREA)
  • Mechanical Engineering (AREA)
  • Control Of Transmission Device (AREA)

Abstract

The invention discloses a kind of intelligent automatic gear shifting control systems of bicycle, including speed changer, further includes: detection module detects the physical condition and bicycle vehicle condition of rider, and generates physical condition signal and vehicle condition signal；ANN Control module has and forms computation model by sample data training, and the computation model obtains the physical condition signal and the variable rate signal of current bicycle is calculated in vehicle condition signal；Physical condition signal and vehicle condition signal update computation model are used simultaneously；Change gear control module controls speed changer according to the variable rate signal；The data of present invention combination bicycle and human body are analyzed by calculating, and realize the accurate speed change of automatic intelligent, improve riding-balance, are improved riding safety, are greatly improved experience of riding；The automatic gear-shifting strategy based on BP neural network is proposed, best gear identification problem can be well solved, and the requirement of automatic speed-transformation of bicycle can be met accurately and in time.

Description

A kind of intelligent automatic gear shifting control system of bicycle
Technical field
The present invention relates to the speed Control technical field of bicycle, in particular to the intelligent fluid drive control of a kind of bicycle System processed.
Background technique
Even to this day, many years are pass by by the mountain bike introducing country, and as time goes by, bicycle becomes increasingly It is advanced, it is more and more professional, it is certain also more and more expensive；Recent years is ridden due to the increase of everybody environmental consciousness and health perception It is increasingly becoming everybody a kind of entertainment selection and hobby.With the development of economy and science and technology, people are for health and move more Value, user for bicycle demand also just from traditional ride instead of walk type vehicles divertical motion, body-building, tourism and leisure etc. it is more Function bicycle.It rides increasingly fashionable, demand for services of the bicyclist in ride tends to diversification, intelligence.In environment Under the dual-pressure of traffic, bicycle also becomes the choice for traveling of more people.At the same time, the development of bike tech Make rapid progress, there are many innovations from the everyways such as power, car body materials and structure, transmission gear shift technology.Began from 2015, Diversified intelligent bicycle enters consumption market, becomes another selection of high-end bicycle purchasing group.
Bicycle shift system is to change speed by the cooperation for changing the toothed disc of chain and different forward and backward sizes Speed.The size of preceding fluted disc and the size of rear tooth disk determine dynamics when bicycle turn pedal.Preceding fluted disc is bigger, rear tooth disk More hour, when pedal, more feel laborious.Preceding fluted disc is smaller, and when rear tooth disk is bigger, when pedal more feels relief.According to different drivers Ability, or the different section of reply, road conditions can adjust the speed of bicycle by adjusting the size of forward and backward fluted disc.Now Manual transmission commonly used by the market, which generally refers to, to be dialled or handle, when speed change, by pulling cable to overcome the bullet of elastic element Power moves chain guide element on rear sprocket wheel, and the bias force for discharging elastic element makes chain guide element in the opposite direction Movement.Inevitably there are some problems using the rear chiain-moving device of cable: 1, structure is complicated, and manufacturing cost is high；2, A biggish power is needed when shift, especially firmly particularly evident when low speed or upward slope, bicyclist is with greatly Power shift, also brings complete hidden danger to riding-balance；3, using the control mode of cable, it generally cannot achieve continuous shift, ride Row experience is poor.
And electrical motor gearshift is used, it is the inexorable trend that speed changer is converted into electrical shift by mechanical speed change.Become in conjunction with electronics Speed and internet realize that the intelligent speed changing of bicycle is current research hotspot.Such as the patent of publication number CN 101037134A Document discloses a kind of automatic speed-changing system, is mounted in bicycle, includes: can detect the heart beat detection list of rider's heartbeat Member is electrically connected with the heart beat detection unit and can execute the heartbeat processing unit for being relevant to heartbeat message operation and the heartbeat Processing unit is electrically connected and can generate the driving unit of driving signal, and is electrically connected with the driving unit and can be according to aforementioned Driving signal drives the automatic transmission unit of the bicycle speed-varying unit speed change.Whereby, enable the riding speed foundation of the bicycle The heartbeat change rate of rider, auto-changing riding speed.
But the above method can not accurately reflect that the speed change situation that people currently needs, intelligence be not still high.
Summary of the invention
The present invention provides a kind of intelligent automatic gear shifting control system of bicycle, real-time update control systems, thus real The intelligent fluid drive of existing bicycle, is more in line with the own bodies situation of driving situation and car owner.
A kind of intelligent automatic gear shifting control system of bicycle, including speed changer, further includes:
Detection module, detects the physical condition and bicycle vehicle condition of rider, and generates physical condition signal and vehicle condition Signal；
ANN Control module, has and forms computation model by sample data training, and the computation model obtains institute It states physical condition signal and the variable rate signal of current bicycle is calculated in vehicle condition signal；Physical condition signal and vehicle are used simultaneously Condition signal update computation model；
Change gear control module controls speed changer according to the variable rate signal, can integrate in mobile device terminal. Change gear control module is integrated in bicycle terminal.
It preferably, further include Cloud Server, the ANN Control module is integrated in the Cloud Server, the cloud clothes Business device further include:
Database, for storing the physical condition data, rider's information and variable rate signal of bicycle vehicle condition, rider；
Access modules, for carrying out data interaction with database.
Bicycle and bicyclist are accessed internet, collect and survey car body and somatic data using big data cloud platform, it is real The intelligent fluid drive of existing bicycle, and by the Human Physiology index of mobile device terminal feedback bicyclist, meet user The experience of riding of property.
BP neural network has the advantages that structure is simple, is easily achieved, and has in terms of the fitting of non-linear continuous function It is widely applied.Therefore, the present invention uses BP neural network.By comprehensively considering, the present invention is by the network design at including eight A input, the network structure of an output and a hidden layer.8 inputs are respectively gender S, age A, height H, weight W, the heart Rate R, vehicle velocity Vi, step on frequency Ci, gradient I repeats, and exports the gear value G for prediction, hidden layer includes 8 neuron nodes.
Network is trained before neural network prediction, make network that there is associative memory and forecast function by training.Face first Product experiment is carried out to different age group and different sexes crowd, (age, gender, height, weight, heart rate, speed are stepped on for acquisition Frequently, the gradient) with the corresponding informance of gear, i.e. sample data is trained neural network model as training set, obtains Obtaining one can be used for predicting the model of gear, and be applied to product.Preferably, calculating is formed by sample data training Specific step is as follows for model:
Step 1, the weight w between input layer and hidden layer is initializedijAnd the weight w between hidden layer and output layerj, Hidden layer threshold value a, output layer threshold value b are initialized, determines that learning rate and neuron excitation function, i indicate i-th of input layer, j Indicate j-th of node of hidden layer；
Step 2, the connection weight weight values w according to input variable X, between input layer and hidden layerijAnd hidden layer threshold value a, It calculates hidden layer and exports H；
Step 3, H, connection weight w are exported according to the hidden layer that step 2 is calculatedjWith threshold value b, it is pre- to calculate neural network Survey output G, that is, the gear value predicted；
Step 4, G is exported according to neural network desired output Y and neural network prediction, calculates neural network prediction error E；
E=Y-G
Step 5, connection weight w is updated according to neural network prediction error Eij, wj；
wij=wij+ηHj(1-Hj)x(i)wjE
wj=wj+ηHjE
In formula, η is learning rate.
Step 6, predict that error E updates network node threshold value a, b according to BP neural network；
aj=aj+ηHj(1-Hj)wjE
B=b+E
Step 7, judge whether algorithm iteration terminates, if being not over, return step 2.
Preferably, in step 2, the expression formula of each node is as follows:
In formula, l is node in hidden layer, l=8 in this formula；F is general hidden layer excitation function, and output area is [- 1,1].Mind Stimulation through member accumulation is the sum of the quantity of stimulus passed over by input layer and corresponding weight.
The general hidden layer excitation function can be the differentiable function being increased monotonically there are many expression-form, Tan-Sigmoid, Flatness is good, and derivative form is simple, and output area is limited, and data are not easy to dissipate during transmitting, it is preferred that institute Stating general hidden layer excitation function is Tan-Sigmoid, and Tan-Sigmoid function expression is as follows:
To realize the optimal speed change of bicycle, need by neural network algorithm in conjunction with machine autonomous learning, i.e., using online The partial derivative (gradient of parameter) in neural network algorithm is constantly updated in study, thus Intelligent Optimal speed change model.It is excellent Choosing, updating computation model using physical condition signal, vehicle condition signal and variable rate signal, specific step is as follows:
(1) current physical condition signal and vehicle condition signal are obtained；
(2) its gradient for corresponding to parameter is calculated to the data in step (1)；
Have for j-th of output neuron:Wherein Δ ω (i, j) indicates i-th of input Layer, the weight w of j-th of nodeijGradient (and increment)；
Then it obtains:
The derived function of known Tan-Sigmoid function:
F ' (x)=1-f (x)2
Then have:
δij=(Gj-Yj)·(1-Yj 2)
(3) gradient is updated according to the result of step (2) and obtains new parameter, and then update computation model；
(5) data of step (1) are abandoned.
It preferably, further include mobile device terminal, the mobile device terminal includes:
Input module inputs personal information for user；
Location identification module, for generating the current location information of bicycle；
Communication module receives the physical condition signal of detection module and vehicle condition signal and is sent to Cloud Server.
Preferably, physical condition includes: age, gender, height, weight and heart rate；Vehicle condition signal includes: speed, steps on frequency And the gradient.
By intelligence rack the system integration sensor and wear the intelligent detection equipments such as heart rate band and carry out car body and human body number According to real-time acquisition, and be uploaded to cloud database, pass through the optimization that following on-line study strategy carries out model.
The prediction of gear value can be carried out using the model of optimization according to real time data when user rides.When When new input data is uploaded to Cloud Server, input data is substituted into the model after optimization, obtains prediction gear value.If prediction Gear value is greater than current gear value, shifts up, otherwise downshift.
The automatic speed-changing system that the present invention describes is combined with Cloud Server, generates an initial model by primary data, Then the ride time series data stream of generation of user is further learnt.For example, for specific user, every production A raw new data just carry out new study, abandon this data after the completion of study again.This way is advantageous in that, we Algorithm can be well adapted for the tendentiousness of user, algorithm can be continuously updated model for the current behavior of user with suitable Using family.In short, traditional model is the variable block strategy for making user go habit, adapt to bicycle, and Neural Network Online learns Model can make the model of variable block strategy be more in line with the habit of user, provide the service of customization for every bicyclist.
Traditional neural network generally uses batch type learning algorithm to be trained network, is limited to be trained to data Scale, can not effectively handle ultra-large data set and online data stream.For bicycle shift system of the present invention System acquires bicycle and somatic data in real time, and data volume is larger, and batch type learning algorithm can not be effectively treated, and neural Network on-line study is constantly to learn to newly generated data, just abandons this data after the completion of study, therefore can handle Big data quantity and on-line training can carry out real-time, quickly model adjustment according to feedback data on line, so that model is anti-in time The variation on line is reflected, the accuracy rate predicted on line is improved.Process substantially includes: to be become with the prediction result control bicycle of model Then speed collects the data of riding of user, then is used to training pattern, the system for forming closed loop.
For cycling road conditions are changeable, human body physiological state is complicated, are difficult to carry out Modeling Research with traditional method The case where, the automatic gear-shifting strategy based on BP neural network is proposed, which can carry out self-teaching according to user data, And the parameter of neural network is corrected, Optimized model efficiently solves the insoluble non-linear and real-time variability of conventional method System control problem.BP neural network can well solve best gear identification problem, and can meet accurately and in time voluntarily The requirement of vehicle fluid drive.
(such as 3G mobile communications network and 4G are moved using mobile network for the mobile device terminal and the Cloud Server Dynamic communication network) it carries out wireless communication.Detection module using intelligent detection equipment can be for example, as it is known that Intelligent bracelet, intelligence Energy wrist-watch, intelligent finger ring, code table, heart rate band, sensor etc., but not limited to this, and intelligent detection equipment can be including institute There is any intelligent detection equipment of above structure.
Beneficial effects of the present invention:
The intelligent automatic gear shifting control system of bicycle of the invention passes through calculating in conjunction with the data of bicycle and human body The accurate speed change of automatic intelligent is realized in analysis, improves riding-balance, is improved riding safety, is greatly improved experience of riding；It mentions The automatic gear-shifting strategy based on BP neural network is gone out, best gear identification problem can have been well solved, and can in time, accurately Ground meets the requirement of automatic speed-transformation of bicycle.
Detailed description of the invention
Fig. 1 is the structure line frame graph of the intelligent automatic gear shifting control system of the bicycle of the present embodiment.
Fig. 2 is the flow diagram of the application method of the intelligent automatic gear shifting control system of the bicycle of the present embodiment.
Fig. 3 is that line neural network model treatment process is shown in the intelligent automatic gear shifting control system of the bicycle of embodiment It is intended to.
Fig. 4 is the neural network model of the intelligent automatic gear shifting control system of the bicycle of embodiment.
Specific embodiment
Below by preferred embodiment, and in conjunction with attached drawing, the technical solutions of the present invention will be further described.
As shown in Figure 1, the intelligent automatic gear shifting control system of the bicycle of the present embodiment include: intelligent detection equipment 10, Mobile device terminal 20, bicycle terminal 30 and Cloud Server 40, wherein intelligent detection equipment 10 is typically integrated in mobile device In terminal 20.Mobile device terminal 20 is connect with intelligent detection equipment 10, bicycle terminal 30 and Cloud Server 40, to use Family can receive data from intelligent detection equipment 10 by mobile device terminal 20, transmit data to Cloud Server 40 and take from cloud Business device 40 obtains control signal, and sends control signal to bicycle terminal 30, is realized by electronic transmission intelligent automatic Speed change, and shown in mobile device terminal 20 to feed back Human Physiology index.
Mobile device terminal 20 includes: communication module 21, information acquisition module 22 and location identification module 23, input mould Block 24 and display module 25.
Bicycle terminal 30 includes: communication module 31 and control module 32.
Cloud Server 40 includes: database, communication module 41, database access module 42 and ANN Control module 43。
As shown in Figures 2 and 3, at runtime, the intelligent automatic gear shifting control system of the bicycle of the present embodiment is using following Method carry out:
1) intelligent detection equipment 10 detect bicycle and human body data, bicycle data include step on frequency, speed, the gradient, The information such as gear, somatic data include the information such as heart rate.
2) mobile device terminal 20 is communicated with intelligent detection equipment 10, acquires intelligent inspection by information acquisition module 22 The data of measurement equipment 10 are stored and are shared, and feed back Human Physiology index by display module 25, and by bicycle and people Volume data, personal information are transferred to Cloud Server 40.
Wherein, according to actual needs, user needs to input personal information to the input module 24 of mobile device terminal 20, with Just the experience of riding for being suitble to itself is provided for user, personal information includes the information such as gender, height, weight, age.
3) Cloud Server 40 is communicated with mobile device terminal 20, receives the data that mobile device terminal 20 is sent Information accesses to database by database access module 42, stores data in database, and neural network control is utilized Molding block 43 is analyzed and processed received data, obtains bicycle speed and controls signal, and control signal is transferred to shifting Dynamic device end 20.
Wherein, ANN Control module 43 uses the automatic gear-shifting strategy based on BP neural network, and analysis is handled voluntarily The data of vehicle and human body, and obtain the prediction gear value for being suitble to each bicyclist.
4) bicycle terminal 30 is communicated with mobile device terminal 20, receives the control that mobile device terminal 20 is sent Signal processed, and send control signal to control module 32, the fluid drive of bicycle intelligence is realized by electrical shift module 33.
The present embodiment selects BP neural network.By comprehensively considering, which is inputted at comprising eight, one defeated Out with the network structure of a hidden layer.8 inputs respectively gender S, age A, height H, weight W, heart rate R, vehicle velocity V are stepped on Frequency C, gradient I are repeated, and export the gear value G for prediction, and hidden layer includes 8 neuron nodes.
Network is trained before neural network prediction, make network that there is associative memory and forecast function by training.Face first Product experiment is carried out to different age group and different sexes crowd, (age, gender, height, weight, heart rate, speed are stepped on for acquisition Frequently, the gradient) with the corresponding informance of gear, i.e. sample data is trained neural network model as training set, obtains Obtaining one can be used for predicting the model of gear, and be applied to product.
As shown in figure 4, mainly including following steps using the process of sample data training neural network:
Step 1: the initialization of neural network.Initialize the weight w between input layer and hidden layerijAnd hidden layer and defeated Weight w between layer outj, hidden layer threshold value a is initialized, output layer threshold value b determines learning rate and neuron excitation function.
Step 2: calculating hidden layer output.Connection weight weight values w according to input variable X, between input layer and hidden layerijWith And hidden layer threshold value a, it calculates hidden layer and exports H, the expression formula of each node is such as shown in (1).
In formula, l is node in hidden layer, l=8 in this formula；F is general hidden layer excitation function, which can be there are many table Up to form, function selected by this paper is Tan-Sigmoid, and output area is [- 1,1], and Tan-Sigmoid function expression is as follows:
Step 3: calculating the output of output layer.H, connection weight w are exported according to hidden layerjWith threshold value b, BP nerve net is calculated Network prediction output G.
Step 4: calculating error.According to neural network desired output Y and prediction output G, neural network prediction error is calculated E。
E=Y-G (4)
Step 5: updating weight.Predict that error E updates connection weight w according to BP neural networkij, wj。
wij=wij+ηHj(1-Hj)x(i)wjE (5)
wj=wj+ηHjE (6)
In formula, η is learning rate.
Step 6: updating threshold value.Predict that error E updates network node threshold value a, b according to BP neural network.
aj=aj+ηHj(1-Hj)wjE (7)
B=b+E (8)
Step 7: judge whether algorithm iteration terminates, if being not over, return step 2.
To realize the optimal speed change of bicycle, need by neural network algorithm in conjunction with machine autonomous learning, i.e., using online The partial derivative (gradient of parameter) in neural network algorithm is constantly updated in study, thus Intelligent Optimal speed change model.
By intelligence rack the system integration sensor and wear the intelligent detection equipments such as heart rate band and carry out car body and human body number According to real-time acquisition, and be uploaded to cloud database, pass through the optimization that following on-line study strategy carries out model:
Recycled (when riding) when bicycle is in operating status
Obtain the data & somatic data and current gear of bicycle at this time；
Above-mentioned data are uploaded to Cloud Server；
Its gradient for corresponding to parameter is calculated to this data；
It updates gradient and obtains new parameter, thus more new model；
Abandon data；
}
The prediction of gear value can be carried out using the model of optimization according to real time data when user rides.When When new input data is uploaded to Cloud Server, input data is substituted into the model after optimization, obtains prediction gear value.If prediction Gear value is greater than current gear value, shifts up, otherwise downshift.
In the present embodiment, mobile device terminal 20 is provided with communication module 21 to examine via cordless communication network and intelligence Measurement equipment 10, bicycle terminal 30 and Cloud Server 40 communicate, and receive data from intelligent detection equipment 10 and Cloud Server 40, will Data are transferred to Cloud Server 40 and bicycle terminal 30.Mobile device terminal 20 is configured to include institute's riding cycle for identification The location identification module 23 of position, for example, GPS module, for generating the riding track of user.
Mobile device terminal 20 can be for example, as it is known that smart phone, tablet computer etc., but not limited to this, and eventually End 20 can be any terminal including all above structures.
It on the other hand, can be with as the communication between terminal 20 and intelligent detection equipment 10, terminal 30, Cloud Server 40 Using wireless communication, such as bluetooth, mobile communications network (such as 3G mobile communications network and 4G mobile communications network).
Bicycle terminal 30 is known multi-speed bicycle, is equipped with electronic transmission, but not limited to this, and bicycle Terminal can be any bicycle terminal including above structure.
The above description is only a preferred embodiment of the present invention, not thereby limits scope of patent protection of the invention, all It is directly or indirectly to be used in other relevant technologies with equivalent structure transformation made by description of the invention and accompanying drawing content Field similarly includes within the scope of the present invention.

Claims (4)

1. a kind of intelligent automatic gear shifting control system of bicycle, including speed changer, which is characterized in that further include:
Detection module, detects the physical condition and bicycle vehicle condition of rider, and generates physical condition signal and vehicle condition signal；
ANN Control module, has and forms computation model by sample data training, and the computation model obtains the body The variable rate signal of current bicycle is calculated in body situation signal and vehicle condition signal；Believed simultaneously using physical condition signal and vehicle condition Number update computation model；
Change gear control module controls speed changer according to the variable rate signal；
Forming computation model by sample data training, specific step is as follows:
Step 1: the initialization of neural network initializes the weight w between input layer and hidden layerijAnd hidden layer and output layer Between weight wj, hidden layer threshold value a, output layer threshold value b are initialized, determines that learning rate and neuron excitation function, i indicate I-th of input layer, j indicate j-th of node of hidden layer；
Step 2: calculating hidden layer output, the weight w according to input variable X, between input layer and hidden layerijAnd hidden layer threshold Value a calculates hidden layer and exports H, and the expression formula of each node such as (1) is shown,
In formula, 1 is node in hidden layer, 1=8 in this formula；F be general hidden layer excitation function Tan-Sigmoid, output area be [- 1,1], Tan-Sigmoid function expression is as follows:
Step 3: calculating the output of output layer, H, weight w are exported according to hidden layerjWith output layer threshold value b, BP neural network is calculated Prediction output G,
Step 4: error is calculated, according to neural network desired output Y and prediction output G, calculates neural network prediction error E,
E=Y-G (4)
Step 5: updating weight, predict that error E updates weight w according to BP neural networkij, wj,
wij=wij+ηHj(1-Hj)x(i)wjE (5)
wj=wj+ηHjE (6)
In formula, η is learning rate；
Step 6: threshold value is updated, predicts that error E updates hidden layer threshold value a, output layer threshold value b according to BP neural network,
aj=aj+ηHj(1-Hj)wjE (7)
B=b+E (8)
Step 7: judge whether algorithm iteration terminates, if being not over, return step 2；
Updating computation model using physical condition signal, vehicle condition signal and variable rate signal, specific step is as follows:
(1) current physical condition signal and vehicle condition signal are obtained；
(2) its gradient for corresponding to parameter is calculated to the data in step (1)；
Have for j-th of output neuron:Wherein Δ ω (i, j) indicates i-th of input layer, the The weight w of j nodeijGradient；
Then it obtains:
The derived function of known Tan-Sigmoid function:
F ' (x)=1-f (x)2
Then have:
δij=(Gj-Yj)·(1-Yj 2)
(3) gradient is updated according to the result of step (2) and obtains new parameter, and then update computation model；
(4) data of step (1) are abandoned.
2. the intelligent automatic gear shifting control system of bicycle as described in claim 1, which is characterized in that further include cloud service Device, the ANN Control module are integrated in the Cloud Server, the Cloud Server further include:
Database, for storing the physical condition data, rider's information and variable rate signal of bicycle vehicle condition, rider；
Access modules, for carrying out data interaction with database.
3. the intelligent automatic gear shifting control system of bicycle as claimed in claim 2, which is characterized in that further include mobile device Terminal, the mobile device terminal include:
Input module inputs personal information for user；
Location identification module, for generating the current location information of bicycle；
Communication module receives the physical condition signal of detection module and vehicle condition signal and is sent to Cloud Server.
4. the intelligent automatic gear shifting control system of bicycle as described in claim 1, which is characterized in that physical condition includes: Age, gender, height, weight and heart rate；Vehicle condition signal includes: speed, steps on frequency and the gradient.
CN201710208591.5A 2017-03-31 2017-03-31 A kind of intelligent automatic gear shifting control system of bicycle Active CN106896723B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710208591.5A CN106896723B (en) 2017-03-31 2017-03-31 A kind of intelligent automatic gear shifting control system of bicycle

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710208591.5A CN106896723B (en) 2017-03-31 2017-03-31 A kind of intelligent automatic gear shifting control system of bicycle

Publications (2)

Publication Number Publication Date
CN106896723A CN106896723A (en) 2017-06-27
CN106896723B true CN106896723B (en) 2019-08-13

Family

ID=59194375

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710208591.5A Active CN106896723B (en) 2017-03-31 2017-03-31 A kind of intelligent automatic gear shifting control system of bicycle

Country Status (1)

Country Link
CN (1) CN106896723B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11866114B2 (en) 2019-01-31 2024-01-09 Shimano Inc. Control device, creation method for learning model, learning model, computer program and storage medium

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107310681A (en) * 2017-07-04 2017-11-03 苏州八方电机科技有限公司 A kind of speed change management system for electric bicycle
CN107423761B (en) * 2017-07-24 2021-01-26 清华大学 Rail locomotive energy-saving optimization operation method based on feature selection and machine learning
CN108126341A (en) * 2017-12-26 2018-06-08 歌尔科技有限公司 A kind of method, apparatus and equipment of the seat parameter for adjusting Virtual Reality Platform
JP6985217B2 (en) * 2018-07-09 2021-12-22 株式会社シマノ Control data creation device, component control device, control data creation method, component control method, and computer program
JP6989452B2 (en) * 2018-07-09 2022-01-05 株式会社シマノ Creation device, component control device, creation method, component control method, and computer program
JP7007244B2 (en) * 2018-07-09 2022-01-24 株式会社シマノ Creation device, component control device, creation method, component control method, computer program and learning model
JP7312549B2 (en) * 2018-12-18 2023-07-21 株式会社シマノ Human-powered vehicle controller, electronic device, and human-powered vehicle control system
US11527981B2 (en) 2018-07-09 2022-12-13 Shimano Inc. Human-powered vehicle control device, electronic device, and human-powered vehicle control system
CN109018184B (en) * 2018-10-23 2020-12-08 深圳市几米物联有限公司 Intelligent speed changing method and device for bicycle and intelligent speed changing bicycle
CN112158285B (en) * 2020-09-14 2022-04-01 杭州电子科技大学 Real-time speed change suggestion system for mountain bike
CN114684321B (en) * 2020-12-25 2024-06-04 株式会社岛野 Control device for man-power driven vehicle
CN112882410A (en) * 2021-01-11 2021-06-01 太原理工大学 Real-time control's cabin system of riding that man-machine environment fuses altogether
JP7316318B2 (en) * 2021-03-31 2023-07-27 本田技研工業株式会社 Sharing system and sharing method

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5059158A (en) * 1990-05-08 1991-10-22 E.B.T., Inc. Electronic transmission control system for a bicycle
IT1310144B1 (en) * 1999-08-24 2002-02-11 Ferrero Spa SYSTEM AND PROCEDURE FOR THE CONTROL OF TRANSMISSIONS WITH VARIABLE RATIO
CN101037134A (en) * 2006-03-17 2007-09-19 捷安特(中国)有限公司 Automatic speed-changing system
CN104044689A (en) * 2014-07-03 2014-09-17 钟明 Bicycle with automatic speed changing function
CN105564584B (en) * 2016-01-27 2018-07-27 深圳市七尚科技有限公司 Automatic transimission and its bicycle, the method for changing speed of multi-speed bicycle
CN106080941B (en) * 2016-08-04 2018-12-18 清华大学 A kind of unmanned bicycle for realizing speed change balance control

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11866114B2 (en) 2019-01-31 2024-01-09 Shimano Inc. Control device, creation method for learning model, learning model, computer program and storage medium

Also Published As

Publication number Publication date
CN106896723A (en) 2017-06-27

Similar Documents

Publication Publication Date Title
CN106896723B (en) A kind of intelligent automatic gear shifting control system of bicycle
CN100504675C (en) Learning apparatus and learning method
CN109567313B (en) Intelligent insole with biological characteristic recognition function
CN109910909A (en) A kind of interactive prediction technique of vehicle track net connection of more vehicle motion states
CN102622418A (en) Prediction device and equipment based on BP (Back Propagation) nerve network
JP7007244B2 (en) Creation device, component control device, creation method, component control method, computer program and learning model
CN102622515A (en) Weather prediction method
DE102020102300A1 (en) CONTROL DEVICE, CREATION METHOD FOR TRAINED MODEL FOR MACHINE LEARNING, TRAINED MODEL FOR MACHINE LEARNING, COMPUTER PROGRAM AND STORAGE MEDIUM
CN106472332A (en) Pet feeding method and system based on dynamic intelligent algorithm
US10967935B2 (en) Creation device, component control device, creation method, component control method and computer program
CN107273969A (en) It is a kind of to parameterize the expansible full articulamentum multilayer interconnection structure of neutral net
CN108973674A (en) Maloperation decision maker
CN107679743A (en) A kind of emergent schedule management method of tracing to the source of environment-protecting intelligent law enforcement based on big data
CN108510126A (en) A kind of Predictive Methods of Road Accidents based on PCA and BP neural network
CN109969329A (en) The control device of human-powered vehicle
CN108021028A (en) A kind of various dimensions cooperative control method converted based on relevant redundancy with strengthening study
CN112158285A (en) Real-time speed change suggestion system for mountain bike
US11521117B2 (en) Control data creation device, component control device, control data creation method, component control method and computer program
CN106447117A (en) Pet feeding method and system based on pet daily data analysis
CN115600741A (en) Urban power distribution network multistage dynamic reconstruction method and system based on deep learning
CN116959665B (en) Motion monitoring method, device, equipment and storage medium
CN105651941B (en) A kind of cigarette sense organ intelligent evaluation system based on decomposition aggregation strategy
CN116214527A (en) Three-body collaborative intelligent decision-making method and system for enhancing man-machine collaborative adaptability
KR102263134B1 (en) System for providing automatic bike fitting service intergrated with realtime spinning contents
Demosthenous et al. Deep reinforcement learning for improving competitive cycling performance

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant