A method, apparatus, and medium for preventing accidental operation of a scooter

Abstract

The application discloses a method, device and medium for preventing accidental operation of a scooter, and relates to the technical field of intelligent safety, and comprises the following steps: collecting a multi-source running state data set, extracting speed change characteristics, acceleration change characteristics, posture tilt characteristics, braking force characteristics, environmental proximity characteristics and temperature characteristics, calculating a real-time risk score value, and comparing the real-time risk score value with a historical scene threshold to obtain a current risk level; obtaining a control parameter set based on the current risk level and the multi-source running state data set and implementing intervention; collecting execution process data to generate event samples, labeling and characterizing to obtain an event feature data set; using an event chain prediction model to deduce a potential risk event chain based on the event feature data set, dynamically reconstructing an intervention window to obtain intervention optimization parameters, and updating the anti-accident control parameter set through a multi-objective optimization strategy to realize continuous optimization. The application can improve the running safety and control response accuracy of the scooter.

Description

Technical Field

[0001] This invention relates to the field of intelligent safety technology, and in particular to a method, device and medium for preventing accidental operation of a scooter. Background Technology

[0002] In recent years, with the rapid growth in urban micro-mobility demand, scooters have become widely used globally as a flexible and environmentally friendly short-distance transportation tool. To improve the safety of scooter use, related technologies have continuously developed, gradually introducing multi-source sensor data acquisition, real-time control strategies, and intelligent auxiliary braking to reduce the risk of accidents caused by driver error, sudden obstacles, or complex environmental conditions. Some advanced accident prevention control methods attempt to combine data such as speed, acceleration, attitude angle, braking force, and surrounding environment perception for comprehensive analysis, and adjust the braking and drive actuators through specific control parameters, thereby achieving active intervention to a certain extent.

[0003] Existing technologies still have the following shortcomings: First, they rely heavily on fixed thresholds to trigger interventions, lacking dynamic prediction and intervention window reconstruction based on future potential risk event chains, resulting in inaccurate response timing and difficulty in balancing safety and comfort. Second, the optimization of anti-accident control parameters has a single objective, lacking multi-objective comprehensive optimization that takes into account safety, comfort, energy consumption, and thermal management, leading to insufficient adaptability under different operating conditions and difficulty in continuously optimizing the intervention effect. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for preventing accidental operation of scooters, which solves the problems of existing methods for preventing accidental operation of scooters lacking a dynamic intervention mechanism based on event chain prediction and having a single objective for optimizing the control parameters for preventing accidents.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for preventing accidental operation of a scooter, comprising,

[0008] Collect multi-source operational status datasets during the operation of scooters, extract multi-source operational status features from the multi-source operational status datasets, and obtain risk score values;

[0009] The current risk level is obtained through the risk score, and the set of control parameters is obtained based on the multi-source operating status dataset and the current risk level.

[0010] Commands are executed by controlling the parameter set to obtain execution process data. Event samples are obtained from the execution process data. Dangerous results are labeled and characterized on the event samples to obtain an event feature dataset.

[0011] Based on the event feature dataset, an event chain prediction model is used to obtain potential risk event chains. Based on the potential risk event chains, the intervention window is reconstructed to obtain intervention optimization parameters.

[0012] Based on the intervention optimization parameters and control parameter sets, a multi-objective optimization strategy is used to calculate and update the anti-accident control parameter set, and obtain the updated control parameters.

[0013] The updated control parameters are transmitted to the scooter via wireless communication, enabling the scooter to continuously optimize its operation to prevent accidents.

[0014] As a preferred embodiment of the method for preventing accidental operation of a scooter according to the present invention, the steps of collecting a multi-source operating status dataset during the operation of the scooter, extracting multi-source operating status features from the multi-source operating status dataset, and obtaining a risk score value are as follows.

[0015] Data on speed, acceleration, attitude angle, braking force, buffer bar position, environmental distance, and braking component temperature of the scooter during operation are collected to obtain a multi-source operating status dataset.

[0016] By using a multi-source operational status dataset, we obtain velocity change features, acceleration change features, attitude tilt features, braking force features, environmental proximity features, and temperature features to obtain multi-source operational status features.

[0017] Based on the characteristics of multi-source operation status, a real-time risk score is calculated by setting weights.

[0018] In a preferred embodiment of the method for preventing accidental operation of a scooter according to the present invention, the step of obtaining the current risk level through a risk score is as follows:

[0019] Real-time risk score data is obtained through real-time risk score values, and historical scenario threshold data is obtained through historical scenario threshold values.

[0020] The real-time risk score data is compared with the historical scenario threshold data to obtain the risk comparison result, and the current risk level is determined based on the risk comparison result.

[0021] As a preferred embodiment of the method for preventing accidental operation of a scooter according to the present invention, the specific steps for obtaining a control parameter set based on a multi-source operating status dataset and the current risk level are as follows:

[0022] The current risk level is matched with the multi-source operational status dataset to calculate the risk feature fusion weight;

[0023] The link reliability coefficient is calculated by combining the risk feature fusion weights with the corresponding feature values ​​in the multi-source operational status dataset.

[0024] Based on the link reliability coefficient and the current risk level, the risk level threshold is determined. Based on multi-source operational status data, risk feature fusion weights, link reliability coefficient, and risk level threshold, a set of control parameters is obtained.

[0025] As a preferred embodiment of the method for preventing accidental operation of a scooter according to the present invention, the steps of executing commands through a control parameter set, obtaining execution process data, obtaining event samples through the execution process data, labeling and characterizing the event samples to obtain an event feature dataset are as follows.

[0026] The control parameter set is sent to the left and right brakes, front and rear drive modules and buffer devices to intervene and obtain execution process data.

[0027] Read the time series information of the execution process data, obtain the execution process time series dataset, and extract the feature change segments related to abnormal operations or potential risk events for feature label matching to obtain event samples;

[0028] Based on event samples, hazard outcome labeling and feature processing are performed to obtain an event feature dataset. As a preferred embodiment of the scooter accident prevention method of the present invention, wherein:

[0029] Based on the event feature dataset, potential risk event chains are obtained. Then, the intervention window is reconstructed according to these chains to obtain optimal intervention parameters. The specific steps are as follows.

[0030] Based on the event feature dataset, potential risk event chains are obtained by performing step-by-step calculations according to a preset time window.

[0031] The intervention window is dynamically reconstructed based on the potential risk event chain to obtain intervention optimization parameters.

[0032] As a preferred embodiment of the method for preventing accidental operation of a scooter according to the present invention, the steps of calculating and updating the anti-accident control parameter set using a multi-objective optimization strategy based on the intervention optimization parameters and control parameter set, and obtaining the updated control parameters, are as follows:

[0033] By mapping the intervention optimization parameters to the control parameter set and synthesizing the weights, the initial solution of the anti-accident control parameter set is obtained;

[0034] Using the initial solution of the anti-accident control parameter set as the optimization variable, the comprehensive optimization objective value is calculated and iteratively solved according to the multi-objective optimization strategy to obtain the candidate anti-accident control parameter set;

[0035] The candidate set of control parameters for accident prevention is constrained and its stability is checked, and then it is fused with the control parameter set to obtain the updated control parameters.

[0036] In a preferred embodiment of the method for preventing accidental operation of a scooter according to the present invention, the updated control parameters are transmitted to the control system via wireless communication for continuous optimization of the anti-accidental operation method. The specific steps are as follows.

[0037] Based on the updated control parameters, the data is transmitted to the scooter via wireless communication. The updated control parameters are then used in subsequent data acquisition and control processes to continuously optimize the anti-accident operation effect.

[0038] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the method for preventing accidental operation of a scooter as described in the first aspect of the present invention.

[0039] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the method for preventing accidental operation of a scooter as described in the first aspect of the present invention.

[0040] The beneficial effects of this invention are as follows: by using an event chain prediction model based on event feature datasets to extrapolate future potential risk event chains, the intervention window can be reconstructed in advance to improve the accuracy of intervention response; by mapping and weighting the intervention optimization parameters and control parameter sets, the adaptability of the anti-accident control parameter set under multiple operating conditions can be improved; by updating the anti-accident control parameter set based on a multi-objective optimization strategy of safety, comfort, energy consumption and thermal management, the continuous optimization of intervention effect and the balanced improvement of comprehensive performance can be achieved. Attached Figure Description

[0041] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 A flowchart of methods to prevent accidental operation of scooters.

[0043] Figure 2A flowchart for predicting event chains from an event feature dataset.

[0044] Figure 3 A flowchart for reconstructing the intervention window and obtaining intervention optimization parameters.

[0045] Figure 4 A flowchart for the mapping and optimization of intervention optimization parameters and control parameter sets. Detailed Implementation

[0046] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0047] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0048] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0049] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a method for preventing accidental operation of a scooter, comprising the following steps:

[0050] S1: Collect multi-source operating status datasets during the operation of the scooter, extract multi-source operating status features from the multi-source operating status datasets, and obtain risk score values.

[0051] Data on speed, acceleration, attitude angle, braking force, buffer bar position, environmental distance, and braking component temperature of the scooter during operation are collected to obtain a multi-source operating status dataset.

[0052] Furthermore, the speed data collected during the scooter's operation, the acceleration data collected by the accelerometer, the attitude angle data collected by the attitude angle sensor, the braking force data collected by the braking force sensor, the buffer rod position data collected by the buffer rod position sensor, the environmental distance data collected by the environmental distance sensor, and the braking component temperature data collected by the temperature sensor are merged according to the same timestamp to obtain a multi-source operating status dataset.

[0053] By using a multi-source operational status dataset, we can obtain characteristics of velocity change, acceleration change, attitude tilt, braking force, environmental proximity, and temperature to obtain multi-source operational status features.

[0054] Furthermore, speed data, acceleration data, attitude angle data, braking force data, buffer rod position data, environmental distance data, and braking component temperature data are collected in time series and differential calculations are performed to obtain the rate of change sequence. From the rate of change sequence, characteristic parameters such as mean, variance, peak value, root mean square value, and change amplitude are extracted to form speed change characteristics, acceleration change characteristics, attitude tilt characteristics, braking force characteristics, environmental proximity characteristics, and temperature characteristics, respectively, thus constructing multi-source operating state characteristics.

[0055] Based on the characteristics of multi-source operation status, a real-time risk score is calculated by setting weights.

[0056] Furthermore, multi-source operational status feature data is read, and preset weight parameters are obtained by calculating the correlation between historical event feature data and accident results and normalizing the data. Multi-source operational status feature data is then matched one-to-one with the preset weight parameters to form matching results. These matching results are then weighted to obtain a weighted feature value set. Finally, the real-time risk score R is calculated using this weighted feature set. s The expression is:

[0057]

[0058] Among them, f i w represents the feature value of the i-th feature in the multi-source operational status feature data. i σ represents the weight of the i-th feature in the multi-source operational status feature data. link denoted as the link difference metric, γ represents the link difference suppression coefficient, and σ(x) represents the normalization function.

[0059] It should be noted that the link difference suppression coefficient γ is calibrated through regression analysis based on historical multi-link operation data and corresponding risk levels. The true risk level is obtained through statistical analysis of historical operation data and accident records, minimizing the error between the risk score and the true risk level. The link difference measure σ... link This reflects the degree of difference between multiple key data acquisition links, at the same time t. k The physical quantities obtained from the two links are measured by the normalized root mean square error, expressed as:

[0060]

[0061] in, This represents the value of link 1 in the j-th feature. This represents the value of link 2 in the j-th feature, where n represents the number of features, and σ j Let represent the standard deviation of the j-th feature in the historical samples.

[0062] S2: Obtain the current risk level through the risk score, and obtain the control parameter set based on the multi-source operating status dataset and the current risk level.

[0063] The current risk level is obtained by comparing the real-time risk score with historical scenario thresholds.

[0064] Furthermore, real-time risk score data is obtained through real-time risk scores, and historical scenario threshold data is obtained through historical scenario thresholds, including low-risk and high-risk thresholds. The real-time risk score data is compared with the historical scenario threshold data to obtain the risk comparison result. Based on the risk comparison result, the current risk level is determined, resulting in the current risk level data L. r .

[0065] For example, when the current risk level is 1, it means that the real-time risk score is less than the low risk threshold; when the current risk level is 2, it means that the real-time risk score is greater than or equal to the low risk threshold and less than the high risk threshold; and when the current risk level is 3, it means that the real-time risk score is greater than or equal to the high risk threshold.

[0066] The low-risk threshold and high-risk threshold are obtained by performing cluster analysis or quantile statistics on the distribution of real-time risk scores in historical operational datasets, selecting the boundary values ​​that can effectively distinguish between low-risk, medium-risk, and high-risk scenarios.

[0067] Obtain the corresponding set of control parameters based on the current risk level and multi-source operational status dataset.

[0068] Furthermore, the numerical variation amplitudes of velocity change features, acceleration change features, attitude tilt features, braking force features, environmental proximity features, and temperature features are obtained from the multi-source operating status dataset. These numerical variation amplitudes are compared with risk level threshold ranges determined based on historical data. Relevant features are selected based on the risk level range they fall into, and feature combinations corresponding to the current risk level are determined. The risk feature fusion weight is calculated, expressed as:

[0069]

[0070] Among them, w i r represents the risk feature fusion weight of the i-th feature. i The stability factor represents the i-th feature. Indicates the current risk level L rThe corresponding feature importance coefficient, where k represents the temperature coefficient.

[0071] It should be noted that, By statistically analyzing historical event characteristics, data is concentrated at risk level L. r The correlation coefficients between the following features and the accident outcome, after normalization, are used to determine the feature importance coefficients. k is obtained by measuring scooter operation data under different temperature conditions, analyzing the impact of temperature changes on risk score values ​​or control parameter deviations, and using regression fitting to obtain the temperature coefficient. i The stability factor of the i-th feature is expressed as:

[0072]

[0073] Where, σ i ε represents the standard deviation of the i-th feature calculated from the multi-source running state dataset within the sliding time window, and ε represents a positive number to prevent the denominator from being zero.

[0074] The link reliability coefficient η is calculated by combining the risk feature fusion weights with the corresponding feature values ​​in the multi-source operational status dataset. link The expression is:

[0075]

[0076] Where T represents the number of sampling points within the sampling time window, w i f represents the risk feature fusion weight of the i-th feature. i (t) represents the i-th feature value collected at time t. F represents the statistical reference value of the i-th feature under historical stable operating conditions. max This represents the theoretical maximum difference among all features after normalization.

[0077] It should be noted that the statistical reference values ​​are obtained by calculating the mean or median of historical operating data within a stable range.

[0078] Based on the link reliability coefficient and the current risk level, the risk level threshold is determined as follows:

[0079]

[0080] Among them, T low T represents the low-risk threshold in the risk level threshold. high This represents the high-risk threshold within the risk level threshold. This represents the high-risk threshold benchmark in historical scenario thresholds. η represents the low-risk threshold benchmark in historical scenario thresholds. linkδ(L) represents the link reliability coefficient, k represents the link reliability coefficient weight, and δ(L) represents the link reliability coefficient weight. r ) represents the current risk level hysteresis compensation term, ε represents the gap amount, and clip(x,a,b) means controlling the value within the interval [a,b].

[0081] It should be noted that η link By comparing the characteristic changes of the link during operation with historical stable characteristic reference values, the average normalized difference of the characteristic deviation is statistically calculated, and the result of subtracting this difference from 1 is used as the link reliability coefficient. The current risk level hysteresis compensation term δ(L) is calculated. r The risk score fluctuation range during different risk level switching is statistically analyzed from historical multi-source operational status datasets and risk level change records. The risk score fluctuation range is set after balancing the avoidance of frequent switching and the timeliness of response. The link reliability coefficient weight k is a fixed value obtained by using a multi-objective optimization method to establish a weighted evaluation function among safety indicators, comfort indicators and energy consumption indicators in historical multi-source operational status datasets and risk event occurrence records. The parameter search is performed with the goal of minimizing the risk misjudgment rate and response delay. The gap quantity ε is obtained by analyzing historical multi-source operational status datasets and risk level determination records, statistically analyzing the minimum difference distribution between high-risk threshold and low-risk threshold under stable operating conditions, and selecting the minimum difference of stable samples as a fixed value while ensuring that risk level switching does not overlap.

[0082] By combining the risk level threshold with the braking force characteristics, buffer rod position characteristics, and temperature characteristics in the multi-source operating status dataset, braking and buffer parameters are determined. These braking and buffer parameters, along with the risk feature fusion weights, link reliability coefficients, risk level thresholds, and braking and buffer parameters, constitute a set of control parameters.

[0083] S3: Execute commands by controlling the parameter set, obtain execution process data, obtain event samples through the execution process data, perform dangerous result labeling and feature processing on the event samples, and obtain event feature dataset.

[0084] The control parameter set is sent to control the left and right brakes, front and rear drive modules and buffer devices for intervention.

[0085] Furthermore, by controlling the parameter set, the current risk level, and the multi-source operational status dataset, the intervention allocation ratio is calculated, expressed as:

[0086]

[0087] Where, π u Indicates the intervention allocation ratio, u represents the channel type identifier, and s represents the channel type identifier. u,0 s represents the allocation reference coefficient for channel u.u,1 L represents the sensitivity coefficient for channel u. r represents the current risk level, and v represents the channel traversal variable in the denominator summation process.

[0088] It should be noted that s u,0 This represents the basic allocation of channel u when the risk level is zero. It is obtained through regression analysis of historical multi-source operational status datasets and actual intervention effect data. u,1 This indicates that the allocation amount of channel u varies with the risk level L. r The increased rate of change was obtained through statistical regression of historical risk levels and intervention allocation ratios. The channel type identifiers include left and right brakes (BRK), front and rear drive modules (DRV), and buffer devices (BUF). The channel traversal variable is used to uniformly weight and sum the allocation baseline coefficients and allocation sensitivity coefficients of the three channels.

[0089] The intervention control target value is obtained by multiplying the parameter in the control parameter set with the intervention allocation ratio corresponding to the parameter. The intervention control target value is then sent to the left and right brakes, front and rear drive modules and buffer devices to complete the intervention operation of the left and right brakes, front and rear drive modules and buffer devices.

[0090] When performing intervention operations, acquire execution process data.

[0091] Furthermore, during the intervention control process, the actual output of the displacement of the left brake, right brake, front drive module, rear drive module and buffer device is recorded in real time by sensors to obtain actuator output data. The speed, acceleration, attitude angle, braking force, buffer rod position, ambient distance and braking component temperature generated during the execution process are collected to obtain the intervention operation data. The intervention operation data is synchronized with the actuator output data in time to form the execution process data.

[0092] Event samples are obtained by executing process data.

[0093] Furthermore, by reading the time-series information of speed, acceleration, attitude angle, braking force, buffer rod position, ambient distance, and braking component temperature recorded in the execution process data, an execution process time-series dataset is obtained. Based on the execution process time-series dataset, statistical analysis is performed on the speed, acceleration, attitude angle, braking force, buffer rod position, ambient distance, and braking component temperature in the execution process time-series dataset to detect their rate of change, deviation value, and out-of-limit situations. When any feature shows a sudden change in a continuous period (e.g., the speed drops by more than 3 m / s within 1 second, or the attitude angle deviates instantaneously by more than 15°), exceeds the normal range (e.g., the braking component temperature exceeds 80°), or is inconsistent with the actuator output data, the continuous period is divided into feature change segments to form a feature change segment set. The feature change segment set is matched with the execution process time-series dataset for feature labels to generate event samples.

[0094] Based on the event samples, dangerous results are labeled and characterized to obtain an event feature dataset.

[0095] Furthermore, the event sample features such as velocity, acceleration, attitude angle, braking force, buffer rod position, ambient distance, and braking component temperature are read to form an event sample feature set. Based on the event sample feature set and the actual hazard results, each event sample is labeled with a hazard result, forming an event sample feature set with hazard result labels. The event sample feature set with hazard result labels is then normalized, time-windowed, and feature vectors are extracted to form an event feature dataset.

[0096] S4: Based on the event feature dataset, obtain the potential risk event chain, reconstruct the intervention window according to the potential risk event chain, and obtain the intervention optimization parameters.

[0097] Based on the event feature dataset, potential risk event chains are obtained by performing step-by-step calculations according to a preset time window.

[0098] Furthermore, based on the event feature dataset, time series of events are formed by aligning with timestamps and resampling at a uniform sampling period. By combining braking response time with the duration of historical accidents and comparing the accuracy and timeliness of risk prediction under different time windows, the interval with the best performance is selected as the preset time window. The event feature time series and the preset time window are truncated to form the event chain prediction input sequence. The event chain prediction input sequence is input into the event chain prediction model for forward calculation to obtain the event chain prediction intensity sequence. The cumulative risk value of the event chain is calculated using the discrete window integral form, as expressed in the following expression:

[0099]

[0100] Where λ[m] represents the intensity of the event chain prediction intensity sequence at the m-th sampling point, x i [m] represents the i-th feature value of the event chain prediction input sequence at the m-th sampling point, d represents the feature dimension, and θ i The feature weight coefficients of the event chain prediction model are represented, with values ​​ranging from [0,1]. μ represents the baseline strength, Δt represents the sampling interval, M represents the number of sampling points within the preset time window, and R... chain This represents the cumulative risk value of the event chain.

[0101] It should be noted that the feature dimension d is determined by the number of independent feature types contained in the event feature dataset. For example, when the event feature dataset contains velocity change features, acceleration change features, attitude tilt features, braking force features, environmental proximity features, and temperature features, d = 6, θ i It is obtained by supervised learning training on the historical event feature dataset. The value range is set to [0,1] because the importance of different features needs to be normalized in the event chain prediction model to ensure that the sum of all feature weight coefficients can be used for probability or proportion calculation, thereby avoiding a certain feature from dominating the prediction result due to excessive value. The baseline strength μ is obtained by statistical calculation of the average occurrence intensity of similar risk events in the historical event feature dataset under no-intervention conditions.

[0102] The cumulative risk value of the event chain is matched one by one with the predicted intensity sequence of the event chain. By statistically analyzing the distribution of the cumulative risk value of the historical event chain, and combining it with the analysis of the frequency of accident occurrence in different risk value intervals using labeled accident samples, when the probability of accident occurrence shows a significant upward trend in a certain value interval, the starting point of that value interval is used as the event chain trigger threshold. Based on the event chain trigger threshold, the analysis is performed. When the cumulative risk value of the event chain is greater than or equal to the event chain trigger threshold and the corresponding event chain predicted intensity sequence value increases continuously in the time dimension, the event chain predicted intensity sequences that meet the conditions are sorted in chronological order to form a potential risk event chain arranged in chronological order and output it for subsequent dynamic reconstruction of the intervention window and adjustment of the accident prevention and control strategy.

[0103] The intervention window is dynamically reconstructed based on the potential risk event chain to obtain intervention optimization parameters.

[0104] Furthermore, based on the temporal distribution and cumulative risk value of events in the potential risk event chain, the start time and duration of the intervention window are adjusted to complete the dynamic reconstruction of the intervention window, expressed as:

[0105]

[0106] Among them, W new W represents the duration of the intervention window after dynamic reconstruction.base R represents the initial intervention window duration, α represents the intervention adjustment sensitivity coefficient, and its value ranges from [0,1]. cmn R represents the cumulative risk value of a potential risk event chain, G represents the risk scoring threshold, and R represents the risk score threshold. max T represents the theoretical maximum value of the cumulative risk. event T represents the expected time of occurrence of the first high-risk event in a potential risk event chain. now Indicates the current time.

[0107] It should be noted that the intervention adjustment sensitivity coefficient α is determined through regression analysis of historical intervention effects and risk response efficiency. If α < 0, it will lead to a reversal of the intervention direction, violating the risk suppression objective. If α > 1, it will lead to the intervention magnitude exceeding the system safety margin, which may cause over-braking or false triggering of intervention, thereby reducing comfort and safety. The risk score threshold G is determined by combining historical risk event data statistics with safety specification requirements. It is the minimum risk score threshold required to trigger the anti-accident control action and is determined after experimental calibration and safety margin correction.

[0108] The dynamically reconstructed intervention window is matched with the risk characteristics of the potential risk event chain to generate intervention optimization parameters.

[0109] S5: Based on the event feature dataset, use a multi-objective optimization strategy to calculate and update the anti-accident control parameter set, and obtain the updated control parameters;

[0110] By mapping the intervention optimization parameters to the control parameter set and synthesizing the weights, an initial solution for the anti-accident control parameter set is obtained.

[0111] Furthermore, by using intervention optimization coefficients, a parameter mapping relationship matrix between intervention optimization parameters and control parameter sets is established. The intervention optimization parameters are mapped to the parameter dimensions corresponding to the control parameter sets to form the mapped control parameter sets. The control parameter sets are then weighted and synthesized according to preset weight coefficients to obtain the initial solution of the anti-accident control parameter set.

[0112] Using the initial solution of the anti-accident control parameter set, and based on a multi-objective optimization strategy considering safety, comfort, energy consumption, and thermal management, the anti-accident control parameter set is adjusted to obtain the updated control parameters.

[0113] Furthermore, by utilizing the velocity change features, acceleration change features, attitude tilt features, braking force features, environmental proximity features, and temperature features from the event feature dataset, an event feature input set is constructed. This input set is then substituted into the multi-objective optimization strategies for safety, comfort, energy consumption, and thermal management, respectively, to calculate the performance index values ​​for each optimization objective. The expressions are as follows:

[0114]

[0115] Among them, J total Let x ∈ {s,c,e,h} represent the comprehensive optimization target value, where x ∈ {s,c,e,h} represents the four targets of safety, comfort, energy consumption, and thermal management, respectively. z represents the target value obtained by normalizing the event feature dataset. x φ represents the reference baseline value for each target. x P represents adaptive weights. cand P represents the candidate set of control parameters for preventing accidents. ori This represents the original set of anti-accident control parameters, and μ represents the parameter update smoothing coefficient.

[0116] Among them, z x This involves statistically analyzing historical event feature datasets across the corresponding target dimensions, extracting the values ​​at designated quantiles for each target, and using these values ​​as a baseline reference for that target. (P) cand It is the combination of multiple sets of anti-accident control parameters obtained after inputting the event feature dataset into a multi-objective optimization strategy based on safety, comfort, energy consumption, and thermal management, P. ori It is the initial control parameter combination determined based on historical operating data, factory safety calibration values, and regulatory requirements during the initial operation of the scooter. φ x Indicates based on the current risk level L r With link reliability coefficient η link The calculated adaptive weights are expressed as follows:

[0117]

[0118] in, Indicates the current risk level L r The priority score of the i-th optimization objective is given below. Indicates the current risk level L r The priority score of the u-th optimization objective.

[0119] It should be noted that the priority score is a statistical measure of the impact of each optimization objective on the probability of an accident under different risk levels.

[0120] Based on the comprehensive optimization objective value, the set of accident prevention control parameters P ori Adjustments are made by statistically analyzing the historical event feature dataset, extracting the corresponding target values ​​at set quantiles (e.g., the 95th percentile) as reference baseline values, calculating the deviations between each optimized target value and the reference baseline value, and combining adaptive weights with the anti-accident control parameter set P. ori Perform weighted correction to obtain the updated control parameter P. new This enables subsequent interventions to achieve a balance between safety, comfort, energy consumption, and thermal management.

[0121] S6: The updated control parameters are transmitted to the control system via wireless communication for continuous optimization to prevent unexpected operations.

[0122] Based on the updated control parameters, the data is written back to the scooter control system via wireless communication.

[0123] Furthermore, the updated control parameters are input into the wireless communication data encapsulation process to generate wireless transmission data packets. The wireless transmission data packets are then encrypted through an encryption encoding process and sent to the receiving end of the scooter control system. The receiving end of the scooter control system obtains the updated control parameters through a decryption and decoding process.

[0124] The updated control parameters are used in subsequent data acquisition and control processes to continuously optimize the effectiveness of preventing accidental operation.

[0125] Furthermore, during subsequent scooter operation, the updated control parameters are continuously used as the core input for the collection of multi-source operating status data and real-time control decisions. The collected multi-source operating status data is combined with the updated control parameters to generate a real-time risk score, which is dynamically compared with historical scenario thresholds to obtain the current risk level. Based on the current risk level, targeted intervention operations are performed on the left and right brakes, front and rear drive modules, and buffer devices. At the same time, the execution process data is recorded and transformed into an event feature dataset. Through a multi-objective optimization strategy, the anti-accident control parameter set is iteratively updated, enabling the control strategy to continuously adapt to changing operating conditions and driving characteristics, thereby achieving continuous optimization and stable improvement of the anti-accident operation effect.

[0126] This embodiment also provides a computer device applicable to the method for preventing accidental operation of a scooter, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for preventing accidental operation of a scooter as proposed in the above embodiment.

[0127] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0128] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for preventing accidental operation of a scooter as proposed in the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0129] In summary, this invention achieves improved accuracy of intervention response by: using an event chain prediction model based on event feature datasets to predict future potential risk event chains; mapping and weighting intervention optimization parameters with control parameter sets to enhance the adaptability of the anti-accident control parameter set under multiple operating conditions; and updating the anti-accident control parameter set based on a multi-objective optimization strategy considering safety, comfort, energy consumption, and thermal management to achieve continuous optimization of intervention effects and balanced improvement of overall performance.

[0130] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method of preventing accidental operation of a scooter, characterized by: include, Collect multi-source operational status datasets during the operation of scooters, extract multi-source operational status features from the multi-source operational status datasets, and obtain risk score values; The current risk level is obtained through the risk score, and the set of control parameters is obtained based on the multi-source operating status dataset and the current risk level. Commands are executed by controlling the parameter set to obtain execution process data. Event samples are obtained from the execution process data. Dangerous results are labeled and characterized on the event samples to obtain an event feature dataset. Based on the event feature dataset, obtain the potential risk event chain, reconstruct the intervention window according to the potential risk event chain, and obtain the intervention optimization parameters; Based on the intervention optimization parameters and control parameter sets, a multi-objective optimization strategy is used to calculate and update the anti-accident control parameter set, and obtain the updated control parameters. The updated control parameters are transmitted to the scooter via wireless communication, enabling the scooter to continuously optimize its operation to prevent accidents.

2. The method of preventing accidental operation of the scooter of claim 1, wherein: The process involves collecting a multi-source operational status dataset during the scooter's operation, extracting multi-source operational status features from this dataset, and obtaining a risk score. The specific steps are as follows: Data on speed, acceleration, attitude angle, braking force, buffer bar position, environmental distance, and braking component temperature of the scooter during operation are collected to obtain a multi-source operating status dataset. By using a multi-source operational status dataset, we obtain velocity change features, acceleration change features, attitude tilt features, braking force features, environmental proximity features, and temperature features to obtain multi-source operational status features. Based on the characteristics of multi-source operation status, a real-time risk score is calculated by setting weights.

3. The method of claim 2, wherein: The specific steps for obtaining the current risk level through a risk score are as follows: Real-time risk score data is obtained through real-time risk score values, and historical scenario threshold data is obtained through historical scenario threshold values. The real-time risk score data is compared with the historical scenario threshold data to obtain the risk comparison result, and the current risk level is determined based on the risk comparison result.

4. The method of preventing accidental operation of the scooter of claim 3, wherein: The specific steps for obtaining the control parameter set based on the multi-source operational status dataset and the current risk level are as follows: The current risk level is matched with the multi-source operational status dataset to calculate the risk feature fusion weight; The link reliability coefficient is calculated by combining the risk feature fusion weights with the corresponding feature values ​​in the multi-source operational status dataset. Based on the link reliability coefficient and the current risk level, the risk level threshold is determined. Based on multi-source operational status data, risk feature fusion weights, link reliability coefficient, and risk level threshold, a set of control parameters is obtained.

5. The method of preventing accidental operation of the scooter of claim 4, wherein: The steps involve executing commands through a control parameter set, obtaining execution process data, acquiring event samples from the execution process data, labeling and characterizing the event samples to obtain an event feature dataset. The control parameter set is sent to the left and right brakes, front and rear drive modules and buffer devices to intervene and obtain execution process data. Read the time series information of the execution process data, obtain the execution process time series dataset, and extract the feature change segments related to abnormal operations or potential risk events for feature label matching to obtain event samples; Based on the event samples, dangerous results are labeled and characterized to obtain an event feature dataset.

6. The method of preventing accidental operation of the scooter of claim 5, wherein: The process involves obtaining potential risk event chains based on event feature datasets, reconstructing intervention windows based on these chains, and obtaining intervention optimization parameters. The specific steps are as follows: Based on the event feature dataset, potential risk event chains are obtained by performing step-by-step calculations according to a preset time window. The intervention window is dynamically reconstructed based on the potential risk event chain to obtain intervention optimization parameters.

7. The method of preventing accidental operation of a scooter of claim 6, wherein: The steps involve calculating and updating the anti-accident control parameter set using a multi-objective optimization strategy based on the intervention optimization parameters and control parameter set, and obtaining the updated control parameters. By mapping the intervention optimization parameters to the control parameter set and synthesizing the weights, the initial solution of the anti-accident control parameter set is obtained; Using the initial solution of the anti-accident control parameter set as the optimization variable, the comprehensive optimization objective value is calculated and iteratively solved according to the multi-objective optimization strategy to obtain the candidate anti-accident control parameter set; The candidate set of control parameters for accident prevention is constrained and its stability is checked, and then it is fused with the control parameter set to obtain the updated control parameters.

8. The method of preventing accidental operation of a scooter of claim 7, wherein: The updated control parameters are transmitted to the scooter via wireless communication for continuous optimization to prevent accidental operation. The specific steps are as follows: Based on the updated control parameters, the data is transmitted to the scooter via wireless communication. The updated control parameters are then used in subsequent data acquisition and control processes to continuously optimize the anti-accident operation effect. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: When the processor executes the computer program, it implements the steps of the method for preventing accidental operation of a scooter as described in any one of claims 1 to 8.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: When the computer program is executed by the processor, it implements the steps of the method for preventing accidental operation of the scooter as described in any one of claims 1 to 8.

Images