Electric bicycle speed regulation method based on torque sensor

Abstract

The invention discloses an electric bicycle speed regulation method based on a torque sensor, and particularly relates to the technical field of electric bicycles, a fuzzy control output coefficient is obtained according to electric signal data of the torque sensor, and fuzzy control complexity is evaluated according to the fuzzy control output coefficient; comparing the fuzzy control complexity obtained by calculation with a preset fuzzy control threshold value, when the fuzzy control complexity exceeds the preset fuzzy control threshold value, constructing a control algorithm by using a neural network to carry out speed regulation control, and carrying out adaptability evaluation according to the nonlinear performance of fuzzy control and neural network control. An adaptability evaluation result is combined with driving route complexity and energy consumption performance to adjust a speed regulation control strategy, fuzzy control and a neural network are combined for application, robustness is effectively enhanced, and high robustness of fuzzy control and self-learning ability of the neural network are utilized to improve speed regulation control accuracy. The optimal control algorithm can be automatically selected according to real-time evaluation to optimize the performance, so that the calculation burden of fuzzy control is reduced.

Description

Technical Field

[0001] The invention relates to the technical field of electric bicycles, and more specifically, to a speed regulating method for electric bicycles based on a torque sensor. Background Art

[0002] The speed regulation method of an electric bicycle based on a torque sensor is to monitor the torque by installing the torque sensor between the motor and the pedals of the electric bicycle, and convert the monitored torque into an electrical signal. The electrical signal data of the sensor is processed by a control algorithm, and the output power of the motor is adjusted according to the torque change of the rider to ensure smooth and efficient riding.

[0003] The existing speed control algorithms based on torque sensors are difficult to cover complex riding environment conditions. Although the control algorithms based on system models or parameter adjustments respond promptly and adjust accurately, they cannot adapt to the nonlinear and uncertain states during riding. The control algorithms based on preset rules adapt well to nonlinear states but lack the control accuracy. In addition, due to the complexity of preset rules, it is difficult to match the speed control to complex riding conditions. Therefore, while improving the nonlinear performance of speed control, solving the problem of the complexity of preset rules has become an urgent problem to be solved.

[0004] In order to solve the above defects, a technical solution is now proposed. Summary of the invention

[0005] The purpose of the present invention is to provide an electric bicycle speed regulation method based on a torque sensor to solve the deficiencies in the background technology.

[0006] In order to achieve the above-mentioned object, the present invention provides the following technical solutions: a speed control method for an electric bicycle based on a torque sensor, obtaining a fuzzy control output coefficient according to electrical signal data of the torque sensor, and evaluating the fuzzy control complexity according to the fuzzy control output coefficient;

[0007] The calculated fuzzy control complexity is compared with the preset fuzzy control threshold. When the fuzzy control complexity exceeds the preset fuzzy control threshold, a control algorithm is constructed using a neural network to perform speed control.

[0008] Conduct adaptability evaluation based on the nonlinear performance of fuzzy control and neural network control;

[0009] The adaptability evaluation results are combined with the complexity of the driving route and the energy consumption performance to adjust the speed control strategy.

[0010] Preferably, the method for collecting the electrical signal data of the torque sensor is:

[0011] The input accumulation coefficient of the electrical signal data of the torque sensor within the cycle time T is obtained, and the input accumulation coefficients of the electrical signal data of the torque sensor within several cycle times T are integrated into a data set, and the data set is marked as DA = {da l}, where l = {1, 2, 3, ..., k}, and k is a positive integer, da l It represents the input accumulation coefficient of the electrical signal data of the lth torque sensor within the cycle time T. The calculation expression of the input accumulation coefficient is: Among them, I v represents the input variable of the fuzzy control algorithm, M f is the number of membership functions of the input variables of the fuzzy control algorithm, where f is the index of the input variable, and f={1, 2, 3, ..., g}, and g is a positive integer.

[0012] Preferably, the method for obtaining the fuzzy control output coefficient according to the electrical signal data of the torque sensor is as follows:

[0013] Calculate the standard deviation of the input accumulation coefficient of the electrical signal data of the torque sensor within the cycle time T. The calculation expression is: Where dav is the average value of the input accumulation coefficient of the electrical signal data of the torque sensor within the cycle time T, and the calculation expression is: Among them, E ac is the fuzzy control output coefficient.

[0014] Preferably, the method for evaluating the complexity of fuzzy control according to the fuzzy control output coefficient is as follows:

[0015] The number of fuzzy rules in the calibration fuzzy rule algorithm is R, the number of output variables is O, and the sum of the number of membership functions of all input variables is M. The expression of M is Then the calculation expression of fuzzy control complexity is In the formula, α1 and α2 are and the weight coefficients of M, and α1 and α2 are both positive numbers.

[0016] Preferably, the logic for comparing the calculated fuzzy control complexity with the preset fuzzy control threshold is:

[0017] Preset fuzzy control threshold F t , the calculated fuzzy control complexity F uc and fuzzy control threshold F t When the calculated fuzzy control complexity F uc Greater than or equal to the fuzzy control threshold F t When , the speed regulation fuzzy control is marked as overloaded and switched to neural network control;

[0018] When the calculated fuzzy control complexity F uc Less than the fuzzy control threshold F t When , mark the speed regulation fuzzy control balance and continue to use fuzzy control.

[0019] Preferably, the method of using a neural network to construct a control algorithm for speed control is:

[0020] Set the input variable to x(t), which includes the moment T p (t) and speed v(t), the output variable is the motor torque T m (t), then the neural network model is T m (t) = f NN (x(t),θ), where f NN is the nonlinear mapping function of the neural network, θ is the weight and bias parameter of the neural network, and the known input variables and output variables are used as data sets to input the neural network model for training. The data set is Da = {(x i , T m，i )}, where i is the index of the input variable and the output variable, and i is a positive integer, then the minimization loss function is Where N is the total number of samples in the data set, x i is the i-th input variable, T m，i is the ith output variable, is the predicted output of the i-th variable, which represents the neural network model using parameter θ for the input variable x i The motor torque at the time of prediction.

[0021] Preferably, the method for adaptively evaluating the nonlinear performance of fuzzy control and neural network control is:

[0022] The calibration control accuracy is P rm , the response time is T r , energy consumption is expressed as C e , then the adaptability evaluation index is In the formula, β1 and β2 are weight coefficients, and both β1 and β2 are positive numbers. Based on the historical data of different riding conditions, the adaptability evaluation indexes of fuzzy control and neural network control are calculated respectively. The adaptability evaluation index based on fuzzy control is E fuzzy , the adaptability evaluation index based on neural network control is E NN .

[0023] Preferably, the logic of adjusting the speed control strategy by combining the adaptability evaluation result with the complexity of the driving route and the energy consumption performance is:

[0024] The complexity of the driving route is calibrated as R c , energy consumption is expressed as E th, then the logic for adjusting the speed control strategy is The calculation expression of driving route complexity is R c =γ1*N r Trs +γ2*S lo , where N r is the number of different road sections included in the driving route, Trs is the traffic flow of each road section, S lo is the slope difference of each road section, γ1 and γ2 are proportional coefficients, and both γ1 and γ2 are positive numbers. The energy consumption performance E th The calculation expression is E th =L ong *P avg *A d , where L ong is the total length of the ride, P avg is the average power consumption of the electric bicycle during riding, A d is the adjustment factor related to wind speed, and A d Is a positive number.

[0025] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0026] This application combines fuzzy control and neural networks. When the system state is simple and the computational complexity is low, fuzzy algorithm control is used, and when the computational complexity is high, it is switched to neural network control. This speed control method effectively enhances the robustness, and utilizes the strong robustness of fuzzy control and the self-learning ability of neural networks to make the speed control perform well in a variety of different environments. It can automatically select the best control algorithm based on real-time evaluation to optimize performance, and use neural networks when the complexity is high, thereby reducing the computational burden of fuzzy control. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0028] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] Example 1: Please refer to Figure 1 As shown, the present invention is a speed regulating method for an electric bicycle based on a torque sensor, and the specific steps include:

[0031] Obtaining a fuzzy control output coefficient according to the electrical signal data of the torque sensor, and evaluating the fuzzy control complexity according to the fuzzy control output coefficient;

[0032] The calculated fuzzy control complexity is compared with the preset fuzzy control threshold. When the fuzzy control complexity exceeds the preset fuzzy control threshold, a control algorithm is constructed using a neural network to perform speed control.

[0033] Conduct adaptability evaluation based on the nonlinear performance of fuzzy control and neural network control;

[0034] The adaptability evaluation results are combined with the complexity of the driving route and the energy consumption performance to adjust the speed control strategy.

[0035] In the speed control process of electric bicycles based on torque sensors, the use of fuzzy algorithm control has a strong adaptability to system uncertainty and noise. Fuzzy algorithm control is based on empirical rules and does not require complex mathematical models. At the same time, it can provide a natural acceleration and deceleration experience and reduce mutations.

[0036] When the preset rules of fuzzy algorithm control rely on the experience and knowledge of experts, the artificially set rules may be incomplete and have certain deviations. In the process of multi-input and multi-output speed control, the number of rules required will increase sharply. In complex operating conditions, it is not as effective as the model-based control method.

[0037] The speed control based on neural network can learn complex nonlinear relationships from data to adapt to different operating conditions. Neural network can capture and predict dynamic behaviors during operation and is suitable for processing high-dimensional and complex input data.

[0038] On the other hand, speed control based on neural networks requires the input of a large amount of data for learning and training, which takes a long time. In addition, due to the black box characteristics of neural networks, the control decision process cannot be explained and it is difficult to understand the internal mechanism. At the same time, the performance of the neural network model is significantly affected by the model architecture and hyperparameter selection.

[0039] In the process of electric bicycle speed control based on torque sensor, nonlinear system behavior usually includes the following aspects:

[0040] Motor characteristics: The relationship between the motor's output torque and input voltage is nonlinear and is affected by factors such as temperature and load;

[0041] Cycling dynamics: changes in rider force, terrain undulations, and wind resistance can cause nonlinear responses;

[0042] Battery Performance: The output characteristics of a battery vary with discharge state and temperature resulting in non-linear behavior.

[0043] In the speed control of electric bicycles based on torque sensors, nonlinear system behavior has the following adverse effects on the speed control of electric bicycles:

[0044] Nonlinear characteristics may cause the control system to react slowly to input changes, thus affecting the smoothness of acceleration and deceleration;

[0045] Nonlinear behavior may cause system oscillation or instability, which may cause the rider to lose control under sudden acceleration or deceleration;

[0046] Nonlinear factors such as load changes and terrain fluctuations may cause deviations in measurement and control signals, affecting the accuracy of speed regulation;

[0047] Ineffective control adjustments for nonlinear behavior may cause the e-bike to consume more electrical energy during driving;

[0048] Nonlinear behavior may cause uneven riding and affect the user's overall riding comfort.

[0049] Obtaining a fuzzy control output coefficient according to the electrical signal data of the torque sensor, and evaluating the fuzzy control complexity according to the fuzzy control output coefficient;

[0050] The method for collecting the electrical signal data of the torque sensor is:

[0051] The input accumulation coefficient of the electrical signal data of the torque sensor within the cycle time T is obtained, and the input accumulation coefficients of the electrical signal data of the torque sensor within several cycle times T are integrated into a data set, and the data set is marked as DA = {da l}, where l = {1, 2, 3, ..., k}, and k is a positive integer, da l It represents the input accumulation coefficient of the electrical signal data of the lth torque sensor within the cycle time T. The calculation expression of the input accumulation coefficient is: Among them, I v represents the input variable of the fuzzy control algorithm, M fis the number of membership functions of the input variables of the fuzzy control algorithm, where f is the index of the input variable, and f={1, 2, 3, ..., g}, g is a positive integer;

[0052] It should be noted that the number of membership functions for each input is the number of fuzzy sets defined for each input variable. The number of membership functions for each input depends on the complexity of the system and the design requirements of the fuzzy control. In a specific embodiment, each fuzzy set maps the input value to a membership degree within a range through a membership function, which is expressed as low, medium, and high different states. For the speed input of the electric bicycle, there are three membership functions: slow, medium, and fast.

[0053] Fuzzy rules are defined based on expert experience and system requirements. Rules are formulated by analyzing the system behavior and design goals of the electric bicycle speed regulation. The number of rules is a combination of the number of input variables and the number of fuzzy sets for each variable.

[0054] The input variables are defined based on the dynamic characteristics of the system and the control objectives. In the speed control of electric bicycles, the input variables include the torque applied by the pedals, the current speed, the rider's needs, etc.

[0055] Output variables are usually related to control objectives. In the speed control of electric bicycles, output variables include the torque output and power assist level of the motor.

[0056] The method for obtaining the fuzzy control output coefficient based on the electrical signal data of the torque sensor is as follows:

[0057] Calculate the standard deviation of the input accumulation coefficient of the electrical signal data of the torque sensor within the cycle time T. The calculation expression is: Where dav is the average value of the input accumulation coefficient of the electrical signal data of the torque sensor within the cycle time T, and the calculation expression is: Among them, E ac is the fuzzy control output coefficient.

[0058] The method for evaluating the complexity of fuzzy control based on the fuzzy control output coefficient is as follows:

[0059] The number of fuzzy rules in the calibration fuzzy rule algorithm is R, the number of output variables is O, and the sum of the number of membership functions of all input variables is M. The expression of M is Then the calculation expression of fuzzy control complexity is In the formula, α1 and α2 are and the weight coefficients of M, and α1 and α2 are both positive numbers.

[0060] It should be noted that α1 and α2 are weight coefficients related to the nonlinear characteristics of the speed control and the hardware performance of the torque sensor, and are set by professional and technical personnel in this field.

[0061] The calculated fuzzy control complexity is compared with the preset fuzzy control threshold. When the fuzzy control complexity exceeds the preset fuzzy control threshold, a control algorithm is constructed using a neural network to perform speed control.

[0062] Preset fuzzy control threshold F t , the calculated fuzzy control complexity F uc and fuzzy control threshold F t When the calculated fuzzy control complexity F uc Greater than or equal to the fuzzy control threshold F t When , the speed regulation fuzzy control is marked as overloaded and switched to neural network control;

[0063] When the calculated fuzzy control complexity F uc Less than the fuzzy control threshold F t When , mark the speed regulation fuzzy control balance and continue to use fuzzy control.

[0064] The method of using neural network to construct control algorithm for speed control is:

[0065] Set the input variable to x(t), which includes the moment T p (t) and speed v(t), the output variable is the motor torque T m (t), then the neural network model is T m (t) = f NN (x(t),θ), where f NN is the nonlinear mapping function of the neural network, θ is the weight and bias parameter of the neural network, and the known input variables and output variables are used as data sets to input the neural network model for training. The data set is Da = {(x i , T m，i )}, where i is the index of the input variable and the output variable, and i is a positive integer, then the minimization loss function is Where N is the total number of samples in the data set, x i is the i-th input variable, T m，i is the ith output variable, is the predicted output of the i-th variable, which represents the neural network model using parameter θ for the input variable x i The motor torque at the time of prediction.

[0066] When the computational complexity of fuzzy control is too high, the significance of using neural network control algorithm to control the speed of electric bicycles is as follows:

[0067] Neural networks can effectively learn and capture complex nonlinear relationships, and adjust control strategies in real time in complex or dynamically changing environments to ensure continuous and stable operation of the system and adapt to the dynamic behavior of electric bicycles under different riding conditions;

[0068] Through a large amount of training data, the neural network can automatically adjust the weights and biases, reducing the reliance on fuzzy control when the computational complexity is high, thereby reducing the computational burden of the system and improving efficiency. There is no need to manually design control rules, which reduces the design complexity.

[0069] Neural networks can process multiple input variables simultaneously and adapt to changing riding conditions, including different speeds, slopes, loads, etc.

[0070] The trained neural network can make control decisions quickly, optimize motor output, improve the speed regulation accuracy of electric bicycles, increase the response speed of the system, enhance riding safety and comfort, and ensure a smooth riding experience.

[0071] Conduct adaptability evaluation based on the nonlinear performance of fuzzy control and neural network control;

[0072] The method for adaptively evaluating the nonlinear performance of fuzzy control and neural network control is:

[0073] The calibration control accuracy is P rm , the response time is T r , energy consumption is expressed as C e , then the adaptability evaluation index is In the formula, β1 and β2 are weight coefficients, and both β1 and β2 are positive numbers. Based on the historical data of different riding conditions, the adaptability evaluation indexes of fuzzy control and neural network control are calculated respectively. The adaptability evaluation index based on fuzzy control is E fuzzy , the adaptability evaluation index based on neural network control is E NN .

[0074] The adaptability evaluation results are combined with the complexity of the driving route and the energy consumption performance to adjust the speed control strategy;

[0075] The complexity of the driving route is calibrated as R c , energy consumption is expressed as E th , then the logic for adjusting the speed control strategy is The calculation expression of driving route complexity is R c =γ1*N r Trs +γ2*S lo , where N r is the number of different road sections included in the driving route, Trs is the traffic flow of each road section, Slo is the slope difference of each road section, γ1 and γ2 are proportional coefficients, and both γ1 and γ2 are positive numbers. The energy consumption performance E th The calculation expression is E th =L ong *P avg *A d , where L ong is the total length of the ride, P avg is the average power consumption of the electric bicycle during riding, A d is the adjustment factor related to wind speed, and A d Is a positive number.

[0076] This application combines fuzzy control and neural networks. When the system state is simple and the computational complexity is low, fuzzy algorithm control is used, and when the computational complexity is high, it is switched to neural network control. This speed control method effectively enhances the robustness, and utilizes the strong robustness of fuzzy control and the self-learning ability of neural networks to make the speed control perform well in a variety of different environments. It can automatically select the best control algorithm based on real-time evaluation to optimize performance, and use neural networks when the complexity is high, thereby reducing the computational burden of fuzzy control.

[0077] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0078] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of computer program goods. The computer program goods include one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0079] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0080] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0081] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0082] If the functions are implemented in the form of software functional units and sold or used as independent goods, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can essentially or in other words, the part that contributes to the prior art or the part of the technical solution can be embodied in the form of software goods, and the computer software goods are stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0083] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A speed control method for an electric bicycle based on a torque sensor, characterized in that: Obtaining a fuzzy control output coefficient according to the electrical signal data of the torque sensor, and evaluating the fuzzy control complexity according to the fuzzy control output coefficient; The calculated fuzzy control complexity is compared with the preset fuzzy control threshold. When the fuzzy control complexity exceeds the preset fuzzy control threshold, a control algorithm is constructed using a neural network to perform speed control. Conduct adaptability evaluation based on the nonlinear performance of fuzzy control and neural network control; The adaptability evaluation results are combined with the complexity of the driving route and the energy consumption performance to adjust the speed control strategy.

2. The method for adjusting the speed of an electric bicycle based on a torque sensor according to claim 1, characterized in that: The method for collecting the electrical signal data of the torque sensor is: The input accumulation coefficient of the electrical signal data of the torque sensor within the cycle time T is obtained, and the input accumulation coefficients of the electrical signal data of the torque sensor within several cycle times T are integrated into a data set, and the data set is marked as FA = {da l }, where l = {1, 2, 3, ..., k}, and k is a positive integer, da l It represents the input accumulation coefficient of the electrical signal data of the lth torque sensor within the cycle time T. The calculation expression of the input accumulation coefficient is: Among them, I v represents the input variable of the fuzzy control algorithm, M f is the number of membership functions of the input variables of the fuzzy control algorithm, where f is the index of the input variable, and f={1,2,3,…,g}, g is a positive integer.

3. The electric bicycle speed regulation method based on torque sensor according to claim 2 is characterized in that: The method for obtaining the fuzzy control output coefficient based on the electrical signal data of the torque sensor is as follows: Calculate the standard deviation of the input accumulation coefficient of the electrical signal data of the torque sensor within the cycle time T. The calculation expression is: Where dav is the average value of the input accumulation coefficient of the electrical signal data of the torque sensor within the cycle time T, and the calculation expression is: Among them, E ac is the fuzzy control output coefficient.

4. The method for adjusting the speed of an electric bicycle based on a torque sensor according to claim 3, characterized in that: The method for evaluating the complexity of fuzzy control based on the fuzzy control output coefficient is as follows: The number of fuzzy rules in the calibration fuzzy rule algorithm is R, the number of output variables is O, and the sum of the number of membership functions of all input variables is M. The expression of M is Then the calculation expression of fuzzy control complexity is In the formula, α1 and α2 are and the weight coefficients of M, and α1 and α2 are both positive numbers.

5. The method for adjusting the speed of an electric bicycle based on a torque sensor according to claim 4, characterized in that: The logic of comparing the calculated fuzzy control complexity with the preset fuzzy control threshold is: Preset fuzzy control threshold F t , the calculated fuzzy control complexity F uc and fuzzy control threshold F t When the calculated fuzzy control complexity F uc Greater than or equal to the fuzzy control threshold F t When , the speed regulation fuzzy control is marked as overloaded and switched to neural network control; When the calculated fuzzy control complexity F uc Less than the fuzzy control threshold F t When , mark the speed regulation fuzzy control balance and continue to use fuzzy control.

6. The electric bicycle speed regulation method based on torque sensor according to claim 1, characterized in that: The method of using neural network to construct control algorithm for speed control is: Set the input variable to x(t), which includes the moment T p (t) and speed v(t), the output variable is the motor torque T m (t), then the neural network model is T m (t) = f NN (x(t),θ), where f NN is the nonlinear mapping function of the neural network, θ is the weight and bias parameter of the neural network, and the known input variables and output variables are input into the neural network model as a data set for training. The data set is Da = P(x i ,T m,i )}, where i is the index of the input variable and the output variable, and i is a positive integer, then the minimization loss function is Where N is the total number of samples in the data set, x i is the i-th input variable, T m,i is the ith output variable, is the predicted output of the i-th variable, which represents the neural network model using parameter θ for the input variable x i The motor torque at the time of prediction.

7. The method for adjusting the speed of an electric bicycle based on a torque sensor according to claim 6, characterized in that: The method for adaptively evaluating the nonlinear performance of fuzzy control and neural network control is: The calibration control accuracy is P rm , the response time is T r , energy consumption is expressed as C e , then the adaptability evaluation index is In the formula, β1 and β2 are weight coefficients, and both β1 and β2 are positive numbers. Based on the historical data of different riding conditions, the adaptability evaluation indexes of fuzzy control and neural network control are calculated respectively. The adaptability evaluation index based on fuzzy control is E fuzzy , the adaptability evaluation index based on neural network control is E NN .

8. The electric bicycle speed regulation method based on torque sensor according to claim 7, characterized in that: The logic of adjusting the speed control strategy by combining the adaptability evaluation results with the complexity of the driving route and the energy consumption performance is as follows; The complexity of the driving route is calibrated as R c , energy consumption is expressed as E th , then the logic for adjusting the speed control strategy is The calculation expression of driving route complexity is R c =γ1*N r Trs +γ2*S lo , where N r is the number of different road sections included in the driving route, Trs is the traffic flow of each road section, S lo is the slope difference of each road section, γ1 and γ2 are proportional coefficients, and both γ1 and γ2 are positive numbers. The energy consumption performance E th The calculation expression is E th =L ong *P avg *A d , where L ong is the total length of the ride, P avg is the average power consumption of the electric bicycle during riding, A d is the adjustment factor related to wind speed, and A d Is a positive number.

Images