An electric bicycle endurance evaluation system based on multi-source data fusion
Through the combination of multi-source data fusion and deep learning models, a circular list of historical data and evaluation modules are built, real-time evaluation of the battery life of the electric bicycle is realized, and the problem of incomplete update of evaluation results in the existing technology is solved, improving the accuracy of evaluation Sex and real-time.
Patent Information
- Application Number
- CN202410297399.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-15
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-03-15
AI Technical Summary
The existing technology cannot update and dynamically adjust the battery life assessment of electric bicycles in real time, resulting in the disconnection of the evaluation results from the actual situation and lack of real-time application value.
Design an electric bicycle endurance evaluation system based on multi-source data fusion, including data acquisition module, data storage module, model building module, multi-source data prediction module, average power consumption evaluation module and endurance evaluation module, and build historical data Ring linked list and deep learning model, real-time update of range evaluation values.
It improves the accuracy and real-time performance of battery life assessment, can promptly reflect changes in the operating status and energy consumption of electric bicycles, adapt to changes in factors, and ensures the accuracy and reliability of the evaluation results.
Smart Images

Figure CN117972644B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of multi-source data fusion, and in particular to an electric bicycle endurance evaluation system based on multi-source data fusion. Background Art
[0002] Prior art CN113406522A "A method and system for predicting and evaluating the life of a lithium battery system in an electric vehicle" randomly simulates the specific working conditions of the whole vehicle based on system parameters and working condition parameters, combined with a set working condition distribution probability model, and statistically calculates the characteristic distribution of each working condition. The battery cell life data and the statistical working condition distribution are input into the battery cell life model. The calendar life curve and cycle life curve of the system under this working condition can be obtained through a weighted algorithm. The vehicle parameters and the statistical working condition distribution are input into the cruising range correction model, and the cruising range is corrected in three aspects: energy efficiency, energy consumption of accessories, and load influence. Finally, the system life iteration model integrates all parameters and data, predicts the life of the system and obtains a confidence interval. The present invention combines the single cell life data in the prediction of the whole vehicle life, considers the impact of the whole vehicle operating conditions on the system life, corrects the cruising range, and realizes the prediction and evaluation of the service life of the lithium battery system.
[0003] Prior art CN111376792A "A method for estimating the cruising range of pure electric vehicles" calculates the total battery consumption and total driving distance in real time after the electric vehicle is powered on, and samples and stores the updated sample values at certain time intervals; uses the total consumption and total driving distance to calculate the average driving energy consumption according to certain rules; uses the battery charge and discharge attenuation characteristics and the uneven energy density of the battery at different charge and discharge depths to construct the battery attenuation coefficient and divide the battery energy interval, and obtains the proportion of each interval energy relative to the rated power of the battery and constructs a function of the battery energy density, and estimates the remaining battery energy in combination with the current SOC value of the battery. The cruising range is estimated from the remaining energy and average energy consumption of the battery, and filtered. In addition, the energy proportion of each interval is updated according to the operation of the vehicle to make the calculation of the cruising range more accurate.
[0004] Multi-source data fusion can integrate information from different sensors, devices and data sources, and integrate diverse data information to improve the accuracy and precision of endurance assessment. By integrating various data, we can have a more comprehensive understanding of factors such as the operating status, driving environment and energy consumption of electric bicycles, reduce the errors and limitations of a single data source, and improve the reliability of prediction and evaluation. The above technology predicts and estimates battery life through preset established parameters, but the above existing technology does not have the ability to update and dynamically adjust in real time. How to continuously collect, process and analyze information from various data sources to more promptly reflect changes in the endurance of electric bicycles and changes in influencing factors, so that the evaluation results are closer to the actual situation and have real-time application value is a problem that we urgently need to solve. We now provide an electric bicycle endurance assessment system based on multi-source data fusion. Summary of the invention
[0005] In order to solve the above technical problems, the object of the present invention is to provide an electric bicycle endurance evaluation system based on multi-source data fusion, including a monitoring center, the monitoring center is communicatively connected with a data acquisition module, a data storage module, a model building module, a multi-source data prediction module, an average power consumption evaluation module and an endurance evaluation module;
[0006] The data acquisition module is used to obtain data information collected by the edge terminal deployed on the target electric bicycle, mark the collection time, and set the collection cycle;
[0007] The data storage module is used to store and count historical data information of the target electric bicycle within several historical collection cycles, and to construct a historical data circular linked list according to the end timestamps of several historical collection cycles;
[0008] The model building module is used to build a route prediction model, an environmental data prediction model and a user characteristic behavior prediction model according to the historical data information included in the historical data circular linked list;
[0009] The multi-source data prediction module is used to obtain multi-source prediction data of the current collection cycle according to the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model;
[0010] The average power consumption evaluation module constructs an average power consumption evaluation model according to the pre-collected average power consumption simulation results;
[0011] The endurance assessment module obtains the endurance assessment value of the electric bicycle based on the multi-source prediction data of the current collection cycle, the average power consumption assessment model and the battery data collected at the current moment, and performs real-time prediction error matching between the multi-source prediction data and the data information collected in real time, and updates the endurance assessment value in real time based on the prediction error matching.
[0012] Furthermore, the data information collected by the edge terminal deployed on the target electric bicycle includes driving route, environmental data, behavioral characteristic data and battery data. The behavioral characteristic data includes the size of a single acceleration, the frequency of acceleration and the driving speed. The battery data includes the battery voltage, the battery discharge rate and the remaining battery power.
[0013] Furthermore, the data storage module stores and counts historical data information of the target electric bicycle within several historical collection cycles, and constructs a historical data circular linked list according to the end timestamps of the several historical collection cycles, including:
[0014] The geographical scope covered by the target electric bicycle's driving route is obtained according to the historical driving route, and all route information within the geographical scope is obtained by GIS means, and the forks on all routes are marked, and the routes between every two forks are marked as sections, and all sections included in the historical driving route are obtained;
[0015] According to the historical environmental data and historical behavior characteristic data of the target electric bicycle, the historical environmental data of each road section and the single acceleration average value, the acceleration frequency average value and the driving speed average value corresponding to the historical environmental data of each road section are obtained;
[0016] Set a historical data circular linked list and a linked list length k of the historical data circular linked list, obtain the end timestamps of several historical collection cycles in the data storage module, obtain the time span between the end timestamps of several historical collection cycles and the current moment, sort each historical collection cycle in reverse order according to the time span, and insert the historical driving routes corresponding to the k historical collection cycles, the historical environmental data of each road section, and the single acceleration average value, the acceleration frequency average value, and the driving speed average value corresponding to the historical environmental data of each road section into the historical data circular linked list in sequence according to the sorting results;
[0017] The smaller the time span, the newer the historical data information within the historical collection period. The smaller the time span, the higher the ranking. The smaller the time span, the better the historical data information. The historical data information within the historical collection period is sorted first. The historical data information within the historical collection period is sorted from front to back according to the time span in the circular linked list. The latest historical collection period will be inserted into the head of the table. When the historical collection period needs to be eliminated, you only need to delete the historical collection period at the end of the linked list, because the historical collection period at the end of the linked list represents the oldest historical collection period.
[0018] Furthermore, the process of constructing the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model according to the historical data information included in the historical data circular linked list by the model construction module includes:
[0019] Based on deep learning, a route prediction model, an environmental data prediction model and a user characteristic behavior prediction model are constructed, and the historical driving routes within k historical collection cycles, the historical environmental data of each road section, and the single acceleration average value, the acceleration frequency average value and the driving speed average value corresponding to the historical environmental data of each road section in the historical data circular linked list are obtained. The historical driving routes within k historical collection cycles, the historical environmental data of each road section, and the single acceleration average value, the acceleration frequency average value and the driving speed average value corresponding to the historical environmental data of each road section are used as the training set and test set of the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model respectively. The route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are trained in real time until the loss function corresponding to the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model is trained stably, and the model parameters are saved. Subsequently, the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are tested through the test set until they meet the preset requirements, and the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are output.
[0020] Furthermore, the process of the multi-source data prediction module acquiring the multi-source prediction data of the current collection cycle according to the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model includes:
[0021] Acquire the driving route collected in the current collection period, input the driving route into the route prediction model, and obtain the predicted driving route of the target electric bicycle in the current collection period;
[0022] Input the predicted driving route of the target electric bicycle in the current collection period into the environmental data prediction model, and obtain the environmental prediction data of each section in the predicted driving route in the current collection period according to the environmental data prediction model;
[0023] The environmental prediction data of each section in the predicted driving route of the current collection period is input into the user characteristic behavior prediction model, and the single acceleration average value, acceleration frequency average value and driving speed average value of each section in the predicted driving route of the current collection period are obtained according to the user characteristic behavior prediction model.
[0024] Furthermore, the process of the average power consumption evaluation module constructing an average power consumption evaluation model according to the pre-collected average power consumption simulation results includes:
[0025] Acquire the structural parameters, configuration parameters and battery parameters of the target electric bicycle in advance, use Simulink technology to build a dynamic three-dimensional model of the target electric bicycle according to the structural parameters, configuration parameters and battery parameters of the electric bicycle, use a simulator to generate a number of behavioral characteristic data consisting of different single accelerations, acceleration frequencies and driving speeds, as well as different environmental data, use Simulink technology to simulate the dynamic three-dimensional model in different behavioral characteristic data and different environmental data, and generate average power consumption simulation results of the dynamic three-dimensional model in different behavioral characteristic data and different environmental data;
[0026] An average power consumption evaluation model is built based on deep learning, and the average power consumption simulation results of the dynamic three-dimensional model in different behavioral feature data and different environmental data are used as the test set and training set;
[0027] The training set is input into the average power consumption evaluation model for training until the loss function training is stable, and the model parameters are saved. The average power consumption evaluation model is tested by the test set until it meets the preset requirements, and the average power consumption evaluation model is output.
[0028] Furthermore, the structural parameters of the electric bicycle include the various components of the electric bicycle, such as the engine, transmission system, wheels, suspension system, etc., and the assembly relationship between the various components of the electric bicycle;
[0029] The configuration parameters of the electric bicycle include the mass of the electric bicycle, the height of the center of gravity, the friction coefficient of the tire, etc.;
[0030] The battery parameters of an electric bicycle include the battery discharge rate of the electric bicycle, battery capacity, battery type (different types of batteries, such as lithium-ion batteries, lead-acid batteries, etc., have different energy densities and performance characteristics, which directly affect the endurance), body weight and battery voltage;
[0031] Environmental data collected by edge terminals include: temperature, wind speed, humidity, and road conditions obtained through GIS;
[0032] It should be further explained that the deep learning model generally uses a loss function to calculate the loss value between the ground truth data and the predicted probability to quantify the gap between the two. The smaller the loss value, the more accurate the classification. The present invention uses a logistic regression loss function to calculate the loss values of the route prediction model, the environmental data prediction model, the user feature behavior prediction model, and the average power consumption evaluation model, respectively. The formula is:
[0033]
[0034] h(x)=wx+b;
[0035] Among them, (x, y) represents samples, x represents input features, y represents output labels, w represents weight parameters, b represents bias parameters, N represents the number of samples, and D represents a data set of multiple samples.
[0036] Furthermore, the process of obtaining the range evaluation value of the electric bicycle by the endurance evaluation module according to the multi-source prediction data of the current collection period, the average power consumption evaluation model and the battery data collected at the current moment includes:
[0037] Input the average single acceleration value, the average acceleration frequency value and the average driving speed value of each section in the predicted driving route in the current collection period and the environmental prediction data of each section in the predicted driving route into the average power consumption evaluation model, and output the average power consumption evaluation value of the target electric bicycle in each section in the predicted driving route according to the average power consumption evaluation model;
[0038] The battery voltage, battery discharge rate and battery remaining power of the electric bicycle at the current moment are obtained, and the range evaluation value of the electric bicycle is obtained according to the average power consumption evaluation value, battery voltage, battery discharge rate and battery remaining power of each road section.
[0039] Among them, the mileage evaluation value Dist of each road section i =(battery capacity C×battery voltage V×discharge rate θ) / (average power consumption of each section W i );
[0040] Estimated range of electric bicycles
[0041] Among them, Dist represents the estimated value of the cruising range of the electric bicycle, C represents the remaining battery power, V represents the battery voltage, θ represents the battery discharge rate, δ represents the conversion coefficient, and W i represents the average power consumption evaluation value of the ith road segment, and N represents the total number of road segments included in the predicted driving route;
[0042] Furthermore, the process in which the cruising range evaluation module matches the multi-source prediction data with the real-time collected data information in real-time prediction error and updates the cruising range evaluation value in real time according to the prediction error matching includes:
[0043] Perform real-time prediction error matching on the predicted driving route of the current collection period, the environmental prediction data of each section in the predicted driving route, the average single acceleration, the average acceleration frequency and the average driving speed of each section in the predicted driving route, and the driving route, environmental data, behavioral feature data and battery data of the current collection period;
[0044] If the driving route of the current collection cycle is inconsistent with the predicted driving route, or the environmental data of the section in the driving route of the current collection cycle is inconsistent with the environmental prediction data of the same section in the predicted driving route, or the single acceleration average value, acceleration frequency average value and driving speed average value of the section in the driving route are inconsistent with the single acceleration average value, acceleration frequency average value and driving speed average value of the same section in the predicted driving route, then the collected data information of the current collection cycle is inserted into the historical data circular linked list, and the historical data information of the longest span historical collection cycle in the historical data circular linked list is overwritten by the data information of the current collection cycle, and then the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are retrained according to the historical data information in the historical data circular linked list, and new multi-source prediction data is obtained according to the re-trained route prediction model, environmental data prediction model and user characteristic behavior prediction model, and a new cruising range evaluation value is obtained according to the new multi-source prediction data, the average power consumption evaluation model and the battery data at the current moment, and the original cruising range evaluation value is updated to the new cruising range evaluation value.
[0045] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention constructs a historical data circular linked list according to the end timestamps of several historical collection cycles, constructs a route prediction model, an environmental data prediction model and a user characteristic behavior prediction model according to the historical data information included in the historical data circular linked list, obtains multi-source prediction data of the current collection cycle according to the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model, obtains the range evaluation value of the electric bicycle according to the multi-source prediction data of the current collection cycle, the average power consumption evaluation model and the battery data collected at the current moment, and performs real-time prediction error matching between the multi-source prediction data and the real-time collected data information, and when the real-time collected data such as the driving route, weather conditions, user characteristic behavior habits, etc. undergoes significant changes, when the multi-source prediction data deviates greatly from the real-time collected data information, the current collection cycle data is used as new training data to ensure that the model can adapt to changes when processing new inputs and that the model prediction results will not continue to deviate from the actual situation, thereby improving the accuracy and real-time nature of the prediction, and at the same time helping riders or managers to make accurate decisions and adjustments during driving, thereby improving driving efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] Figure 1 This is a schematic diagram of an electric bicycle endurance assessment system based on multi-source data fusion according to an embodiment of the present application. DETAILED DESCRIPTION
[0047] The following is a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.
[0048] like Figure 1 As shown, an electric bicycle endurance assessment system based on multi-source data fusion includes a cloud server, and the cloud server is communicatively connected with a data acquisition module, a data storage module, a model building module, a multi-source data prediction module, an average power consumption assessment module and an endurance assessment module;
[0049] The data acquisition module is used to obtain data information collected by the edge terminal deployed on the target electric bicycle, mark the collection time, and set the collection cycle;
[0050] The data storage module is used to store and count historical data information of the target electric bicycle within several historical collection cycles, and to construct a historical data circular linked list according to the end timestamps of several historical collection cycles;
[0051] The model building module is used to build a route prediction model, an environmental data prediction model and a user characteristic behavior prediction model according to the historical data information included in the historical data circular linked list;
[0052] The multi-source data prediction module is used to obtain multi-source prediction data of the current collection cycle according to the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model;
[0053] The average power consumption evaluation module constructs an average power consumption evaluation model according to the pre-collected average power consumption simulation results;
[0054] The endurance assessment module obtains the endurance assessment value of the electric bicycle based on the multi-source prediction data of the current collection cycle, the average power consumption assessment model and the battery data collected at the current moment, and performs real-time prediction error matching between the multi-source prediction data and the data information collected in real time, and updates the endurance assessment value in real time based on the prediction error matching.
[0055] It should be further explained that, in the specific implementation process, the data information collected by the edge terminal deployed on the target electric bicycle includes driving route, environmental data, behavioral characteristic data and battery data. The behavioral characteristic data includes the size of a single acceleration, the frequency of acceleration and the driving speed. The battery data includes the battery voltage, battery discharge rate and the remaining battery power.
[0056] It should be further explained that, in the specific implementation process, the data storage module stores and counts the historical data information of the target electric bicycle within several historical collection cycles, and constructs a historical data circular linked list according to the end timestamps of several historical collection cycles, including:
[0057] The geographical scope covered by the target electric bicycle's driving route is obtained according to the historical driving route, and all route information within the geographical scope is obtained by GIS means, and the forks on all routes are marked, and the routes between every two forks are marked as sections, and all sections included in the historical driving route are obtained;
[0058] According to the historical environmental data and historical behavior characteristic data of the target electric bicycle, the historical environmental data of each road section and the single acceleration average value, the acceleration frequency average value and the driving speed average value corresponding to the historical environmental data of each road section are obtained;
[0059] Set a historical data circular linked list and its linked list length k, obtain the end timestamps of several historical collection cycles in the data storage module, obtain the time span between the end timestamps of several historical collection cycles and the current moment, sort each historical collection cycle in reverse order according to the time span, and according to the sorting results, insert the historical driving routes corresponding to the k historical collection cycles, the historical environmental data of each road section, and the single acceleration average, acceleration frequency average, and driving speed average corresponding to the historical environmental data of each road section into the historical data circular linked list in sequence.
[0060] The smaller the time span, the newer the historical data information within the historical collection period. The smaller the time span, the higher the ranking. The smaller the time span, the better the historical data information. The historical data information within the historical collection period is sorted first. The historical data information within the historical collection period is sorted from front to back according to the time span in the circular linked list. The latest historical collection period will be inserted into the head of the table. When the historical collection period needs to be eliminated, you only need to delete the historical collection period at the end of the linked list, because the historical collection period at the end of the linked list represents the oldest historical collection period.
[0061] It should be further explained that, in the specific implementation process, the model building module constructs the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model according to the historical data information included in the historical data circular linked list, including:
[0062] Based on deep learning, a route prediction model, an environmental data prediction model and a user characteristic behavior prediction model are constructed, and the historical driving routes within k historical collection cycles, the historical environmental data of each road section, and the single acceleration average value, the acceleration frequency average value and the driving speed average value corresponding to the historical environmental data of each road section in the historical data circular linked list are obtained. The historical driving routes within k historical collection cycles, the historical environmental data of each road section, and the single acceleration average value, the acceleration frequency average value and the driving speed average value corresponding to the historical environmental data of each road section are used as the training set and test set of the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model respectively. The route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are trained in real time until the loss function corresponding to the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model is trained stably, and the model parameters are saved. Subsequently, the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are tested through the test set until they meet the preset requirements, and the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are output.
[0063] It should be further explained that, in the specific implementation process, the multi-source data prediction module obtains the multi-source prediction data of the current collection cycle according to the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model, and the process includes:
[0064] Acquire the driving route collected in the current collection period, input the driving route into the route prediction model, and obtain the predicted driving route of the target electric bicycle in the current collection period;
[0065] Input the predicted driving route of the target electric bicycle in the current collection period into the environmental data prediction model, and obtain the environmental prediction data of each section in the predicted driving route in the current collection period according to the environmental data prediction model;
[0066] The environmental prediction data of each section in the predicted driving route of the current collection period is input into the user characteristic behavior prediction model, and the single acceleration average value, acceleration frequency average value and driving speed average value of each section in the predicted driving route of the current collection period are obtained according to the user characteristic behavior prediction model.
[0067] It should be further explained that, in a specific implementation process, the process of the average power consumption evaluation module constructing an average power consumption evaluation model according to the pre-collected average power consumption simulation results includes:
[0068] Acquire the structural parameters, configuration parameters and battery parameters of the target electric bicycle in advance, use Simulink technology to build a dynamic three-dimensional model of the target electric bicycle according to the structural parameters, configuration parameters and battery parameters of the electric bicycle, use a simulator to generate a number of behavioral characteristic data consisting of different single accelerations, acceleration frequencies and driving speeds, as well as different environmental data, use Simulink technology to simulate the dynamic three-dimensional model in different behavioral characteristic data and different environmental data, and generate average power consumption simulation results of the dynamic three-dimensional model in different behavioral characteristic data and different environmental data;
[0069] An average power consumption evaluation model is built based on deep learning, and the average power consumption simulation results of the dynamic three-dimensional model in different behavioral feature data and different environmental data are used as the test set and training set;
[0070] The training set is input into the average power consumption evaluation model for training until the loss function training is stable, and the model parameters are saved. The average power consumption evaluation model is tested by the test set until it meets the preset requirements, and the average power consumption evaluation model is output.
[0071] It should be further explained that, in the specific implementation process, the structural parameters of the electric bicycle include the various components of the electric bicycle, such as the engine, transmission system, wheels, suspension system, etc., and the assembly relationship between the various components of the electric bicycle;
[0072] The configuration parameters of the electric bicycle include the mass of the electric bicycle, the height of the center of gravity, the friction coefficient of the tire, etc.;
[0073] The battery parameters of an electric bicycle include the battery discharge rate of the electric bicycle, battery capacity, battery type (different types of batteries, such as lithium-ion batteries, lead-acid batteries, etc., have different energy densities and performance characteristics, which directly affect the endurance), body weight and battery voltage;
[0074] Environmental data collected by edge terminals include: temperature, wind speed, humidity, and road conditions obtained through GIS;
[0075] It should be further explained that, in the specific implementation process, the deep learning model generally uses a loss function to calculate the loss value between the ground truth data and the predicted probability to quantify the gap between the two. When the loss value is smaller, the classification is more accurate. The present invention uses a logistic regression loss function to calculate the loss values of the route prediction model, the environmental data prediction model, the user characteristic behavior prediction model, and the average power consumption evaluation model, respectively. The formula is:
[0076]
[0077] h(x)=wx+b;
[0078] Among them, (x, y) represents samples, x represents input features, y represents output labels, w represents weight parameters, b represents bias parameters, N represents the number of samples, and D represents a data set of multiple samples.
[0079] It should be further explained that, in the specific implementation process, the endurance evaluation module obtains the endurance mileage evaluation value of the electric bicycle according to the multi-source prediction data of the current collection period, the average power consumption evaluation model and the battery data collected at the current moment, and the process includes:
[0080] Input the average single acceleration value, the average acceleration frequency value and the average driving speed value of each section in the predicted driving route in the current collection period and the environmental prediction data of each section in the predicted driving route into the average power consumption evaluation model, and output the average power consumption evaluation value of the target electric bicycle in each section in the predicted driving route according to the average power consumption evaluation model;
[0081] The battery voltage, battery discharge rate and battery remaining power of the electric bicycle at the current moment are obtained, and the range evaluation value of the electric bicycle is obtained according to the average power consumption evaluation value, battery voltage, battery discharge rate and battery remaining power of each road section.
[0082] Among them, the mileage evaluation value Dist of each road section i =(battery capacity C×battery voltage V×discharge rate θ) / (average power consumption of each section W i );
[0083] Estimated range of electric bicycles
[0084] Among them, Dist represents the estimated value of the cruising range of the electric bicycle, C represents the remaining battery power, V represents the battery voltage, θ represents the battery discharge rate, δ represents the conversion coefficient, and W i represents the average power consumption evaluation value of the i-th road segment, and N represents the total number of road segments included in the predicted driving route.
[0085] It should be further explained that, in the specific implementation process, the endurance evaluation module matches the multi-source prediction data with the real-time collected data information in real-time prediction error, and updates the endurance evaluation value in real time according to the prediction error matching, including:
[0086] Perform real-time prediction error matching on the predicted driving route of the current collection period, the environmental prediction data of each section in the predicted driving route, the average single acceleration, the average acceleration frequency and the average driving speed of each section in the predicted driving route, and the driving route, environmental data, behavioral feature data and battery data of the current collection period;
[0087] If the driving route of the current collection cycle is inconsistent with the predicted driving route, or the environmental data of the section in the driving route of the current collection cycle is inconsistent with the environmental prediction data of the same section in the predicted driving route, or the single acceleration average value, acceleration frequency average value and driving speed average value of the section in the driving route are inconsistent with the single acceleration average value, acceleration frequency average value and driving speed average value of the same section in the predicted driving route, then the collected data information of the current collection cycle is inserted into the historical data circular linked list, and the historical data information of the longest span historical collection cycle in the historical data circular linked list is overwritten by the data information of the current collection cycle, and then the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are retrained according to the historical data information in the historical data circular linked list, and new multi-source prediction data is obtained according to the re-trained route prediction model, environmental data prediction model and user characteristic behavior prediction model, and a new cruising range evaluation value is obtained according to the new multi-source prediction data, the average power consumption evaluation model and the battery data at the current moment, and the original cruising range evaluation value is updated to the new cruising range evaluation value.
[0088] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.
Claims
1. An electric bicycle endurance evaluation system based on multi-source data fusion, comprising a cloud server, characterized in that: The cloud server is communicatively connected to a data acquisition module, a data storage module, a model building module, a multi-source data prediction module, an average power consumption evaluation module, and a battery life evaluation module; The data acquisition module is used to obtain data information collected by the edge terminal deployed on the target electric bicycle, mark the collection time, and set the collection cycle; The data storage module is used to store and count historical data information of the target electric bicycle within several historical collection cycles, and to construct a historical data circular linked list according to the end timestamps of several historical collection cycles; The data storage module stores and counts historical data information of the target electric bicycle within several historical collection cycles, and constructs a historical data circular linked list according to the end timestamps of several historical collection cycles, including: The geographical scope covered by the target electric bicycle's driving route is obtained according to the historical driving route, and all route information within the geographical scope is obtained by GIS means, and the forks on all routes are marked, and the routes between every two forks are marked as sections, and all sections included in the historical driving route are obtained; According to the historical environmental data and historical behavior characteristic data of the target electric bicycle, the historical environmental data of each road section and the single acceleration average value, the acceleration frequency average value and the driving speed average value corresponding to the historical environmental data of each road section are obtained; Set a historical data circular linked list and a linked list length k of the historical data circular linked list, obtain the end timestamps of several historical collection cycles in the data storage module, obtain the time span between the end timestamps of several historical collection cycles and the current moment, sort each historical collection cycle in reverse order according to the time span, and insert the historical driving routes corresponding to the k historical collection cycles, the historical environmental data of each road section, and the single acceleration average value, the acceleration frequency average value, and the driving speed average value corresponding to the historical environmental data of each road section into the historical data circular linked list in sequence according to the sorting results; The model building module is used to build a route prediction model, an environmental data prediction model and a user characteristic behavior prediction model according to the historical data information included in the historical data circular linked list; The process of constructing a route prediction model, an environmental data prediction model and a user characteristic behavior prediction model according to the historical data information included in the historical data circular linked list by the model construction module includes: Based on deep learning, a route prediction model, an environmental data prediction model and a user characteristic behavior prediction model are constructed, and the historical driving routes within k historical collection cycles, the historical environmental data of each road section and the single acceleration average value, the acceleration frequency average value and the driving speed average value corresponding to the historical environmental data of each road section in the historical data circular linked list are obtained. The historical driving routes within k historical collection cycles, the historical environmental data of each road section and the single acceleration average value, the acceleration frequency average value and the driving speed average value corresponding to the historical environmental data of each road section are used as the training set and test set of the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model respectively, and the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are trained in real time until the loss function training corresponding to the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model is stable, and the model parameters are saved, and then the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are tested through the test set until they meet the preset requirements, and the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are output; The multi-source data prediction module is used to obtain multi-source prediction data of the current collection cycle according to the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model; The process of obtaining multi-source prediction data of the current collection period by the multi-source data prediction module according to the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model includes: Acquire the driving route collected in the current collection period, input the driving route into the route prediction model, and obtain the predicted driving route of the target electric bicycle in the current collection period; Input the predicted driving route of the target electric bicycle in the current collection period into the environmental data prediction model, and obtain the environmental prediction data of each section in the predicted driving route in the current collection period according to the environmental data prediction model; Input the environmental prediction data of each section in the predicted driving route of the current collection period into the user characteristic behavior prediction model, and obtain the single acceleration average value, acceleration frequency average value and driving speed average value of each section in the predicted driving route of the current collection period according to the user characteristic behavior prediction model; The average power consumption evaluation module is used to construct an average power consumption evaluation model according to the pre-collected average power consumption simulation results; The process of constructing an average power consumption evaluation model according to the pre-collected average power consumption simulation results by the average power consumption evaluation module includes: Acquire the structural parameters, configuration parameters and battery parameters of the target electric bicycle in advance, use Simulink technology to build a dynamic three-dimensional model of the target electric bicycle according to the structural parameters, configuration parameters and battery parameters of the electric bicycle, use a simulator to generate a number of behavioral characteristic data consisting of different single accelerations, acceleration frequencies and driving speeds, as well as different environmental data, use Simulink technology to simulate the dynamic three-dimensional model in different behavioral characteristic data and different environmental data, and generate average power consumption simulation results of the dynamic three-dimensional model in different behavioral characteristic data and different environmental data; An average power consumption evaluation model is built based on deep learning, and the average power consumption simulation results of the dynamic three-dimensional model in different behavioral feature data and different environmental data are used as the test set and training set; Input the training set into the average power consumption evaluation model for training until the loss function training is stable, and save the model parameters, test the average power consumption evaluation model through the test set until it meets the preset requirements, and output the average power consumption evaluation model; The endurance evaluation module is used to obtain the endurance evaluation value of the electric bicycle according to the multi-source prediction data of the current collection period, the average power consumption evaluation model and the battery data collected at the current moment, and to perform real-time prediction error matching between the multi-source prediction data and the data information collected in real time, wherein the data information collected in real time is the driving route, environmental data, behavioral characteristic data and battery data of the current collection period, and to update the endurance evaluation value in real time according to the prediction error matching; The process of obtaining the range evaluation value of the electric bicycle by the endurance evaluation module according to the multi-source prediction data of the current collection period, the average power consumption evaluation model and the battery data collected at the current moment includes: Input the average single acceleration value, the average acceleration frequency value and the average driving speed value of each section in the predicted driving route in the current collection period and the environmental prediction data of each section in the predicted driving route into the average power consumption evaluation model, and output the average power consumption evaluation value of the target electric bicycle in each section in the predicted driving route according to the average power consumption evaluation model; The battery voltage, battery discharge rate and battery remaining power of the electric bicycle at the current moment are obtained, and the range evaluation value of the electric bicycle is obtained according to the average power consumption evaluation value, battery voltage, battery discharge rate and battery remaining power of each road section.
2. The electric bicycle endurance evaluation system based on multi-source data fusion according to claim 1 is characterized in that: The data information collected by the edge terminal deployed on the target electric bicycle includes the driving route, environmental data, behavioral characteristic data and battery data. The behavioral characteristic data includes the single acceleration size, acceleration frequency and driving speed. The battery data includes battery voltage, battery discharge rate and battery remaining power.
3. The electric bicycle endurance evaluation system based on multi-source data fusion according to claim 2 is characterized in that: The process of the cruising range evaluation module matching the multi-source prediction data with the real-time collected data information with real-time prediction errors and updating the cruising range evaluation value in real time according to the prediction error matching includes: Perform real-time prediction error matching on the predicted driving route of the current collection period, the environmental prediction data of each section in the predicted driving route, the average single acceleration, the average acceleration frequency and the average driving speed of each section in the predicted driving route, and the driving route, environmental data, behavioral feature data and battery data of the current collection period; If the driving route of the current collection cycle is inconsistent with the predicted driving route, or the environmental data of the section in the driving route of the current collection cycle is inconsistent with the environmental prediction data of the same section in the predicted driving route, or the single acceleration average value, acceleration frequency average value and driving speed average value of the section in the driving route are inconsistent with the single acceleration average value, acceleration frequency average value and driving speed average value of the same section in the predicted driving route, then the collected data information of the current collection cycle is inserted into the historical data circular linked list, and the historical data information of the longest span historical collection cycle in the historical data circular linked list is overwritten by the data information of the current collection cycle, and then the route prediction model, the environmental data prediction model and the user characteristic behavior prediction model are retrained according to the historical data information in the historical data circular linked list, and new multi-source prediction data is obtained according to the re-trained route prediction model, environmental data prediction model and user characteristic behavior prediction model, and a new cruising range evaluation value is obtained according to the new multi-source prediction data, the average power consumption evaluation model and the battery data at the current moment, and the original cruising range evaluation value is updated to the new cruising range evaluation value.
