Energy management method of electric bicycle system and electric bicycle system

Abstract

The invention discloses an energy management method of an electric bicycle system and the electric bicycle system.The method comprises the steps that battery parameters, motor parameters and whole vehicle kinetic parameters of an electric bicycle are obtained, and digital mapping processing is carried out to obtain a digital twin system; acquiring operation parameters of the battery, the motor and the whole vehicle, and performing distributed feature extraction processing to obtain feature data; training the bionic optimization model fusing the genetic algorithm and the ant colony algorithm to obtain a scenarized energy optimization strategy; performing hierarchical regulation and control processing on the control system to obtain a hierarchical control instruction sequence; and performing dynamic regulation and control processing on energy distribution to obtain a real-time energy optimization control instruction. According to the technical scheme of the invention, intelligent prediction and dynamic optimization distribution of energy demands are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of electric bicycle system control, and in particular to an energy management method for an electric bicycle system and an electric bicycle system. Background Art

[0002] The electric bicycle system is a new type of transportation tool that integrates core components such as batteries, motors, and controllers. It uses electric energy to assist or replace human riding. With the increasing application of electric bicycles in diverse scenarios such as urban commuting, mountain biking, and express delivery, the importance of energy management has become increasingly prominent. Reasonable energy management is not only related to the length of the driving range, but also directly affects the battery life and the performance of the entire vehicle. Therefore, how to efficiently manage the energy of the electric bicycle system has become an important research topic in this field.

[0003] However, the energy management methods in the prior art mainly adopt control strategies based on fixed rules, which are difficult to dynamically optimize according to the characteristics of different scenarios. For example, in urban commuting scenarios, due to frequent starting, stopping, acceleration and deceleration processes, fixed energy control strategies may cause excessive discharge of batteries or low energy recovery efficiency; in mountain biking scenarios, because climbing conditions have high power output requirements, simple control strategies may cause insufficient power output; and in express delivery scenarios, due to large fluctuations in load weight, traditional energy management methods cannot be adjusted in time according to load changes. The root cause of these problems is that the existing technology lacks the ability to accurately model and intelligently optimize multiple scenarios, and cannot achieve intelligent prediction and dynamic optimization allocation of energy demand. Summary of the invention

[0004] The main purpose of the present invention is to solve the technical problem that the existing electric bicycle system cannot realize the intelligent prediction and dynamic optimization allocation of energy demand.

[0005] A first aspect of the present invention provides an energy management method for an electric bicycle system, the energy management method for an electric bicycle system comprising: Obtaining battery parameters, motor parameters, and vehicle dynamics parameters of the electric bicycle, and performing digital mapping processing on the electric bicycle according to the battery parameters, motor parameters, and vehicle dynamics parameters to obtain a digital twin system including a base layer physical model, a data interaction layer, and a scene recognition layer; According to the digital twin system, the voltage, current and temperature parameters of the battery, the speed, torque and temperature parameters of the motor, and the speed, acceleration and slope parameters of the vehicle are obtained, and distributed feature extraction processing is performed on the obtained parameters to obtain feature data corresponding to the current operating scenario; According to the characteristic data, a bionic optimization model integrating a genetic algorithm and an ant colony algorithm is trained, wherein the bionic optimization model includes a control parameter encoding layer, a scene parameter encoding layer and a user parameter encoding layer, and a scene-based energy optimization strategy is obtained; According to the scenario-based energy optimization strategy, the control system of the electric bicycle is subjected to hierarchical control processing, wherein the control system includes a strategic layer for scenario recognition and strategy formulation, a tactical layer for strategy decomposition, and an execution layer for command execution, to obtain a hierarchical control command sequence; According to the hierarchical control instruction sequence, the energy distribution of the electric bicycle is dynamically regulated to obtain real-time energy optimization control instructions, wherein the energy optimization control instructions include a battery discharge strategy, a motor output strategy and an energy recovery strategy.

[0006] Optionally, the battery parameters, motor parameters and vehicle dynamics parameters of the electric bicycle are obtained, and the electric bicycle is digitally mapped according to the battery parameters, motor parameters and vehicle dynamics parameters to obtain a digital twin system including a base layer physical model, a data interaction layer and a scene recognition layer, including: Conducting sample tests on electric bicycles on preset working condition sections to obtain battery parameters, motor parameters and vehicle dynamics parameters, wherein the battery parameters include nominal voltage, capacity decay rate and internal resistance change rate, the motor parameters include peak torque, rated power and torque coefficient, and the vehicle dynamics parameters include unloaded mass, wind resistance coefficient and rolling resistance coefficient; Establish a battery performance model according to the battery parameters, establish a motor characteristic model according to the motor parameters, and establish a vehicle dynamics model according to the vehicle dynamics parameters to obtain a basic layer physical model; Constructing a data flow relationship matrix including state variables, control variables and environment variables according to the physical model of the basic layer, and performing time series processing on the data flow relationship matrix to obtain a data interaction layer; The data interaction layer is subjected to multi-scenario classification processing, wherein: the energy distribution coefficient of the acceleration and deceleration conditions is calculated for the urban commuting scenario, the power output coefficient of the climbing condition is calculated for the mountain biking scenario, and the energy adjustment coefficient of the load change is calculated for the express delivery scenario, to obtain the scene recognition layer; The basic layer physical model, data interaction layer and scene recognition layer are hierarchically integrated to obtain a digital twin system.

[0007] Optionally, the data interaction layer is subjected to multi-scenario classification processing, wherein: the energy distribution coefficient of acceleration and deceleration conditions is calculated for the urban commuting scenario, the power output coefficient of the climbing condition is calculated for the mountain biking scenario, and the energy adjustment coefficient of the load change is calculated for the express delivery scenario, to obtain the scene recognition layer, including: Constructing an initial scene feature vector according to the state variables, control variables and environmental variables in the data interaction layer, performing time-series sliding processing on the initial scene feature vector to obtain a dynamic scene feature sequence, performing multi-scale decomposition and reconstruction on the dynamic scene feature sequence to obtain scene recognition models for urban commuting scenes, mountain biking scenes and express delivery scenes; The scene recognition model is used to calculate the acceleration and deceleration energy characteristics, including performing piecewise integration on the instantaneous power curve of the starting section to obtain the starting acceleration energy consumption coefficient, performing spectrum analysis on the speed fluctuation curve of the cruising section to obtain the cruising energy consumption coefficient, performing probability density estimation on the kinetic energy loss curve of the braking section to obtain the braking energy recovery coefficient, and obtaining the energy distribution coefficient of the acceleration and deceleration working conditions; The scene recognition model of the mountain biking scene is used to calculate the climbing power characteristics, including performing wavelet decomposition on the slope change curve to obtain a slope compensation coefficient, performing peak detection on the power demand curve to obtain a power response coefficient, performing fractal analysis on the battery discharge curve to obtain a range balance coefficient, and obtaining a power output coefficient of the climbing condition; The scene recognition model of the express delivery scene is used to calculate the load characteristics, including performing modal decomposition on the load change curve to obtain the load compensation coefficient, performing entropy analysis on the power loss curve to obtain the power loss coefficient, performing cluster analysis on the energy consumption distribution curve to obtain the energy consumption balance coefficient, and obtaining the energy regulation coefficient of the load change; A scene probability distribution matrix is ​​constructed according to the energy allocation coefficient of the acceleration and deceleration conditions, the power output coefficient of the climbing condition, and the energy adjustment coefficient of the load change. Fuzzy adaptive calculation is performed on the scene probability distribution matrix to obtain the scene transition weight coefficient. According to the scene transition weight coefficient, dynamic weighted fusion is performed on each scene coefficient to obtain the scene recognition layer.

[0008] Optionally, the voltage, current and temperature parameters of the battery, the speed, torque and temperature parameters of the motor, and the speed, acceleration and slope parameters of the vehicle are obtained according to the digital twin system, and distributed feature extraction processing is performed on the obtained parameters to obtain feature data corresponding to the current operating scenario, including: Distributed nodes for three scenarios, namely, urban commuting, mountain biking, and express delivery, are set according to the digital twin system. At the distributed nodes, the voltage, current, and temperature parameters of the battery are obtained to form a battery parameter vector, the speed, torque, and temperature parameters of the motor are obtained to form a motor parameter vector, and the speed, acceleration, and slope parameters of the vehicle are obtained to form a vehicle parameter vector; Extracting parameter features of distributed nodes in the urban commuting scenario, including calculating the charge-discharge cycle features of the battery parameter vector, calculating the start-stop transition features of the motor parameter vector, and calculating the acceleration-deceleration switching features of the vehicle parameter vector, to obtain a feature subset of the urban commuting scenario; Extracting parameter features of distributed nodes of a mountain biking scene, including calculating a continuous discharge feature of the battery parameter vector, calculating a high power output feature of the motor parameter vector, and calculating a climbing resistance feature of the vehicle parameter vector, to obtain a feature subset of the mountain biking scene; Parameter feature extraction is performed on the distributed nodes of the express delivery scenario, including calculating the variable load discharge feature of the battery parameter vector, calculating the load compensation feature of the motor parameter vector, and calculating the road condition adaptation feature of the vehicle parameter vector, to obtain a feature subset of the express delivery scenario; Distributed data fusion is performed on the feature subsets of the urban commuting scenario, the feature subsets of the mountain biking scenario, and the feature subsets of the express delivery scenario, including time alignment processing of the feature subsets, weight calculation of the feature subsets, and dynamic fusion of the feature subsets, to obtain feature data corresponding to the current operating scenario.

[0009] Optionally, the parameter feature extraction of the distributed nodes of the urban commuting scenario includes calculating the charge and discharge cycle features of the battery parameter vector, calculating the start-stop transition features of the motor parameter vector, and calculating the acceleration and deceleration switching features of the vehicle parameter vector, to obtain a feature subset of the urban commuting scenario, including: The battery parameter vector is segmented according to urban road sections, expressway sections and residential road sections, and the charging and discharging characteristics of parking at a traffic light intersection are calculated for the parameters of the urban road sections, the energy consumption characteristics of continuous cruising are calculated for the parameters of the expressway sections, and the power fluctuation characteristics of low-speed driving are calculated for the parameters of the residential road sections, so as to obtain the charging and discharging cycle characteristics; The motor parameter vector is divided into time series according to the morning peak period, the off-peak period and the evening peak period, the parameters of the morning peak period are used to calculate the power fluctuation characteristics of stop-and-go, the parameters of the off-peak period are used to calculate the energy consumption characteristics of constant speed cruising, and the parameters of the evening peak period are used to calculate the torque change characteristics of the traffic jam condition, so as to obtain the start-stop transition characteristics; The vehicle parameter vector is classified into working conditions according to a straight section, a turning section and a slope section, a cruising speed distribution characteristic is calculated for the parameters of the straight section, a steering deceleration characteristic is calculated for the parameters of the turning section, and a climbing power characteristic is calculated for the parameters of the slope section, so as to obtain an acceleration / deceleration switching characteristic; A road section energy consumption characteristic matrix is ​​constructed according to the charge and discharge cycle characteristics, a time period energy consumption characteristic matrix is ​​constructed according to the start-stop transition characteristics, and an operating condition energy consumption characteristic matrix is ​​constructed according to the acceleration and deceleration switching characteristics. A multi-dimensional fusion operation is performed on the road section energy consumption characteristic matrix, the time period energy consumption characteristic matrix, and the operating condition energy consumption characteristic matrix to obtain a characteristic subset of the urban commuting scenario.

[0010] Optionally, according to the feature data, a bionic optimization model integrating a genetic algorithm and an ant colony algorithm is trained, wherein the bionic optimization model includes a control parameter encoding layer, a scene parameter encoding layer, and a user parameter encoding layer, to obtain a scene-based energy optimization strategy, including: Constructing a chromosome encoding matrix according to the characteristic data, wherein the control parameter encoding layer sets battery power curve parameters, motor efficiency diagram parameters and energy recovery curve parameters, the scene parameter encoding layer sets road condition distribution parameters, environmental factor parameters and load characteristic parameters, and the user parameter encoding layer sets riding mode parameters, control habit parameters and experience preference parameters; Performing pheromone initialization processing on the chromosome coding matrix, including setting a scene pheromone concentration matrix and a path pheromone concentration matrix, wherein the scene pheromone concentration matrix is ​​used to characterize the energy distribution characteristics in different scenes, and the path pheromone concentration matrix is ​​used to characterize the energy consumption distribution characteristics of different road sections, to obtain a mixed pheromone space; A double-layer crossover operation is performed according to the hybrid pheromone space, wherein the first layer performs a genetic crossover on the control parameters and the scene parameters to obtain a working condition optimization sequence, and the second layer performs a pheromone-guided crossover on the working condition optimization sequence and the user parameters to obtain a strategy optimization sequence; The strategy optimization sequence is subjected to scenario-based variation processing, including variation of traffic light intersections in urban commuting scenarios, variation of slope changes in mountain biking scenarios, and variation of load switching in express delivery scenarios, and the pheromone concentration in the mixed pheromone space is updated at the same time to obtain an iterative optimization sequence; The iterative optimization sequence is classified according to the scenario, and the energy loss evaluation index is calculated according to the multi-point start-stop characteristics of the urban commuting scenario, the continuous climbing characteristics of the mountain cycling scenario, and the load fluctuation characteristics of the express delivery scenario. The solution with the minimum energy loss is locally searched according to the ant colony path selection rule to obtain the scenario-based energy optimization strategy.

[0011] Optionally, according to the scenario-based energy optimization strategy, the control system of the electric bicycle is subjected to hierarchical control processing, wherein the control system includes a strategic layer for scenario recognition and strategy formulation, a tactical layer for strategy decomposition, and an execution layer for instruction execution, and a hierarchical control instruction sequence is obtained, including: According to the scenario-based energy optimization strategy, probability estimation is performed on the characteristics of three scenarios, namely, urban commuting, mountain biking, and express delivery, to obtain a scenario probability matrix, and multi-threshold fuzzy judgment is performed on the scenario probability matrix to obtain the multi-scenario fusion result of the strategic layer; Performing strategy matching processing on the multi-scenario fusion results of the strategy layer, including the energy balance strategy in the urban commuting scenario, the power allocation strategy in the mountain biking scenario, and the load compensation strategy in the express delivery scenario, performing forecast period optimization calculation on the matching strategies, and obtaining the strategy optimization sequence of the strategy layer; Decomposing the control parameter matrix according to the strategy optimization sequence of the strategic layer, including decomposing the power limit matrix, efficiency control matrix and recovery control matrix for the energy balance strategy, decomposing the torque control matrix, speed control matrix and temperature control matrix for the power allocation strategy, and decomposing the load control matrix, road condition control matrix and state control matrix for the load compensation strategy, to obtain the matrix decomposition result of the tactical layer; Performing parameter constraint optimization on the matrix decomposition result of the tactical layer, including setting power limit boundary, torque limit boundary and load limit boundary, and performing dynamic boundary correction on each control matrix to obtain the optimized control parameters of the tactical layer; According to the optimized control parameters of the tactical layer, execution layer instructions are generated, including a battery control instruction sequence, a motor control instruction sequence and an energy recovery instruction sequence, and the instruction sequences are subjected to timing synchronization and switching smoothing processing to obtain a hierarchical control instruction sequence.

[0012] Optionally, the energy distribution of the electric bicycle is dynamically regulated according to the hierarchical control instruction sequence to obtain a real-time energy optimization control instruction, wherein the energy optimization control instruction includes a battery discharge strategy, a motor output strategy and an energy recovery strategy, including: The hierarchical control instruction sequence is segmented into time windows, the urban commuting scene weight, the mountain biking scene weight and the express delivery scene weight are calculated in each time window, the scene weights are dynamically updated, and a real-time scene weight matrix is ​​obtained; The battery discharge strategy is calculated according to the real-time scenario weight matrix, including setting a constant current mode for a discharge power less than a threshold value T1, setting a pulse mode for a discharge power between the threshold values ​​T1 and T2, and setting a derating mode for a discharge power greater than the threshold value T2, and dynamically adjusting the thresholds of the three modes in combination with the remaining battery capacity to obtain a battery discharge control sequence; The motor output strategy is calculated according to the real-time scenario weight matrix, including setting a torque priority control mode for a low-speed interval, setting an efficiency priority control mode for a medium-speed interval, and setting a power priority control mode for a high-speed interval, and dynamically adjusting the speed demarcation points of the three modes according to the current load state to obtain a motor output control sequence; Calculating the energy recovery strategy according to the real-time scenario weight matrix, including setting a maximum recovery mode for the braking deceleration segment, setting a medium recovery mode for the downhill gliding segment, and setting a slight recovery mode for the coasting segment, and dynamically limiting the recovery power of the three modes according to the battery charging state to obtain an energy recovery control sequence; A multi-objective optimization matrix is ​​constructed according to the battery discharge control sequence, the motor output control sequence and the energy recovery control sequence, the multi-objective optimization matrix is ​​solved in the rolling time domain, and the solution result is corrected online to obtain a real-time energy optimization control instruction.

[0013] A second aspect of the present invention provides an electric bicycle system, the electric bicycle system comprising: A digital twin modeling module is used to obtain battery parameters, motor parameters and vehicle dynamics parameters of the electric bicycle, and digitally map the electric bicycle according to the battery parameters, motor parameters and vehicle dynamics parameters to obtain a digital twin system including a basic layer physical model, a data interaction layer and a scene recognition layer; A feature data extraction module is used to obtain the voltage, current and temperature parameters of the battery, the speed, torque and temperature parameters of the motor, and the speed, acceleration and slope parameters of the vehicle according to the digital twin system, and perform distributed feature extraction processing on the obtained parameters to obtain feature data corresponding to the current operating scenario; A bionic optimization training module, used to train a bionic optimization model integrating a genetic algorithm and an ant colony algorithm according to the feature data, wherein the bionic optimization model includes a control parameter coding layer, a scene parameter coding layer and a user parameter coding layer, and obtain a scene-based energy optimization strategy; A hierarchical control processing module is used to perform hierarchical control processing on the control system of the electric bicycle according to the scenario-based energy optimization strategy, wherein the control system includes a strategic layer for scenario recognition and strategy formulation, a tactical layer for strategy decomposition, and an execution layer for command execution, to obtain a hierarchical control command sequence; The energy optimization control module is used to dynamically regulate the energy distribution of the electric bicycle according to the hierarchical control instruction sequence to obtain real-time energy optimization control instructions, wherein the energy optimization control instructions include battery discharge strategy, motor output strategy and energy recovery strategy.

[0014] The electric bicycle system energy management method provided by the present invention realizes accurate modeling of the electric bicycle system by constructing a multi-level digital twin system. The digital twin system includes a basic layer physical model, a data interaction layer and a scene recognition layer, wherein the basic layer physical model describes the core characteristics of the battery, motor and vehicle dynamics through mathematical equations, the data interaction layer ensures the data flow and state synchronization between the models, and the scene recognition layer is responsible for identifying the current operating scene. This layered architecture enables the system to fully and accurately reflect the operating status of the electric bicycle under different working conditions.

[0015] On this basis, the present invention adopts a distributed feature extraction method to obtain and process operating parameters from three dimensions: battery, motor and vehicle. This distributed processing method not only improves the efficiency of data processing, but more importantly, it can capture characteristic patterns in different scenarios. For example, in urban commuting scenarios, the system can identify the characteristics of frequent starts and stops; in mountain biking scenarios, it can capture the characteristics of continuous climbing; in express delivery scenarios, it can identify the characteristics of load changes.

[0016] In order to realize intelligent prediction and optimization of energy demand, the present invention innovatively integrates genetic algorithm and ant colony algorithm to construct a multi-level bionic optimization model. The model converts complex energy management problems into optimizable coding sequences by controlling parameter coding layer, scenario parameter coding layer and user parameter coding layer. Among them, the evolutionary mechanism of genetic algorithm ensures the globality of optimization results, while the pheromone mechanism of ant colony algorithm enhances local search capability. The advantages of the two algorithms complement each other, enabling the system to better cope with energy optimization needs in different scenarios.

[0017] At the execution level of the control strategy, the present invention adopts a hierarchical control architecture design. The strategic layer is responsible for scene identification and strategy formulation, and can select appropriate control strategies according to the characteristics of the current scene; the tactical layer is responsible for strategy decomposition, converting high-level strategies into specific control parameters; the execution layer is responsible for command execution, ensuring the smooth implementation of control commands. This hierarchical control architecture ensures both the accuracy of control and the real-time and reliability of the system.

[0018] Finally, the present invention generates real-time energy optimization control instructions including battery discharge strategy, motor output strategy and energy recovery strategy through dynamic control processing. This method based on multi-level optimization and dynamic control can dynamically adjust the energy allocation strategy according to the characteristics of different scenarios, effectively solving the problem that traditional fixed control strategies are difficult to adapt to complex scenarios. Through detailed scenario modeling and intelligent optimization calculations, the system can maximize energy utilization efficiency while ensuring power performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0020] Figure 1 A schematic diagram of an embodiment of an energy management method for an electric bicycle system according to an embodiment of the present invention; Figure 2 FIG. 1 is a schematic diagram of an embodiment of an electric bicycle system in an embodiment of the present invention.

[0021] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0022] 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.

[0023] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back...), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0024] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, "and / or" in the full text includes three solutions. Taking A and / or B as an example, it includes technical solution A, technical solution B, and technical solution that satisfies both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, which must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0025] An embodiment of the present application provides an energy management method for an electric bicycle system. Figure 1A flow chart of an energy management method for an electric bicycle system provided in one embodiment of the present application. In this embodiment, the method includes: See also Figure 1 , obtaining battery parameters, motor parameters and vehicle dynamics parameters of the electric bicycle, performing digital mapping processing on the electric bicycle according to the battery parameters, motor parameters and vehicle dynamics parameters, and obtaining a digital twin system including a basic layer physical model, a data interaction layer and a scene recognition layer; In one embodiment of the present invention, the battery parameters, motor parameters and vehicle dynamics parameters of the electric bicycle are obtained, and the electric bicycle is digitally mapped according to the battery parameters, motor parameters and vehicle dynamics parameters to obtain a digital twin system including a base layer physical model, a data interaction layer and a scene recognition layer, including: Conducting sample tests on electric bicycles on preset working condition sections to obtain battery parameters, motor parameters and vehicle dynamics parameters, wherein the battery parameters include nominal voltage, capacity decay rate and internal resistance change rate, the motor parameters include peak torque, rated power and torque coefficient, and the vehicle dynamics parameters include unloaded mass, wind resistance coefficient and rolling resistance coefficient; Establish a battery performance model according to the battery parameters, establish a motor characteristic model according to the motor parameters, and establish a vehicle dynamics model according to the vehicle dynamics parameters to obtain a basic layer physical model; Constructing a data flow relationship matrix including state variables, control variables and environment variables according to the physical model of the basic layer, and performing time series processing on the data flow relationship matrix to obtain a data interaction layer; The data interaction layer is subjected to multi-scenario classification processing, wherein: the energy distribution coefficient of the acceleration and deceleration conditions is calculated for the urban commuting scenario, the power output coefficient of the climbing condition is calculated for the mountain biking scenario, and the energy adjustment coefficient of the load change is calculated for the express delivery scenario, to obtain the scene recognition layer; The basic layer physical model, data interaction layer and scene recognition layer are hierarchically integrated to obtain a digital twin system.

[0026] Specifically, in the process of implementing the system, sample testing must first be carried out to collect key parameters of the battery, motor and vehicle. The core of this step is to ensure the accuracy and representativeness of the data. First, a section with typical working conditions needs to be selected. The test section can be selected according to specific needs, such as urban commuting routes, mountain biking sections or mixed road conditions. During the test, a series of sensors will be installed on the electric bicycle, such as battery voltage sensors, current sensors, acceleration sensors, GPS positioning systems, and motor speed sensors. These devices are connected to the data acquisition system wirelessly to monitor and record relevant parameters in real time. Specifically, the battery voltage sensor can accurately collect the changes in battery voltage during riding, while the current sensor measures the current fluctuations during battery discharge in real time, and further calculates the power output of the battery. The accelerometer and GPS system help obtain the acceleration, speed and driving trajectory data of the electric bicycle. The motor speed sensor provides the working speed data of the motor, which helps to infer the output power of the motor.

[0027] After collecting these real-time data, the next step is to extract the dynamic parameters of the battery, motor and vehicle. In terms of battery parameters, the main focus is on the nominal voltage, capacity decay rate and internal resistance change rate. The nominal voltage of the battery is directly read by the battery voltage sensor; the capacity decay rate is measured through multiple charge and discharge experiments, and the decay rate is calculated by comparing the difference between the maximum capacity and the initial capacity after each charge; the change in internal resistance is obtained by measuring the voltage change of the battery under different current loads. In the process of obtaining motor parameters, the peak torque and rated power are obtained by measuring the output torque of the motor at different speeds and loads, and the torque coefficient is calculated based on the relationship between the motor speed and torque output. The dynamic parameters of the whole vehicle mainly include unloaded mass, drag coefficient and rolling resistance coefficient. The unloaded mass is obtained by precise weighing; the drag coefficient can be calculated by testing in a wind tunnel or measuring the relationship between wind speed and vehicle speed in actual riding; the rolling resistance coefficient is obtained by experimental measurements under different ground conditions.

[0028] After the basic parameters of the battery, motor and vehicle are obtained, the next step is to establish the corresponding performance model. This process uses physical modeling and data fitting methods. The battery performance model is mainly constructed based on the voltage-current characteristics of the battery. Combined with the battery's discharge curve, internal resistance change and capacity attenuation characteristics, an equivalent circuit model is used to simulate the actual working state of the battery to ensure that the battery performance can be accurately predicted under different discharge conditions. The motor characteristic model is modeled through motor speed and torque data, combined with the motor's efficiency curve, to calculate the power output and torque characteristics of the motor under different working conditions. The vehicle dynamics model relies on data such as vehicle mass, air resistance, rolling resistance, etc., combined with a mechanical model to simulate the vehicle's acceleration, braking, climbing and other dynamic behaviors, and then evaluate the performance of the vehicle under different road conditions.

[0029] After establishing these basic models, the system enters the stage of constructing and timing the data flow relationship matrix. By organizing the state variables, control variables and environmental variables of the battery, motor and vehicle, a multidimensional data matrix is ​​formed. These variables include the voltage and current of the battery; the speed and torque of the motor; the acceleration and speed of the vehicle; and the external environment such as slope and temperature. Through the time series method, the system tracks and updates the changes of these variables in real time. Specifically, the system will first calculate the energy flow between the battery, motor and vehicle in real time to ensure that the working state of each part remains in the optimal range. Over time, the system will update the control strategy in real time according to the remaining power of the battery, the output power of the motor, the driving status of the vehicle, etc., so as to achieve intelligent optimization of the system.

[0030] After that, the system needs to perform multi-scenario classification processing. At this time, first for the urban commuting scenario, the system calculates the energy distribution coefficient of the acceleration and deceleration conditions. For example, on urban roads, due to frequent stopping, starting and acceleration, the energy consumption of electric bicycles is mainly concentrated in the acceleration stage. According to the test data, the system adjusts the energy distribution of the battery and the motor in real time for each acceleration stage to ensure that energy loss is minimized. In the mountain biking scenario, the system calculates the power output coefficient of the climbing condition. For climbing conditions, the motor needs to provide greater output power. Therefore, the system will adjust the power output of the motor in real time according to the changes in the slope and the load conditions to ensure the power demand when climbing. In addition, in the express delivery scenario, due to the unstable load of the vehicle, the system will dynamically adjust the battery discharge and motor output requirements according to different loads, thereby optimizing energy consumption and ensuring endurance during delivery.

[0031] Finally, the system integrates the models at each layer to form a complete digital twin system. The integration of battery performance model, motor characteristic model, vehicle dynamics model, data flow relationship matrix and scene recognition layer enables the system to monitor and adjust the performance of electric bicycles in various usage scenarios in real time. Through this system, users can accurately predict the performance of electric bicycles under different working conditions. The system will also adjust the battery power, motor power and vehicle status based on real-time data feedback, optimize energy use, extend battery life, and enhance the riding experience.

[0032] This digital twin system can solve many problems in existing technologies, especially the energy management problem of electric bicycles in complex environments. Through accurate model construction and real-time data feedback, the system can dynamically adjust the working status of each subsystem, thereby achieving efficient energy utilization under various working conditions and significantly improving the overall performance and endurance of electric bicycles.

[0033] In one embodiment of the present invention, the data interaction layer is subjected to multi-scenario classification processing, wherein: the energy distribution coefficient of the acceleration and deceleration conditions is calculated for the urban commuting scenario, the power output coefficient of the climbing condition is calculated for the mountain biking scenario, and the energy adjustment coefficient of the load change is calculated for the express delivery scenario, to obtain the scene recognition layer, including: Constructing an initial scene feature vector according to the state variables, control variables and environmental variables in the data interaction layer, performing time-series sliding processing on the initial scene feature vector to obtain a dynamic scene feature sequence, performing multi-scale decomposition and reconstruction on the dynamic scene feature sequence to obtain scene recognition models for urban commuting scenes, mountain biking scenes and express delivery scenes; The scene recognition model is used to calculate the acceleration and deceleration energy characteristics, including performing piecewise integration on the instantaneous power curve of the starting section to obtain the starting acceleration energy consumption coefficient, performing spectrum analysis on the speed fluctuation curve of the cruising section to obtain the cruising energy consumption coefficient, performing probability density estimation on the kinetic energy loss curve of the braking section to obtain the braking energy recovery coefficient, and obtaining the energy distribution coefficient of the acceleration and deceleration working conditions; The scene recognition model of the mountain biking scene is used to calculate the climbing power characteristics, including performing wavelet decomposition on the slope change curve to obtain a slope compensation coefficient, performing peak detection on the power demand curve to obtain a power response coefficient, performing fractal analysis on the battery discharge curve to obtain a range balance coefficient, and obtaining a power output coefficient of the climbing condition; The scene recognition model of the express delivery scene is used to calculate the load characteristics, including performing modal decomposition on the load change curve to obtain the load compensation coefficient, performing entropy analysis on the power loss curve to obtain the power loss coefficient, performing cluster analysis on the energy consumption distribution curve to obtain the energy consumption balance coefficient, and obtaining the energy regulation coefficient of the load change; A scene probability distribution matrix is ​​constructed according to the energy allocation coefficient of the acceleration and deceleration conditions, the power output coefficient of the climbing condition, and the energy adjustment coefficient of the load change. Fuzzy adaptive calculation is performed on the scene probability distribution matrix to obtain the scene transition weight coefficient. According to the scene transition weight coefficient, dynamic weighted fusion is performed on each scene coefficient to obtain the scene recognition layer.

[0034] Specifically, first, when constructing the initial scene feature vector, the system generates a complete vector based on the state variables, control variables, and environmental variables collected in real time. These variables are collected in real time through on-board sensors (such as accelerometers, GPS, gyroscopes, temperature and humidity sensors, etc.). For example, vehicle speed and acceleration data are collected through acceleration sensors, while battery voltage and current are obtained through battery management systems. Environmental variables such as road slope are obtained through GPS positioning data and map information. These variables are combined to generate a time series feature vector that fully reflects the current driving status and external environmental conditions of the electric bicycle.

[0035] After obtaining the initial scene feature vector, the next step is to perform time-series sliding processing to obtain a dynamic scene feature sequence. Time-series sliding processing is to collect data within a certain time window and continuously slide the window on the time axis to update the data to form a new scene feature vector. Specifically, assuming that the size of each sliding window is 1 second, the scene feature vector will be recalculated every 1 second. The data in each window includes information such as vehicle speed, acceleration, battery voltage, etc. at the current moment. Over time, these data will reflect the behavior patterns and energy consumption of electric bicycles in different time periods. For example, during urban commuting, the energy consumption of electric bicycles when stopping and starting at traffic lights will be significantly different from the energy consumption in a stable driving state. Through time-series sliding, the system can capture these changes in real time and convert them into new scene features.

[0036] The process of multi-scale decomposition and reconstruction of dynamic scene feature sequences is to capture key features at different time scales and identify fast-changing and slow-changing patterns. Wavelet transform is commonly used for this multi-scale analysis, which decomposes dynamic scene feature sequences into components of different frequencies. By analyzing each component, a more detailed understanding of the behavior of electric bicycles can be obtained. For example, short-term fluctuations may reflect the rider's rapid acceleration or deceleration, while long-term trends reflect the overall riding pattern. In this way, the system can extract specific operating conditions from a relatively rough feature sequence to help identify different riding scenarios.

[0037] After multi-scale analysis, the system will reconstruct the characteristic model of each riding scenario (such as urban commuting, mountain biking, and express delivery). This model can not only accurately identify riding scenarios under specific working conditions, but also provide real-time energy management strategies for each scenario. By reconstructing the characteristics of urban commuting, the system can capture high-frequency operations such as acceleration, deceleration, and parking, while in mountain biking scenarios, it pays more attention to the power output and battery usage of electric bicycles. The express delivery scenario needs to pay special attention to the impact of load changes on energy consumption. Therefore, the reconstructed characteristic model can specifically reflect the key behaviors in each riding scenario, helping the system to dynamically adjust the energy distribution and power output of electric bicycles.

[0038] The next step is to calculate the energy characteristics under different working conditions. Under acceleration and deceleration conditions, the system first integrates the instantaneous power curve of the starting section in sections. This process calculates the power output of the starting stage section by section to obtain the energy consumption of the battery during the starting process. Specifically, the system monitors the battery voltage and current of the electric bicycle at the start, calculates the instantaneous power from these data and integrates them to obtain the energy consumption coefficient of starting acceleration. For the cruising section, the system obtains the cruising energy consumption coefficient by spectral analysis of the speed fluctuation curve. Spectral analysis can reveal energy consumption information in different frequency bands, helping the system to evaluate energy loss during steady riding. For the braking section, the system analyzes the kinetic energy loss curve through the probability density estimation method to obtain the braking energy recovery coefficient, which helps to identify the energy recovery effect of the electric bicycle during deceleration.

[0039] In mountain biking scenarios, it is necessary to calculate the power output characteristics of climbing conditions. The slope change curve is first decomposed by wavelet to obtain the slope compensation coefficient to ensure that the system can adjust the power output of the motor according to the actual road conditions. During the climbing process, the power demand of the electric bicycle will change with the change of the slope, so it is necessary to perform peak detection on the power demand curve to obtain the power response coefficient. This coefficient can help adjust the power output of the electric bicycle to ensure that the rider can smoothly cope with different slopes. In terms of battery discharge, the battery discharge curve is analyzed through fractal analysis to obtain the endurance balance coefficient to ensure that the battery can maintain reasonable power usage during long-term climbing and avoid premature depletion of power.

[0040] For express delivery scenarios, load changes have an important impact on the energy management of electric bicycles. The system first decomposes the load change curve through modal analysis to obtain the load compensation coefficient, which reflects the energy consumption under different load conditions. Subsequently, the power loss curve is analyzed through entropy value to obtain the power loss coefficient to evaluate the additional energy loss caused by load changes. After clustering the energy consumption distribution curve, the system obtains the energy consumption balance coefficient to help adjust the energy distribution and ensure that the electric bicycle can achieve a more balanced energy use during the load change process.

[0041] After obtaining the energy allocation coefficient for acceleration and deceleration conditions, the power output coefficient for climbing conditions, and the energy adjustment coefficient for load changes, the system uses these coefficients to construct a scenario probability distribution matrix. The matrix obtains the scenario transition weight coefficients through fuzzy adaptive calculation. These weight coefficients reflect the transition between different riding scenarios. For example, when switching from urban commuting to mountain riding, the system will dynamically adjust the energy allocation strategy based on the scenario transition weight coefficient. During the scene transition process, the system will perform weighted fusion of the energy characteristics of different scenes according to the weight coefficients to generate the final scene recognition layer, helping the system make smarter energy management and control decisions.

[0042] Through these precise calculations and dynamic adjustments, the digital twin system of the electric bicycle can respond to the needs of different riding scenarios in real time, optimize the energy use of batteries and motors, improve riding efficiency and endurance, and meet performance requirements in a variety of usage scenarios.

[0043] Please continue reading Figure 1 , according to the digital twin system, the voltage, current and temperature parameters of the battery, the speed, torque and temperature parameters of the motor, and the speed, acceleration and slope parameters of the whole vehicle are obtained, and distributed feature extraction processing is performed on the obtained parameters to obtain feature data corresponding to the current operation scenario; In one embodiment of the present invention, the voltage, current and temperature parameters of the battery, the speed, torque and temperature parameters of the motor, and the speed, acceleration and slope parameters of the vehicle are obtained according to the digital twin system, and distributed feature extraction processing is performed on the obtained parameters to obtain feature data corresponding to the current operating scenario, including: Distributed nodes for three scenarios, namely, urban commuting, mountain biking, and express delivery, are set according to the digital twin system. At the distributed nodes, the voltage, current, and temperature parameters of the battery are obtained to form a battery parameter vector, the speed, torque, and temperature parameters of the motor are obtained to form a motor parameter vector, and the speed, acceleration, and slope parameters of the vehicle are obtained to form a vehicle parameter vector; Extracting parameter features of distributed nodes in the urban commuting scenario, including calculating the charge-discharge cycle features of the battery parameter vector, calculating the start-stop transition features of the motor parameter vector, and calculating the acceleration-deceleration switching features of the vehicle parameter vector, to obtain a feature subset of the urban commuting scenario; Extracting parameter features of distributed nodes of a mountain biking scene, including calculating a continuous discharge feature of the battery parameter vector, calculating a high power output feature of the motor parameter vector, and calculating a climbing resistance feature of the vehicle parameter vector, to obtain a feature subset of the mountain biking scene; Parameter feature extraction is performed on the distributed nodes of the express delivery scenario, including calculating the variable load discharge feature of the battery parameter vector, calculating the load compensation feature of the motor parameter vector, and calculating the road condition adaptation feature of the vehicle parameter vector, to obtain a feature subset of the express delivery scenario; Distributed data fusion is performed on the feature subsets of the urban commuting scenario, the feature subsets of the mountain biking scenario, and the feature subsets of the express delivery scenario, including time alignment processing of the feature subsets, weight calculation of the feature subsets, and dynamic fusion of the feature subsets, to obtain feature data corresponding to the current operating scenario.

[0044] Specifically, when implementing the digital twin system, it is first necessary to set up distributed nodes in three scenarios: urban commuting, mountain biking, and express delivery, to ensure that the performance of electric bicycles in different environments can be monitored in real time. These distributed nodes will be deployed in key parts of batteries, motors, and vehicles, and relevant parameters will be collected through a variety of sensors. These sensors will obtain the voltage, current, and temperature of the battery, the speed, torque, and temperature of the motor, as well as the speed, acceleration, and slope parameters of the vehicle in real time. With these data, the system can establish battery parameter vectors, motor parameter vectors, and vehicle parameter vectors, which together describe the current working status of the electric bicycle and provide basic data for subsequent feature extraction and scenario analysis.

[0045] In the distributed nodes of the urban commuting scenario, the parameter vectors of the battery, motor and vehicle will be processed by feature extraction to obtain specific parameter features. For the battery, the charge and discharge cycle characteristics can be calculated by analyzing the trend of the battery voltage, current and temperature. These characteristics can help analyze the power consumption and battery health of the battery during actual use. For example, in the urban commuting scenario, the frequent start and stop of electric bicycles will cause the battery to charge and discharge frequently. Therefore, the extraction of charge and discharge cycle characteristics helps to evaluate the performance and health of the battery in such frequent use scenarios. For the motor, the start-stop transition characteristics can be obtained by analyzing the speed, torque and temperature of the motor. These characteristics describe the performance of the motor when it starts and stops during urban commuting, and reflect the efficiency and stability of the motor under low-speed and high-frequency start-stop conditions. For example, in urban commuting with traffic congestion, electric bicycles frequently accelerate and decelerate. At this time, the start-stop characteristics of the motor will directly affect the response speed and stability of the overall power system. The vehicle parameters include speed, acceleration and slope. By analyzing these data, the acceleration and deceleration switching characteristics can be extracted. These characteristics reflect the ever-changing driving conditions of electric bicycles on urban roads, especially the responsiveness of electric bicycles to acceleration and deceleration when traffic lights change and on undulating roads.

[0046] In mountain biking scenarios, the parameter feature extraction process of batteries, motors and vehicles will be different. Since mountain biking often involves long-term climbing, the continuous discharge characteristics of the battery are particularly important. By tracking the battery voltage and current data for a long time, the characteristics of the battery under continuous high-load discharge can be calculated to help determine the battery's endurance and discharge stability during mountain biking. For example, when climbing for a long time, the battery is heavily loaded and needs to be discharged continuously. During this process, the temperature rise and voltage drop of the battery will affect the riding performance. The extraction of continuous discharge characteristics can help the battery management system make corresponding adjustments. The parameter extraction of the motor focuses on high-power output characteristics, especially in mountainous environments with large slopes, where the motor needs to provide greater power output to maintain a stable riding speed. By analyzing the motor's speed, torque and temperature, the system can evaluate the efficiency and stability of the motor under high-power output to ensure that the electric bicycle can adapt to the complex mountain biking environment. In terms of the whole vehicle, electric bicycles often face greater climbing resistance during mountain biking, especially on steeper sections. The acceleration and slope data of the whole vehicle can be used to calculate the climbing resistance characteristics, helping to analyze the power output and battery consumption required by the electric bicycle under different slope and load conditions.

[0047] For express delivery scenarios, the focus of feature extraction is how to deal with frequent load changes and diverse road conditions. In the battery parameter vector, variable load discharge features can be extracted by analyzing the changing trends of battery voltage, current and temperature data. Since express delivery scenarios often involve the transportation of different items, the load of electric bicycles will change frequently, which will affect the discharge characteristics and endurance of the battery. By analyzing the discharge performance of the battery under different loads, the working efficiency of the battery under high and low load conditions can be evaluated, thereby optimizing battery management and energy distribution. The speed, torque and temperature of the motor can be used to extract load compensation features, which reflect the power requirements and output capacity of the motor when facing different loads. For the whole vehicle, since express delivery usually involves different road conditions, the speed, acceleration and slope data of the whole vehicle can be used to extract road condition adaptation features, which help evaluate the stability and adaptability of electric bicycles under variable road conditions. For example, when encountering steep slopes or uneven roads, electric bicycles need stronger power and higher traction. The extraction of road condition adaptation features can help the system optimize riding strategies and ensure the smooth progress of the delivery process.

[0048] After completing the feature extraction in each scenario, the system enters the data fusion stage. First, the feature subsets of different scenarios collected by distributed nodes will be time-aligned to ensure that the data from different nodes have a consistent time base. For example, the parameter vectors of the battery, motor, and vehicle need to be synchronized at the same time interval so that their performance in different scenarios can be compared and analyzed. Next, the weight calculation of the feature subset will be adjusted according to the characteristics of each scenario. For example, in the urban commuting scenario, due to frequent starts and stops, the charging and discharging and start-stop transition characteristics of the battery and motor will be given a higher weight, while in mountain riding, the continuous discharge characteristics of the battery and the high-power output characteristics of the motor will receive more attention. Finally, the system will obtain feature data that matches the current operating scenario through the dynamic fusion of feature subsets. This process not only ensures the accuracy of scene recognition, but also optimizes the energy management and performance of electric bicycles according to the needs of different scenarios.

[0049] This distributed data fusion and feature extraction process provides full support for the dynamic adjustment of electric bicycles in different scenarios. Through accurate scene recognition and real-time parameter analysis, electric bicycles can intelligently adjust battery power, motor power and energy recovery strategy according to the current riding environment and load conditions, thereby maximizing endurance and riding comfort, ensuring the rider's experience in different scenarios.

[0050] In one embodiment of the present invention, the parameter feature extraction of the distributed nodes of the urban commuting scene includes calculating the charge and discharge cycle features of the battery parameter vector, calculating the start-stop transition features of the motor parameter vector, and calculating the acceleration and deceleration switching features of the vehicle parameter vector, to obtain a feature subset of the urban commuting scene, including: The battery parameter vector is segmented according to urban road sections, expressway sections and residential road sections, and the charging and discharging characteristics of parking at a traffic light intersection are calculated for the parameters of the urban road sections, the energy consumption characteristics of continuous cruising are calculated for the parameters of the expressway sections, and the power fluctuation characteristics of low-speed driving are calculated for the parameters of the residential road sections, so as to obtain the charging and discharging cycle characteristics; The motor parameter vector is divided into time series according to the morning peak period, the off-peak period and the evening peak period, the parameters of the morning peak period are used to calculate the power fluctuation characteristics of stop-and-go, the parameters of the off-peak period are used to calculate the energy consumption characteristics of constant speed cruising, and the parameters of the evening peak period are used to calculate the torque change characteristics of the traffic jam condition, so as to obtain the start-stop transition characteristics; The vehicle parameter vector is classified into working conditions according to a straight section, a turning section and a slope section, a cruising speed distribution characteristic is calculated for the parameters of the straight section, a steering deceleration characteristic is calculated for the parameters of the turning section, and a climbing power characteristic is calculated for the parameters of the slope section, so as to obtain an acceleration / deceleration switching characteristic; A road section energy consumption characteristic matrix is ​​constructed according to the charge and discharge cycle characteristics, a time period energy consumption characteristic matrix is ​​constructed according to the start-stop transition characteristics, and an operating condition energy consumption characteristic matrix is ​​constructed according to the acceleration and deceleration switching characteristics. A multi-dimensional fusion operation is performed on the road section energy consumption characteristic matrix, the time period energy consumption characteristic matrix, and the operating condition energy consumption characteristic matrix to obtain a characteristic subset of the urban commuting scenario.

[0051] Specifically, when calculating the energy consumption characteristics of the urban commuting scenario, first of all, the parameter data of the battery, motor and vehicle need to be appropriately segmented. For the parameter vector of the battery, it is first divided according to different road section types, including urban sections, expressways and residential sections. The characteristics of each section determine the working mode and energy consumption mode of the battery. Specifically, in urban sections, due to the frequent traffic lights, electric bicycles usually need to stop and start frequently, which will cause large fluctuations in the battery charging and discharging process. Therefore, in the battery parameter data of the urban section, the extraction of charging and discharging characteristics is mainly through analyzing the current and voltage changes of the battery when stopping and starting, so as to obtain the charging and discharging characteristics of the battery when stopping at the traffic light intersection. These characteristics help to judge the frequent power fluctuations of the battery in a short period of time and their impact on battery health. In the expressway section, electric bicycles usually maintain a constant speed cruise at a higher speed, which means that the load of the battery is relatively stable and mainly discharges continuously. Therefore, on expressways, the battery's charge and discharge characteristics focus on the energy consumption mode under continuous cruising. By analyzing the changing trends of the battery voltage and current, the energy consumption characteristics under this mode can be extracted. For residential roads, due to the low driving speed and frequent speed changes, the battery power fluctuations are more significant. Therefore, the battery will have more obvious power fluctuations during low-speed driving, forming the power fluctuation characteristics of low-speed driving. These characteristics can help evaluate the efficiency of the battery during low-speed driving, as well as the impact of power fluctuations on overall endurance.

[0052] Similarly, the parameter vector of the motor also needs to be divided into time series according to different time periods, including morning peak period, off-peak period and evening peak period. The morning peak period is usually the most congested time, and the driving state of the electric bicycle is stop-and-go. In this case, the power output of the motor shows a more obvious fluctuation. By analyzing the parameters such as the speed, torque and temperature of the motor, the power fluctuation characteristics of the motor during the stop-and-go process can be calculated. These characteristics help to evaluate the law of power changes of the motor during frequent start-stop processes and provide a reference for the optimization of the motor control system. During the off-peak period, the traffic is relatively smooth, and the electric bicycle can maintain a relatively stable cruising speed. Therefore, the power demand of the motor is relatively stable, which is mainly reflected in the energy consumption characteristics under constant speed cruising. By analyzing the changes in the speed and torque of the motor, the energy consumption pattern of the motor during this period can be accurately calculated, and a reference can be provided for the energy management system. During the evening peak period, traffic is congested again, and electric bicycles usually face a long period of traffic jams. At this time, the torque output of the motor may change significantly, so it is necessary to analyze the torque change of the motor in detail and extract the torque change characteristics, which is crucial for evaluating the motor's workload and energy consumption in traffic jams.

[0053] The parameter vector of the whole vehicle is classified according to the working conditions of different road sections, including straight sections, turning sections and slope sections. On straight sections, electric bicycles usually maintain a relatively stable cruising speed, so it is necessary to calculate the cruising speed distribution characteristics. By analyzing the speed and acceleration data of the whole vehicle on the straight section, a characteristic describing the speed distribution of the electric bicycle on the section can be obtained. These characteristics can reflect the stability of the electric bicycle and its power demand during the straight process. On the turning section, since the electric bicycle needs to slow down when turning, the turning deceleration characteristic becomes an important indicator. By analyzing the changes in the speed and acceleration of the electric bicycle when turning, the deceleration characteristics of the whole vehicle during the turning process can be extracted. These characteristics help to evaluate the dynamic performance and energy consumption pattern of the electric bicycle when turning. Slope sections usually require electric bicycles to overcome a large slope, so the climbing power characteristics become particularly important. By analyzing the relationship between the acceleration and slope of the whole vehicle on the slope section, the power requirements of the electric bicycle at different slopes can be calculated, thereby obtaining the climbing power characteristics, which can help evaluate the power output required by the electric bicycle during the climbing process.

[0054] After these features are extracted, the next step is to fuse the feature matrices. First, a road section energy consumption feature matrix is ​​constructed based on the battery charge and discharge cycle characteristics. By analyzing the battery charge and discharge characteristics on different sections, the energy consumption of each section can be obtained. Secondly, the start-stop transition characteristics of the motor will be used to construct a time period energy consumption feature matrix. By analyzing the energy consumption mode of the motor in different time periods (such as morning peak, flat peak and evening peak), the energy consumption in the period can be obtained. Finally, the acceleration and deceleration switching characteristics of the whole vehicle will be constructed into a working condition energy consumption feature matrix. By calculating the energy consumption characteristics of the whole vehicle under different working conditions (such as straight driving, turning, climbing, etc.), the corresponding energy consumption data can be obtained. After performing multi-dimensional fusion operations on these three feature matrices, the result is the feature subset of the urban commuting scene. This fusion process can not only integrate the energy consumption characteristics of each section, time period and working condition, but also dynamically adjust the energy management strategy of the electric bicycle according to the actual riding situation, thereby improving energy efficiency and endurance.

[0055] Please continue reading Figure 1 , according to the characteristic data, training a bionic optimization model integrating a genetic algorithm and an ant colony algorithm, wherein the bionic optimization model includes a control parameter encoding layer, a scene parameter encoding layer and a user parameter encoding layer, and obtaining a scene-based energy optimization strategy; In one embodiment of the present invention, the bionic optimization model integrating the genetic algorithm and the ant colony algorithm is trained according to the feature data, wherein the bionic optimization model includes a control parameter encoding layer, a scene parameter encoding layer and a user parameter encoding layer, and a scene-based energy optimization strategy is obtained, including: Constructing a chromosome encoding matrix according to the characteristic data, wherein the control parameter encoding layer sets battery power curve parameters, motor efficiency diagram parameters and energy recovery curve parameters, the scene parameter encoding layer sets road condition distribution parameters, environmental factor parameters and load characteristic parameters, and the user parameter encoding layer sets riding mode parameters, control habit parameters and experience preference parameters; Performing pheromone initialization processing on the chromosome coding matrix, including setting a scene pheromone concentration matrix and a path pheromone concentration matrix, wherein the scene pheromone concentration matrix is ​​used to characterize the energy distribution characteristics in different scenes, and the path pheromone concentration matrix is ​​used to characterize the energy consumption distribution characteristics of different road sections, to obtain a mixed pheromone space; A double-layer crossover operation is performed according to the hybrid pheromone space, wherein the first layer performs a genetic crossover on the control parameters and the scene parameters to obtain a working condition optimization sequence, and the second layer performs a pheromone-guided crossover on the working condition optimization sequence and the user parameters to obtain a strategy optimization sequence; The strategy optimization sequence is subjected to scenario-based variation processing, including variation of traffic light intersections in urban commuting scenarios, variation of slope changes in mountain biking scenarios, and variation of load switching in express delivery scenarios, and the pheromone concentration in the mixed pheromone space is updated at the same time to obtain an iterative optimization sequence; The iterative optimization sequence is classified according to the scenario, and the energy loss evaluation index is calculated according to the multi-point start-stop characteristics of the urban commuting scenario, the continuous climbing characteristics of the mountain cycling scenario, and the load fluctuation characteristics of the express delivery scenario. The solution with the minimum energy loss is locally searched according to the ant colony path selection rule to obtain the scenario-based energy optimization strategy.

[0056] Specifically, first, through the construction of feature data, a chromosome encoding matrix can be established. The design of this matrix involves three different encoding layers, corresponding to control parameters, scenario parameters and user parameters. The control parameter encoding layer includes battery power curve parameters, motor efficiency diagram parameters and energy recovery curve parameters. The battery power curve reflects the output capacity of the battery under different loads, while the motor efficiency diagram describes the efficiency performance of the motor under different working conditions, and the energy recovery curve represents the energy recovery characteristics of the motor during braking or deceleration. The battery power curve, motor efficiency diagram and energy recovery curve are the key to optimizing the energy efficiency of electric bicycles, and they affect the energy consumption and recovery efficiency. The scenario parameter encoding layer includes road condition distribution parameters, environmental factor parameters and load characteristic parameters. The road condition distribution parameters describe the resistance and energy consumption characteristics encountered by electric bicycles when driving on different road types (such as urban roads, expressways and mountains), the environmental factor parameters take into account the impact of external environments such as temperature and humidity on the performance of electric bicycles, and the load characteristic parameters reflect the power requirements of electric bicycles under different loads (such as load changes). The user parameter encoding layer includes riding mode parameters, control habit parameters and experience preference parameters. The riding mode parameters can reflect the user's riding habits (such as speed preference, starting acceleration preference, etc.), the control habit parameters describe the user's operation methods during riding (such as braking intensity, turning method, etc.), and the experience preference parameters focus on the user's preferences for the power performance and comfort of electric bicycles. These three levels of coding can comprehensively describe the performance of electric bicycles under different conditions, forming a multi-dimensional chromosome coding matrix.

[0057] After the chromosome encoding matrix is ​​constructed, it is necessary to perform pheromone initialization on it. This process is achieved by setting two pheromone concentration matrices, namely the scene pheromone concentration matrix and the path pheromone concentration matrix. The scene pheromone concentration matrix is ​​used to characterize the energy distribution characteristics in different scenarios. Different riding scenarios (such as urban commuting, mountain biking, and express delivery) have different energy requirements for batteries, motors, and vehicles, so the energy allocation strategies will also be different in different scenarios. For example, urban commuting scenarios usually experience frequent starting and stopping, and the energy consumption and recovery characteristics are significantly different from the climbing conditions in mountain biking scenarios. The path pheromone concentration matrix is ​​used to characterize the energy consumption distribution characteristics of different sections. The energy consumption characteristics on each section are different. For example, frequent traffic light stops on urban roads will result in higher energy consumption, while expressways focus more on energy consumption during continuous cruising. Therefore, the path pheromone concentration matrix reflects the energy consumption patterns of different sections, which can provide a reference for subsequent path selection. By initializing these two pheromone concentration matrices, preliminary guidance can be provided for subsequent algorithm operations.

[0058] Once the pheromone initialization is completed, the next step is to enter the double-layer crossover operation stage. The first layer of crossover operation is to genetically cross the control parameters and scenario parameters. The genetic crossover between the control parameters and the scenario parameters can adapt to different riding scenarios by adjusting the working modes of key components such as batteries and motors. For example, in the urban commuting scenario, the battery needs to respond quickly during frequent stops and starts, while the motor needs to adapt to frequent start-stop processes. The genetic crossover operation merges the control parameters and scenario parameters to form an optimized operating condition sequence that adapts to the scenario requirements. The second layer of crossover operation is to perform a pheromone-guided crossover of the operating condition optimization sequence with the user parameters. User parameters can reflect the user's personalized needs and preferences. Therefore, in this layer, the operating condition optimization sequence can be adjusted through pheromone-guided crossover to better match the user's riding habits and experience needs. For example, some users prefer fast acceleration, while others pay more attention to a smooth riding experience. Through this crossover operation, the strategy can be dynamically adjusted according to the user's needs, thereby achieving personalized optimization.

[0059] After the strategy optimization sequence is generated, the next step is to perform scenario-based mutation processing. The goal of scenario-based mutation processing is to adjust the optimization strategy according to the characteristics of different scenarios. Specifically, in the urban commuting scenario, due to the influence of traffic factors such as traffic lights, the riding mode of electric bicycles often changes, so it is necessary to mutate the traffic light intersections in the urban commuting scenario. This mutation process adjusts the working mode of the battery and motor according to the actual traffic flow and the time interval of the traffic light to improve energy utilization efficiency. In the mountain biking scenario, the slope change is a key factor affecting the riding efficiency, so the slope change needs to be mutated. This process automatically adjusts the power output of the electric bicycle to cope with different climbing requirements by analyzing the slope changes of the road section in real time. In the express delivery scenario, since the change of load will have a significant impact on the energy consumption of the electric bicycle, it is necessary to mutate the load switching to adapt to different delivery needs. All these mutation operations will update the pheromone concentration in real time to ensure that the optimization strategy in each scenario can reflect the latest energy requirements and riding characteristics.

[0060] After these mutation operations, the resulting iterative optimization sequence will be classified according to different scenarios. According to the multi-point start-stop characteristics of urban commuting scenarios, the continuous climbing characteristics of mountain biking scenarios, and the load fluctuation characteristics of express delivery scenarios, the energy loss evaluation indicators are calculated separately. The energy loss evaluation indicators in each scenario reflect the energy consumption under different riding conditions. On this basis, according to the ant colony path selection rules, a local search is performed for the solution with the minimum energy loss to further optimize the strategy. This process can not only find the optimal energy allocation plan in each scenario, but also dynamically adapt to the actual needs of users and changes in the external environment, and finally obtain a scenario-based energy optimization strategy.

[0061] Through this series of operations, the energy optimization strategy finally obtained can fully reflect the energy distribution and consumption characteristics under different riding scenarios, road sections and user needs. This strategy can effectively optimize the energy utilization of batteries, motors and the entire vehicle, improve the endurance and riding efficiency of electric bicycles, and also meet the personalized needs of different users.

[0062] Please continue reading Figure 1 , according to the scenario-based energy optimization strategy, the control system of the electric bicycle is hierarchically regulated, wherein the control system includes a strategic layer for scenario recognition and strategy formulation, a tactical layer for strategy decomposition, and an execution layer for command execution, to obtain a hierarchical control command sequence; In one embodiment of the present invention, the control system of the electric bicycle is hierarchically regulated according to the scenario-based energy optimization strategy, wherein the control system includes a strategic layer for scenario recognition and strategy formulation, a tactical layer for strategy decomposition, and an execution layer for instruction execution, and a hierarchical control instruction sequence is obtained, including: According to the scenario-based energy optimization strategy, probability estimation is performed on the characteristics of three scenarios, namely, urban commuting, mountain biking, and express delivery, to obtain a scenario probability matrix, and multi-threshold fuzzy judgment is performed on the scenario probability matrix to obtain the multi-scenario fusion result of the strategic layer; Performing strategy matching processing on the multi-scenario fusion results of the strategy layer, including the energy balance strategy in the urban commuting scenario, the power allocation strategy in the mountain biking scenario, and the load compensation strategy in the express delivery scenario, performing forecast period optimization calculation on the matching strategies, and obtaining the strategy optimization sequence of the strategy layer; Decomposing the control parameter matrix according to the strategy optimization sequence of the strategic layer, including decomposing the power limit matrix, efficiency control matrix and recovery control matrix for the energy balance strategy, decomposing the torque control matrix, speed control matrix and temperature control matrix for the power allocation strategy, and decomposing the load control matrix, road condition control matrix and state control matrix for the load compensation strategy, to obtain the matrix decomposition result of the tactical layer; Performing parameter constraint optimization on the matrix decomposition result of the tactical layer, including setting power limit boundary, torque limit boundary and load limit boundary, and performing dynamic boundary correction on each control matrix to obtain the optimized control parameters of the tactical layer; According to the optimized control parameters of the tactical layer, execution layer instructions are generated, including a battery control instruction sequence, a motor control instruction sequence and an energy recovery instruction sequence, and the instruction sequences are subjected to timing synchronization and switching smoothing processing to obtain a hierarchical control instruction sequence.

[0063] Specifically, in this embodiment, the implementation process of the scenario-based energy optimization strategy starts with the probability estimation of the scenario features. First, it is necessary to estimate the probability of the three types of scenario features, namely urban commuting, mountain biking and express delivery, and obtain the probability matrix of the scenario. Specifically, the characteristics of each scenario are modeled as a probability distribution, which can reflect the different operating conditions that may occur in different scenarios and their corresponding energy consumption. For example, in the urban commuting scenario, considering factors such as traffic lights, starting acceleration and parking, the system will calculate the impact of traffic flow and road conditions on energy distribution in different time periods. The mountain biking scenario needs to consider the slope, the roughness of the road, and the requirements of the battery and motor energy efficiency under the climbing condition, while the express delivery scenario pays more attention to the energy consumption mode under different loads and transportation tasks. Through the calculation of these characteristics, a comprehensive scenario probability matrix can be obtained, reflecting the probability of each scenario and the corresponding energy demand.

[0064] Next, the scenario probability matrix is ​​processed with multi-threshold fuzzy judgment to obtain the multi-scenario fusion result at the strategic level. Multi-threshold fuzzy judgment is a decision-making method based on fuzzy logic, which is mainly used to make decisions in environments with high uncertainty. Here, the purpose of threshold setting is to make dynamic adjustments under different probabilities of occurrence of different scenarios. For example, in urban commuting scenarios, the complexity of road conditions makes the demand for energy more volatile. Therefore, through multi-threshold judgment, it is possible to dynamically evaluate which working conditions are most important for the current scenario, thereby determining the most appropriate energy allocation method. Similarly, the energy allocation of mountain biking and express delivery scenarios also needs to be adjusted through similar logical judgments. Ultimately, through this fuzzy judgment, the energy requirements and working condition characteristics of all scenarios can be integrated to form a multi-scenario fusion decision result, which provides a basis for subsequent strategy matching and optimization.

[0065] After obtaining the multi-scenario fusion results, the next step is to perform strategy matching. At this point, it is necessary to formulate appropriate energy strategies according to the characteristics of different scenarios. For urban commuting scenarios, the focus is on energy balance strategies. The main goal is to ensure that the battery can maintain a reasonable charge and discharge state under frequent starts and stops, thereby improving overall energy efficiency; for mountain biking scenarios, it is necessary to formulate power allocation strategies, especially in climbing conditions, how to reasonably allocate battery output power to meet climbing needs while avoiding excessive power consumption; and in express delivery scenarios, the focus is on load compensation strategies. When carrying different cargo weights, the energy consumption of electric bicycles is adjusted to make the delivery process both efficient and energy-saving. In actual applications, these strategies will be dynamically adjusted according to the characteristics of different scenarios to adapt to changing riding needs.

[0066] After the strategy matching process is completed, the forecast period optimization calculation phase begins, with the goal of further optimizing the strategy to cope with future changes in riding conditions. The forecast period optimization calculation adjusts the current strategy by analyzing historical data and predicting future environmental changes, thereby achieving more accurate energy management. For example, by predicting traffic flow, weather changes, and user riding habits, the strategy can be better adjusted so that the battery and motor efficiency are always kept at the best state. The main purpose of this step is to improve the enforceability and long-term effectiveness of the strategy by predicting future conditions.

[0067] Next, based on the strategy optimization sequence at the strategic level, the control parameter matrix needs to be decomposed. The control parameters in the strategy optimization sequence will be deconstructed according to different working condition characteristics to obtain different control matrices. For example, for the energy balance strategy, the power limit matrix is ​​first decomposed. This matrix is ​​mainly used to limit the power output of the battery to prevent the shortening of the battery life due to excessive discharge; at the same time, the efficiency control matrix needs to be decomposed to ensure that the motor can maintain the best efficiency when working; in addition, the recovery control matrix is ​​used to optimize the energy recovery process to ensure that the energy can be effectively recovered and fed back to the battery during deceleration or braking. For the power allocation strategy, the torque control matrix, speed control matrix and temperature control matrix are obtained by decomposition, and the output torque, speed and temperature of the motor are adjusted respectively to achieve the best power distribution. The load compensation strategy is decomposed to obtain the load control matrix, road condition control matrix and state control matrix. These matrices can dynamically adjust the energy demand of the vehicle according to different loads and road conditions. Each control matrix accurately corresponds to the energy optimization requirements under different working conditions. Through the decomposition of these matrices, the energy management capability of electric bicycles in complex environments can be further improved.

[0068] The decomposed control matrix also needs to be optimized by parameter constraints. Specifically, boundary conditions need to be set for each control matrix. For example, the power limit boundary can avoid excessive discharge by setting the maximum power output value, and the torque limit boundary ensures the stability of the electric bicycle under various working conditions by setting the torque range of the motor; the load limit boundary takes into account that the power output of the vehicle should change under different loads to avoid affecting the performance of the vehicle due to overload. For these boundaries, dynamic adjustments need to be made according to real-time data and changes in working conditions to ensure that all control matrices can be kept within the optimal range in actual operation, thereby effectively improving energy efficiency.

[0069] Finally, the execution layer instructions are generated based on the optimized control parameters of the tactical layer. The execution layer instructions include battery control instruction sequences, motor control instruction sequences, and energy recovery instruction sequences. These instructions directly act on hardware such as batteries and motors, and achieve the optimal performance of the system in different riding scenarios by accurately controlling the battery's charge and discharge status, the motor's speed and torque, and the energy recovery process. After the execution layer instruction sequence is generated, timing synchronization and switching smoothing are required to ensure that the transition between the various control instructions is smooth and to avoid mutations or discontinuities in the control process. This process can ensure that the vehicle runs smoothly in a complex environment and avoid performance degradation or energy waste caused by improper instruction switching.

[0070] Through this series of steps, the resulting hierarchical control instruction sequence can ensure that the energy efficiency of electric bicycles is maximized in different scenarios and working conditions, while meeting the personalized needs of users and changes in the external environment. The entire process, from the formulation of scenario-based energy optimization strategies to the final generation of execution-layer instructions, constitutes a complete adaptive energy management system that enables electric bicycles to operate efficiently and intelligently in a changing environment.

[0071] Please continue reading Figure 1 According to the hierarchical control instruction sequence, the energy distribution of the electric bicycle is dynamically regulated to obtain real-time energy optimization control instructions, wherein the energy optimization control instructions include battery discharge strategy, motor output strategy and energy recovery strategy.

[0072] In one embodiment of the present invention, the energy distribution of the electric bicycle is dynamically regulated according to the hierarchical control instruction sequence to obtain real-time energy optimization control instructions, wherein the energy optimization control instructions include a battery discharge strategy, a motor output strategy and an energy recovery strategy, including: The hierarchical control instruction sequence is segmented into time windows, the urban commuting scene weight, the mountain biking scene weight and the express delivery scene weight are calculated in each time window, the scene weights are dynamically updated, and a real-time scene weight matrix is ​​obtained; The battery discharge strategy is calculated according to the real-time scenario weight matrix, including setting a constant current mode for a discharge power less than a threshold value T1, setting a pulse mode for a discharge power between the threshold values ​​T1 and T2, and setting a derating mode for a discharge power greater than the threshold value T2, and dynamically adjusting the thresholds of the three modes in combination with the remaining battery capacity to obtain a battery discharge control sequence; The motor output strategy is calculated according to the real-time scenario weight matrix, including setting a torque priority control mode for a low-speed interval, setting an efficiency priority control mode for a medium-speed interval, and setting a power priority control mode for a high-speed interval, and dynamically adjusting the speed demarcation points of the three modes according to the current load state to obtain a motor output control sequence; Calculating the energy recovery strategy according to the real-time scenario weight matrix, including setting a maximum recovery mode for the braking deceleration segment, setting a medium recovery mode for the downhill gliding segment, and setting a slight recovery mode for the coasting segment, and dynamically limiting the recovery power of the three modes according to the battery charging state to obtain an energy recovery control sequence; A multi-objective optimization matrix is ​​constructed according to the battery discharge control sequence, the motor output control sequence and the energy recovery control sequence, the multi-objective optimization matrix is ​​solved in the rolling time domain, and the solution result is corrected online to obtain a real-time energy optimization control instruction.

[0073] Specifically, the weights of urban commuting scenarios, mountain biking scenarios, and express delivery scenarios are calculated in each time window, and these weights are dynamically updated to obtain a real-time scenario weight matrix. Time window segmentation processing is to divide the entire operation cycle into several hourly time windows. The control strategy in each window is adjusted according to the weight of the current scenario to ensure accurate energy management for different scenarios in different time periods. For example, in urban commuting scenarios, the traffic conditions during the morning rush hour and the evening rush hour are different, and the impact of traffic flow and road conditions on energy consumption is also different. Therefore, the real-time scenario weight matrix can adjust the weights of urban commuting scenarios in real time according to traffic changes and optimize the control strategy.

[0074] Next, based on the real-time scenario weight matrix, the battery discharge strategy is calculated. The key to the battery discharge strategy is to ensure that the battery provides appropriate power under different scenario requirements while maximizing the battery's service life and energy recovery efficiency. Specifically, for discharge power less than the threshold T1, the constant current mode is used to stabilize the battery's discharge current and prevent the battery from being damaged by excessive instantaneous power; for discharge power between thresholds T1 and T2, the pulse mode is used to periodically control battery discharge so that the battery's temperature and internal pressure remain within a safe range while improving energy efficiency; and for discharge power greater than the threshold T2, the derating mode is used to limit the battery's output power to avoid excessive load causing battery damage. In the calculation process of the battery discharge strategy, the thresholds of the three modes also need to be dynamically adjusted in combination with the remaining capacity of the battery to ensure that the discharge strategy can adapt to different power requirements under different battery conditions.

[0075] After the battery discharge strategy is determined, the motor output strategy is calculated next. The purpose of the motor output strategy is to optimize the output power of the motor according to different riding scenarios and load conditions to meet riding needs and improve energy efficiency. Specifically, in the low-speed range, the motor adopts a torque priority control mode to ensure that the motor can provide sufficient torque at low speeds, thereby improving the smoothness and responsiveness of starting acceleration; in the medium-speed range, the efficiency priority control mode is adopted to maintain the high-efficiency operation of the motor during cruising, thereby maximizing the battery life; in the high-speed range, the power priority control mode is adopted to ensure that the motor can provide sufficient power to meet the needs of high-speed driving. At the same time, the speed cutoff point in the motor output strategy needs to be dynamically adjusted according to the current load state. For example, when the load is light, the speed cutoff point can be increased to reduce the power consumption of the motor, and when the load is heavy, the speed cutoff point can be lowered to avoid excessive operation of the motor and consume too much energy.

[0076] The goal of the energy recovery strategy is to recover energy under appropriate working conditions and store it in the battery to maximize the energy recovery efficiency of the system. In the braking deceleration stage, the maximum recovery mode is used, so that as much energy as possible can be recovered by the reverse work of the motor during braking; in the downhill gliding stage, the medium recovery mode is used, because the kinetic energy of the vehicle is large at this time, and the recovery mode needs to be properly controlled to avoid energy overload caused by excessive recovery; and in the coasting stage, the slight recovery mode is used. At this time, the vehicle mainly relies on inertial movement, and the recovery power should not be too large to avoid causing discomfort to the driving experience. In addition, the recovery power in the recovery strategy needs to be dynamically limited according to the battery's state of charge. When the battery is close to full, the recovery power will be limited to avoid overcharging and affecting the battery life.

[0077] After the battery discharge control sequence, motor output control sequence, and energy recovery control sequence are calculated, a multi-objective optimization matrix is ​​constructed. In the process of constructing the multi-objective optimization matrix, the interaction between different strategies needs to be considered to ensure that the optimization directions of each control sequence are consistent and can work together. The goal of the multi-objective optimization matrix is ​​to optimize the overall energy efficiency while ensuring the performance of the battery and motor by reasonably adjusting various control parameters. This process requires the use of the rolling time domain solution method to optimize the various parameters in the matrix according to real-time data to ensure that it can adapt to changes in energy demand under different working conditions. In this solution process, the algorithm is continuously iterated to correct the various parameters in the model in real time to ensure that the system always operates in the best state.

[0078] Finally, after rolling time domain solution and online correction, real-time energy optimization control instructions are obtained. These instructions are processed according to different control sequences through timing synchronization and switching smoothing to ensure smooth switching of various control instructions without sudden changes. In this way, control operations such as battery discharge, motor output and energy recovery can be accurately executed in different scenarios to ensure the optimal performance and energy efficiency of electric bicycles. For example, in urban commuting scenarios, the system may frequently adjust the battery discharge mode and motor output strategy to cope with frequent start-stop and traffic conditions; in mountain riding scenarios, the system needs to pay more attention to the power allocation and energy recovery efficiency of the motor to cope with the needs of climbing and steep sections.

[0079] Through these precise strategy calculations and optimization processing, the entire system can respond to changes in the external environment in real time, ensuring that electric bicycles can always provide the best energy management strategy in complex and changeable riding scenarios, extending battery life, improving energy efficiency, and ensuring that the user's riding experience is always at its best.

[0080] The energy management method of the electric bicycle system in the embodiment of the present invention is described above. The electric bicycle system in the embodiment of the present invention is described below. Figure 2 In one embodiment of the present invention, an electric bicycle system includes: The digital twin modeling module 101 is used to obtain the battery parameters, motor parameters and vehicle dynamics parameters of the electric bicycle, and perform digital mapping processing on the electric bicycle according to the battery parameters, motor parameters and vehicle dynamics parameters to obtain a digital twin system including a basic layer physical model, a data interaction layer and a scene recognition layer; The feature data extraction module 102 is used to obtain the voltage, current and temperature parameters of the battery, the speed, torque and temperature parameters of the motor, and the speed, acceleration and slope parameters of the vehicle according to the digital twin system, and perform distributed feature extraction processing on the obtained parameters to obtain feature data corresponding to the current operation scenario; A bionic optimization training module 103 is used to train a bionic optimization model integrating a genetic algorithm and an ant colony algorithm according to the feature data, wherein the bionic optimization model includes a control parameter coding layer, a scene parameter coding layer and a user parameter coding layer, and obtain a scene-based energy optimization strategy; A hierarchical control processing module 104 is used to perform hierarchical control processing on the control system of the electric bicycle according to the scenario-based energy optimization strategy, wherein the control system includes a strategic layer for scenario recognition and strategy formulation, a tactical layer for strategy decomposition, and an execution layer for command execution, to obtain a hierarchical control command sequence; The energy optimization control module 105 is used to dynamically regulate the energy distribution of the electric bicycle according to the hierarchical control instruction sequence to obtain real-time energy optimization control instructions, wherein the energy optimization control instructions include battery discharge strategy, motor output strategy and energy recovery strategy.

[0081] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the inventive concept of the present invention, or directly / indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. An energy management method for an electric bicycle system, characterized in that: include: Obtaining battery parameters, motor parameters, and vehicle dynamics parameters of the electric bicycle, and performing digital mapping processing on the electric bicycle according to the battery parameters, motor parameters, and vehicle dynamics parameters to obtain a digital twin system including a base layer physical model, a data interaction layer, and a scene recognition layer; According to the digital twin system, the voltage, current and temperature parameters of the battery, the speed, torque and temperature parameters of the motor, and the speed, acceleration and slope parameters of the vehicle are obtained, and distributed feature extraction processing is performed on the obtained parameters to obtain feature data corresponding to the current operating scenario; According to the characteristic data, a bionic optimization model integrating a genetic algorithm and an ant colony algorithm is trained, wherein the bionic optimization model includes a control parameter encoding layer, a scene parameter encoding layer and a user parameter encoding layer, and a scene-based energy optimization strategy is obtained; According to the scenario-based energy optimization strategy, the control system of the electric bicycle is subjected to hierarchical control processing, wherein the control system includes a strategic layer for scenario recognition and strategy formulation, a tactical layer for strategy decomposition, and an execution layer for command execution, to obtain a hierarchical control command sequence; According to the hierarchical control instruction sequence, the energy distribution of the electric bicycle is dynamically regulated to obtain real-time energy optimization control instructions, wherein the energy optimization control instructions include a battery discharge strategy, a motor output strategy and an energy recovery strategy.

2. The energy management method of the electric bicycle system according to claim 1, characterized in that: The battery parameters, motor parameters and vehicle dynamics parameters of the electric bicycle are obtained, and the electric bicycle is digitally mapped according to the battery parameters, motor parameters and vehicle dynamics parameters to obtain a digital twin system including a basic layer physical model, a data interaction layer and a scene recognition layer, including: Conducting sample tests on electric bicycles on preset working condition sections to obtain battery parameters, motor parameters and vehicle dynamics parameters, wherein the battery parameters include nominal voltage, capacity decay rate and internal resistance change rate, the motor parameters include peak torque, rated power and torque coefficient, and the vehicle dynamics parameters include unloaded mass, wind resistance coefficient and rolling resistance coefficient; Establish a battery performance model according to the battery parameters, establish a motor characteristic model according to the motor parameters, and establish a vehicle dynamics model according to the vehicle dynamics parameters to obtain a basic layer physical model; Constructing a data flow relationship matrix including state variables, control variables and environment variables according to the physical model of the basic layer, and performing time series processing on the data flow relationship matrix to obtain a data interaction layer; The data interaction layer is subjected to multi-scenario classification processing, wherein: the energy distribution coefficient of the acceleration and deceleration conditions is calculated for the urban commuting scenario, the power output coefficient of the climbing condition is calculated for the mountain biking scenario, and the energy adjustment coefficient of the load change is calculated for the express delivery scenario, to obtain the scene recognition layer; The basic layer physical model, data interaction layer and scene recognition layer are hierarchically integrated to obtain a digital twin system.

3. The energy management method of the electric bicycle system according to claim 2, characterized in that: The data interaction layer is subjected to multi-scenario classification processing, wherein: the energy distribution coefficient of the acceleration and deceleration conditions is calculated for the urban commuting scenario, the power output coefficient of the climbing condition is calculated for the mountain biking scenario, and the energy adjustment coefficient of the load change is calculated for the express delivery scenario, to obtain the scene recognition layer, including: Constructing an initial scene feature vector according to the state variables, control variables and environmental variables in the data interaction layer, performing time-series sliding processing on the initial scene feature vector to obtain a dynamic scene feature sequence, performing multi-scale decomposition and reconstruction on the dynamic scene feature sequence to obtain scene recognition models for urban commuting scenes, mountain biking scenes and express delivery scenes; The scene recognition model is used to calculate the acceleration and deceleration energy characteristics, including performing piecewise integration on the instantaneous power curve of the starting section to obtain the starting acceleration energy consumption coefficient, performing spectrum analysis on the speed fluctuation curve of the cruising section to obtain the cruising energy consumption coefficient, performing probability density estimation on the kinetic energy loss curve of the braking section to obtain the braking energy recovery coefficient, and obtaining the energy distribution coefficient of the acceleration and deceleration working conditions; The scene recognition model of the mountain biking scene is used to calculate the climbing power characteristics, including performing wavelet decomposition on the slope change curve to obtain a slope compensation coefficient, performing peak detection on the power demand curve to obtain a power response coefficient, performing fractal analysis on the battery discharge curve to obtain a range balance coefficient, and obtaining a power output coefficient of the climbing condition; The scene recognition model of the express delivery scene is used to calculate the load characteristics, including performing modal decomposition on the load change curve to obtain the load compensation coefficient, performing entropy analysis on the power loss curve to obtain the power loss coefficient, performing cluster analysis on the energy consumption distribution curve to obtain the energy consumption balance coefficient, and obtaining the energy regulation coefficient of the load change; A scene probability distribution matrix is ​​constructed according to the energy allocation coefficient of the acceleration and deceleration conditions, the power output coefficient of the climbing condition, and the energy adjustment coefficient of the load change. Fuzzy adaptive calculation is performed on the scene probability distribution matrix to obtain the scene transition weight coefficient. According to the scene transition weight coefficient, dynamic weighted fusion is performed on each scene coefficient to obtain the scene recognition layer.

4. The energy management method of the electric bicycle system according to claim 1, characterized in that: The voltage, current and temperature parameters of the battery, the speed, torque and temperature parameters of the motor, and the speed, acceleration and slope parameters of the vehicle are obtained according to the digital twin system, and distributed feature extraction processing is performed on the obtained parameters to obtain feature data corresponding to the current operating scenario, including: Distributed nodes for three scenarios, namely, urban commuting, mountain biking, and express delivery, are set according to the digital twin system. At the distributed nodes, the voltage, current, and temperature parameters of the battery are obtained to form a battery parameter vector, the speed, torque, and temperature parameters of the motor are obtained to form a motor parameter vector, and the speed, acceleration, and slope parameters of the vehicle are obtained to form a vehicle parameter vector; Extracting parameter features of distributed nodes in the urban commuting scenario, including calculating the charge-discharge cycle features of the battery parameter vector, calculating the start-stop transition features of the motor parameter vector, and calculating the acceleration-deceleration switching features of the vehicle parameter vector, to obtain a feature subset of the urban commuting scenario; Extracting parameter features of distributed nodes of a mountain biking scene, including calculating a continuous discharge feature of the battery parameter vector, calculating a high power output feature of the motor parameter vector, and calculating a climbing resistance feature of the vehicle parameter vector, to obtain a feature subset of the mountain biking scene; Parameter feature extraction is performed on the distributed nodes of the express delivery scenario, including calculating the variable load discharge feature of the battery parameter vector, calculating the load compensation feature of the motor parameter vector, and calculating the road condition adaptation feature of the vehicle parameter vector, to obtain a feature subset of the express delivery scenario; Distributed data fusion is performed on the feature subsets of the urban commuting scenario, the feature subsets of the mountain biking scenario, and the feature subsets of the express delivery scenario, including time alignment processing of the feature subsets, weight calculation of the feature subsets, and dynamic fusion of the feature subsets, to obtain feature data corresponding to the current operating scenario.

5. The energy management method of the electric bicycle system according to claim 4, characterized in that: The parameter feature extraction of the distributed nodes of the urban commuting scene includes calculating the charge and discharge cycle features of the battery parameter vector, calculating the start-stop transition features of the motor parameter vector, and calculating the acceleration and deceleration switching features of the vehicle parameter vector, to obtain a feature subset of the urban commuting scene, including: The battery parameter vector is segmented according to urban road sections, expressway sections and residential road sections, and the charging and discharging characteristics of parking at a traffic light intersection are calculated for the parameters of the urban road sections, the energy consumption characteristics of continuous cruising are calculated for the parameters of the expressway sections, and the power fluctuation characteristics of low-speed driving are calculated for the parameters of the residential road sections, so as to obtain the charging and discharging cycle characteristics; The motor parameter vector is divided into time series according to the morning peak period, the off-peak period and the evening peak period, the parameters of the morning peak period are used to calculate the power fluctuation characteristics of stop-and-go, the parameters of the off-peak period are used to calculate the energy consumption characteristics of constant speed cruising, and the parameters of the evening peak period are used to calculate the torque change characteristics of the traffic jam condition, so as to obtain the start-stop transition characteristics; The vehicle parameter vector is classified into working conditions according to a straight section, a turning section and a slope section, a cruising speed distribution characteristic is calculated for the parameters of the straight section, a steering deceleration characteristic is calculated for the parameters of the turning section, and a climbing power characteristic is calculated for the parameters of the slope section, so as to obtain an acceleration / deceleration switching characteristic; A road section energy consumption characteristic matrix is ​​constructed according to the charge and discharge cycle characteristics, a time period energy consumption characteristic matrix is ​​constructed according to the start-stop transition characteristics, and an operating condition energy consumption characteristic matrix is ​​constructed according to the acceleration and deceleration switching characteristics. A multi-dimensional fusion operation is performed on the road section energy consumption characteristic matrix, the time period energy consumption characteristic matrix, and the operating condition energy consumption characteristic matrix to obtain a characteristic subset of the urban commuting scenario.

6. The energy management method of the electric bicycle system according to claim 1, characterized in that: The bionic optimization model integrating the genetic algorithm and the ant colony algorithm is trained according to the feature data, wherein the bionic optimization model includes a control parameter coding layer, a scene parameter coding layer and a user parameter coding layer, and a scene-based energy optimization strategy is obtained, including: Constructing a chromosome encoding matrix according to the characteristic data, wherein the control parameter encoding layer sets battery power curve parameters, motor efficiency diagram parameters and energy recovery curve parameters, the scene parameter encoding layer sets road condition distribution parameters, environmental factor parameters and load characteristic parameters, and the user parameter encoding layer sets riding mode parameters, control habit parameters and experience preference parameters; Performing pheromone initialization processing on the chromosome coding matrix, including setting a scene pheromone concentration matrix and a path pheromone concentration matrix, wherein the scene pheromone concentration matrix is ​​used to characterize the energy distribution characteristics in different scenes, and the path pheromone concentration matrix is ​​used to characterize the energy consumption distribution characteristics of different road sections, to obtain a mixed pheromone space; A double-layer crossover operation is performed according to the hybrid pheromone space, wherein the first layer performs a genetic crossover on the control parameters and the scene parameters to obtain a working condition optimization sequence, and the second layer performs a pheromone-guided crossover on the working condition optimization sequence and the user parameters to obtain a strategy optimization sequence; The strategy optimization sequence is subjected to scenario-based variation processing, including variation of traffic light intersections in urban commuting scenarios, variation of slope changes in mountain biking scenarios, and variation of load switching in express delivery scenarios, and the pheromone concentration in the mixed pheromone space is updated at the same time to obtain an iterative optimization sequence; The iterative optimization sequence is classified according to the scenario, and the energy loss evaluation index is calculated according to the multi-point start-stop characteristics of the urban commuting scenario, the continuous climbing characteristics of the mountain cycling scenario, and the load fluctuation characteristics of the express delivery scenario. The solution with the minimum energy loss is locally searched according to the ant colony path selection rule to obtain the scenario-based energy optimization strategy.

7. The energy management method of the electric bicycle system according to claim 1, characterized in that: According to the scenario-based energy optimization strategy, the control system of the electric bicycle is subjected to hierarchical control processing, wherein the control system includes a strategic layer for scenario recognition and strategy formulation, a tactical layer for strategy decomposition, and an execution layer for command execution, and a hierarchical control command sequence is obtained, including: According to the scenario-based energy optimization strategy, probability estimation is performed on the characteristics of three scenarios, namely, urban commuting, mountain biking, and express delivery, to obtain a scenario probability matrix, and multi-threshold fuzzy judgment is performed on the scenario probability matrix to obtain the multi-scenario fusion result of the strategic layer; Performing strategy matching processing on the multi-scenario fusion results of the strategy layer, including the energy balance strategy in the urban commuting scenario, the power allocation strategy in the mountain biking scenario, and the load compensation strategy in the express delivery scenario, performing forecast period optimization calculation on the matching strategies, and obtaining the strategy optimization sequence of the strategy layer; Decomposing the control parameter matrix according to the strategy optimization sequence of the strategic layer, including decomposing the power limit matrix, efficiency control matrix and recovery control matrix for the energy balance strategy, decomposing the torque control matrix, speed control matrix and temperature control matrix for the power allocation strategy, and decomposing the load control matrix, road condition control matrix and state control matrix for the load compensation strategy, to obtain the matrix decomposition result of the tactical layer; Performing parameter constraint optimization on the matrix decomposition result of the tactical layer, including setting power limit boundary, torque limit boundary and load limit boundary, and performing dynamic boundary correction on each control matrix to obtain the optimized control parameters of the tactical layer; According to the optimized control parameters of the tactical layer, execution layer instructions are generated, including a battery control instruction sequence, a motor control instruction sequence and an energy recovery instruction sequence, and the instruction sequences are subjected to timing synchronization and switching smoothing processing to obtain a hierarchical control instruction sequence.

8. The energy management method of the electric bicycle system according to claim 1, characterized in that: According to the hierarchical control instruction sequence, the energy distribution of the electric bicycle is dynamically regulated to obtain a real-time energy optimization control instruction, wherein the energy optimization control instruction includes a battery discharge strategy, a motor output strategy and an energy recovery strategy, including: The hierarchical control instruction sequence is segmented into time windows, the urban commuting scene weight, the mountain biking scene weight and the express delivery scene weight are calculated in each time window, the scene weights are dynamically updated, and a real-time scene weight matrix is ​​obtained; The battery discharge strategy is calculated according to the real-time scenario weight matrix, including setting a constant current mode for a discharge power less than a threshold value T1, setting a pulse mode for a discharge power between the threshold values ​​T1 and T2, and setting a derating mode for a discharge power greater than the threshold value T2, and dynamically adjusting the thresholds of the three modes in combination with the remaining battery capacity to obtain a battery discharge control sequence; The motor output strategy is calculated according to the real-time scenario weight matrix, including setting a torque priority control mode for a low-speed interval, setting an efficiency priority control mode for a medium-speed interval, and setting a power priority control mode for a high-speed interval, and dynamically adjusting the speed demarcation points of the three modes according to the current load state to obtain a motor output control sequence; Calculating the energy recovery strategy according to the real-time scenario weight matrix, including setting a maximum recovery mode for the braking deceleration segment, setting a medium recovery mode for the downhill gliding segment, and setting a slight recovery mode for the coasting segment, and dynamically limiting the recovery power of the three modes according to the battery charging state to obtain an energy recovery control sequence; A multi-objective optimization matrix is ​​constructed according to the battery discharge control sequence, the motor output control sequence and the energy recovery control sequence, the multi-objective optimization matrix is ​​solved in the rolling time domain, and the solution result is corrected online to obtain a real-time energy optimization control instruction.

9. An electric bicycle system, characterized in that: include: A digital twin modeling module is used to obtain battery parameters, motor parameters and vehicle dynamics parameters of the electric bicycle, and digitally map the electric bicycle according to the battery parameters, motor parameters and vehicle dynamics parameters to obtain a digital twin system including a basic layer physical model, a data interaction layer and a scene recognition layer; A feature data extraction module is used to obtain the voltage, current and temperature parameters of the battery, the speed, torque and temperature parameters of the motor, and the speed, acceleration and slope parameters of the vehicle according to the digital twin system, and perform distributed feature extraction processing on the obtained parameters to obtain feature data corresponding to the current operating scenario; A bionic optimization training module, used to train a bionic optimization model integrating a genetic algorithm and an ant colony algorithm according to the feature data, wherein the bionic optimization model includes a control parameter encoding layer, a scene parameter encoding layer and a user parameter encoding layer, and obtain a scene-based energy optimization strategy; A hierarchical control processing module is used to perform hierarchical control processing on the control system of the electric bicycle according to the scenario-based energy optimization strategy, wherein the control system includes a strategic layer for scenario recognition and strategy formulation, a tactical layer for strategy decomposition, and an execution layer for command execution, to obtain a hierarchical control command sequence; The energy optimization control module is used to dynamically regulate the energy distribution of the electric bicycle according to the hierarchical control instruction sequence to obtain real-time energy optimization control instructions, wherein the energy optimization control instructions include battery discharge strategy, motor output strategy and energy recovery strategy.

Images